instruction stringlengths 34 186 | input stringlengths 2.02k 93.8k | output stringlengths 2 418 | meta_questiontype stringclasses 6
values | meta_inputlicense stringclasses 6
values | meta_pmid stringlengths 8 8 | meta_safetyreportid int64 9.51M 21M | meta_articlepubdate stringlengths 4 10 |
|---|---|---|---|---|---|---|---|
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Disease recurrence'. | Case Report: Successful Treatment of Cutaneous Squamous Cell Carcinoma in Three Patients With a Combination of Acitretin and Clarithromycin.
Few studies have evaluated acitretin as a primary treatment for cutaneous squamous cell carcinoma (CSCC). We report, for the first time, three cases of CSCC successfully treated with acitretin and clarithromycin. A literature review on this subject was also was performed. This case report included three patients with CSCC treated with acitretin and clarithromycin at the First Affiliated Hospital of Fujian Medical University (2008-2019). Patient 1 (83-year-old woman, ulcerated mass on the left cheek), patient 2 (97-year-old woman, painful mass on the left cheek) and patient 3 (76-year-old woman, large mass on the right ankle) received 8, 6, and 30 courses of combination therapy. All patients tolerated the adverse effects (pseudotumor cerebri and mucocutaneous dryness) and achieved complete regression within 6 months. Patients 1, 2, and 3 have not experienced recurrence during a 10-, 3-, and 6-year follow-up. Acitretin has limited efficacy as a monotherapy for CSCC. Our experience indicates that combination therapy with acitretin and clarithromycin may be an effective and well-tolerated treatment for unresectable CSCC.
Introduction
Given its increasing incidence and potential for poor outcomes (1–3), cutaneous squamous cell carcinoma (CSCC) is emerging as a public health problem. High-risk CSCC can be difficult to treat by surgical excision because the neoplasm can extend or infiltrate beyond the visible borders of the lesion. A wide range of non-surgical treatments are available for high-risk CSCC. Acitretin is a second generation retinoid shown to reduce the incidence of new primary non-melanoma skin cancers in immunocompromised transplant recipients (4) and patients receiving BRAF inhibitor treatment (5). Acitretin is considered to have limited efficacy as a monotherapy for CSCC but seems more effective when used in combination with other agents (6, 7). Clarithromycin is an antibacterial drug that has been reported to enhance the anti-tumor activity of chemotherapeutic drugs in vitro and in vivo (8, 9). However, no previous study has evaluated the combination of acitretin and clarithromycin in the treatment of CSCC. We present 3 cases illustrating that combination therapy with acitretin and clarithromycin might be a promising option for patients with CSCC who are unwilling or unable to undergo surgery.
Case Description
Ethical Statement
The study was approved by the Medical Technology Clinical Application Ethics Committee of the First Affiliated Hospital of Fujian (No. [2015]084). Before treatment, the patients were fully informed about the potential advantages and disadvantages of combination therapy with acitretin and clarithromycin, including the possible side effects of these drugs. All patients provided written informed consent for treatment and agreed to attend follow-up visits every 2 weeks. All patients provided written informed consent for data analysis and publication.
Patients
The present case report represents a retrospective analysis of the medical records of patients with CSCC who were treated with a combination of acitretin and clarithromycin at the Dermatology Department of the First Affiliated Hospital of Fujian between 2008 and 2018. Inclusion criteria consisted of a diagnosis of SCC by pathological biopsy and a history as well as tumor diameter ≥1 cm (10).
A total of 3 patients were involved in our study. Acitretin and clarithromycin were administered together to the patients from the time when skin biopsy confirmed the diagnosis of invasive SCC. We defined oral 30 mg/day of acitretin for 4 weeks and oral 500 mg/day of clarithromycin for 3 weeks as 1-month treatment course. There was no interval time between courses.
Once biochemical investigations, including measurements of serum creatinine, cholesterol, triglycerides, and potassium appeared abnormal, the treatment was stopped or the dosage was reduced.
Patient 1
Patient 1 was an 83-year-old Chinese woman with a 2-year history of an asymptomatic ulcerated mass on her left cheek. Physical examination in March 2008 revealed a 4 × 3.5 cm crater-like mass with central ulceration on her left cheek (Figure 1A). Skin biopsy revealed the presence of atypical epithelial tumor cell formations extending beyond the epidermis into the underlying dermis, cornified cells and horny pearls, thereby confirming the diagnosis of invasive SCC. The patient showed a favorable response to acitretin and clarithromycin within 2 months after the first oral administration. The patient declined long-term treatment due to the development of adverse effects (pseudotumor cerebri and dryness of the lips and skin). Therefore, CO2 laser resection of the CSCC was performed after a total of 6 courses of combination therapy, and the patient then received an additional two courses of combination therapy. This patient has lived without tumor recurrence for 10 years.
Figure 1 Photographs showing the regression of cutaneous squamous cell carcinoma (CSCC) in three patients treated with acitretin and clarithromycin. (A) Patient 1. The CSCC on the left cheek is shown at month 0 (before initiation of combination therapy), 5 and 12. (B) Patient 2. The CSCC on the left cheek is shown at month 0, 3, and 29. (C) Patient 3. The CSCC on the right ankle is shown at month 0, 6, 30, 42, and 60.
Patient 2
Patient 2 was a 97-year-old Chinese woman with a 1-year history of a painful mass on her left cheek. Physical examination in November 2015 demonstrated the presence of a 2 × 2 cm mass with an irregular edge and signs of erosion on her left cheek (Figure 1B). Skin biopsy confirmed the diagnosis of invasive SCC. The patient was treated with 6 courses of combination therapy. Darkening and regression of the tumor lesions were first detected in 2 months after the initiation of combination therapy, and a complete response was observed at 6 months. The only adverse effect reported by the patient was mucocutaneous dryness, but this resolved following cessation of treatment with acitretin. The patient has not experienced recurrence of CSCC during the 3-year follow-up.
Patient 3
Patient 3 was a 76-year-old Chinese woman with a 1-year history of a large mass on her right ankle that had progressively increased in size. Physical examination in September 2012 revealed a 4 × 4 cm mass with erosion on her right ankle (Figure 1C). Skin biopsy showed that the overlying epidermis was ulcerated with atypical keratinocytes invading the dermis. The keratinocytes were larger in the central parts of the tumor nests and formed eosinophilic horny pearls. The neoplastic keratinocytes had abundant eosinophilic and dyskeratotic cytoplasm and large vesicular nuclei with visible nucleoli and mitoses. An inflammatory reaction was evident, with many lymphocytes surrounding the stromal vessels and penetrating the tumor nests (Figure 2). Immunohistochemical analyses demonstrated that the tumor cells were positive for AE1/AE3 and P63, negative for Ber-EP4, and positive for Ki67 in about 40% of cells, confirming the diagnosis of invasive SCC (Figure 3). The patient was initially treated with 6 courses of combination therapy, which led to substantial regression of the tumor such that it resembled a small residual ulcer, and the patient then decided to discontinue the treatment. A local tumor relapse was detected 2 years later, so the patient resumed combination therapy and received an additional 24 courses of treatment, which resulted in complete remission. The only adverse effect experienced by the patient was mucocutaneous dryness. The patient remains alive and without tumor recurrence after a 6-year follow-up.
Figure 2 Representative images showing the histologic appearance of a tumor specimen from patient 3. (A) Section stained with hematoxylin-eosin (magnification, ×25). (B) Section stained with hematoxylin-eosin (magnification, ×200). (C) Section stained with hematoxylin-eosin (magnification, ×400).
Figure 3 Immunohistology of tumor specimens from patient 3. (A) Tumor cells were positive for AE1/AE3 (magnification, ×25). (B) Tumor cells were positive for P63 (magnification, ×25). (C) Tumor cells were negative for Ber-EP4 (magnification, ×25). (D) Around 40% of tumor cells were positive for Ki67 (magnification, ×400).
Discussion
A notable finding of this case report was that combination therapy achieved CSCC regression in all three patients within 2–6 months after the first administered dose of acitretin and clarithromycin. Furthermore, two of the patients exhibited complete remission of the tumor without scaring, thereby avoiding the need for surgery, while the other patient was successfully treated by tumor excision after combination therapy. Importantly, all three patients remained free of recurrence during a follow-up of 3–10 years. In addition, all three patients tolerated the treatment and experienced only minimal adverse effects (pseudotumor cerebri and mucocutaneous dryness) that resolved after discontinuation of acitretin. To our knowledge, this is the first study to report that combination therapy with acitretin and clarithromycin may be an effective treatment for CSCC.
The majority of CSCCs can be successfully eradicated by surgical excision with post-operative margin assessment or microscopically controlled surgery, although the cosmetic results are not always satisfactory. When patients cannot or refuse to undergo surgery, alternative treatment options include radiotherapy, intralesional chemotherapy and photodynamic therapy. Radiotherapy is a potentially curative treatment in cases of early-stage CSCC where resection would cause substantial cosmetic or functional deficits. However, radiotherapy alone has limited effectiveness as a treatment for advanced CSCC and is associated with adverse effects. Anti-PD-1 antibodies are considered a first-line systemic treatment for patients with metastatic or locally advanced CSCC who are not candidates for curative surgery or radiotherapy, although its use in elderly patients is limited by the toxicity of this chemotherapeutic agent (11, 12). Recently, epidermal growth factor receptor inhibitors combined with chemotherapy or radiotherapy have been used as second-line systemic treatments for advanced CSCC (11, 13), but further research is needed to fully characterize the efficacy and safety of these strategies (11, 14). As awareness grows regarding the increasing number of cases of post-surgical CSSC relapse, it is becoming recognized that improved non-surgical treatments for CSCC are urgently needed.
Retinoids are established drugs and have been shown to reduce the development of non-melanoma skin cancer in high risk individuals. Previous randomized controlled trials have demonstrated that systemic retinoids can prevent new primary CSCC or new precancerous lesions in immunocompromised patients after transplantation (15). Typically, retinoids are considered for patients who develop multiple tumors each year, have multiple CSCCs in high-risk areas, or have numerous actinic keratoses. Acitretin is a second-generation retinoid reported to exhibit anti-tumor activity against actinic keratosis, verrucous carcinomas and SCC. Previous studies of the use of acitretin in the management of CSCC are rare and summarized in Table 1 (5–7). Two of these three studies were case reports describing the use of acitretin in the chemoprevention of CSCC in 8 patients receiving BRAF inhibitors (5) and one patient who had been treated with phototherapy (6). Only one prior study has described the use of acitretin to treat CSCC: the combination of acitretin, intralesional 5-fluorouracil and 5-fluorouracil chemowraps was found to cause complete regression of widespread CSCCs on the legs of two patients (7). To our knowledge, the present case report is the first to report the successful use of acitretin in combination with clarithromycin as a treatment for CSCC.
Table 1 Summary of publications reporting cases and our cases of cutaneous squamous cell carcinoma (CSCC) treated with acitretin.
Reference Patients Treatment Response
Manalo et al. (7) Two patients (74-year-old female and 60-year-old female) with widespread CSCCs on the legs Acitretin (25 mg/d) combined with 5-fluorouracil and chemowraps for 6 months Complete clinical resolution
Anforth et al. (5) Eight patients treated with a BRAF inhibitor Acitretin (10–15 mg/d) for 1–13 months Suppression of CSCC development
Lebwohl et al. (6) A 40-year-old male patient with psoriasis who developed CSCC following PUVA phototherapy Acitretin (25 mg/d) for 25 months Suppression of CSCC development
Patient 1 An 83-year-old Chinese woman with invasive SCC on the left cheek Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 8 courses Complete remission (6
months); 10 years without
recurrence
Patient 2 A 97-year-old Chinese woman with invasive SCC on the left cheek Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 6 courses Complete remission (6
months); 3 years without
recurrence
Patient 3 An 76-year-old Chinese woman with invasive SCC on the right ankle Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 6 plus 24 courses resumed* Complete remission (24 months resumed); 6 years without recurrence
* After 6 courses of the combination therapy, this patient self-discontinued the treatment. Two years later, tumor recurrence was detected and the patient resumed the combination treatment for another 24 courses.
Acitretin monotherapy is generally considered to have a low efficacy and is associated with dose-dependent side effects such as mucocutaneous dryness, hair loss, elevated triglycerides, transient elevations in liver function tests and (with long-term treatment) diffuse idiopathic skeletal hyperostosis. In view of the above issues, the patients in this case report were also administered clarithromycin, a macrolide antibiotic that has been reported to enhance the activity of conventional chemotherapy agents in cell lines and animal experiments (8, 9). Furthermore, long-term administration of clarithromycin was reported to prolong the survival time of patients with lung cancer (16). We speculate that the excellent responses to combination therapy observed in all three patients in the present report were due to the antiangiogenic and antitumor effects of clarithromycin, which synergized with the anti-tumor actions of acitretin.
Combination therapy with acitretin and clarithromycin was well-tolerated by all three patients in this study. We summarized the common characteristics of the three patients in our study which may somehow contributed to the successful treatment outcomes. We found that all of them were female older than 75 and refused large-scale surgical treatment. The pathological diagnoses were poorly to well-differentiated invasive SCC. Their tumors were all single and without any metastases. We believe that elderly patients with slow tumor growth respond well to drug treatment and this can help them to avoid the risks associated with surgery. For young patients with multiple or larger tumors, the drug treatment can also be used to reduce the size of the tumor, furthermore it will be beneficial to surgical resection. The only adverse effects reported were pseudotumor cerebri and mucocutaneous dryness, which are known side-effects of acitretin. Therefore, we suggest that the combination of acitretin and clarithromycin may represent an effective and safe therapy for CSCC.
As far as we are aware, this is the first study to report that combination therapy with acitretin and clarithromycin may be an effective and safe treatment for CSCC. Additional studies are required to further assess the efficacy and safety of combination therapy with acitretin and clarithromycin in the treatment of CSCC in larger populations and to investigate the underlying molecular mechanisms.
Data Availability Statement
The original contributions generated in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.
Ethics Statement
The studies involving human participants were reviewed and approved by The study was approved by the Medical Technology Clinical Application Ethics Committee of the First Affiliated Hospital of Fujian (No. [2015]084). The patients/participants provided their written informed consent to participate in this study. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.
Author Contributions
YZha: conception and design. CJ: administrative support. YZha, YZhu, and HW: provision of study materials or patients. CJ and YZhu: data analysis and interpretation. All authors: collection, assembly of data, manuscript writing, and final approval of manuscript.
Conflict of Interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The authors acknowledge the help of all patients who participated in our study. | ACITRETIN, CLARITHROMYCIN | DrugsGivenReaction | CC BY | 33718249 | 20,874,907 | 2021 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Drug effective for unapproved indication'. | Case Report: Successful Treatment of Cutaneous Squamous Cell Carcinoma in Three Patients With a Combination of Acitretin and Clarithromycin.
Few studies have evaluated acitretin as a primary treatment for cutaneous squamous cell carcinoma (CSCC). We report, for the first time, three cases of CSCC successfully treated with acitretin and clarithromycin. A literature review on this subject was also was performed. This case report included three patients with CSCC treated with acitretin and clarithromycin at the First Affiliated Hospital of Fujian Medical University (2008-2019). Patient 1 (83-year-old woman, ulcerated mass on the left cheek), patient 2 (97-year-old woman, painful mass on the left cheek) and patient 3 (76-year-old woman, large mass on the right ankle) received 8, 6, and 30 courses of combination therapy. All patients tolerated the adverse effects (pseudotumor cerebri and mucocutaneous dryness) and achieved complete regression within 6 months. Patients 1, 2, and 3 have not experienced recurrence during a 10-, 3-, and 6-year follow-up. Acitretin has limited efficacy as a monotherapy for CSCC. Our experience indicates that combination therapy with acitretin and clarithromycin may be an effective and well-tolerated treatment for unresectable CSCC.
Introduction
Given its increasing incidence and potential for poor outcomes (1–3), cutaneous squamous cell carcinoma (CSCC) is emerging as a public health problem. High-risk CSCC can be difficult to treat by surgical excision because the neoplasm can extend or infiltrate beyond the visible borders of the lesion. A wide range of non-surgical treatments are available for high-risk CSCC. Acitretin is a second generation retinoid shown to reduce the incidence of new primary non-melanoma skin cancers in immunocompromised transplant recipients (4) and patients receiving BRAF inhibitor treatment (5). Acitretin is considered to have limited efficacy as a monotherapy for CSCC but seems more effective when used in combination with other agents (6, 7). Clarithromycin is an antibacterial drug that has been reported to enhance the anti-tumor activity of chemotherapeutic drugs in vitro and in vivo (8, 9). However, no previous study has evaluated the combination of acitretin and clarithromycin in the treatment of CSCC. We present 3 cases illustrating that combination therapy with acitretin and clarithromycin might be a promising option for patients with CSCC who are unwilling or unable to undergo surgery.
Case Description
Ethical Statement
The study was approved by the Medical Technology Clinical Application Ethics Committee of the First Affiliated Hospital of Fujian (No. [2015]084). Before treatment, the patients were fully informed about the potential advantages and disadvantages of combination therapy with acitretin and clarithromycin, including the possible side effects of these drugs. All patients provided written informed consent for treatment and agreed to attend follow-up visits every 2 weeks. All patients provided written informed consent for data analysis and publication.
Patients
The present case report represents a retrospective analysis of the medical records of patients with CSCC who were treated with a combination of acitretin and clarithromycin at the Dermatology Department of the First Affiliated Hospital of Fujian between 2008 and 2018. Inclusion criteria consisted of a diagnosis of SCC by pathological biopsy and a history as well as tumor diameter ≥1 cm (10).
A total of 3 patients were involved in our study. Acitretin and clarithromycin were administered together to the patients from the time when skin biopsy confirmed the diagnosis of invasive SCC. We defined oral 30 mg/day of acitretin for 4 weeks and oral 500 mg/day of clarithromycin for 3 weeks as 1-month treatment course. There was no interval time between courses.
Once biochemical investigations, including measurements of serum creatinine, cholesterol, triglycerides, and potassium appeared abnormal, the treatment was stopped or the dosage was reduced.
Patient 1
Patient 1 was an 83-year-old Chinese woman with a 2-year history of an asymptomatic ulcerated mass on her left cheek. Physical examination in March 2008 revealed a 4 × 3.5 cm crater-like mass with central ulceration on her left cheek (Figure 1A). Skin biopsy revealed the presence of atypical epithelial tumor cell formations extending beyond the epidermis into the underlying dermis, cornified cells and horny pearls, thereby confirming the diagnosis of invasive SCC. The patient showed a favorable response to acitretin and clarithromycin within 2 months after the first oral administration. The patient declined long-term treatment due to the development of adverse effects (pseudotumor cerebri and dryness of the lips and skin). Therefore, CO2 laser resection of the CSCC was performed after a total of 6 courses of combination therapy, and the patient then received an additional two courses of combination therapy. This patient has lived without tumor recurrence for 10 years.
Figure 1 Photographs showing the regression of cutaneous squamous cell carcinoma (CSCC) in three patients treated with acitretin and clarithromycin. (A) Patient 1. The CSCC on the left cheek is shown at month 0 (before initiation of combination therapy), 5 and 12. (B) Patient 2. The CSCC on the left cheek is shown at month 0, 3, and 29. (C) Patient 3. The CSCC on the right ankle is shown at month 0, 6, 30, 42, and 60.
Patient 2
Patient 2 was a 97-year-old Chinese woman with a 1-year history of a painful mass on her left cheek. Physical examination in November 2015 demonstrated the presence of a 2 × 2 cm mass with an irregular edge and signs of erosion on her left cheek (Figure 1B). Skin biopsy confirmed the diagnosis of invasive SCC. The patient was treated with 6 courses of combination therapy. Darkening and regression of the tumor lesions were first detected in 2 months after the initiation of combination therapy, and a complete response was observed at 6 months. The only adverse effect reported by the patient was mucocutaneous dryness, but this resolved following cessation of treatment with acitretin. The patient has not experienced recurrence of CSCC during the 3-year follow-up.
Patient 3
Patient 3 was a 76-year-old Chinese woman with a 1-year history of a large mass on her right ankle that had progressively increased in size. Physical examination in September 2012 revealed a 4 × 4 cm mass with erosion on her right ankle (Figure 1C). Skin biopsy showed that the overlying epidermis was ulcerated with atypical keratinocytes invading the dermis. The keratinocytes were larger in the central parts of the tumor nests and formed eosinophilic horny pearls. The neoplastic keratinocytes had abundant eosinophilic and dyskeratotic cytoplasm and large vesicular nuclei with visible nucleoli and mitoses. An inflammatory reaction was evident, with many lymphocytes surrounding the stromal vessels and penetrating the tumor nests (Figure 2). Immunohistochemical analyses demonstrated that the tumor cells were positive for AE1/AE3 and P63, negative for Ber-EP4, and positive for Ki67 in about 40% of cells, confirming the diagnosis of invasive SCC (Figure 3). The patient was initially treated with 6 courses of combination therapy, which led to substantial regression of the tumor such that it resembled a small residual ulcer, and the patient then decided to discontinue the treatment. A local tumor relapse was detected 2 years later, so the patient resumed combination therapy and received an additional 24 courses of treatment, which resulted in complete remission. The only adverse effect experienced by the patient was mucocutaneous dryness. The patient remains alive and without tumor recurrence after a 6-year follow-up.
Figure 2 Representative images showing the histologic appearance of a tumor specimen from patient 3. (A) Section stained with hematoxylin-eosin (magnification, ×25). (B) Section stained with hematoxylin-eosin (magnification, ×200). (C) Section stained with hematoxylin-eosin (magnification, ×400).
Figure 3 Immunohistology of tumor specimens from patient 3. (A) Tumor cells were positive for AE1/AE3 (magnification, ×25). (B) Tumor cells were positive for P63 (magnification, ×25). (C) Tumor cells were negative for Ber-EP4 (magnification, ×25). (D) Around 40% of tumor cells were positive for Ki67 (magnification, ×400).
Discussion
A notable finding of this case report was that combination therapy achieved CSCC regression in all three patients within 2–6 months after the first administered dose of acitretin and clarithromycin. Furthermore, two of the patients exhibited complete remission of the tumor without scaring, thereby avoiding the need for surgery, while the other patient was successfully treated by tumor excision after combination therapy. Importantly, all three patients remained free of recurrence during a follow-up of 3–10 years. In addition, all three patients tolerated the treatment and experienced only minimal adverse effects (pseudotumor cerebri and mucocutaneous dryness) that resolved after discontinuation of acitretin. To our knowledge, this is the first study to report that combination therapy with acitretin and clarithromycin may be an effective treatment for CSCC.
The majority of CSCCs can be successfully eradicated by surgical excision with post-operative margin assessment or microscopically controlled surgery, although the cosmetic results are not always satisfactory. When patients cannot or refuse to undergo surgery, alternative treatment options include radiotherapy, intralesional chemotherapy and photodynamic therapy. Radiotherapy is a potentially curative treatment in cases of early-stage CSCC where resection would cause substantial cosmetic or functional deficits. However, radiotherapy alone has limited effectiveness as a treatment for advanced CSCC and is associated with adverse effects. Anti-PD-1 antibodies are considered a first-line systemic treatment for patients with metastatic or locally advanced CSCC who are not candidates for curative surgery or radiotherapy, although its use in elderly patients is limited by the toxicity of this chemotherapeutic agent (11, 12). Recently, epidermal growth factor receptor inhibitors combined with chemotherapy or radiotherapy have been used as second-line systemic treatments for advanced CSCC (11, 13), but further research is needed to fully characterize the efficacy and safety of these strategies (11, 14). As awareness grows regarding the increasing number of cases of post-surgical CSSC relapse, it is becoming recognized that improved non-surgical treatments for CSCC are urgently needed.
Retinoids are established drugs and have been shown to reduce the development of non-melanoma skin cancer in high risk individuals. Previous randomized controlled trials have demonstrated that systemic retinoids can prevent new primary CSCC or new precancerous lesions in immunocompromised patients after transplantation (15). Typically, retinoids are considered for patients who develop multiple tumors each year, have multiple CSCCs in high-risk areas, or have numerous actinic keratoses. Acitretin is a second-generation retinoid reported to exhibit anti-tumor activity against actinic keratosis, verrucous carcinomas and SCC. Previous studies of the use of acitretin in the management of CSCC are rare and summarized in Table 1 (5–7). Two of these three studies were case reports describing the use of acitretin in the chemoprevention of CSCC in 8 patients receiving BRAF inhibitors (5) and one patient who had been treated with phototherapy (6). Only one prior study has described the use of acitretin to treat CSCC: the combination of acitretin, intralesional 5-fluorouracil and 5-fluorouracil chemowraps was found to cause complete regression of widespread CSCCs on the legs of two patients (7). To our knowledge, the present case report is the first to report the successful use of acitretin in combination with clarithromycin as a treatment for CSCC.
Table 1 Summary of publications reporting cases and our cases of cutaneous squamous cell carcinoma (CSCC) treated with acitretin.
Reference Patients Treatment Response
Manalo et al. (7) Two patients (74-year-old female and 60-year-old female) with widespread CSCCs on the legs Acitretin (25 mg/d) combined with 5-fluorouracil and chemowraps for 6 months Complete clinical resolution
Anforth et al. (5) Eight patients treated with a BRAF inhibitor Acitretin (10–15 mg/d) for 1–13 months Suppression of CSCC development
Lebwohl et al. (6) A 40-year-old male patient with psoriasis who developed CSCC following PUVA phototherapy Acitretin (25 mg/d) for 25 months Suppression of CSCC development
Patient 1 An 83-year-old Chinese woman with invasive SCC on the left cheek Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 8 courses Complete remission (6
months); 10 years without
recurrence
Patient 2 A 97-year-old Chinese woman with invasive SCC on the left cheek Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 6 courses Complete remission (6
months); 3 years without
recurrence
Patient 3 An 76-year-old Chinese woman with invasive SCC on the right ankle Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 6 plus 24 courses resumed* Complete remission (24 months resumed); 6 years without recurrence
* After 6 courses of the combination therapy, this patient self-discontinued the treatment. Two years later, tumor recurrence was detected and the patient resumed the combination treatment for another 24 courses.
Acitretin monotherapy is generally considered to have a low efficacy and is associated with dose-dependent side effects such as mucocutaneous dryness, hair loss, elevated triglycerides, transient elevations in liver function tests and (with long-term treatment) diffuse idiopathic skeletal hyperostosis. In view of the above issues, the patients in this case report were also administered clarithromycin, a macrolide antibiotic that has been reported to enhance the activity of conventional chemotherapy agents in cell lines and animal experiments (8, 9). Furthermore, long-term administration of clarithromycin was reported to prolong the survival time of patients with lung cancer (16). We speculate that the excellent responses to combination therapy observed in all three patients in the present report were due to the antiangiogenic and antitumor effects of clarithromycin, which synergized with the anti-tumor actions of acitretin.
Combination therapy with acitretin and clarithromycin was well-tolerated by all three patients in this study. We summarized the common characteristics of the three patients in our study which may somehow contributed to the successful treatment outcomes. We found that all of them were female older than 75 and refused large-scale surgical treatment. The pathological diagnoses were poorly to well-differentiated invasive SCC. Their tumors were all single and without any metastases. We believe that elderly patients with slow tumor growth respond well to drug treatment and this can help them to avoid the risks associated with surgery. For young patients with multiple or larger tumors, the drug treatment can also be used to reduce the size of the tumor, furthermore it will be beneficial to surgical resection. The only adverse effects reported were pseudotumor cerebri and mucocutaneous dryness, which are known side-effects of acitretin. Therefore, we suggest that the combination of acitretin and clarithromycin may represent an effective and safe therapy for CSCC.
As far as we are aware, this is the first study to report that combination therapy with acitretin and clarithromycin may be an effective and safe treatment for CSCC. Additional studies are required to further assess the efficacy and safety of combination therapy with acitretin and clarithromycin in the treatment of CSCC in larger populations and to investigate the underlying molecular mechanisms.
Data Availability Statement
The original contributions generated in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.
Ethics Statement
The studies involving human participants were reviewed and approved by The study was approved by the Medical Technology Clinical Application Ethics Committee of the First Affiliated Hospital of Fujian (No. [2015]084). The patients/participants provided their written informed consent to participate in this study. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.
Author Contributions
YZha: conception and design. CJ: administrative support. YZha, YZhu, and HW: provision of study materials or patients. CJ and YZhu: data analysis and interpretation. All authors: collection, assembly of data, manuscript writing, and final approval of manuscript.
Conflict of Interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The authors acknowledge the help of all patients who participated in our study. | ACITRETIN, CLARITHROMYCIN | DrugsGivenReaction | CC BY | 33718249 | 19,083,671 | 2021 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Dry skin'. | Case Report: Successful Treatment of Cutaneous Squamous Cell Carcinoma in Three Patients With a Combination of Acitretin and Clarithromycin.
Few studies have evaluated acitretin as a primary treatment for cutaneous squamous cell carcinoma (CSCC). We report, for the first time, three cases of CSCC successfully treated with acitretin and clarithromycin. A literature review on this subject was also was performed. This case report included three patients with CSCC treated with acitretin and clarithromycin at the First Affiliated Hospital of Fujian Medical University (2008-2019). Patient 1 (83-year-old woman, ulcerated mass on the left cheek), patient 2 (97-year-old woman, painful mass on the left cheek) and patient 3 (76-year-old woman, large mass on the right ankle) received 8, 6, and 30 courses of combination therapy. All patients tolerated the adverse effects (pseudotumor cerebri and mucocutaneous dryness) and achieved complete regression within 6 months. Patients 1, 2, and 3 have not experienced recurrence during a 10-, 3-, and 6-year follow-up. Acitretin has limited efficacy as a monotherapy for CSCC. Our experience indicates that combination therapy with acitretin and clarithromycin may be an effective and well-tolerated treatment for unresectable CSCC.
Introduction
Given its increasing incidence and potential for poor outcomes (1–3), cutaneous squamous cell carcinoma (CSCC) is emerging as a public health problem. High-risk CSCC can be difficult to treat by surgical excision because the neoplasm can extend or infiltrate beyond the visible borders of the lesion. A wide range of non-surgical treatments are available for high-risk CSCC. Acitretin is a second generation retinoid shown to reduce the incidence of new primary non-melanoma skin cancers in immunocompromised transplant recipients (4) and patients receiving BRAF inhibitor treatment (5). Acitretin is considered to have limited efficacy as a monotherapy for CSCC but seems more effective when used in combination with other agents (6, 7). Clarithromycin is an antibacterial drug that has been reported to enhance the anti-tumor activity of chemotherapeutic drugs in vitro and in vivo (8, 9). However, no previous study has evaluated the combination of acitretin and clarithromycin in the treatment of CSCC. We present 3 cases illustrating that combination therapy with acitretin and clarithromycin might be a promising option for patients with CSCC who are unwilling or unable to undergo surgery.
Case Description
Ethical Statement
The study was approved by the Medical Technology Clinical Application Ethics Committee of the First Affiliated Hospital of Fujian (No. [2015]084). Before treatment, the patients were fully informed about the potential advantages and disadvantages of combination therapy with acitretin and clarithromycin, including the possible side effects of these drugs. All patients provided written informed consent for treatment and agreed to attend follow-up visits every 2 weeks. All patients provided written informed consent for data analysis and publication.
Patients
The present case report represents a retrospective analysis of the medical records of patients with CSCC who were treated with a combination of acitretin and clarithromycin at the Dermatology Department of the First Affiliated Hospital of Fujian between 2008 and 2018. Inclusion criteria consisted of a diagnosis of SCC by pathological biopsy and a history as well as tumor diameter ≥1 cm (10).
A total of 3 patients were involved in our study. Acitretin and clarithromycin were administered together to the patients from the time when skin biopsy confirmed the diagnosis of invasive SCC. We defined oral 30 mg/day of acitretin for 4 weeks and oral 500 mg/day of clarithromycin for 3 weeks as 1-month treatment course. There was no interval time between courses.
Once biochemical investigations, including measurements of serum creatinine, cholesterol, triglycerides, and potassium appeared abnormal, the treatment was stopped or the dosage was reduced.
Patient 1
Patient 1 was an 83-year-old Chinese woman with a 2-year history of an asymptomatic ulcerated mass on her left cheek. Physical examination in March 2008 revealed a 4 × 3.5 cm crater-like mass with central ulceration on her left cheek (Figure 1A). Skin biopsy revealed the presence of atypical epithelial tumor cell formations extending beyond the epidermis into the underlying dermis, cornified cells and horny pearls, thereby confirming the diagnosis of invasive SCC. The patient showed a favorable response to acitretin and clarithromycin within 2 months after the first oral administration. The patient declined long-term treatment due to the development of adverse effects (pseudotumor cerebri and dryness of the lips and skin). Therefore, CO2 laser resection of the CSCC was performed after a total of 6 courses of combination therapy, and the patient then received an additional two courses of combination therapy. This patient has lived without tumor recurrence for 10 years.
Figure 1 Photographs showing the regression of cutaneous squamous cell carcinoma (CSCC) in three patients treated with acitretin and clarithromycin. (A) Patient 1. The CSCC on the left cheek is shown at month 0 (before initiation of combination therapy), 5 and 12. (B) Patient 2. The CSCC on the left cheek is shown at month 0, 3, and 29. (C) Patient 3. The CSCC on the right ankle is shown at month 0, 6, 30, 42, and 60.
Patient 2
Patient 2 was a 97-year-old Chinese woman with a 1-year history of a painful mass on her left cheek. Physical examination in November 2015 demonstrated the presence of a 2 × 2 cm mass with an irregular edge and signs of erosion on her left cheek (Figure 1B). Skin biopsy confirmed the diagnosis of invasive SCC. The patient was treated with 6 courses of combination therapy. Darkening and regression of the tumor lesions were first detected in 2 months after the initiation of combination therapy, and a complete response was observed at 6 months. The only adverse effect reported by the patient was mucocutaneous dryness, but this resolved following cessation of treatment with acitretin. The patient has not experienced recurrence of CSCC during the 3-year follow-up.
Patient 3
Patient 3 was a 76-year-old Chinese woman with a 1-year history of a large mass on her right ankle that had progressively increased in size. Physical examination in September 2012 revealed a 4 × 4 cm mass with erosion on her right ankle (Figure 1C). Skin biopsy showed that the overlying epidermis was ulcerated with atypical keratinocytes invading the dermis. The keratinocytes were larger in the central parts of the tumor nests and formed eosinophilic horny pearls. The neoplastic keratinocytes had abundant eosinophilic and dyskeratotic cytoplasm and large vesicular nuclei with visible nucleoli and mitoses. An inflammatory reaction was evident, with many lymphocytes surrounding the stromal vessels and penetrating the tumor nests (Figure 2). Immunohistochemical analyses demonstrated that the tumor cells were positive for AE1/AE3 and P63, negative for Ber-EP4, and positive for Ki67 in about 40% of cells, confirming the diagnosis of invasive SCC (Figure 3). The patient was initially treated with 6 courses of combination therapy, which led to substantial regression of the tumor such that it resembled a small residual ulcer, and the patient then decided to discontinue the treatment. A local tumor relapse was detected 2 years later, so the patient resumed combination therapy and received an additional 24 courses of treatment, which resulted in complete remission. The only adverse effect experienced by the patient was mucocutaneous dryness. The patient remains alive and without tumor recurrence after a 6-year follow-up.
Figure 2 Representative images showing the histologic appearance of a tumor specimen from patient 3. (A) Section stained with hematoxylin-eosin (magnification, ×25). (B) Section stained with hematoxylin-eosin (magnification, ×200). (C) Section stained with hematoxylin-eosin (magnification, ×400).
Figure 3 Immunohistology of tumor specimens from patient 3. (A) Tumor cells were positive for AE1/AE3 (magnification, ×25). (B) Tumor cells were positive for P63 (magnification, ×25). (C) Tumor cells were negative for Ber-EP4 (magnification, ×25). (D) Around 40% of tumor cells were positive for Ki67 (magnification, ×400).
Discussion
A notable finding of this case report was that combination therapy achieved CSCC regression in all three patients within 2–6 months after the first administered dose of acitretin and clarithromycin. Furthermore, two of the patients exhibited complete remission of the tumor without scaring, thereby avoiding the need for surgery, while the other patient was successfully treated by tumor excision after combination therapy. Importantly, all three patients remained free of recurrence during a follow-up of 3–10 years. In addition, all three patients tolerated the treatment and experienced only minimal adverse effects (pseudotumor cerebri and mucocutaneous dryness) that resolved after discontinuation of acitretin. To our knowledge, this is the first study to report that combination therapy with acitretin and clarithromycin may be an effective treatment for CSCC.
The majority of CSCCs can be successfully eradicated by surgical excision with post-operative margin assessment or microscopically controlled surgery, although the cosmetic results are not always satisfactory. When patients cannot or refuse to undergo surgery, alternative treatment options include radiotherapy, intralesional chemotherapy and photodynamic therapy. Radiotherapy is a potentially curative treatment in cases of early-stage CSCC where resection would cause substantial cosmetic or functional deficits. However, radiotherapy alone has limited effectiveness as a treatment for advanced CSCC and is associated with adverse effects. Anti-PD-1 antibodies are considered a first-line systemic treatment for patients with metastatic or locally advanced CSCC who are not candidates for curative surgery or radiotherapy, although its use in elderly patients is limited by the toxicity of this chemotherapeutic agent (11, 12). Recently, epidermal growth factor receptor inhibitors combined with chemotherapy or radiotherapy have been used as second-line systemic treatments for advanced CSCC (11, 13), but further research is needed to fully characterize the efficacy and safety of these strategies (11, 14). As awareness grows regarding the increasing number of cases of post-surgical CSSC relapse, it is becoming recognized that improved non-surgical treatments for CSCC are urgently needed.
Retinoids are established drugs and have been shown to reduce the development of non-melanoma skin cancer in high risk individuals. Previous randomized controlled trials have demonstrated that systemic retinoids can prevent new primary CSCC or new precancerous lesions in immunocompromised patients after transplantation (15). Typically, retinoids are considered for patients who develop multiple tumors each year, have multiple CSCCs in high-risk areas, or have numerous actinic keratoses. Acitretin is a second-generation retinoid reported to exhibit anti-tumor activity against actinic keratosis, verrucous carcinomas and SCC. Previous studies of the use of acitretin in the management of CSCC are rare and summarized in Table 1 (5–7). Two of these three studies were case reports describing the use of acitretin in the chemoprevention of CSCC in 8 patients receiving BRAF inhibitors (5) and one patient who had been treated with phototherapy (6). Only one prior study has described the use of acitretin to treat CSCC: the combination of acitretin, intralesional 5-fluorouracil and 5-fluorouracil chemowraps was found to cause complete regression of widespread CSCCs on the legs of two patients (7). To our knowledge, the present case report is the first to report the successful use of acitretin in combination with clarithromycin as a treatment for CSCC.
Table 1 Summary of publications reporting cases and our cases of cutaneous squamous cell carcinoma (CSCC) treated with acitretin.
Reference Patients Treatment Response
Manalo et al. (7) Two patients (74-year-old female and 60-year-old female) with widespread CSCCs on the legs Acitretin (25 mg/d) combined with 5-fluorouracil and chemowraps for 6 months Complete clinical resolution
Anforth et al. (5) Eight patients treated with a BRAF inhibitor Acitretin (10–15 mg/d) for 1–13 months Suppression of CSCC development
Lebwohl et al. (6) A 40-year-old male patient with psoriasis who developed CSCC following PUVA phototherapy Acitretin (25 mg/d) for 25 months Suppression of CSCC development
Patient 1 An 83-year-old Chinese woman with invasive SCC on the left cheek Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 8 courses Complete remission (6
months); 10 years without
recurrence
Patient 2 A 97-year-old Chinese woman with invasive SCC on the left cheek Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 6 courses Complete remission (6
months); 3 years without
recurrence
Patient 3 An 76-year-old Chinese woman with invasive SCC on the right ankle Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 6 plus 24 courses resumed* Complete remission (24 months resumed); 6 years without recurrence
* After 6 courses of the combination therapy, this patient self-discontinued the treatment. Two years later, tumor recurrence was detected and the patient resumed the combination treatment for another 24 courses.
Acitretin monotherapy is generally considered to have a low efficacy and is associated with dose-dependent side effects such as mucocutaneous dryness, hair loss, elevated triglycerides, transient elevations in liver function tests and (with long-term treatment) diffuse idiopathic skeletal hyperostosis. In view of the above issues, the patients in this case report were also administered clarithromycin, a macrolide antibiotic that has been reported to enhance the activity of conventional chemotherapy agents in cell lines and animal experiments (8, 9). Furthermore, long-term administration of clarithromycin was reported to prolong the survival time of patients with lung cancer (16). We speculate that the excellent responses to combination therapy observed in all three patients in the present report were due to the antiangiogenic and antitumor effects of clarithromycin, which synergized with the anti-tumor actions of acitretin.
Combination therapy with acitretin and clarithromycin was well-tolerated by all three patients in this study. We summarized the common characteristics of the three patients in our study which may somehow contributed to the successful treatment outcomes. We found that all of them were female older than 75 and refused large-scale surgical treatment. The pathological diagnoses were poorly to well-differentiated invasive SCC. Their tumors were all single and without any metastases. We believe that elderly patients with slow tumor growth respond well to drug treatment and this can help them to avoid the risks associated with surgery. For young patients with multiple or larger tumors, the drug treatment can also be used to reduce the size of the tumor, furthermore it will be beneficial to surgical resection. The only adverse effects reported were pseudotumor cerebri and mucocutaneous dryness, which are known side-effects of acitretin. Therefore, we suggest that the combination of acitretin and clarithromycin may represent an effective and safe therapy for CSCC.
As far as we are aware, this is the first study to report that combination therapy with acitretin and clarithromycin may be an effective and safe treatment for CSCC. Additional studies are required to further assess the efficacy and safety of combination therapy with acitretin and clarithromycin in the treatment of CSCC in larger populations and to investigate the underlying molecular mechanisms.
Data Availability Statement
The original contributions generated in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.
Ethics Statement
The studies involving human participants were reviewed and approved by The study was approved by the Medical Technology Clinical Application Ethics Committee of the First Affiliated Hospital of Fujian (No. [2015]084). The patients/participants provided their written informed consent to participate in this study. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.
Author Contributions
YZha: conception and design. CJ: administrative support. YZha, YZhu, and HW: provision of study materials or patients. CJ and YZhu: data analysis and interpretation. All authors: collection, assembly of data, manuscript writing, and final approval of manuscript.
Conflict of Interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The authors acknowledge the help of all patients who participated in our study. | ACITRETIN, CLARITHROMYCIN | DrugsGivenReaction | CC BY | 33718249 | 19,083,671 | 2021 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Idiopathic intracranial hypertension'. | Case Report: Successful Treatment of Cutaneous Squamous Cell Carcinoma in Three Patients With a Combination of Acitretin and Clarithromycin.
Few studies have evaluated acitretin as a primary treatment for cutaneous squamous cell carcinoma (CSCC). We report, for the first time, three cases of CSCC successfully treated with acitretin and clarithromycin. A literature review on this subject was also was performed. This case report included three patients with CSCC treated with acitretin and clarithromycin at the First Affiliated Hospital of Fujian Medical University (2008-2019). Patient 1 (83-year-old woman, ulcerated mass on the left cheek), patient 2 (97-year-old woman, painful mass on the left cheek) and patient 3 (76-year-old woman, large mass on the right ankle) received 8, 6, and 30 courses of combination therapy. All patients tolerated the adverse effects (pseudotumor cerebri and mucocutaneous dryness) and achieved complete regression within 6 months. Patients 1, 2, and 3 have not experienced recurrence during a 10-, 3-, and 6-year follow-up. Acitretin has limited efficacy as a monotherapy for CSCC. Our experience indicates that combination therapy with acitretin and clarithromycin may be an effective and well-tolerated treatment for unresectable CSCC.
Introduction
Given its increasing incidence and potential for poor outcomes (1–3), cutaneous squamous cell carcinoma (CSCC) is emerging as a public health problem. High-risk CSCC can be difficult to treat by surgical excision because the neoplasm can extend or infiltrate beyond the visible borders of the lesion. A wide range of non-surgical treatments are available for high-risk CSCC. Acitretin is a second generation retinoid shown to reduce the incidence of new primary non-melanoma skin cancers in immunocompromised transplant recipients (4) and patients receiving BRAF inhibitor treatment (5). Acitretin is considered to have limited efficacy as a monotherapy for CSCC but seems more effective when used in combination with other agents (6, 7). Clarithromycin is an antibacterial drug that has been reported to enhance the anti-tumor activity of chemotherapeutic drugs in vitro and in vivo (8, 9). However, no previous study has evaluated the combination of acitretin and clarithromycin in the treatment of CSCC. We present 3 cases illustrating that combination therapy with acitretin and clarithromycin might be a promising option for patients with CSCC who are unwilling or unable to undergo surgery.
Case Description
Ethical Statement
The study was approved by the Medical Technology Clinical Application Ethics Committee of the First Affiliated Hospital of Fujian (No. [2015]084). Before treatment, the patients were fully informed about the potential advantages and disadvantages of combination therapy with acitretin and clarithromycin, including the possible side effects of these drugs. All patients provided written informed consent for treatment and agreed to attend follow-up visits every 2 weeks. All patients provided written informed consent for data analysis and publication.
Patients
The present case report represents a retrospective analysis of the medical records of patients with CSCC who were treated with a combination of acitretin and clarithromycin at the Dermatology Department of the First Affiliated Hospital of Fujian between 2008 and 2018. Inclusion criteria consisted of a diagnosis of SCC by pathological biopsy and a history as well as tumor diameter ≥1 cm (10).
A total of 3 patients were involved in our study. Acitretin and clarithromycin were administered together to the patients from the time when skin biopsy confirmed the diagnosis of invasive SCC. We defined oral 30 mg/day of acitretin for 4 weeks and oral 500 mg/day of clarithromycin for 3 weeks as 1-month treatment course. There was no interval time between courses.
Once biochemical investigations, including measurements of serum creatinine, cholesterol, triglycerides, and potassium appeared abnormal, the treatment was stopped or the dosage was reduced.
Patient 1
Patient 1 was an 83-year-old Chinese woman with a 2-year history of an asymptomatic ulcerated mass on her left cheek. Physical examination in March 2008 revealed a 4 × 3.5 cm crater-like mass with central ulceration on her left cheek (Figure 1A). Skin biopsy revealed the presence of atypical epithelial tumor cell formations extending beyond the epidermis into the underlying dermis, cornified cells and horny pearls, thereby confirming the diagnosis of invasive SCC. The patient showed a favorable response to acitretin and clarithromycin within 2 months after the first oral administration. The patient declined long-term treatment due to the development of adverse effects (pseudotumor cerebri and dryness of the lips and skin). Therefore, CO2 laser resection of the CSCC was performed after a total of 6 courses of combination therapy, and the patient then received an additional two courses of combination therapy. This patient has lived without tumor recurrence for 10 years.
Figure 1 Photographs showing the regression of cutaneous squamous cell carcinoma (CSCC) in three patients treated with acitretin and clarithromycin. (A) Patient 1. The CSCC on the left cheek is shown at month 0 (before initiation of combination therapy), 5 and 12. (B) Patient 2. The CSCC on the left cheek is shown at month 0, 3, and 29. (C) Patient 3. The CSCC on the right ankle is shown at month 0, 6, 30, 42, and 60.
Patient 2
Patient 2 was a 97-year-old Chinese woman with a 1-year history of a painful mass on her left cheek. Physical examination in November 2015 demonstrated the presence of a 2 × 2 cm mass with an irregular edge and signs of erosion on her left cheek (Figure 1B). Skin biopsy confirmed the diagnosis of invasive SCC. The patient was treated with 6 courses of combination therapy. Darkening and regression of the tumor lesions were first detected in 2 months after the initiation of combination therapy, and a complete response was observed at 6 months. The only adverse effect reported by the patient was mucocutaneous dryness, but this resolved following cessation of treatment with acitretin. The patient has not experienced recurrence of CSCC during the 3-year follow-up.
Patient 3
Patient 3 was a 76-year-old Chinese woman with a 1-year history of a large mass on her right ankle that had progressively increased in size. Physical examination in September 2012 revealed a 4 × 4 cm mass with erosion on her right ankle (Figure 1C). Skin biopsy showed that the overlying epidermis was ulcerated with atypical keratinocytes invading the dermis. The keratinocytes were larger in the central parts of the tumor nests and formed eosinophilic horny pearls. The neoplastic keratinocytes had abundant eosinophilic and dyskeratotic cytoplasm and large vesicular nuclei with visible nucleoli and mitoses. An inflammatory reaction was evident, with many lymphocytes surrounding the stromal vessels and penetrating the tumor nests (Figure 2). Immunohistochemical analyses demonstrated that the tumor cells were positive for AE1/AE3 and P63, negative for Ber-EP4, and positive for Ki67 in about 40% of cells, confirming the diagnosis of invasive SCC (Figure 3). The patient was initially treated with 6 courses of combination therapy, which led to substantial regression of the tumor such that it resembled a small residual ulcer, and the patient then decided to discontinue the treatment. A local tumor relapse was detected 2 years later, so the patient resumed combination therapy and received an additional 24 courses of treatment, which resulted in complete remission. The only adverse effect experienced by the patient was mucocutaneous dryness. The patient remains alive and without tumor recurrence after a 6-year follow-up.
Figure 2 Representative images showing the histologic appearance of a tumor specimen from patient 3. (A) Section stained with hematoxylin-eosin (magnification, ×25). (B) Section stained with hematoxylin-eosin (magnification, ×200). (C) Section stained with hematoxylin-eosin (magnification, ×400).
Figure 3 Immunohistology of tumor specimens from patient 3. (A) Tumor cells were positive for AE1/AE3 (magnification, ×25). (B) Tumor cells were positive for P63 (magnification, ×25). (C) Tumor cells were negative for Ber-EP4 (magnification, ×25). (D) Around 40% of tumor cells were positive for Ki67 (magnification, ×400).
Discussion
A notable finding of this case report was that combination therapy achieved CSCC regression in all three patients within 2–6 months after the first administered dose of acitretin and clarithromycin. Furthermore, two of the patients exhibited complete remission of the tumor without scaring, thereby avoiding the need for surgery, while the other patient was successfully treated by tumor excision after combination therapy. Importantly, all three patients remained free of recurrence during a follow-up of 3–10 years. In addition, all three patients tolerated the treatment and experienced only minimal adverse effects (pseudotumor cerebri and mucocutaneous dryness) that resolved after discontinuation of acitretin. To our knowledge, this is the first study to report that combination therapy with acitretin and clarithromycin may be an effective treatment for CSCC.
The majority of CSCCs can be successfully eradicated by surgical excision with post-operative margin assessment or microscopically controlled surgery, although the cosmetic results are not always satisfactory. When patients cannot or refuse to undergo surgery, alternative treatment options include radiotherapy, intralesional chemotherapy and photodynamic therapy. Radiotherapy is a potentially curative treatment in cases of early-stage CSCC where resection would cause substantial cosmetic or functional deficits. However, radiotherapy alone has limited effectiveness as a treatment for advanced CSCC and is associated with adverse effects. Anti-PD-1 antibodies are considered a first-line systemic treatment for patients with metastatic or locally advanced CSCC who are not candidates for curative surgery or radiotherapy, although its use in elderly patients is limited by the toxicity of this chemotherapeutic agent (11, 12). Recently, epidermal growth factor receptor inhibitors combined with chemotherapy or radiotherapy have been used as second-line systemic treatments for advanced CSCC (11, 13), but further research is needed to fully characterize the efficacy and safety of these strategies (11, 14). As awareness grows regarding the increasing number of cases of post-surgical CSSC relapse, it is becoming recognized that improved non-surgical treatments for CSCC are urgently needed.
Retinoids are established drugs and have been shown to reduce the development of non-melanoma skin cancer in high risk individuals. Previous randomized controlled trials have demonstrated that systemic retinoids can prevent new primary CSCC or new precancerous lesions in immunocompromised patients after transplantation (15). Typically, retinoids are considered for patients who develop multiple tumors each year, have multiple CSCCs in high-risk areas, or have numerous actinic keratoses. Acitretin is a second-generation retinoid reported to exhibit anti-tumor activity against actinic keratosis, verrucous carcinomas and SCC. Previous studies of the use of acitretin in the management of CSCC are rare and summarized in Table 1 (5–7). Two of these three studies were case reports describing the use of acitretin in the chemoprevention of CSCC in 8 patients receiving BRAF inhibitors (5) and one patient who had been treated with phototherapy (6). Only one prior study has described the use of acitretin to treat CSCC: the combination of acitretin, intralesional 5-fluorouracil and 5-fluorouracil chemowraps was found to cause complete regression of widespread CSCCs on the legs of two patients (7). To our knowledge, the present case report is the first to report the successful use of acitretin in combination with clarithromycin as a treatment for CSCC.
Table 1 Summary of publications reporting cases and our cases of cutaneous squamous cell carcinoma (CSCC) treated with acitretin.
Reference Patients Treatment Response
Manalo et al. (7) Two patients (74-year-old female and 60-year-old female) with widespread CSCCs on the legs Acitretin (25 mg/d) combined with 5-fluorouracil and chemowraps for 6 months Complete clinical resolution
Anforth et al. (5) Eight patients treated with a BRAF inhibitor Acitretin (10–15 mg/d) for 1–13 months Suppression of CSCC development
Lebwohl et al. (6) A 40-year-old male patient with psoriasis who developed CSCC following PUVA phototherapy Acitretin (25 mg/d) for 25 months Suppression of CSCC development
Patient 1 An 83-year-old Chinese woman with invasive SCC on the left cheek Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 8 courses Complete remission (6
months); 10 years without
recurrence
Patient 2 A 97-year-old Chinese woman with invasive SCC on the left cheek Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 6 courses Complete remission (6
months); 3 years without
recurrence
Patient 3 An 76-year-old Chinese woman with invasive SCC on the right ankle Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 6 plus 24 courses resumed* Complete remission (24 months resumed); 6 years without recurrence
* After 6 courses of the combination therapy, this patient self-discontinued the treatment. Two years later, tumor recurrence was detected and the patient resumed the combination treatment for another 24 courses.
Acitretin monotherapy is generally considered to have a low efficacy and is associated with dose-dependent side effects such as mucocutaneous dryness, hair loss, elevated triglycerides, transient elevations in liver function tests and (with long-term treatment) diffuse idiopathic skeletal hyperostosis. In view of the above issues, the patients in this case report were also administered clarithromycin, a macrolide antibiotic that has been reported to enhance the activity of conventional chemotherapy agents in cell lines and animal experiments (8, 9). Furthermore, long-term administration of clarithromycin was reported to prolong the survival time of patients with lung cancer (16). We speculate that the excellent responses to combination therapy observed in all three patients in the present report were due to the antiangiogenic and antitumor effects of clarithromycin, which synergized with the anti-tumor actions of acitretin.
Combination therapy with acitretin and clarithromycin was well-tolerated by all three patients in this study. We summarized the common characteristics of the three patients in our study which may somehow contributed to the successful treatment outcomes. We found that all of them were female older than 75 and refused large-scale surgical treatment. The pathological diagnoses were poorly to well-differentiated invasive SCC. Their tumors were all single and without any metastases. We believe that elderly patients with slow tumor growth respond well to drug treatment and this can help them to avoid the risks associated with surgery. For young patients with multiple or larger tumors, the drug treatment can also be used to reduce the size of the tumor, furthermore it will be beneficial to surgical resection. The only adverse effects reported were pseudotumor cerebri and mucocutaneous dryness, which are known side-effects of acitretin. Therefore, we suggest that the combination of acitretin and clarithromycin may represent an effective and safe therapy for CSCC.
As far as we are aware, this is the first study to report that combination therapy with acitretin and clarithromycin may be an effective and safe treatment for CSCC. Additional studies are required to further assess the efficacy and safety of combination therapy with acitretin and clarithromycin in the treatment of CSCC in larger populations and to investigate the underlying molecular mechanisms.
Data Availability Statement
The original contributions generated in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.
Ethics Statement
The studies involving human participants were reviewed and approved by The study was approved by the Medical Technology Clinical Application Ethics Committee of the First Affiliated Hospital of Fujian (No. [2015]084). The patients/participants provided their written informed consent to participate in this study. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.
Author Contributions
YZha: conception and design. CJ: administrative support. YZha, YZhu, and HW: provision of study materials or patients. CJ and YZhu: data analysis and interpretation. All authors: collection, assembly of data, manuscript writing, and final approval of manuscript.
Conflict of Interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The authors acknowledge the help of all patients who participated in our study. | ACITRETIN, CLARITHROMYCIN | DrugsGivenReaction | CC BY | 33718249 | 19,083,671 | 2021 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Lip dry'. | Case Report: Successful Treatment of Cutaneous Squamous Cell Carcinoma in Three Patients With a Combination of Acitretin and Clarithromycin.
Few studies have evaluated acitretin as a primary treatment for cutaneous squamous cell carcinoma (CSCC). We report, for the first time, three cases of CSCC successfully treated with acitretin and clarithromycin. A literature review on this subject was also was performed. This case report included three patients with CSCC treated with acitretin and clarithromycin at the First Affiliated Hospital of Fujian Medical University (2008-2019). Patient 1 (83-year-old woman, ulcerated mass on the left cheek), patient 2 (97-year-old woman, painful mass on the left cheek) and patient 3 (76-year-old woman, large mass on the right ankle) received 8, 6, and 30 courses of combination therapy. All patients tolerated the adverse effects (pseudotumor cerebri and mucocutaneous dryness) and achieved complete regression within 6 months. Patients 1, 2, and 3 have not experienced recurrence during a 10-, 3-, and 6-year follow-up. Acitretin has limited efficacy as a monotherapy for CSCC. Our experience indicates that combination therapy with acitretin and clarithromycin may be an effective and well-tolerated treatment for unresectable CSCC.
Introduction
Given its increasing incidence and potential for poor outcomes (1–3), cutaneous squamous cell carcinoma (CSCC) is emerging as a public health problem. High-risk CSCC can be difficult to treat by surgical excision because the neoplasm can extend or infiltrate beyond the visible borders of the lesion. A wide range of non-surgical treatments are available for high-risk CSCC. Acitretin is a second generation retinoid shown to reduce the incidence of new primary non-melanoma skin cancers in immunocompromised transplant recipients (4) and patients receiving BRAF inhibitor treatment (5). Acitretin is considered to have limited efficacy as a monotherapy for CSCC but seems more effective when used in combination with other agents (6, 7). Clarithromycin is an antibacterial drug that has been reported to enhance the anti-tumor activity of chemotherapeutic drugs in vitro and in vivo (8, 9). However, no previous study has evaluated the combination of acitretin and clarithromycin in the treatment of CSCC. We present 3 cases illustrating that combination therapy with acitretin and clarithromycin might be a promising option for patients with CSCC who are unwilling or unable to undergo surgery.
Case Description
Ethical Statement
The study was approved by the Medical Technology Clinical Application Ethics Committee of the First Affiliated Hospital of Fujian (No. [2015]084). Before treatment, the patients were fully informed about the potential advantages and disadvantages of combination therapy with acitretin and clarithromycin, including the possible side effects of these drugs. All patients provided written informed consent for treatment and agreed to attend follow-up visits every 2 weeks. All patients provided written informed consent for data analysis and publication.
Patients
The present case report represents a retrospective analysis of the medical records of patients with CSCC who were treated with a combination of acitretin and clarithromycin at the Dermatology Department of the First Affiliated Hospital of Fujian between 2008 and 2018. Inclusion criteria consisted of a diagnosis of SCC by pathological biopsy and a history as well as tumor diameter ≥1 cm (10).
A total of 3 patients were involved in our study. Acitretin and clarithromycin were administered together to the patients from the time when skin biopsy confirmed the diagnosis of invasive SCC. We defined oral 30 mg/day of acitretin for 4 weeks and oral 500 mg/day of clarithromycin for 3 weeks as 1-month treatment course. There was no interval time between courses.
Once biochemical investigations, including measurements of serum creatinine, cholesterol, triglycerides, and potassium appeared abnormal, the treatment was stopped or the dosage was reduced.
Patient 1
Patient 1 was an 83-year-old Chinese woman with a 2-year history of an asymptomatic ulcerated mass on her left cheek. Physical examination in March 2008 revealed a 4 × 3.5 cm crater-like mass with central ulceration on her left cheek (Figure 1A). Skin biopsy revealed the presence of atypical epithelial tumor cell formations extending beyond the epidermis into the underlying dermis, cornified cells and horny pearls, thereby confirming the diagnosis of invasive SCC. The patient showed a favorable response to acitretin and clarithromycin within 2 months after the first oral administration. The patient declined long-term treatment due to the development of adverse effects (pseudotumor cerebri and dryness of the lips and skin). Therefore, CO2 laser resection of the CSCC was performed after a total of 6 courses of combination therapy, and the patient then received an additional two courses of combination therapy. This patient has lived without tumor recurrence for 10 years.
Figure 1 Photographs showing the regression of cutaneous squamous cell carcinoma (CSCC) in three patients treated with acitretin and clarithromycin. (A) Patient 1. The CSCC on the left cheek is shown at month 0 (before initiation of combination therapy), 5 and 12. (B) Patient 2. The CSCC on the left cheek is shown at month 0, 3, and 29. (C) Patient 3. The CSCC on the right ankle is shown at month 0, 6, 30, 42, and 60.
Patient 2
Patient 2 was a 97-year-old Chinese woman with a 1-year history of a painful mass on her left cheek. Physical examination in November 2015 demonstrated the presence of a 2 × 2 cm mass with an irregular edge and signs of erosion on her left cheek (Figure 1B). Skin biopsy confirmed the diagnosis of invasive SCC. The patient was treated with 6 courses of combination therapy. Darkening and regression of the tumor lesions were first detected in 2 months after the initiation of combination therapy, and a complete response was observed at 6 months. The only adverse effect reported by the patient was mucocutaneous dryness, but this resolved following cessation of treatment with acitretin. The patient has not experienced recurrence of CSCC during the 3-year follow-up.
Patient 3
Patient 3 was a 76-year-old Chinese woman with a 1-year history of a large mass on her right ankle that had progressively increased in size. Physical examination in September 2012 revealed a 4 × 4 cm mass with erosion on her right ankle (Figure 1C). Skin biopsy showed that the overlying epidermis was ulcerated with atypical keratinocytes invading the dermis. The keratinocytes were larger in the central parts of the tumor nests and formed eosinophilic horny pearls. The neoplastic keratinocytes had abundant eosinophilic and dyskeratotic cytoplasm and large vesicular nuclei with visible nucleoli and mitoses. An inflammatory reaction was evident, with many lymphocytes surrounding the stromal vessels and penetrating the tumor nests (Figure 2). Immunohistochemical analyses demonstrated that the tumor cells were positive for AE1/AE3 and P63, negative for Ber-EP4, and positive for Ki67 in about 40% of cells, confirming the diagnosis of invasive SCC (Figure 3). The patient was initially treated with 6 courses of combination therapy, which led to substantial regression of the tumor such that it resembled a small residual ulcer, and the patient then decided to discontinue the treatment. A local tumor relapse was detected 2 years later, so the patient resumed combination therapy and received an additional 24 courses of treatment, which resulted in complete remission. The only adverse effect experienced by the patient was mucocutaneous dryness. The patient remains alive and without tumor recurrence after a 6-year follow-up.
Figure 2 Representative images showing the histologic appearance of a tumor specimen from patient 3. (A) Section stained with hematoxylin-eosin (magnification, ×25). (B) Section stained with hematoxylin-eosin (magnification, ×200). (C) Section stained with hematoxylin-eosin (magnification, ×400).
Figure 3 Immunohistology of tumor specimens from patient 3. (A) Tumor cells were positive for AE1/AE3 (magnification, ×25). (B) Tumor cells were positive for P63 (magnification, ×25). (C) Tumor cells were negative for Ber-EP4 (magnification, ×25). (D) Around 40% of tumor cells were positive for Ki67 (magnification, ×400).
Discussion
A notable finding of this case report was that combination therapy achieved CSCC regression in all three patients within 2–6 months after the first administered dose of acitretin and clarithromycin. Furthermore, two of the patients exhibited complete remission of the tumor without scaring, thereby avoiding the need for surgery, while the other patient was successfully treated by tumor excision after combination therapy. Importantly, all three patients remained free of recurrence during a follow-up of 3–10 years. In addition, all three patients tolerated the treatment and experienced only minimal adverse effects (pseudotumor cerebri and mucocutaneous dryness) that resolved after discontinuation of acitretin. To our knowledge, this is the first study to report that combination therapy with acitretin and clarithromycin may be an effective treatment for CSCC.
The majority of CSCCs can be successfully eradicated by surgical excision with post-operative margin assessment or microscopically controlled surgery, although the cosmetic results are not always satisfactory. When patients cannot or refuse to undergo surgery, alternative treatment options include radiotherapy, intralesional chemotherapy and photodynamic therapy. Radiotherapy is a potentially curative treatment in cases of early-stage CSCC where resection would cause substantial cosmetic or functional deficits. However, radiotherapy alone has limited effectiveness as a treatment for advanced CSCC and is associated with adverse effects. Anti-PD-1 antibodies are considered a first-line systemic treatment for patients with metastatic or locally advanced CSCC who are not candidates for curative surgery or radiotherapy, although its use in elderly patients is limited by the toxicity of this chemotherapeutic agent (11, 12). Recently, epidermal growth factor receptor inhibitors combined with chemotherapy or radiotherapy have been used as second-line systemic treatments for advanced CSCC (11, 13), but further research is needed to fully characterize the efficacy and safety of these strategies (11, 14). As awareness grows regarding the increasing number of cases of post-surgical CSSC relapse, it is becoming recognized that improved non-surgical treatments for CSCC are urgently needed.
Retinoids are established drugs and have been shown to reduce the development of non-melanoma skin cancer in high risk individuals. Previous randomized controlled trials have demonstrated that systemic retinoids can prevent new primary CSCC or new precancerous lesions in immunocompromised patients after transplantation (15). Typically, retinoids are considered for patients who develop multiple tumors each year, have multiple CSCCs in high-risk areas, or have numerous actinic keratoses. Acitretin is a second-generation retinoid reported to exhibit anti-tumor activity against actinic keratosis, verrucous carcinomas and SCC. Previous studies of the use of acitretin in the management of CSCC are rare and summarized in Table 1 (5–7). Two of these three studies were case reports describing the use of acitretin in the chemoprevention of CSCC in 8 patients receiving BRAF inhibitors (5) and one patient who had been treated with phototherapy (6). Only one prior study has described the use of acitretin to treat CSCC: the combination of acitretin, intralesional 5-fluorouracil and 5-fluorouracil chemowraps was found to cause complete regression of widespread CSCCs on the legs of two patients (7). To our knowledge, the present case report is the first to report the successful use of acitretin in combination with clarithromycin as a treatment for CSCC.
Table 1 Summary of publications reporting cases and our cases of cutaneous squamous cell carcinoma (CSCC) treated with acitretin.
Reference Patients Treatment Response
Manalo et al. (7) Two patients (74-year-old female and 60-year-old female) with widespread CSCCs on the legs Acitretin (25 mg/d) combined with 5-fluorouracil and chemowraps for 6 months Complete clinical resolution
Anforth et al. (5) Eight patients treated with a BRAF inhibitor Acitretin (10–15 mg/d) for 1–13 months Suppression of CSCC development
Lebwohl et al. (6) A 40-year-old male patient with psoriasis who developed CSCC following PUVA phototherapy Acitretin (25 mg/d) for 25 months Suppression of CSCC development
Patient 1 An 83-year-old Chinese woman with invasive SCC on the left cheek Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 8 courses Complete remission (6
months); 10 years without
recurrence
Patient 2 A 97-year-old Chinese woman with invasive SCC on the left cheek Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 6 courses Complete remission (6
months); 3 years without
recurrence
Patient 3 An 76-year-old Chinese woman with invasive SCC on the right ankle Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 6 plus 24 courses resumed* Complete remission (24 months resumed); 6 years without recurrence
* After 6 courses of the combination therapy, this patient self-discontinued the treatment. Two years later, tumor recurrence was detected and the patient resumed the combination treatment for another 24 courses.
Acitretin monotherapy is generally considered to have a low efficacy and is associated with dose-dependent side effects such as mucocutaneous dryness, hair loss, elevated triglycerides, transient elevations in liver function tests and (with long-term treatment) diffuse idiopathic skeletal hyperostosis. In view of the above issues, the patients in this case report were also administered clarithromycin, a macrolide antibiotic that has been reported to enhance the activity of conventional chemotherapy agents in cell lines and animal experiments (8, 9). Furthermore, long-term administration of clarithromycin was reported to prolong the survival time of patients with lung cancer (16). We speculate that the excellent responses to combination therapy observed in all three patients in the present report were due to the antiangiogenic and antitumor effects of clarithromycin, which synergized with the anti-tumor actions of acitretin.
Combination therapy with acitretin and clarithromycin was well-tolerated by all three patients in this study. We summarized the common characteristics of the three patients in our study which may somehow contributed to the successful treatment outcomes. We found that all of them were female older than 75 and refused large-scale surgical treatment. The pathological diagnoses were poorly to well-differentiated invasive SCC. Their tumors were all single and without any metastases. We believe that elderly patients with slow tumor growth respond well to drug treatment and this can help them to avoid the risks associated with surgery. For young patients with multiple or larger tumors, the drug treatment can also be used to reduce the size of the tumor, furthermore it will be beneficial to surgical resection. The only adverse effects reported were pseudotumor cerebri and mucocutaneous dryness, which are known side-effects of acitretin. Therefore, we suggest that the combination of acitretin and clarithromycin may represent an effective and safe therapy for CSCC.
As far as we are aware, this is the first study to report that combination therapy with acitretin and clarithromycin may be an effective and safe treatment for CSCC. Additional studies are required to further assess the efficacy and safety of combination therapy with acitretin and clarithromycin in the treatment of CSCC in larger populations and to investigate the underlying molecular mechanisms.
Data Availability Statement
The original contributions generated in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.
Ethics Statement
The studies involving human participants were reviewed and approved by The study was approved by the Medical Technology Clinical Application Ethics Committee of the First Affiliated Hospital of Fujian (No. [2015]084). The patients/participants provided their written informed consent to participate in this study. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.
Author Contributions
YZha: conception and design. CJ: administrative support. YZha, YZhu, and HW: provision of study materials or patients. CJ and YZhu: data analysis and interpretation. All authors: collection, assembly of data, manuscript writing, and final approval of manuscript.
Conflict of Interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The authors acknowledge the help of all patients who participated in our study. | ACITRETIN, CLARITHROMYCIN | DrugsGivenReaction | CC BY | 33718249 | 19,083,671 | 2021 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Off label use'. | Case Report: Successful Treatment of Cutaneous Squamous Cell Carcinoma in Three Patients With a Combination of Acitretin and Clarithromycin.
Few studies have evaluated acitretin as a primary treatment for cutaneous squamous cell carcinoma (CSCC). We report, for the first time, three cases of CSCC successfully treated with acitretin and clarithromycin. A literature review on this subject was also was performed. This case report included three patients with CSCC treated with acitretin and clarithromycin at the First Affiliated Hospital of Fujian Medical University (2008-2019). Patient 1 (83-year-old woman, ulcerated mass on the left cheek), patient 2 (97-year-old woman, painful mass on the left cheek) and patient 3 (76-year-old woman, large mass on the right ankle) received 8, 6, and 30 courses of combination therapy. All patients tolerated the adverse effects (pseudotumor cerebri and mucocutaneous dryness) and achieved complete regression within 6 months. Patients 1, 2, and 3 have not experienced recurrence during a 10-, 3-, and 6-year follow-up. Acitretin has limited efficacy as a monotherapy for CSCC. Our experience indicates that combination therapy with acitretin and clarithromycin may be an effective and well-tolerated treatment for unresectable CSCC.
Introduction
Given its increasing incidence and potential for poor outcomes (1–3), cutaneous squamous cell carcinoma (CSCC) is emerging as a public health problem. High-risk CSCC can be difficult to treat by surgical excision because the neoplasm can extend or infiltrate beyond the visible borders of the lesion. A wide range of non-surgical treatments are available for high-risk CSCC. Acitretin is a second generation retinoid shown to reduce the incidence of new primary non-melanoma skin cancers in immunocompromised transplant recipients (4) and patients receiving BRAF inhibitor treatment (5). Acitretin is considered to have limited efficacy as a monotherapy for CSCC but seems more effective when used in combination with other agents (6, 7). Clarithromycin is an antibacterial drug that has been reported to enhance the anti-tumor activity of chemotherapeutic drugs in vitro and in vivo (8, 9). However, no previous study has evaluated the combination of acitretin and clarithromycin in the treatment of CSCC. We present 3 cases illustrating that combination therapy with acitretin and clarithromycin might be a promising option for patients with CSCC who are unwilling or unable to undergo surgery.
Case Description
Ethical Statement
The study was approved by the Medical Technology Clinical Application Ethics Committee of the First Affiliated Hospital of Fujian (No. [2015]084). Before treatment, the patients were fully informed about the potential advantages and disadvantages of combination therapy with acitretin and clarithromycin, including the possible side effects of these drugs. All patients provided written informed consent for treatment and agreed to attend follow-up visits every 2 weeks. All patients provided written informed consent for data analysis and publication.
Patients
The present case report represents a retrospective analysis of the medical records of patients with CSCC who were treated with a combination of acitretin and clarithromycin at the Dermatology Department of the First Affiliated Hospital of Fujian between 2008 and 2018. Inclusion criteria consisted of a diagnosis of SCC by pathological biopsy and a history as well as tumor diameter ≥1 cm (10).
A total of 3 patients were involved in our study. Acitretin and clarithromycin were administered together to the patients from the time when skin biopsy confirmed the diagnosis of invasive SCC. We defined oral 30 mg/day of acitretin for 4 weeks and oral 500 mg/day of clarithromycin for 3 weeks as 1-month treatment course. There was no interval time between courses.
Once biochemical investigations, including measurements of serum creatinine, cholesterol, triglycerides, and potassium appeared abnormal, the treatment was stopped or the dosage was reduced.
Patient 1
Patient 1 was an 83-year-old Chinese woman with a 2-year history of an asymptomatic ulcerated mass on her left cheek. Physical examination in March 2008 revealed a 4 × 3.5 cm crater-like mass with central ulceration on her left cheek (Figure 1A). Skin biopsy revealed the presence of atypical epithelial tumor cell formations extending beyond the epidermis into the underlying dermis, cornified cells and horny pearls, thereby confirming the diagnosis of invasive SCC. The patient showed a favorable response to acitretin and clarithromycin within 2 months after the first oral administration. The patient declined long-term treatment due to the development of adverse effects (pseudotumor cerebri and dryness of the lips and skin). Therefore, CO2 laser resection of the CSCC was performed after a total of 6 courses of combination therapy, and the patient then received an additional two courses of combination therapy. This patient has lived without tumor recurrence for 10 years.
Figure 1 Photographs showing the regression of cutaneous squamous cell carcinoma (CSCC) in three patients treated with acitretin and clarithromycin. (A) Patient 1. The CSCC on the left cheek is shown at month 0 (before initiation of combination therapy), 5 and 12. (B) Patient 2. The CSCC on the left cheek is shown at month 0, 3, and 29. (C) Patient 3. The CSCC on the right ankle is shown at month 0, 6, 30, 42, and 60.
Patient 2
Patient 2 was a 97-year-old Chinese woman with a 1-year history of a painful mass on her left cheek. Physical examination in November 2015 demonstrated the presence of a 2 × 2 cm mass with an irregular edge and signs of erosion on her left cheek (Figure 1B). Skin biopsy confirmed the diagnosis of invasive SCC. The patient was treated with 6 courses of combination therapy. Darkening and regression of the tumor lesions were first detected in 2 months after the initiation of combination therapy, and a complete response was observed at 6 months. The only adverse effect reported by the patient was mucocutaneous dryness, but this resolved following cessation of treatment with acitretin. The patient has not experienced recurrence of CSCC during the 3-year follow-up.
Patient 3
Patient 3 was a 76-year-old Chinese woman with a 1-year history of a large mass on her right ankle that had progressively increased in size. Physical examination in September 2012 revealed a 4 × 4 cm mass with erosion on her right ankle (Figure 1C). Skin biopsy showed that the overlying epidermis was ulcerated with atypical keratinocytes invading the dermis. The keratinocytes were larger in the central parts of the tumor nests and formed eosinophilic horny pearls. The neoplastic keratinocytes had abundant eosinophilic and dyskeratotic cytoplasm and large vesicular nuclei with visible nucleoli and mitoses. An inflammatory reaction was evident, with many lymphocytes surrounding the stromal vessels and penetrating the tumor nests (Figure 2). Immunohistochemical analyses demonstrated that the tumor cells were positive for AE1/AE3 and P63, negative for Ber-EP4, and positive for Ki67 in about 40% of cells, confirming the diagnosis of invasive SCC (Figure 3). The patient was initially treated with 6 courses of combination therapy, which led to substantial regression of the tumor such that it resembled a small residual ulcer, and the patient then decided to discontinue the treatment. A local tumor relapse was detected 2 years later, so the patient resumed combination therapy and received an additional 24 courses of treatment, which resulted in complete remission. The only adverse effect experienced by the patient was mucocutaneous dryness. The patient remains alive and without tumor recurrence after a 6-year follow-up.
Figure 2 Representative images showing the histologic appearance of a tumor specimen from patient 3. (A) Section stained with hematoxylin-eosin (magnification, ×25). (B) Section stained with hematoxylin-eosin (magnification, ×200). (C) Section stained with hematoxylin-eosin (magnification, ×400).
Figure 3 Immunohistology of tumor specimens from patient 3. (A) Tumor cells were positive for AE1/AE3 (magnification, ×25). (B) Tumor cells were positive for P63 (magnification, ×25). (C) Tumor cells were negative for Ber-EP4 (magnification, ×25). (D) Around 40% of tumor cells were positive for Ki67 (magnification, ×400).
Discussion
A notable finding of this case report was that combination therapy achieved CSCC regression in all three patients within 2–6 months after the first administered dose of acitretin and clarithromycin. Furthermore, two of the patients exhibited complete remission of the tumor without scaring, thereby avoiding the need for surgery, while the other patient was successfully treated by tumor excision after combination therapy. Importantly, all three patients remained free of recurrence during a follow-up of 3–10 years. In addition, all three patients tolerated the treatment and experienced only minimal adverse effects (pseudotumor cerebri and mucocutaneous dryness) that resolved after discontinuation of acitretin. To our knowledge, this is the first study to report that combination therapy with acitretin and clarithromycin may be an effective treatment for CSCC.
The majority of CSCCs can be successfully eradicated by surgical excision with post-operative margin assessment or microscopically controlled surgery, although the cosmetic results are not always satisfactory. When patients cannot or refuse to undergo surgery, alternative treatment options include radiotherapy, intralesional chemotherapy and photodynamic therapy. Radiotherapy is a potentially curative treatment in cases of early-stage CSCC where resection would cause substantial cosmetic or functional deficits. However, radiotherapy alone has limited effectiveness as a treatment for advanced CSCC and is associated with adverse effects. Anti-PD-1 antibodies are considered a first-line systemic treatment for patients with metastatic or locally advanced CSCC who are not candidates for curative surgery or radiotherapy, although its use in elderly patients is limited by the toxicity of this chemotherapeutic agent (11, 12). Recently, epidermal growth factor receptor inhibitors combined with chemotherapy or radiotherapy have been used as second-line systemic treatments for advanced CSCC (11, 13), but further research is needed to fully characterize the efficacy and safety of these strategies (11, 14). As awareness grows regarding the increasing number of cases of post-surgical CSSC relapse, it is becoming recognized that improved non-surgical treatments for CSCC are urgently needed.
Retinoids are established drugs and have been shown to reduce the development of non-melanoma skin cancer in high risk individuals. Previous randomized controlled trials have demonstrated that systemic retinoids can prevent new primary CSCC or new precancerous lesions in immunocompromised patients after transplantation (15). Typically, retinoids are considered for patients who develop multiple tumors each year, have multiple CSCCs in high-risk areas, or have numerous actinic keratoses. Acitretin is a second-generation retinoid reported to exhibit anti-tumor activity against actinic keratosis, verrucous carcinomas and SCC. Previous studies of the use of acitretin in the management of CSCC are rare and summarized in Table 1 (5–7). Two of these three studies were case reports describing the use of acitretin in the chemoprevention of CSCC in 8 patients receiving BRAF inhibitors (5) and one patient who had been treated with phototherapy (6). Only one prior study has described the use of acitretin to treat CSCC: the combination of acitretin, intralesional 5-fluorouracil and 5-fluorouracil chemowraps was found to cause complete regression of widespread CSCCs on the legs of two patients (7). To our knowledge, the present case report is the first to report the successful use of acitretin in combination with clarithromycin as a treatment for CSCC.
Table 1 Summary of publications reporting cases and our cases of cutaneous squamous cell carcinoma (CSCC) treated with acitretin.
Reference Patients Treatment Response
Manalo et al. (7) Two patients (74-year-old female and 60-year-old female) with widespread CSCCs on the legs Acitretin (25 mg/d) combined with 5-fluorouracil and chemowraps for 6 months Complete clinical resolution
Anforth et al. (5) Eight patients treated with a BRAF inhibitor Acitretin (10–15 mg/d) for 1–13 months Suppression of CSCC development
Lebwohl et al. (6) A 40-year-old male patient with psoriasis who developed CSCC following PUVA phototherapy Acitretin (25 mg/d) for 25 months Suppression of CSCC development
Patient 1 An 83-year-old Chinese woman with invasive SCC on the left cheek Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 8 courses Complete remission (6
months); 10 years without
recurrence
Patient 2 A 97-year-old Chinese woman with invasive SCC on the left cheek Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 6 courses Complete remission (6
months); 3 years without
recurrence
Patient 3 An 76-year-old Chinese woman with invasive SCC on the right ankle Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 6 plus 24 courses resumed* Complete remission (24 months resumed); 6 years without recurrence
* After 6 courses of the combination therapy, this patient self-discontinued the treatment. Two years later, tumor recurrence was detected and the patient resumed the combination treatment for another 24 courses.
Acitretin monotherapy is generally considered to have a low efficacy and is associated with dose-dependent side effects such as mucocutaneous dryness, hair loss, elevated triglycerides, transient elevations in liver function tests and (with long-term treatment) diffuse idiopathic skeletal hyperostosis. In view of the above issues, the patients in this case report were also administered clarithromycin, a macrolide antibiotic that has been reported to enhance the activity of conventional chemotherapy agents in cell lines and animal experiments (8, 9). Furthermore, long-term administration of clarithromycin was reported to prolong the survival time of patients with lung cancer (16). We speculate that the excellent responses to combination therapy observed in all three patients in the present report were due to the antiangiogenic and antitumor effects of clarithromycin, which synergized with the anti-tumor actions of acitretin.
Combination therapy with acitretin and clarithromycin was well-tolerated by all three patients in this study. We summarized the common characteristics of the three patients in our study which may somehow contributed to the successful treatment outcomes. We found that all of them were female older than 75 and refused large-scale surgical treatment. The pathological diagnoses were poorly to well-differentiated invasive SCC. Their tumors were all single and without any metastases. We believe that elderly patients with slow tumor growth respond well to drug treatment and this can help them to avoid the risks associated with surgery. For young patients with multiple or larger tumors, the drug treatment can also be used to reduce the size of the tumor, furthermore it will be beneficial to surgical resection. The only adverse effects reported were pseudotumor cerebri and mucocutaneous dryness, which are known side-effects of acitretin. Therefore, we suggest that the combination of acitretin and clarithromycin may represent an effective and safe therapy for CSCC.
As far as we are aware, this is the first study to report that combination therapy with acitretin and clarithromycin may be an effective and safe treatment for CSCC. Additional studies are required to further assess the efficacy and safety of combination therapy with acitretin and clarithromycin in the treatment of CSCC in larger populations and to investigate the underlying molecular mechanisms.
Data Availability Statement
The original contributions generated in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.
Ethics Statement
The studies involving human participants were reviewed and approved by The study was approved by the Medical Technology Clinical Application Ethics Committee of the First Affiliated Hospital of Fujian (No. [2015]084). The patients/participants provided their written informed consent to participate in this study. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.
Author Contributions
YZha: conception and design. CJ: administrative support. YZha, YZhu, and HW: provision of study materials or patients. CJ and YZhu: data analysis and interpretation. All authors: collection, assembly of data, manuscript writing, and final approval of manuscript.
Conflict of Interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The authors acknowledge the help of all patients who participated in our study. | ACITRETIN, CLARITHROMYCIN | DrugsGivenReaction | CC BY | 33718249 | 19,083,671 | 2021 |
What was the administration route of drug 'ACITRETIN'? | Case Report: Successful Treatment of Cutaneous Squamous Cell Carcinoma in Three Patients With a Combination of Acitretin and Clarithromycin.
Few studies have evaluated acitretin as a primary treatment for cutaneous squamous cell carcinoma (CSCC). We report, for the first time, three cases of CSCC successfully treated with acitretin and clarithromycin. A literature review on this subject was also was performed. This case report included three patients with CSCC treated with acitretin and clarithromycin at the First Affiliated Hospital of Fujian Medical University (2008-2019). Patient 1 (83-year-old woman, ulcerated mass on the left cheek), patient 2 (97-year-old woman, painful mass on the left cheek) and patient 3 (76-year-old woman, large mass on the right ankle) received 8, 6, and 30 courses of combination therapy. All patients tolerated the adverse effects (pseudotumor cerebri and mucocutaneous dryness) and achieved complete regression within 6 months. Patients 1, 2, and 3 have not experienced recurrence during a 10-, 3-, and 6-year follow-up. Acitretin has limited efficacy as a monotherapy for CSCC. Our experience indicates that combination therapy with acitretin and clarithromycin may be an effective and well-tolerated treatment for unresectable CSCC.
Introduction
Given its increasing incidence and potential for poor outcomes (1–3), cutaneous squamous cell carcinoma (CSCC) is emerging as a public health problem. High-risk CSCC can be difficult to treat by surgical excision because the neoplasm can extend or infiltrate beyond the visible borders of the lesion. A wide range of non-surgical treatments are available for high-risk CSCC. Acitretin is a second generation retinoid shown to reduce the incidence of new primary non-melanoma skin cancers in immunocompromised transplant recipients (4) and patients receiving BRAF inhibitor treatment (5). Acitretin is considered to have limited efficacy as a monotherapy for CSCC but seems more effective when used in combination with other agents (6, 7). Clarithromycin is an antibacterial drug that has been reported to enhance the anti-tumor activity of chemotherapeutic drugs in vitro and in vivo (8, 9). However, no previous study has evaluated the combination of acitretin and clarithromycin in the treatment of CSCC. We present 3 cases illustrating that combination therapy with acitretin and clarithromycin might be a promising option for patients with CSCC who are unwilling or unable to undergo surgery.
Case Description
Ethical Statement
The study was approved by the Medical Technology Clinical Application Ethics Committee of the First Affiliated Hospital of Fujian (No. [2015]084). Before treatment, the patients were fully informed about the potential advantages and disadvantages of combination therapy with acitretin and clarithromycin, including the possible side effects of these drugs. All patients provided written informed consent for treatment and agreed to attend follow-up visits every 2 weeks. All patients provided written informed consent for data analysis and publication.
Patients
The present case report represents a retrospective analysis of the medical records of patients with CSCC who were treated with a combination of acitretin and clarithromycin at the Dermatology Department of the First Affiliated Hospital of Fujian between 2008 and 2018. Inclusion criteria consisted of a diagnosis of SCC by pathological biopsy and a history as well as tumor diameter ≥1 cm (10).
A total of 3 patients were involved in our study. Acitretin and clarithromycin were administered together to the patients from the time when skin biopsy confirmed the diagnosis of invasive SCC. We defined oral 30 mg/day of acitretin for 4 weeks and oral 500 mg/day of clarithromycin for 3 weeks as 1-month treatment course. There was no interval time between courses.
Once biochemical investigations, including measurements of serum creatinine, cholesterol, triglycerides, and potassium appeared abnormal, the treatment was stopped or the dosage was reduced.
Patient 1
Patient 1 was an 83-year-old Chinese woman with a 2-year history of an asymptomatic ulcerated mass on her left cheek. Physical examination in March 2008 revealed a 4 × 3.5 cm crater-like mass with central ulceration on her left cheek (Figure 1A). Skin biopsy revealed the presence of atypical epithelial tumor cell formations extending beyond the epidermis into the underlying dermis, cornified cells and horny pearls, thereby confirming the diagnosis of invasive SCC. The patient showed a favorable response to acitretin and clarithromycin within 2 months after the first oral administration. The patient declined long-term treatment due to the development of adverse effects (pseudotumor cerebri and dryness of the lips and skin). Therefore, CO2 laser resection of the CSCC was performed after a total of 6 courses of combination therapy, and the patient then received an additional two courses of combination therapy. This patient has lived without tumor recurrence for 10 years.
Figure 1 Photographs showing the regression of cutaneous squamous cell carcinoma (CSCC) in three patients treated with acitretin and clarithromycin. (A) Patient 1. The CSCC on the left cheek is shown at month 0 (before initiation of combination therapy), 5 and 12. (B) Patient 2. The CSCC on the left cheek is shown at month 0, 3, and 29. (C) Patient 3. The CSCC on the right ankle is shown at month 0, 6, 30, 42, and 60.
Patient 2
Patient 2 was a 97-year-old Chinese woman with a 1-year history of a painful mass on her left cheek. Physical examination in November 2015 demonstrated the presence of a 2 × 2 cm mass with an irregular edge and signs of erosion on her left cheek (Figure 1B). Skin biopsy confirmed the diagnosis of invasive SCC. The patient was treated with 6 courses of combination therapy. Darkening and regression of the tumor lesions were first detected in 2 months after the initiation of combination therapy, and a complete response was observed at 6 months. The only adverse effect reported by the patient was mucocutaneous dryness, but this resolved following cessation of treatment with acitretin. The patient has not experienced recurrence of CSCC during the 3-year follow-up.
Patient 3
Patient 3 was a 76-year-old Chinese woman with a 1-year history of a large mass on her right ankle that had progressively increased in size. Physical examination in September 2012 revealed a 4 × 4 cm mass with erosion on her right ankle (Figure 1C). Skin biopsy showed that the overlying epidermis was ulcerated with atypical keratinocytes invading the dermis. The keratinocytes were larger in the central parts of the tumor nests and formed eosinophilic horny pearls. The neoplastic keratinocytes had abundant eosinophilic and dyskeratotic cytoplasm and large vesicular nuclei with visible nucleoli and mitoses. An inflammatory reaction was evident, with many lymphocytes surrounding the stromal vessels and penetrating the tumor nests (Figure 2). Immunohistochemical analyses demonstrated that the tumor cells were positive for AE1/AE3 and P63, negative for Ber-EP4, and positive for Ki67 in about 40% of cells, confirming the diagnosis of invasive SCC (Figure 3). The patient was initially treated with 6 courses of combination therapy, which led to substantial regression of the tumor such that it resembled a small residual ulcer, and the patient then decided to discontinue the treatment. A local tumor relapse was detected 2 years later, so the patient resumed combination therapy and received an additional 24 courses of treatment, which resulted in complete remission. The only adverse effect experienced by the patient was mucocutaneous dryness. The patient remains alive and without tumor recurrence after a 6-year follow-up.
Figure 2 Representative images showing the histologic appearance of a tumor specimen from patient 3. (A) Section stained with hematoxylin-eosin (magnification, ×25). (B) Section stained with hematoxylin-eosin (magnification, ×200). (C) Section stained with hematoxylin-eosin (magnification, ×400).
Figure 3 Immunohistology of tumor specimens from patient 3. (A) Tumor cells were positive for AE1/AE3 (magnification, ×25). (B) Tumor cells were positive for P63 (magnification, ×25). (C) Tumor cells were negative for Ber-EP4 (magnification, ×25). (D) Around 40% of tumor cells were positive for Ki67 (magnification, ×400).
Discussion
A notable finding of this case report was that combination therapy achieved CSCC regression in all three patients within 2–6 months after the first administered dose of acitretin and clarithromycin. Furthermore, two of the patients exhibited complete remission of the tumor without scaring, thereby avoiding the need for surgery, while the other patient was successfully treated by tumor excision after combination therapy. Importantly, all three patients remained free of recurrence during a follow-up of 3–10 years. In addition, all three patients tolerated the treatment and experienced only minimal adverse effects (pseudotumor cerebri and mucocutaneous dryness) that resolved after discontinuation of acitretin. To our knowledge, this is the first study to report that combination therapy with acitretin and clarithromycin may be an effective treatment for CSCC.
The majority of CSCCs can be successfully eradicated by surgical excision with post-operative margin assessment or microscopically controlled surgery, although the cosmetic results are not always satisfactory. When patients cannot or refuse to undergo surgery, alternative treatment options include radiotherapy, intralesional chemotherapy and photodynamic therapy. Radiotherapy is a potentially curative treatment in cases of early-stage CSCC where resection would cause substantial cosmetic or functional deficits. However, radiotherapy alone has limited effectiveness as a treatment for advanced CSCC and is associated with adverse effects. Anti-PD-1 antibodies are considered a first-line systemic treatment for patients with metastatic or locally advanced CSCC who are not candidates for curative surgery or radiotherapy, although its use in elderly patients is limited by the toxicity of this chemotherapeutic agent (11, 12). Recently, epidermal growth factor receptor inhibitors combined with chemotherapy or radiotherapy have been used as second-line systemic treatments for advanced CSCC (11, 13), but further research is needed to fully characterize the efficacy and safety of these strategies (11, 14). As awareness grows regarding the increasing number of cases of post-surgical CSSC relapse, it is becoming recognized that improved non-surgical treatments for CSCC are urgently needed.
Retinoids are established drugs and have been shown to reduce the development of non-melanoma skin cancer in high risk individuals. Previous randomized controlled trials have demonstrated that systemic retinoids can prevent new primary CSCC or new precancerous lesions in immunocompromised patients after transplantation (15). Typically, retinoids are considered for patients who develop multiple tumors each year, have multiple CSCCs in high-risk areas, or have numerous actinic keratoses. Acitretin is a second-generation retinoid reported to exhibit anti-tumor activity against actinic keratosis, verrucous carcinomas and SCC. Previous studies of the use of acitretin in the management of CSCC are rare and summarized in Table 1 (5–7). Two of these three studies were case reports describing the use of acitretin in the chemoprevention of CSCC in 8 patients receiving BRAF inhibitors (5) and one patient who had been treated with phototherapy (6). Only one prior study has described the use of acitretin to treat CSCC: the combination of acitretin, intralesional 5-fluorouracil and 5-fluorouracil chemowraps was found to cause complete regression of widespread CSCCs on the legs of two patients (7). To our knowledge, the present case report is the first to report the successful use of acitretin in combination with clarithromycin as a treatment for CSCC.
Table 1 Summary of publications reporting cases and our cases of cutaneous squamous cell carcinoma (CSCC) treated with acitretin.
Reference Patients Treatment Response
Manalo et al. (7) Two patients (74-year-old female and 60-year-old female) with widespread CSCCs on the legs Acitretin (25 mg/d) combined with 5-fluorouracil and chemowraps for 6 months Complete clinical resolution
Anforth et al. (5) Eight patients treated with a BRAF inhibitor Acitretin (10–15 mg/d) for 1–13 months Suppression of CSCC development
Lebwohl et al. (6) A 40-year-old male patient with psoriasis who developed CSCC following PUVA phototherapy Acitretin (25 mg/d) for 25 months Suppression of CSCC development
Patient 1 An 83-year-old Chinese woman with invasive SCC on the left cheek Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 8 courses Complete remission (6
months); 10 years without
recurrence
Patient 2 A 97-year-old Chinese woman with invasive SCC on the left cheek Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 6 courses Complete remission (6
months); 3 years without
recurrence
Patient 3 An 76-year-old Chinese woman with invasive SCC on the right ankle Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 6 plus 24 courses resumed* Complete remission (24 months resumed); 6 years without recurrence
* After 6 courses of the combination therapy, this patient self-discontinued the treatment. Two years later, tumor recurrence was detected and the patient resumed the combination treatment for another 24 courses.
Acitretin monotherapy is generally considered to have a low efficacy and is associated with dose-dependent side effects such as mucocutaneous dryness, hair loss, elevated triglycerides, transient elevations in liver function tests and (with long-term treatment) diffuse idiopathic skeletal hyperostosis. In view of the above issues, the patients in this case report were also administered clarithromycin, a macrolide antibiotic that has been reported to enhance the activity of conventional chemotherapy agents in cell lines and animal experiments (8, 9). Furthermore, long-term administration of clarithromycin was reported to prolong the survival time of patients with lung cancer (16). We speculate that the excellent responses to combination therapy observed in all three patients in the present report were due to the antiangiogenic and antitumor effects of clarithromycin, which synergized with the anti-tumor actions of acitretin.
Combination therapy with acitretin and clarithromycin was well-tolerated by all three patients in this study. We summarized the common characteristics of the three patients in our study which may somehow contributed to the successful treatment outcomes. We found that all of them were female older than 75 and refused large-scale surgical treatment. The pathological diagnoses were poorly to well-differentiated invasive SCC. Their tumors were all single and without any metastases. We believe that elderly patients with slow tumor growth respond well to drug treatment and this can help them to avoid the risks associated with surgery. For young patients with multiple or larger tumors, the drug treatment can also be used to reduce the size of the tumor, furthermore it will be beneficial to surgical resection. The only adverse effects reported were pseudotumor cerebri and mucocutaneous dryness, which are known side-effects of acitretin. Therefore, we suggest that the combination of acitretin and clarithromycin may represent an effective and safe therapy for CSCC.
As far as we are aware, this is the first study to report that combination therapy with acitretin and clarithromycin may be an effective and safe treatment for CSCC. Additional studies are required to further assess the efficacy and safety of combination therapy with acitretin and clarithromycin in the treatment of CSCC in larger populations and to investigate the underlying molecular mechanisms.
Data Availability Statement
The original contributions generated in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.
Ethics Statement
The studies involving human participants were reviewed and approved by The study was approved by the Medical Technology Clinical Application Ethics Committee of the First Affiliated Hospital of Fujian (No. [2015]084). The patients/participants provided their written informed consent to participate in this study. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.
Author Contributions
YZha: conception and design. CJ: administrative support. YZha, YZhu, and HW: provision of study materials or patients. CJ and YZhu: data analysis and interpretation. All authors: collection, assembly of data, manuscript writing, and final approval of manuscript.
Conflict of Interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The authors acknowledge the help of all patients who participated in our study. | Oral | DrugAdministrationRoute | CC BY | 33718249 | 19,083,671 | 2021 |
What was the administration route of drug 'CLARITHROMYCIN'? | Case Report: Successful Treatment of Cutaneous Squamous Cell Carcinoma in Three Patients With a Combination of Acitretin and Clarithromycin.
Few studies have evaluated acitretin as a primary treatment for cutaneous squamous cell carcinoma (CSCC). We report, for the first time, three cases of CSCC successfully treated with acitretin and clarithromycin. A literature review on this subject was also was performed. This case report included three patients with CSCC treated with acitretin and clarithromycin at the First Affiliated Hospital of Fujian Medical University (2008-2019). Patient 1 (83-year-old woman, ulcerated mass on the left cheek), patient 2 (97-year-old woman, painful mass on the left cheek) and patient 3 (76-year-old woman, large mass on the right ankle) received 8, 6, and 30 courses of combination therapy. All patients tolerated the adverse effects (pseudotumor cerebri and mucocutaneous dryness) and achieved complete regression within 6 months. Patients 1, 2, and 3 have not experienced recurrence during a 10-, 3-, and 6-year follow-up. Acitretin has limited efficacy as a monotherapy for CSCC. Our experience indicates that combination therapy with acitretin and clarithromycin may be an effective and well-tolerated treatment for unresectable CSCC.
Introduction
Given its increasing incidence and potential for poor outcomes (1–3), cutaneous squamous cell carcinoma (CSCC) is emerging as a public health problem. High-risk CSCC can be difficult to treat by surgical excision because the neoplasm can extend or infiltrate beyond the visible borders of the lesion. A wide range of non-surgical treatments are available for high-risk CSCC. Acitretin is a second generation retinoid shown to reduce the incidence of new primary non-melanoma skin cancers in immunocompromised transplant recipients (4) and patients receiving BRAF inhibitor treatment (5). Acitretin is considered to have limited efficacy as a monotherapy for CSCC but seems more effective when used in combination with other agents (6, 7). Clarithromycin is an antibacterial drug that has been reported to enhance the anti-tumor activity of chemotherapeutic drugs in vitro and in vivo (8, 9). However, no previous study has evaluated the combination of acitretin and clarithromycin in the treatment of CSCC. We present 3 cases illustrating that combination therapy with acitretin and clarithromycin might be a promising option for patients with CSCC who are unwilling or unable to undergo surgery.
Case Description
Ethical Statement
The study was approved by the Medical Technology Clinical Application Ethics Committee of the First Affiliated Hospital of Fujian (No. [2015]084). Before treatment, the patients were fully informed about the potential advantages and disadvantages of combination therapy with acitretin and clarithromycin, including the possible side effects of these drugs. All patients provided written informed consent for treatment and agreed to attend follow-up visits every 2 weeks. All patients provided written informed consent for data analysis and publication.
Patients
The present case report represents a retrospective analysis of the medical records of patients with CSCC who were treated with a combination of acitretin and clarithromycin at the Dermatology Department of the First Affiliated Hospital of Fujian between 2008 and 2018. Inclusion criteria consisted of a diagnosis of SCC by pathological biopsy and a history as well as tumor diameter ≥1 cm (10).
A total of 3 patients were involved in our study. Acitretin and clarithromycin were administered together to the patients from the time when skin biopsy confirmed the diagnosis of invasive SCC. We defined oral 30 mg/day of acitretin for 4 weeks and oral 500 mg/day of clarithromycin for 3 weeks as 1-month treatment course. There was no interval time between courses.
Once biochemical investigations, including measurements of serum creatinine, cholesterol, triglycerides, and potassium appeared abnormal, the treatment was stopped or the dosage was reduced.
Patient 1
Patient 1 was an 83-year-old Chinese woman with a 2-year history of an asymptomatic ulcerated mass on her left cheek. Physical examination in March 2008 revealed a 4 × 3.5 cm crater-like mass with central ulceration on her left cheek (Figure 1A). Skin biopsy revealed the presence of atypical epithelial tumor cell formations extending beyond the epidermis into the underlying dermis, cornified cells and horny pearls, thereby confirming the diagnosis of invasive SCC. The patient showed a favorable response to acitretin and clarithromycin within 2 months after the first oral administration. The patient declined long-term treatment due to the development of adverse effects (pseudotumor cerebri and dryness of the lips and skin). Therefore, CO2 laser resection of the CSCC was performed after a total of 6 courses of combination therapy, and the patient then received an additional two courses of combination therapy. This patient has lived without tumor recurrence for 10 years.
Figure 1 Photographs showing the regression of cutaneous squamous cell carcinoma (CSCC) in three patients treated with acitretin and clarithromycin. (A) Patient 1. The CSCC on the left cheek is shown at month 0 (before initiation of combination therapy), 5 and 12. (B) Patient 2. The CSCC on the left cheek is shown at month 0, 3, and 29. (C) Patient 3. The CSCC on the right ankle is shown at month 0, 6, 30, 42, and 60.
Patient 2
Patient 2 was a 97-year-old Chinese woman with a 1-year history of a painful mass on her left cheek. Physical examination in November 2015 demonstrated the presence of a 2 × 2 cm mass with an irregular edge and signs of erosion on her left cheek (Figure 1B). Skin biopsy confirmed the diagnosis of invasive SCC. The patient was treated with 6 courses of combination therapy. Darkening and regression of the tumor lesions were first detected in 2 months after the initiation of combination therapy, and a complete response was observed at 6 months. The only adverse effect reported by the patient was mucocutaneous dryness, but this resolved following cessation of treatment with acitretin. The patient has not experienced recurrence of CSCC during the 3-year follow-up.
Patient 3
Patient 3 was a 76-year-old Chinese woman with a 1-year history of a large mass on her right ankle that had progressively increased in size. Physical examination in September 2012 revealed a 4 × 4 cm mass with erosion on her right ankle (Figure 1C). Skin biopsy showed that the overlying epidermis was ulcerated with atypical keratinocytes invading the dermis. The keratinocytes were larger in the central parts of the tumor nests and formed eosinophilic horny pearls. The neoplastic keratinocytes had abundant eosinophilic and dyskeratotic cytoplasm and large vesicular nuclei with visible nucleoli and mitoses. An inflammatory reaction was evident, with many lymphocytes surrounding the stromal vessels and penetrating the tumor nests (Figure 2). Immunohistochemical analyses demonstrated that the tumor cells were positive for AE1/AE3 and P63, negative for Ber-EP4, and positive for Ki67 in about 40% of cells, confirming the diagnosis of invasive SCC (Figure 3). The patient was initially treated with 6 courses of combination therapy, which led to substantial regression of the tumor such that it resembled a small residual ulcer, and the patient then decided to discontinue the treatment. A local tumor relapse was detected 2 years later, so the patient resumed combination therapy and received an additional 24 courses of treatment, which resulted in complete remission. The only adverse effect experienced by the patient was mucocutaneous dryness. The patient remains alive and without tumor recurrence after a 6-year follow-up.
Figure 2 Representative images showing the histologic appearance of a tumor specimen from patient 3. (A) Section stained with hematoxylin-eosin (magnification, ×25). (B) Section stained with hematoxylin-eosin (magnification, ×200). (C) Section stained with hematoxylin-eosin (magnification, ×400).
Figure 3 Immunohistology of tumor specimens from patient 3. (A) Tumor cells were positive for AE1/AE3 (magnification, ×25). (B) Tumor cells were positive for P63 (magnification, ×25). (C) Tumor cells were negative for Ber-EP4 (magnification, ×25). (D) Around 40% of tumor cells were positive for Ki67 (magnification, ×400).
Discussion
A notable finding of this case report was that combination therapy achieved CSCC regression in all three patients within 2–6 months after the first administered dose of acitretin and clarithromycin. Furthermore, two of the patients exhibited complete remission of the tumor without scaring, thereby avoiding the need for surgery, while the other patient was successfully treated by tumor excision after combination therapy. Importantly, all three patients remained free of recurrence during a follow-up of 3–10 years. In addition, all three patients tolerated the treatment and experienced only minimal adverse effects (pseudotumor cerebri and mucocutaneous dryness) that resolved after discontinuation of acitretin. To our knowledge, this is the first study to report that combination therapy with acitretin and clarithromycin may be an effective treatment for CSCC.
The majority of CSCCs can be successfully eradicated by surgical excision with post-operative margin assessment or microscopically controlled surgery, although the cosmetic results are not always satisfactory. When patients cannot or refuse to undergo surgery, alternative treatment options include radiotherapy, intralesional chemotherapy and photodynamic therapy. Radiotherapy is a potentially curative treatment in cases of early-stage CSCC where resection would cause substantial cosmetic or functional deficits. However, radiotherapy alone has limited effectiveness as a treatment for advanced CSCC and is associated with adverse effects. Anti-PD-1 antibodies are considered a first-line systemic treatment for patients with metastatic or locally advanced CSCC who are not candidates for curative surgery or radiotherapy, although its use in elderly patients is limited by the toxicity of this chemotherapeutic agent (11, 12). Recently, epidermal growth factor receptor inhibitors combined with chemotherapy or radiotherapy have been used as second-line systemic treatments for advanced CSCC (11, 13), but further research is needed to fully characterize the efficacy and safety of these strategies (11, 14). As awareness grows regarding the increasing number of cases of post-surgical CSSC relapse, it is becoming recognized that improved non-surgical treatments for CSCC are urgently needed.
Retinoids are established drugs and have been shown to reduce the development of non-melanoma skin cancer in high risk individuals. Previous randomized controlled trials have demonstrated that systemic retinoids can prevent new primary CSCC or new precancerous lesions in immunocompromised patients after transplantation (15). Typically, retinoids are considered for patients who develop multiple tumors each year, have multiple CSCCs in high-risk areas, or have numerous actinic keratoses. Acitretin is a second-generation retinoid reported to exhibit anti-tumor activity against actinic keratosis, verrucous carcinomas and SCC. Previous studies of the use of acitretin in the management of CSCC are rare and summarized in Table 1 (5–7). Two of these three studies were case reports describing the use of acitretin in the chemoprevention of CSCC in 8 patients receiving BRAF inhibitors (5) and one patient who had been treated with phototherapy (6). Only one prior study has described the use of acitretin to treat CSCC: the combination of acitretin, intralesional 5-fluorouracil and 5-fluorouracil chemowraps was found to cause complete regression of widespread CSCCs on the legs of two patients (7). To our knowledge, the present case report is the first to report the successful use of acitretin in combination with clarithromycin as a treatment for CSCC.
Table 1 Summary of publications reporting cases and our cases of cutaneous squamous cell carcinoma (CSCC) treated with acitretin.
Reference Patients Treatment Response
Manalo et al. (7) Two patients (74-year-old female and 60-year-old female) with widespread CSCCs on the legs Acitretin (25 mg/d) combined with 5-fluorouracil and chemowraps for 6 months Complete clinical resolution
Anforth et al. (5) Eight patients treated with a BRAF inhibitor Acitretin (10–15 mg/d) for 1–13 months Suppression of CSCC development
Lebwohl et al. (6) A 40-year-old male patient with psoriasis who developed CSCC following PUVA phototherapy Acitretin (25 mg/d) for 25 months Suppression of CSCC development
Patient 1 An 83-year-old Chinese woman with invasive SCC on the left cheek Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 8 courses Complete remission (6
months); 10 years without
recurrence
Patient 2 A 97-year-old Chinese woman with invasive SCC on the left cheek Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 6 courses Complete remission (6
months); 3 years without
recurrence
Patient 3 An 76-year-old Chinese woman with invasive SCC on the right ankle Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 6 plus 24 courses resumed* Complete remission (24 months resumed); 6 years without recurrence
* After 6 courses of the combination therapy, this patient self-discontinued the treatment. Two years later, tumor recurrence was detected and the patient resumed the combination treatment for another 24 courses.
Acitretin monotherapy is generally considered to have a low efficacy and is associated with dose-dependent side effects such as mucocutaneous dryness, hair loss, elevated triglycerides, transient elevations in liver function tests and (with long-term treatment) diffuse idiopathic skeletal hyperostosis. In view of the above issues, the patients in this case report were also administered clarithromycin, a macrolide antibiotic that has been reported to enhance the activity of conventional chemotherapy agents in cell lines and animal experiments (8, 9). Furthermore, long-term administration of clarithromycin was reported to prolong the survival time of patients with lung cancer (16). We speculate that the excellent responses to combination therapy observed in all three patients in the present report were due to the antiangiogenic and antitumor effects of clarithromycin, which synergized with the anti-tumor actions of acitretin.
Combination therapy with acitretin and clarithromycin was well-tolerated by all three patients in this study. We summarized the common characteristics of the three patients in our study which may somehow contributed to the successful treatment outcomes. We found that all of them were female older than 75 and refused large-scale surgical treatment. The pathological diagnoses were poorly to well-differentiated invasive SCC. Their tumors were all single and without any metastases. We believe that elderly patients with slow tumor growth respond well to drug treatment and this can help them to avoid the risks associated with surgery. For young patients with multiple or larger tumors, the drug treatment can also be used to reduce the size of the tumor, furthermore it will be beneficial to surgical resection. The only adverse effects reported were pseudotumor cerebri and mucocutaneous dryness, which are known side-effects of acitretin. Therefore, we suggest that the combination of acitretin and clarithromycin may represent an effective and safe therapy for CSCC.
As far as we are aware, this is the first study to report that combination therapy with acitretin and clarithromycin may be an effective and safe treatment for CSCC. Additional studies are required to further assess the efficacy and safety of combination therapy with acitretin and clarithromycin in the treatment of CSCC in larger populations and to investigate the underlying molecular mechanisms.
Data Availability Statement
The original contributions generated in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.
Ethics Statement
The studies involving human participants were reviewed and approved by The study was approved by the Medical Technology Clinical Application Ethics Committee of the First Affiliated Hospital of Fujian (No. [2015]084). The patients/participants provided their written informed consent to participate in this study. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.
Author Contributions
YZha: conception and design. CJ: administrative support. YZha, YZhu, and HW: provision of study materials or patients. CJ and YZhu: data analysis and interpretation. All authors: collection, assembly of data, manuscript writing, and final approval of manuscript.
Conflict of Interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The authors acknowledge the help of all patients who participated in our study. | Oral | DrugAdministrationRoute | CC BY | 33718249 | 19,083,671 | 2021 |
What was the dosage of drug 'ACITRETIN'? | Case Report: Successful Treatment of Cutaneous Squamous Cell Carcinoma in Three Patients With a Combination of Acitretin and Clarithromycin.
Few studies have evaluated acitretin as a primary treatment for cutaneous squamous cell carcinoma (CSCC). We report, for the first time, three cases of CSCC successfully treated with acitretin and clarithromycin. A literature review on this subject was also was performed. This case report included three patients with CSCC treated with acitretin and clarithromycin at the First Affiliated Hospital of Fujian Medical University (2008-2019). Patient 1 (83-year-old woman, ulcerated mass on the left cheek), patient 2 (97-year-old woman, painful mass on the left cheek) and patient 3 (76-year-old woman, large mass on the right ankle) received 8, 6, and 30 courses of combination therapy. All patients tolerated the adverse effects (pseudotumor cerebri and mucocutaneous dryness) and achieved complete regression within 6 months. Patients 1, 2, and 3 have not experienced recurrence during a 10-, 3-, and 6-year follow-up. Acitretin has limited efficacy as a monotherapy for CSCC. Our experience indicates that combination therapy with acitretin and clarithromycin may be an effective and well-tolerated treatment for unresectable CSCC.
Introduction
Given its increasing incidence and potential for poor outcomes (1–3), cutaneous squamous cell carcinoma (CSCC) is emerging as a public health problem. High-risk CSCC can be difficult to treat by surgical excision because the neoplasm can extend or infiltrate beyond the visible borders of the lesion. A wide range of non-surgical treatments are available for high-risk CSCC. Acitretin is a second generation retinoid shown to reduce the incidence of new primary non-melanoma skin cancers in immunocompromised transplant recipients (4) and patients receiving BRAF inhibitor treatment (5). Acitretin is considered to have limited efficacy as a monotherapy for CSCC but seems more effective when used in combination with other agents (6, 7). Clarithromycin is an antibacterial drug that has been reported to enhance the anti-tumor activity of chemotherapeutic drugs in vitro and in vivo (8, 9). However, no previous study has evaluated the combination of acitretin and clarithromycin in the treatment of CSCC. We present 3 cases illustrating that combination therapy with acitretin and clarithromycin might be a promising option for patients with CSCC who are unwilling or unable to undergo surgery.
Case Description
Ethical Statement
The study was approved by the Medical Technology Clinical Application Ethics Committee of the First Affiliated Hospital of Fujian (No. [2015]084). Before treatment, the patients were fully informed about the potential advantages and disadvantages of combination therapy with acitretin and clarithromycin, including the possible side effects of these drugs. All patients provided written informed consent for treatment and agreed to attend follow-up visits every 2 weeks. All patients provided written informed consent for data analysis and publication.
Patients
The present case report represents a retrospective analysis of the medical records of patients with CSCC who were treated with a combination of acitretin and clarithromycin at the Dermatology Department of the First Affiliated Hospital of Fujian between 2008 and 2018. Inclusion criteria consisted of a diagnosis of SCC by pathological biopsy and a history as well as tumor diameter ≥1 cm (10).
A total of 3 patients were involved in our study. Acitretin and clarithromycin were administered together to the patients from the time when skin biopsy confirmed the diagnosis of invasive SCC. We defined oral 30 mg/day of acitretin for 4 weeks and oral 500 mg/day of clarithromycin for 3 weeks as 1-month treatment course. There was no interval time between courses.
Once biochemical investigations, including measurements of serum creatinine, cholesterol, triglycerides, and potassium appeared abnormal, the treatment was stopped or the dosage was reduced.
Patient 1
Patient 1 was an 83-year-old Chinese woman with a 2-year history of an asymptomatic ulcerated mass on her left cheek. Physical examination in March 2008 revealed a 4 × 3.5 cm crater-like mass with central ulceration on her left cheek (Figure 1A). Skin biopsy revealed the presence of atypical epithelial tumor cell formations extending beyond the epidermis into the underlying dermis, cornified cells and horny pearls, thereby confirming the diagnosis of invasive SCC. The patient showed a favorable response to acitretin and clarithromycin within 2 months after the first oral administration. The patient declined long-term treatment due to the development of adverse effects (pseudotumor cerebri and dryness of the lips and skin). Therefore, CO2 laser resection of the CSCC was performed after a total of 6 courses of combination therapy, and the patient then received an additional two courses of combination therapy. This patient has lived without tumor recurrence for 10 years.
Figure 1 Photographs showing the regression of cutaneous squamous cell carcinoma (CSCC) in three patients treated with acitretin and clarithromycin. (A) Patient 1. The CSCC on the left cheek is shown at month 0 (before initiation of combination therapy), 5 and 12. (B) Patient 2. The CSCC on the left cheek is shown at month 0, 3, and 29. (C) Patient 3. The CSCC on the right ankle is shown at month 0, 6, 30, 42, and 60.
Patient 2
Patient 2 was a 97-year-old Chinese woman with a 1-year history of a painful mass on her left cheek. Physical examination in November 2015 demonstrated the presence of a 2 × 2 cm mass with an irregular edge and signs of erosion on her left cheek (Figure 1B). Skin biopsy confirmed the diagnosis of invasive SCC. The patient was treated with 6 courses of combination therapy. Darkening and regression of the tumor lesions were first detected in 2 months after the initiation of combination therapy, and a complete response was observed at 6 months. The only adverse effect reported by the patient was mucocutaneous dryness, but this resolved following cessation of treatment with acitretin. The patient has not experienced recurrence of CSCC during the 3-year follow-up.
Patient 3
Patient 3 was a 76-year-old Chinese woman with a 1-year history of a large mass on her right ankle that had progressively increased in size. Physical examination in September 2012 revealed a 4 × 4 cm mass with erosion on her right ankle (Figure 1C). Skin biopsy showed that the overlying epidermis was ulcerated with atypical keratinocytes invading the dermis. The keratinocytes were larger in the central parts of the tumor nests and formed eosinophilic horny pearls. The neoplastic keratinocytes had abundant eosinophilic and dyskeratotic cytoplasm and large vesicular nuclei with visible nucleoli and mitoses. An inflammatory reaction was evident, with many lymphocytes surrounding the stromal vessels and penetrating the tumor nests (Figure 2). Immunohistochemical analyses demonstrated that the tumor cells were positive for AE1/AE3 and P63, negative for Ber-EP4, and positive for Ki67 in about 40% of cells, confirming the diagnosis of invasive SCC (Figure 3). The patient was initially treated with 6 courses of combination therapy, which led to substantial regression of the tumor such that it resembled a small residual ulcer, and the patient then decided to discontinue the treatment. A local tumor relapse was detected 2 years later, so the patient resumed combination therapy and received an additional 24 courses of treatment, which resulted in complete remission. The only adverse effect experienced by the patient was mucocutaneous dryness. The patient remains alive and without tumor recurrence after a 6-year follow-up.
Figure 2 Representative images showing the histologic appearance of a tumor specimen from patient 3. (A) Section stained with hematoxylin-eosin (magnification, ×25). (B) Section stained with hematoxylin-eosin (magnification, ×200). (C) Section stained with hematoxylin-eosin (magnification, ×400).
Figure 3 Immunohistology of tumor specimens from patient 3. (A) Tumor cells were positive for AE1/AE3 (magnification, ×25). (B) Tumor cells were positive for P63 (magnification, ×25). (C) Tumor cells were negative for Ber-EP4 (magnification, ×25). (D) Around 40% of tumor cells were positive for Ki67 (magnification, ×400).
Discussion
A notable finding of this case report was that combination therapy achieved CSCC regression in all three patients within 2–6 months after the first administered dose of acitretin and clarithromycin. Furthermore, two of the patients exhibited complete remission of the tumor without scaring, thereby avoiding the need for surgery, while the other patient was successfully treated by tumor excision after combination therapy. Importantly, all three patients remained free of recurrence during a follow-up of 3–10 years. In addition, all three patients tolerated the treatment and experienced only minimal adverse effects (pseudotumor cerebri and mucocutaneous dryness) that resolved after discontinuation of acitretin. To our knowledge, this is the first study to report that combination therapy with acitretin and clarithromycin may be an effective treatment for CSCC.
The majority of CSCCs can be successfully eradicated by surgical excision with post-operative margin assessment or microscopically controlled surgery, although the cosmetic results are not always satisfactory. When patients cannot or refuse to undergo surgery, alternative treatment options include radiotherapy, intralesional chemotherapy and photodynamic therapy. Radiotherapy is a potentially curative treatment in cases of early-stage CSCC where resection would cause substantial cosmetic or functional deficits. However, radiotherapy alone has limited effectiveness as a treatment for advanced CSCC and is associated with adverse effects. Anti-PD-1 antibodies are considered a first-line systemic treatment for patients with metastatic or locally advanced CSCC who are not candidates for curative surgery or radiotherapy, although its use in elderly patients is limited by the toxicity of this chemotherapeutic agent (11, 12). Recently, epidermal growth factor receptor inhibitors combined with chemotherapy or radiotherapy have been used as second-line systemic treatments for advanced CSCC (11, 13), but further research is needed to fully characterize the efficacy and safety of these strategies (11, 14). As awareness grows regarding the increasing number of cases of post-surgical CSSC relapse, it is becoming recognized that improved non-surgical treatments for CSCC are urgently needed.
Retinoids are established drugs and have been shown to reduce the development of non-melanoma skin cancer in high risk individuals. Previous randomized controlled trials have demonstrated that systemic retinoids can prevent new primary CSCC or new precancerous lesions in immunocompromised patients after transplantation (15). Typically, retinoids are considered for patients who develop multiple tumors each year, have multiple CSCCs in high-risk areas, or have numerous actinic keratoses. Acitretin is a second-generation retinoid reported to exhibit anti-tumor activity against actinic keratosis, verrucous carcinomas and SCC. Previous studies of the use of acitretin in the management of CSCC are rare and summarized in Table 1 (5–7). Two of these three studies were case reports describing the use of acitretin in the chemoprevention of CSCC in 8 patients receiving BRAF inhibitors (5) and one patient who had been treated with phototherapy (6). Only one prior study has described the use of acitretin to treat CSCC: the combination of acitretin, intralesional 5-fluorouracil and 5-fluorouracil chemowraps was found to cause complete regression of widespread CSCCs on the legs of two patients (7). To our knowledge, the present case report is the first to report the successful use of acitretin in combination with clarithromycin as a treatment for CSCC.
Table 1 Summary of publications reporting cases and our cases of cutaneous squamous cell carcinoma (CSCC) treated with acitretin.
Reference Patients Treatment Response
Manalo et al. (7) Two patients (74-year-old female and 60-year-old female) with widespread CSCCs on the legs Acitretin (25 mg/d) combined with 5-fluorouracil and chemowraps for 6 months Complete clinical resolution
Anforth et al. (5) Eight patients treated with a BRAF inhibitor Acitretin (10–15 mg/d) for 1–13 months Suppression of CSCC development
Lebwohl et al. (6) A 40-year-old male patient with psoriasis who developed CSCC following PUVA phototherapy Acitretin (25 mg/d) for 25 months Suppression of CSCC development
Patient 1 An 83-year-old Chinese woman with invasive SCC on the left cheek Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 8 courses Complete remission (6
months); 10 years without
recurrence
Patient 2 A 97-year-old Chinese woman with invasive SCC on the left cheek Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 6 courses Complete remission (6
months); 3 years without
recurrence
Patient 3 An 76-year-old Chinese woman with invasive SCC on the right ankle Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 6 plus 24 courses resumed* Complete remission (24 months resumed); 6 years without recurrence
* After 6 courses of the combination therapy, this patient self-discontinued the treatment. Two years later, tumor recurrence was detected and the patient resumed the combination treatment for another 24 courses.
Acitretin monotherapy is generally considered to have a low efficacy and is associated with dose-dependent side effects such as mucocutaneous dryness, hair loss, elevated triglycerides, transient elevations in liver function tests and (with long-term treatment) diffuse idiopathic skeletal hyperostosis. In view of the above issues, the patients in this case report were also administered clarithromycin, a macrolide antibiotic that has been reported to enhance the activity of conventional chemotherapy agents in cell lines and animal experiments (8, 9). Furthermore, long-term administration of clarithromycin was reported to prolong the survival time of patients with lung cancer (16). We speculate that the excellent responses to combination therapy observed in all three patients in the present report were due to the antiangiogenic and antitumor effects of clarithromycin, which synergized with the anti-tumor actions of acitretin.
Combination therapy with acitretin and clarithromycin was well-tolerated by all three patients in this study. We summarized the common characteristics of the three patients in our study which may somehow contributed to the successful treatment outcomes. We found that all of them were female older than 75 and refused large-scale surgical treatment. The pathological diagnoses were poorly to well-differentiated invasive SCC. Their tumors were all single and without any metastases. We believe that elderly patients with slow tumor growth respond well to drug treatment and this can help them to avoid the risks associated with surgery. For young patients with multiple or larger tumors, the drug treatment can also be used to reduce the size of the tumor, furthermore it will be beneficial to surgical resection. The only adverse effects reported were pseudotumor cerebri and mucocutaneous dryness, which are known side-effects of acitretin. Therefore, we suggest that the combination of acitretin and clarithromycin may represent an effective and safe therapy for CSCC.
As far as we are aware, this is the first study to report that combination therapy with acitretin and clarithromycin may be an effective and safe treatment for CSCC. Additional studies are required to further assess the efficacy and safety of combination therapy with acitretin and clarithromycin in the treatment of CSCC in larger populations and to investigate the underlying molecular mechanisms.
Data Availability Statement
The original contributions generated in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.
Ethics Statement
The studies involving human participants were reviewed and approved by The study was approved by the Medical Technology Clinical Application Ethics Committee of the First Affiliated Hospital of Fujian (No. [2015]084). The patients/participants provided their written informed consent to participate in this study. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.
Author Contributions
YZha: conception and design. CJ: administrative support. YZha, YZhu, and HW: provision of study materials or patients. CJ and YZhu: data analysis and interpretation. All authors: collection, assembly of data, manuscript writing, and final approval of manuscript.
Conflict of Interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The authors acknowledge the help of all patients who participated in our study. | 30 MILLIGRAM DAILY; | DrugDosageText | CC BY | 33718249 | 19,083,671 | 2021 |
What was the dosage of drug 'CLARITHROMYCIN'? | Case Report: Successful Treatment of Cutaneous Squamous Cell Carcinoma in Three Patients With a Combination of Acitretin and Clarithromycin.
Few studies have evaluated acitretin as a primary treatment for cutaneous squamous cell carcinoma (CSCC). We report, for the first time, three cases of CSCC successfully treated with acitretin and clarithromycin. A literature review on this subject was also was performed. This case report included three patients with CSCC treated with acitretin and clarithromycin at the First Affiliated Hospital of Fujian Medical University (2008-2019). Patient 1 (83-year-old woman, ulcerated mass on the left cheek), patient 2 (97-year-old woman, painful mass on the left cheek) and patient 3 (76-year-old woman, large mass on the right ankle) received 8, 6, and 30 courses of combination therapy. All patients tolerated the adverse effects (pseudotumor cerebri and mucocutaneous dryness) and achieved complete regression within 6 months. Patients 1, 2, and 3 have not experienced recurrence during a 10-, 3-, and 6-year follow-up. Acitretin has limited efficacy as a monotherapy for CSCC. Our experience indicates that combination therapy with acitretin and clarithromycin may be an effective and well-tolerated treatment for unresectable CSCC.
Introduction
Given its increasing incidence and potential for poor outcomes (1–3), cutaneous squamous cell carcinoma (CSCC) is emerging as a public health problem. High-risk CSCC can be difficult to treat by surgical excision because the neoplasm can extend or infiltrate beyond the visible borders of the lesion. A wide range of non-surgical treatments are available for high-risk CSCC. Acitretin is a second generation retinoid shown to reduce the incidence of new primary non-melanoma skin cancers in immunocompromised transplant recipients (4) and patients receiving BRAF inhibitor treatment (5). Acitretin is considered to have limited efficacy as a monotherapy for CSCC but seems more effective when used in combination with other agents (6, 7). Clarithromycin is an antibacterial drug that has been reported to enhance the anti-tumor activity of chemotherapeutic drugs in vitro and in vivo (8, 9). However, no previous study has evaluated the combination of acitretin and clarithromycin in the treatment of CSCC. We present 3 cases illustrating that combination therapy with acitretin and clarithromycin might be a promising option for patients with CSCC who are unwilling or unable to undergo surgery.
Case Description
Ethical Statement
The study was approved by the Medical Technology Clinical Application Ethics Committee of the First Affiliated Hospital of Fujian (No. [2015]084). Before treatment, the patients were fully informed about the potential advantages and disadvantages of combination therapy with acitretin and clarithromycin, including the possible side effects of these drugs. All patients provided written informed consent for treatment and agreed to attend follow-up visits every 2 weeks. All patients provided written informed consent for data analysis and publication.
Patients
The present case report represents a retrospective analysis of the medical records of patients with CSCC who were treated with a combination of acitretin and clarithromycin at the Dermatology Department of the First Affiliated Hospital of Fujian between 2008 and 2018. Inclusion criteria consisted of a diagnosis of SCC by pathological biopsy and a history as well as tumor diameter ≥1 cm (10).
A total of 3 patients were involved in our study. Acitretin and clarithromycin were administered together to the patients from the time when skin biopsy confirmed the diagnosis of invasive SCC. We defined oral 30 mg/day of acitretin for 4 weeks and oral 500 mg/day of clarithromycin for 3 weeks as 1-month treatment course. There was no interval time between courses.
Once biochemical investigations, including measurements of serum creatinine, cholesterol, triglycerides, and potassium appeared abnormal, the treatment was stopped or the dosage was reduced.
Patient 1
Patient 1 was an 83-year-old Chinese woman with a 2-year history of an asymptomatic ulcerated mass on her left cheek. Physical examination in March 2008 revealed a 4 × 3.5 cm crater-like mass with central ulceration on her left cheek (Figure 1A). Skin biopsy revealed the presence of atypical epithelial tumor cell formations extending beyond the epidermis into the underlying dermis, cornified cells and horny pearls, thereby confirming the diagnosis of invasive SCC. The patient showed a favorable response to acitretin and clarithromycin within 2 months after the first oral administration. The patient declined long-term treatment due to the development of adverse effects (pseudotumor cerebri and dryness of the lips and skin). Therefore, CO2 laser resection of the CSCC was performed after a total of 6 courses of combination therapy, and the patient then received an additional two courses of combination therapy. This patient has lived without tumor recurrence for 10 years.
Figure 1 Photographs showing the regression of cutaneous squamous cell carcinoma (CSCC) in three patients treated with acitretin and clarithromycin. (A) Patient 1. The CSCC on the left cheek is shown at month 0 (before initiation of combination therapy), 5 and 12. (B) Patient 2. The CSCC on the left cheek is shown at month 0, 3, and 29. (C) Patient 3. The CSCC on the right ankle is shown at month 0, 6, 30, 42, and 60.
Patient 2
Patient 2 was a 97-year-old Chinese woman with a 1-year history of a painful mass on her left cheek. Physical examination in November 2015 demonstrated the presence of a 2 × 2 cm mass with an irregular edge and signs of erosion on her left cheek (Figure 1B). Skin biopsy confirmed the diagnosis of invasive SCC. The patient was treated with 6 courses of combination therapy. Darkening and regression of the tumor lesions were first detected in 2 months after the initiation of combination therapy, and a complete response was observed at 6 months. The only adverse effect reported by the patient was mucocutaneous dryness, but this resolved following cessation of treatment with acitretin. The patient has not experienced recurrence of CSCC during the 3-year follow-up.
Patient 3
Patient 3 was a 76-year-old Chinese woman with a 1-year history of a large mass on her right ankle that had progressively increased in size. Physical examination in September 2012 revealed a 4 × 4 cm mass with erosion on her right ankle (Figure 1C). Skin biopsy showed that the overlying epidermis was ulcerated with atypical keratinocytes invading the dermis. The keratinocytes were larger in the central parts of the tumor nests and formed eosinophilic horny pearls. The neoplastic keratinocytes had abundant eosinophilic and dyskeratotic cytoplasm and large vesicular nuclei with visible nucleoli and mitoses. An inflammatory reaction was evident, with many lymphocytes surrounding the stromal vessels and penetrating the tumor nests (Figure 2). Immunohistochemical analyses demonstrated that the tumor cells were positive for AE1/AE3 and P63, negative for Ber-EP4, and positive for Ki67 in about 40% of cells, confirming the diagnosis of invasive SCC (Figure 3). The patient was initially treated with 6 courses of combination therapy, which led to substantial regression of the tumor such that it resembled a small residual ulcer, and the patient then decided to discontinue the treatment. A local tumor relapse was detected 2 years later, so the patient resumed combination therapy and received an additional 24 courses of treatment, which resulted in complete remission. The only adverse effect experienced by the patient was mucocutaneous dryness. The patient remains alive and without tumor recurrence after a 6-year follow-up.
Figure 2 Representative images showing the histologic appearance of a tumor specimen from patient 3. (A) Section stained with hematoxylin-eosin (magnification, ×25). (B) Section stained with hematoxylin-eosin (magnification, ×200). (C) Section stained with hematoxylin-eosin (magnification, ×400).
Figure 3 Immunohistology of tumor specimens from patient 3. (A) Tumor cells were positive for AE1/AE3 (magnification, ×25). (B) Tumor cells were positive for P63 (magnification, ×25). (C) Tumor cells were negative for Ber-EP4 (magnification, ×25). (D) Around 40% of tumor cells were positive for Ki67 (magnification, ×400).
Discussion
A notable finding of this case report was that combination therapy achieved CSCC regression in all three patients within 2–6 months after the first administered dose of acitretin and clarithromycin. Furthermore, two of the patients exhibited complete remission of the tumor without scaring, thereby avoiding the need for surgery, while the other patient was successfully treated by tumor excision after combination therapy. Importantly, all three patients remained free of recurrence during a follow-up of 3–10 years. In addition, all three patients tolerated the treatment and experienced only minimal adverse effects (pseudotumor cerebri and mucocutaneous dryness) that resolved after discontinuation of acitretin. To our knowledge, this is the first study to report that combination therapy with acitretin and clarithromycin may be an effective treatment for CSCC.
The majority of CSCCs can be successfully eradicated by surgical excision with post-operative margin assessment or microscopically controlled surgery, although the cosmetic results are not always satisfactory. When patients cannot or refuse to undergo surgery, alternative treatment options include radiotherapy, intralesional chemotherapy and photodynamic therapy. Radiotherapy is a potentially curative treatment in cases of early-stage CSCC where resection would cause substantial cosmetic or functional deficits. However, radiotherapy alone has limited effectiveness as a treatment for advanced CSCC and is associated with adverse effects. Anti-PD-1 antibodies are considered a first-line systemic treatment for patients with metastatic or locally advanced CSCC who are not candidates for curative surgery or radiotherapy, although its use in elderly patients is limited by the toxicity of this chemotherapeutic agent (11, 12). Recently, epidermal growth factor receptor inhibitors combined with chemotherapy or radiotherapy have been used as second-line systemic treatments for advanced CSCC (11, 13), but further research is needed to fully characterize the efficacy and safety of these strategies (11, 14). As awareness grows regarding the increasing number of cases of post-surgical CSSC relapse, it is becoming recognized that improved non-surgical treatments for CSCC are urgently needed.
Retinoids are established drugs and have been shown to reduce the development of non-melanoma skin cancer in high risk individuals. Previous randomized controlled trials have demonstrated that systemic retinoids can prevent new primary CSCC or new precancerous lesions in immunocompromised patients after transplantation (15). Typically, retinoids are considered for patients who develop multiple tumors each year, have multiple CSCCs in high-risk areas, or have numerous actinic keratoses. Acitretin is a second-generation retinoid reported to exhibit anti-tumor activity against actinic keratosis, verrucous carcinomas and SCC. Previous studies of the use of acitretin in the management of CSCC are rare and summarized in Table 1 (5–7). Two of these three studies were case reports describing the use of acitretin in the chemoprevention of CSCC in 8 patients receiving BRAF inhibitors (5) and one patient who had been treated with phototherapy (6). Only one prior study has described the use of acitretin to treat CSCC: the combination of acitretin, intralesional 5-fluorouracil and 5-fluorouracil chemowraps was found to cause complete regression of widespread CSCCs on the legs of two patients (7). To our knowledge, the present case report is the first to report the successful use of acitretin in combination with clarithromycin as a treatment for CSCC.
Table 1 Summary of publications reporting cases and our cases of cutaneous squamous cell carcinoma (CSCC) treated with acitretin.
Reference Patients Treatment Response
Manalo et al. (7) Two patients (74-year-old female and 60-year-old female) with widespread CSCCs on the legs Acitretin (25 mg/d) combined with 5-fluorouracil and chemowraps for 6 months Complete clinical resolution
Anforth et al. (5) Eight patients treated with a BRAF inhibitor Acitretin (10–15 mg/d) for 1–13 months Suppression of CSCC development
Lebwohl et al. (6) A 40-year-old male patient with psoriasis who developed CSCC following PUVA phototherapy Acitretin (25 mg/d) for 25 months Suppression of CSCC development
Patient 1 An 83-year-old Chinese woman with invasive SCC on the left cheek Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 8 courses Complete remission (6
months); 10 years without
recurrence
Patient 2 A 97-year-old Chinese woman with invasive SCC on the left cheek Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 6 courses Complete remission (6
months); 3 years without
recurrence
Patient 3 An 76-year-old Chinese woman with invasive SCC on the right ankle Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 6 plus 24 courses resumed* Complete remission (24 months resumed); 6 years without recurrence
* After 6 courses of the combination therapy, this patient self-discontinued the treatment. Two years later, tumor recurrence was detected and the patient resumed the combination treatment for another 24 courses.
Acitretin monotherapy is generally considered to have a low efficacy and is associated with dose-dependent side effects such as mucocutaneous dryness, hair loss, elevated triglycerides, transient elevations in liver function tests and (with long-term treatment) diffuse idiopathic skeletal hyperostosis. In view of the above issues, the patients in this case report were also administered clarithromycin, a macrolide antibiotic that has been reported to enhance the activity of conventional chemotherapy agents in cell lines and animal experiments (8, 9). Furthermore, long-term administration of clarithromycin was reported to prolong the survival time of patients with lung cancer (16). We speculate that the excellent responses to combination therapy observed in all three patients in the present report were due to the antiangiogenic and antitumor effects of clarithromycin, which synergized with the anti-tumor actions of acitretin.
Combination therapy with acitretin and clarithromycin was well-tolerated by all three patients in this study. We summarized the common characteristics of the three patients in our study which may somehow contributed to the successful treatment outcomes. We found that all of them were female older than 75 and refused large-scale surgical treatment. The pathological diagnoses were poorly to well-differentiated invasive SCC. Their tumors were all single and without any metastases. We believe that elderly patients with slow tumor growth respond well to drug treatment and this can help them to avoid the risks associated with surgery. For young patients with multiple or larger tumors, the drug treatment can also be used to reduce the size of the tumor, furthermore it will be beneficial to surgical resection. The only adverse effects reported were pseudotumor cerebri and mucocutaneous dryness, which are known side-effects of acitretin. Therefore, we suggest that the combination of acitretin and clarithromycin may represent an effective and safe therapy for CSCC.
As far as we are aware, this is the first study to report that combination therapy with acitretin and clarithromycin may be an effective and safe treatment for CSCC. Additional studies are required to further assess the efficacy and safety of combination therapy with acitretin and clarithromycin in the treatment of CSCC in larger populations and to investigate the underlying molecular mechanisms.
Data Availability Statement
The original contributions generated in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.
Ethics Statement
The studies involving human participants were reviewed and approved by The study was approved by the Medical Technology Clinical Application Ethics Committee of the First Affiliated Hospital of Fujian (No. [2015]084). The patients/participants provided their written informed consent to participate in this study. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.
Author Contributions
YZha: conception and design. CJ: administrative support. YZha, YZhu, and HW: provision of study materials or patients. CJ and YZhu: data analysis and interpretation. All authors: collection, assembly of data, manuscript writing, and final approval of manuscript.
Conflict of Interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The authors acknowledge the help of all patients who participated in our study. | 500 MILLIGRAM DAILY; | DrugDosageText | CC BY | 33718249 | 19,083,671 | 2021 |
What was the outcome of reaction 'Dry skin'? | Case Report: Successful Treatment of Cutaneous Squamous Cell Carcinoma in Three Patients With a Combination of Acitretin and Clarithromycin.
Few studies have evaluated acitretin as a primary treatment for cutaneous squamous cell carcinoma (CSCC). We report, for the first time, three cases of CSCC successfully treated with acitretin and clarithromycin. A literature review on this subject was also was performed. This case report included three patients with CSCC treated with acitretin and clarithromycin at the First Affiliated Hospital of Fujian Medical University (2008-2019). Patient 1 (83-year-old woman, ulcerated mass on the left cheek), patient 2 (97-year-old woman, painful mass on the left cheek) and patient 3 (76-year-old woman, large mass on the right ankle) received 8, 6, and 30 courses of combination therapy. All patients tolerated the adverse effects (pseudotumor cerebri and mucocutaneous dryness) and achieved complete regression within 6 months. Patients 1, 2, and 3 have not experienced recurrence during a 10-, 3-, and 6-year follow-up. Acitretin has limited efficacy as a monotherapy for CSCC. Our experience indicates that combination therapy with acitretin and clarithromycin may be an effective and well-tolerated treatment for unresectable CSCC.
Introduction
Given its increasing incidence and potential for poor outcomes (1–3), cutaneous squamous cell carcinoma (CSCC) is emerging as a public health problem. High-risk CSCC can be difficult to treat by surgical excision because the neoplasm can extend or infiltrate beyond the visible borders of the lesion. A wide range of non-surgical treatments are available for high-risk CSCC. Acitretin is a second generation retinoid shown to reduce the incidence of new primary non-melanoma skin cancers in immunocompromised transplant recipients (4) and patients receiving BRAF inhibitor treatment (5). Acitretin is considered to have limited efficacy as a monotherapy for CSCC but seems more effective when used in combination with other agents (6, 7). Clarithromycin is an antibacterial drug that has been reported to enhance the anti-tumor activity of chemotherapeutic drugs in vitro and in vivo (8, 9). However, no previous study has evaluated the combination of acitretin and clarithromycin in the treatment of CSCC. We present 3 cases illustrating that combination therapy with acitretin and clarithromycin might be a promising option for patients with CSCC who are unwilling or unable to undergo surgery.
Case Description
Ethical Statement
The study was approved by the Medical Technology Clinical Application Ethics Committee of the First Affiliated Hospital of Fujian (No. [2015]084). Before treatment, the patients were fully informed about the potential advantages and disadvantages of combination therapy with acitretin and clarithromycin, including the possible side effects of these drugs. All patients provided written informed consent for treatment and agreed to attend follow-up visits every 2 weeks. All patients provided written informed consent for data analysis and publication.
Patients
The present case report represents a retrospective analysis of the medical records of patients with CSCC who were treated with a combination of acitretin and clarithromycin at the Dermatology Department of the First Affiliated Hospital of Fujian between 2008 and 2018. Inclusion criteria consisted of a diagnosis of SCC by pathological biopsy and a history as well as tumor diameter ≥1 cm (10).
A total of 3 patients were involved in our study. Acitretin and clarithromycin were administered together to the patients from the time when skin biopsy confirmed the diagnosis of invasive SCC. We defined oral 30 mg/day of acitretin for 4 weeks and oral 500 mg/day of clarithromycin for 3 weeks as 1-month treatment course. There was no interval time between courses.
Once biochemical investigations, including measurements of serum creatinine, cholesterol, triglycerides, and potassium appeared abnormal, the treatment was stopped or the dosage was reduced.
Patient 1
Patient 1 was an 83-year-old Chinese woman with a 2-year history of an asymptomatic ulcerated mass on her left cheek. Physical examination in March 2008 revealed a 4 × 3.5 cm crater-like mass with central ulceration on her left cheek (Figure 1A). Skin biopsy revealed the presence of atypical epithelial tumor cell formations extending beyond the epidermis into the underlying dermis, cornified cells and horny pearls, thereby confirming the diagnosis of invasive SCC. The patient showed a favorable response to acitretin and clarithromycin within 2 months after the first oral administration. The patient declined long-term treatment due to the development of adverse effects (pseudotumor cerebri and dryness of the lips and skin). Therefore, CO2 laser resection of the CSCC was performed after a total of 6 courses of combination therapy, and the patient then received an additional two courses of combination therapy. This patient has lived without tumor recurrence for 10 years.
Figure 1 Photographs showing the regression of cutaneous squamous cell carcinoma (CSCC) in three patients treated with acitretin and clarithromycin. (A) Patient 1. The CSCC on the left cheek is shown at month 0 (before initiation of combination therapy), 5 and 12. (B) Patient 2. The CSCC on the left cheek is shown at month 0, 3, and 29. (C) Patient 3. The CSCC on the right ankle is shown at month 0, 6, 30, 42, and 60.
Patient 2
Patient 2 was a 97-year-old Chinese woman with a 1-year history of a painful mass on her left cheek. Physical examination in November 2015 demonstrated the presence of a 2 × 2 cm mass with an irregular edge and signs of erosion on her left cheek (Figure 1B). Skin biopsy confirmed the diagnosis of invasive SCC. The patient was treated with 6 courses of combination therapy. Darkening and regression of the tumor lesions were first detected in 2 months after the initiation of combination therapy, and a complete response was observed at 6 months. The only adverse effect reported by the patient was mucocutaneous dryness, but this resolved following cessation of treatment with acitretin. The patient has not experienced recurrence of CSCC during the 3-year follow-up.
Patient 3
Patient 3 was a 76-year-old Chinese woman with a 1-year history of a large mass on her right ankle that had progressively increased in size. Physical examination in September 2012 revealed a 4 × 4 cm mass with erosion on her right ankle (Figure 1C). Skin biopsy showed that the overlying epidermis was ulcerated with atypical keratinocytes invading the dermis. The keratinocytes were larger in the central parts of the tumor nests and formed eosinophilic horny pearls. The neoplastic keratinocytes had abundant eosinophilic and dyskeratotic cytoplasm and large vesicular nuclei with visible nucleoli and mitoses. An inflammatory reaction was evident, with many lymphocytes surrounding the stromal vessels and penetrating the tumor nests (Figure 2). Immunohistochemical analyses demonstrated that the tumor cells were positive for AE1/AE3 and P63, negative for Ber-EP4, and positive for Ki67 in about 40% of cells, confirming the diagnosis of invasive SCC (Figure 3). The patient was initially treated with 6 courses of combination therapy, which led to substantial regression of the tumor such that it resembled a small residual ulcer, and the patient then decided to discontinue the treatment. A local tumor relapse was detected 2 years later, so the patient resumed combination therapy and received an additional 24 courses of treatment, which resulted in complete remission. The only adverse effect experienced by the patient was mucocutaneous dryness. The patient remains alive and without tumor recurrence after a 6-year follow-up.
Figure 2 Representative images showing the histologic appearance of a tumor specimen from patient 3. (A) Section stained with hematoxylin-eosin (magnification, ×25). (B) Section stained with hematoxylin-eosin (magnification, ×200). (C) Section stained with hematoxylin-eosin (magnification, ×400).
Figure 3 Immunohistology of tumor specimens from patient 3. (A) Tumor cells were positive for AE1/AE3 (magnification, ×25). (B) Tumor cells were positive for P63 (magnification, ×25). (C) Tumor cells were negative for Ber-EP4 (magnification, ×25). (D) Around 40% of tumor cells were positive for Ki67 (magnification, ×400).
Discussion
A notable finding of this case report was that combination therapy achieved CSCC regression in all three patients within 2–6 months after the first administered dose of acitretin and clarithromycin. Furthermore, two of the patients exhibited complete remission of the tumor without scaring, thereby avoiding the need for surgery, while the other patient was successfully treated by tumor excision after combination therapy. Importantly, all three patients remained free of recurrence during a follow-up of 3–10 years. In addition, all three patients tolerated the treatment and experienced only minimal adverse effects (pseudotumor cerebri and mucocutaneous dryness) that resolved after discontinuation of acitretin. To our knowledge, this is the first study to report that combination therapy with acitretin and clarithromycin may be an effective treatment for CSCC.
The majority of CSCCs can be successfully eradicated by surgical excision with post-operative margin assessment or microscopically controlled surgery, although the cosmetic results are not always satisfactory. When patients cannot or refuse to undergo surgery, alternative treatment options include radiotherapy, intralesional chemotherapy and photodynamic therapy. Radiotherapy is a potentially curative treatment in cases of early-stage CSCC where resection would cause substantial cosmetic or functional deficits. However, radiotherapy alone has limited effectiveness as a treatment for advanced CSCC and is associated with adverse effects. Anti-PD-1 antibodies are considered a first-line systemic treatment for patients with metastatic or locally advanced CSCC who are not candidates for curative surgery or radiotherapy, although its use in elderly patients is limited by the toxicity of this chemotherapeutic agent (11, 12). Recently, epidermal growth factor receptor inhibitors combined with chemotherapy or radiotherapy have been used as second-line systemic treatments for advanced CSCC (11, 13), but further research is needed to fully characterize the efficacy and safety of these strategies (11, 14). As awareness grows regarding the increasing number of cases of post-surgical CSSC relapse, it is becoming recognized that improved non-surgical treatments for CSCC are urgently needed.
Retinoids are established drugs and have been shown to reduce the development of non-melanoma skin cancer in high risk individuals. Previous randomized controlled trials have demonstrated that systemic retinoids can prevent new primary CSCC or new precancerous lesions in immunocompromised patients after transplantation (15). Typically, retinoids are considered for patients who develop multiple tumors each year, have multiple CSCCs in high-risk areas, or have numerous actinic keratoses. Acitretin is a second-generation retinoid reported to exhibit anti-tumor activity against actinic keratosis, verrucous carcinomas and SCC. Previous studies of the use of acitretin in the management of CSCC are rare and summarized in Table 1 (5–7). Two of these three studies were case reports describing the use of acitretin in the chemoprevention of CSCC in 8 patients receiving BRAF inhibitors (5) and one patient who had been treated with phototherapy (6). Only one prior study has described the use of acitretin to treat CSCC: the combination of acitretin, intralesional 5-fluorouracil and 5-fluorouracil chemowraps was found to cause complete regression of widespread CSCCs on the legs of two patients (7). To our knowledge, the present case report is the first to report the successful use of acitretin in combination with clarithromycin as a treatment for CSCC.
Table 1 Summary of publications reporting cases and our cases of cutaneous squamous cell carcinoma (CSCC) treated with acitretin.
Reference Patients Treatment Response
Manalo et al. (7) Two patients (74-year-old female and 60-year-old female) with widespread CSCCs on the legs Acitretin (25 mg/d) combined with 5-fluorouracil and chemowraps for 6 months Complete clinical resolution
Anforth et al. (5) Eight patients treated with a BRAF inhibitor Acitretin (10–15 mg/d) for 1–13 months Suppression of CSCC development
Lebwohl et al. (6) A 40-year-old male patient with psoriasis who developed CSCC following PUVA phototherapy Acitretin (25 mg/d) for 25 months Suppression of CSCC development
Patient 1 An 83-year-old Chinese woman with invasive SCC on the left cheek Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 8 courses Complete remission (6
months); 10 years without
recurrence
Patient 2 A 97-year-old Chinese woman with invasive SCC on the left cheek Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 6 courses Complete remission (6
months); 3 years without
recurrence
Patient 3 An 76-year-old Chinese woman with invasive SCC on the right ankle Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 6 plus 24 courses resumed* Complete remission (24 months resumed); 6 years without recurrence
* After 6 courses of the combination therapy, this patient self-discontinued the treatment. Two years later, tumor recurrence was detected and the patient resumed the combination treatment for another 24 courses.
Acitretin monotherapy is generally considered to have a low efficacy and is associated with dose-dependent side effects such as mucocutaneous dryness, hair loss, elevated triglycerides, transient elevations in liver function tests and (with long-term treatment) diffuse idiopathic skeletal hyperostosis. In view of the above issues, the patients in this case report were also administered clarithromycin, a macrolide antibiotic that has been reported to enhance the activity of conventional chemotherapy agents in cell lines and animal experiments (8, 9). Furthermore, long-term administration of clarithromycin was reported to prolong the survival time of patients with lung cancer (16). We speculate that the excellent responses to combination therapy observed in all three patients in the present report were due to the antiangiogenic and antitumor effects of clarithromycin, which synergized with the anti-tumor actions of acitretin.
Combination therapy with acitretin and clarithromycin was well-tolerated by all three patients in this study. We summarized the common characteristics of the three patients in our study which may somehow contributed to the successful treatment outcomes. We found that all of them were female older than 75 and refused large-scale surgical treatment. The pathological diagnoses were poorly to well-differentiated invasive SCC. Their tumors were all single and without any metastases. We believe that elderly patients with slow tumor growth respond well to drug treatment and this can help them to avoid the risks associated with surgery. For young patients with multiple or larger tumors, the drug treatment can also be used to reduce the size of the tumor, furthermore it will be beneficial to surgical resection. The only adverse effects reported were pseudotumor cerebri and mucocutaneous dryness, which are known side-effects of acitretin. Therefore, we suggest that the combination of acitretin and clarithromycin may represent an effective and safe therapy for CSCC.
As far as we are aware, this is the first study to report that combination therapy with acitretin and clarithromycin may be an effective and safe treatment for CSCC. Additional studies are required to further assess the efficacy and safety of combination therapy with acitretin and clarithromycin in the treatment of CSCC in larger populations and to investigate the underlying molecular mechanisms.
Data Availability Statement
The original contributions generated in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.
Ethics Statement
The studies involving human participants were reviewed and approved by The study was approved by the Medical Technology Clinical Application Ethics Committee of the First Affiliated Hospital of Fujian (No. [2015]084). The patients/participants provided their written informed consent to participate in this study. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.
Author Contributions
YZha: conception and design. CJ: administrative support. YZha, YZhu, and HW: provision of study materials or patients. CJ and YZhu: data analysis and interpretation. All authors: collection, assembly of data, manuscript writing, and final approval of manuscript.
Conflict of Interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The authors acknowledge the help of all patients who participated in our study. | Recovered | ReactionOutcome | CC BY | 33718249 | 19,083,671 | 2021 |
What was the outcome of reaction 'Idiopathic intracranial hypertension'? | Case Report: Successful Treatment of Cutaneous Squamous Cell Carcinoma in Three Patients With a Combination of Acitretin and Clarithromycin.
Few studies have evaluated acitretin as a primary treatment for cutaneous squamous cell carcinoma (CSCC). We report, for the first time, three cases of CSCC successfully treated with acitretin and clarithromycin. A literature review on this subject was also was performed. This case report included three patients with CSCC treated with acitretin and clarithromycin at the First Affiliated Hospital of Fujian Medical University (2008-2019). Patient 1 (83-year-old woman, ulcerated mass on the left cheek), patient 2 (97-year-old woman, painful mass on the left cheek) and patient 3 (76-year-old woman, large mass on the right ankle) received 8, 6, and 30 courses of combination therapy. All patients tolerated the adverse effects (pseudotumor cerebri and mucocutaneous dryness) and achieved complete regression within 6 months. Patients 1, 2, and 3 have not experienced recurrence during a 10-, 3-, and 6-year follow-up. Acitretin has limited efficacy as a monotherapy for CSCC. Our experience indicates that combination therapy with acitretin and clarithromycin may be an effective and well-tolerated treatment for unresectable CSCC.
Introduction
Given its increasing incidence and potential for poor outcomes (1–3), cutaneous squamous cell carcinoma (CSCC) is emerging as a public health problem. High-risk CSCC can be difficult to treat by surgical excision because the neoplasm can extend or infiltrate beyond the visible borders of the lesion. A wide range of non-surgical treatments are available for high-risk CSCC. Acitretin is a second generation retinoid shown to reduce the incidence of new primary non-melanoma skin cancers in immunocompromised transplant recipients (4) and patients receiving BRAF inhibitor treatment (5). Acitretin is considered to have limited efficacy as a monotherapy for CSCC but seems more effective when used in combination with other agents (6, 7). Clarithromycin is an antibacterial drug that has been reported to enhance the anti-tumor activity of chemotherapeutic drugs in vitro and in vivo (8, 9). However, no previous study has evaluated the combination of acitretin and clarithromycin in the treatment of CSCC. We present 3 cases illustrating that combination therapy with acitretin and clarithromycin might be a promising option for patients with CSCC who are unwilling or unable to undergo surgery.
Case Description
Ethical Statement
The study was approved by the Medical Technology Clinical Application Ethics Committee of the First Affiliated Hospital of Fujian (No. [2015]084). Before treatment, the patients were fully informed about the potential advantages and disadvantages of combination therapy with acitretin and clarithromycin, including the possible side effects of these drugs. All patients provided written informed consent for treatment and agreed to attend follow-up visits every 2 weeks. All patients provided written informed consent for data analysis and publication.
Patients
The present case report represents a retrospective analysis of the medical records of patients with CSCC who were treated with a combination of acitretin and clarithromycin at the Dermatology Department of the First Affiliated Hospital of Fujian between 2008 and 2018. Inclusion criteria consisted of a diagnosis of SCC by pathological biopsy and a history as well as tumor diameter ≥1 cm (10).
A total of 3 patients were involved in our study. Acitretin and clarithromycin were administered together to the patients from the time when skin biopsy confirmed the diagnosis of invasive SCC. We defined oral 30 mg/day of acitretin for 4 weeks and oral 500 mg/day of clarithromycin for 3 weeks as 1-month treatment course. There was no interval time between courses.
Once biochemical investigations, including measurements of serum creatinine, cholesterol, triglycerides, and potassium appeared abnormal, the treatment was stopped or the dosage was reduced.
Patient 1
Patient 1 was an 83-year-old Chinese woman with a 2-year history of an asymptomatic ulcerated mass on her left cheek. Physical examination in March 2008 revealed a 4 × 3.5 cm crater-like mass with central ulceration on her left cheek (Figure 1A). Skin biopsy revealed the presence of atypical epithelial tumor cell formations extending beyond the epidermis into the underlying dermis, cornified cells and horny pearls, thereby confirming the diagnosis of invasive SCC. The patient showed a favorable response to acitretin and clarithromycin within 2 months after the first oral administration. The patient declined long-term treatment due to the development of adverse effects (pseudotumor cerebri and dryness of the lips and skin). Therefore, CO2 laser resection of the CSCC was performed after a total of 6 courses of combination therapy, and the patient then received an additional two courses of combination therapy. This patient has lived without tumor recurrence for 10 years.
Figure 1 Photographs showing the regression of cutaneous squamous cell carcinoma (CSCC) in three patients treated with acitretin and clarithromycin. (A) Patient 1. The CSCC on the left cheek is shown at month 0 (before initiation of combination therapy), 5 and 12. (B) Patient 2. The CSCC on the left cheek is shown at month 0, 3, and 29. (C) Patient 3. The CSCC on the right ankle is shown at month 0, 6, 30, 42, and 60.
Patient 2
Patient 2 was a 97-year-old Chinese woman with a 1-year history of a painful mass on her left cheek. Physical examination in November 2015 demonstrated the presence of a 2 × 2 cm mass with an irregular edge and signs of erosion on her left cheek (Figure 1B). Skin biopsy confirmed the diagnosis of invasive SCC. The patient was treated with 6 courses of combination therapy. Darkening and regression of the tumor lesions were first detected in 2 months after the initiation of combination therapy, and a complete response was observed at 6 months. The only adverse effect reported by the patient was mucocutaneous dryness, but this resolved following cessation of treatment with acitretin. The patient has not experienced recurrence of CSCC during the 3-year follow-up.
Patient 3
Patient 3 was a 76-year-old Chinese woman with a 1-year history of a large mass on her right ankle that had progressively increased in size. Physical examination in September 2012 revealed a 4 × 4 cm mass with erosion on her right ankle (Figure 1C). Skin biopsy showed that the overlying epidermis was ulcerated with atypical keratinocytes invading the dermis. The keratinocytes were larger in the central parts of the tumor nests and formed eosinophilic horny pearls. The neoplastic keratinocytes had abundant eosinophilic and dyskeratotic cytoplasm and large vesicular nuclei with visible nucleoli and mitoses. An inflammatory reaction was evident, with many lymphocytes surrounding the stromal vessels and penetrating the tumor nests (Figure 2). Immunohistochemical analyses demonstrated that the tumor cells were positive for AE1/AE3 and P63, negative for Ber-EP4, and positive for Ki67 in about 40% of cells, confirming the diagnosis of invasive SCC (Figure 3). The patient was initially treated with 6 courses of combination therapy, which led to substantial regression of the tumor such that it resembled a small residual ulcer, and the patient then decided to discontinue the treatment. A local tumor relapse was detected 2 years later, so the patient resumed combination therapy and received an additional 24 courses of treatment, which resulted in complete remission. The only adverse effect experienced by the patient was mucocutaneous dryness. The patient remains alive and without tumor recurrence after a 6-year follow-up.
Figure 2 Representative images showing the histologic appearance of a tumor specimen from patient 3. (A) Section stained with hematoxylin-eosin (magnification, ×25). (B) Section stained with hematoxylin-eosin (magnification, ×200). (C) Section stained with hematoxylin-eosin (magnification, ×400).
Figure 3 Immunohistology of tumor specimens from patient 3. (A) Tumor cells were positive for AE1/AE3 (magnification, ×25). (B) Tumor cells were positive for P63 (magnification, ×25). (C) Tumor cells were negative for Ber-EP4 (magnification, ×25). (D) Around 40% of tumor cells were positive for Ki67 (magnification, ×400).
Discussion
A notable finding of this case report was that combination therapy achieved CSCC regression in all three patients within 2–6 months after the first administered dose of acitretin and clarithromycin. Furthermore, two of the patients exhibited complete remission of the tumor without scaring, thereby avoiding the need for surgery, while the other patient was successfully treated by tumor excision after combination therapy. Importantly, all three patients remained free of recurrence during a follow-up of 3–10 years. In addition, all three patients tolerated the treatment and experienced only minimal adverse effects (pseudotumor cerebri and mucocutaneous dryness) that resolved after discontinuation of acitretin. To our knowledge, this is the first study to report that combination therapy with acitretin and clarithromycin may be an effective treatment for CSCC.
The majority of CSCCs can be successfully eradicated by surgical excision with post-operative margin assessment or microscopically controlled surgery, although the cosmetic results are not always satisfactory. When patients cannot or refuse to undergo surgery, alternative treatment options include radiotherapy, intralesional chemotherapy and photodynamic therapy. Radiotherapy is a potentially curative treatment in cases of early-stage CSCC where resection would cause substantial cosmetic or functional deficits. However, radiotherapy alone has limited effectiveness as a treatment for advanced CSCC and is associated with adverse effects. Anti-PD-1 antibodies are considered a first-line systemic treatment for patients with metastatic or locally advanced CSCC who are not candidates for curative surgery or radiotherapy, although its use in elderly patients is limited by the toxicity of this chemotherapeutic agent (11, 12). Recently, epidermal growth factor receptor inhibitors combined with chemotherapy or radiotherapy have been used as second-line systemic treatments for advanced CSCC (11, 13), but further research is needed to fully characterize the efficacy and safety of these strategies (11, 14). As awareness grows regarding the increasing number of cases of post-surgical CSSC relapse, it is becoming recognized that improved non-surgical treatments for CSCC are urgently needed.
Retinoids are established drugs and have been shown to reduce the development of non-melanoma skin cancer in high risk individuals. Previous randomized controlled trials have demonstrated that systemic retinoids can prevent new primary CSCC or new precancerous lesions in immunocompromised patients after transplantation (15). Typically, retinoids are considered for patients who develop multiple tumors each year, have multiple CSCCs in high-risk areas, or have numerous actinic keratoses. Acitretin is a second-generation retinoid reported to exhibit anti-tumor activity against actinic keratosis, verrucous carcinomas and SCC. Previous studies of the use of acitretin in the management of CSCC are rare and summarized in Table 1 (5–7). Two of these three studies were case reports describing the use of acitretin in the chemoprevention of CSCC in 8 patients receiving BRAF inhibitors (5) and one patient who had been treated with phototherapy (6). Only one prior study has described the use of acitretin to treat CSCC: the combination of acitretin, intralesional 5-fluorouracil and 5-fluorouracil chemowraps was found to cause complete regression of widespread CSCCs on the legs of two patients (7). To our knowledge, the present case report is the first to report the successful use of acitretin in combination with clarithromycin as a treatment for CSCC.
Table 1 Summary of publications reporting cases and our cases of cutaneous squamous cell carcinoma (CSCC) treated with acitretin.
Reference Patients Treatment Response
Manalo et al. (7) Two patients (74-year-old female and 60-year-old female) with widespread CSCCs on the legs Acitretin (25 mg/d) combined with 5-fluorouracil and chemowraps for 6 months Complete clinical resolution
Anforth et al. (5) Eight patients treated with a BRAF inhibitor Acitretin (10–15 mg/d) for 1–13 months Suppression of CSCC development
Lebwohl et al. (6) A 40-year-old male patient with psoriasis who developed CSCC following PUVA phototherapy Acitretin (25 mg/d) for 25 months Suppression of CSCC development
Patient 1 An 83-year-old Chinese woman with invasive SCC on the left cheek Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 8 courses Complete remission (6
months); 10 years without
recurrence
Patient 2 A 97-year-old Chinese woman with invasive SCC on the left cheek Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 6 courses Complete remission (6
months); 3 years without
recurrence
Patient 3 An 76-year-old Chinese woman with invasive SCC on the right ankle Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 6 plus 24 courses resumed* Complete remission (24 months resumed); 6 years without recurrence
* After 6 courses of the combination therapy, this patient self-discontinued the treatment. Two years later, tumor recurrence was detected and the patient resumed the combination treatment for another 24 courses.
Acitretin monotherapy is generally considered to have a low efficacy and is associated with dose-dependent side effects such as mucocutaneous dryness, hair loss, elevated triglycerides, transient elevations in liver function tests and (with long-term treatment) diffuse idiopathic skeletal hyperostosis. In view of the above issues, the patients in this case report were also administered clarithromycin, a macrolide antibiotic that has been reported to enhance the activity of conventional chemotherapy agents in cell lines and animal experiments (8, 9). Furthermore, long-term administration of clarithromycin was reported to prolong the survival time of patients with lung cancer (16). We speculate that the excellent responses to combination therapy observed in all three patients in the present report were due to the antiangiogenic and antitumor effects of clarithromycin, which synergized with the anti-tumor actions of acitretin.
Combination therapy with acitretin and clarithromycin was well-tolerated by all three patients in this study. We summarized the common characteristics of the three patients in our study which may somehow contributed to the successful treatment outcomes. We found that all of them were female older than 75 and refused large-scale surgical treatment. The pathological diagnoses were poorly to well-differentiated invasive SCC. Their tumors were all single and without any metastases. We believe that elderly patients with slow tumor growth respond well to drug treatment and this can help them to avoid the risks associated with surgery. For young patients with multiple or larger tumors, the drug treatment can also be used to reduce the size of the tumor, furthermore it will be beneficial to surgical resection. The only adverse effects reported were pseudotumor cerebri and mucocutaneous dryness, which are known side-effects of acitretin. Therefore, we suggest that the combination of acitretin and clarithromycin may represent an effective and safe therapy for CSCC.
As far as we are aware, this is the first study to report that combination therapy with acitretin and clarithromycin may be an effective and safe treatment for CSCC. Additional studies are required to further assess the efficacy and safety of combination therapy with acitretin and clarithromycin in the treatment of CSCC in larger populations and to investigate the underlying molecular mechanisms.
Data Availability Statement
The original contributions generated in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.
Ethics Statement
The studies involving human participants were reviewed and approved by The study was approved by the Medical Technology Clinical Application Ethics Committee of the First Affiliated Hospital of Fujian (No. [2015]084). The patients/participants provided their written informed consent to participate in this study. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.
Author Contributions
YZha: conception and design. CJ: administrative support. YZha, YZhu, and HW: provision of study materials or patients. CJ and YZhu: data analysis and interpretation. All authors: collection, assembly of data, manuscript writing, and final approval of manuscript.
Conflict of Interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The authors acknowledge the help of all patients who participated in our study. | Recovered | ReactionOutcome | CC BY | 33718249 | 19,083,671 | 2021 |
What was the outcome of reaction 'Lip dry'? | Case Report: Successful Treatment of Cutaneous Squamous Cell Carcinoma in Three Patients With a Combination of Acitretin and Clarithromycin.
Few studies have evaluated acitretin as a primary treatment for cutaneous squamous cell carcinoma (CSCC). We report, for the first time, three cases of CSCC successfully treated with acitretin and clarithromycin. A literature review on this subject was also was performed. This case report included three patients with CSCC treated with acitretin and clarithromycin at the First Affiliated Hospital of Fujian Medical University (2008-2019). Patient 1 (83-year-old woman, ulcerated mass on the left cheek), patient 2 (97-year-old woman, painful mass on the left cheek) and patient 3 (76-year-old woman, large mass on the right ankle) received 8, 6, and 30 courses of combination therapy. All patients tolerated the adverse effects (pseudotumor cerebri and mucocutaneous dryness) and achieved complete regression within 6 months. Patients 1, 2, and 3 have not experienced recurrence during a 10-, 3-, and 6-year follow-up. Acitretin has limited efficacy as a monotherapy for CSCC. Our experience indicates that combination therapy with acitretin and clarithromycin may be an effective and well-tolerated treatment for unresectable CSCC.
Introduction
Given its increasing incidence and potential for poor outcomes (1–3), cutaneous squamous cell carcinoma (CSCC) is emerging as a public health problem. High-risk CSCC can be difficult to treat by surgical excision because the neoplasm can extend or infiltrate beyond the visible borders of the lesion. A wide range of non-surgical treatments are available for high-risk CSCC. Acitretin is a second generation retinoid shown to reduce the incidence of new primary non-melanoma skin cancers in immunocompromised transplant recipients (4) and patients receiving BRAF inhibitor treatment (5). Acitretin is considered to have limited efficacy as a monotherapy for CSCC but seems more effective when used in combination with other agents (6, 7). Clarithromycin is an antibacterial drug that has been reported to enhance the anti-tumor activity of chemotherapeutic drugs in vitro and in vivo (8, 9). However, no previous study has evaluated the combination of acitretin and clarithromycin in the treatment of CSCC. We present 3 cases illustrating that combination therapy with acitretin and clarithromycin might be a promising option for patients with CSCC who are unwilling or unable to undergo surgery.
Case Description
Ethical Statement
The study was approved by the Medical Technology Clinical Application Ethics Committee of the First Affiliated Hospital of Fujian (No. [2015]084). Before treatment, the patients were fully informed about the potential advantages and disadvantages of combination therapy with acitretin and clarithromycin, including the possible side effects of these drugs. All patients provided written informed consent for treatment and agreed to attend follow-up visits every 2 weeks. All patients provided written informed consent for data analysis and publication.
Patients
The present case report represents a retrospective analysis of the medical records of patients with CSCC who were treated with a combination of acitretin and clarithromycin at the Dermatology Department of the First Affiliated Hospital of Fujian between 2008 and 2018. Inclusion criteria consisted of a diagnosis of SCC by pathological biopsy and a history as well as tumor diameter ≥1 cm (10).
A total of 3 patients were involved in our study. Acitretin and clarithromycin were administered together to the patients from the time when skin biopsy confirmed the diagnosis of invasive SCC. We defined oral 30 mg/day of acitretin for 4 weeks and oral 500 mg/day of clarithromycin for 3 weeks as 1-month treatment course. There was no interval time between courses.
Once biochemical investigations, including measurements of serum creatinine, cholesterol, triglycerides, and potassium appeared abnormal, the treatment was stopped or the dosage was reduced.
Patient 1
Patient 1 was an 83-year-old Chinese woman with a 2-year history of an asymptomatic ulcerated mass on her left cheek. Physical examination in March 2008 revealed a 4 × 3.5 cm crater-like mass with central ulceration on her left cheek (Figure 1A). Skin biopsy revealed the presence of atypical epithelial tumor cell formations extending beyond the epidermis into the underlying dermis, cornified cells and horny pearls, thereby confirming the diagnosis of invasive SCC. The patient showed a favorable response to acitretin and clarithromycin within 2 months after the first oral administration. The patient declined long-term treatment due to the development of adverse effects (pseudotumor cerebri and dryness of the lips and skin). Therefore, CO2 laser resection of the CSCC was performed after a total of 6 courses of combination therapy, and the patient then received an additional two courses of combination therapy. This patient has lived without tumor recurrence for 10 years.
Figure 1 Photographs showing the regression of cutaneous squamous cell carcinoma (CSCC) in three patients treated with acitretin and clarithromycin. (A) Patient 1. The CSCC on the left cheek is shown at month 0 (before initiation of combination therapy), 5 and 12. (B) Patient 2. The CSCC on the left cheek is shown at month 0, 3, and 29. (C) Patient 3. The CSCC on the right ankle is shown at month 0, 6, 30, 42, and 60.
Patient 2
Patient 2 was a 97-year-old Chinese woman with a 1-year history of a painful mass on her left cheek. Physical examination in November 2015 demonstrated the presence of a 2 × 2 cm mass with an irregular edge and signs of erosion on her left cheek (Figure 1B). Skin biopsy confirmed the diagnosis of invasive SCC. The patient was treated with 6 courses of combination therapy. Darkening and regression of the tumor lesions were first detected in 2 months after the initiation of combination therapy, and a complete response was observed at 6 months. The only adverse effect reported by the patient was mucocutaneous dryness, but this resolved following cessation of treatment with acitretin. The patient has not experienced recurrence of CSCC during the 3-year follow-up.
Patient 3
Patient 3 was a 76-year-old Chinese woman with a 1-year history of a large mass on her right ankle that had progressively increased in size. Physical examination in September 2012 revealed a 4 × 4 cm mass with erosion on her right ankle (Figure 1C). Skin biopsy showed that the overlying epidermis was ulcerated with atypical keratinocytes invading the dermis. The keratinocytes were larger in the central parts of the tumor nests and formed eosinophilic horny pearls. The neoplastic keratinocytes had abundant eosinophilic and dyskeratotic cytoplasm and large vesicular nuclei with visible nucleoli and mitoses. An inflammatory reaction was evident, with many lymphocytes surrounding the stromal vessels and penetrating the tumor nests (Figure 2). Immunohistochemical analyses demonstrated that the tumor cells were positive for AE1/AE3 and P63, negative for Ber-EP4, and positive for Ki67 in about 40% of cells, confirming the diagnosis of invasive SCC (Figure 3). The patient was initially treated with 6 courses of combination therapy, which led to substantial regression of the tumor such that it resembled a small residual ulcer, and the patient then decided to discontinue the treatment. A local tumor relapse was detected 2 years later, so the patient resumed combination therapy and received an additional 24 courses of treatment, which resulted in complete remission. The only adverse effect experienced by the patient was mucocutaneous dryness. The patient remains alive and without tumor recurrence after a 6-year follow-up.
Figure 2 Representative images showing the histologic appearance of a tumor specimen from patient 3. (A) Section stained with hematoxylin-eosin (magnification, ×25). (B) Section stained with hematoxylin-eosin (magnification, ×200). (C) Section stained with hematoxylin-eosin (magnification, ×400).
Figure 3 Immunohistology of tumor specimens from patient 3. (A) Tumor cells were positive for AE1/AE3 (magnification, ×25). (B) Tumor cells were positive for P63 (magnification, ×25). (C) Tumor cells were negative for Ber-EP4 (magnification, ×25). (D) Around 40% of tumor cells were positive for Ki67 (magnification, ×400).
Discussion
A notable finding of this case report was that combination therapy achieved CSCC regression in all three patients within 2–6 months after the first administered dose of acitretin and clarithromycin. Furthermore, two of the patients exhibited complete remission of the tumor without scaring, thereby avoiding the need for surgery, while the other patient was successfully treated by tumor excision after combination therapy. Importantly, all three patients remained free of recurrence during a follow-up of 3–10 years. In addition, all three patients tolerated the treatment and experienced only minimal adverse effects (pseudotumor cerebri and mucocutaneous dryness) that resolved after discontinuation of acitretin. To our knowledge, this is the first study to report that combination therapy with acitretin and clarithromycin may be an effective treatment for CSCC.
The majority of CSCCs can be successfully eradicated by surgical excision with post-operative margin assessment or microscopically controlled surgery, although the cosmetic results are not always satisfactory. When patients cannot or refuse to undergo surgery, alternative treatment options include radiotherapy, intralesional chemotherapy and photodynamic therapy. Radiotherapy is a potentially curative treatment in cases of early-stage CSCC where resection would cause substantial cosmetic or functional deficits. However, radiotherapy alone has limited effectiveness as a treatment for advanced CSCC and is associated with adverse effects. Anti-PD-1 antibodies are considered a first-line systemic treatment for patients with metastatic or locally advanced CSCC who are not candidates for curative surgery or radiotherapy, although its use in elderly patients is limited by the toxicity of this chemotherapeutic agent (11, 12). Recently, epidermal growth factor receptor inhibitors combined with chemotherapy or radiotherapy have been used as second-line systemic treatments for advanced CSCC (11, 13), but further research is needed to fully characterize the efficacy and safety of these strategies (11, 14). As awareness grows regarding the increasing number of cases of post-surgical CSSC relapse, it is becoming recognized that improved non-surgical treatments for CSCC are urgently needed.
Retinoids are established drugs and have been shown to reduce the development of non-melanoma skin cancer in high risk individuals. Previous randomized controlled trials have demonstrated that systemic retinoids can prevent new primary CSCC or new precancerous lesions in immunocompromised patients after transplantation (15). Typically, retinoids are considered for patients who develop multiple tumors each year, have multiple CSCCs in high-risk areas, or have numerous actinic keratoses. Acitretin is a second-generation retinoid reported to exhibit anti-tumor activity against actinic keratosis, verrucous carcinomas and SCC. Previous studies of the use of acitretin in the management of CSCC are rare and summarized in Table 1 (5–7). Two of these three studies were case reports describing the use of acitretin in the chemoprevention of CSCC in 8 patients receiving BRAF inhibitors (5) and one patient who had been treated with phototherapy (6). Only one prior study has described the use of acitretin to treat CSCC: the combination of acitretin, intralesional 5-fluorouracil and 5-fluorouracil chemowraps was found to cause complete regression of widespread CSCCs on the legs of two patients (7). To our knowledge, the present case report is the first to report the successful use of acitretin in combination with clarithromycin as a treatment for CSCC.
Table 1 Summary of publications reporting cases and our cases of cutaneous squamous cell carcinoma (CSCC) treated with acitretin.
Reference Patients Treatment Response
Manalo et al. (7) Two patients (74-year-old female and 60-year-old female) with widespread CSCCs on the legs Acitretin (25 mg/d) combined with 5-fluorouracil and chemowraps for 6 months Complete clinical resolution
Anforth et al. (5) Eight patients treated with a BRAF inhibitor Acitretin (10–15 mg/d) for 1–13 months Suppression of CSCC development
Lebwohl et al. (6) A 40-year-old male patient with psoriasis who developed CSCC following PUVA phototherapy Acitretin (25 mg/d) for 25 months Suppression of CSCC development
Patient 1 An 83-year-old Chinese woman with invasive SCC on the left cheek Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 8 courses Complete remission (6
months); 10 years without
recurrence
Patient 2 A 97-year-old Chinese woman with invasive SCC on the left cheek Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 6 courses Complete remission (6
months); 3 years without
recurrence
Patient 3 An 76-year-old Chinese woman with invasive SCC on the right ankle Acitretin (30 mg/d) for 4 weeks and clarithromycin (500 mg/d) for 3 weeks in each course; 6 plus 24 courses resumed* Complete remission (24 months resumed); 6 years without recurrence
* After 6 courses of the combination therapy, this patient self-discontinued the treatment. Two years later, tumor recurrence was detected and the patient resumed the combination treatment for another 24 courses.
Acitretin monotherapy is generally considered to have a low efficacy and is associated with dose-dependent side effects such as mucocutaneous dryness, hair loss, elevated triglycerides, transient elevations in liver function tests and (with long-term treatment) diffuse idiopathic skeletal hyperostosis. In view of the above issues, the patients in this case report were also administered clarithromycin, a macrolide antibiotic that has been reported to enhance the activity of conventional chemotherapy agents in cell lines and animal experiments (8, 9). Furthermore, long-term administration of clarithromycin was reported to prolong the survival time of patients with lung cancer (16). We speculate that the excellent responses to combination therapy observed in all three patients in the present report were due to the antiangiogenic and antitumor effects of clarithromycin, which synergized with the anti-tumor actions of acitretin.
Combination therapy with acitretin and clarithromycin was well-tolerated by all three patients in this study. We summarized the common characteristics of the three patients in our study which may somehow contributed to the successful treatment outcomes. We found that all of them were female older than 75 and refused large-scale surgical treatment. The pathological diagnoses were poorly to well-differentiated invasive SCC. Their tumors were all single and without any metastases. We believe that elderly patients with slow tumor growth respond well to drug treatment and this can help them to avoid the risks associated with surgery. For young patients with multiple or larger tumors, the drug treatment can also be used to reduce the size of the tumor, furthermore it will be beneficial to surgical resection. The only adverse effects reported were pseudotumor cerebri and mucocutaneous dryness, which are known side-effects of acitretin. Therefore, we suggest that the combination of acitretin and clarithromycin may represent an effective and safe therapy for CSCC.
As far as we are aware, this is the first study to report that combination therapy with acitretin and clarithromycin may be an effective and safe treatment for CSCC. Additional studies are required to further assess the efficacy and safety of combination therapy with acitretin and clarithromycin in the treatment of CSCC in larger populations and to investigate the underlying molecular mechanisms.
Data Availability Statement
The original contributions generated in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.
Ethics Statement
The studies involving human participants were reviewed and approved by The study was approved by the Medical Technology Clinical Application Ethics Committee of the First Affiliated Hospital of Fujian (No. [2015]084). The patients/participants provided their written informed consent to participate in this study. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.
Author Contributions
YZha: conception and design. CJ: administrative support. YZha, YZhu, and HW: provision of study materials or patients. CJ and YZhu: data analysis and interpretation. All authors: collection, assembly of data, manuscript writing, and final approval of manuscript.
Conflict of Interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The authors acknowledge the help of all patients who participated in our study. | Recovered | ReactionOutcome | CC BY | 33718249 | 19,083,671 | 2021 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Melaena'. | Accessory Splenic Artery Causing Massive Gastrointestinal Bleed.
The presentation of an upper gastrointestinal bleed secondary to an accessory splenic artery is a rare circumstance described only in 2 previous case reports. This report is the first to describe an upper gastrointestinal bleed consequent of a submucosal accessory splenic artery arising from the left phrenic artery, requiring multiple endoscopies and endovascular embolization. Vascular anatomic variants can pose a challenge to treatment, especially when they are unknown. This case adds to the limited number of case reports involving accessory splenic arteries.
INTRODUCTION
Upper gastrointestinal (GI) bleeds are defined as intraluminal bleeding originating above the ligament of Treitz. There can be significant variation in the presentation of the bleed depending on the etiology. Upper GI bleeding has been reported to have an incidence of approximately 0.08%–0.16% and is a potentially life-threatening condition with a mortality rate of approximately 10%.1,2 Endoscopy and multiphase computed tomography (CT) are the 2 most common modalities initially used to identify the source of GI bleeding. Arterial phase CT can help identify arterial bleeds, while portal-venous phase CT is more beneficial to identify venous or variceal bleeds. Accessory splenic arteries are an uncommon anatomic variant described in few previous reports. There are many theories on the origin of the accessory arteries, including failure of regression or fusion of primitive vitelline arteries during embryological development.3 The presentation of an upper GI bleed secondary to an accessory splenic artery is a rare circumstance, described in only 2 previous case reports.4,5 Our report is the first to describe a massive upper GI bleed consequent of a submucosal accessory splenic artery arising from the left phrenic artery.
CASE REPORT
A 69-year-old man with a medical history significant for chronic neck pain requiring daily ibuprofen use along with the addition of aspirin and methylprednisolone for 1 week before admission presented with a complaint of black tarry stool. On arrival, he was slightly tachycardic and had a hemoglobin of 12 g/dL. However, he subsequently had an episode of large-volume hematemesis and a drop in hemoglobin to 8.6 g/dL. Urgent esophagogastroduodenoscopy (EGD) demonstrated old blood and clots in the gastric fundus without active bleeding. Attempts to clear the clots with suction and Roth Net were unsuccessful.
A repeat EGD the following morning allowed for better visualization of the gastric fundus. A prominent tortuous vessel was identified within the submucosa of the gastric fundus initially believed to be a varix, but no active bleed or stigmata of recent bleeding was identified. Abdominal computed tomography was reviewed, and a prominent artery was identified arising from the left phrenic artery and coursing through the gastric fundus (Figure 1). No gastric varices were identified.
Figure 1. Axial (left) and coronal (right) abdominal computed tomography (CT) showing a prominent artery arising from the left phrenic artery and coursing through the gastric fundus (arrows).
Overnight the patient began to display evidence of a recurrent bleed and underwent visceral angiography, which demonstrated an accessory splenic artery arising from the left phrenic artery and supplying the superior third of the spleen (Figure 2). A conventional splenic artery was also demonstrated in the expected position, supplying the inferior two-thirds of the spleen. Angiography of the remaining mesenteric vessels did not show any active bleeding, and therefore, no intervention was performed.
Figure 2. Digital subtraction angiography image (left) demonstrates the right and left (red arrows) phrenic arteries. The accessory splenic artery is identified (blue arrows), branching off of the left phrenic. Delayed digital subtraction angiography image (right) demonstrates accessory splenic artery supplying the upper third of the spleen.
A few days later, the patient had recurrent bleeding with a drop in hemoglobin level to 5.9 g/dL. The patient was transfused and underwent a third EGD which revealed a serpiginous vessel with a fibrin plug in the gastric fundus. This abnormal artery was believed to be the source of the patient's melena, and 3 hemoclips were placed on the vessel (Figure 3). Although the bleeding initially stopped, the patient had another episode of bleeding the following day. The patient was brought to interventional radiology, and a second angiogram was performed which demonstrated the accessory splenic artery as the clipped culprit vessel (Figure 4). Although no active extravasation was identified, glue embolization of the accessory splenic artery was performed, given it was the likely source of the recurrent bleed.
Figure 3. The third esophagogastroduodenoscopy demonstrating the tortuous submucosal vessel with fibrin clot (left). Endoscopic clipping of the suspected source of bleed (right).
Figure 4. Accessory splenic artery (left) demonstrated coursing adjacent to the endoscopic clips. Digitally subtracted angiogram of the conventional splenic artery (right) confirming course.
After the procedure, the patient did well clinically with no further bleeding. Considering the lack of data, the risk of recurrent bleeding with this anomaly remains largely unknown. As such, he was advised permanent cessation of nonsteroidal anti-inflammatory drugs and steroids. The patient continues to do well, without evidence of recurrent GI bleed.
DISCUSSION
Multiple variants in the origin and course of the splenic artery have been described in the literature, including aberrant origin from the aorta, common hepatic, left gastric, and superior mesenteric arteries. An intrahepatic and intrapancreatic courses, as well as congenital absence or total duplication of the splenic artery, have also been described.5–7 There is, however, a paucity of data regarding accessory splenic arteries.
The presence of an accessory splenic artery is rare, and its presentation as an upper GI bleed has only been described twice previously in the literature. Our case describes a GI bleed consequent of a submucosal accessory splenic artery arising as a branch of the left phrenic artery. This large caliber artery does not meet the criteria to be classified as a Dieulafoy lesion, which is defined as a prominent arteriole measuring 1-3 mm, without tapering and coursing through the submucosa.8 Previous case reports by Kervancioglu et al and Patel and Lowe have both described an accessory splenic artery arising from the left gastric artery and presenting as a gastric bleed.4,5 Cadaveric dissections have also demonstrated the presence of accessory splenic arteries. Padmalath et al reported an accessory splenic artery in a cadaver, arising from the left gastroepiploic artery.9 From our literature review, accessory splenic arteries have not been described as part of congenital syndromes or known to exist specifically with other anomalies.
Our patient ultimately underwent angiographic embolization of the accessory splenic artery due to recurrent bleeding. The available surgical option was more complicated and cumbersome because it involves splenectomy with partial gastrectomy. As such, the lesser invasive option, endovascular embolization, was the preferred modality. After the procedure, the patient was discharged on day 3. The patient continues to do well and remains asymptomatic. Vascular anatomic variants can pose a challenge to treatment, especially when they are unknown. This case adds to the currently limited number of case reports involving accessory splenic arteries.
DISCLOSURES
Author contributions: P. Patel and P. Chadalavada wrote this manuscript. A. Singh provided the endoscopy images. RK Gurajala approved the final manuscript. J-P Achkar revised the manuscript for intellectual content. P. Patel is the article guarantor.
Financial disclosure: None to report.
Informed consent was obtained for this case report. | ASPIRIN, IBUPROFEN, METHYLPREDNISOLONE | DrugsGivenReaction | CC BY-NC-ND | 33718509 | 20,904,986 | 2021-03 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Upper gastrointestinal haemorrhage'. | Accessory Splenic Artery Causing Massive Gastrointestinal Bleed.
The presentation of an upper gastrointestinal bleed secondary to an accessory splenic artery is a rare circumstance described only in 2 previous case reports. This report is the first to describe an upper gastrointestinal bleed consequent of a submucosal accessory splenic artery arising from the left phrenic artery, requiring multiple endoscopies and endovascular embolization. Vascular anatomic variants can pose a challenge to treatment, especially when they are unknown. This case adds to the limited number of case reports involving accessory splenic arteries.
INTRODUCTION
Upper gastrointestinal (GI) bleeds are defined as intraluminal bleeding originating above the ligament of Treitz. There can be significant variation in the presentation of the bleed depending on the etiology. Upper GI bleeding has been reported to have an incidence of approximately 0.08%–0.16% and is a potentially life-threatening condition with a mortality rate of approximately 10%.1,2 Endoscopy and multiphase computed tomography (CT) are the 2 most common modalities initially used to identify the source of GI bleeding. Arterial phase CT can help identify arterial bleeds, while portal-venous phase CT is more beneficial to identify venous or variceal bleeds. Accessory splenic arteries are an uncommon anatomic variant described in few previous reports. There are many theories on the origin of the accessory arteries, including failure of regression or fusion of primitive vitelline arteries during embryological development.3 The presentation of an upper GI bleed secondary to an accessory splenic artery is a rare circumstance, described in only 2 previous case reports.4,5 Our report is the first to describe a massive upper GI bleed consequent of a submucosal accessory splenic artery arising from the left phrenic artery.
CASE REPORT
A 69-year-old man with a medical history significant for chronic neck pain requiring daily ibuprofen use along with the addition of aspirin and methylprednisolone for 1 week before admission presented with a complaint of black tarry stool. On arrival, he was slightly tachycardic and had a hemoglobin of 12 g/dL. However, he subsequently had an episode of large-volume hematemesis and a drop in hemoglobin to 8.6 g/dL. Urgent esophagogastroduodenoscopy (EGD) demonstrated old blood and clots in the gastric fundus without active bleeding. Attempts to clear the clots with suction and Roth Net were unsuccessful.
A repeat EGD the following morning allowed for better visualization of the gastric fundus. A prominent tortuous vessel was identified within the submucosa of the gastric fundus initially believed to be a varix, but no active bleed or stigmata of recent bleeding was identified. Abdominal computed tomography was reviewed, and a prominent artery was identified arising from the left phrenic artery and coursing through the gastric fundus (Figure 1). No gastric varices were identified.
Figure 1. Axial (left) and coronal (right) abdominal computed tomography (CT) showing a prominent artery arising from the left phrenic artery and coursing through the gastric fundus (arrows).
Overnight the patient began to display evidence of a recurrent bleed and underwent visceral angiography, which demonstrated an accessory splenic artery arising from the left phrenic artery and supplying the superior third of the spleen (Figure 2). A conventional splenic artery was also demonstrated in the expected position, supplying the inferior two-thirds of the spleen. Angiography of the remaining mesenteric vessels did not show any active bleeding, and therefore, no intervention was performed.
Figure 2. Digital subtraction angiography image (left) demonstrates the right and left (red arrows) phrenic arteries. The accessory splenic artery is identified (blue arrows), branching off of the left phrenic. Delayed digital subtraction angiography image (right) demonstrates accessory splenic artery supplying the upper third of the spleen.
A few days later, the patient had recurrent bleeding with a drop in hemoglobin level to 5.9 g/dL. The patient was transfused and underwent a third EGD which revealed a serpiginous vessel with a fibrin plug in the gastric fundus. This abnormal artery was believed to be the source of the patient's melena, and 3 hemoclips were placed on the vessel (Figure 3). Although the bleeding initially stopped, the patient had another episode of bleeding the following day. The patient was brought to interventional radiology, and a second angiogram was performed which demonstrated the accessory splenic artery as the clipped culprit vessel (Figure 4). Although no active extravasation was identified, glue embolization of the accessory splenic artery was performed, given it was the likely source of the recurrent bleed.
Figure 3. The third esophagogastroduodenoscopy demonstrating the tortuous submucosal vessel with fibrin clot (left). Endoscopic clipping of the suspected source of bleed (right).
Figure 4. Accessory splenic artery (left) demonstrated coursing adjacent to the endoscopic clips. Digitally subtracted angiogram of the conventional splenic artery (right) confirming course.
After the procedure, the patient did well clinically with no further bleeding. Considering the lack of data, the risk of recurrent bleeding with this anomaly remains largely unknown. As such, he was advised permanent cessation of nonsteroidal anti-inflammatory drugs and steroids. The patient continues to do well, without evidence of recurrent GI bleed.
DISCUSSION
Multiple variants in the origin and course of the splenic artery have been described in the literature, including aberrant origin from the aorta, common hepatic, left gastric, and superior mesenteric arteries. An intrahepatic and intrapancreatic courses, as well as congenital absence or total duplication of the splenic artery, have also been described.5–7 There is, however, a paucity of data regarding accessory splenic arteries.
The presence of an accessory splenic artery is rare, and its presentation as an upper GI bleed has only been described twice previously in the literature. Our case describes a GI bleed consequent of a submucosal accessory splenic artery arising as a branch of the left phrenic artery. This large caliber artery does not meet the criteria to be classified as a Dieulafoy lesion, which is defined as a prominent arteriole measuring 1-3 mm, without tapering and coursing through the submucosa.8 Previous case reports by Kervancioglu et al and Patel and Lowe have both described an accessory splenic artery arising from the left gastric artery and presenting as a gastric bleed.4,5 Cadaveric dissections have also demonstrated the presence of accessory splenic arteries. Padmalath et al reported an accessory splenic artery in a cadaver, arising from the left gastroepiploic artery.9 From our literature review, accessory splenic arteries have not been described as part of congenital syndromes or known to exist specifically with other anomalies.
Our patient ultimately underwent angiographic embolization of the accessory splenic artery due to recurrent bleeding. The available surgical option was more complicated and cumbersome because it involves splenectomy with partial gastrectomy. As such, the lesser invasive option, endovascular embolization, was the preferred modality. After the procedure, the patient was discharged on day 3. The patient continues to do well and remains asymptomatic. Vascular anatomic variants can pose a challenge to treatment, especially when they are unknown. This case adds to the currently limited number of case reports involving accessory splenic arteries.
DISCLOSURES
Author contributions: P. Patel and P. Chadalavada wrote this manuscript. A. Singh provided the endoscopy images. RK Gurajala approved the final manuscript. J-P Achkar revised the manuscript for intellectual content. P. Patel is the article guarantor.
Financial disclosure: None to report.
Informed consent was obtained for this case report. | ASPIRIN, IBUPROFEN, METHYLPREDNISOLONE | DrugsGivenReaction | CC BY-NC-ND | 33718509 | 20,904,986 | 2021-03 |
What was the outcome of reaction 'Melaena'? | Accessory Splenic Artery Causing Massive Gastrointestinal Bleed.
The presentation of an upper gastrointestinal bleed secondary to an accessory splenic artery is a rare circumstance described only in 2 previous case reports. This report is the first to describe an upper gastrointestinal bleed consequent of a submucosal accessory splenic artery arising from the left phrenic artery, requiring multiple endoscopies and endovascular embolization. Vascular anatomic variants can pose a challenge to treatment, especially when they are unknown. This case adds to the limited number of case reports involving accessory splenic arteries.
INTRODUCTION
Upper gastrointestinal (GI) bleeds are defined as intraluminal bleeding originating above the ligament of Treitz. There can be significant variation in the presentation of the bleed depending on the etiology. Upper GI bleeding has been reported to have an incidence of approximately 0.08%–0.16% and is a potentially life-threatening condition with a mortality rate of approximately 10%.1,2 Endoscopy and multiphase computed tomography (CT) are the 2 most common modalities initially used to identify the source of GI bleeding. Arterial phase CT can help identify arterial bleeds, while portal-venous phase CT is more beneficial to identify venous or variceal bleeds. Accessory splenic arteries are an uncommon anatomic variant described in few previous reports. There are many theories on the origin of the accessory arteries, including failure of regression or fusion of primitive vitelline arteries during embryological development.3 The presentation of an upper GI bleed secondary to an accessory splenic artery is a rare circumstance, described in only 2 previous case reports.4,5 Our report is the first to describe a massive upper GI bleed consequent of a submucosal accessory splenic artery arising from the left phrenic artery.
CASE REPORT
A 69-year-old man with a medical history significant for chronic neck pain requiring daily ibuprofen use along with the addition of aspirin and methylprednisolone for 1 week before admission presented with a complaint of black tarry stool. On arrival, he was slightly tachycardic and had a hemoglobin of 12 g/dL. However, he subsequently had an episode of large-volume hematemesis and a drop in hemoglobin to 8.6 g/dL. Urgent esophagogastroduodenoscopy (EGD) demonstrated old blood and clots in the gastric fundus without active bleeding. Attempts to clear the clots with suction and Roth Net were unsuccessful.
A repeat EGD the following morning allowed for better visualization of the gastric fundus. A prominent tortuous vessel was identified within the submucosa of the gastric fundus initially believed to be a varix, but no active bleed or stigmata of recent bleeding was identified. Abdominal computed tomography was reviewed, and a prominent artery was identified arising from the left phrenic artery and coursing through the gastric fundus (Figure 1). No gastric varices were identified.
Figure 1. Axial (left) and coronal (right) abdominal computed tomography (CT) showing a prominent artery arising from the left phrenic artery and coursing through the gastric fundus (arrows).
Overnight the patient began to display evidence of a recurrent bleed and underwent visceral angiography, which demonstrated an accessory splenic artery arising from the left phrenic artery and supplying the superior third of the spleen (Figure 2). A conventional splenic artery was also demonstrated in the expected position, supplying the inferior two-thirds of the spleen. Angiography of the remaining mesenteric vessels did not show any active bleeding, and therefore, no intervention was performed.
Figure 2. Digital subtraction angiography image (left) demonstrates the right and left (red arrows) phrenic arteries. The accessory splenic artery is identified (blue arrows), branching off of the left phrenic. Delayed digital subtraction angiography image (right) demonstrates accessory splenic artery supplying the upper third of the spleen.
A few days later, the patient had recurrent bleeding with a drop in hemoglobin level to 5.9 g/dL. The patient was transfused and underwent a third EGD which revealed a serpiginous vessel with a fibrin plug in the gastric fundus. This abnormal artery was believed to be the source of the patient's melena, and 3 hemoclips were placed on the vessel (Figure 3). Although the bleeding initially stopped, the patient had another episode of bleeding the following day. The patient was brought to interventional radiology, and a second angiogram was performed which demonstrated the accessory splenic artery as the clipped culprit vessel (Figure 4). Although no active extravasation was identified, glue embolization of the accessory splenic artery was performed, given it was the likely source of the recurrent bleed.
Figure 3. The third esophagogastroduodenoscopy demonstrating the tortuous submucosal vessel with fibrin clot (left). Endoscopic clipping of the suspected source of bleed (right).
Figure 4. Accessory splenic artery (left) demonstrated coursing adjacent to the endoscopic clips. Digitally subtracted angiogram of the conventional splenic artery (right) confirming course.
After the procedure, the patient did well clinically with no further bleeding. Considering the lack of data, the risk of recurrent bleeding with this anomaly remains largely unknown. As such, he was advised permanent cessation of nonsteroidal anti-inflammatory drugs and steroids. The patient continues to do well, without evidence of recurrent GI bleed.
DISCUSSION
Multiple variants in the origin and course of the splenic artery have been described in the literature, including aberrant origin from the aorta, common hepatic, left gastric, and superior mesenteric arteries. An intrahepatic and intrapancreatic courses, as well as congenital absence or total duplication of the splenic artery, have also been described.5–7 There is, however, a paucity of data regarding accessory splenic arteries.
The presence of an accessory splenic artery is rare, and its presentation as an upper GI bleed has only been described twice previously in the literature. Our case describes a GI bleed consequent of a submucosal accessory splenic artery arising as a branch of the left phrenic artery. This large caliber artery does not meet the criteria to be classified as a Dieulafoy lesion, which is defined as a prominent arteriole measuring 1-3 mm, without tapering and coursing through the submucosa.8 Previous case reports by Kervancioglu et al and Patel and Lowe have both described an accessory splenic artery arising from the left gastric artery and presenting as a gastric bleed.4,5 Cadaveric dissections have also demonstrated the presence of accessory splenic arteries. Padmalath et al reported an accessory splenic artery in a cadaver, arising from the left gastroepiploic artery.9 From our literature review, accessory splenic arteries have not been described as part of congenital syndromes or known to exist specifically with other anomalies.
Our patient ultimately underwent angiographic embolization of the accessory splenic artery due to recurrent bleeding. The available surgical option was more complicated and cumbersome because it involves splenectomy with partial gastrectomy. As such, the lesser invasive option, endovascular embolization, was the preferred modality. After the procedure, the patient was discharged on day 3. The patient continues to do well and remains asymptomatic. Vascular anatomic variants can pose a challenge to treatment, especially when they are unknown. This case adds to the currently limited number of case reports involving accessory splenic arteries.
DISCLOSURES
Author contributions: P. Patel and P. Chadalavada wrote this manuscript. A. Singh provided the endoscopy images. RK Gurajala approved the final manuscript. J-P Achkar revised the manuscript for intellectual content. P. Patel is the article guarantor.
Financial disclosure: None to report.
Informed consent was obtained for this case report. | Recovered | ReactionOutcome | CC BY-NC-ND | 33718509 | 20,904,986 | 2021-03 |
What was the outcome of reaction 'Upper gastrointestinal haemorrhage'? | Accessory Splenic Artery Causing Massive Gastrointestinal Bleed.
The presentation of an upper gastrointestinal bleed secondary to an accessory splenic artery is a rare circumstance described only in 2 previous case reports. This report is the first to describe an upper gastrointestinal bleed consequent of a submucosal accessory splenic artery arising from the left phrenic artery, requiring multiple endoscopies and endovascular embolization. Vascular anatomic variants can pose a challenge to treatment, especially when they are unknown. This case adds to the limited number of case reports involving accessory splenic arteries.
INTRODUCTION
Upper gastrointestinal (GI) bleeds are defined as intraluminal bleeding originating above the ligament of Treitz. There can be significant variation in the presentation of the bleed depending on the etiology. Upper GI bleeding has been reported to have an incidence of approximately 0.08%–0.16% and is a potentially life-threatening condition with a mortality rate of approximately 10%.1,2 Endoscopy and multiphase computed tomography (CT) are the 2 most common modalities initially used to identify the source of GI bleeding. Arterial phase CT can help identify arterial bleeds, while portal-venous phase CT is more beneficial to identify venous or variceal bleeds. Accessory splenic arteries are an uncommon anatomic variant described in few previous reports. There are many theories on the origin of the accessory arteries, including failure of regression or fusion of primitive vitelline arteries during embryological development.3 The presentation of an upper GI bleed secondary to an accessory splenic artery is a rare circumstance, described in only 2 previous case reports.4,5 Our report is the first to describe a massive upper GI bleed consequent of a submucosal accessory splenic artery arising from the left phrenic artery.
CASE REPORT
A 69-year-old man with a medical history significant for chronic neck pain requiring daily ibuprofen use along with the addition of aspirin and methylprednisolone for 1 week before admission presented with a complaint of black tarry stool. On arrival, he was slightly tachycardic and had a hemoglobin of 12 g/dL. However, he subsequently had an episode of large-volume hematemesis and a drop in hemoglobin to 8.6 g/dL. Urgent esophagogastroduodenoscopy (EGD) demonstrated old blood and clots in the gastric fundus without active bleeding. Attempts to clear the clots with suction and Roth Net were unsuccessful.
A repeat EGD the following morning allowed for better visualization of the gastric fundus. A prominent tortuous vessel was identified within the submucosa of the gastric fundus initially believed to be a varix, but no active bleed or stigmata of recent bleeding was identified. Abdominal computed tomography was reviewed, and a prominent artery was identified arising from the left phrenic artery and coursing through the gastric fundus (Figure 1). No gastric varices were identified.
Figure 1. Axial (left) and coronal (right) abdominal computed tomography (CT) showing a prominent artery arising from the left phrenic artery and coursing through the gastric fundus (arrows).
Overnight the patient began to display evidence of a recurrent bleed and underwent visceral angiography, which demonstrated an accessory splenic artery arising from the left phrenic artery and supplying the superior third of the spleen (Figure 2). A conventional splenic artery was also demonstrated in the expected position, supplying the inferior two-thirds of the spleen. Angiography of the remaining mesenteric vessels did not show any active bleeding, and therefore, no intervention was performed.
Figure 2. Digital subtraction angiography image (left) demonstrates the right and left (red arrows) phrenic arteries. The accessory splenic artery is identified (blue arrows), branching off of the left phrenic. Delayed digital subtraction angiography image (right) demonstrates accessory splenic artery supplying the upper third of the spleen.
A few days later, the patient had recurrent bleeding with a drop in hemoglobin level to 5.9 g/dL. The patient was transfused and underwent a third EGD which revealed a serpiginous vessel with a fibrin plug in the gastric fundus. This abnormal artery was believed to be the source of the patient's melena, and 3 hemoclips were placed on the vessel (Figure 3). Although the bleeding initially stopped, the patient had another episode of bleeding the following day. The patient was brought to interventional radiology, and a second angiogram was performed which demonstrated the accessory splenic artery as the clipped culprit vessel (Figure 4). Although no active extravasation was identified, glue embolization of the accessory splenic artery was performed, given it was the likely source of the recurrent bleed.
Figure 3. The third esophagogastroduodenoscopy demonstrating the tortuous submucosal vessel with fibrin clot (left). Endoscopic clipping of the suspected source of bleed (right).
Figure 4. Accessory splenic artery (left) demonstrated coursing adjacent to the endoscopic clips. Digitally subtracted angiogram of the conventional splenic artery (right) confirming course.
After the procedure, the patient did well clinically with no further bleeding. Considering the lack of data, the risk of recurrent bleeding with this anomaly remains largely unknown. As such, he was advised permanent cessation of nonsteroidal anti-inflammatory drugs and steroids. The patient continues to do well, without evidence of recurrent GI bleed.
DISCUSSION
Multiple variants in the origin and course of the splenic artery have been described in the literature, including aberrant origin from the aorta, common hepatic, left gastric, and superior mesenteric arteries. An intrahepatic and intrapancreatic courses, as well as congenital absence or total duplication of the splenic artery, have also been described.5–7 There is, however, a paucity of data regarding accessory splenic arteries.
The presence of an accessory splenic artery is rare, and its presentation as an upper GI bleed has only been described twice previously in the literature. Our case describes a GI bleed consequent of a submucosal accessory splenic artery arising as a branch of the left phrenic artery. This large caliber artery does not meet the criteria to be classified as a Dieulafoy lesion, which is defined as a prominent arteriole measuring 1-3 mm, without tapering and coursing through the submucosa.8 Previous case reports by Kervancioglu et al and Patel and Lowe have both described an accessory splenic artery arising from the left gastric artery and presenting as a gastric bleed.4,5 Cadaveric dissections have also demonstrated the presence of accessory splenic arteries. Padmalath et al reported an accessory splenic artery in a cadaver, arising from the left gastroepiploic artery.9 From our literature review, accessory splenic arteries have not been described as part of congenital syndromes or known to exist specifically with other anomalies.
Our patient ultimately underwent angiographic embolization of the accessory splenic artery due to recurrent bleeding. The available surgical option was more complicated and cumbersome because it involves splenectomy with partial gastrectomy. As such, the lesser invasive option, endovascular embolization, was the preferred modality. After the procedure, the patient was discharged on day 3. The patient continues to do well and remains asymptomatic. Vascular anatomic variants can pose a challenge to treatment, especially when they are unknown. This case adds to the currently limited number of case reports involving accessory splenic arteries.
DISCLOSURES
Author contributions: P. Patel and P. Chadalavada wrote this manuscript. A. Singh provided the endoscopy images. RK Gurajala approved the final manuscript. J-P Achkar revised the manuscript for intellectual content. P. Patel is the article guarantor.
Financial disclosure: None to report.
Informed consent was obtained for this case report. | Recovered | ReactionOutcome | CC BY-NC-ND | 33718509 | 20,904,986 | 2021-03 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Hypocalcaemia'. | Pembrolizumab-Associated Hypoparathyroidism: A Single Case Report.
To evaluate a case of pembrolizumab-induced hypoparathyroidism leading to hypocalcemia.
The diagnostic tests performed included calcium and parathyroid hormone level detection and calcium-sensing receptor gene analysis.
A 71-year-old Caucasian man was diagnosed with stage IIIB adenocarcinoma of the lung and received radiation therapy but had no other exposure to radiation. Pembrolizumab 200 mg intravenous every 3 weeks was started 5 years after the initial diagnosis. The patient's corrected calcium level was 9.2 mg/dL (normal, 8.5-10.5 mg/dL) at the start of pembrolizumab therapy. The calcium level after the 13th dose of pembrolizumab was 8.1 mg/dL (normal, 8.5-10.2 mg/dL), leading to endocrinology referral. The patient's parathyroid hormone and corrected calcium levels after the 22nd dose were 4.3 mg/dL (normal, 14-72 pg/mL) and 6.5 mg/dL (normal, 8.5-10.2 mg/dL), respectively. He denied symptoms of latent tetany on presentation while on pembrolizumab for 15 months but complained of fatigue and weakness. The patient had no history of autoimmune diseases or neck injuries. Calcium-sensing receptor gene analysis was negative for genetic mutations. Immunotherapy-mediated hypoparathyroidism was diagnosed. He was treated with daily oral calcium carbonate (2000 mg), calcitriol 0.5 μg, 1 dose of calcium gluconate 2 g intravenous, and 3 doses of calcium chloride 1 g intravenous. His fatigue, weakness, and calcium levels improved with therapy.
Pembrolizumab treatment may have resulted in immune-mediated hypoparathyroidism, leading to hypocalcemia. It is important to report such cases to understand its presentation and timing in relation to pembrolizumab, which further facilitates its timely treatment.
Introduction
Most cases of acquired hypoparathyroidism occur after thyroidectomy or are autoimmune mediated. In adults, most cases occur postsurgically. In a retrospective analysis of adults evaluated in a metabolic bone unit of an endocrinology service in Brazil, over 80% of cases of acquired hypoparathyroidism were postsurgical whereas <20% were autoimmune.1 Drug-induced hypoparathyroidism is extremely rare. Here we report a case of hypoparathyroidism and hypocalcemia with pembrolizumab, an immune checkpoint inhibitor used for the treatment of advanced lung cancer.
Case Report
A 71-year-old Caucasian man was diagnosed with a malignant neoplasm of an unspecified part of the right bronchus or right lung (stage IIIB) and a malignant neoplasm of the upper lobe of the left lung and left bronchus. Right lower and upper lobe wedge resection performed soon after diagnosis demonstrated an adenocarcinoma. Two different courses of radiation treatment were performed, one in the left lung and a subsequent treatment in the right lobe of the patient’s lung. Pembrolizumab, 200 mg intravenous every 3 weeks, was started 5 years after the initial diagnosis of cancer after the patient demonstrated a high (100%) expression of programmed death-ligand 1 in an adenosquamous pathology core sample. The patient’s corrected calcium level was 9.2 mg/dL (normal, 8.5-10.2 mg/dL) at the start of pembrolizumab therapy as shown in Figure 1 and the Table. He had no past medical history or other risk factors for hypocalcemia. The calcium level after the 13th dose of pembrolizumab was 8.1 mg/dL (normal, 8.5-10.2 mg/dL), leading to endocrinology referral.Fig. 1 Trend of corrected calcium levels during pembrolizumab therapy and interventions for correction of low calcium levels.
Table Trend of Parathyroid Hormone, Corrected Calcium Levels, and 25-OH Vitamin D Levels in Relation to Pemobrolizumab Therapy
Date Parathyroid Hormone (14-72 pg/ml) Corrected calcium (8.5-10.5 mg/dL) Pemobrolizumab dose No. 25-OH vitamin D levels (30-100 ng/ml)
Jun 2017 … 9.2 1 …
March 2018 … 8.5 13 …
April 2018 … 8.1 14 …
May 2018 … 7.8 15 and 16 …
July 2018 … 7.5 18 and 19 …
August 2018 … 7.2 20 …
September 2018 20.4 7.7 21 …
October 2018 … 7.9 22 35
December 2018 … 6.9 25 …
January 2019 4.3 6.5 26 …
February 2019 7.2 6.4 27 …
February 2019 … 7.2 28 21
March 2019 … 8.4 29 …
April 2019 … 7.7 30 …
April 2019 … 7.9 31 …
May 2019 … 8.2 32 …
July 2019 2.3 8.7 33 …
November 2019 1.2 7.4 34 96
The patient’s parathyroid hormone (PTH) and corrected calcium levels after the 18th dose of pembrolizumab were 20.4 pg/mL (normal, 14-72 pg/mL) and 7.7 mg/dL (normal, 8.5-10.2 mg/dL), respectively (Table). After the 22nd dose, the levels were 4.3 mg/dL (normal, 14-72 pg/mL) and 6.5 (normal, 8.5-10.2 mg/dL), respectively (Table). At the time of presentation, the patient had been on pembrolizumab for 15 months. He denied symptoms of latent tetany but complained of fatigue and weakness. His PTH and calcium levels remained low as shown in Figure 2. The patient denied a history of autoimmune diseases, neck injuries, or exposure to neck radiation. Calcium-sensing receptor gene analysis (LabCorp Specialty Testing Group) was negative for any kind of genetic mutation. Pembrolizumab-associated hypoparathyroidism was diagnosed. The patient was started on daily oral calcium carbonate (2000 mg), calcitriol 0.5 μg, 1 dose of calcium gluconate 2 g intravenous, and 3 doses of calcium chloride 1 g intravenous. He continued to receive vitamin D 50 000 IU weekly for 6 months, followed by 50 000 IU every 2 weeks. His fatigue, weakness, and calcium levels improved with therapy (Fig. 2).Fig. 2 Trend of parathyroid hormone level during pembrolizumab therapy.
Discussion
To the best of our knowledge, this is the third reported case of pembrolizumab-induced hypoparathyroidism and hypocalcemia. In our case, both PTH and calcium levels started to decrease after the 13th dose of pembrolizumab. In the 2 previously reported cases, the decrease in serum calcium levels was reported after the 15th and the second doses of pembrolizumab, respectively.2,3 Similar events have been recently reported after 1.5 months of concurrent treatment with iplimumab and nivolumab, which are checkpoint inhibitors used for treating melanoma.4 Hypocalcemia and hypoparathyroidism were noted in our patient 1.5 months after starting the therapy. It is therefore clear that the dose-response relationship and the timing of pembrolizumab treatment and hypoparathyroidism can vary on a case-to-case basis, emphasizing the need for frequent monitoring of calcium and parathyroid levels.
Fatigue secondary to hypocalcemia was commonly reported in all previously published cases. Fatigue and weakness, if reported, resolved as the calcium level started to normalize in response to treatment. In our patient, calcium levels were stable on oral calcium supplementation, calcitriol, and 50 000 IU per week of vitamin D supplementation for 6 months, followed by 50 000 IU of vitamin D every 2 weeks. A higher requirement of calcium and calcitriol with failure to normalize calcium or the occurrence of unwanted side effects affected tolerability and long-term adherence. Treatment with recombinant human PTH can be considered. One of the limitations of our case report is that we did not have ionized calcium levels available prior to the start of pembrolizumab treatment; however, with the available data on corrected calcium and PTH levels, the clinical message is very clear.
Pembrolizumab is effective for the treatment of various advanced malignancies such as lung cancer as monotherapy and in combination with other chemotherapy agents.5 However, it has been shown to induce immune-related side effects, including disturbances of the endocrine system, and most commonly leads to adrenal insufficiency, pituitary hypophysitis, and thyroid dysfunction.5 Hypoparathyroidism from immune checkpoint inhibitors is rare. So far, a total of 3 cases, including our case, describing hypoparathyroidism associated with pembrolizumab have been reported. Proposed mechanisms include autoimmune hypoparathyroidism caused by increased T-cell activity against parathyroid tissue or the development of calcium-sensing receptor autoantibodies that inhibit PTH secretion. Vigilance is warranted to facilitate timely recognition and the treatment of immune checkpoint inhibitor-related hypoparathyroidism. It is not clear whether the hypoparathyroidism will be permanent after the discontinuation of pembrolizumab treatment or will resolve over time. Further research and/or case studies are needed to advance our understanding of this relationship and its mechanisms.
Conclusion
Pembrolizumab may result in immune-mediated hypoparathyroidism leading to hypocalcemia. We recommend timely recognition and treatment of this endocrine disturbance.
Author Contributions
I.M. and N.D.K. contributed equally to this work.
Disclosure
The authors have no multiplicity of interest to disclose. | PEMBROLIZUMAB | DrugsGivenReaction | CC BY | 33718603 | 20,140,584 | 2021 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Hypoparathyroidism'. | Pembrolizumab-Associated Hypoparathyroidism: A Single Case Report.
To evaluate a case of pembrolizumab-induced hypoparathyroidism leading to hypocalcemia.
The diagnostic tests performed included calcium and parathyroid hormone level detection and calcium-sensing receptor gene analysis.
A 71-year-old Caucasian man was diagnosed with stage IIIB adenocarcinoma of the lung and received radiation therapy but had no other exposure to radiation. Pembrolizumab 200 mg intravenous every 3 weeks was started 5 years after the initial diagnosis. The patient's corrected calcium level was 9.2 mg/dL (normal, 8.5-10.5 mg/dL) at the start of pembrolizumab therapy. The calcium level after the 13th dose of pembrolizumab was 8.1 mg/dL (normal, 8.5-10.2 mg/dL), leading to endocrinology referral. The patient's parathyroid hormone and corrected calcium levels after the 22nd dose were 4.3 mg/dL (normal, 14-72 pg/mL) and 6.5 mg/dL (normal, 8.5-10.2 mg/dL), respectively. He denied symptoms of latent tetany on presentation while on pembrolizumab for 15 months but complained of fatigue and weakness. The patient had no history of autoimmune diseases or neck injuries. Calcium-sensing receptor gene analysis was negative for genetic mutations. Immunotherapy-mediated hypoparathyroidism was diagnosed. He was treated with daily oral calcium carbonate (2000 mg), calcitriol 0.5 μg, 1 dose of calcium gluconate 2 g intravenous, and 3 doses of calcium chloride 1 g intravenous. His fatigue, weakness, and calcium levels improved with therapy.
Pembrolizumab treatment may have resulted in immune-mediated hypoparathyroidism, leading to hypocalcemia. It is important to report such cases to understand its presentation and timing in relation to pembrolizumab, which further facilitates its timely treatment.
Introduction
Most cases of acquired hypoparathyroidism occur after thyroidectomy or are autoimmune mediated. In adults, most cases occur postsurgically. In a retrospective analysis of adults evaluated in a metabolic bone unit of an endocrinology service in Brazil, over 80% of cases of acquired hypoparathyroidism were postsurgical whereas <20% were autoimmune.1 Drug-induced hypoparathyroidism is extremely rare. Here we report a case of hypoparathyroidism and hypocalcemia with pembrolizumab, an immune checkpoint inhibitor used for the treatment of advanced lung cancer.
Case Report
A 71-year-old Caucasian man was diagnosed with a malignant neoplasm of an unspecified part of the right bronchus or right lung (stage IIIB) and a malignant neoplasm of the upper lobe of the left lung and left bronchus. Right lower and upper lobe wedge resection performed soon after diagnosis demonstrated an adenocarcinoma. Two different courses of radiation treatment were performed, one in the left lung and a subsequent treatment in the right lobe of the patient’s lung. Pembrolizumab, 200 mg intravenous every 3 weeks, was started 5 years after the initial diagnosis of cancer after the patient demonstrated a high (100%) expression of programmed death-ligand 1 in an adenosquamous pathology core sample. The patient’s corrected calcium level was 9.2 mg/dL (normal, 8.5-10.2 mg/dL) at the start of pembrolizumab therapy as shown in Figure 1 and the Table. He had no past medical history or other risk factors for hypocalcemia. The calcium level after the 13th dose of pembrolizumab was 8.1 mg/dL (normal, 8.5-10.2 mg/dL), leading to endocrinology referral.Fig. 1 Trend of corrected calcium levels during pembrolizumab therapy and interventions for correction of low calcium levels.
Table Trend of Parathyroid Hormone, Corrected Calcium Levels, and 25-OH Vitamin D Levels in Relation to Pemobrolizumab Therapy
Date Parathyroid Hormone (14-72 pg/ml) Corrected calcium (8.5-10.5 mg/dL) Pemobrolizumab dose No. 25-OH vitamin D levels (30-100 ng/ml)
Jun 2017 … 9.2 1 …
March 2018 … 8.5 13 …
April 2018 … 8.1 14 …
May 2018 … 7.8 15 and 16 …
July 2018 … 7.5 18 and 19 …
August 2018 … 7.2 20 …
September 2018 20.4 7.7 21 …
October 2018 … 7.9 22 35
December 2018 … 6.9 25 …
January 2019 4.3 6.5 26 …
February 2019 7.2 6.4 27 …
February 2019 … 7.2 28 21
March 2019 … 8.4 29 …
April 2019 … 7.7 30 …
April 2019 … 7.9 31 …
May 2019 … 8.2 32 …
July 2019 2.3 8.7 33 …
November 2019 1.2 7.4 34 96
The patient’s parathyroid hormone (PTH) and corrected calcium levels after the 18th dose of pembrolizumab were 20.4 pg/mL (normal, 14-72 pg/mL) and 7.7 mg/dL (normal, 8.5-10.2 mg/dL), respectively (Table). After the 22nd dose, the levels were 4.3 mg/dL (normal, 14-72 pg/mL) and 6.5 (normal, 8.5-10.2 mg/dL), respectively (Table). At the time of presentation, the patient had been on pembrolizumab for 15 months. He denied symptoms of latent tetany but complained of fatigue and weakness. His PTH and calcium levels remained low as shown in Figure 2. The patient denied a history of autoimmune diseases, neck injuries, or exposure to neck radiation. Calcium-sensing receptor gene analysis (LabCorp Specialty Testing Group) was negative for any kind of genetic mutation. Pembrolizumab-associated hypoparathyroidism was diagnosed. The patient was started on daily oral calcium carbonate (2000 mg), calcitriol 0.5 μg, 1 dose of calcium gluconate 2 g intravenous, and 3 doses of calcium chloride 1 g intravenous. He continued to receive vitamin D 50 000 IU weekly for 6 months, followed by 50 000 IU every 2 weeks. His fatigue, weakness, and calcium levels improved with therapy (Fig. 2).Fig. 2 Trend of parathyroid hormone level during pembrolizumab therapy.
Discussion
To the best of our knowledge, this is the third reported case of pembrolizumab-induced hypoparathyroidism and hypocalcemia. In our case, both PTH and calcium levels started to decrease after the 13th dose of pembrolizumab. In the 2 previously reported cases, the decrease in serum calcium levels was reported after the 15th and the second doses of pembrolizumab, respectively.2,3 Similar events have been recently reported after 1.5 months of concurrent treatment with iplimumab and nivolumab, which are checkpoint inhibitors used for treating melanoma.4 Hypocalcemia and hypoparathyroidism were noted in our patient 1.5 months after starting the therapy. It is therefore clear that the dose-response relationship and the timing of pembrolizumab treatment and hypoparathyroidism can vary on a case-to-case basis, emphasizing the need for frequent monitoring of calcium and parathyroid levels.
Fatigue secondary to hypocalcemia was commonly reported in all previously published cases. Fatigue and weakness, if reported, resolved as the calcium level started to normalize in response to treatment. In our patient, calcium levels were stable on oral calcium supplementation, calcitriol, and 50 000 IU per week of vitamin D supplementation for 6 months, followed by 50 000 IU of vitamin D every 2 weeks. A higher requirement of calcium and calcitriol with failure to normalize calcium or the occurrence of unwanted side effects affected tolerability and long-term adherence. Treatment with recombinant human PTH can be considered. One of the limitations of our case report is that we did not have ionized calcium levels available prior to the start of pembrolizumab treatment; however, with the available data on corrected calcium and PTH levels, the clinical message is very clear.
Pembrolizumab is effective for the treatment of various advanced malignancies such as lung cancer as monotherapy and in combination with other chemotherapy agents.5 However, it has been shown to induce immune-related side effects, including disturbances of the endocrine system, and most commonly leads to adrenal insufficiency, pituitary hypophysitis, and thyroid dysfunction.5 Hypoparathyroidism from immune checkpoint inhibitors is rare. So far, a total of 3 cases, including our case, describing hypoparathyroidism associated with pembrolizumab have been reported. Proposed mechanisms include autoimmune hypoparathyroidism caused by increased T-cell activity against parathyroid tissue or the development of calcium-sensing receptor autoantibodies that inhibit PTH secretion. Vigilance is warranted to facilitate timely recognition and the treatment of immune checkpoint inhibitor-related hypoparathyroidism. It is not clear whether the hypoparathyroidism will be permanent after the discontinuation of pembrolizumab treatment or will resolve over time. Further research and/or case studies are needed to advance our understanding of this relationship and its mechanisms.
Conclusion
Pembrolizumab may result in immune-mediated hypoparathyroidism leading to hypocalcemia. We recommend timely recognition and treatment of this endocrine disturbance.
Author Contributions
I.M. and N.D.K. contributed equally to this work.
Disclosure
The authors have no multiplicity of interest to disclose. | PEMBROLIZUMAB | DrugsGivenReaction | CC BY | 33718603 | 20,140,584 | 2021 |
What was the administration route of drug 'PEMBROLIZUMAB'? | Pembrolizumab-Associated Hypoparathyroidism: A Single Case Report.
To evaluate a case of pembrolizumab-induced hypoparathyroidism leading to hypocalcemia.
The diagnostic tests performed included calcium and parathyroid hormone level detection and calcium-sensing receptor gene analysis.
A 71-year-old Caucasian man was diagnosed with stage IIIB adenocarcinoma of the lung and received radiation therapy but had no other exposure to radiation. Pembrolizumab 200 mg intravenous every 3 weeks was started 5 years after the initial diagnosis. The patient's corrected calcium level was 9.2 mg/dL (normal, 8.5-10.5 mg/dL) at the start of pembrolizumab therapy. The calcium level after the 13th dose of pembrolizumab was 8.1 mg/dL (normal, 8.5-10.2 mg/dL), leading to endocrinology referral. The patient's parathyroid hormone and corrected calcium levels after the 22nd dose were 4.3 mg/dL (normal, 14-72 pg/mL) and 6.5 mg/dL (normal, 8.5-10.2 mg/dL), respectively. He denied symptoms of latent tetany on presentation while on pembrolizumab for 15 months but complained of fatigue and weakness. The patient had no history of autoimmune diseases or neck injuries. Calcium-sensing receptor gene analysis was negative for genetic mutations. Immunotherapy-mediated hypoparathyroidism was diagnosed. He was treated with daily oral calcium carbonate (2000 mg), calcitriol 0.5 μg, 1 dose of calcium gluconate 2 g intravenous, and 3 doses of calcium chloride 1 g intravenous. His fatigue, weakness, and calcium levels improved with therapy.
Pembrolizumab treatment may have resulted in immune-mediated hypoparathyroidism, leading to hypocalcemia. It is important to report such cases to understand its presentation and timing in relation to pembrolizumab, which further facilitates its timely treatment.
Introduction
Most cases of acquired hypoparathyroidism occur after thyroidectomy or are autoimmune mediated. In adults, most cases occur postsurgically. In a retrospective analysis of adults evaluated in a metabolic bone unit of an endocrinology service in Brazil, over 80% of cases of acquired hypoparathyroidism were postsurgical whereas <20% were autoimmune.1 Drug-induced hypoparathyroidism is extremely rare. Here we report a case of hypoparathyroidism and hypocalcemia with pembrolizumab, an immune checkpoint inhibitor used for the treatment of advanced lung cancer.
Case Report
A 71-year-old Caucasian man was diagnosed with a malignant neoplasm of an unspecified part of the right bronchus or right lung (stage IIIB) and a malignant neoplasm of the upper lobe of the left lung and left bronchus. Right lower and upper lobe wedge resection performed soon after diagnosis demonstrated an adenocarcinoma. Two different courses of radiation treatment were performed, one in the left lung and a subsequent treatment in the right lobe of the patient’s lung. Pembrolizumab, 200 mg intravenous every 3 weeks, was started 5 years after the initial diagnosis of cancer after the patient demonstrated a high (100%) expression of programmed death-ligand 1 in an adenosquamous pathology core sample. The patient’s corrected calcium level was 9.2 mg/dL (normal, 8.5-10.2 mg/dL) at the start of pembrolizumab therapy as shown in Figure 1 and the Table. He had no past medical history or other risk factors for hypocalcemia. The calcium level after the 13th dose of pembrolizumab was 8.1 mg/dL (normal, 8.5-10.2 mg/dL), leading to endocrinology referral.Fig. 1 Trend of corrected calcium levels during pembrolizumab therapy and interventions for correction of low calcium levels.
Table Trend of Parathyroid Hormone, Corrected Calcium Levels, and 25-OH Vitamin D Levels in Relation to Pemobrolizumab Therapy
Date Parathyroid Hormone (14-72 pg/ml) Corrected calcium (8.5-10.5 mg/dL) Pemobrolizumab dose No. 25-OH vitamin D levels (30-100 ng/ml)
Jun 2017 … 9.2 1 …
March 2018 … 8.5 13 …
April 2018 … 8.1 14 …
May 2018 … 7.8 15 and 16 …
July 2018 … 7.5 18 and 19 …
August 2018 … 7.2 20 …
September 2018 20.4 7.7 21 …
October 2018 … 7.9 22 35
December 2018 … 6.9 25 …
January 2019 4.3 6.5 26 …
February 2019 7.2 6.4 27 …
February 2019 … 7.2 28 21
March 2019 … 8.4 29 …
April 2019 … 7.7 30 …
April 2019 … 7.9 31 …
May 2019 … 8.2 32 …
July 2019 2.3 8.7 33 …
November 2019 1.2 7.4 34 96
The patient’s parathyroid hormone (PTH) and corrected calcium levels after the 18th dose of pembrolizumab were 20.4 pg/mL (normal, 14-72 pg/mL) and 7.7 mg/dL (normal, 8.5-10.2 mg/dL), respectively (Table). After the 22nd dose, the levels were 4.3 mg/dL (normal, 14-72 pg/mL) and 6.5 (normal, 8.5-10.2 mg/dL), respectively (Table). At the time of presentation, the patient had been on pembrolizumab for 15 months. He denied symptoms of latent tetany but complained of fatigue and weakness. His PTH and calcium levels remained low as shown in Figure 2. The patient denied a history of autoimmune diseases, neck injuries, or exposure to neck radiation. Calcium-sensing receptor gene analysis (LabCorp Specialty Testing Group) was negative for any kind of genetic mutation. Pembrolizumab-associated hypoparathyroidism was diagnosed. The patient was started on daily oral calcium carbonate (2000 mg), calcitriol 0.5 μg, 1 dose of calcium gluconate 2 g intravenous, and 3 doses of calcium chloride 1 g intravenous. He continued to receive vitamin D 50 000 IU weekly for 6 months, followed by 50 000 IU every 2 weeks. His fatigue, weakness, and calcium levels improved with therapy (Fig. 2).Fig. 2 Trend of parathyroid hormone level during pembrolizumab therapy.
Discussion
To the best of our knowledge, this is the third reported case of pembrolizumab-induced hypoparathyroidism and hypocalcemia. In our case, both PTH and calcium levels started to decrease after the 13th dose of pembrolizumab. In the 2 previously reported cases, the decrease in serum calcium levels was reported after the 15th and the second doses of pembrolizumab, respectively.2,3 Similar events have been recently reported after 1.5 months of concurrent treatment with iplimumab and nivolumab, which are checkpoint inhibitors used for treating melanoma.4 Hypocalcemia and hypoparathyroidism were noted in our patient 1.5 months after starting the therapy. It is therefore clear that the dose-response relationship and the timing of pembrolizumab treatment and hypoparathyroidism can vary on a case-to-case basis, emphasizing the need for frequent monitoring of calcium and parathyroid levels.
Fatigue secondary to hypocalcemia was commonly reported in all previously published cases. Fatigue and weakness, if reported, resolved as the calcium level started to normalize in response to treatment. In our patient, calcium levels were stable on oral calcium supplementation, calcitriol, and 50 000 IU per week of vitamin D supplementation for 6 months, followed by 50 000 IU of vitamin D every 2 weeks. A higher requirement of calcium and calcitriol with failure to normalize calcium or the occurrence of unwanted side effects affected tolerability and long-term adherence. Treatment with recombinant human PTH can be considered. One of the limitations of our case report is that we did not have ionized calcium levels available prior to the start of pembrolizumab treatment; however, with the available data on corrected calcium and PTH levels, the clinical message is very clear.
Pembrolizumab is effective for the treatment of various advanced malignancies such as lung cancer as monotherapy and in combination with other chemotherapy agents.5 However, it has been shown to induce immune-related side effects, including disturbances of the endocrine system, and most commonly leads to adrenal insufficiency, pituitary hypophysitis, and thyroid dysfunction.5 Hypoparathyroidism from immune checkpoint inhibitors is rare. So far, a total of 3 cases, including our case, describing hypoparathyroidism associated with pembrolizumab have been reported. Proposed mechanisms include autoimmune hypoparathyroidism caused by increased T-cell activity against parathyroid tissue or the development of calcium-sensing receptor autoantibodies that inhibit PTH secretion. Vigilance is warranted to facilitate timely recognition and the treatment of immune checkpoint inhibitor-related hypoparathyroidism. It is not clear whether the hypoparathyroidism will be permanent after the discontinuation of pembrolizumab treatment or will resolve over time. Further research and/or case studies are needed to advance our understanding of this relationship and its mechanisms.
Conclusion
Pembrolizumab may result in immune-mediated hypoparathyroidism leading to hypocalcemia. We recommend timely recognition and treatment of this endocrine disturbance.
Author Contributions
I.M. and N.D.K. contributed equally to this work.
Disclosure
The authors have no multiplicity of interest to disclose. | Intravenous (not otherwise specified) | DrugAdministrationRoute | CC BY | 33718603 | 20,140,584 | 2021 |
What was the outcome of reaction 'Hypocalcaemia'? | Pembrolizumab-Associated Hypoparathyroidism: A Single Case Report.
To evaluate a case of pembrolizumab-induced hypoparathyroidism leading to hypocalcemia.
The diagnostic tests performed included calcium and parathyroid hormone level detection and calcium-sensing receptor gene analysis.
A 71-year-old Caucasian man was diagnosed with stage IIIB adenocarcinoma of the lung and received radiation therapy but had no other exposure to radiation. Pembrolizumab 200 mg intravenous every 3 weeks was started 5 years after the initial diagnosis. The patient's corrected calcium level was 9.2 mg/dL (normal, 8.5-10.5 mg/dL) at the start of pembrolizumab therapy. The calcium level after the 13th dose of pembrolizumab was 8.1 mg/dL (normal, 8.5-10.2 mg/dL), leading to endocrinology referral. The patient's parathyroid hormone and corrected calcium levels after the 22nd dose were 4.3 mg/dL (normal, 14-72 pg/mL) and 6.5 mg/dL (normal, 8.5-10.2 mg/dL), respectively. He denied symptoms of latent tetany on presentation while on pembrolizumab for 15 months but complained of fatigue and weakness. The patient had no history of autoimmune diseases or neck injuries. Calcium-sensing receptor gene analysis was negative for genetic mutations. Immunotherapy-mediated hypoparathyroidism was diagnosed. He was treated with daily oral calcium carbonate (2000 mg), calcitriol 0.5 μg, 1 dose of calcium gluconate 2 g intravenous, and 3 doses of calcium chloride 1 g intravenous. His fatigue, weakness, and calcium levels improved with therapy.
Pembrolizumab treatment may have resulted in immune-mediated hypoparathyroidism, leading to hypocalcemia. It is important to report such cases to understand its presentation and timing in relation to pembrolizumab, which further facilitates its timely treatment.
Introduction
Most cases of acquired hypoparathyroidism occur after thyroidectomy or are autoimmune mediated. In adults, most cases occur postsurgically. In a retrospective analysis of adults evaluated in a metabolic bone unit of an endocrinology service in Brazil, over 80% of cases of acquired hypoparathyroidism were postsurgical whereas <20% were autoimmune.1 Drug-induced hypoparathyroidism is extremely rare. Here we report a case of hypoparathyroidism and hypocalcemia with pembrolizumab, an immune checkpoint inhibitor used for the treatment of advanced lung cancer.
Case Report
A 71-year-old Caucasian man was diagnosed with a malignant neoplasm of an unspecified part of the right bronchus or right lung (stage IIIB) and a malignant neoplasm of the upper lobe of the left lung and left bronchus. Right lower and upper lobe wedge resection performed soon after diagnosis demonstrated an adenocarcinoma. Two different courses of radiation treatment were performed, one in the left lung and a subsequent treatment in the right lobe of the patient’s lung. Pembrolizumab, 200 mg intravenous every 3 weeks, was started 5 years after the initial diagnosis of cancer after the patient demonstrated a high (100%) expression of programmed death-ligand 1 in an adenosquamous pathology core sample. The patient’s corrected calcium level was 9.2 mg/dL (normal, 8.5-10.2 mg/dL) at the start of pembrolizumab therapy as shown in Figure 1 and the Table. He had no past medical history or other risk factors for hypocalcemia. The calcium level after the 13th dose of pembrolizumab was 8.1 mg/dL (normal, 8.5-10.2 mg/dL), leading to endocrinology referral.Fig. 1 Trend of corrected calcium levels during pembrolizumab therapy and interventions for correction of low calcium levels.
Table Trend of Parathyroid Hormone, Corrected Calcium Levels, and 25-OH Vitamin D Levels in Relation to Pemobrolizumab Therapy
Date Parathyroid Hormone (14-72 pg/ml) Corrected calcium (8.5-10.5 mg/dL) Pemobrolizumab dose No. 25-OH vitamin D levels (30-100 ng/ml)
Jun 2017 … 9.2 1 …
March 2018 … 8.5 13 …
April 2018 … 8.1 14 …
May 2018 … 7.8 15 and 16 …
July 2018 … 7.5 18 and 19 …
August 2018 … 7.2 20 …
September 2018 20.4 7.7 21 …
October 2018 … 7.9 22 35
December 2018 … 6.9 25 …
January 2019 4.3 6.5 26 …
February 2019 7.2 6.4 27 …
February 2019 … 7.2 28 21
March 2019 … 8.4 29 …
April 2019 … 7.7 30 …
April 2019 … 7.9 31 …
May 2019 … 8.2 32 …
July 2019 2.3 8.7 33 …
November 2019 1.2 7.4 34 96
The patient’s parathyroid hormone (PTH) and corrected calcium levels after the 18th dose of pembrolizumab were 20.4 pg/mL (normal, 14-72 pg/mL) and 7.7 mg/dL (normal, 8.5-10.2 mg/dL), respectively (Table). After the 22nd dose, the levels were 4.3 mg/dL (normal, 14-72 pg/mL) and 6.5 (normal, 8.5-10.2 mg/dL), respectively (Table). At the time of presentation, the patient had been on pembrolizumab for 15 months. He denied symptoms of latent tetany but complained of fatigue and weakness. His PTH and calcium levels remained low as shown in Figure 2. The patient denied a history of autoimmune diseases, neck injuries, or exposure to neck radiation. Calcium-sensing receptor gene analysis (LabCorp Specialty Testing Group) was negative for any kind of genetic mutation. Pembrolizumab-associated hypoparathyroidism was diagnosed. The patient was started on daily oral calcium carbonate (2000 mg), calcitriol 0.5 μg, 1 dose of calcium gluconate 2 g intravenous, and 3 doses of calcium chloride 1 g intravenous. He continued to receive vitamin D 50 000 IU weekly for 6 months, followed by 50 000 IU every 2 weeks. His fatigue, weakness, and calcium levels improved with therapy (Fig. 2).Fig. 2 Trend of parathyroid hormone level during pembrolizumab therapy.
Discussion
To the best of our knowledge, this is the third reported case of pembrolizumab-induced hypoparathyroidism and hypocalcemia. In our case, both PTH and calcium levels started to decrease after the 13th dose of pembrolizumab. In the 2 previously reported cases, the decrease in serum calcium levels was reported after the 15th and the second doses of pembrolizumab, respectively.2,3 Similar events have been recently reported after 1.5 months of concurrent treatment with iplimumab and nivolumab, which are checkpoint inhibitors used for treating melanoma.4 Hypocalcemia and hypoparathyroidism were noted in our patient 1.5 months after starting the therapy. It is therefore clear that the dose-response relationship and the timing of pembrolizumab treatment and hypoparathyroidism can vary on a case-to-case basis, emphasizing the need for frequent monitoring of calcium and parathyroid levels.
Fatigue secondary to hypocalcemia was commonly reported in all previously published cases. Fatigue and weakness, if reported, resolved as the calcium level started to normalize in response to treatment. In our patient, calcium levels were stable on oral calcium supplementation, calcitriol, and 50 000 IU per week of vitamin D supplementation for 6 months, followed by 50 000 IU of vitamin D every 2 weeks. A higher requirement of calcium and calcitriol with failure to normalize calcium or the occurrence of unwanted side effects affected tolerability and long-term adherence. Treatment with recombinant human PTH can be considered. One of the limitations of our case report is that we did not have ionized calcium levels available prior to the start of pembrolizumab treatment; however, with the available data on corrected calcium and PTH levels, the clinical message is very clear.
Pembrolizumab is effective for the treatment of various advanced malignancies such as lung cancer as monotherapy and in combination with other chemotherapy agents.5 However, it has been shown to induce immune-related side effects, including disturbances of the endocrine system, and most commonly leads to adrenal insufficiency, pituitary hypophysitis, and thyroid dysfunction.5 Hypoparathyroidism from immune checkpoint inhibitors is rare. So far, a total of 3 cases, including our case, describing hypoparathyroidism associated with pembrolizumab have been reported. Proposed mechanisms include autoimmune hypoparathyroidism caused by increased T-cell activity against parathyroid tissue or the development of calcium-sensing receptor autoantibodies that inhibit PTH secretion. Vigilance is warranted to facilitate timely recognition and the treatment of immune checkpoint inhibitor-related hypoparathyroidism. It is not clear whether the hypoparathyroidism will be permanent after the discontinuation of pembrolizumab treatment or will resolve over time. Further research and/or case studies are needed to advance our understanding of this relationship and its mechanisms.
Conclusion
Pembrolizumab may result in immune-mediated hypoparathyroidism leading to hypocalcemia. We recommend timely recognition and treatment of this endocrine disturbance.
Author Contributions
I.M. and N.D.K. contributed equally to this work.
Disclosure
The authors have no multiplicity of interest to disclose. | Recovering | ReactionOutcome | CC BY | 33718603 | 20,140,584 | 2021 |
What was the outcome of reaction 'Hypoparathyroidism'? | Pembrolizumab-Associated Hypoparathyroidism: A Single Case Report.
To evaluate a case of pembrolizumab-induced hypoparathyroidism leading to hypocalcemia.
The diagnostic tests performed included calcium and parathyroid hormone level detection and calcium-sensing receptor gene analysis.
A 71-year-old Caucasian man was diagnosed with stage IIIB adenocarcinoma of the lung and received radiation therapy but had no other exposure to radiation. Pembrolizumab 200 mg intravenous every 3 weeks was started 5 years after the initial diagnosis. The patient's corrected calcium level was 9.2 mg/dL (normal, 8.5-10.5 mg/dL) at the start of pembrolizumab therapy. The calcium level after the 13th dose of pembrolizumab was 8.1 mg/dL (normal, 8.5-10.2 mg/dL), leading to endocrinology referral. The patient's parathyroid hormone and corrected calcium levels after the 22nd dose were 4.3 mg/dL (normal, 14-72 pg/mL) and 6.5 mg/dL (normal, 8.5-10.2 mg/dL), respectively. He denied symptoms of latent tetany on presentation while on pembrolizumab for 15 months but complained of fatigue and weakness. The patient had no history of autoimmune diseases or neck injuries. Calcium-sensing receptor gene analysis was negative for genetic mutations. Immunotherapy-mediated hypoparathyroidism was diagnosed. He was treated with daily oral calcium carbonate (2000 mg), calcitriol 0.5 μg, 1 dose of calcium gluconate 2 g intravenous, and 3 doses of calcium chloride 1 g intravenous. His fatigue, weakness, and calcium levels improved with therapy.
Pembrolizumab treatment may have resulted in immune-mediated hypoparathyroidism, leading to hypocalcemia. It is important to report such cases to understand its presentation and timing in relation to pembrolizumab, which further facilitates its timely treatment.
Introduction
Most cases of acquired hypoparathyroidism occur after thyroidectomy or are autoimmune mediated. In adults, most cases occur postsurgically. In a retrospective analysis of adults evaluated in a metabolic bone unit of an endocrinology service in Brazil, over 80% of cases of acquired hypoparathyroidism were postsurgical whereas <20% were autoimmune.1 Drug-induced hypoparathyroidism is extremely rare. Here we report a case of hypoparathyroidism and hypocalcemia with pembrolizumab, an immune checkpoint inhibitor used for the treatment of advanced lung cancer.
Case Report
A 71-year-old Caucasian man was diagnosed with a malignant neoplasm of an unspecified part of the right bronchus or right lung (stage IIIB) and a malignant neoplasm of the upper lobe of the left lung and left bronchus. Right lower and upper lobe wedge resection performed soon after diagnosis demonstrated an adenocarcinoma. Two different courses of radiation treatment were performed, one in the left lung and a subsequent treatment in the right lobe of the patient’s lung. Pembrolizumab, 200 mg intravenous every 3 weeks, was started 5 years after the initial diagnosis of cancer after the patient demonstrated a high (100%) expression of programmed death-ligand 1 in an adenosquamous pathology core sample. The patient’s corrected calcium level was 9.2 mg/dL (normal, 8.5-10.2 mg/dL) at the start of pembrolizumab therapy as shown in Figure 1 and the Table. He had no past medical history or other risk factors for hypocalcemia. The calcium level after the 13th dose of pembrolizumab was 8.1 mg/dL (normal, 8.5-10.2 mg/dL), leading to endocrinology referral.Fig. 1 Trend of corrected calcium levels during pembrolizumab therapy and interventions for correction of low calcium levels.
Table Trend of Parathyroid Hormone, Corrected Calcium Levels, and 25-OH Vitamin D Levels in Relation to Pemobrolizumab Therapy
Date Parathyroid Hormone (14-72 pg/ml) Corrected calcium (8.5-10.5 mg/dL) Pemobrolizumab dose No. 25-OH vitamin D levels (30-100 ng/ml)
Jun 2017 … 9.2 1 …
March 2018 … 8.5 13 …
April 2018 … 8.1 14 …
May 2018 … 7.8 15 and 16 …
July 2018 … 7.5 18 and 19 …
August 2018 … 7.2 20 …
September 2018 20.4 7.7 21 …
October 2018 … 7.9 22 35
December 2018 … 6.9 25 …
January 2019 4.3 6.5 26 …
February 2019 7.2 6.4 27 …
February 2019 … 7.2 28 21
March 2019 … 8.4 29 …
April 2019 … 7.7 30 …
April 2019 … 7.9 31 …
May 2019 … 8.2 32 …
July 2019 2.3 8.7 33 …
November 2019 1.2 7.4 34 96
The patient’s parathyroid hormone (PTH) and corrected calcium levels after the 18th dose of pembrolizumab were 20.4 pg/mL (normal, 14-72 pg/mL) and 7.7 mg/dL (normal, 8.5-10.2 mg/dL), respectively (Table). After the 22nd dose, the levels were 4.3 mg/dL (normal, 14-72 pg/mL) and 6.5 (normal, 8.5-10.2 mg/dL), respectively (Table). At the time of presentation, the patient had been on pembrolizumab for 15 months. He denied symptoms of latent tetany but complained of fatigue and weakness. His PTH and calcium levels remained low as shown in Figure 2. The patient denied a history of autoimmune diseases, neck injuries, or exposure to neck radiation. Calcium-sensing receptor gene analysis (LabCorp Specialty Testing Group) was negative for any kind of genetic mutation. Pembrolizumab-associated hypoparathyroidism was diagnosed. The patient was started on daily oral calcium carbonate (2000 mg), calcitriol 0.5 μg, 1 dose of calcium gluconate 2 g intravenous, and 3 doses of calcium chloride 1 g intravenous. He continued to receive vitamin D 50 000 IU weekly for 6 months, followed by 50 000 IU every 2 weeks. His fatigue, weakness, and calcium levels improved with therapy (Fig. 2).Fig. 2 Trend of parathyroid hormone level during pembrolizumab therapy.
Discussion
To the best of our knowledge, this is the third reported case of pembrolizumab-induced hypoparathyroidism and hypocalcemia. In our case, both PTH and calcium levels started to decrease after the 13th dose of pembrolizumab. In the 2 previously reported cases, the decrease in serum calcium levels was reported after the 15th and the second doses of pembrolizumab, respectively.2,3 Similar events have been recently reported after 1.5 months of concurrent treatment with iplimumab and nivolumab, which are checkpoint inhibitors used for treating melanoma.4 Hypocalcemia and hypoparathyroidism were noted in our patient 1.5 months after starting the therapy. It is therefore clear that the dose-response relationship and the timing of pembrolizumab treatment and hypoparathyroidism can vary on a case-to-case basis, emphasizing the need for frequent monitoring of calcium and parathyroid levels.
Fatigue secondary to hypocalcemia was commonly reported in all previously published cases. Fatigue and weakness, if reported, resolved as the calcium level started to normalize in response to treatment. In our patient, calcium levels were stable on oral calcium supplementation, calcitriol, and 50 000 IU per week of vitamin D supplementation for 6 months, followed by 50 000 IU of vitamin D every 2 weeks. A higher requirement of calcium and calcitriol with failure to normalize calcium or the occurrence of unwanted side effects affected tolerability and long-term adherence. Treatment with recombinant human PTH can be considered. One of the limitations of our case report is that we did not have ionized calcium levels available prior to the start of pembrolizumab treatment; however, with the available data on corrected calcium and PTH levels, the clinical message is very clear.
Pembrolizumab is effective for the treatment of various advanced malignancies such as lung cancer as monotherapy and in combination with other chemotherapy agents.5 However, it has been shown to induce immune-related side effects, including disturbances of the endocrine system, and most commonly leads to adrenal insufficiency, pituitary hypophysitis, and thyroid dysfunction.5 Hypoparathyroidism from immune checkpoint inhibitors is rare. So far, a total of 3 cases, including our case, describing hypoparathyroidism associated with pembrolizumab have been reported. Proposed mechanisms include autoimmune hypoparathyroidism caused by increased T-cell activity against parathyroid tissue or the development of calcium-sensing receptor autoantibodies that inhibit PTH secretion. Vigilance is warranted to facilitate timely recognition and the treatment of immune checkpoint inhibitor-related hypoparathyroidism. It is not clear whether the hypoparathyroidism will be permanent after the discontinuation of pembrolizumab treatment or will resolve over time. Further research and/or case studies are needed to advance our understanding of this relationship and its mechanisms.
Conclusion
Pembrolizumab may result in immune-mediated hypoparathyroidism leading to hypocalcemia. We recommend timely recognition and treatment of this endocrine disturbance.
Author Contributions
I.M. and N.D.K. contributed equally to this work.
Disclosure
The authors have no multiplicity of interest to disclose. | Recovering | ReactionOutcome | CC BY | 33718603 | 20,140,584 | 2021 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Angiosarcoma metastatic'. | Angiosarcoma in arteriovenous fistula after kidney transplantation.
Angiosarcoma is a rare complication of both functioning and nonfunctioning fistulas. It is an aggressive soft tissue sarcoma arising from vascular or lymphatic endothelial cells. We report a case of angiosarcoma from a nonfunctional fistula in a kidney transplantation patient receiving immunosuppressive drugs. The patient had presented with arm pain mimicking a thrombosed arteriovenous fistula.
Angiosarcoma is a rare aggressive soft-tissue sarcoma characterized by rapidly proliferating and extensively infiltrating anaplastic cells, which derives from blood vessels and lining of irregular blood-filled spaces.1 It is a rare complication of arteriovenous fistulas.2 In renal transplant recipients, angiosarcoma is an extremely rare type of cancer, accounting for <1% of all cancers and 2% of localized soft tissue cancer.3 Because of its rarity, most clinicians are unaware of this condition, resulting in a delayed diagnosis and treatment. The institutional review board and ethics committee approved the present study, and a retrospective analysis of the medical records was conducted. The related reported data were also reviewed.
Case report
A 58-year-old man had presented with right arm pain at a brachiocephalic arteriovenous fistula (BCAVF). He had end-stage renal disease and had begun hemodialysis via a right BCAVF 5 years earlier. After 1 year of hemodialysis, he had undergone living related kidney transplantation (KT). He had been receiving immunosuppression therapy with prednisolone, mycophenolate mofetil, and tacrolimus. His baseline creatinine was 0.6 mg/dL. Later, he had complained of right arm pain, and a partially thrombosed AVF was diagnosed. After 6 months of conservative treatment, his pain and weakness had become severe. Also, paresthesia in his right arm and bleeding erosion from the BCAVF were observed. The diagnosis was impending rupture of an AVF aneurysm. He subsequently underwent aneurysmal resection. Preoperative chest radiograph showed a right lung mass. During the intraoperative period, no mass in the intraluminal cephalic vein could be identified (Fig 1). However, his severe arm pain had persisted postoperatively, and a radiograph showed multiple osteolytic lesions in the right humeral shaft, proximal ulna, and radius (Fig 2). Chest computed tomography revealed a 2-cm, lobulated pulmonary nodule in the right upper lung. Therefore, we suspected primary lung cancer with bone metastases. Fifteen days later, he had returned with acute limb ischemia of the right arm. Computed tomography angiography showed no contrast filling in the proximal radial and ulnar arteries at the distal brachial to brachial bifurcation. An eccentric filling defect was seen at the right brachial artery (Fig 3). Intraoperatively, soft tan-brown tissue was recovered by brachial arterial embolectomy. Cytologic examination of the blood clot showed atypical cells. In addition, pathologic examination of the tissue revealed epithelioid angiosarcoma. Examination of a bone biopsy specimen from the right humerus revealed angiosarcoma. At 3 months after aneurysm resection, the patient underwent right shoulder disarticulation. Histopathologic examination of the right shoulder disarticulation revealed angiosarcoma involving the AVF, soft tissue of the cubital fossa, and proximal humerus (Figs 4 and 5). The final diagnosis was angiosarcoma at the AVF with bone and pulmonary metastasis. Therefore, we planned palliative chemotherapy with paclitaxel (Taxol; Bristol-Myers Squibb, New York, NY) weekly. However, the patient developed hemoptysis. He died of massive hemothorax 10 months after symptom onset and 1 month after diagnosis. Because the patient had died and we were unable to contact his relatives, consent could not be obtained from the patient or his family.Fig 1 Thrombosis of cephalic vein found intraoperatively. After removal of the clot, no mass and no abnormality was found in the cephalic vein.
Fig 2 Radiograph of the right elbow showing an osteolytic lesion at the ulna (white arrow).
Fig 3 Computed tomography angiogram showing total thrombosis of the cephalic vein of the brachiocephalic arteriovenous fistula. No contrast filling had occurred in the proximal radial and ulnar arteries at the distal brachial to brachial bifurcation. An eccentric filling defect was seen at the right brachial artery (white arrow).
Fig 4 High-grade pleomorphic epithelioid cells with amphophilic cytoplasm and vasoformative growth.
Fig 5 Positivity to ERG immunohistochemistry and other vascular markers such as FLI1, CD31, and CD34 (data not shown) confirmed the diagnosis of angiosarcoma.
Discussion
According to our review of the reported data4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23 (Table), 23 cases of angiosarcoma occurring from vascular access had been described in 20 English language reports. Most of the patients were men (83.3%), and their mean age was 52.7 ± 15.2 years. The most often used vascular access was the AVF (95.6%). Of the 20 patients, 13 (65%) had undergone deceased donor KT, 5 (25%) had undergone living related KT, and 2 (10%) had not undergone KT. Of the 23 kidney grafts, 20 (93%) were functional. Only five patients had received a second KT.Table Summary of reported cases of angiosarcoma at arteriovenous fistula sites
Investigator Sex; age, years Vascular access AVF status Transplantation IST Presentation Initial Dx Interval to Dx (weeks) Interval (years) from Metastasis before Tx Tx Recurrence Survival
Type Status VA to AS KT to AS
Byers et al,4 1992; Parrott et al,5 1993 M; 36 RC-AVF Functional DDKT Functional Azathioprine, cyclosporine, prednisolone Swelling Thrombotic VA, infection 20 12 8 No Above elbow amputation Postnasal space, lung and axillary lymph node metastasis 1 Year
Wehrli et al,6 1998 M; 64 RC-AVF Thrombosed DDKT Functional Azathioprine, cyclosporine, prednisone Swelling, pain Thrombotic VA 4 10 7 No RT, followed by below elbow amputation Local and lung metastasis Alive 6 months after amputation
Álvarez et al,3 2013 F; 47 AVF Thrombosed DDKT Functional Cyclosporine, MMF Nail-bed splinter lesions, pain, hand retraction Thrombotic or aneurysmal VA 12 14 8 Bone, lung CMT NR 2 Months
Conlon et al,7 1993 M; 40 BC-AVF Thrombosed (ligated aneurysm) First DDKT
Second DDKT Rejection
Functional Treated for rejection
Prednisolone, azathioprine, prednisone Pain and swelling NA 4 40 16
7 No Skeletalization of anterior compartment of arm; RT Local recurrence and lung metastasis 5 Months
Keane et al,8 1993 M; 11 AVF Thrombosed (ligated aneurysm) First DDKT
Second DDKT Rejection
Functional Treated for rejection
Corticosteroids, azathioprine Mass, induration arising from AVF NA 8 NR NR
7 No Wide excision, RT Local recurrence, lung and brain metastasis 8 Months
Bessis et al,9 1998 M; 61 BC-AVF Thrombosed DDKT Functional Azathioprine, cyclosporine, prednisolone Pain, swelling, tenderness Thrombotic or aneurysmal VA NR (>6) 7 5 No Above elbow amputation Lung metastasis 7 Months
Farag et al,10 2005 M; 26 AVF Thrombosed (ligated aneurysm) First DDKT plus PT Second DDKT Not functional
Functional NA
Azathioprine, cyclosporine, prednisolone Mass Thrombotic or aneurysmal VA NR 13 NR
11 Lung CMT Local recurrence 11 Months
Webster et al,11 2011 M; 59 BC-AVF Thrombosed LRKT Functional Rituximab daclizumab, tacrolimus Pain, swelling Thrombotic and infection VA NR 3 2 No Above elbow amputation No Survived
Webster et al,11 2011 F; 41 BC-AVF Thrombosed LRKT Functional Alemtuzumab, prednisolone, tacrolimus Pain Thrombotic VA 4 NR NR NR NR Lung metastasis 4 Months
Webster et al,11 2011 M; 44 BC-AVF NR DDKT Functional Prednisolone, tacrolimus, MMF Pain, mass NR 12 13 8 NR NR Lung metastasis 6 Months
Demey et al,12 2014 M; 80 AVF Functional No KT No KT No KT Pain, swelling Aneurysmal VA 36 4.5 No KT Bone Supportive care NR 1 Month
Roy et al,13 2018 M; 63 AVF Thrombosed (ligated aneurysm) KT Functional MMF, tacrolimus Mass Mass NR NR 6 Lung NR NR 10 Days
Aldaabil et al,14 2016 M; 46 BC-AVF Thrombosed LRKT Functional MMF, tacrolimus, prednisone Pain, swelling NR NR NR 3 Lung, bone CMT, above elbow amputation NR 6 Months
Qureshi et al,15 2010 M; 48 AVF Functional First KT
Second LRKT Not functional
Functional NA
Prednisolone, MMF, rituximab, IVIG, tacrolimus Pain Infection NR 3 NR
2 No “High” above elbow amputation No recurrence or metastasis 9 Months
Costa et al,16 2017 F; 70 BC-AVF Thrombosed (ligated aneurysm) KT Functional Tacrolimus, MMF, prednisolone Dyspnea Pneumonia NR 11 9 Lung NR NR NR
Chanyaputhipong et al,17 2011 M; 57 RC-AVF Thrombosed (ligated aneurysm) DDKT Rejection None Pain, mass Granuloma, infection 8 17 NR NR Wide excision, RT, CMT Local recurrence, lung and bone metastasis 11 Months
Chanyaputhipong et al,17 2011 M; 63 BC-AVF Thrombosed (aneurysm) No KT No KT No KT Pain, mass with bleeding AVF malformation or AS 3 9 No KT No Wide excision, RT, CMT Lung metastasis 9.5 Months
Gale et al,18 2017 M; 44 RC-AVF Functional DDKT NR NR Necrotic fingers Hand ischemia NR 6 NR No Forequarter amputation NR NR
Kakisis et al,19 2019 M; 60 RC-AVF Thrombosed DDKT Functional Prednisolone, cyclosporine, MMF Bleeding from AVF NR NR 11 6 No Above elbow amputation Lung metastasis 6 Months
Jansen et al,20 2013 M; 58 NR NR DDKT Functional Yes (NR) Nodule, pain, hand dysfunction NR NR NR NR NR NR NR NR
Kleman et al,21 2016 M; 71 AVF NR LRKT Functional Alemtuzumab, tacrolimus, MMF Edema, erythema, non-PTH hypercalcemia Cancer NR NR NR No AVF resection, RT, CMT NR NR
Andre et al,22 2012 M; 62 AVF Functional First KT
Second KT Not functional
Functional NR
Prednisone, tacrolimus, MMF Nonhealed wound at AVF removal site NR NR NR NR
NR Lung RT, CMT NR NR
Figueiredo et al,23 2019 F; 55 AVG Thrombosed (ligated aneurysm) KT Functional Tacrolimus, prednisolone, MMF Mass NR 8 NR 15 Lung Palliative care, CMT NR 4 Months
Present case, 2020 M; 58 BC-AVF Thrombosed LRKT Functional Prednisolone, tacrolimus, MMF Pain Thrombotic VA 24 5 4 Lung, bone Shoulder disarticulation, CMT NR 1 Month
AS, Angiosarcoma; AVF, arteriovenous fistula; BC, brachiocephalic; CMT, chemotherapy; DDKT, deceased donor kidney transplantation; Dx, diagnosis; F, female; IST, immunosuppressive therapy; KT, kidney transplantation; LRKT, living related kidney transplantation; M, male; MMF, mycophenolate mofetil; NA, not applicable; NR, not reported; PT, pancreatic transplantation; PTH, parathyroid hormone; RC, radiocephalic; RT, radiation therapy; Tx, treatment; VA, vascular access.
The pathophysiology of angiosarcoma at an AVF after KT remains unclear. The possible mechanisms include the following11: (1) impairment of local immune response caused by the increased lymphatic workload and decreased lymphatic drainage from the increased venous pressure at the AVF site24; (2) turbulent blood flow in the AVF associated with the increased production of reactive oxygen species and matrix metalloproteinase-9, which results in a proliferative response and leads to vascular remodeling25; and (3) a large hypoxic area of the tumor caused by rapid cell proliferation exceeding the capacity of the oxygen supply. These mechanisms induce T-cell suppression, which leads to a diminished immune response.26 Furthermore, KT patients have a three- to fivefold increased risk of any cancer.27 Also, prolonged immunosuppression therapy has been associated with an increased cancer risk.28
According to our review, the most common clinical presentations were nonspecific pain (58.3%), swelling (29.2%), and a mass (29.2%). The initial diagnosis for pain at the AVF included a thrombosed AVF (44%), aneurysm (25%), infection (25%), and cancer (6.3%). Therefore, conservative treatment, including antibiotics and ligation without biopsy, was commonly implemented. These treatments usually led to a delayed diagnosis. The median interval from presentation to diagnosis was 10.8 weeks (range, 4-12 weeks). Thus, a high index of suspicion for angiosarcoma is required.
In our patient who had undergone brachial arterial embolectomy, we sent the clot and tissue for histopathologic examination. Cytologic examination of the blood clot showed atypical cells, and pathologic examination of the tissue showed epithelioid angiosarcoma. Pathologic examination of the shoulder disarticulation showed extensive tumor invasion; however, the origin of the tumor could not be identified. It might be that the origin of angiosarcoma came from the arterial side. Thus, we believe that any patient who has undergone KT and presents with pain or aneurysmal changes should have the blood clot from the aneurysm of an AVF examined cytologically and the tissue examined pathologically.
The tumor histologic features in angiosarcoma have varied widely from well to poorly differentiated. Abnormal, pleomorphic, and malignant endothelium are hallmarks of angiosarcoma. The cell appearance can be rounded, polygonal, fusiform, and, possibly, epithelioid.2 Epithelioid angiosarcoma is a variant of angiosarcoma composed of neoplastic cells with an epithelioid appearance. The most sensitive and specific marker for endothelial differentiation is CD31, which indicates platelet–endothelial cell adhesion molecule expression.29 From our review, all the patients were positive for CD31 expression. For our patient, the pathologic findings showed that the tumor was composed of high-grade pleomorphic epithelioid cells with positivity to ERG using immunohistochemistry and other vascular markers such as FLI1, CD31, and CD34.
Angiosarcoma behaves aggressively, recurs locally, and spreads widely. Moreover, it has a high rate of lymph node and systemic metastases. Our review showed that nearly one half of the patients had had metastases at diagnosis. Also, 85% of the patients had experienced recurrence, with 7.75% in the bone and 76.95% in the lungs. The mean survival period was 24.2 ± 14.4 weeks. Therefore, the tumor-related death rate is considerably high.
The ideal primary treatment is radical surgery with complete resection. However, difficulties ensue with wide excision owing to the tumor size and location, which will result in functional impairment and difficulty in wound reconstruction. Because of the high local recurrence rate, adjuvant radiotherapy has been recommended. Angiosarcoma spreads mainly through hematogenously, and the most common metastatic site has been the lungs. Chemotherapy is the primary option for metastatic angiosarcoma. Biologic therapies, in particular, antiangiogenic therapies, are considered interesting options for angiosarcoma-specific treatment.
Conclusion
Angiosarcoma at vascular access sites can occur in both KT and non-KT patients. Because the presentation is not specific, a high index of suspicion is required to prevent a delayed diagnosis. Angiosarcoma has aggressive behavior and high local recurrence and metastatic rates. The primary treatment of angiosarcoma is radical surgery with complete resection. Chemotherapy is the primary treatment of metastatic angiosarcoma, with radiotherapy for local recurrence prevention. However, despite multimodal treatment, the prognosis of these patients has remained poor.
The authors gratefully acknowledge Nipapan Choonu for the statistical analysis.
Author conflict of interest: none.
The editors and reviewers of this article have no relevant financial relationships to disclose per the Journal policy that requires reviewers to decline review of any manuscript for which they may have a conflict of interest. | MYCOPHENOLATE MOFETIL, PREDNISOLONE, TACROLIMUS | DrugsGivenReaction | CC BY-NC-ND | 33718686 | 19,581,913 | 2021-03 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Angiosarcoma'. | Angiosarcoma in arteriovenous fistula after kidney transplantation.
Angiosarcoma is a rare complication of both functioning and nonfunctioning fistulas. It is an aggressive soft tissue sarcoma arising from vascular or lymphatic endothelial cells. We report a case of angiosarcoma from a nonfunctional fistula in a kidney transplantation patient receiving immunosuppressive drugs. The patient had presented with arm pain mimicking a thrombosed arteriovenous fistula.
Angiosarcoma is a rare aggressive soft-tissue sarcoma characterized by rapidly proliferating and extensively infiltrating anaplastic cells, which derives from blood vessels and lining of irregular blood-filled spaces.1 It is a rare complication of arteriovenous fistulas.2 In renal transplant recipients, angiosarcoma is an extremely rare type of cancer, accounting for <1% of all cancers and 2% of localized soft tissue cancer.3 Because of its rarity, most clinicians are unaware of this condition, resulting in a delayed diagnosis and treatment. The institutional review board and ethics committee approved the present study, and a retrospective analysis of the medical records was conducted. The related reported data were also reviewed.
Case report
A 58-year-old man had presented with right arm pain at a brachiocephalic arteriovenous fistula (BCAVF). He had end-stage renal disease and had begun hemodialysis via a right BCAVF 5 years earlier. After 1 year of hemodialysis, he had undergone living related kidney transplantation (KT). He had been receiving immunosuppression therapy with prednisolone, mycophenolate mofetil, and tacrolimus. His baseline creatinine was 0.6 mg/dL. Later, he had complained of right arm pain, and a partially thrombosed AVF was diagnosed. After 6 months of conservative treatment, his pain and weakness had become severe. Also, paresthesia in his right arm and bleeding erosion from the BCAVF were observed. The diagnosis was impending rupture of an AVF aneurysm. He subsequently underwent aneurysmal resection. Preoperative chest radiograph showed a right lung mass. During the intraoperative period, no mass in the intraluminal cephalic vein could be identified (Fig 1). However, his severe arm pain had persisted postoperatively, and a radiograph showed multiple osteolytic lesions in the right humeral shaft, proximal ulna, and radius (Fig 2). Chest computed tomography revealed a 2-cm, lobulated pulmonary nodule in the right upper lung. Therefore, we suspected primary lung cancer with bone metastases. Fifteen days later, he had returned with acute limb ischemia of the right arm. Computed tomography angiography showed no contrast filling in the proximal radial and ulnar arteries at the distal brachial to brachial bifurcation. An eccentric filling defect was seen at the right brachial artery (Fig 3). Intraoperatively, soft tan-brown tissue was recovered by brachial arterial embolectomy. Cytologic examination of the blood clot showed atypical cells. In addition, pathologic examination of the tissue revealed epithelioid angiosarcoma. Examination of a bone biopsy specimen from the right humerus revealed angiosarcoma. At 3 months after aneurysm resection, the patient underwent right shoulder disarticulation. Histopathologic examination of the right shoulder disarticulation revealed angiosarcoma involving the AVF, soft tissue of the cubital fossa, and proximal humerus (Figs 4 and 5). The final diagnosis was angiosarcoma at the AVF with bone and pulmonary metastasis. Therefore, we planned palliative chemotherapy with paclitaxel (Taxol; Bristol-Myers Squibb, New York, NY) weekly. However, the patient developed hemoptysis. He died of massive hemothorax 10 months after symptom onset and 1 month after diagnosis. Because the patient had died and we were unable to contact his relatives, consent could not be obtained from the patient or his family.Fig 1 Thrombosis of cephalic vein found intraoperatively. After removal of the clot, no mass and no abnormality was found in the cephalic vein.
Fig 2 Radiograph of the right elbow showing an osteolytic lesion at the ulna (white arrow).
Fig 3 Computed tomography angiogram showing total thrombosis of the cephalic vein of the brachiocephalic arteriovenous fistula. No contrast filling had occurred in the proximal radial and ulnar arteries at the distal brachial to brachial bifurcation. An eccentric filling defect was seen at the right brachial artery (white arrow).
Fig 4 High-grade pleomorphic epithelioid cells with amphophilic cytoplasm and vasoformative growth.
Fig 5 Positivity to ERG immunohistochemistry and other vascular markers such as FLI1, CD31, and CD34 (data not shown) confirmed the diagnosis of angiosarcoma.
Discussion
According to our review of the reported data4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23 (Table), 23 cases of angiosarcoma occurring from vascular access had been described in 20 English language reports. Most of the patients were men (83.3%), and their mean age was 52.7 ± 15.2 years. The most often used vascular access was the AVF (95.6%). Of the 20 patients, 13 (65%) had undergone deceased donor KT, 5 (25%) had undergone living related KT, and 2 (10%) had not undergone KT. Of the 23 kidney grafts, 20 (93%) were functional. Only five patients had received a second KT.Table Summary of reported cases of angiosarcoma at arteriovenous fistula sites
Investigator Sex; age, years Vascular access AVF status Transplantation IST Presentation Initial Dx Interval to Dx (weeks) Interval (years) from Metastasis before Tx Tx Recurrence Survival
Type Status VA to AS KT to AS
Byers et al,4 1992; Parrott et al,5 1993 M; 36 RC-AVF Functional DDKT Functional Azathioprine, cyclosporine, prednisolone Swelling Thrombotic VA, infection 20 12 8 No Above elbow amputation Postnasal space, lung and axillary lymph node metastasis 1 Year
Wehrli et al,6 1998 M; 64 RC-AVF Thrombosed DDKT Functional Azathioprine, cyclosporine, prednisone Swelling, pain Thrombotic VA 4 10 7 No RT, followed by below elbow amputation Local and lung metastasis Alive 6 months after amputation
Álvarez et al,3 2013 F; 47 AVF Thrombosed DDKT Functional Cyclosporine, MMF Nail-bed splinter lesions, pain, hand retraction Thrombotic or aneurysmal VA 12 14 8 Bone, lung CMT NR 2 Months
Conlon et al,7 1993 M; 40 BC-AVF Thrombosed (ligated aneurysm) First DDKT
Second DDKT Rejection
Functional Treated for rejection
Prednisolone, azathioprine, prednisone Pain and swelling NA 4 40 16
7 No Skeletalization of anterior compartment of arm; RT Local recurrence and lung metastasis 5 Months
Keane et al,8 1993 M; 11 AVF Thrombosed (ligated aneurysm) First DDKT
Second DDKT Rejection
Functional Treated for rejection
Corticosteroids, azathioprine Mass, induration arising from AVF NA 8 NR NR
7 No Wide excision, RT Local recurrence, lung and brain metastasis 8 Months
Bessis et al,9 1998 M; 61 BC-AVF Thrombosed DDKT Functional Azathioprine, cyclosporine, prednisolone Pain, swelling, tenderness Thrombotic or aneurysmal VA NR (>6) 7 5 No Above elbow amputation Lung metastasis 7 Months
Farag et al,10 2005 M; 26 AVF Thrombosed (ligated aneurysm) First DDKT plus PT Second DDKT Not functional
Functional NA
Azathioprine, cyclosporine, prednisolone Mass Thrombotic or aneurysmal VA NR 13 NR
11 Lung CMT Local recurrence 11 Months
Webster et al,11 2011 M; 59 BC-AVF Thrombosed LRKT Functional Rituximab daclizumab, tacrolimus Pain, swelling Thrombotic and infection VA NR 3 2 No Above elbow amputation No Survived
Webster et al,11 2011 F; 41 BC-AVF Thrombosed LRKT Functional Alemtuzumab, prednisolone, tacrolimus Pain Thrombotic VA 4 NR NR NR NR Lung metastasis 4 Months
Webster et al,11 2011 M; 44 BC-AVF NR DDKT Functional Prednisolone, tacrolimus, MMF Pain, mass NR 12 13 8 NR NR Lung metastasis 6 Months
Demey et al,12 2014 M; 80 AVF Functional No KT No KT No KT Pain, swelling Aneurysmal VA 36 4.5 No KT Bone Supportive care NR 1 Month
Roy et al,13 2018 M; 63 AVF Thrombosed (ligated aneurysm) KT Functional MMF, tacrolimus Mass Mass NR NR 6 Lung NR NR 10 Days
Aldaabil et al,14 2016 M; 46 BC-AVF Thrombosed LRKT Functional MMF, tacrolimus, prednisone Pain, swelling NR NR NR 3 Lung, bone CMT, above elbow amputation NR 6 Months
Qureshi et al,15 2010 M; 48 AVF Functional First KT
Second LRKT Not functional
Functional NA
Prednisolone, MMF, rituximab, IVIG, tacrolimus Pain Infection NR 3 NR
2 No “High” above elbow amputation No recurrence or metastasis 9 Months
Costa et al,16 2017 F; 70 BC-AVF Thrombosed (ligated aneurysm) KT Functional Tacrolimus, MMF, prednisolone Dyspnea Pneumonia NR 11 9 Lung NR NR NR
Chanyaputhipong et al,17 2011 M; 57 RC-AVF Thrombosed (ligated aneurysm) DDKT Rejection None Pain, mass Granuloma, infection 8 17 NR NR Wide excision, RT, CMT Local recurrence, lung and bone metastasis 11 Months
Chanyaputhipong et al,17 2011 M; 63 BC-AVF Thrombosed (aneurysm) No KT No KT No KT Pain, mass with bleeding AVF malformation or AS 3 9 No KT No Wide excision, RT, CMT Lung metastasis 9.5 Months
Gale et al,18 2017 M; 44 RC-AVF Functional DDKT NR NR Necrotic fingers Hand ischemia NR 6 NR No Forequarter amputation NR NR
Kakisis et al,19 2019 M; 60 RC-AVF Thrombosed DDKT Functional Prednisolone, cyclosporine, MMF Bleeding from AVF NR NR 11 6 No Above elbow amputation Lung metastasis 6 Months
Jansen et al,20 2013 M; 58 NR NR DDKT Functional Yes (NR) Nodule, pain, hand dysfunction NR NR NR NR NR NR NR NR
Kleman et al,21 2016 M; 71 AVF NR LRKT Functional Alemtuzumab, tacrolimus, MMF Edema, erythema, non-PTH hypercalcemia Cancer NR NR NR No AVF resection, RT, CMT NR NR
Andre et al,22 2012 M; 62 AVF Functional First KT
Second KT Not functional
Functional NR
Prednisone, tacrolimus, MMF Nonhealed wound at AVF removal site NR NR NR NR
NR Lung RT, CMT NR NR
Figueiredo et al,23 2019 F; 55 AVG Thrombosed (ligated aneurysm) KT Functional Tacrolimus, prednisolone, MMF Mass NR 8 NR 15 Lung Palliative care, CMT NR 4 Months
Present case, 2020 M; 58 BC-AVF Thrombosed LRKT Functional Prednisolone, tacrolimus, MMF Pain Thrombotic VA 24 5 4 Lung, bone Shoulder disarticulation, CMT NR 1 Month
AS, Angiosarcoma; AVF, arteriovenous fistula; BC, brachiocephalic; CMT, chemotherapy; DDKT, deceased donor kidney transplantation; Dx, diagnosis; F, female; IST, immunosuppressive therapy; KT, kidney transplantation; LRKT, living related kidney transplantation; M, male; MMF, mycophenolate mofetil; NA, not applicable; NR, not reported; PT, pancreatic transplantation; PTH, parathyroid hormone; RC, radiocephalic; RT, radiation therapy; Tx, treatment; VA, vascular access.
The pathophysiology of angiosarcoma at an AVF after KT remains unclear. The possible mechanisms include the following11: (1) impairment of local immune response caused by the increased lymphatic workload and decreased lymphatic drainage from the increased venous pressure at the AVF site24; (2) turbulent blood flow in the AVF associated with the increased production of reactive oxygen species and matrix metalloproteinase-9, which results in a proliferative response and leads to vascular remodeling25; and (3) a large hypoxic area of the tumor caused by rapid cell proliferation exceeding the capacity of the oxygen supply. These mechanisms induce T-cell suppression, which leads to a diminished immune response.26 Furthermore, KT patients have a three- to fivefold increased risk of any cancer.27 Also, prolonged immunosuppression therapy has been associated with an increased cancer risk.28
According to our review, the most common clinical presentations were nonspecific pain (58.3%), swelling (29.2%), and a mass (29.2%). The initial diagnosis for pain at the AVF included a thrombosed AVF (44%), aneurysm (25%), infection (25%), and cancer (6.3%). Therefore, conservative treatment, including antibiotics and ligation without biopsy, was commonly implemented. These treatments usually led to a delayed diagnosis. The median interval from presentation to diagnosis was 10.8 weeks (range, 4-12 weeks). Thus, a high index of suspicion for angiosarcoma is required.
In our patient who had undergone brachial arterial embolectomy, we sent the clot and tissue for histopathologic examination. Cytologic examination of the blood clot showed atypical cells, and pathologic examination of the tissue showed epithelioid angiosarcoma. Pathologic examination of the shoulder disarticulation showed extensive tumor invasion; however, the origin of the tumor could not be identified. It might be that the origin of angiosarcoma came from the arterial side. Thus, we believe that any patient who has undergone KT and presents with pain or aneurysmal changes should have the blood clot from the aneurysm of an AVF examined cytologically and the tissue examined pathologically.
The tumor histologic features in angiosarcoma have varied widely from well to poorly differentiated. Abnormal, pleomorphic, and malignant endothelium are hallmarks of angiosarcoma. The cell appearance can be rounded, polygonal, fusiform, and, possibly, epithelioid.2 Epithelioid angiosarcoma is a variant of angiosarcoma composed of neoplastic cells with an epithelioid appearance. The most sensitive and specific marker for endothelial differentiation is CD31, which indicates platelet–endothelial cell adhesion molecule expression.29 From our review, all the patients were positive for CD31 expression. For our patient, the pathologic findings showed that the tumor was composed of high-grade pleomorphic epithelioid cells with positivity to ERG using immunohistochemistry and other vascular markers such as FLI1, CD31, and CD34.
Angiosarcoma behaves aggressively, recurs locally, and spreads widely. Moreover, it has a high rate of lymph node and systemic metastases. Our review showed that nearly one half of the patients had had metastases at diagnosis. Also, 85% of the patients had experienced recurrence, with 7.75% in the bone and 76.95% in the lungs. The mean survival period was 24.2 ± 14.4 weeks. Therefore, the tumor-related death rate is considerably high.
The ideal primary treatment is radical surgery with complete resection. However, difficulties ensue with wide excision owing to the tumor size and location, which will result in functional impairment and difficulty in wound reconstruction. Because of the high local recurrence rate, adjuvant radiotherapy has been recommended. Angiosarcoma spreads mainly through hematogenously, and the most common metastatic site has been the lungs. Chemotherapy is the primary option for metastatic angiosarcoma. Biologic therapies, in particular, antiangiogenic therapies, are considered interesting options for angiosarcoma-specific treatment.
Conclusion
Angiosarcoma at vascular access sites can occur in both KT and non-KT patients. Because the presentation is not specific, a high index of suspicion is required to prevent a delayed diagnosis. Angiosarcoma has aggressive behavior and high local recurrence and metastatic rates. The primary treatment of angiosarcoma is radical surgery with complete resection. Chemotherapy is the primary treatment of metastatic angiosarcoma, with radiotherapy for local recurrence prevention. However, despite multimodal treatment, the prognosis of these patients has remained poor.
The authors gratefully acknowledge Nipapan Choonu for the statistical analysis.
Author conflict of interest: none.
The editors and reviewers of this article have no relevant financial relationships to disclose per the Journal policy that requires reviewers to decline review of any manuscript for which they may have a conflict of interest. | CYCLOSPORINE, MYCOPHENOLIC ACID | DrugsGivenReaction | CC BY-NC-ND | 33718686 | 9,506,483 | 2021-03 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Haemoptysis'. | Angiosarcoma in arteriovenous fistula after kidney transplantation.
Angiosarcoma is a rare complication of both functioning and nonfunctioning fistulas. It is an aggressive soft tissue sarcoma arising from vascular or lymphatic endothelial cells. We report a case of angiosarcoma from a nonfunctional fistula in a kidney transplantation patient receiving immunosuppressive drugs. The patient had presented with arm pain mimicking a thrombosed arteriovenous fistula.
Angiosarcoma is a rare aggressive soft-tissue sarcoma characterized by rapidly proliferating and extensively infiltrating anaplastic cells, which derives from blood vessels and lining of irregular blood-filled spaces.1 It is a rare complication of arteriovenous fistulas.2 In renal transplant recipients, angiosarcoma is an extremely rare type of cancer, accounting for <1% of all cancers and 2% of localized soft tissue cancer.3 Because of its rarity, most clinicians are unaware of this condition, resulting in a delayed diagnosis and treatment. The institutional review board and ethics committee approved the present study, and a retrospective analysis of the medical records was conducted. The related reported data were also reviewed.
Case report
A 58-year-old man had presented with right arm pain at a brachiocephalic arteriovenous fistula (BCAVF). He had end-stage renal disease and had begun hemodialysis via a right BCAVF 5 years earlier. After 1 year of hemodialysis, he had undergone living related kidney transplantation (KT). He had been receiving immunosuppression therapy with prednisolone, mycophenolate mofetil, and tacrolimus. His baseline creatinine was 0.6 mg/dL. Later, he had complained of right arm pain, and a partially thrombosed AVF was diagnosed. After 6 months of conservative treatment, his pain and weakness had become severe. Also, paresthesia in his right arm and bleeding erosion from the BCAVF were observed. The diagnosis was impending rupture of an AVF aneurysm. He subsequently underwent aneurysmal resection. Preoperative chest radiograph showed a right lung mass. During the intraoperative period, no mass in the intraluminal cephalic vein could be identified (Fig 1). However, his severe arm pain had persisted postoperatively, and a radiograph showed multiple osteolytic lesions in the right humeral shaft, proximal ulna, and radius (Fig 2). Chest computed tomography revealed a 2-cm, lobulated pulmonary nodule in the right upper lung. Therefore, we suspected primary lung cancer with bone metastases. Fifteen days later, he had returned with acute limb ischemia of the right arm. Computed tomography angiography showed no contrast filling in the proximal radial and ulnar arteries at the distal brachial to brachial bifurcation. An eccentric filling defect was seen at the right brachial artery (Fig 3). Intraoperatively, soft tan-brown tissue was recovered by brachial arterial embolectomy. Cytologic examination of the blood clot showed atypical cells. In addition, pathologic examination of the tissue revealed epithelioid angiosarcoma. Examination of a bone biopsy specimen from the right humerus revealed angiosarcoma. At 3 months after aneurysm resection, the patient underwent right shoulder disarticulation. Histopathologic examination of the right shoulder disarticulation revealed angiosarcoma involving the AVF, soft tissue of the cubital fossa, and proximal humerus (Figs 4 and 5). The final diagnosis was angiosarcoma at the AVF with bone and pulmonary metastasis. Therefore, we planned palliative chemotherapy with paclitaxel (Taxol; Bristol-Myers Squibb, New York, NY) weekly. However, the patient developed hemoptysis. He died of massive hemothorax 10 months after symptom onset and 1 month after diagnosis. Because the patient had died and we were unable to contact his relatives, consent could not be obtained from the patient or his family.Fig 1 Thrombosis of cephalic vein found intraoperatively. After removal of the clot, no mass and no abnormality was found in the cephalic vein.
Fig 2 Radiograph of the right elbow showing an osteolytic lesion at the ulna (white arrow).
Fig 3 Computed tomography angiogram showing total thrombosis of the cephalic vein of the brachiocephalic arteriovenous fistula. No contrast filling had occurred in the proximal radial and ulnar arteries at the distal brachial to brachial bifurcation. An eccentric filling defect was seen at the right brachial artery (white arrow).
Fig 4 High-grade pleomorphic epithelioid cells with amphophilic cytoplasm and vasoformative growth.
Fig 5 Positivity to ERG immunohistochemistry and other vascular markers such as FLI1, CD31, and CD34 (data not shown) confirmed the diagnosis of angiosarcoma.
Discussion
According to our review of the reported data4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23 (Table), 23 cases of angiosarcoma occurring from vascular access had been described in 20 English language reports. Most of the patients were men (83.3%), and their mean age was 52.7 ± 15.2 years. The most often used vascular access was the AVF (95.6%). Of the 20 patients, 13 (65%) had undergone deceased donor KT, 5 (25%) had undergone living related KT, and 2 (10%) had not undergone KT. Of the 23 kidney grafts, 20 (93%) were functional. Only five patients had received a second KT.Table Summary of reported cases of angiosarcoma at arteriovenous fistula sites
Investigator Sex; age, years Vascular access AVF status Transplantation IST Presentation Initial Dx Interval to Dx (weeks) Interval (years) from Metastasis before Tx Tx Recurrence Survival
Type Status VA to AS KT to AS
Byers et al,4 1992; Parrott et al,5 1993 M; 36 RC-AVF Functional DDKT Functional Azathioprine, cyclosporine, prednisolone Swelling Thrombotic VA, infection 20 12 8 No Above elbow amputation Postnasal space, lung and axillary lymph node metastasis 1 Year
Wehrli et al,6 1998 M; 64 RC-AVF Thrombosed DDKT Functional Azathioprine, cyclosporine, prednisone Swelling, pain Thrombotic VA 4 10 7 No RT, followed by below elbow amputation Local and lung metastasis Alive 6 months after amputation
Álvarez et al,3 2013 F; 47 AVF Thrombosed DDKT Functional Cyclosporine, MMF Nail-bed splinter lesions, pain, hand retraction Thrombotic or aneurysmal VA 12 14 8 Bone, lung CMT NR 2 Months
Conlon et al,7 1993 M; 40 BC-AVF Thrombosed (ligated aneurysm) First DDKT
Second DDKT Rejection
Functional Treated for rejection
Prednisolone, azathioprine, prednisone Pain and swelling NA 4 40 16
7 No Skeletalization of anterior compartment of arm; RT Local recurrence and lung metastasis 5 Months
Keane et al,8 1993 M; 11 AVF Thrombosed (ligated aneurysm) First DDKT
Second DDKT Rejection
Functional Treated for rejection
Corticosteroids, azathioprine Mass, induration arising from AVF NA 8 NR NR
7 No Wide excision, RT Local recurrence, lung and brain metastasis 8 Months
Bessis et al,9 1998 M; 61 BC-AVF Thrombosed DDKT Functional Azathioprine, cyclosporine, prednisolone Pain, swelling, tenderness Thrombotic or aneurysmal VA NR (>6) 7 5 No Above elbow amputation Lung metastasis 7 Months
Farag et al,10 2005 M; 26 AVF Thrombosed (ligated aneurysm) First DDKT plus PT Second DDKT Not functional
Functional NA
Azathioprine, cyclosporine, prednisolone Mass Thrombotic or aneurysmal VA NR 13 NR
11 Lung CMT Local recurrence 11 Months
Webster et al,11 2011 M; 59 BC-AVF Thrombosed LRKT Functional Rituximab daclizumab, tacrolimus Pain, swelling Thrombotic and infection VA NR 3 2 No Above elbow amputation No Survived
Webster et al,11 2011 F; 41 BC-AVF Thrombosed LRKT Functional Alemtuzumab, prednisolone, tacrolimus Pain Thrombotic VA 4 NR NR NR NR Lung metastasis 4 Months
Webster et al,11 2011 M; 44 BC-AVF NR DDKT Functional Prednisolone, tacrolimus, MMF Pain, mass NR 12 13 8 NR NR Lung metastasis 6 Months
Demey et al,12 2014 M; 80 AVF Functional No KT No KT No KT Pain, swelling Aneurysmal VA 36 4.5 No KT Bone Supportive care NR 1 Month
Roy et al,13 2018 M; 63 AVF Thrombosed (ligated aneurysm) KT Functional MMF, tacrolimus Mass Mass NR NR 6 Lung NR NR 10 Days
Aldaabil et al,14 2016 M; 46 BC-AVF Thrombosed LRKT Functional MMF, tacrolimus, prednisone Pain, swelling NR NR NR 3 Lung, bone CMT, above elbow amputation NR 6 Months
Qureshi et al,15 2010 M; 48 AVF Functional First KT
Second LRKT Not functional
Functional NA
Prednisolone, MMF, rituximab, IVIG, tacrolimus Pain Infection NR 3 NR
2 No “High” above elbow amputation No recurrence or metastasis 9 Months
Costa et al,16 2017 F; 70 BC-AVF Thrombosed (ligated aneurysm) KT Functional Tacrolimus, MMF, prednisolone Dyspnea Pneumonia NR 11 9 Lung NR NR NR
Chanyaputhipong et al,17 2011 M; 57 RC-AVF Thrombosed (ligated aneurysm) DDKT Rejection None Pain, mass Granuloma, infection 8 17 NR NR Wide excision, RT, CMT Local recurrence, lung and bone metastasis 11 Months
Chanyaputhipong et al,17 2011 M; 63 BC-AVF Thrombosed (aneurysm) No KT No KT No KT Pain, mass with bleeding AVF malformation or AS 3 9 No KT No Wide excision, RT, CMT Lung metastasis 9.5 Months
Gale et al,18 2017 M; 44 RC-AVF Functional DDKT NR NR Necrotic fingers Hand ischemia NR 6 NR No Forequarter amputation NR NR
Kakisis et al,19 2019 M; 60 RC-AVF Thrombosed DDKT Functional Prednisolone, cyclosporine, MMF Bleeding from AVF NR NR 11 6 No Above elbow amputation Lung metastasis 6 Months
Jansen et al,20 2013 M; 58 NR NR DDKT Functional Yes (NR) Nodule, pain, hand dysfunction NR NR NR NR NR NR NR NR
Kleman et al,21 2016 M; 71 AVF NR LRKT Functional Alemtuzumab, tacrolimus, MMF Edema, erythema, non-PTH hypercalcemia Cancer NR NR NR No AVF resection, RT, CMT NR NR
Andre et al,22 2012 M; 62 AVF Functional First KT
Second KT Not functional
Functional NR
Prednisone, tacrolimus, MMF Nonhealed wound at AVF removal site NR NR NR NR
NR Lung RT, CMT NR NR
Figueiredo et al,23 2019 F; 55 AVG Thrombosed (ligated aneurysm) KT Functional Tacrolimus, prednisolone, MMF Mass NR 8 NR 15 Lung Palliative care, CMT NR 4 Months
Present case, 2020 M; 58 BC-AVF Thrombosed LRKT Functional Prednisolone, tacrolimus, MMF Pain Thrombotic VA 24 5 4 Lung, bone Shoulder disarticulation, CMT NR 1 Month
AS, Angiosarcoma; AVF, arteriovenous fistula; BC, brachiocephalic; CMT, chemotherapy; DDKT, deceased donor kidney transplantation; Dx, diagnosis; F, female; IST, immunosuppressive therapy; KT, kidney transplantation; LRKT, living related kidney transplantation; M, male; MMF, mycophenolate mofetil; NA, not applicable; NR, not reported; PT, pancreatic transplantation; PTH, parathyroid hormone; RC, radiocephalic; RT, radiation therapy; Tx, treatment; VA, vascular access.
The pathophysiology of angiosarcoma at an AVF after KT remains unclear. The possible mechanisms include the following11: (1) impairment of local immune response caused by the increased lymphatic workload and decreased lymphatic drainage from the increased venous pressure at the AVF site24; (2) turbulent blood flow in the AVF associated with the increased production of reactive oxygen species and matrix metalloproteinase-9, which results in a proliferative response and leads to vascular remodeling25; and (3) a large hypoxic area of the tumor caused by rapid cell proliferation exceeding the capacity of the oxygen supply. These mechanisms induce T-cell suppression, which leads to a diminished immune response.26 Furthermore, KT patients have a three- to fivefold increased risk of any cancer.27 Also, prolonged immunosuppression therapy has been associated with an increased cancer risk.28
According to our review, the most common clinical presentations were nonspecific pain (58.3%), swelling (29.2%), and a mass (29.2%). The initial diagnosis for pain at the AVF included a thrombosed AVF (44%), aneurysm (25%), infection (25%), and cancer (6.3%). Therefore, conservative treatment, including antibiotics and ligation without biopsy, was commonly implemented. These treatments usually led to a delayed diagnosis. The median interval from presentation to diagnosis was 10.8 weeks (range, 4-12 weeks). Thus, a high index of suspicion for angiosarcoma is required.
In our patient who had undergone brachial arterial embolectomy, we sent the clot and tissue for histopathologic examination. Cytologic examination of the blood clot showed atypical cells, and pathologic examination of the tissue showed epithelioid angiosarcoma. Pathologic examination of the shoulder disarticulation showed extensive tumor invasion; however, the origin of the tumor could not be identified. It might be that the origin of angiosarcoma came from the arterial side. Thus, we believe that any patient who has undergone KT and presents with pain or aneurysmal changes should have the blood clot from the aneurysm of an AVF examined cytologically and the tissue examined pathologically.
The tumor histologic features in angiosarcoma have varied widely from well to poorly differentiated. Abnormal, pleomorphic, and malignant endothelium are hallmarks of angiosarcoma. The cell appearance can be rounded, polygonal, fusiform, and, possibly, epithelioid.2 Epithelioid angiosarcoma is a variant of angiosarcoma composed of neoplastic cells with an epithelioid appearance. The most sensitive and specific marker for endothelial differentiation is CD31, which indicates platelet–endothelial cell adhesion molecule expression.29 From our review, all the patients were positive for CD31 expression. For our patient, the pathologic findings showed that the tumor was composed of high-grade pleomorphic epithelioid cells with positivity to ERG using immunohistochemistry and other vascular markers such as FLI1, CD31, and CD34.
Angiosarcoma behaves aggressively, recurs locally, and spreads widely. Moreover, it has a high rate of lymph node and systemic metastases. Our review showed that nearly one half of the patients had had metastases at diagnosis. Also, 85% of the patients had experienced recurrence, with 7.75% in the bone and 76.95% in the lungs. The mean survival period was 24.2 ± 14.4 weeks. Therefore, the tumor-related death rate is considerably high.
The ideal primary treatment is radical surgery with complete resection. However, difficulties ensue with wide excision owing to the tumor size and location, which will result in functional impairment and difficulty in wound reconstruction. Because of the high local recurrence rate, adjuvant radiotherapy has been recommended. Angiosarcoma spreads mainly through hematogenously, and the most common metastatic site has been the lungs. Chemotherapy is the primary option for metastatic angiosarcoma. Biologic therapies, in particular, antiangiogenic therapies, are considered interesting options for angiosarcoma-specific treatment.
Conclusion
Angiosarcoma at vascular access sites can occur in both KT and non-KT patients. Because the presentation is not specific, a high index of suspicion is required to prevent a delayed diagnosis. Angiosarcoma has aggressive behavior and high local recurrence and metastatic rates. The primary treatment of angiosarcoma is radical surgery with complete resection. Chemotherapy is the primary treatment of metastatic angiosarcoma, with radiotherapy for local recurrence prevention. However, despite multimodal treatment, the prognosis of these patients has remained poor.
The authors gratefully acknowledge Nipapan Choonu for the statistical analysis.
Author conflict of interest: none.
The editors and reviewers of this article have no relevant financial relationships to disclose per the Journal policy that requires reviewers to decline review of any manuscript for which they may have a conflict of interest. | MYCOPHENOLATE MOFETIL, PREDNISOLONE, TACROLIMUS | DrugsGivenReaction | CC BY-NC-ND | 33718686 | 19,581,913 | 2021-03 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Haemothorax'. | Angiosarcoma in arteriovenous fistula after kidney transplantation.
Angiosarcoma is a rare complication of both functioning and nonfunctioning fistulas. It is an aggressive soft tissue sarcoma arising from vascular or lymphatic endothelial cells. We report a case of angiosarcoma from a nonfunctional fistula in a kidney transplantation patient receiving immunosuppressive drugs. The patient had presented with arm pain mimicking a thrombosed arteriovenous fistula.
Angiosarcoma is a rare aggressive soft-tissue sarcoma characterized by rapidly proliferating and extensively infiltrating anaplastic cells, which derives from blood vessels and lining of irregular blood-filled spaces.1 It is a rare complication of arteriovenous fistulas.2 In renal transplant recipients, angiosarcoma is an extremely rare type of cancer, accounting for <1% of all cancers and 2% of localized soft tissue cancer.3 Because of its rarity, most clinicians are unaware of this condition, resulting in a delayed diagnosis and treatment. The institutional review board and ethics committee approved the present study, and a retrospective analysis of the medical records was conducted. The related reported data were also reviewed.
Case report
A 58-year-old man had presented with right arm pain at a brachiocephalic arteriovenous fistula (BCAVF). He had end-stage renal disease and had begun hemodialysis via a right BCAVF 5 years earlier. After 1 year of hemodialysis, he had undergone living related kidney transplantation (KT). He had been receiving immunosuppression therapy with prednisolone, mycophenolate mofetil, and tacrolimus. His baseline creatinine was 0.6 mg/dL. Later, he had complained of right arm pain, and a partially thrombosed AVF was diagnosed. After 6 months of conservative treatment, his pain and weakness had become severe. Also, paresthesia in his right arm and bleeding erosion from the BCAVF were observed. The diagnosis was impending rupture of an AVF aneurysm. He subsequently underwent aneurysmal resection. Preoperative chest radiograph showed a right lung mass. During the intraoperative period, no mass in the intraluminal cephalic vein could be identified (Fig 1). However, his severe arm pain had persisted postoperatively, and a radiograph showed multiple osteolytic lesions in the right humeral shaft, proximal ulna, and radius (Fig 2). Chest computed tomography revealed a 2-cm, lobulated pulmonary nodule in the right upper lung. Therefore, we suspected primary lung cancer with bone metastases. Fifteen days later, he had returned with acute limb ischemia of the right arm. Computed tomography angiography showed no contrast filling in the proximal radial and ulnar arteries at the distal brachial to brachial bifurcation. An eccentric filling defect was seen at the right brachial artery (Fig 3). Intraoperatively, soft tan-brown tissue was recovered by brachial arterial embolectomy. Cytologic examination of the blood clot showed atypical cells. In addition, pathologic examination of the tissue revealed epithelioid angiosarcoma. Examination of a bone biopsy specimen from the right humerus revealed angiosarcoma. At 3 months after aneurysm resection, the patient underwent right shoulder disarticulation. Histopathologic examination of the right shoulder disarticulation revealed angiosarcoma involving the AVF, soft tissue of the cubital fossa, and proximal humerus (Figs 4 and 5). The final diagnosis was angiosarcoma at the AVF with bone and pulmonary metastasis. Therefore, we planned palliative chemotherapy with paclitaxel (Taxol; Bristol-Myers Squibb, New York, NY) weekly. However, the patient developed hemoptysis. He died of massive hemothorax 10 months after symptom onset and 1 month after diagnosis. Because the patient had died and we were unable to contact his relatives, consent could not be obtained from the patient or his family.Fig 1 Thrombosis of cephalic vein found intraoperatively. After removal of the clot, no mass and no abnormality was found in the cephalic vein.
Fig 2 Radiograph of the right elbow showing an osteolytic lesion at the ulna (white arrow).
Fig 3 Computed tomography angiogram showing total thrombosis of the cephalic vein of the brachiocephalic arteriovenous fistula. No contrast filling had occurred in the proximal radial and ulnar arteries at the distal brachial to brachial bifurcation. An eccentric filling defect was seen at the right brachial artery (white arrow).
Fig 4 High-grade pleomorphic epithelioid cells with amphophilic cytoplasm and vasoformative growth.
Fig 5 Positivity to ERG immunohistochemistry and other vascular markers such as FLI1, CD31, and CD34 (data not shown) confirmed the diagnosis of angiosarcoma.
Discussion
According to our review of the reported data4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23 (Table), 23 cases of angiosarcoma occurring from vascular access had been described in 20 English language reports. Most of the patients were men (83.3%), and their mean age was 52.7 ± 15.2 years. The most often used vascular access was the AVF (95.6%). Of the 20 patients, 13 (65%) had undergone deceased donor KT, 5 (25%) had undergone living related KT, and 2 (10%) had not undergone KT. Of the 23 kidney grafts, 20 (93%) were functional. Only five patients had received a second KT.Table Summary of reported cases of angiosarcoma at arteriovenous fistula sites
Investigator Sex; age, years Vascular access AVF status Transplantation IST Presentation Initial Dx Interval to Dx (weeks) Interval (years) from Metastasis before Tx Tx Recurrence Survival
Type Status VA to AS KT to AS
Byers et al,4 1992; Parrott et al,5 1993 M; 36 RC-AVF Functional DDKT Functional Azathioprine, cyclosporine, prednisolone Swelling Thrombotic VA, infection 20 12 8 No Above elbow amputation Postnasal space, lung and axillary lymph node metastasis 1 Year
Wehrli et al,6 1998 M; 64 RC-AVF Thrombosed DDKT Functional Azathioprine, cyclosporine, prednisone Swelling, pain Thrombotic VA 4 10 7 No RT, followed by below elbow amputation Local and lung metastasis Alive 6 months after amputation
Álvarez et al,3 2013 F; 47 AVF Thrombosed DDKT Functional Cyclosporine, MMF Nail-bed splinter lesions, pain, hand retraction Thrombotic or aneurysmal VA 12 14 8 Bone, lung CMT NR 2 Months
Conlon et al,7 1993 M; 40 BC-AVF Thrombosed (ligated aneurysm) First DDKT
Second DDKT Rejection
Functional Treated for rejection
Prednisolone, azathioprine, prednisone Pain and swelling NA 4 40 16
7 No Skeletalization of anterior compartment of arm; RT Local recurrence and lung metastasis 5 Months
Keane et al,8 1993 M; 11 AVF Thrombosed (ligated aneurysm) First DDKT
Second DDKT Rejection
Functional Treated for rejection
Corticosteroids, azathioprine Mass, induration arising from AVF NA 8 NR NR
7 No Wide excision, RT Local recurrence, lung and brain metastasis 8 Months
Bessis et al,9 1998 M; 61 BC-AVF Thrombosed DDKT Functional Azathioprine, cyclosporine, prednisolone Pain, swelling, tenderness Thrombotic or aneurysmal VA NR (>6) 7 5 No Above elbow amputation Lung metastasis 7 Months
Farag et al,10 2005 M; 26 AVF Thrombosed (ligated aneurysm) First DDKT plus PT Second DDKT Not functional
Functional NA
Azathioprine, cyclosporine, prednisolone Mass Thrombotic or aneurysmal VA NR 13 NR
11 Lung CMT Local recurrence 11 Months
Webster et al,11 2011 M; 59 BC-AVF Thrombosed LRKT Functional Rituximab daclizumab, tacrolimus Pain, swelling Thrombotic and infection VA NR 3 2 No Above elbow amputation No Survived
Webster et al,11 2011 F; 41 BC-AVF Thrombosed LRKT Functional Alemtuzumab, prednisolone, tacrolimus Pain Thrombotic VA 4 NR NR NR NR Lung metastasis 4 Months
Webster et al,11 2011 M; 44 BC-AVF NR DDKT Functional Prednisolone, tacrolimus, MMF Pain, mass NR 12 13 8 NR NR Lung metastasis 6 Months
Demey et al,12 2014 M; 80 AVF Functional No KT No KT No KT Pain, swelling Aneurysmal VA 36 4.5 No KT Bone Supportive care NR 1 Month
Roy et al,13 2018 M; 63 AVF Thrombosed (ligated aneurysm) KT Functional MMF, tacrolimus Mass Mass NR NR 6 Lung NR NR 10 Days
Aldaabil et al,14 2016 M; 46 BC-AVF Thrombosed LRKT Functional MMF, tacrolimus, prednisone Pain, swelling NR NR NR 3 Lung, bone CMT, above elbow amputation NR 6 Months
Qureshi et al,15 2010 M; 48 AVF Functional First KT
Second LRKT Not functional
Functional NA
Prednisolone, MMF, rituximab, IVIG, tacrolimus Pain Infection NR 3 NR
2 No “High” above elbow amputation No recurrence or metastasis 9 Months
Costa et al,16 2017 F; 70 BC-AVF Thrombosed (ligated aneurysm) KT Functional Tacrolimus, MMF, prednisolone Dyspnea Pneumonia NR 11 9 Lung NR NR NR
Chanyaputhipong et al,17 2011 M; 57 RC-AVF Thrombosed (ligated aneurysm) DDKT Rejection None Pain, mass Granuloma, infection 8 17 NR NR Wide excision, RT, CMT Local recurrence, lung and bone metastasis 11 Months
Chanyaputhipong et al,17 2011 M; 63 BC-AVF Thrombosed (aneurysm) No KT No KT No KT Pain, mass with bleeding AVF malformation or AS 3 9 No KT No Wide excision, RT, CMT Lung metastasis 9.5 Months
Gale et al,18 2017 M; 44 RC-AVF Functional DDKT NR NR Necrotic fingers Hand ischemia NR 6 NR No Forequarter amputation NR NR
Kakisis et al,19 2019 M; 60 RC-AVF Thrombosed DDKT Functional Prednisolone, cyclosporine, MMF Bleeding from AVF NR NR 11 6 No Above elbow amputation Lung metastasis 6 Months
Jansen et al,20 2013 M; 58 NR NR DDKT Functional Yes (NR) Nodule, pain, hand dysfunction NR NR NR NR NR NR NR NR
Kleman et al,21 2016 M; 71 AVF NR LRKT Functional Alemtuzumab, tacrolimus, MMF Edema, erythema, non-PTH hypercalcemia Cancer NR NR NR No AVF resection, RT, CMT NR NR
Andre et al,22 2012 M; 62 AVF Functional First KT
Second KT Not functional
Functional NR
Prednisone, tacrolimus, MMF Nonhealed wound at AVF removal site NR NR NR NR
NR Lung RT, CMT NR NR
Figueiredo et al,23 2019 F; 55 AVG Thrombosed (ligated aneurysm) KT Functional Tacrolimus, prednisolone, MMF Mass NR 8 NR 15 Lung Palliative care, CMT NR 4 Months
Present case, 2020 M; 58 BC-AVF Thrombosed LRKT Functional Prednisolone, tacrolimus, MMF Pain Thrombotic VA 24 5 4 Lung, bone Shoulder disarticulation, CMT NR 1 Month
AS, Angiosarcoma; AVF, arteriovenous fistula; BC, brachiocephalic; CMT, chemotherapy; DDKT, deceased donor kidney transplantation; Dx, diagnosis; F, female; IST, immunosuppressive therapy; KT, kidney transplantation; LRKT, living related kidney transplantation; M, male; MMF, mycophenolate mofetil; NA, not applicable; NR, not reported; PT, pancreatic transplantation; PTH, parathyroid hormone; RC, radiocephalic; RT, radiation therapy; Tx, treatment; VA, vascular access.
The pathophysiology of angiosarcoma at an AVF after KT remains unclear. The possible mechanisms include the following11: (1) impairment of local immune response caused by the increased lymphatic workload and decreased lymphatic drainage from the increased venous pressure at the AVF site24; (2) turbulent blood flow in the AVF associated with the increased production of reactive oxygen species and matrix metalloproteinase-9, which results in a proliferative response and leads to vascular remodeling25; and (3) a large hypoxic area of the tumor caused by rapid cell proliferation exceeding the capacity of the oxygen supply. These mechanisms induce T-cell suppression, which leads to a diminished immune response.26 Furthermore, KT patients have a three- to fivefold increased risk of any cancer.27 Also, prolonged immunosuppression therapy has been associated with an increased cancer risk.28
According to our review, the most common clinical presentations were nonspecific pain (58.3%), swelling (29.2%), and a mass (29.2%). The initial diagnosis for pain at the AVF included a thrombosed AVF (44%), aneurysm (25%), infection (25%), and cancer (6.3%). Therefore, conservative treatment, including antibiotics and ligation without biopsy, was commonly implemented. These treatments usually led to a delayed diagnosis. The median interval from presentation to diagnosis was 10.8 weeks (range, 4-12 weeks). Thus, a high index of suspicion for angiosarcoma is required.
In our patient who had undergone brachial arterial embolectomy, we sent the clot and tissue for histopathologic examination. Cytologic examination of the blood clot showed atypical cells, and pathologic examination of the tissue showed epithelioid angiosarcoma. Pathologic examination of the shoulder disarticulation showed extensive tumor invasion; however, the origin of the tumor could not be identified. It might be that the origin of angiosarcoma came from the arterial side. Thus, we believe that any patient who has undergone KT and presents with pain or aneurysmal changes should have the blood clot from the aneurysm of an AVF examined cytologically and the tissue examined pathologically.
The tumor histologic features in angiosarcoma have varied widely from well to poorly differentiated. Abnormal, pleomorphic, and malignant endothelium are hallmarks of angiosarcoma. The cell appearance can be rounded, polygonal, fusiform, and, possibly, epithelioid.2 Epithelioid angiosarcoma is a variant of angiosarcoma composed of neoplastic cells with an epithelioid appearance. The most sensitive and specific marker for endothelial differentiation is CD31, which indicates platelet–endothelial cell adhesion molecule expression.29 From our review, all the patients were positive for CD31 expression. For our patient, the pathologic findings showed that the tumor was composed of high-grade pleomorphic epithelioid cells with positivity to ERG using immunohistochemistry and other vascular markers such as FLI1, CD31, and CD34.
Angiosarcoma behaves aggressively, recurs locally, and spreads widely. Moreover, it has a high rate of lymph node and systemic metastases. Our review showed that nearly one half of the patients had had metastases at diagnosis. Also, 85% of the patients had experienced recurrence, with 7.75% in the bone and 76.95% in the lungs. The mean survival period was 24.2 ± 14.4 weeks. Therefore, the tumor-related death rate is considerably high.
The ideal primary treatment is radical surgery with complete resection. However, difficulties ensue with wide excision owing to the tumor size and location, which will result in functional impairment and difficulty in wound reconstruction. Because of the high local recurrence rate, adjuvant radiotherapy has been recommended. Angiosarcoma spreads mainly through hematogenously, and the most common metastatic site has been the lungs. Chemotherapy is the primary option for metastatic angiosarcoma. Biologic therapies, in particular, antiangiogenic therapies, are considered interesting options for angiosarcoma-specific treatment.
Conclusion
Angiosarcoma at vascular access sites can occur in both KT and non-KT patients. Because the presentation is not specific, a high index of suspicion is required to prevent a delayed diagnosis. Angiosarcoma has aggressive behavior and high local recurrence and metastatic rates. The primary treatment of angiosarcoma is radical surgery with complete resection. Chemotherapy is the primary treatment of metastatic angiosarcoma, with radiotherapy for local recurrence prevention. However, despite multimodal treatment, the prognosis of these patients has remained poor.
The authors gratefully acknowledge Nipapan Choonu for the statistical analysis.
Author conflict of interest: none.
The editors and reviewers of this article have no relevant financial relationships to disclose per the Journal policy that requires reviewers to decline review of any manuscript for which they may have a conflict of interest. | MYCOPHENOLATE MOFETIL, PREDNISOLONE, TACROLIMUS | DrugsGivenReaction | CC BY-NC-ND | 33718686 | 19,581,913 | 2021-03 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Metastases to bone'. | Angiosarcoma in arteriovenous fistula after kidney transplantation.
Angiosarcoma is a rare complication of both functioning and nonfunctioning fistulas. It is an aggressive soft tissue sarcoma arising from vascular or lymphatic endothelial cells. We report a case of angiosarcoma from a nonfunctional fistula in a kidney transplantation patient receiving immunosuppressive drugs. The patient had presented with arm pain mimicking a thrombosed arteriovenous fistula.
Angiosarcoma is a rare aggressive soft-tissue sarcoma characterized by rapidly proliferating and extensively infiltrating anaplastic cells, which derives from blood vessels and lining of irregular blood-filled spaces.1 It is a rare complication of arteriovenous fistulas.2 In renal transplant recipients, angiosarcoma is an extremely rare type of cancer, accounting for <1% of all cancers and 2% of localized soft tissue cancer.3 Because of its rarity, most clinicians are unaware of this condition, resulting in a delayed diagnosis and treatment. The institutional review board and ethics committee approved the present study, and a retrospective analysis of the medical records was conducted. The related reported data were also reviewed.
Case report
A 58-year-old man had presented with right arm pain at a brachiocephalic arteriovenous fistula (BCAVF). He had end-stage renal disease and had begun hemodialysis via a right BCAVF 5 years earlier. After 1 year of hemodialysis, he had undergone living related kidney transplantation (KT). He had been receiving immunosuppression therapy with prednisolone, mycophenolate mofetil, and tacrolimus. His baseline creatinine was 0.6 mg/dL. Later, he had complained of right arm pain, and a partially thrombosed AVF was diagnosed. After 6 months of conservative treatment, his pain and weakness had become severe. Also, paresthesia in his right arm and bleeding erosion from the BCAVF were observed. The diagnosis was impending rupture of an AVF aneurysm. He subsequently underwent aneurysmal resection. Preoperative chest radiograph showed a right lung mass. During the intraoperative period, no mass in the intraluminal cephalic vein could be identified (Fig 1). However, his severe arm pain had persisted postoperatively, and a radiograph showed multiple osteolytic lesions in the right humeral shaft, proximal ulna, and radius (Fig 2). Chest computed tomography revealed a 2-cm, lobulated pulmonary nodule in the right upper lung. Therefore, we suspected primary lung cancer with bone metastases. Fifteen days later, he had returned with acute limb ischemia of the right arm. Computed tomography angiography showed no contrast filling in the proximal radial and ulnar arteries at the distal brachial to brachial bifurcation. An eccentric filling defect was seen at the right brachial artery (Fig 3). Intraoperatively, soft tan-brown tissue was recovered by brachial arterial embolectomy. Cytologic examination of the blood clot showed atypical cells. In addition, pathologic examination of the tissue revealed epithelioid angiosarcoma. Examination of a bone biopsy specimen from the right humerus revealed angiosarcoma. At 3 months after aneurysm resection, the patient underwent right shoulder disarticulation. Histopathologic examination of the right shoulder disarticulation revealed angiosarcoma involving the AVF, soft tissue of the cubital fossa, and proximal humerus (Figs 4 and 5). The final diagnosis was angiosarcoma at the AVF with bone and pulmonary metastasis. Therefore, we planned palliative chemotherapy with paclitaxel (Taxol; Bristol-Myers Squibb, New York, NY) weekly. However, the patient developed hemoptysis. He died of massive hemothorax 10 months after symptom onset and 1 month after diagnosis. Because the patient had died and we were unable to contact his relatives, consent could not be obtained from the patient or his family.Fig 1 Thrombosis of cephalic vein found intraoperatively. After removal of the clot, no mass and no abnormality was found in the cephalic vein.
Fig 2 Radiograph of the right elbow showing an osteolytic lesion at the ulna (white arrow).
Fig 3 Computed tomography angiogram showing total thrombosis of the cephalic vein of the brachiocephalic arteriovenous fistula. No contrast filling had occurred in the proximal radial and ulnar arteries at the distal brachial to brachial bifurcation. An eccentric filling defect was seen at the right brachial artery (white arrow).
Fig 4 High-grade pleomorphic epithelioid cells with amphophilic cytoplasm and vasoformative growth.
Fig 5 Positivity to ERG immunohistochemistry and other vascular markers such as FLI1, CD31, and CD34 (data not shown) confirmed the diagnosis of angiosarcoma.
Discussion
According to our review of the reported data4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23 (Table), 23 cases of angiosarcoma occurring from vascular access had been described in 20 English language reports. Most of the patients were men (83.3%), and their mean age was 52.7 ± 15.2 years. The most often used vascular access was the AVF (95.6%). Of the 20 patients, 13 (65%) had undergone deceased donor KT, 5 (25%) had undergone living related KT, and 2 (10%) had not undergone KT. Of the 23 kidney grafts, 20 (93%) were functional. Only five patients had received a second KT.Table Summary of reported cases of angiosarcoma at arteriovenous fistula sites
Investigator Sex; age, years Vascular access AVF status Transplantation IST Presentation Initial Dx Interval to Dx (weeks) Interval (years) from Metastasis before Tx Tx Recurrence Survival
Type Status VA to AS KT to AS
Byers et al,4 1992; Parrott et al,5 1993 M; 36 RC-AVF Functional DDKT Functional Azathioprine, cyclosporine, prednisolone Swelling Thrombotic VA, infection 20 12 8 No Above elbow amputation Postnasal space, lung and axillary lymph node metastasis 1 Year
Wehrli et al,6 1998 M; 64 RC-AVF Thrombosed DDKT Functional Azathioprine, cyclosporine, prednisone Swelling, pain Thrombotic VA 4 10 7 No RT, followed by below elbow amputation Local and lung metastasis Alive 6 months after amputation
Álvarez et al,3 2013 F; 47 AVF Thrombosed DDKT Functional Cyclosporine, MMF Nail-bed splinter lesions, pain, hand retraction Thrombotic or aneurysmal VA 12 14 8 Bone, lung CMT NR 2 Months
Conlon et al,7 1993 M; 40 BC-AVF Thrombosed (ligated aneurysm) First DDKT
Second DDKT Rejection
Functional Treated for rejection
Prednisolone, azathioprine, prednisone Pain and swelling NA 4 40 16
7 No Skeletalization of anterior compartment of arm; RT Local recurrence and lung metastasis 5 Months
Keane et al,8 1993 M; 11 AVF Thrombosed (ligated aneurysm) First DDKT
Second DDKT Rejection
Functional Treated for rejection
Corticosteroids, azathioprine Mass, induration arising from AVF NA 8 NR NR
7 No Wide excision, RT Local recurrence, lung and brain metastasis 8 Months
Bessis et al,9 1998 M; 61 BC-AVF Thrombosed DDKT Functional Azathioprine, cyclosporine, prednisolone Pain, swelling, tenderness Thrombotic or aneurysmal VA NR (>6) 7 5 No Above elbow amputation Lung metastasis 7 Months
Farag et al,10 2005 M; 26 AVF Thrombosed (ligated aneurysm) First DDKT plus PT Second DDKT Not functional
Functional NA
Azathioprine, cyclosporine, prednisolone Mass Thrombotic or aneurysmal VA NR 13 NR
11 Lung CMT Local recurrence 11 Months
Webster et al,11 2011 M; 59 BC-AVF Thrombosed LRKT Functional Rituximab daclizumab, tacrolimus Pain, swelling Thrombotic and infection VA NR 3 2 No Above elbow amputation No Survived
Webster et al,11 2011 F; 41 BC-AVF Thrombosed LRKT Functional Alemtuzumab, prednisolone, tacrolimus Pain Thrombotic VA 4 NR NR NR NR Lung metastasis 4 Months
Webster et al,11 2011 M; 44 BC-AVF NR DDKT Functional Prednisolone, tacrolimus, MMF Pain, mass NR 12 13 8 NR NR Lung metastasis 6 Months
Demey et al,12 2014 M; 80 AVF Functional No KT No KT No KT Pain, swelling Aneurysmal VA 36 4.5 No KT Bone Supportive care NR 1 Month
Roy et al,13 2018 M; 63 AVF Thrombosed (ligated aneurysm) KT Functional MMF, tacrolimus Mass Mass NR NR 6 Lung NR NR 10 Days
Aldaabil et al,14 2016 M; 46 BC-AVF Thrombosed LRKT Functional MMF, tacrolimus, prednisone Pain, swelling NR NR NR 3 Lung, bone CMT, above elbow amputation NR 6 Months
Qureshi et al,15 2010 M; 48 AVF Functional First KT
Second LRKT Not functional
Functional NA
Prednisolone, MMF, rituximab, IVIG, tacrolimus Pain Infection NR 3 NR
2 No “High” above elbow amputation No recurrence or metastasis 9 Months
Costa et al,16 2017 F; 70 BC-AVF Thrombosed (ligated aneurysm) KT Functional Tacrolimus, MMF, prednisolone Dyspnea Pneumonia NR 11 9 Lung NR NR NR
Chanyaputhipong et al,17 2011 M; 57 RC-AVF Thrombosed (ligated aneurysm) DDKT Rejection None Pain, mass Granuloma, infection 8 17 NR NR Wide excision, RT, CMT Local recurrence, lung and bone metastasis 11 Months
Chanyaputhipong et al,17 2011 M; 63 BC-AVF Thrombosed (aneurysm) No KT No KT No KT Pain, mass with bleeding AVF malformation or AS 3 9 No KT No Wide excision, RT, CMT Lung metastasis 9.5 Months
Gale et al,18 2017 M; 44 RC-AVF Functional DDKT NR NR Necrotic fingers Hand ischemia NR 6 NR No Forequarter amputation NR NR
Kakisis et al,19 2019 M; 60 RC-AVF Thrombosed DDKT Functional Prednisolone, cyclosporine, MMF Bleeding from AVF NR NR 11 6 No Above elbow amputation Lung metastasis 6 Months
Jansen et al,20 2013 M; 58 NR NR DDKT Functional Yes (NR) Nodule, pain, hand dysfunction NR NR NR NR NR NR NR NR
Kleman et al,21 2016 M; 71 AVF NR LRKT Functional Alemtuzumab, tacrolimus, MMF Edema, erythema, non-PTH hypercalcemia Cancer NR NR NR No AVF resection, RT, CMT NR NR
Andre et al,22 2012 M; 62 AVF Functional First KT
Second KT Not functional
Functional NR
Prednisone, tacrolimus, MMF Nonhealed wound at AVF removal site NR NR NR NR
NR Lung RT, CMT NR NR
Figueiredo et al,23 2019 F; 55 AVG Thrombosed (ligated aneurysm) KT Functional Tacrolimus, prednisolone, MMF Mass NR 8 NR 15 Lung Palliative care, CMT NR 4 Months
Present case, 2020 M; 58 BC-AVF Thrombosed LRKT Functional Prednisolone, tacrolimus, MMF Pain Thrombotic VA 24 5 4 Lung, bone Shoulder disarticulation, CMT NR 1 Month
AS, Angiosarcoma; AVF, arteriovenous fistula; BC, brachiocephalic; CMT, chemotherapy; DDKT, deceased donor kidney transplantation; Dx, diagnosis; F, female; IST, immunosuppressive therapy; KT, kidney transplantation; LRKT, living related kidney transplantation; M, male; MMF, mycophenolate mofetil; NA, not applicable; NR, not reported; PT, pancreatic transplantation; PTH, parathyroid hormone; RC, radiocephalic; RT, radiation therapy; Tx, treatment; VA, vascular access.
The pathophysiology of angiosarcoma at an AVF after KT remains unclear. The possible mechanisms include the following11: (1) impairment of local immune response caused by the increased lymphatic workload and decreased lymphatic drainage from the increased venous pressure at the AVF site24; (2) turbulent blood flow in the AVF associated with the increased production of reactive oxygen species and matrix metalloproteinase-9, which results in a proliferative response and leads to vascular remodeling25; and (3) a large hypoxic area of the tumor caused by rapid cell proliferation exceeding the capacity of the oxygen supply. These mechanisms induce T-cell suppression, which leads to a diminished immune response.26 Furthermore, KT patients have a three- to fivefold increased risk of any cancer.27 Also, prolonged immunosuppression therapy has been associated with an increased cancer risk.28
According to our review, the most common clinical presentations were nonspecific pain (58.3%), swelling (29.2%), and a mass (29.2%). The initial diagnosis for pain at the AVF included a thrombosed AVF (44%), aneurysm (25%), infection (25%), and cancer (6.3%). Therefore, conservative treatment, including antibiotics and ligation without biopsy, was commonly implemented. These treatments usually led to a delayed diagnosis. The median interval from presentation to diagnosis was 10.8 weeks (range, 4-12 weeks). Thus, a high index of suspicion for angiosarcoma is required.
In our patient who had undergone brachial arterial embolectomy, we sent the clot and tissue for histopathologic examination. Cytologic examination of the blood clot showed atypical cells, and pathologic examination of the tissue showed epithelioid angiosarcoma. Pathologic examination of the shoulder disarticulation showed extensive tumor invasion; however, the origin of the tumor could not be identified. It might be that the origin of angiosarcoma came from the arterial side. Thus, we believe that any patient who has undergone KT and presents with pain or aneurysmal changes should have the blood clot from the aneurysm of an AVF examined cytologically and the tissue examined pathologically.
The tumor histologic features in angiosarcoma have varied widely from well to poorly differentiated. Abnormal, pleomorphic, and malignant endothelium are hallmarks of angiosarcoma. The cell appearance can be rounded, polygonal, fusiform, and, possibly, epithelioid.2 Epithelioid angiosarcoma is a variant of angiosarcoma composed of neoplastic cells with an epithelioid appearance. The most sensitive and specific marker for endothelial differentiation is CD31, which indicates platelet–endothelial cell adhesion molecule expression.29 From our review, all the patients were positive for CD31 expression. For our patient, the pathologic findings showed that the tumor was composed of high-grade pleomorphic epithelioid cells with positivity to ERG using immunohistochemistry and other vascular markers such as FLI1, CD31, and CD34.
Angiosarcoma behaves aggressively, recurs locally, and spreads widely. Moreover, it has a high rate of lymph node and systemic metastases. Our review showed that nearly one half of the patients had had metastases at diagnosis. Also, 85% of the patients had experienced recurrence, with 7.75% in the bone and 76.95% in the lungs. The mean survival period was 24.2 ± 14.4 weeks. Therefore, the tumor-related death rate is considerably high.
The ideal primary treatment is radical surgery with complete resection. However, difficulties ensue with wide excision owing to the tumor size and location, which will result in functional impairment and difficulty in wound reconstruction. Because of the high local recurrence rate, adjuvant radiotherapy has been recommended. Angiosarcoma spreads mainly through hematogenously, and the most common metastatic site has been the lungs. Chemotherapy is the primary option for metastatic angiosarcoma. Biologic therapies, in particular, antiangiogenic therapies, are considered interesting options for angiosarcoma-specific treatment.
Conclusion
Angiosarcoma at vascular access sites can occur in both KT and non-KT patients. Because the presentation is not specific, a high index of suspicion is required to prevent a delayed diagnosis. Angiosarcoma has aggressive behavior and high local recurrence and metastatic rates. The primary treatment of angiosarcoma is radical surgery with complete resection. Chemotherapy is the primary treatment of metastatic angiosarcoma, with radiotherapy for local recurrence prevention. However, despite multimodal treatment, the prognosis of these patients has remained poor.
The authors gratefully acknowledge Nipapan Choonu for the statistical analysis.
Author conflict of interest: none.
The editors and reviewers of this article have no relevant financial relationships to disclose per the Journal policy that requires reviewers to decline review of any manuscript for which they may have a conflict of interest. | MYCOPHENOLATE MOFETIL, PREDNISOLONE, TACROLIMUS | DrugsGivenReaction | CC BY-NC-ND | 33718686 | 19,581,913 | 2021-03 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Metastases to lung'. | Angiosarcoma in arteriovenous fistula after kidney transplantation.
Angiosarcoma is a rare complication of both functioning and nonfunctioning fistulas. It is an aggressive soft tissue sarcoma arising from vascular or lymphatic endothelial cells. We report a case of angiosarcoma from a nonfunctional fistula in a kidney transplantation patient receiving immunosuppressive drugs. The patient had presented with arm pain mimicking a thrombosed arteriovenous fistula.
Angiosarcoma is a rare aggressive soft-tissue sarcoma characterized by rapidly proliferating and extensively infiltrating anaplastic cells, which derives from blood vessels and lining of irregular blood-filled spaces.1 It is a rare complication of arteriovenous fistulas.2 In renal transplant recipients, angiosarcoma is an extremely rare type of cancer, accounting for <1% of all cancers and 2% of localized soft tissue cancer.3 Because of its rarity, most clinicians are unaware of this condition, resulting in a delayed diagnosis and treatment. The institutional review board and ethics committee approved the present study, and a retrospective analysis of the medical records was conducted. The related reported data were also reviewed.
Case report
A 58-year-old man had presented with right arm pain at a brachiocephalic arteriovenous fistula (BCAVF). He had end-stage renal disease and had begun hemodialysis via a right BCAVF 5 years earlier. After 1 year of hemodialysis, he had undergone living related kidney transplantation (KT). He had been receiving immunosuppression therapy with prednisolone, mycophenolate mofetil, and tacrolimus. His baseline creatinine was 0.6 mg/dL. Later, he had complained of right arm pain, and a partially thrombosed AVF was diagnosed. After 6 months of conservative treatment, his pain and weakness had become severe. Also, paresthesia in his right arm and bleeding erosion from the BCAVF were observed. The diagnosis was impending rupture of an AVF aneurysm. He subsequently underwent aneurysmal resection. Preoperative chest radiograph showed a right lung mass. During the intraoperative period, no mass in the intraluminal cephalic vein could be identified (Fig 1). However, his severe arm pain had persisted postoperatively, and a radiograph showed multiple osteolytic lesions in the right humeral shaft, proximal ulna, and radius (Fig 2). Chest computed tomography revealed a 2-cm, lobulated pulmonary nodule in the right upper lung. Therefore, we suspected primary lung cancer with bone metastases. Fifteen days later, he had returned with acute limb ischemia of the right arm. Computed tomography angiography showed no contrast filling in the proximal radial and ulnar arteries at the distal brachial to brachial bifurcation. An eccentric filling defect was seen at the right brachial artery (Fig 3). Intraoperatively, soft tan-brown tissue was recovered by brachial arterial embolectomy. Cytologic examination of the blood clot showed atypical cells. In addition, pathologic examination of the tissue revealed epithelioid angiosarcoma. Examination of a bone biopsy specimen from the right humerus revealed angiosarcoma. At 3 months after aneurysm resection, the patient underwent right shoulder disarticulation. Histopathologic examination of the right shoulder disarticulation revealed angiosarcoma involving the AVF, soft tissue of the cubital fossa, and proximal humerus (Figs 4 and 5). The final diagnosis was angiosarcoma at the AVF with bone and pulmonary metastasis. Therefore, we planned palliative chemotherapy with paclitaxel (Taxol; Bristol-Myers Squibb, New York, NY) weekly. However, the patient developed hemoptysis. He died of massive hemothorax 10 months after symptom onset and 1 month after diagnosis. Because the patient had died and we were unable to contact his relatives, consent could not be obtained from the patient or his family.Fig 1 Thrombosis of cephalic vein found intraoperatively. After removal of the clot, no mass and no abnormality was found in the cephalic vein.
Fig 2 Radiograph of the right elbow showing an osteolytic lesion at the ulna (white arrow).
Fig 3 Computed tomography angiogram showing total thrombosis of the cephalic vein of the brachiocephalic arteriovenous fistula. No contrast filling had occurred in the proximal radial and ulnar arteries at the distal brachial to brachial bifurcation. An eccentric filling defect was seen at the right brachial artery (white arrow).
Fig 4 High-grade pleomorphic epithelioid cells with amphophilic cytoplasm and vasoformative growth.
Fig 5 Positivity to ERG immunohistochemistry and other vascular markers such as FLI1, CD31, and CD34 (data not shown) confirmed the diagnosis of angiosarcoma.
Discussion
According to our review of the reported data4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23 (Table), 23 cases of angiosarcoma occurring from vascular access had been described in 20 English language reports. Most of the patients were men (83.3%), and their mean age was 52.7 ± 15.2 years. The most often used vascular access was the AVF (95.6%). Of the 20 patients, 13 (65%) had undergone deceased donor KT, 5 (25%) had undergone living related KT, and 2 (10%) had not undergone KT. Of the 23 kidney grafts, 20 (93%) were functional. Only five patients had received a second KT.Table Summary of reported cases of angiosarcoma at arteriovenous fistula sites
Investigator Sex; age, years Vascular access AVF status Transplantation IST Presentation Initial Dx Interval to Dx (weeks) Interval (years) from Metastasis before Tx Tx Recurrence Survival
Type Status VA to AS KT to AS
Byers et al,4 1992; Parrott et al,5 1993 M; 36 RC-AVF Functional DDKT Functional Azathioprine, cyclosporine, prednisolone Swelling Thrombotic VA, infection 20 12 8 No Above elbow amputation Postnasal space, lung and axillary lymph node metastasis 1 Year
Wehrli et al,6 1998 M; 64 RC-AVF Thrombosed DDKT Functional Azathioprine, cyclosporine, prednisone Swelling, pain Thrombotic VA 4 10 7 No RT, followed by below elbow amputation Local and lung metastasis Alive 6 months after amputation
Álvarez et al,3 2013 F; 47 AVF Thrombosed DDKT Functional Cyclosporine, MMF Nail-bed splinter lesions, pain, hand retraction Thrombotic or aneurysmal VA 12 14 8 Bone, lung CMT NR 2 Months
Conlon et al,7 1993 M; 40 BC-AVF Thrombosed (ligated aneurysm) First DDKT
Second DDKT Rejection
Functional Treated for rejection
Prednisolone, azathioprine, prednisone Pain and swelling NA 4 40 16
7 No Skeletalization of anterior compartment of arm; RT Local recurrence and lung metastasis 5 Months
Keane et al,8 1993 M; 11 AVF Thrombosed (ligated aneurysm) First DDKT
Second DDKT Rejection
Functional Treated for rejection
Corticosteroids, azathioprine Mass, induration arising from AVF NA 8 NR NR
7 No Wide excision, RT Local recurrence, lung and brain metastasis 8 Months
Bessis et al,9 1998 M; 61 BC-AVF Thrombosed DDKT Functional Azathioprine, cyclosporine, prednisolone Pain, swelling, tenderness Thrombotic or aneurysmal VA NR (>6) 7 5 No Above elbow amputation Lung metastasis 7 Months
Farag et al,10 2005 M; 26 AVF Thrombosed (ligated aneurysm) First DDKT plus PT Second DDKT Not functional
Functional NA
Azathioprine, cyclosporine, prednisolone Mass Thrombotic or aneurysmal VA NR 13 NR
11 Lung CMT Local recurrence 11 Months
Webster et al,11 2011 M; 59 BC-AVF Thrombosed LRKT Functional Rituximab daclizumab, tacrolimus Pain, swelling Thrombotic and infection VA NR 3 2 No Above elbow amputation No Survived
Webster et al,11 2011 F; 41 BC-AVF Thrombosed LRKT Functional Alemtuzumab, prednisolone, tacrolimus Pain Thrombotic VA 4 NR NR NR NR Lung metastasis 4 Months
Webster et al,11 2011 M; 44 BC-AVF NR DDKT Functional Prednisolone, tacrolimus, MMF Pain, mass NR 12 13 8 NR NR Lung metastasis 6 Months
Demey et al,12 2014 M; 80 AVF Functional No KT No KT No KT Pain, swelling Aneurysmal VA 36 4.5 No KT Bone Supportive care NR 1 Month
Roy et al,13 2018 M; 63 AVF Thrombosed (ligated aneurysm) KT Functional MMF, tacrolimus Mass Mass NR NR 6 Lung NR NR 10 Days
Aldaabil et al,14 2016 M; 46 BC-AVF Thrombosed LRKT Functional MMF, tacrolimus, prednisone Pain, swelling NR NR NR 3 Lung, bone CMT, above elbow amputation NR 6 Months
Qureshi et al,15 2010 M; 48 AVF Functional First KT
Second LRKT Not functional
Functional NA
Prednisolone, MMF, rituximab, IVIG, tacrolimus Pain Infection NR 3 NR
2 No “High” above elbow amputation No recurrence or metastasis 9 Months
Costa et al,16 2017 F; 70 BC-AVF Thrombosed (ligated aneurysm) KT Functional Tacrolimus, MMF, prednisolone Dyspnea Pneumonia NR 11 9 Lung NR NR NR
Chanyaputhipong et al,17 2011 M; 57 RC-AVF Thrombosed (ligated aneurysm) DDKT Rejection None Pain, mass Granuloma, infection 8 17 NR NR Wide excision, RT, CMT Local recurrence, lung and bone metastasis 11 Months
Chanyaputhipong et al,17 2011 M; 63 BC-AVF Thrombosed (aneurysm) No KT No KT No KT Pain, mass with bleeding AVF malformation or AS 3 9 No KT No Wide excision, RT, CMT Lung metastasis 9.5 Months
Gale et al,18 2017 M; 44 RC-AVF Functional DDKT NR NR Necrotic fingers Hand ischemia NR 6 NR No Forequarter amputation NR NR
Kakisis et al,19 2019 M; 60 RC-AVF Thrombosed DDKT Functional Prednisolone, cyclosporine, MMF Bleeding from AVF NR NR 11 6 No Above elbow amputation Lung metastasis 6 Months
Jansen et al,20 2013 M; 58 NR NR DDKT Functional Yes (NR) Nodule, pain, hand dysfunction NR NR NR NR NR NR NR NR
Kleman et al,21 2016 M; 71 AVF NR LRKT Functional Alemtuzumab, tacrolimus, MMF Edema, erythema, non-PTH hypercalcemia Cancer NR NR NR No AVF resection, RT, CMT NR NR
Andre et al,22 2012 M; 62 AVF Functional First KT
Second KT Not functional
Functional NR
Prednisone, tacrolimus, MMF Nonhealed wound at AVF removal site NR NR NR NR
NR Lung RT, CMT NR NR
Figueiredo et al,23 2019 F; 55 AVG Thrombosed (ligated aneurysm) KT Functional Tacrolimus, prednisolone, MMF Mass NR 8 NR 15 Lung Palliative care, CMT NR 4 Months
Present case, 2020 M; 58 BC-AVF Thrombosed LRKT Functional Prednisolone, tacrolimus, MMF Pain Thrombotic VA 24 5 4 Lung, bone Shoulder disarticulation, CMT NR 1 Month
AS, Angiosarcoma; AVF, arteriovenous fistula; BC, brachiocephalic; CMT, chemotherapy; DDKT, deceased donor kidney transplantation; Dx, diagnosis; F, female; IST, immunosuppressive therapy; KT, kidney transplantation; LRKT, living related kidney transplantation; M, male; MMF, mycophenolate mofetil; NA, not applicable; NR, not reported; PT, pancreatic transplantation; PTH, parathyroid hormone; RC, radiocephalic; RT, radiation therapy; Tx, treatment; VA, vascular access.
The pathophysiology of angiosarcoma at an AVF after KT remains unclear. The possible mechanisms include the following11: (1) impairment of local immune response caused by the increased lymphatic workload and decreased lymphatic drainage from the increased venous pressure at the AVF site24; (2) turbulent blood flow in the AVF associated with the increased production of reactive oxygen species and matrix metalloproteinase-9, which results in a proliferative response and leads to vascular remodeling25; and (3) a large hypoxic area of the tumor caused by rapid cell proliferation exceeding the capacity of the oxygen supply. These mechanisms induce T-cell suppression, which leads to a diminished immune response.26 Furthermore, KT patients have a three- to fivefold increased risk of any cancer.27 Also, prolonged immunosuppression therapy has been associated with an increased cancer risk.28
According to our review, the most common clinical presentations were nonspecific pain (58.3%), swelling (29.2%), and a mass (29.2%). The initial diagnosis for pain at the AVF included a thrombosed AVF (44%), aneurysm (25%), infection (25%), and cancer (6.3%). Therefore, conservative treatment, including antibiotics and ligation without biopsy, was commonly implemented. These treatments usually led to a delayed diagnosis. The median interval from presentation to diagnosis was 10.8 weeks (range, 4-12 weeks). Thus, a high index of suspicion for angiosarcoma is required.
In our patient who had undergone brachial arterial embolectomy, we sent the clot and tissue for histopathologic examination. Cytologic examination of the blood clot showed atypical cells, and pathologic examination of the tissue showed epithelioid angiosarcoma. Pathologic examination of the shoulder disarticulation showed extensive tumor invasion; however, the origin of the tumor could not be identified. It might be that the origin of angiosarcoma came from the arterial side. Thus, we believe that any patient who has undergone KT and presents with pain or aneurysmal changes should have the blood clot from the aneurysm of an AVF examined cytologically and the tissue examined pathologically.
The tumor histologic features in angiosarcoma have varied widely from well to poorly differentiated. Abnormal, pleomorphic, and malignant endothelium are hallmarks of angiosarcoma. The cell appearance can be rounded, polygonal, fusiform, and, possibly, epithelioid.2 Epithelioid angiosarcoma is a variant of angiosarcoma composed of neoplastic cells with an epithelioid appearance. The most sensitive and specific marker for endothelial differentiation is CD31, which indicates platelet–endothelial cell adhesion molecule expression.29 From our review, all the patients were positive for CD31 expression. For our patient, the pathologic findings showed that the tumor was composed of high-grade pleomorphic epithelioid cells with positivity to ERG using immunohistochemistry and other vascular markers such as FLI1, CD31, and CD34.
Angiosarcoma behaves aggressively, recurs locally, and spreads widely. Moreover, it has a high rate of lymph node and systemic metastases. Our review showed that nearly one half of the patients had had metastases at diagnosis. Also, 85% of the patients had experienced recurrence, with 7.75% in the bone and 76.95% in the lungs. The mean survival period was 24.2 ± 14.4 weeks. Therefore, the tumor-related death rate is considerably high.
The ideal primary treatment is radical surgery with complete resection. However, difficulties ensue with wide excision owing to the tumor size and location, which will result in functional impairment and difficulty in wound reconstruction. Because of the high local recurrence rate, adjuvant radiotherapy has been recommended. Angiosarcoma spreads mainly through hematogenously, and the most common metastatic site has been the lungs. Chemotherapy is the primary option for metastatic angiosarcoma. Biologic therapies, in particular, antiangiogenic therapies, are considered interesting options for angiosarcoma-specific treatment.
Conclusion
Angiosarcoma at vascular access sites can occur in both KT and non-KT patients. Because the presentation is not specific, a high index of suspicion is required to prevent a delayed diagnosis. Angiosarcoma has aggressive behavior and high local recurrence and metastatic rates. The primary treatment of angiosarcoma is radical surgery with complete resection. Chemotherapy is the primary treatment of metastatic angiosarcoma, with radiotherapy for local recurrence prevention. However, despite multimodal treatment, the prognosis of these patients has remained poor.
The authors gratefully acknowledge Nipapan Choonu for the statistical analysis.
Author conflict of interest: none.
The editors and reviewers of this article have no relevant financial relationships to disclose per the Journal policy that requires reviewers to decline review of any manuscript for which they may have a conflict of interest. | MYCOPHENOLATE MOFETIL, PREDNISOLONE, TACROLIMUS | DrugsGivenReaction | CC BY-NC-ND | 33718686 | 19,581,913 | 2021-03 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Peripheral ischaemia'. | Angiosarcoma in arteriovenous fistula after kidney transplantation.
Angiosarcoma is a rare complication of both functioning and nonfunctioning fistulas. It is an aggressive soft tissue sarcoma arising from vascular or lymphatic endothelial cells. We report a case of angiosarcoma from a nonfunctional fistula in a kidney transplantation patient receiving immunosuppressive drugs. The patient had presented with arm pain mimicking a thrombosed arteriovenous fistula.
Angiosarcoma is a rare aggressive soft-tissue sarcoma characterized by rapidly proliferating and extensively infiltrating anaplastic cells, which derives from blood vessels and lining of irregular blood-filled spaces.1 It is a rare complication of arteriovenous fistulas.2 In renal transplant recipients, angiosarcoma is an extremely rare type of cancer, accounting for <1% of all cancers and 2% of localized soft tissue cancer.3 Because of its rarity, most clinicians are unaware of this condition, resulting in a delayed diagnosis and treatment. The institutional review board and ethics committee approved the present study, and a retrospective analysis of the medical records was conducted. The related reported data were also reviewed.
Case report
A 58-year-old man had presented with right arm pain at a brachiocephalic arteriovenous fistula (BCAVF). He had end-stage renal disease and had begun hemodialysis via a right BCAVF 5 years earlier. After 1 year of hemodialysis, he had undergone living related kidney transplantation (KT). He had been receiving immunosuppression therapy with prednisolone, mycophenolate mofetil, and tacrolimus. His baseline creatinine was 0.6 mg/dL. Later, he had complained of right arm pain, and a partially thrombosed AVF was diagnosed. After 6 months of conservative treatment, his pain and weakness had become severe. Also, paresthesia in his right arm and bleeding erosion from the BCAVF were observed. The diagnosis was impending rupture of an AVF aneurysm. He subsequently underwent aneurysmal resection. Preoperative chest radiograph showed a right lung mass. During the intraoperative period, no mass in the intraluminal cephalic vein could be identified (Fig 1). However, his severe arm pain had persisted postoperatively, and a radiograph showed multiple osteolytic lesions in the right humeral shaft, proximal ulna, and radius (Fig 2). Chest computed tomography revealed a 2-cm, lobulated pulmonary nodule in the right upper lung. Therefore, we suspected primary lung cancer with bone metastases. Fifteen days later, he had returned with acute limb ischemia of the right arm. Computed tomography angiography showed no contrast filling in the proximal radial and ulnar arteries at the distal brachial to brachial bifurcation. An eccentric filling defect was seen at the right brachial artery (Fig 3). Intraoperatively, soft tan-brown tissue was recovered by brachial arterial embolectomy. Cytologic examination of the blood clot showed atypical cells. In addition, pathologic examination of the tissue revealed epithelioid angiosarcoma. Examination of a bone biopsy specimen from the right humerus revealed angiosarcoma. At 3 months after aneurysm resection, the patient underwent right shoulder disarticulation. Histopathologic examination of the right shoulder disarticulation revealed angiosarcoma involving the AVF, soft tissue of the cubital fossa, and proximal humerus (Figs 4 and 5). The final diagnosis was angiosarcoma at the AVF with bone and pulmonary metastasis. Therefore, we planned palliative chemotherapy with paclitaxel (Taxol; Bristol-Myers Squibb, New York, NY) weekly. However, the patient developed hemoptysis. He died of massive hemothorax 10 months after symptom onset and 1 month after diagnosis. Because the patient had died and we were unable to contact his relatives, consent could not be obtained from the patient or his family.Fig 1 Thrombosis of cephalic vein found intraoperatively. After removal of the clot, no mass and no abnormality was found in the cephalic vein.
Fig 2 Radiograph of the right elbow showing an osteolytic lesion at the ulna (white arrow).
Fig 3 Computed tomography angiogram showing total thrombosis of the cephalic vein of the brachiocephalic arteriovenous fistula. No contrast filling had occurred in the proximal radial and ulnar arteries at the distal brachial to brachial bifurcation. An eccentric filling defect was seen at the right brachial artery (white arrow).
Fig 4 High-grade pleomorphic epithelioid cells with amphophilic cytoplasm and vasoformative growth.
Fig 5 Positivity to ERG immunohistochemistry and other vascular markers such as FLI1, CD31, and CD34 (data not shown) confirmed the diagnosis of angiosarcoma.
Discussion
According to our review of the reported data4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23 (Table), 23 cases of angiosarcoma occurring from vascular access had been described in 20 English language reports. Most of the patients were men (83.3%), and their mean age was 52.7 ± 15.2 years. The most often used vascular access was the AVF (95.6%). Of the 20 patients, 13 (65%) had undergone deceased donor KT, 5 (25%) had undergone living related KT, and 2 (10%) had not undergone KT. Of the 23 kidney grafts, 20 (93%) were functional. Only five patients had received a second KT.Table Summary of reported cases of angiosarcoma at arteriovenous fistula sites
Investigator Sex; age, years Vascular access AVF status Transplantation IST Presentation Initial Dx Interval to Dx (weeks) Interval (years) from Metastasis before Tx Tx Recurrence Survival
Type Status VA to AS KT to AS
Byers et al,4 1992; Parrott et al,5 1993 M; 36 RC-AVF Functional DDKT Functional Azathioprine, cyclosporine, prednisolone Swelling Thrombotic VA, infection 20 12 8 No Above elbow amputation Postnasal space, lung and axillary lymph node metastasis 1 Year
Wehrli et al,6 1998 M; 64 RC-AVF Thrombosed DDKT Functional Azathioprine, cyclosporine, prednisone Swelling, pain Thrombotic VA 4 10 7 No RT, followed by below elbow amputation Local and lung metastasis Alive 6 months after amputation
Álvarez et al,3 2013 F; 47 AVF Thrombosed DDKT Functional Cyclosporine, MMF Nail-bed splinter lesions, pain, hand retraction Thrombotic or aneurysmal VA 12 14 8 Bone, lung CMT NR 2 Months
Conlon et al,7 1993 M; 40 BC-AVF Thrombosed (ligated aneurysm) First DDKT
Second DDKT Rejection
Functional Treated for rejection
Prednisolone, azathioprine, prednisone Pain and swelling NA 4 40 16
7 No Skeletalization of anterior compartment of arm; RT Local recurrence and lung metastasis 5 Months
Keane et al,8 1993 M; 11 AVF Thrombosed (ligated aneurysm) First DDKT
Second DDKT Rejection
Functional Treated for rejection
Corticosteroids, azathioprine Mass, induration arising from AVF NA 8 NR NR
7 No Wide excision, RT Local recurrence, lung and brain metastasis 8 Months
Bessis et al,9 1998 M; 61 BC-AVF Thrombosed DDKT Functional Azathioprine, cyclosporine, prednisolone Pain, swelling, tenderness Thrombotic or aneurysmal VA NR (>6) 7 5 No Above elbow amputation Lung metastasis 7 Months
Farag et al,10 2005 M; 26 AVF Thrombosed (ligated aneurysm) First DDKT plus PT Second DDKT Not functional
Functional NA
Azathioprine, cyclosporine, prednisolone Mass Thrombotic or aneurysmal VA NR 13 NR
11 Lung CMT Local recurrence 11 Months
Webster et al,11 2011 M; 59 BC-AVF Thrombosed LRKT Functional Rituximab daclizumab, tacrolimus Pain, swelling Thrombotic and infection VA NR 3 2 No Above elbow amputation No Survived
Webster et al,11 2011 F; 41 BC-AVF Thrombosed LRKT Functional Alemtuzumab, prednisolone, tacrolimus Pain Thrombotic VA 4 NR NR NR NR Lung metastasis 4 Months
Webster et al,11 2011 M; 44 BC-AVF NR DDKT Functional Prednisolone, tacrolimus, MMF Pain, mass NR 12 13 8 NR NR Lung metastasis 6 Months
Demey et al,12 2014 M; 80 AVF Functional No KT No KT No KT Pain, swelling Aneurysmal VA 36 4.5 No KT Bone Supportive care NR 1 Month
Roy et al,13 2018 M; 63 AVF Thrombosed (ligated aneurysm) KT Functional MMF, tacrolimus Mass Mass NR NR 6 Lung NR NR 10 Days
Aldaabil et al,14 2016 M; 46 BC-AVF Thrombosed LRKT Functional MMF, tacrolimus, prednisone Pain, swelling NR NR NR 3 Lung, bone CMT, above elbow amputation NR 6 Months
Qureshi et al,15 2010 M; 48 AVF Functional First KT
Second LRKT Not functional
Functional NA
Prednisolone, MMF, rituximab, IVIG, tacrolimus Pain Infection NR 3 NR
2 No “High” above elbow amputation No recurrence or metastasis 9 Months
Costa et al,16 2017 F; 70 BC-AVF Thrombosed (ligated aneurysm) KT Functional Tacrolimus, MMF, prednisolone Dyspnea Pneumonia NR 11 9 Lung NR NR NR
Chanyaputhipong et al,17 2011 M; 57 RC-AVF Thrombosed (ligated aneurysm) DDKT Rejection None Pain, mass Granuloma, infection 8 17 NR NR Wide excision, RT, CMT Local recurrence, lung and bone metastasis 11 Months
Chanyaputhipong et al,17 2011 M; 63 BC-AVF Thrombosed (aneurysm) No KT No KT No KT Pain, mass with bleeding AVF malformation or AS 3 9 No KT No Wide excision, RT, CMT Lung metastasis 9.5 Months
Gale et al,18 2017 M; 44 RC-AVF Functional DDKT NR NR Necrotic fingers Hand ischemia NR 6 NR No Forequarter amputation NR NR
Kakisis et al,19 2019 M; 60 RC-AVF Thrombosed DDKT Functional Prednisolone, cyclosporine, MMF Bleeding from AVF NR NR 11 6 No Above elbow amputation Lung metastasis 6 Months
Jansen et al,20 2013 M; 58 NR NR DDKT Functional Yes (NR) Nodule, pain, hand dysfunction NR NR NR NR NR NR NR NR
Kleman et al,21 2016 M; 71 AVF NR LRKT Functional Alemtuzumab, tacrolimus, MMF Edema, erythema, non-PTH hypercalcemia Cancer NR NR NR No AVF resection, RT, CMT NR NR
Andre et al,22 2012 M; 62 AVF Functional First KT
Second KT Not functional
Functional NR
Prednisone, tacrolimus, MMF Nonhealed wound at AVF removal site NR NR NR NR
NR Lung RT, CMT NR NR
Figueiredo et al,23 2019 F; 55 AVG Thrombosed (ligated aneurysm) KT Functional Tacrolimus, prednisolone, MMF Mass NR 8 NR 15 Lung Palliative care, CMT NR 4 Months
Present case, 2020 M; 58 BC-AVF Thrombosed LRKT Functional Prednisolone, tacrolimus, MMF Pain Thrombotic VA 24 5 4 Lung, bone Shoulder disarticulation, CMT NR 1 Month
AS, Angiosarcoma; AVF, arteriovenous fistula; BC, brachiocephalic; CMT, chemotherapy; DDKT, deceased donor kidney transplantation; Dx, diagnosis; F, female; IST, immunosuppressive therapy; KT, kidney transplantation; LRKT, living related kidney transplantation; M, male; MMF, mycophenolate mofetil; NA, not applicable; NR, not reported; PT, pancreatic transplantation; PTH, parathyroid hormone; RC, radiocephalic; RT, radiation therapy; Tx, treatment; VA, vascular access.
The pathophysiology of angiosarcoma at an AVF after KT remains unclear. The possible mechanisms include the following11: (1) impairment of local immune response caused by the increased lymphatic workload and decreased lymphatic drainage from the increased venous pressure at the AVF site24; (2) turbulent blood flow in the AVF associated with the increased production of reactive oxygen species and matrix metalloproteinase-9, which results in a proliferative response and leads to vascular remodeling25; and (3) a large hypoxic area of the tumor caused by rapid cell proliferation exceeding the capacity of the oxygen supply. These mechanisms induce T-cell suppression, which leads to a diminished immune response.26 Furthermore, KT patients have a three- to fivefold increased risk of any cancer.27 Also, prolonged immunosuppression therapy has been associated with an increased cancer risk.28
According to our review, the most common clinical presentations were nonspecific pain (58.3%), swelling (29.2%), and a mass (29.2%). The initial diagnosis for pain at the AVF included a thrombosed AVF (44%), aneurysm (25%), infection (25%), and cancer (6.3%). Therefore, conservative treatment, including antibiotics and ligation without biopsy, was commonly implemented. These treatments usually led to a delayed diagnosis. The median interval from presentation to diagnosis was 10.8 weeks (range, 4-12 weeks). Thus, a high index of suspicion for angiosarcoma is required.
In our patient who had undergone brachial arterial embolectomy, we sent the clot and tissue for histopathologic examination. Cytologic examination of the blood clot showed atypical cells, and pathologic examination of the tissue showed epithelioid angiosarcoma. Pathologic examination of the shoulder disarticulation showed extensive tumor invasion; however, the origin of the tumor could not be identified. It might be that the origin of angiosarcoma came from the arterial side. Thus, we believe that any patient who has undergone KT and presents with pain or aneurysmal changes should have the blood clot from the aneurysm of an AVF examined cytologically and the tissue examined pathologically.
The tumor histologic features in angiosarcoma have varied widely from well to poorly differentiated. Abnormal, pleomorphic, and malignant endothelium are hallmarks of angiosarcoma. The cell appearance can be rounded, polygonal, fusiform, and, possibly, epithelioid.2 Epithelioid angiosarcoma is a variant of angiosarcoma composed of neoplastic cells with an epithelioid appearance. The most sensitive and specific marker for endothelial differentiation is CD31, which indicates platelet–endothelial cell adhesion molecule expression.29 From our review, all the patients were positive for CD31 expression. For our patient, the pathologic findings showed that the tumor was composed of high-grade pleomorphic epithelioid cells with positivity to ERG using immunohistochemistry and other vascular markers such as FLI1, CD31, and CD34.
Angiosarcoma behaves aggressively, recurs locally, and spreads widely. Moreover, it has a high rate of lymph node and systemic metastases. Our review showed that nearly one half of the patients had had metastases at diagnosis. Also, 85% of the patients had experienced recurrence, with 7.75% in the bone and 76.95% in the lungs. The mean survival period was 24.2 ± 14.4 weeks. Therefore, the tumor-related death rate is considerably high.
The ideal primary treatment is radical surgery with complete resection. However, difficulties ensue with wide excision owing to the tumor size and location, which will result in functional impairment and difficulty in wound reconstruction. Because of the high local recurrence rate, adjuvant radiotherapy has been recommended. Angiosarcoma spreads mainly through hematogenously, and the most common metastatic site has been the lungs. Chemotherapy is the primary option for metastatic angiosarcoma. Biologic therapies, in particular, antiangiogenic therapies, are considered interesting options for angiosarcoma-specific treatment.
Conclusion
Angiosarcoma at vascular access sites can occur in both KT and non-KT patients. Because the presentation is not specific, a high index of suspicion is required to prevent a delayed diagnosis. Angiosarcoma has aggressive behavior and high local recurrence and metastatic rates. The primary treatment of angiosarcoma is radical surgery with complete resection. Chemotherapy is the primary treatment of metastatic angiosarcoma, with radiotherapy for local recurrence prevention. However, despite multimodal treatment, the prognosis of these patients has remained poor.
The authors gratefully acknowledge Nipapan Choonu for the statistical analysis.
Author conflict of interest: none.
The editors and reviewers of this article have no relevant financial relationships to disclose per the Journal policy that requires reviewers to decline review of any manuscript for which they may have a conflict of interest. | MYCOPHENOLATE MOFETIL, PREDNISOLONE, TACROLIMUS | DrugsGivenReaction | CC BY-NC-ND | 33718686 | 19,581,913 | 2021-03 |
What was the outcome of reaction 'Angiosarcoma metastatic'? | Angiosarcoma in arteriovenous fistula after kidney transplantation.
Angiosarcoma is a rare complication of both functioning and nonfunctioning fistulas. It is an aggressive soft tissue sarcoma arising from vascular or lymphatic endothelial cells. We report a case of angiosarcoma from a nonfunctional fistula in a kidney transplantation patient receiving immunosuppressive drugs. The patient had presented with arm pain mimicking a thrombosed arteriovenous fistula.
Angiosarcoma is a rare aggressive soft-tissue sarcoma characterized by rapidly proliferating and extensively infiltrating anaplastic cells, which derives from blood vessels and lining of irregular blood-filled spaces.1 It is a rare complication of arteriovenous fistulas.2 In renal transplant recipients, angiosarcoma is an extremely rare type of cancer, accounting for <1% of all cancers and 2% of localized soft tissue cancer.3 Because of its rarity, most clinicians are unaware of this condition, resulting in a delayed diagnosis and treatment. The institutional review board and ethics committee approved the present study, and a retrospective analysis of the medical records was conducted. The related reported data were also reviewed.
Case report
A 58-year-old man had presented with right arm pain at a brachiocephalic arteriovenous fistula (BCAVF). He had end-stage renal disease and had begun hemodialysis via a right BCAVF 5 years earlier. After 1 year of hemodialysis, he had undergone living related kidney transplantation (KT). He had been receiving immunosuppression therapy with prednisolone, mycophenolate mofetil, and tacrolimus. His baseline creatinine was 0.6 mg/dL. Later, he had complained of right arm pain, and a partially thrombosed AVF was diagnosed. After 6 months of conservative treatment, his pain and weakness had become severe. Also, paresthesia in his right arm and bleeding erosion from the BCAVF were observed. The diagnosis was impending rupture of an AVF aneurysm. He subsequently underwent aneurysmal resection. Preoperative chest radiograph showed a right lung mass. During the intraoperative period, no mass in the intraluminal cephalic vein could be identified (Fig 1). However, his severe arm pain had persisted postoperatively, and a radiograph showed multiple osteolytic lesions in the right humeral shaft, proximal ulna, and radius (Fig 2). Chest computed tomography revealed a 2-cm, lobulated pulmonary nodule in the right upper lung. Therefore, we suspected primary lung cancer with bone metastases. Fifteen days later, he had returned with acute limb ischemia of the right arm. Computed tomography angiography showed no contrast filling in the proximal radial and ulnar arteries at the distal brachial to brachial bifurcation. An eccentric filling defect was seen at the right brachial artery (Fig 3). Intraoperatively, soft tan-brown tissue was recovered by brachial arterial embolectomy. Cytologic examination of the blood clot showed atypical cells. In addition, pathologic examination of the tissue revealed epithelioid angiosarcoma. Examination of a bone biopsy specimen from the right humerus revealed angiosarcoma. At 3 months after aneurysm resection, the patient underwent right shoulder disarticulation. Histopathologic examination of the right shoulder disarticulation revealed angiosarcoma involving the AVF, soft tissue of the cubital fossa, and proximal humerus (Figs 4 and 5). The final diagnosis was angiosarcoma at the AVF with bone and pulmonary metastasis. Therefore, we planned palliative chemotherapy with paclitaxel (Taxol; Bristol-Myers Squibb, New York, NY) weekly. However, the patient developed hemoptysis. He died of massive hemothorax 10 months after symptom onset and 1 month after diagnosis. Because the patient had died and we were unable to contact his relatives, consent could not be obtained from the patient or his family.Fig 1 Thrombosis of cephalic vein found intraoperatively. After removal of the clot, no mass and no abnormality was found in the cephalic vein.
Fig 2 Radiograph of the right elbow showing an osteolytic lesion at the ulna (white arrow).
Fig 3 Computed tomography angiogram showing total thrombosis of the cephalic vein of the brachiocephalic arteriovenous fistula. No contrast filling had occurred in the proximal radial and ulnar arteries at the distal brachial to brachial bifurcation. An eccentric filling defect was seen at the right brachial artery (white arrow).
Fig 4 High-grade pleomorphic epithelioid cells with amphophilic cytoplasm and vasoformative growth.
Fig 5 Positivity to ERG immunohistochemistry and other vascular markers such as FLI1, CD31, and CD34 (data not shown) confirmed the diagnosis of angiosarcoma.
Discussion
According to our review of the reported data4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23 (Table), 23 cases of angiosarcoma occurring from vascular access had been described in 20 English language reports. Most of the patients were men (83.3%), and their mean age was 52.7 ± 15.2 years. The most often used vascular access was the AVF (95.6%). Of the 20 patients, 13 (65%) had undergone deceased donor KT, 5 (25%) had undergone living related KT, and 2 (10%) had not undergone KT. Of the 23 kidney grafts, 20 (93%) were functional. Only five patients had received a second KT.Table Summary of reported cases of angiosarcoma at arteriovenous fistula sites
Investigator Sex; age, years Vascular access AVF status Transplantation IST Presentation Initial Dx Interval to Dx (weeks) Interval (years) from Metastasis before Tx Tx Recurrence Survival
Type Status VA to AS KT to AS
Byers et al,4 1992; Parrott et al,5 1993 M; 36 RC-AVF Functional DDKT Functional Azathioprine, cyclosporine, prednisolone Swelling Thrombotic VA, infection 20 12 8 No Above elbow amputation Postnasal space, lung and axillary lymph node metastasis 1 Year
Wehrli et al,6 1998 M; 64 RC-AVF Thrombosed DDKT Functional Azathioprine, cyclosporine, prednisone Swelling, pain Thrombotic VA 4 10 7 No RT, followed by below elbow amputation Local and lung metastasis Alive 6 months after amputation
Álvarez et al,3 2013 F; 47 AVF Thrombosed DDKT Functional Cyclosporine, MMF Nail-bed splinter lesions, pain, hand retraction Thrombotic or aneurysmal VA 12 14 8 Bone, lung CMT NR 2 Months
Conlon et al,7 1993 M; 40 BC-AVF Thrombosed (ligated aneurysm) First DDKT
Second DDKT Rejection
Functional Treated for rejection
Prednisolone, azathioprine, prednisone Pain and swelling NA 4 40 16
7 No Skeletalization of anterior compartment of arm; RT Local recurrence and lung metastasis 5 Months
Keane et al,8 1993 M; 11 AVF Thrombosed (ligated aneurysm) First DDKT
Second DDKT Rejection
Functional Treated for rejection
Corticosteroids, azathioprine Mass, induration arising from AVF NA 8 NR NR
7 No Wide excision, RT Local recurrence, lung and brain metastasis 8 Months
Bessis et al,9 1998 M; 61 BC-AVF Thrombosed DDKT Functional Azathioprine, cyclosporine, prednisolone Pain, swelling, tenderness Thrombotic or aneurysmal VA NR (>6) 7 5 No Above elbow amputation Lung metastasis 7 Months
Farag et al,10 2005 M; 26 AVF Thrombosed (ligated aneurysm) First DDKT plus PT Second DDKT Not functional
Functional NA
Azathioprine, cyclosporine, prednisolone Mass Thrombotic or aneurysmal VA NR 13 NR
11 Lung CMT Local recurrence 11 Months
Webster et al,11 2011 M; 59 BC-AVF Thrombosed LRKT Functional Rituximab daclizumab, tacrolimus Pain, swelling Thrombotic and infection VA NR 3 2 No Above elbow amputation No Survived
Webster et al,11 2011 F; 41 BC-AVF Thrombosed LRKT Functional Alemtuzumab, prednisolone, tacrolimus Pain Thrombotic VA 4 NR NR NR NR Lung metastasis 4 Months
Webster et al,11 2011 M; 44 BC-AVF NR DDKT Functional Prednisolone, tacrolimus, MMF Pain, mass NR 12 13 8 NR NR Lung metastasis 6 Months
Demey et al,12 2014 M; 80 AVF Functional No KT No KT No KT Pain, swelling Aneurysmal VA 36 4.5 No KT Bone Supportive care NR 1 Month
Roy et al,13 2018 M; 63 AVF Thrombosed (ligated aneurysm) KT Functional MMF, tacrolimus Mass Mass NR NR 6 Lung NR NR 10 Days
Aldaabil et al,14 2016 M; 46 BC-AVF Thrombosed LRKT Functional MMF, tacrolimus, prednisone Pain, swelling NR NR NR 3 Lung, bone CMT, above elbow amputation NR 6 Months
Qureshi et al,15 2010 M; 48 AVF Functional First KT
Second LRKT Not functional
Functional NA
Prednisolone, MMF, rituximab, IVIG, tacrolimus Pain Infection NR 3 NR
2 No “High” above elbow amputation No recurrence or metastasis 9 Months
Costa et al,16 2017 F; 70 BC-AVF Thrombosed (ligated aneurysm) KT Functional Tacrolimus, MMF, prednisolone Dyspnea Pneumonia NR 11 9 Lung NR NR NR
Chanyaputhipong et al,17 2011 M; 57 RC-AVF Thrombosed (ligated aneurysm) DDKT Rejection None Pain, mass Granuloma, infection 8 17 NR NR Wide excision, RT, CMT Local recurrence, lung and bone metastasis 11 Months
Chanyaputhipong et al,17 2011 M; 63 BC-AVF Thrombosed (aneurysm) No KT No KT No KT Pain, mass with bleeding AVF malformation or AS 3 9 No KT No Wide excision, RT, CMT Lung metastasis 9.5 Months
Gale et al,18 2017 M; 44 RC-AVF Functional DDKT NR NR Necrotic fingers Hand ischemia NR 6 NR No Forequarter amputation NR NR
Kakisis et al,19 2019 M; 60 RC-AVF Thrombosed DDKT Functional Prednisolone, cyclosporine, MMF Bleeding from AVF NR NR 11 6 No Above elbow amputation Lung metastasis 6 Months
Jansen et al,20 2013 M; 58 NR NR DDKT Functional Yes (NR) Nodule, pain, hand dysfunction NR NR NR NR NR NR NR NR
Kleman et al,21 2016 M; 71 AVF NR LRKT Functional Alemtuzumab, tacrolimus, MMF Edema, erythema, non-PTH hypercalcemia Cancer NR NR NR No AVF resection, RT, CMT NR NR
Andre et al,22 2012 M; 62 AVF Functional First KT
Second KT Not functional
Functional NR
Prednisone, tacrolimus, MMF Nonhealed wound at AVF removal site NR NR NR NR
NR Lung RT, CMT NR NR
Figueiredo et al,23 2019 F; 55 AVG Thrombosed (ligated aneurysm) KT Functional Tacrolimus, prednisolone, MMF Mass NR 8 NR 15 Lung Palliative care, CMT NR 4 Months
Present case, 2020 M; 58 BC-AVF Thrombosed LRKT Functional Prednisolone, tacrolimus, MMF Pain Thrombotic VA 24 5 4 Lung, bone Shoulder disarticulation, CMT NR 1 Month
AS, Angiosarcoma; AVF, arteriovenous fistula; BC, brachiocephalic; CMT, chemotherapy; DDKT, deceased donor kidney transplantation; Dx, diagnosis; F, female; IST, immunosuppressive therapy; KT, kidney transplantation; LRKT, living related kidney transplantation; M, male; MMF, mycophenolate mofetil; NA, not applicable; NR, not reported; PT, pancreatic transplantation; PTH, parathyroid hormone; RC, radiocephalic; RT, radiation therapy; Tx, treatment; VA, vascular access.
The pathophysiology of angiosarcoma at an AVF after KT remains unclear. The possible mechanisms include the following11: (1) impairment of local immune response caused by the increased lymphatic workload and decreased lymphatic drainage from the increased venous pressure at the AVF site24; (2) turbulent blood flow in the AVF associated with the increased production of reactive oxygen species and matrix metalloproteinase-9, which results in a proliferative response and leads to vascular remodeling25; and (3) a large hypoxic area of the tumor caused by rapid cell proliferation exceeding the capacity of the oxygen supply. These mechanisms induce T-cell suppression, which leads to a diminished immune response.26 Furthermore, KT patients have a three- to fivefold increased risk of any cancer.27 Also, prolonged immunosuppression therapy has been associated with an increased cancer risk.28
According to our review, the most common clinical presentations were nonspecific pain (58.3%), swelling (29.2%), and a mass (29.2%). The initial diagnosis for pain at the AVF included a thrombosed AVF (44%), aneurysm (25%), infection (25%), and cancer (6.3%). Therefore, conservative treatment, including antibiotics and ligation without biopsy, was commonly implemented. These treatments usually led to a delayed diagnosis. The median interval from presentation to diagnosis was 10.8 weeks (range, 4-12 weeks). Thus, a high index of suspicion for angiosarcoma is required.
In our patient who had undergone brachial arterial embolectomy, we sent the clot and tissue for histopathologic examination. Cytologic examination of the blood clot showed atypical cells, and pathologic examination of the tissue showed epithelioid angiosarcoma. Pathologic examination of the shoulder disarticulation showed extensive tumor invasion; however, the origin of the tumor could not be identified. It might be that the origin of angiosarcoma came from the arterial side. Thus, we believe that any patient who has undergone KT and presents with pain or aneurysmal changes should have the blood clot from the aneurysm of an AVF examined cytologically and the tissue examined pathologically.
The tumor histologic features in angiosarcoma have varied widely from well to poorly differentiated. Abnormal, pleomorphic, and malignant endothelium are hallmarks of angiosarcoma. The cell appearance can be rounded, polygonal, fusiform, and, possibly, epithelioid.2 Epithelioid angiosarcoma is a variant of angiosarcoma composed of neoplastic cells with an epithelioid appearance. The most sensitive and specific marker for endothelial differentiation is CD31, which indicates platelet–endothelial cell adhesion molecule expression.29 From our review, all the patients were positive for CD31 expression. For our patient, the pathologic findings showed that the tumor was composed of high-grade pleomorphic epithelioid cells with positivity to ERG using immunohistochemistry and other vascular markers such as FLI1, CD31, and CD34.
Angiosarcoma behaves aggressively, recurs locally, and spreads widely. Moreover, it has a high rate of lymph node and systemic metastases. Our review showed that nearly one half of the patients had had metastases at diagnosis. Also, 85% of the patients had experienced recurrence, with 7.75% in the bone and 76.95% in the lungs. The mean survival period was 24.2 ± 14.4 weeks. Therefore, the tumor-related death rate is considerably high.
The ideal primary treatment is radical surgery with complete resection. However, difficulties ensue with wide excision owing to the tumor size and location, which will result in functional impairment and difficulty in wound reconstruction. Because of the high local recurrence rate, adjuvant radiotherapy has been recommended. Angiosarcoma spreads mainly through hematogenously, and the most common metastatic site has been the lungs. Chemotherapy is the primary option for metastatic angiosarcoma. Biologic therapies, in particular, antiangiogenic therapies, are considered interesting options for angiosarcoma-specific treatment.
Conclusion
Angiosarcoma at vascular access sites can occur in both KT and non-KT patients. Because the presentation is not specific, a high index of suspicion is required to prevent a delayed diagnosis. Angiosarcoma has aggressive behavior and high local recurrence and metastatic rates. The primary treatment of angiosarcoma is radical surgery with complete resection. Chemotherapy is the primary treatment of metastatic angiosarcoma, with radiotherapy for local recurrence prevention. However, despite multimodal treatment, the prognosis of these patients has remained poor.
The authors gratefully acknowledge Nipapan Choonu for the statistical analysis.
Author conflict of interest: none.
The editors and reviewers of this article have no relevant financial relationships to disclose per the Journal policy that requires reviewers to decline review of any manuscript for which they may have a conflict of interest. | Fatal | ReactionOutcome | CC BY-NC-ND | 33718686 | 19,581,913 | 2021-03 |
What was the outcome of reaction 'Angiosarcoma'? | Angiosarcoma in arteriovenous fistula after kidney transplantation.
Angiosarcoma is a rare complication of both functioning and nonfunctioning fistulas. It is an aggressive soft tissue sarcoma arising from vascular or lymphatic endothelial cells. We report a case of angiosarcoma from a nonfunctional fistula in a kidney transplantation patient receiving immunosuppressive drugs. The patient had presented with arm pain mimicking a thrombosed arteriovenous fistula.
Angiosarcoma is a rare aggressive soft-tissue sarcoma characterized by rapidly proliferating and extensively infiltrating anaplastic cells, which derives from blood vessels and lining of irregular blood-filled spaces.1 It is a rare complication of arteriovenous fistulas.2 In renal transplant recipients, angiosarcoma is an extremely rare type of cancer, accounting for <1% of all cancers and 2% of localized soft tissue cancer.3 Because of its rarity, most clinicians are unaware of this condition, resulting in a delayed diagnosis and treatment. The institutional review board and ethics committee approved the present study, and a retrospective analysis of the medical records was conducted. The related reported data were also reviewed.
Case report
A 58-year-old man had presented with right arm pain at a brachiocephalic arteriovenous fistula (BCAVF). He had end-stage renal disease and had begun hemodialysis via a right BCAVF 5 years earlier. After 1 year of hemodialysis, he had undergone living related kidney transplantation (KT). He had been receiving immunosuppression therapy with prednisolone, mycophenolate mofetil, and tacrolimus. His baseline creatinine was 0.6 mg/dL. Later, he had complained of right arm pain, and a partially thrombosed AVF was diagnosed. After 6 months of conservative treatment, his pain and weakness had become severe. Also, paresthesia in his right arm and bleeding erosion from the BCAVF were observed. The diagnosis was impending rupture of an AVF aneurysm. He subsequently underwent aneurysmal resection. Preoperative chest radiograph showed a right lung mass. During the intraoperative period, no mass in the intraluminal cephalic vein could be identified (Fig 1). However, his severe arm pain had persisted postoperatively, and a radiograph showed multiple osteolytic lesions in the right humeral shaft, proximal ulna, and radius (Fig 2). Chest computed tomography revealed a 2-cm, lobulated pulmonary nodule in the right upper lung. Therefore, we suspected primary lung cancer with bone metastases. Fifteen days later, he had returned with acute limb ischemia of the right arm. Computed tomography angiography showed no contrast filling in the proximal radial and ulnar arteries at the distal brachial to brachial bifurcation. An eccentric filling defect was seen at the right brachial artery (Fig 3). Intraoperatively, soft tan-brown tissue was recovered by brachial arterial embolectomy. Cytologic examination of the blood clot showed atypical cells. In addition, pathologic examination of the tissue revealed epithelioid angiosarcoma. Examination of a bone biopsy specimen from the right humerus revealed angiosarcoma. At 3 months after aneurysm resection, the patient underwent right shoulder disarticulation. Histopathologic examination of the right shoulder disarticulation revealed angiosarcoma involving the AVF, soft tissue of the cubital fossa, and proximal humerus (Figs 4 and 5). The final diagnosis was angiosarcoma at the AVF with bone and pulmonary metastasis. Therefore, we planned palliative chemotherapy with paclitaxel (Taxol; Bristol-Myers Squibb, New York, NY) weekly. However, the patient developed hemoptysis. He died of massive hemothorax 10 months after symptom onset and 1 month after diagnosis. Because the patient had died and we were unable to contact his relatives, consent could not be obtained from the patient or his family.Fig 1 Thrombosis of cephalic vein found intraoperatively. After removal of the clot, no mass and no abnormality was found in the cephalic vein.
Fig 2 Radiograph of the right elbow showing an osteolytic lesion at the ulna (white arrow).
Fig 3 Computed tomography angiogram showing total thrombosis of the cephalic vein of the brachiocephalic arteriovenous fistula. No contrast filling had occurred in the proximal radial and ulnar arteries at the distal brachial to brachial bifurcation. An eccentric filling defect was seen at the right brachial artery (white arrow).
Fig 4 High-grade pleomorphic epithelioid cells with amphophilic cytoplasm and vasoformative growth.
Fig 5 Positivity to ERG immunohistochemistry and other vascular markers such as FLI1, CD31, and CD34 (data not shown) confirmed the diagnosis of angiosarcoma.
Discussion
According to our review of the reported data4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23 (Table), 23 cases of angiosarcoma occurring from vascular access had been described in 20 English language reports. Most of the patients were men (83.3%), and their mean age was 52.7 ± 15.2 years. The most often used vascular access was the AVF (95.6%). Of the 20 patients, 13 (65%) had undergone deceased donor KT, 5 (25%) had undergone living related KT, and 2 (10%) had not undergone KT. Of the 23 kidney grafts, 20 (93%) were functional. Only five patients had received a second KT.Table Summary of reported cases of angiosarcoma at arteriovenous fistula sites
Investigator Sex; age, years Vascular access AVF status Transplantation IST Presentation Initial Dx Interval to Dx (weeks) Interval (years) from Metastasis before Tx Tx Recurrence Survival
Type Status VA to AS KT to AS
Byers et al,4 1992; Parrott et al,5 1993 M; 36 RC-AVF Functional DDKT Functional Azathioprine, cyclosporine, prednisolone Swelling Thrombotic VA, infection 20 12 8 No Above elbow amputation Postnasal space, lung and axillary lymph node metastasis 1 Year
Wehrli et al,6 1998 M; 64 RC-AVF Thrombosed DDKT Functional Azathioprine, cyclosporine, prednisone Swelling, pain Thrombotic VA 4 10 7 No RT, followed by below elbow amputation Local and lung metastasis Alive 6 months after amputation
Álvarez et al,3 2013 F; 47 AVF Thrombosed DDKT Functional Cyclosporine, MMF Nail-bed splinter lesions, pain, hand retraction Thrombotic or aneurysmal VA 12 14 8 Bone, lung CMT NR 2 Months
Conlon et al,7 1993 M; 40 BC-AVF Thrombosed (ligated aneurysm) First DDKT
Second DDKT Rejection
Functional Treated for rejection
Prednisolone, azathioprine, prednisone Pain and swelling NA 4 40 16
7 No Skeletalization of anterior compartment of arm; RT Local recurrence and lung metastasis 5 Months
Keane et al,8 1993 M; 11 AVF Thrombosed (ligated aneurysm) First DDKT
Second DDKT Rejection
Functional Treated for rejection
Corticosteroids, azathioprine Mass, induration arising from AVF NA 8 NR NR
7 No Wide excision, RT Local recurrence, lung and brain metastasis 8 Months
Bessis et al,9 1998 M; 61 BC-AVF Thrombosed DDKT Functional Azathioprine, cyclosporine, prednisolone Pain, swelling, tenderness Thrombotic or aneurysmal VA NR (>6) 7 5 No Above elbow amputation Lung metastasis 7 Months
Farag et al,10 2005 M; 26 AVF Thrombosed (ligated aneurysm) First DDKT plus PT Second DDKT Not functional
Functional NA
Azathioprine, cyclosporine, prednisolone Mass Thrombotic or aneurysmal VA NR 13 NR
11 Lung CMT Local recurrence 11 Months
Webster et al,11 2011 M; 59 BC-AVF Thrombosed LRKT Functional Rituximab daclizumab, tacrolimus Pain, swelling Thrombotic and infection VA NR 3 2 No Above elbow amputation No Survived
Webster et al,11 2011 F; 41 BC-AVF Thrombosed LRKT Functional Alemtuzumab, prednisolone, tacrolimus Pain Thrombotic VA 4 NR NR NR NR Lung metastasis 4 Months
Webster et al,11 2011 M; 44 BC-AVF NR DDKT Functional Prednisolone, tacrolimus, MMF Pain, mass NR 12 13 8 NR NR Lung metastasis 6 Months
Demey et al,12 2014 M; 80 AVF Functional No KT No KT No KT Pain, swelling Aneurysmal VA 36 4.5 No KT Bone Supportive care NR 1 Month
Roy et al,13 2018 M; 63 AVF Thrombosed (ligated aneurysm) KT Functional MMF, tacrolimus Mass Mass NR NR 6 Lung NR NR 10 Days
Aldaabil et al,14 2016 M; 46 BC-AVF Thrombosed LRKT Functional MMF, tacrolimus, prednisone Pain, swelling NR NR NR 3 Lung, bone CMT, above elbow amputation NR 6 Months
Qureshi et al,15 2010 M; 48 AVF Functional First KT
Second LRKT Not functional
Functional NA
Prednisolone, MMF, rituximab, IVIG, tacrolimus Pain Infection NR 3 NR
2 No “High” above elbow amputation No recurrence or metastasis 9 Months
Costa et al,16 2017 F; 70 BC-AVF Thrombosed (ligated aneurysm) KT Functional Tacrolimus, MMF, prednisolone Dyspnea Pneumonia NR 11 9 Lung NR NR NR
Chanyaputhipong et al,17 2011 M; 57 RC-AVF Thrombosed (ligated aneurysm) DDKT Rejection None Pain, mass Granuloma, infection 8 17 NR NR Wide excision, RT, CMT Local recurrence, lung and bone metastasis 11 Months
Chanyaputhipong et al,17 2011 M; 63 BC-AVF Thrombosed (aneurysm) No KT No KT No KT Pain, mass with bleeding AVF malformation or AS 3 9 No KT No Wide excision, RT, CMT Lung metastasis 9.5 Months
Gale et al,18 2017 M; 44 RC-AVF Functional DDKT NR NR Necrotic fingers Hand ischemia NR 6 NR No Forequarter amputation NR NR
Kakisis et al,19 2019 M; 60 RC-AVF Thrombosed DDKT Functional Prednisolone, cyclosporine, MMF Bleeding from AVF NR NR 11 6 No Above elbow amputation Lung metastasis 6 Months
Jansen et al,20 2013 M; 58 NR NR DDKT Functional Yes (NR) Nodule, pain, hand dysfunction NR NR NR NR NR NR NR NR
Kleman et al,21 2016 M; 71 AVF NR LRKT Functional Alemtuzumab, tacrolimus, MMF Edema, erythema, non-PTH hypercalcemia Cancer NR NR NR No AVF resection, RT, CMT NR NR
Andre et al,22 2012 M; 62 AVF Functional First KT
Second KT Not functional
Functional NR
Prednisone, tacrolimus, MMF Nonhealed wound at AVF removal site NR NR NR NR
NR Lung RT, CMT NR NR
Figueiredo et al,23 2019 F; 55 AVG Thrombosed (ligated aneurysm) KT Functional Tacrolimus, prednisolone, MMF Mass NR 8 NR 15 Lung Palliative care, CMT NR 4 Months
Present case, 2020 M; 58 BC-AVF Thrombosed LRKT Functional Prednisolone, tacrolimus, MMF Pain Thrombotic VA 24 5 4 Lung, bone Shoulder disarticulation, CMT NR 1 Month
AS, Angiosarcoma; AVF, arteriovenous fistula; BC, brachiocephalic; CMT, chemotherapy; DDKT, deceased donor kidney transplantation; Dx, diagnosis; F, female; IST, immunosuppressive therapy; KT, kidney transplantation; LRKT, living related kidney transplantation; M, male; MMF, mycophenolate mofetil; NA, not applicable; NR, not reported; PT, pancreatic transplantation; PTH, parathyroid hormone; RC, radiocephalic; RT, radiation therapy; Tx, treatment; VA, vascular access.
The pathophysiology of angiosarcoma at an AVF after KT remains unclear. The possible mechanisms include the following11: (1) impairment of local immune response caused by the increased lymphatic workload and decreased lymphatic drainage from the increased venous pressure at the AVF site24; (2) turbulent blood flow in the AVF associated with the increased production of reactive oxygen species and matrix metalloproteinase-9, which results in a proliferative response and leads to vascular remodeling25; and (3) a large hypoxic area of the tumor caused by rapid cell proliferation exceeding the capacity of the oxygen supply. These mechanisms induce T-cell suppression, which leads to a diminished immune response.26 Furthermore, KT patients have a three- to fivefold increased risk of any cancer.27 Also, prolonged immunosuppression therapy has been associated with an increased cancer risk.28
According to our review, the most common clinical presentations were nonspecific pain (58.3%), swelling (29.2%), and a mass (29.2%). The initial diagnosis for pain at the AVF included a thrombosed AVF (44%), aneurysm (25%), infection (25%), and cancer (6.3%). Therefore, conservative treatment, including antibiotics and ligation without biopsy, was commonly implemented. These treatments usually led to a delayed diagnosis. The median interval from presentation to diagnosis was 10.8 weeks (range, 4-12 weeks). Thus, a high index of suspicion for angiosarcoma is required.
In our patient who had undergone brachial arterial embolectomy, we sent the clot and tissue for histopathologic examination. Cytologic examination of the blood clot showed atypical cells, and pathologic examination of the tissue showed epithelioid angiosarcoma. Pathologic examination of the shoulder disarticulation showed extensive tumor invasion; however, the origin of the tumor could not be identified. It might be that the origin of angiosarcoma came from the arterial side. Thus, we believe that any patient who has undergone KT and presents with pain or aneurysmal changes should have the blood clot from the aneurysm of an AVF examined cytologically and the tissue examined pathologically.
The tumor histologic features in angiosarcoma have varied widely from well to poorly differentiated. Abnormal, pleomorphic, and malignant endothelium are hallmarks of angiosarcoma. The cell appearance can be rounded, polygonal, fusiform, and, possibly, epithelioid.2 Epithelioid angiosarcoma is a variant of angiosarcoma composed of neoplastic cells with an epithelioid appearance. The most sensitive and specific marker for endothelial differentiation is CD31, which indicates platelet–endothelial cell adhesion molecule expression.29 From our review, all the patients were positive for CD31 expression. For our patient, the pathologic findings showed that the tumor was composed of high-grade pleomorphic epithelioid cells with positivity to ERG using immunohistochemistry and other vascular markers such as FLI1, CD31, and CD34.
Angiosarcoma behaves aggressively, recurs locally, and spreads widely. Moreover, it has a high rate of lymph node and systemic metastases. Our review showed that nearly one half of the patients had had metastases at diagnosis. Also, 85% of the patients had experienced recurrence, with 7.75% in the bone and 76.95% in the lungs. The mean survival period was 24.2 ± 14.4 weeks. Therefore, the tumor-related death rate is considerably high.
The ideal primary treatment is radical surgery with complete resection. However, difficulties ensue with wide excision owing to the tumor size and location, which will result in functional impairment and difficulty in wound reconstruction. Because of the high local recurrence rate, adjuvant radiotherapy has been recommended. Angiosarcoma spreads mainly through hematogenously, and the most common metastatic site has been the lungs. Chemotherapy is the primary option for metastatic angiosarcoma. Biologic therapies, in particular, antiangiogenic therapies, are considered interesting options for angiosarcoma-specific treatment.
Conclusion
Angiosarcoma at vascular access sites can occur in both KT and non-KT patients. Because the presentation is not specific, a high index of suspicion is required to prevent a delayed diagnosis. Angiosarcoma has aggressive behavior and high local recurrence and metastatic rates. The primary treatment of angiosarcoma is radical surgery with complete resection. Chemotherapy is the primary treatment of metastatic angiosarcoma, with radiotherapy for local recurrence prevention. However, despite multimodal treatment, the prognosis of these patients has remained poor.
The authors gratefully acknowledge Nipapan Choonu for the statistical analysis.
Author conflict of interest: none.
The editors and reviewers of this article have no relevant financial relationships to disclose per the Journal policy that requires reviewers to decline review of any manuscript for which they may have a conflict of interest. | Fatal | ReactionOutcome | CC BY-NC-ND | 33718686 | 9,506,483 | 2021-03 |
What was the outcome of reaction 'Haemothorax'? | Angiosarcoma in arteriovenous fistula after kidney transplantation.
Angiosarcoma is a rare complication of both functioning and nonfunctioning fistulas. It is an aggressive soft tissue sarcoma arising from vascular or lymphatic endothelial cells. We report a case of angiosarcoma from a nonfunctional fistula in a kidney transplantation patient receiving immunosuppressive drugs. The patient had presented with arm pain mimicking a thrombosed arteriovenous fistula.
Angiosarcoma is a rare aggressive soft-tissue sarcoma characterized by rapidly proliferating and extensively infiltrating anaplastic cells, which derives from blood vessels and lining of irregular blood-filled spaces.1 It is a rare complication of arteriovenous fistulas.2 In renal transplant recipients, angiosarcoma is an extremely rare type of cancer, accounting for <1% of all cancers and 2% of localized soft tissue cancer.3 Because of its rarity, most clinicians are unaware of this condition, resulting in a delayed diagnosis and treatment. The institutional review board and ethics committee approved the present study, and a retrospective analysis of the medical records was conducted. The related reported data were also reviewed.
Case report
A 58-year-old man had presented with right arm pain at a brachiocephalic arteriovenous fistula (BCAVF). He had end-stage renal disease and had begun hemodialysis via a right BCAVF 5 years earlier. After 1 year of hemodialysis, he had undergone living related kidney transplantation (KT). He had been receiving immunosuppression therapy with prednisolone, mycophenolate mofetil, and tacrolimus. His baseline creatinine was 0.6 mg/dL. Later, he had complained of right arm pain, and a partially thrombosed AVF was diagnosed. After 6 months of conservative treatment, his pain and weakness had become severe. Also, paresthesia in his right arm and bleeding erosion from the BCAVF were observed. The diagnosis was impending rupture of an AVF aneurysm. He subsequently underwent aneurysmal resection. Preoperative chest radiograph showed a right lung mass. During the intraoperative period, no mass in the intraluminal cephalic vein could be identified (Fig 1). However, his severe arm pain had persisted postoperatively, and a radiograph showed multiple osteolytic lesions in the right humeral shaft, proximal ulna, and radius (Fig 2). Chest computed tomography revealed a 2-cm, lobulated pulmonary nodule in the right upper lung. Therefore, we suspected primary lung cancer with bone metastases. Fifteen days later, he had returned with acute limb ischemia of the right arm. Computed tomography angiography showed no contrast filling in the proximal radial and ulnar arteries at the distal brachial to brachial bifurcation. An eccentric filling defect was seen at the right brachial artery (Fig 3). Intraoperatively, soft tan-brown tissue was recovered by brachial arterial embolectomy. Cytologic examination of the blood clot showed atypical cells. In addition, pathologic examination of the tissue revealed epithelioid angiosarcoma. Examination of a bone biopsy specimen from the right humerus revealed angiosarcoma. At 3 months after aneurysm resection, the patient underwent right shoulder disarticulation. Histopathologic examination of the right shoulder disarticulation revealed angiosarcoma involving the AVF, soft tissue of the cubital fossa, and proximal humerus (Figs 4 and 5). The final diagnosis was angiosarcoma at the AVF with bone and pulmonary metastasis. Therefore, we planned palliative chemotherapy with paclitaxel (Taxol; Bristol-Myers Squibb, New York, NY) weekly. However, the patient developed hemoptysis. He died of massive hemothorax 10 months after symptom onset and 1 month after diagnosis. Because the patient had died and we were unable to contact his relatives, consent could not be obtained from the patient or his family.Fig 1 Thrombosis of cephalic vein found intraoperatively. After removal of the clot, no mass and no abnormality was found in the cephalic vein.
Fig 2 Radiograph of the right elbow showing an osteolytic lesion at the ulna (white arrow).
Fig 3 Computed tomography angiogram showing total thrombosis of the cephalic vein of the brachiocephalic arteriovenous fistula. No contrast filling had occurred in the proximal radial and ulnar arteries at the distal brachial to brachial bifurcation. An eccentric filling defect was seen at the right brachial artery (white arrow).
Fig 4 High-grade pleomorphic epithelioid cells with amphophilic cytoplasm and vasoformative growth.
Fig 5 Positivity to ERG immunohistochemistry and other vascular markers such as FLI1, CD31, and CD34 (data not shown) confirmed the diagnosis of angiosarcoma.
Discussion
According to our review of the reported data4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23 (Table), 23 cases of angiosarcoma occurring from vascular access had been described in 20 English language reports. Most of the patients were men (83.3%), and their mean age was 52.7 ± 15.2 years. The most often used vascular access was the AVF (95.6%). Of the 20 patients, 13 (65%) had undergone deceased donor KT, 5 (25%) had undergone living related KT, and 2 (10%) had not undergone KT. Of the 23 kidney grafts, 20 (93%) were functional. Only five patients had received a second KT.Table Summary of reported cases of angiosarcoma at arteriovenous fistula sites
Investigator Sex; age, years Vascular access AVF status Transplantation IST Presentation Initial Dx Interval to Dx (weeks) Interval (years) from Metastasis before Tx Tx Recurrence Survival
Type Status VA to AS KT to AS
Byers et al,4 1992; Parrott et al,5 1993 M; 36 RC-AVF Functional DDKT Functional Azathioprine, cyclosporine, prednisolone Swelling Thrombotic VA, infection 20 12 8 No Above elbow amputation Postnasal space, lung and axillary lymph node metastasis 1 Year
Wehrli et al,6 1998 M; 64 RC-AVF Thrombosed DDKT Functional Azathioprine, cyclosporine, prednisone Swelling, pain Thrombotic VA 4 10 7 No RT, followed by below elbow amputation Local and lung metastasis Alive 6 months after amputation
Álvarez et al,3 2013 F; 47 AVF Thrombosed DDKT Functional Cyclosporine, MMF Nail-bed splinter lesions, pain, hand retraction Thrombotic or aneurysmal VA 12 14 8 Bone, lung CMT NR 2 Months
Conlon et al,7 1993 M; 40 BC-AVF Thrombosed (ligated aneurysm) First DDKT
Second DDKT Rejection
Functional Treated for rejection
Prednisolone, azathioprine, prednisone Pain and swelling NA 4 40 16
7 No Skeletalization of anterior compartment of arm; RT Local recurrence and lung metastasis 5 Months
Keane et al,8 1993 M; 11 AVF Thrombosed (ligated aneurysm) First DDKT
Second DDKT Rejection
Functional Treated for rejection
Corticosteroids, azathioprine Mass, induration arising from AVF NA 8 NR NR
7 No Wide excision, RT Local recurrence, lung and brain metastasis 8 Months
Bessis et al,9 1998 M; 61 BC-AVF Thrombosed DDKT Functional Azathioprine, cyclosporine, prednisolone Pain, swelling, tenderness Thrombotic or aneurysmal VA NR (>6) 7 5 No Above elbow amputation Lung metastasis 7 Months
Farag et al,10 2005 M; 26 AVF Thrombosed (ligated aneurysm) First DDKT plus PT Second DDKT Not functional
Functional NA
Azathioprine, cyclosporine, prednisolone Mass Thrombotic or aneurysmal VA NR 13 NR
11 Lung CMT Local recurrence 11 Months
Webster et al,11 2011 M; 59 BC-AVF Thrombosed LRKT Functional Rituximab daclizumab, tacrolimus Pain, swelling Thrombotic and infection VA NR 3 2 No Above elbow amputation No Survived
Webster et al,11 2011 F; 41 BC-AVF Thrombosed LRKT Functional Alemtuzumab, prednisolone, tacrolimus Pain Thrombotic VA 4 NR NR NR NR Lung metastasis 4 Months
Webster et al,11 2011 M; 44 BC-AVF NR DDKT Functional Prednisolone, tacrolimus, MMF Pain, mass NR 12 13 8 NR NR Lung metastasis 6 Months
Demey et al,12 2014 M; 80 AVF Functional No KT No KT No KT Pain, swelling Aneurysmal VA 36 4.5 No KT Bone Supportive care NR 1 Month
Roy et al,13 2018 M; 63 AVF Thrombosed (ligated aneurysm) KT Functional MMF, tacrolimus Mass Mass NR NR 6 Lung NR NR 10 Days
Aldaabil et al,14 2016 M; 46 BC-AVF Thrombosed LRKT Functional MMF, tacrolimus, prednisone Pain, swelling NR NR NR 3 Lung, bone CMT, above elbow amputation NR 6 Months
Qureshi et al,15 2010 M; 48 AVF Functional First KT
Second LRKT Not functional
Functional NA
Prednisolone, MMF, rituximab, IVIG, tacrolimus Pain Infection NR 3 NR
2 No “High” above elbow amputation No recurrence or metastasis 9 Months
Costa et al,16 2017 F; 70 BC-AVF Thrombosed (ligated aneurysm) KT Functional Tacrolimus, MMF, prednisolone Dyspnea Pneumonia NR 11 9 Lung NR NR NR
Chanyaputhipong et al,17 2011 M; 57 RC-AVF Thrombosed (ligated aneurysm) DDKT Rejection None Pain, mass Granuloma, infection 8 17 NR NR Wide excision, RT, CMT Local recurrence, lung and bone metastasis 11 Months
Chanyaputhipong et al,17 2011 M; 63 BC-AVF Thrombosed (aneurysm) No KT No KT No KT Pain, mass with bleeding AVF malformation or AS 3 9 No KT No Wide excision, RT, CMT Lung metastasis 9.5 Months
Gale et al,18 2017 M; 44 RC-AVF Functional DDKT NR NR Necrotic fingers Hand ischemia NR 6 NR No Forequarter amputation NR NR
Kakisis et al,19 2019 M; 60 RC-AVF Thrombosed DDKT Functional Prednisolone, cyclosporine, MMF Bleeding from AVF NR NR 11 6 No Above elbow amputation Lung metastasis 6 Months
Jansen et al,20 2013 M; 58 NR NR DDKT Functional Yes (NR) Nodule, pain, hand dysfunction NR NR NR NR NR NR NR NR
Kleman et al,21 2016 M; 71 AVF NR LRKT Functional Alemtuzumab, tacrolimus, MMF Edema, erythema, non-PTH hypercalcemia Cancer NR NR NR No AVF resection, RT, CMT NR NR
Andre et al,22 2012 M; 62 AVF Functional First KT
Second KT Not functional
Functional NR
Prednisone, tacrolimus, MMF Nonhealed wound at AVF removal site NR NR NR NR
NR Lung RT, CMT NR NR
Figueiredo et al,23 2019 F; 55 AVG Thrombosed (ligated aneurysm) KT Functional Tacrolimus, prednisolone, MMF Mass NR 8 NR 15 Lung Palliative care, CMT NR 4 Months
Present case, 2020 M; 58 BC-AVF Thrombosed LRKT Functional Prednisolone, tacrolimus, MMF Pain Thrombotic VA 24 5 4 Lung, bone Shoulder disarticulation, CMT NR 1 Month
AS, Angiosarcoma; AVF, arteriovenous fistula; BC, brachiocephalic; CMT, chemotherapy; DDKT, deceased donor kidney transplantation; Dx, diagnosis; F, female; IST, immunosuppressive therapy; KT, kidney transplantation; LRKT, living related kidney transplantation; M, male; MMF, mycophenolate mofetil; NA, not applicable; NR, not reported; PT, pancreatic transplantation; PTH, parathyroid hormone; RC, radiocephalic; RT, radiation therapy; Tx, treatment; VA, vascular access.
The pathophysiology of angiosarcoma at an AVF after KT remains unclear. The possible mechanisms include the following11: (1) impairment of local immune response caused by the increased lymphatic workload and decreased lymphatic drainage from the increased venous pressure at the AVF site24; (2) turbulent blood flow in the AVF associated with the increased production of reactive oxygen species and matrix metalloproteinase-9, which results in a proliferative response and leads to vascular remodeling25; and (3) a large hypoxic area of the tumor caused by rapid cell proliferation exceeding the capacity of the oxygen supply. These mechanisms induce T-cell suppression, which leads to a diminished immune response.26 Furthermore, KT patients have a three- to fivefold increased risk of any cancer.27 Also, prolonged immunosuppression therapy has been associated with an increased cancer risk.28
According to our review, the most common clinical presentations were nonspecific pain (58.3%), swelling (29.2%), and a mass (29.2%). The initial diagnosis for pain at the AVF included a thrombosed AVF (44%), aneurysm (25%), infection (25%), and cancer (6.3%). Therefore, conservative treatment, including antibiotics and ligation without biopsy, was commonly implemented. These treatments usually led to a delayed diagnosis. The median interval from presentation to diagnosis was 10.8 weeks (range, 4-12 weeks). Thus, a high index of suspicion for angiosarcoma is required.
In our patient who had undergone brachial arterial embolectomy, we sent the clot and tissue for histopathologic examination. Cytologic examination of the blood clot showed atypical cells, and pathologic examination of the tissue showed epithelioid angiosarcoma. Pathologic examination of the shoulder disarticulation showed extensive tumor invasion; however, the origin of the tumor could not be identified. It might be that the origin of angiosarcoma came from the arterial side. Thus, we believe that any patient who has undergone KT and presents with pain or aneurysmal changes should have the blood clot from the aneurysm of an AVF examined cytologically and the tissue examined pathologically.
The tumor histologic features in angiosarcoma have varied widely from well to poorly differentiated. Abnormal, pleomorphic, and malignant endothelium are hallmarks of angiosarcoma. The cell appearance can be rounded, polygonal, fusiform, and, possibly, epithelioid.2 Epithelioid angiosarcoma is a variant of angiosarcoma composed of neoplastic cells with an epithelioid appearance. The most sensitive and specific marker for endothelial differentiation is CD31, which indicates platelet–endothelial cell adhesion molecule expression.29 From our review, all the patients were positive for CD31 expression. For our patient, the pathologic findings showed that the tumor was composed of high-grade pleomorphic epithelioid cells with positivity to ERG using immunohistochemistry and other vascular markers such as FLI1, CD31, and CD34.
Angiosarcoma behaves aggressively, recurs locally, and spreads widely. Moreover, it has a high rate of lymph node and systemic metastases. Our review showed that nearly one half of the patients had had metastases at diagnosis. Also, 85% of the patients had experienced recurrence, with 7.75% in the bone and 76.95% in the lungs. The mean survival period was 24.2 ± 14.4 weeks. Therefore, the tumor-related death rate is considerably high.
The ideal primary treatment is radical surgery with complete resection. However, difficulties ensue with wide excision owing to the tumor size and location, which will result in functional impairment and difficulty in wound reconstruction. Because of the high local recurrence rate, adjuvant radiotherapy has been recommended. Angiosarcoma spreads mainly through hematogenously, and the most common metastatic site has been the lungs. Chemotherapy is the primary option for metastatic angiosarcoma. Biologic therapies, in particular, antiangiogenic therapies, are considered interesting options for angiosarcoma-specific treatment.
Conclusion
Angiosarcoma at vascular access sites can occur in both KT and non-KT patients. Because the presentation is not specific, a high index of suspicion is required to prevent a delayed diagnosis. Angiosarcoma has aggressive behavior and high local recurrence and metastatic rates. The primary treatment of angiosarcoma is radical surgery with complete resection. Chemotherapy is the primary treatment of metastatic angiosarcoma, with radiotherapy for local recurrence prevention. However, despite multimodal treatment, the prognosis of these patients has remained poor.
The authors gratefully acknowledge Nipapan Choonu for the statistical analysis.
Author conflict of interest: none.
The editors and reviewers of this article have no relevant financial relationships to disclose per the Journal policy that requires reviewers to decline review of any manuscript for which they may have a conflict of interest. | Fatal | ReactionOutcome | CC BY-NC-ND | 33718686 | 19,581,913 | 2021-03 |
What was the outcome of reaction 'Metastases to lung'? | Angiosarcoma in arteriovenous fistula after kidney transplantation.
Angiosarcoma is a rare complication of both functioning and nonfunctioning fistulas. It is an aggressive soft tissue sarcoma arising from vascular or lymphatic endothelial cells. We report a case of angiosarcoma from a nonfunctional fistula in a kidney transplantation patient receiving immunosuppressive drugs. The patient had presented with arm pain mimicking a thrombosed arteriovenous fistula.
Angiosarcoma is a rare aggressive soft-tissue sarcoma characterized by rapidly proliferating and extensively infiltrating anaplastic cells, which derives from blood vessels and lining of irregular blood-filled spaces.1 It is a rare complication of arteriovenous fistulas.2 In renal transplant recipients, angiosarcoma is an extremely rare type of cancer, accounting for <1% of all cancers and 2% of localized soft tissue cancer.3 Because of its rarity, most clinicians are unaware of this condition, resulting in a delayed diagnosis and treatment. The institutional review board and ethics committee approved the present study, and a retrospective analysis of the medical records was conducted. The related reported data were also reviewed.
Case report
A 58-year-old man had presented with right arm pain at a brachiocephalic arteriovenous fistula (BCAVF). He had end-stage renal disease and had begun hemodialysis via a right BCAVF 5 years earlier. After 1 year of hemodialysis, he had undergone living related kidney transplantation (KT). He had been receiving immunosuppression therapy with prednisolone, mycophenolate mofetil, and tacrolimus. His baseline creatinine was 0.6 mg/dL. Later, he had complained of right arm pain, and a partially thrombosed AVF was diagnosed. After 6 months of conservative treatment, his pain and weakness had become severe. Also, paresthesia in his right arm and bleeding erosion from the BCAVF were observed. The diagnosis was impending rupture of an AVF aneurysm. He subsequently underwent aneurysmal resection. Preoperative chest radiograph showed a right lung mass. During the intraoperative period, no mass in the intraluminal cephalic vein could be identified (Fig 1). However, his severe arm pain had persisted postoperatively, and a radiograph showed multiple osteolytic lesions in the right humeral shaft, proximal ulna, and radius (Fig 2). Chest computed tomography revealed a 2-cm, lobulated pulmonary nodule in the right upper lung. Therefore, we suspected primary lung cancer with bone metastases. Fifteen days later, he had returned with acute limb ischemia of the right arm. Computed tomography angiography showed no contrast filling in the proximal radial and ulnar arteries at the distal brachial to brachial bifurcation. An eccentric filling defect was seen at the right brachial artery (Fig 3). Intraoperatively, soft tan-brown tissue was recovered by brachial arterial embolectomy. Cytologic examination of the blood clot showed atypical cells. In addition, pathologic examination of the tissue revealed epithelioid angiosarcoma. Examination of a bone biopsy specimen from the right humerus revealed angiosarcoma. At 3 months after aneurysm resection, the patient underwent right shoulder disarticulation. Histopathologic examination of the right shoulder disarticulation revealed angiosarcoma involving the AVF, soft tissue of the cubital fossa, and proximal humerus (Figs 4 and 5). The final diagnosis was angiosarcoma at the AVF with bone and pulmonary metastasis. Therefore, we planned palliative chemotherapy with paclitaxel (Taxol; Bristol-Myers Squibb, New York, NY) weekly. However, the patient developed hemoptysis. He died of massive hemothorax 10 months after symptom onset and 1 month after diagnosis. Because the patient had died and we were unable to contact his relatives, consent could not be obtained from the patient or his family.Fig 1 Thrombosis of cephalic vein found intraoperatively. After removal of the clot, no mass and no abnormality was found in the cephalic vein.
Fig 2 Radiograph of the right elbow showing an osteolytic lesion at the ulna (white arrow).
Fig 3 Computed tomography angiogram showing total thrombosis of the cephalic vein of the brachiocephalic arteriovenous fistula. No contrast filling had occurred in the proximal radial and ulnar arteries at the distal brachial to brachial bifurcation. An eccentric filling defect was seen at the right brachial artery (white arrow).
Fig 4 High-grade pleomorphic epithelioid cells with amphophilic cytoplasm and vasoformative growth.
Fig 5 Positivity to ERG immunohistochemistry and other vascular markers such as FLI1, CD31, and CD34 (data not shown) confirmed the diagnosis of angiosarcoma.
Discussion
According to our review of the reported data4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23 (Table), 23 cases of angiosarcoma occurring from vascular access had been described in 20 English language reports. Most of the patients were men (83.3%), and their mean age was 52.7 ± 15.2 years. The most often used vascular access was the AVF (95.6%). Of the 20 patients, 13 (65%) had undergone deceased donor KT, 5 (25%) had undergone living related KT, and 2 (10%) had not undergone KT. Of the 23 kidney grafts, 20 (93%) were functional. Only five patients had received a second KT.Table Summary of reported cases of angiosarcoma at arteriovenous fistula sites
Investigator Sex; age, years Vascular access AVF status Transplantation IST Presentation Initial Dx Interval to Dx (weeks) Interval (years) from Metastasis before Tx Tx Recurrence Survival
Type Status VA to AS KT to AS
Byers et al,4 1992; Parrott et al,5 1993 M; 36 RC-AVF Functional DDKT Functional Azathioprine, cyclosporine, prednisolone Swelling Thrombotic VA, infection 20 12 8 No Above elbow amputation Postnasal space, lung and axillary lymph node metastasis 1 Year
Wehrli et al,6 1998 M; 64 RC-AVF Thrombosed DDKT Functional Azathioprine, cyclosporine, prednisone Swelling, pain Thrombotic VA 4 10 7 No RT, followed by below elbow amputation Local and lung metastasis Alive 6 months after amputation
Álvarez et al,3 2013 F; 47 AVF Thrombosed DDKT Functional Cyclosporine, MMF Nail-bed splinter lesions, pain, hand retraction Thrombotic or aneurysmal VA 12 14 8 Bone, lung CMT NR 2 Months
Conlon et al,7 1993 M; 40 BC-AVF Thrombosed (ligated aneurysm) First DDKT
Second DDKT Rejection
Functional Treated for rejection
Prednisolone, azathioprine, prednisone Pain and swelling NA 4 40 16
7 No Skeletalization of anterior compartment of arm; RT Local recurrence and lung metastasis 5 Months
Keane et al,8 1993 M; 11 AVF Thrombosed (ligated aneurysm) First DDKT
Second DDKT Rejection
Functional Treated for rejection
Corticosteroids, azathioprine Mass, induration arising from AVF NA 8 NR NR
7 No Wide excision, RT Local recurrence, lung and brain metastasis 8 Months
Bessis et al,9 1998 M; 61 BC-AVF Thrombosed DDKT Functional Azathioprine, cyclosporine, prednisolone Pain, swelling, tenderness Thrombotic or aneurysmal VA NR (>6) 7 5 No Above elbow amputation Lung metastasis 7 Months
Farag et al,10 2005 M; 26 AVF Thrombosed (ligated aneurysm) First DDKT plus PT Second DDKT Not functional
Functional NA
Azathioprine, cyclosporine, prednisolone Mass Thrombotic or aneurysmal VA NR 13 NR
11 Lung CMT Local recurrence 11 Months
Webster et al,11 2011 M; 59 BC-AVF Thrombosed LRKT Functional Rituximab daclizumab, tacrolimus Pain, swelling Thrombotic and infection VA NR 3 2 No Above elbow amputation No Survived
Webster et al,11 2011 F; 41 BC-AVF Thrombosed LRKT Functional Alemtuzumab, prednisolone, tacrolimus Pain Thrombotic VA 4 NR NR NR NR Lung metastasis 4 Months
Webster et al,11 2011 M; 44 BC-AVF NR DDKT Functional Prednisolone, tacrolimus, MMF Pain, mass NR 12 13 8 NR NR Lung metastasis 6 Months
Demey et al,12 2014 M; 80 AVF Functional No KT No KT No KT Pain, swelling Aneurysmal VA 36 4.5 No KT Bone Supportive care NR 1 Month
Roy et al,13 2018 M; 63 AVF Thrombosed (ligated aneurysm) KT Functional MMF, tacrolimus Mass Mass NR NR 6 Lung NR NR 10 Days
Aldaabil et al,14 2016 M; 46 BC-AVF Thrombosed LRKT Functional MMF, tacrolimus, prednisone Pain, swelling NR NR NR 3 Lung, bone CMT, above elbow amputation NR 6 Months
Qureshi et al,15 2010 M; 48 AVF Functional First KT
Second LRKT Not functional
Functional NA
Prednisolone, MMF, rituximab, IVIG, tacrolimus Pain Infection NR 3 NR
2 No “High” above elbow amputation No recurrence or metastasis 9 Months
Costa et al,16 2017 F; 70 BC-AVF Thrombosed (ligated aneurysm) KT Functional Tacrolimus, MMF, prednisolone Dyspnea Pneumonia NR 11 9 Lung NR NR NR
Chanyaputhipong et al,17 2011 M; 57 RC-AVF Thrombosed (ligated aneurysm) DDKT Rejection None Pain, mass Granuloma, infection 8 17 NR NR Wide excision, RT, CMT Local recurrence, lung and bone metastasis 11 Months
Chanyaputhipong et al,17 2011 M; 63 BC-AVF Thrombosed (aneurysm) No KT No KT No KT Pain, mass with bleeding AVF malformation or AS 3 9 No KT No Wide excision, RT, CMT Lung metastasis 9.5 Months
Gale et al,18 2017 M; 44 RC-AVF Functional DDKT NR NR Necrotic fingers Hand ischemia NR 6 NR No Forequarter amputation NR NR
Kakisis et al,19 2019 M; 60 RC-AVF Thrombosed DDKT Functional Prednisolone, cyclosporine, MMF Bleeding from AVF NR NR 11 6 No Above elbow amputation Lung metastasis 6 Months
Jansen et al,20 2013 M; 58 NR NR DDKT Functional Yes (NR) Nodule, pain, hand dysfunction NR NR NR NR NR NR NR NR
Kleman et al,21 2016 M; 71 AVF NR LRKT Functional Alemtuzumab, tacrolimus, MMF Edema, erythema, non-PTH hypercalcemia Cancer NR NR NR No AVF resection, RT, CMT NR NR
Andre et al,22 2012 M; 62 AVF Functional First KT
Second KT Not functional
Functional NR
Prednisone, tacrolimus, MMF Nonhealed wound at AVF removal site NR NR NR NR
NR Lung RT, CMT NR NR
Figueiredo et al,23 2019 F; 55 AVG Thrombosed (ligated aneurysm) KT Functional Tacrolimus, prednisolone, MMF Mass NR 8 NR 15 Lung Palliative care, CMT NR 4 Months
Present case, 2020 M; 58 BC-AVF Thrombosed LRKT Functional Prednisolone, tacrolimus, MMF Pain Thrombotic VA 24 5 4 Lung, bone Shoulder disarticulation, CMT NR 1 Month
AS, Angiosarcoma; AVF, arteriovenous fistula; BC, brachiocephalic; CMT, chemotherapy; DDKT, deceased donor kidney transplantation; Dx, diagnosis; F, female; IST, immunosuppressive therapy; KT, kidney transplantation; LRKT, living related kidney transplantation; M, male; MMF, mycophenolate mofetil; NA, not applicable; NR, not reported; PT, pancreatic transplantation; PTH, parathyroid hormone; RC, radiocephalic; RT, radiation therapy; Tx, treatment; VA, vascular access.
The pathophysiology of angiosarcoma at an AVF after KT remains unclear. The possible mechanisms include the following11: (1) impairment of local immune response caused by the increased lymphatic workload and decreased lymphatic drainage from the increased venous pressure at the AVF site24; (2) turbulent blood flow in the AVF associated with the increased production of reactive oxygen species and matrix metalloproteinase-9, which results in a proliferative response and leads to vascular remodeling25; and (3) a large hypoxic area of the tumor caused by rapid cell proliferation exceeding the capacity of the oxygen supply. These mechanisms induce T-cell suppression, which leads to a diminished immune response.26 Furthermore, KT patients have a three- to fivefold increased risk of any cancer.27 Also, prolonged immunosuppression therapy has been associated with an increased cancer risk.28
According to our review, the most common clinical presentations were nonspecific pain (58.3%), swelling (29.2%), and a mass (29.2%). The initial diagnosis for pain at the AVF included a thrombosed AVF (44%), aneurysm (25%), infection (25%), and cancer (6.3%). Therefore, conservative treatment, including antibiotics and ligation without biopsy, was commonly implemented. These treatments usually led to a delayed diagnosis. The median interval from presentation to diagnosis was 10.8 weeks (range, 4-12 weeks). Thus, a high index of suspicion for angiosarcoma is required.
In our patient who had undergone brachial arterial embolectomy, we sent the clot and tissue for histopathologic examination. Cytologic examination of the blood clot showed atypical cells, and pathologic examination of the tissue showed epithelioid angiosarcoma. Pathologic examination of the shoulder disarticulation showed extensive tumor invasion; however, the origin of the tumor could not be identified. It might be that the origin of angiosarcoma came from the arterial side. Thus, we believe that any patient who has undergone KT and presents with pain or aneurysmal changes should have the blood clot from the aneurysm of an AVF examined cytologically and the tissue examined pathologically.
The tumor histologic features in angiosarcoma have varied widely from well to poorly differentiated. Abnormal, pleomorphic, and malignant endothelium are hallmarks of angiosarcoma. The cell appearance can be rounded, polygonal, fusiform, and, possibly, epithelioid.2 Epithelioid angiosarcoma is a variant of angiosarcoma composed of neoplastic cells with an epithelioid appearance. The most sensitive and specific marker for endothelial differentiation is CD31, which indicates platelet–endothelial cell adhesion molecule expression.29 From our review, all the patients were positive for CD31 expression. For our patient, the pathologic findings showed that the tumor was composed of high-grade pleomorphic epithelioid cells with positivity to ERG using immunohistochemistry and other vascular markers such as FLI1, CD31, and CD34.
Angiosarcoma behaves aggressively, recurs locally, and spreads widely. Moreover, it has a high rate of lymph node and systemic metastases. Our review showed that nearly one half of the patients had had metastases at diagnosis. Also, 85% of the patients had experienced recurrence, with 7.75% in the bone and 76.95% in the lungs. The mean survival period was 24.2 ± 14.4 weeks. Therefore, the tumor-related death rate is considerably high.
The ideal primary treatment is radical surgery with complete resection. However, difficulties ensue with wide excision owing to the tumor size and location, which will result in functional impairment and difficulty in wound reconstruction. Because of the high local recurrence rate, adjuvant radiotherapy has been recommended. Angiosarcoma spreads mainly through hematogenously, and the most common metastatic site has been the lungs. Chemotherapy is the primary option for metastatic angiosarcoma. Biologic therapies, in particular, antiangiogenic therapies, are considered interesting options for angiosarcoma-specific treatment.
Conclusion
Angiosarcoma at vascular access sites can occur in both KT and non-KT patients. Because the presentation is not specific, a high index of suspicion is required to prevent a delayed diagnosis. Angiosarcoma has aggressive behavior and high local recurrence and metastatic rates. The primary treatment of angiosarcoma is radical surgery with complete resection. Chemotherapy is the primary treatment of metastatic angiosarcoma, with radiotherapy for local recurrence prevention. However, despite multimodal treatment, the prognosis of these patients has remained poor.
The authors gratefully acknowledge Nipapan Choonu for the statistical analysis.
Author conflict of interest: none.
The editors and reviewers of this article have no relevant financial relationships to disclose per the Journal policy that requires reviewers to decline review of any manuscript for which they may have a conflict of interest. | Fatal | ReactionOutcome | CC BY-NC-ND | 33718686 | 19,581,913 | 2021-03 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Adverse event'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Angioedema'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Anxiety'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Confusional state'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Constipation'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Decreased appetite'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Depression'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Dizziness'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Drug intolerance'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Dry mouth'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Fatigue'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Feeling jittery'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Formication'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Gastrooesophageal reflux disease'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Hallucination'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Headache'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Inadequate analgesia'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Influenza like illness'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Insomnia'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Malaise'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Muscle twitching'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Nausea'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Nightmare'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Orthostatic hypotension'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Paraesthesia'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Pruritus'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Pyrexia'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Rash'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Somnolence'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Suicidal ideation'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Tinnitus'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Urinary retention'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Vision blurred'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Vomiting'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Withdrawal syndrome'. | Impact of Pharmacogenomic Information on Values of Care and Quality of Life Associated with Codeine and Tramadol-Related Adverse Drug Events.
To assess the potential impact of Pharmacogenomic (PGx) variation in cytochrome P450 2D6 (CYP2D6) enzyme function, using loss in quality-adjusted life years (QALYs) associated with treatment problems, and the willingness to pay to avoid treatment problems from patients' and payers' perspectives.
The study included patients prescribed tramadol or codeine, or both, between January 1, 2005, and December 31, 2017. Demographic information and adverse drug events, including adverse drug events and poor pain control, were collected from the electronic health records using natural language processing techniques and review by trained abstractors. Patients' willingness to pay and QALY estimates were based on comprehensive literature review. The CYP2D6 phenotypes were divided into 4 groups: ultra-rapid metabolizers, normal metabolizers, intermediate metabolizers, and poor metabolizers.
Among the 2860 identified patients, 63 (2%) were ultrarapid metabolizers, 1449 (50%) were normal metabolizers, 1155 (40%) were intermediate metabolizers, and 193 (7%) were poor metabolizers. The patients' average estimated willingness-to-pay value to avoid treatment problems was $23 per month; poor metabolizers developed problems with the highest estimated willingness-to-pay value ($32 per month). The mean QALY loss among all patients was 0.024 QALYs (8.8 healthy days); poor metabolizers had the highest loss (0.027 QALYs, 9.9 healthy days).
Patients with various phenotypes developed different treatment problem profiles. Poor CYP2D6 metabolizers developed problems with highest willingness to pay, and they might potentially benefit most from PGx-guided treatment and problem prevention.
Opioid analgesics are frequently used for management of both acute and chronic pain.1,2 Opioid prescriptions in the United States are common, with a rate of 59 per 100 persons in 2017.3 Codeine and tramadol are two of the most commonly used opioids. Codeine yields adverse effects of gastrointestinal discomfort, vertigo, sleepiness, and rash4; tramadol has been reported to cause nausea, vomiting, fatigue, sedation, sweating, and serotonin toxicity.5,6 Many of these treatment problems, including adverse drug events (ADEs) and poor pain control, are not life-threatening, and they can require limited management by health care providers. Therefore, the direct financial effects of these problems may be limited; however, these problems also affect quality of life, and information quantifying the effects of such outcomes on patient quality of life is lacking.
Cytochrome P450 2D6 (CYP2D6) is the primary enzyme that bioactivates codeine, and variation in CYP2D6 function has been associated with treatment problems related to use of tramadol and codeine.7 At the genetic level, persons exposed to codeine and tramadol with the CYP2D6 poor metabolizer phenotype have lower levels of the active opioid metabolites, while patients with ultrarapid CYP2D6 phenotypes experience higher systemic levels of the metabolites.8 Therefore, poor CYP2D6 metabolizers might have less pain control compared with normal metabolizers.9 In contrast, ultrarapid CYP2D6 metabolizers require less pain medication to achieve pain control, but they may be at greater risk of other ADEs compared with normal metabolizers.10, 11, 12
Consideration of patient pharmacogenomics (PGx) information at the time of drug prescriptions holds the potential to avoid ADEs and to maximize drug effectiveness.13, 14, 15 Ideally, access to CYP2D6 phenotype information at the time of drug prescription would enable genotype-guided drug and dose selection.13, 14, 15 Such information could improve the patient experience and reduce health care costs. However, clinical guidelines regarding the implementation of PGx information into clinical practice were mainly established using data from pharmacokinetic studies. Data regarding value of care or improvements in quality of life owing to the implementation of PGx information for opioid prescriptions is currently limited.16,17
Therefore, the goal of our study was to understand to what extent PGx-guided treatment might decrease potential costs associated with treatment problems or improve the quality of life for patients who receive codeine and tramadol prescriptions. To accomplish this goal, we described the types of treatment problems experienced by patients who received codeine or tramadol prescriptions. Second, using existing literature, we assigned monetary values to the problems based on what patients have reported that they are willing to pay to avoid these events; we also assigned quality-adjusted life years (QALYs) to each ADE or case of poor pain control. QALY is a generic measure of disease burden that aggregates both health-related quality of life and length of life into a single measure.18 QALYs are frequently used in cost-effectiveness analyses to measure health outcomes related to treatment effectiveness. One QALY reflects the treatment benefit of gaining the patient 1 full year of life with good health, and it was designed for interpersonal comparison.19 Finally, we estimated health insurance’s (payer’s) willingness to pay to treat the problems. Together, our study provides an innovative approach to examine the monetary effects that PGx information could provide on treatment problems and loss in QALYs that patients might experience with codeine or tramadol prescriptions.
Patients and Methods
Study Population
This study included patients who participated in the Right Drug, Right Dose, Right Time-Using Genomic Data to Individualize Treatment (RIGHT Study), and who were prescribed tramadol and codeine between January 1, 2005, and December 31, 2017.20 In brief, the RIGHT study enrolled 10,074 participants with self-reported demographic information and blood samples sequenced for CYP2D6.20 The initiation of tramadol and codeine treatment was captured using the Rochester Epidemiology Project (REP) research infrastructure, with the details reported previously.21 Normalized names for clinical drugs provided by RxNorm were used to identify the ingredients that included tramadol or codeine.22 We excluded persons who were found to have prior tramadol or codeine use between January 1, 2004, and December 31, 2004, and those who used codeine as a cough suppressant. We also excluded persons who were prescribed strong or moderately strong CYP2D6 inhibitors (Supplemental Table 1, available online at http://mcpiqojournal.org) during this period23 to minimize the effect of phenoconversion, which is a change in apparent drug metabolism phenotype owing to drug–drug interaction rather than to genetic variation.24 The study was approved by the Mayo Clinic Institutional Review Board (IRB# 16-000189), and all the study subjects gave informed consent.
Pharmacogenomic Phenotyping
The CYP2D6 phenotypes were divided into 4 groups according to the drug metabolism rates estimated from patients’ genotypes: (1) ultra-rapid metabolizers, including ultra-rapid and rapid metabolizers; (2) normal metabolizers, including normal and intermediate to normal metabolizers; (3) intermediate metabolizers, including intermediate and poor to intermediate metabolizers; and (4) poor metabolizers. The processes of genotyping and phenotype prediction have been reported previously.25
Identifying Treatment Problems and Their Values
Treatment Problems
Treatment problems included ADEs and poor pain control. These data were captured from reviewing the electronic health records. The reviewing process included two steps. First, a natural language processing technique was applied in the initial screening for opioid-related adverse outcomes. This process captured all electronic health record sentences with normalized opioid names and key words related to treatment problems (eg, nausea, vomiting, constipation and poor-pain-control). Second, two abstractors (J.L.S. and a trained nurse abstractor) reviewed the screened sentences and recorded any adverse symptom or poor pain control event attributed to codeine or tramadol within 6 weeks after the first codeine or tramadol prescription. Adverse symptoms were further grouped in 6 system categories (eg, nausea and vomiting were classified as “gastrointestinal symptoms”). Our final definition of treatment problems included the presence of either an adverse symptom attributed to codeine or tramadol or documentation of poor pain control (eg, “patient has been taking Ultram but continues to complain of severe neck pain”) after prescription of these medications. If there was no mention of adverse symptoms or poor pain control in the medical record notes, the patient was classified as not having treatment problems. Figure 1 lists all treatment problem types and their system categories.Figure 1 Patients’ willingness-to-pay values, loss of QALYs for each type of adverse drug outcome. Bubble size represents the proportion of the study population that experienced the specific treatment problem. QALY, quality-adjusted life year.
Patients’ Willingness to Pay to Avoid Treatment Problems
We estimated patients’ willingness to pay to avoid treatment problems resulting from a codeine or tramadol prescription. Willingness to pay provides a useful estimate of a patient’s perceived value of care, because monetary value is a proxy for the trade-off and priority a person places on receiving a particular outcome.26 To obtain willingness-to-pay monetary values for each problem, we conducted a comprehensive literature review across multiple databases from inception through October 11, 2019, and the inception varies by database. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Ovid PsycINFO, and Scopus. Controlled vocabulary supplemented with keywords was used to search for patients’ willingness to pay in monetary values of opioid-induced symptoms from the patient’s perspective, in the United States and Canada. We found 21 relevant studies (of 1294 possible studies) reporting the monetary values patients that were willing to pay to treat the problem types identified in our study. The literature review process and the search strategies are shown in the Supplemental Table 2 and Supplemental Figure 1 (available online at http://mcpiqojournal.org). All the monetary values were converted from the study year to 2018 US dollars using the gross domestic product price deflator from the Bureau of Economic Analysis.27
Health-Related Quality of Life Loss Owing to Treatment Problems
We also used loss in QALYs to quantitatively measure the effect of individual treatment problems on both length of life and health-related quality of life. The decrease in QALYs for each of the problem types was estimated using the available literature from the best possible similar patient populations within the most recent 20 years (1998-2018). If we were unable to find estimated QALYs for ADEs using studies from US or Canadian populations, we used results from European countries as close estimations.
Payer’s Willingness to Pay for the Improvement of Health-Related Quality of Life
The widely accepted threshold for payers’ willingness to cover the costs associated with decreases in quality of life is $50,000-10,000/QALY.28 We estimated the payer’s monthly willingness to pay by multiplying the QALYs by $50,000/QALY (Equation 1).29,30 This threshold has been adopted by most of the recent cost-effectiveness studies for most countries.31 Payers’ willingness to pay was calculated using the following function:Willingness_to_Payi/month=Uiyears×$50000/year12 months(Equation1)
where Ui is the number of QALY lost for each adverse outcome.
Patients’ and payers’ willingness to pay estimates and decreases in QALYs associated with each of the ADEs are listed in Supplemental Table 3 (available online at http://mcpiqojournal.org).
Analysis
This study examined the baseline patient characteristics for the overall sample and by the 4 CYP2D6 phenotypes. Average values of patients’ willingness to pay, QALYs, and payers’ willingness to pay were calculated for the overall sample and by phenotypes. The timeframe for QALYs and willingness to pay was calculated by month, because the opioid can be used as postoperational pain control, which is relatively short period. P for trend values were calculated from ordered logistic regression, and P < .05 were considered statistically significant. The study results were achieved under multiple assumptions (Supplemental Table 4, available online at http://mcpiqojournal.org).
Results
This study included 2877 patients with new tramadol and codeine prescriptions from January 1, 2005, to December 31, 2017. Characteristics of the study population are shown in Table. Overall, 792 patients (28%) were prescribed codeine, 2396 patients (83%) were prescribed tramadol, and 311 patients (11%) were prescribed both of the medications. Overall, 61 patients (2%) were CYP2D6 ultrarapid metabolizers, 1448 patients (50%) were normal metabolizers, 1175 patients (41%) were intermediate metabolizers, and 193 patients (7%) were poor metabolizers.Table Patient Characteristics for Full Study Sample and by CYP2D6 Phenotypea
Variable Full sample Ultra-rapid metabolizer Normal metabolizer Intermediate metabolizer Poor metabolizer
Number of patients 2860 63 (2%) 1449 (51%) 1155 (40%) 193 (7%)
Mean BMI, kg/m2 (SD) 29.74 (7.36) 29.27 (5.66) 29.53 (6.53) 30.02 (8.58) 29.68 (5.47)
Mean age, years (SD) 61.27 (13.58) 63.84 (11.33) 60.83 (13.94) 61.39 (13.22) 62.99 (13.49)
Sex, n (%)
Female 1680 (59%) 37 (59%) 869 (60%) 666 (58%) 108 (56%)
Male 1180 (41%) 26 (41%) 580 (40%) 489 (42%) 85 (44%)
Race, n (%)
White 2690 (94%) 56 (89%) 1354 (93%) 1101 (95%) 179 (93%)
African American 11 (0%) 0 (0%) 7 (0%) 4 (0%) 0 (0%)
Asian/Native 22 (1%) 0 (0%) 18 (1%) 4 (0%) 0 (0%)
Othersb 137 (5%) 7 (11%) 70 (5%) 46 (4%) 14 (7%)
Ethnicity, n (%)
Non-Hispanic 2823 (99%) 62 (98%) 1428 (99%) 1140 (99%) 193 (100%)
Hispanic 33 (1%) 1 (2%) 18 (1%) 14 (1%) 0 (0%)
Unknown 4 (0%) 0 (0%) 3 (0%) 1 (0%) 0 (0%)
Marital status, n (%)
Married 2259 (79%) 55 (87%) 1131 (78%) 917 (79%) 156 (81%)
Previously marriedc 409 (14%) 6 (10%) 209 (14%) 172 (15%) 22 (11%)
Never married 191 (7%) 2 (3%) 108 (7%) 66 (6%) 15 (8%)
Unknown 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%)
Education, n (%)
≤High school 423 (15%) 7 (11%) 227 (16%) 154 (13%) 35 (18%)
College 1363 (48%) 28 (44%) 678 (47%) 570 (49%) 87 (45%)
Postgraduate 1070 (37%) 28 (44%) 542 (37%) 429 (37%) 71 (37%)
Unknown 4 (0%) 0 (0%) 2 (0%) 2 (0%) 0 (0%)
Prescription, n (%)d
Codeine 785 (27.4%) 14 (22.2%) 394 (27.2%) 314 (27.2%) 63 (32.6%)
Tramadol 2384 (83.4%) 55 (87.3%) 1206 (83.2%) 969 (83.9%) 154 (79.8%)
Number of ADEs
1 ADE type 241 (8.4%) 3 (4.8%) 118 (8.1%) 103 (8.9%) 17 (8.8%)
2 ADE types 44 (1.5%) 3 (4.8%) 27 (1.9%) 11 (1.0%) 3 (1.6%)
≥3 ADE types 8 (0.3%) 0 (0.0%) 5 (0.3%) 3 (0.3%) 0 (0.0%)
Any ADE 301 (10.5%) 6 (9.5%) 155 (10.7%) 120 (10.4%) 20 (10.4%)
No ADE 2567 (89.8%) 57 (90.5%) 1299 (89.6%) 1038 (89.9%) 173 (89.6%)
a ADE, adverse drug event; BMI, body mass index.
b Others including race reported by patients as “other,” mixed, or unknown.
c Includes widowed or divorced but currently not married.
d A total of 311 (10.8%) patients were prescribed of both codeine and tramadol.
Two hundred forty-three patients (8.4%) had 1 type of ADE, 44 patients (1.5%) had 2 types, and 8 patients (0.3%) experienced 3 or more types of ADEs recorded in their medical records (Table). Type of ADE did not differ significantly by CYP2D6 phenotype, with the exception of neurologic reactions (Figure 2). Neurologic reactions were most frequent among ultra-rapid metabolizers and least frequent among poor metabolizers (P = .006). However, when ADEs were combined, the proportion of persons experiencing at least one ADE did not differ significantly by phenotype. Poor pain control was highest in poor metabolizers and lowest among ultrarapid metabolizers.Figure 2 Differences in types of treatment problems (including drug adverse effects and poor pain control) due to codeine and tramadol prescriptions by CYP2D6 phenotype. P values were from logistic regressions of genetic groups on adverse events.
Figure 1 illustrates the patients’ estimated willingness to pay to avoid individual treatment problems and the estimated QALYs for each problem. Bubble size represents the proportion of the study population that experienced the problem. For example, tinnitus was rare in our study population (n = 2; small bubble), but it had the highest estimated patient willingness-to-pay value ($1084 per month; Figure 1). Pain, nausea, and vomiting were more common problems (n = 117 and 79, respectively), but the corresponding estimated willingness to pay was lower ($333 and $138 per month, respectively). Pain had the highest QALY value, with an estimated loss of 0.33 QALYs owing to pain (Figure 1, right panel).
On average, patients’ average willingness to pay for avoiding a problem associated with codeine or tramadol treatment was estimated at $23.16 per month (Figure 3A). The mean QALY loss owing to the treatment problems among all the patients was 0.024 QALYs (8.8 quality-adjusted days; Figure 3B). Finally, the overall estimated payers’ willingness to pay was $100 per month (Figure 3C). Patients with a poor metabolizer phenotype had the highest estimated willingness to pay to avoid their treatment problems ($32 per month), worst QALYs (10 quality-adjusted days), and the highest estimated payer willingness to pay ($113 per month). These results were not statistically significant (all P values >.05). We used the threshold of $100,000 per QALY to examine the cost-effectiveness of genetic testing. The results suggested that if the testing costs were less than $88.86 for each patient, it would be cost-effective for all the genetic groups. The testing value changed among genetic variant groups (Figure 4). We found that health care payers’ willingness to pay was $75-$87 per month higher than patients’ willingness to pay for each phenotype group. Figure 4 illustrates the differences in patients’ willingness to pay versus payers’ willingness to pay for treatment problems owing to codeine and tramadol prescriptions.Figure 3 Estimated patient willingness to pay, QALYs, and payer willingness to pay overall and by CYP2D6 phenotype. (A) Estimated amount patients would be willing to pay to avoid treatment problems. (B) Estimated loss of QALYS owing to treatment problems. (C) Estimated willingness to pay from payer’s perspective. P values are from logistic regressions of genetic groups on adverse events. ADE, adverse drug events; QALY, quality-adjusted life-years.
Figure 4 Estimated patient willingness to pay and payer willingness to pay overall and by CYP2D6 phenotype with patients’ total values for opioid treatments, including all adverse symptoms and poor pain control. Payers’ willingness to pay was calculated from $50,000 per QALY multiplied by the QALYs lost owing to treatment problems. All the values were converted to 2018 US dollars using the gross domestic product price deflator.
Discussion
This study provided important information that the number of problems from treatment does not provide enough information in evaluating the adverse effects. When weighted by patients’ values in either monetary values or QALYs, differences appeared among groups of patients with different PGx backgrounds.
Overall, estimated amounts that patients would be willing to pay to avoid treatment problems, and the estimated QALYs lost owing to these problems, were modest. However, estimated payer willingness to pay to improve QALYs was nearly 5 times greater than patient willingness to pay, suggesting the substantially large benefit that PGx information could bring to both stakeholders. Finally, patients with poor CYP2D6 metabolizer phenotypes had the highest willingness to pay and the worst QALY estimates, suggesting that this population is the most likely to benefit from PGx-directed prescribing.
The significance of this study is that we estimated the amount patients would be willing to pay to avoid specific problems, which was captured from previous studies in similar populations. This study provided information that could facilitate clinical communication on treatment outcomes, based on the theory that people use the subjective value of the treatment outcome to generate expectations for the treatment.32 Overall, we found that patients would be willing to pay modest amounts ($23 per month) to avoid treatment problems and that the number of QALYs lost owing to codeine and tramadol were also modest (9 quality-adjusted days). These figures reflect the relatively limited proportion of persons who had treatment problem information documented in their medical records, and the fact that most effects of codeine and tramadol are relatively short term. One important reason is that only 10% of the patients in our study had documentation in their medical record indicating a treatment problem resulting from codeine or tramadol, which is consistent with other studies on tramadol and codeine.4,6 Similar to other studies, the ADEs observed in our study could reflect the real-world nature of our study, as patients are not routinely asked about medication responses. We expect that some patients who were prescribed these medications experienced treatment problems, coped with the problems at home, and did not report problems to their health care providers. Therefore, our estimates are likely to be an underestimate of both the true proportion of persons who experienced treatment problems following codeine and tramadol use, and an underestimate of the costs of these problem. Another important phenomenon discovered by this study was that tramadol was predominantly used in clinical practice than codeine (80% vs. 20%), and the problem profiles were similar to what developed from tramadol use (Supplemental Figure 2, available online at http://mcpiqojournal.org).
We also found that persons with poor metabolizer CYP2D6 phenotypes had the highest estimated values of willingness to pay owing to codeine and tramadol. For example, the average estimated willingness to pay for a poor metabolizer to avoid a treatment problem was $32 per month compared with an estimated $19 per month for an ultrarapid metabolizer. We also found that treatment problems can decrease patients’ health-related quality of life by approximately 8-10 days during 1 healthy year. If these could be avoided, poor metabolizers would receive the biggest benefit (gaining 9.9 healthy days) compared with normal metabolizer patients (gaining 8.6 healthy days). We also compared this benefit to health care payers’ willingness to pay and found that the threshold was $113 per month for poor metabolizer patients and $98 per month for normal metabolizer patients, which suggests that the costs below this threshold would be cost-effective from the health care payers’ perspective. In summary, although cost differences among metabolizer phenotypes were not statistically significant, our results suggest that poor metabolizer patients could benefit more from PGx-guided prescribing compared with normal metabolizers.
Strengths of our study include our adoption of innovative methods of conducting cost-effectiveness analysis by assigning each problem value captured from a literature review. This method is advantageous because it can increase study generalizability to inform decision making in a wide range of the population.33 The values assigned to treatment problems were previously reported by patient populations who actually developed the symptoms, and were from the US patients’ perspective. Therefore, these values theoretically provide a practical and accurate view of the treatment experience and treatment burden following opioid prescriptions in the United States.
An additional strength of our study included our examination of willingness to pay from both patients’ and payers’ perspectives. We followed the recommendation from the Second Panel on Cost-effectiveness in Health and Medicine that more studies are needed to investigate the health care costs from societal, patient’s and payer’s perspectives.34 In comparing both perspectives, we are able to identify potential benefits that PGx information would bring to both of the stakeholders. In particular, the health care payers had 5-fold higher willingness to pay than patients did; therefore, avoiding ADEs could potentially bring a much greater benefit from health care payers’ perspective than from patients’ perspective. However, we suggest that PGx implementation needs to be evaluated at both levels separately. In addition, we note that the health care payers’ willingness-to-pay threshold used in this study ($50,000 per QALY) was relatively low compared with the threshold of $100,000 per QALY that has been used recently in other cost-effectiveness analyses in developed countries.31 Therefore, the differences between patients perceived value of care and payers’ willingness to pay may be underestimated.
Our study results were limited by the scarce results from the current literature. We adopted the best possible evidence from the available literature, but some values could still be an underestimation or overestimation of true costs. For example, the value we assigned to tinnitus was $1085 per month. This value was adopted from a survey study in which the patient population was self-registered for a “Tinnitus Update” email listserv. Therefore, the responders could be patients with relatively severe tinnitus that impaired their social functions, who tend to look for more treatment information.35 Although we adopted the values reported from patients with the mildest tinnitus level, and only 2 of the patients in our population reported tinnitus in our study, this value of tinnitus could overestimate the average patient’s experience of tinnitus, and it might overestimate the average treatment value for tinnitus.
This study is also limited by the treatment problems captured through medical record review. As such, the problem needed to be shared with the health care provider and documented in the medical record. We therefore expect that our capture of treatment problems is likely an underestimate of the true number of problems that occur in patients after using tramadol or codeine. Fewer outcomes results in more limited power and a reduced ability to detect associations should they actually exist. Our results are therefore conservative, and we might have missed weaker associations. We also note that the patient is the best source of information for treatment problems, and future studies that collect treatment problem information at the time of treatment are needed to obtain complete information. Despite these limitations, we note that when compared with other studies of tramadol and codeine, our ADE rates are comparable. For example, the study by Nossol et al6 found that patients received tramadol developed nausea (3.4%), dizziness (1.5%), and vomiting (1.1%).
Finally, we conducted multiple tests of association, and considering associations as significant at P < .05 may be too permissive; however, P values are completely dependent on sample size, and they should be used only as a guide to highlight potentially interesting patterns.
Conclusion
We estimated willingness to pay related to treatment problems resulting from treatment with codeine and tramadol from both the patient and the payer perspective. Although overall costs were modest, we found that patients with a poor metabolizer CYP2D6 phenotype are likely to benefit most from PGx-guided prescribing designed to reduce treatment problems resulting from these prescription medications.
Supplemental Online Material
Supplemental Material
Acknowledgments
This research was partially supported by a research grant from 10.13039/100000871 Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery and the Mayo Clinic Center for Individualized Medicine. The views expressed in the article are those of the authors and do not necessarily reflect the views of Mayo Clinic.
Potential Competing Interests: Drs. Wang and Weinshilboum are cofounders and stockholders of OneOme, LLC, a pharmacogenomic decision support company.
All authors made substantial contributions to this study. Y.Z. and J.L.S. conceptualized and designed the study; Y.Z. performed the analysis, and drafted the manuscript; Y.Z., G.S.L., and J.L.S. finalized the results, interpretation of the data and the first draft of the manuscript; S.J.B., B.J.B., N.B.L., A.M.M., J.E.O., L.W., and R.W. provided critical comments for significant intellectual content; J.L.S. provided acquisition, supervision and funding support for this study, and acts as the study guarantor. All authors read and approved the manuscript for publication.
Supplemental material can be found online at http://mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. | CODEINE, TRAMADOL HYDROCHLORIDE | DrugsGivenReaction | CC BY-NC-ND | 33718782 | 18,874,449 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Oedema peripheral'. | Pulmonary Vascular Disease Due to Plasma Cell Dyscrasia.
Pulmonary hypertension (PH) has been described in myeloproliferative disorders; monoclonal plasma cell disorder such as polyneuropathy, organomegaly, endocrinopathy, monoclonal gammopathy, and skin changes syndrome; and plasma cell dyscrasias such as multiple myeloma and amyloidosis. We describe 4 cases of PH likely due to pulmonary vascular involvement and myocardial deposition from light chain deposition disease, amyloidosis, and multiple myeloma. On the basis of our clinical experience and literature review, we propose screening for plasma cell dyscrasia in patients with heart failure with preserved ejection fraction, unexplained PH, and hematological abnormalities. We also recommend inclusion of cardiopulmonary screening in patients with monoclonal gammopathy of undetermined significance.
Pulmonary hypertension (PH) is associated with myeloproliferative disorders, polyneuropathy, organomegaly, endocrinopathy, monoclonal gammopathy, and skin changes syndrome, multiple myeloma (MM), plasma cell leukemia, and amyloidosis.1, 2, 3, 4, 5, 6 The pathophysiology of PH in plasma cell dyscrasias could be due to pulmonary vascular deposition of amyloid fibrils or light chains, pulmonary parenchymal involvement (World Health Organization [WHO] group III), or due to left-sided heart disease from infiltrative cardiomyopathy (WHO group II).7 Reports of the reversibility of PH with the treatment of MM and polyneuropathy, organomegaly, endocrinopathy, monoclonal gammopathy, and skin changes syndrome have been reported.1,8
We report 4 cases of PH associated with amyloid light chain (AL) amyloidosis and MM. The study was approved by the University of Alabama at Birmingham Institutional Review Board, and a waiver of informed consent was obtained. The aim of this case series is 2-fold. First, we propose the incorporation of cardiopulmonary screening in patients with monoclonal gammopathy of undetermined significance (MGUS). Conversely, patients with heart failure with preserved ejection fraction or unexplained PH should be screened for plasma cell dyscrasia. Second, we propose the term “monoclonal gammopathy of cardiac or pulmonary significance” be coined to describe the cardiac and pulmonary manifestations for early initiation of treatment before end-organ function ensues. The term “monoclonal gammopathy of renal significance” was coined by the International Kidney and Monoclonal Gammopathy Research Group to describe renal dysfunction due to the deposition of monoclonal proteins in the kidneys. Despite the lack of criteria for the diagnosis of MM in patients with MGUS, recommendations are to treat the plasma cell disorder with monoclonal gammopathy of renal significance.
Case Reports
Case 1
A 43-year-old Black man with no medical history was referred for the management of severe PH. He developed exertional dyspnea 3 months before presentation. Right heart catheterization (RHC) details are summarized in Table 1, indicating severely elevated mean pulmonary artery pressure (PAP) and elevated right atrial (RA) pressures but normal cardiac index. Laboratory testing revealed normal renal function, elevated total bilirubin level, and elevated brain natriuretic peptide level. Transthoracic echocardiography (TTE) revealed dilated right ventricle (RV) with severely depressed RV systolic function, normal left ventricular (LV) systolic function, normal LV wall thickness, and moderate-sized pericardial effusion (Figure 1A). A computed tomography (CT) scan of the chest did not reveal interstitial lung disease (ILD) or pulmonary embolism. Connective tissue disease work-up was unrevealing. Infectious serologies for human immunodeficiency virus (HIV) and hepatitis were negative. Genetic testing did not reveal pathogenic sequence variation. Cardiac magnetic resonance imaging (MRI) performed during the initial diagnosis of PH did not find evidence of infiltrative disease. Treatment with phosphodiesterase-5 inhibitor and endothelin receptor antagonist for idiopathic pulmonary artery hypertension was initiated with improvement in functional capacity and hemodynamics. One year after diagnosis, there was evidence of worsening PAPs and RV systolic dysfunction by invasive testing even though the patient denied worsening functional capacity. Inhaled treprostinil was initiated. Repeat TTE revealed a mild increase in LV wall thickness, preserved LV systolic function, indeterminate LV diastolic function, severely depressed RV systolic function, and moderate-sized pericardial effusion (Figure 1B). Because of suspicion of infiltrative disease, serum free light chain (FLC) testing was ordered that found an elevated kappa FLC (κ FLC) of 184.6 mg/L and a normal lambda FLC (λ FLC) of 23.8, with an abnormal FLC ratio of 7.76 (normal range, 0.26-1.65). Serum monoclonal-spike was 1.81 g/dL. A bone marrow biopsy (BMB) revealed monoclonal plasma cells more than 20%. Congo red staining of BMB specimens was negative. The patient developed worsening renal dysfunction and nephrotic range proteinuria (Table 1, case 1). There was no laboratory evidence of anemia, hypercalcemia, or osteolytic lesions. A renal biopsy revealed global and segmental glomerulosclerosis, interstitial fibrosis, and tubular atrophy, with immunofluorescence staining positive for kappa light chain and Congo red staining negative for amyloid. Diagnosis of light chain (IgG κ) MM with light chain deposition disease (LCDD) involving kidneys, lungs, and heart was made. Treatment was initiated with bortezomib. He developed end-stage renal disease and began receiving hemodialysis. The patient transferred his care to another facility for the management of MM.Table 1 Summary of Clinical and Laboratory Data for the Described Clinical Casesa
Characteristic Case 1 Case 2 Case 3 Case 4
Age (y) 46 75 71 54
Sex Male Male Female Female
Race Black White White Black
Onset of symptoms before the diagnosis of plasma cell dyscrasia 2 y (Dyspnea)
Pulmonary artery hypertension 2 y (Dyspnea)
Pulmonary artery hypertension 1 y (Dyspnea)
Pulmonary artery hypertension 4 mo (Dyspnea, lower extremity edema)
Restrictive cardiomyopathy
Pulmonary artery hypertension
Diagnosis of MGUS No No Yes No
Laboratory data
BNP level (<100 pg/mL) 1433 979 1125 805
Troponin I level (<0.02 ng/mL) 0.05 <0.02 0.085 0.074
NT-proBNP level (<300 pg/mL) NA NA 23,899 NA
Troponin T level (<0.01 ng/mL) <0.01 NA 0.08 NA
Hemoglobin level (g/dL) 13 11.7 13.3 10.6
Calcium level (mg/dL) 9.7 9.4 9.8 9.6
Serum creatinine level (mg/dL) 1.5 2.2 1.7 0.8
Serum monoclonal-spike Present (1.81 g/dL) Present (4.1 g/dL) Absent Present (2 g/dL)
Kappa light chain (mg/L) 184 2266 19.7 1.5
Lambda light chain (mg/L) 23.8 3.1 948 519
Serum free light chain ratio 7.76 731.23 0.02 0.002
Urine monoclonal-spike Present Present Present Present
Urine protein level (24 h) 10 g 937 mg 248 mg NA
Immunoglobulins
IgA (66-436) 150 22 129 24
IgG (694-1618) 2693 5326 647 3067
IgM (45-281) 72 <10 29 <25
CRAB criteria Absent Present
Renal insufficiency Present
Renal insufficiency (eGFR <30 mL/min per 1.73 m2) Present
Anemia
Electrocardiography Normal sinus rhythm
RVH Atrial fibrillation
Right bundle branch block Normal sinus rhythm (82 beats/min)
Low QRS voltage
Right axis deviation
Poor R-wave progression Normal sinus rhythm
Poor R-wave progression
2D echocardiography Normal LV systolic function
Moderate concentric LVH
RV dilated and severely depressed function Normal LV systolic function and wall thickness
RV dilated and function severely depressed Moderate concentric LVH
LVEF 35%-40%
Small pericardial effusion Mild concentric LVH
LVEF >55%
RVSP 72 mm Hg
Normal RV systolic function
Right heart catheterization
RA pressure (mm Hg) 19 21 20 11
PA pressure (systolic/diastolic/mean) 87/30/50 88/41/57 79/34/50 61/25/40
PCWP (mm Hg) 10 16 26 18
Fick cardiac index (L/min per m2) 2.46 1.43 1.8 2.16
PVR (WU) 7.6
Precapillary 14
Combined pre- and postcapillary 5.7
Combined pre- and postcapillary 6
Combined pre- and postcapillary
Bone marrow biopsy
Plasma cells 20% monoclonal plasma cells 30% monoclonal plasma cells 10%-15% monoclonal plasma cells 90% plasma cells
Congo red stain Negative Negative Negative Not performed
Endomyocardial biopsy Not done Not done Congo red–positive deposits
Mass spectrometry: AL (lambda) type Not performed
Renal biopsy Global and segmental glomerulosclerosis, interstitial fibrosis, tubular atrophy Immunofluorescence staining positive for kappa LC; Congo red staining negative for amyloid NA NA NA
Treatment Bortezomib Bortezomib Bortezomib Induction chemotherapy (RVD)
Autologous bone marrow transplant
a 2D = 2-dimensional; BNP = brain natriuretic peptide; CRAB = criteria hypercalcemia (>11.5 mg/dL), renal insufficiency (serum creatinine level > 2 mg/dL or eGFR < 40 mL/min per 1.73 m2, anemia (hemoglobin level < 10 g/dL), bony lesion >1 site osteolytic lesions on skeletal survey; D = diastolic; eGFR = estimated glomerular filtration rate; LC = light chain; LV = left ventricular; LVEF = left ventricular ejection fraction; LVH = left ventricular hypertrophy; M = mean; MGUS = monoclonal gammopathy of undetermined significance; NA = not available; NT-proBNP = N-terminal pro–hormone brain natriuretic peptide; PA = pulmonary artery; PCWP = pulmonary capillary wedge pressure; PVR = pulmonary vascular resistance; RA = right atrium; RV = right ventricular; RVD = Revlimid + Velcade + dexamethasone; RVH = right ventricular hypertrophy; RVSP = right ventricular systolic pressure; S = systolic; WU = wood unit.
Figure 1 A, A 2-dimensional echocardiogram revealing right ventricular dilation and pericardial effusion. B, A year later, a 2-dimensional echocardiogram revealing severe left ventricular hypertrophy suggestive of infiltrative disease.
Case 2
A 75-year-old White man with a medical history of hypertension, chronic kidney disease (CKD) stage IV, and atrial fibrillation was referred to our PH clinic. Transthoracic echocardiography revealed normal LV wall thickness and systolic function, an elevated estimated RV systolic pressure of 90 mm Hg, severely dilated RV with depressed systolic function, and a small pericardial effusion. Right heart catheterization revealed severely elevated mean PAP, RA pressure, and low Fick cardiac index. because of pancytopenia and CKD, serum FLC testing was ordered. Serum FLC testing found an elevated κ FLC of 2266 mg/L and an λ FLC of 3.1 mg/L, with an elevated FLC ratio of 731. Serum monoclonal-spike was 4.10 g/dL (Table 1, case 2). Twenty-four–hour urine protein testing revealed approximately 1 g of protein. A bone marrow biopsy revealed monoclonal plasma cells with kappa light chain restriction more than 39% (Figure 2). Congo red staining was negative for amyloid. A CT scan of the chest did not find evidence of ILD. Connective tissue disease work-up and infectious serologies for HIV and hepatitis were negative. A ventilation/perfusion scan was not performed to rule out chronic thromboembolic pulmonary disease. Treatment with bortezomib was initiated for IgG κ light chain MM with LCDD involving kidneys and pulmonary vasculature. The patient also received dexamethasone because of suspicion of renal myeloma involvement. The patient developed acute on chronic renal failure, but because of family and patient’s wishes, hemodialysis was not initiated. Supportive care measures were instituted, and the patient died in hospice.Figure 2 Sections reveal a hypercellular marrow containing a large number of plasma cells (arrows).
Case 3
A 71-year-old White woman presented to our PH clinic for a second opinion on progressive exertional dyspnea over the last year. Family history was significant for MM in her mother. Her medical history included MGUS, hypertension, CKD (estimated glomerular filtration rate, 30 mL/min per 1.73 m2), obstructive sleep apnea adhering to nightly continuous positive airway pressure, Hashimoto disease, and celiac disease.
Diagnosis of lambda light chain MGUS (LC-MGUS) was first made in 2010; at that time, BMB was unremarkable and a fat pad biopsy was negative for amyloid. Serum electropheresis was performed every 6 months with stable kappa/lambda light chain ratio (0.02-0.03). The skeletal survey found no lytic lesions, and renal function was normal.
Her initial work-up for exertional dyspnea included chest CT angiography, which revealed a small pulmonary embolism for which she began receiving anticoagulation therapy. Despite periodic diuresis for volume overload and participation in cardiac rehabilitation, her functional capacity declined. Transthoracic echocardiography revealed an estimated pulmonary artery systolic pressure of 71 mm Hg, and she empirically began receiving riociguat by the referring physician.
Our assessment included an electrocardiogram that revealed sinus rhythm, left atrial enlargement, and low voltage. Transthoracic echocardiography revealed thickened myocardium with infiltrative appearance, dilated RV, preserved LV ejection fraction, and grade III diastolic dysfunction.
She was admitted to our hospital for expedited evaluation after outpatient RHC that exhibited elevated RA pressures and PAPs with low cardiac output (Table 1, case 3). Cardiac MRI found biventricular systolic dysfunction with diffuse mid-wall late gadolinium enhancement involving nearly all the LV myocardium and biatrial enlargement. Serum FLC testing found a normal κ FLC of 20 mg/L and an elevated λ FLC of 948 mg/L, with a ratio of 0.02. Repeat BMB revealed plasma cell dyscrasia and mildly hypocellular marrow, and Congo red staining was negative for amyloid.
With a high suspicion of plasma cell dyscrasia leading to an infiltrative cardiomyopathy, she then underwent endomyocardial biopsy. Histopathological evaluation revealed mild interstitial widening, no significant fibrosis, and Congo red staining suggestive of amyloid (Figure 3). Liquid chromatography tandem mass spectrometry confirmed AL (lambda)–type amyloid deposition. She was offered bortezomib, which, over the next few weeks, led to a 50% reduction in λ FLC. Unfortunately, despite this, her clinical condition deteriorated, and she was readmitted multiple times for heart failure and volume overload. She was ultimately discharged home with hospice 2 months after her initial diagnosis and died.Figure 3 A, Congo red–stained section reveals amorphous material consistent with amyloid in the myocardial interstitium (arrows). B, Under polarized light, apple green birefringence is seen (arrows).
Case 4
A 54-year-old Black woman with a history of hyperthyroidism received a diagnosis of IgG λ MM after routine blood work revealed elevated serum protein. The ventilation/perfusion scan revealed low probability for pulmonary embolism. There was no evidence of ILD on the CT scan of the chest. She underwent induction chemotherapy with lenalidomide, bortezomib, and dexamethasone. She had to begin receiving diuretics after the initiation of chemotherapy because of lower extremity edema. Transthoracic echocardiography performed before autologous stem cell transplant (ASCT) revealed mild concentric LV hypertrophy with preserved ejection fraction, grade III diastolic dysfunction, an elevated estimated PAP of greater than 70 mm Hg, and normal RV systolic function (Table 1, case 4). Cardiac MRI did not find any evidence of infiltrative disease. She underwent ASCT 5 months after diagnosis. She had episodes of volume overload that were treated with an increased dose of diuretics. The patient began receiving phosphodiesterase-5 inhibitor a few months after ASCT because of persistent PH on the basis of RHC (Table 1, case 4). Six years after ASCT, the patient still has good function capacity without any admissions for volume overload. Repeat TTE 5 years after ASCT still revealed persistent PH on the basis of the estimated RV systolic pressure of approximately 70 mm Hg without RV dysfunction.
Discussion
The 4 cases described here highlight the cardiopulmonary manifestations of plasma cell dyscrasia. Patients described in cases 1 and 2 received a diagnosis of idiopathic pulmonary artery hypertension before a clinical diagnosis of MM. The patient in case 3 had a history of MGUS for 10 years and PH diagnosis for a year preceding the diagnosis of AL amyloidosis. The fourth patient had received a diagnosis of PH after MM diagnosis. Work-up for secondary causes of PH, such as connective tissue disease, infectious diseases such as HIV and chronic hepatitis C, and ILD, was negative in all 4 patients. Three patients had cardiac, pulmonary, and renal involvement, but the clinical presentation was consistent with PH and RV failure.
The pulmonary manifestations of AL amyloidosis or LCDD include diffuse alveolar septal or localized nodular involvement. Pulmonary hypertension due to AL amyloidosis results from restrictive cardiomyopathy, pulmonary parenchymal involvement, or deposition of amyloid fibrils in the pulmonary vasculature. Pulmonary amyloidosis is seen predominantly owing to AL type, and the diagnosis was made postmortem in more than 90% of cases in an autopsy series. Pulmonary vascular involvement was seen in 97% of cases followed by alveolar septal deposition (78%).9 Deposition of amyloid fibrils in the media of small arteries, arterioles, and veins in multiple organs has been described.10 However, the pathogenicity of deposition of amyloid fibrils in the pulmonary vasculature is unclear. There have been reported cases of PH due to AL amyloidosis or MM and is a rare clinical presentation.3,4,7 Our case series add to the existing scarce literature on the pulmonary vascular involvement in plasma cell dyscrasias. The revised WHO classification of PH does not include monoclonal gammopathy, AL amyloidosis, or LCDD as a possible etiology for PH.2
Light chain MGUS accounts for 20% of cases of MM and 65% of AL amyloidosis.11 The prevalence of LC-MGUS is 0.8% in Whites 50 years or older, which is much lower than that of MGUS with heavy chain expression.12 Although the prevalence of MGUS in Blacks is 3 times that in Whites, the exact prevalence of LC-MGUS in Blacks is unknown.13 The progression of LC-MGUS to MM is only 0.3% per year. The evolution of LC-MGUS to MM, AL amyloidosis, or LCDD may depend on the genetics of plasma cell clone, amyloidogenesis, and tissue microenvironment.14 The plasma cell clone secreting light chains may have low malignant potential but can be of clinical significance, with damage to kidneys, bone, skin, and peripheral nervous system.15 The most common organs involved in AL amyloidosis is the heart (71%), kidney (58%), and peripheral nervous system (23%).16 Cardiac involvement portends a poor prognosis with a median survival of 6 months but often diagnosis is delayed.17 Ten to fifteen percent of patients with MM and 9% of patients with MGUS develop AL amyloidosis.18 Patients with LCDD may have coexistent MM (10%-30%) and AL amyloidosis (17%).19 Light chain deposition disease with cardiac involvement (33%) has overall poor survival and is related to increased transplant-related mortality. Only 18% of patients had LV ejection fraction less than 50%, and most frequent manifestations are atrial arrhythmias, sinus bradycardia, and diastolic dysfunction. Light chain deposition disease with coexisting amyloid is commonly associated with increased production of lambda light chains (40%), whereas LCDD is predominantly kappa restricted.
The transformation of LC-MGUS to AL amyloidosis or LCDD may occur over a period during which time occult damage to kidneys, heart, lung, and peripheral nervous system may occur. Proposed high-risk criteria for progression include involved/uninvolved FLC ratio less than 0.125 or greater than 8.0 with progression rates of more than 8% per year.20 The risk of progression is 3 times higher with an abnormal FLC ratio irrespective of the type and size of serum monoclonal M protein. Presence of combined abnormal serum FLC ratio, non-IgG MGUS, and elevated M protein level (>15 g/L) predicts the highest risk of progression (58% at 20 years).21
We propose that LC-MGUS screening should include cardiopulmonary testing. We recommend screening for cardiac involvement with routine biomarkers such as N-terminal pro-brain natriuretic peptide, brain natriuretic peptide, cardiac troponin I or T, and electrocardiography followed by TTE if serum biomarkers are abnormal. Detailed TTE evaluation of LV morphology, LV systolic and diastolic function, RV systolic function, and estimated pulmonary artery systolic pressure should be performed. Diagnosis of pulmonary involvement can be challenging because of the nonspecific radiological findings. A transbronchial or open lung biopsy may be required for confirmation of pulmonary amyloidosis. Thoracentesis with pleural fluid FLC assay can be performed for patients with recurrent pleural effusions (Table 2).Table 2 Diagnostic Criteria for LC-MGUS
Serum free light chain ratio (<0.26 or >1.65)
Elevated monoclonal kappa or lambda light chains
Monoclonal bone marrow plasma cells <10%
Urinary monoclonal protein level < 500 mg/24 h
Absence of immunoglobulin heavy chain expression on immunofixation
No evidence of end-organ damage (hypercalcemia, renal insufficiency, anemia, and bone lesions)
Proposed organ inclusion for screening with noninvasive testing:
Heart (restrictive cardiomyopathy, HFpEF, and conduction system abnormalities)—ECG; cardiac biomarkers—NT-proBNP or BNP and cardiac troponin T or I, 2D echocardiography with speckle tracking
Pulmonary (vascular—PH) (parenchymal—nodular, diffuse alveolar septal, intrathoracic lymphadenopathy, laryngeal, and tracheobronchial)
Chest radiography. CT scan of the chest if indicated or suspicion high (dyspnea, cough, or abnormal chest radiography). Bronchoscopy with biopsy. Thoracentesis and pleural fluid free light chain assay
2D = 2-dimensional; BNP = brain natriuretic peptide; CT = omputed tomography; ECG = electrocardiogram; HFpEF = heart failure with preserved ejection fraction; LC-MGUS = light chain monoclonal gammopathy of undetermined significance; NT-proBNP = N-terminal pro–B-type natriuretic peptide; PH = pulmonary hypertension.
Conclusion
We describe our experience of 2 cases of IgG κ MM, 1 case of IgG λ MM, and 1 case of AL amyloidosis as etiologies for PH. Early diagnosis and timely treatment of plasma cell dyscrasia will improve survival, and select patients may be candidates for solid organ transplantation. We recommend serum FLC testing in patients with heart failure with preserved ejection fraction or unexplained PH. Incorporation of cardiopulmonary involvement to the “CRAB (Hypercalcemia, renal insufficiency, Anemia and Lytic Bone lesions)” criteria may detect early disease and timely treatment to prolong survival. We propose the term “monoclonal gammopathy of cardiac or pulmonary significance” be coined to describe the cardiac and pulmonary manifestations for early initiation of treatment before end-organ function ensues.
Acknowledgments
Drs Rajapreyar and Joly contributed equally to this work.
Potential Competing Interests: The authors report no competing interests. | BORTEZOMIB, DEXAMETHASONE, LENALIDOMIDE | DrugsGivenReaction | CC BY-NC-ND | 33718795 | 19,459,856 | 2021-02 |
What was the administration route of drug 'LENALIDOMIDE'? | Pulmonary Vascular Disease Due to Plasma Cell Dyscrasia.
Pulmonary hypertension (PH) has been described in myeloproliferative disorders; monoclonal plasma cell disorder such as polyneuropathy, organomegaly, endocrinopathy, monoclonal gammopathy, and skin changes syndrome; and plasma cell dyscrasias such as multiple myeloma and amyloidosis. We describe 4 cases of PH likely due to pulmonary vascular involvement and myocardial deposition from light chain deposition disease, amyloidosis, and multiple myeloma. On the basis of our clinical experience and literature review, we propose screening for plasma cell dyscrasia in patients with heart failure with preserved ejection fraction, unexplained PH, and hematological abnormalities. We also recommend inclusion of cardiopulmonary screening in patients with monoclonal gammopathy of undetermined significance.
Pulmonary hypertension (PH) is associated with myeloproliferative disorders, polyneuropathy, organomegaly, endocrinopathy, monoclonal gammopathy, and skin changes syndrome, multiple myeloma (MM), plasma cell leukemia, and amyloidosis.1, 2, 3, 4, 5, 6 The pathophysiology of PH in plasma cell dyscrasias could be due to pulmonary vascular deposition of amyloid fibrils or light chains, pulmonary parenchymal involvement (World Health Organization [WHO] group III), or due to left-sided heart disease from infiltrative cardiomyopathy (WHO group II).7 Reports of the reversibility of PH with the treatment of MM and polyneuropathy, organomegaly, endocrinopathy, monoclonal gammopathy, and skin changes syndrome have been reported.1,8
We report 4 cases of PH associated with amyloid light chain (AL) amyloidosis and MM. The study was approved by the University of Alabama at Birmingham Institutional Review Board, and a waiver of informed consent was obtained. The aim of this case series is 2-fold. First, we propose the incorporation of cardiopulmonary screening in patients with monoclonal gammopathy of undetermined significance (MGUS). Conversely, patients with heart failure with preserved ejection fraction or unexplained PH should be screened for plasma cell dyscrasia. Second, we propose the term “monoclonal gammopathy of cardiac or pulmonary significance” be coined to describe the cardiac and pulmonary manifestations for early initiation of treatment before end-organ function ensues. The term “monoclonal gammopathy of renal significance” was coined by the International Kidney and Monoclonal Gammopathy Research Group to describe renal dysfunction due to the deposition of monoclonal proteins in the kidneys. Despite the lack of criteria for the diagnosis of MM in patients with MGUS, recommendations are to treat the plasma cell disorder with monoclonal gammopathy of renal significance.
Case Reports
Case 1
A 43-year-old Black man with no medical history was referred for the management of severe PH. He developed exertional dyspnea 3 months before presentation. Right heart catheterization (RHC) details are summarized in Table 1, indicating severely elevated mean pulmonary artery pressure (PAP) and elevated right atrial (RA) pressures but normal cardiac index. Laboratory testing revealed normal renal function, elevated total bilirubin level, and elevated brain natriuretic peptide level. Transthoracic echocardiography (TTE) revealed dilated right ventricle (RV) with severely depressed RV systolic function, normal left ventricular (LV) systolic function, normal LV wall thickness, and moderate-sized pericardial effusion (Figure 1A). A computed tomography (CT) scan of the chest did not reveal interstitial lung disease (ILD) or pulmonary embolism. Connective tissue disease work-up was unrevealing. Infectious serologies for human immunodeficiency virus (HIV) and hepatitis were negative. Genetic testing did not reveal pathogenic sequence variation. Cardiac magnetic resonance imaging (MRI) performed during the initial diagnosis of PH did not find evidence of infiltrative disease. Treatment with phosphodiesterase-5 inhibitor and endothelin receptor antagonist for idiopathic pulmonary artery hypertension was initiated with improvement in functional capacity and hemodynamics. One year after diagnosis, there was evidence of worsening PAPs and RV systolic dysfunction by invasive testing even though the patient denied worsening functional capacity. Inhaled treprostinil was initiated. Repeat TTE revealed a mild increase in LV wall thickness, preserved LV systolic function, indeterminate LV diastolic function, severely depressed RV systolic function, and moderate-sized pericardial effusion (Figure 1B). Because of suspicion of infiltrative disease, serum free light chain (FLC) testing was ordered that found an elevated kappa FLC (κ FLC) of 184.6 mg/L and a normal lambda FLC (λ FLC) of 23.8, with an abnormal FLC ratio of 7.76 (normal range, 0.26-1.65). Serum monoclonal-spike was 1.81 g/dL. A bone marrow biopsy (BMB) revealed monoclonal plasma cells more than 20%. Congo red staining of BMB specimens was negative. The patient developed worsening renal dysfunction and nephrotic range proteinuria (Table 1, case 1). There was no laboratory evidence of anemia, hypercalcemia, or osteolytic lesions. A renal biopsy revealed global and segmental glomerulosclerosis, interstitial fibrosis, and tubular atrophy, with immunofluorescence staining positive for kappa light chain and Congo red staining negative for amyloid. Diagnosis of light chain (IgG κ) MM with light chain deposition disease (LCDD) involving kidneys, lungs, and heart was made. Treatment was initiated with bortezomib. He developed end-stage renal disease and began receiving hemodialysis. The patient transferred his care to another facility for the management of MM.Table 1 Summary of Clinical and Laboratory Data for the Described Clinical Casesa
Characteristic Case 1 Case 2 Case 3 Case 4
Age (y) 46 75 71 54
Sex Male Male Female Female
Race Black White White Black
Onset of symptoms before the diagnosis of plasma cell dyscrasia 2 y (Dyspnea)
Pulmonary artery hypertension 2 y (Dyspnea)
Pulmonary artery hypertension 1 y (Dyspnea)
Pulmonary artery hypertension 4 mo (Dyspnea, lower extremity edema)
Restrictive cardiomyopathy
Pulmonary artery hypertension
Diagnosis of MGUS No No Yes No
Laboratory data
BNP level (<100 pg/mL) 1433 979 1125 805
Troponin I level (<0.02 ng/mL) 0.05 <0.02 0.085 0.074
NT-proBNP level (<300 pg/mL) NA NA 23,899 NA
Troponin T level (<0.01 ng/mL) <0.01 NA 0.08 NA
Hemoglobin level (g/dL) 13 11.7 13.3 10.6
Calcium level (mg/dL) 9.7 9.4 9.8 9.6
Serum creatinine level (mg/dL) 1.5 2.2 1.7 0.8
Serum monoclonal-spike Present (1.81 g/dL) Present (4.1 g/dL) Absent Present (2 g/dL)
Kappa light chain (mg/L) 184 2266 19.7 1.5
Lambda light chain (mg/L) 23.8 3.1 948 519
Serum free light chain ratio 7.76 731.23 0.02 0.002
Urine monoclonal-spike Present Present Present Present
Urine protein level (24 h) 10 g 937 mg 248 mg NA
Immunoglobulins
IgA (66-436) 150 22 129 24
IgG (694-1618) 2693 5326 647 3067
IgM (45-281) 72 <10 29 <25
CRAB criteria Absent Present
Renal insufficiency Present
Renal insufficiency (eGFR <30 mL/min per 1.73 m2) Present
Anemia
Electrocardiography Normal sinus rhythm
RVH Atrial fibrillation
Right bundle branch block Normal sinus rhythm (82 beats/min)
Low QRS voltage
Right axis deviation
Poor R-wave progression Normal sinus rhythm
Poor R-wave progression
2D echocardiography Normal LV systolic function
Moderate concentric LVH
RV dilated and severely depressed function Normal LV systolic function and wall thickness
RV dilated and function severely depressed Moderate concentric LVH
LVEF 35%-40%
Small pericardial effusion Mild concentric LVH
LVEF >55%
RVSP 72 mm Hg
Normal RV systolic function
Right heart catheterization
RA pressure (mm Hg) 19 21 20 11
PA pressure (systolic/diastolic/mean) 87/30/50 88/41/57 79/34/50 61/25/40
PCWP (mm Hg) 10 16 26 18
Fick cardiac index (L/min per m2) 2.46 1.43 1.8 2.16
PVR (WU) 7.6
Precapillary 14
Combined pre- and postcapillary 5.7
Combined pre- and postcapillary 6
Combined pre- and postcapillary
Bone marrow biopsy
Plasma cells 20% monoclonal plasma cells 30% monoclonal plasma cells 10%-15% monoclonal plasma cells 90% plasma cells
Congo red stain Negative Negative Negative Not performed
Endomyocardial biopsy Not done Not done Congo red–positive deposits
Mass spectrometry: AL (lambda) type Not performed
Renal biopsy Global and segmental glomerulosclerosis, interstitial fibrosis, tubular atrophy Immunofluorescence staining positive for kappa LC; Congo red staining negative for amyloid NA NA NA
Treatment Bortezomib Bortezomib Bortezomib Induction chemotherapy (RVD)
Autologous bone marrow transplant
a 2D = 2-dimensional; BNP = brain natriuretic peptide; CRAB = criteria hypercalcemia (>11.5 mg/dL), renal insufficiency (serum creatinine level > 2 mg/dL or eGFR < 40 mL/min per 1.73 m2, anemia (hemoglobin level < 10 g/dL), bony lesion >1 site osteolytic lesions on skeletal survey; D = diastolic; eGFR = estimated glomerular filtration rate; LC = light chain; LV = left ventricular; LVEF = left ventricular ejection fraction; LVH = left ventricular hypertrophy; M = mean; MGUS = monoclonal gammopathy of undetermined significance; NA = not available; NT-proBNP = N-terminal pro–hormone brain natriuretic peptide; PA = pulmonary artery; PCWP = pulmonary capillary wedge pressure; PVR = pulmonary vascular resistance; RA = right atrium; RV = right ventricular; RVD = Revlimid + Velcade + dexamethasone; RVH = right ventricular hypertrophy; RVSP = right ventricular systolic pressure; S = systolic; WU = wood unit.
Figure 1 A, A 2-dimensional echocardiogram revealing right ventricular dilation and pericardial effusion. B, A year later, a 2-dimensional echocardiogram revealing severe left ventricular hypertrophy suggestive of infiltrative disease.
Case 2
A 75-year-old White man with a medical history of hypertension, chronic kidney disease (CKD) stage IV, and atrial fibrillation was referred to our PH clinic. Transthoracic echocardiography revealed normal LV wall thickness and systolic function, an elevated estimated RV systolic pressure of 90 mm Hg, severely dilated RV with depressed systolic function, and a small pericardial effusion. Right heart catheterization revealed severely elevated mean PAP, RA pressure, and low Fick cardiac index. because of pancytopenia and CKD, serum FLC testing was ordered. Serum FLC testing found an elevated κ FLC of 2266 mg/L and an λ FLC of 3.1 mg/L, with an elevated FLC ratio of 731. Serum monoclonal-spike was 4.10 g/dL (Table 1, case 2). Twenty-four–hour urine protein testing revealed approximately 1 g of protein. A bone marrow biopsy revealed monoclonal plasma cells with kappa light chain restriction more than 39% (Figure 2). Congo red staining was negative for amyloid. A CT scan of the chest did not find evidence of ILD. Connective tissue disease work-up and infectious serologies for HIV and hepatitis were negative. A ventilation/perfusion scan was not performed to rule out chronic thromboembolic pulmonary disease. Treatment with bortezomib was initiated for IgG κ light chain MM with LCDD involving kidneys and pulmonary vasculature. The patient also received dexamethasone because of suspicion of renal myeloma involvement. The patient developed acute on chronic renal failure, but because of family and patient’s wishes, hemodialysis was not initiated. Supportive care measures were instituted, and the patient died in hospice.Figure 2 Sections reveal a hypercellular marrow containing a large number of plasma cells (arrows).
Case 3
A 71-year-old White woman presented to our PH clinic for a second opinion on progressive exertional dyspnea over the last year. Family history was significant for MM in her mother. Her medical history included MGUS, hypertension, CKD (estimated glomerular filtration rate, 30 mL/min per 1.73 m2), obstructive sleep apnea adhering to nightly continuous positive airway pressure, Hashimoto disease, and celiac disease.
Diagnosis of lambda light chain MGUS (LC-MGUS) was first made in 2010; at that time, BMB was unremarkable and a fat pad biopsy was negative for amyloid. Serum electropheresis was performed every 6 months with stable kappa/lambda light chain ratio (0.02-0.03). The skeletal survey found no lytic lesions, and renal function was normal.
Her initial work-up for exertional dyspnea included chest CT angiography, which revealed a small pulmonary embolism for which she began receiving anticoagulation therapy. Despite periodic diuresis for volume overload and participation in cardiac rehabilitation, her functional capacity declined. Transthoracic echocardiography revealed an estimated pulmonary artery systolic pressure of 71 mm Hg, and she empirically began receiving riociguat by the referring physician.
Our assessment included an electrocardiogram that revealed sinus rhythm, left atrial enlargement, and low voltage. Transthoracic echocardiography revealed thickened myocardium with infiltrative appearance, dilated RV, preserved LV ejection fraction, and grade III diastolic dysfunction.
She was admitted to our hospital for expedited evaluation after outpatient RHC that exhibited elevated RA pressures and PAPs with low cardiac output (Table 1, case 3). Cardiac MRI found biventricular systolic dysfunction with diffuse mid-wall late gadolinium enhancement involving nearly all the LV myocardium and biatrial enlargement. Serum FLC testing found a normal κ FLC of 20 mg/L and an elevated λ FLC of 948 mg/L, with a ratio of 0.02. Repeat BMB revealed plasma cell dyscrasia and mildly hypocellular marrow, and Congo red staining was negative for amyloid.
With a high suspicion of plasma cell dyscrasia leading to an infiltrative cardiomyopathy, she then underwent endomyocardial biopsy. Histopathological evaluation revealed mild interstitial widening, no significant fibrosis, and Congo red staining suggestive of amyloid (Figure 3). Liquid chromatography tandem mass spectrometry confirmed AL (lambda)–type amyloid deposition. She was offered bortezomib, which, over the next few weeks, led to a 50% reduction in λ FLC. Unfortunately, despite this, her clinical condition deteriorated, and she was readmitted multiple times for heart failure and volume overload. She was ultimately discharged home with hospice 2 months after her initial diagnosis and died.Figure 3 A, Congo red–stained section reveals amorphous material consistent with amyloid in the myocardial interstitium (arrows). B, Under polarized light, apple green birefringence is seen (arrows).
Case 4
A 54-year-old Black woman with a history of hyperthyroidism received a diagnosis of IgG λ MM after routine blood work revealed elevated serum protein. The ventilation/perfusion scan revealed low probability for pulmonary embolism. There was no evidence of ILD on the CT scan of the chest. She underwent induction chemotherapy with lenalidomide, bortezomib, and dexamethasone. She had to begin receiving diuretics after the initiation of chemotherapy because of lower extremity edema. Transthoracic echocardiography performed before autologous stem cell transplant (ASCT) revealed mild concentric LV hypertrophy with preserved ejection fraction, grade III diastolic dysfunction, an elevated estimated PAP of greater than 70 mm Hg, and normal RV systolic function (Table 1, case 4). Cardiac MRI did not find any evidence of infiltrative disease. She underwent ASCT 5 months after diagnosis. She had episodes of volume overload that were treated with an increased dose of diuretics. The patient began receiving phosphodiesterase-5 inhibitor a few months after ASCT because of persistent PH on the basis of RHC (Table 1, case 4). Six years after ASCT, the patient still has good function capacity without any admissions for volume overload. Repeat TTE 5 years after ASCT still revealed persistent PH on the basis of the estimated RV systolic pressure of approximately 70 mm Hg without RV dysfunction.
Discussion
The 4 cases described here highlight the cardiopulmonary manifestations of plasma cell dyscrasia. Patients described in cases 1 and 2 received a diagnosis of idiopathic pulmonary artery hypertension before a clinical diagnosis of MM. The patient in case 3 had a history of MGUS for 10 years and PH diagnosis for a year preceding the diagnosis of AL amyloidosis. The fourth patient had received a diagnosis of PH after MM diagnosis. Work-up for secondary causes of PH, such as connective tissue disease, infectious diseases such as HIV and chronic hepatitis C, and ILD, was negative in all 4 patients. Three patients had cardiac, pulmonary, and renal involvement, but the clinical presentation was consistent with PH and RV failure.
The pulmonary manifestations of AL amyloidosis or LCDD include diffuse alveolar septal or localized nodular involvement. Pulmonary hypertension due to AL amyloidosis results from restrictive cardiomyopathy, pulmonary parenchymal involvement, or deposition of amyloid fibrils in the pulmonary vasculature. Pulmonary amyloidosis is seen predominantly owing to AL type, and the diagnosis was made postmortem in more than 90% of cases in an autopsy series. Pulmonary vascular involvement was seen in 97% of cases followed by alveolar septal deposition (78%).9 Deposition of amyloid fibrils in the media of small arteries, arterioles, and veins in multiple organs has been described.10 However, the pathogenicity of deposition of amyloid fibrils in the pulmonary vasculature is unclear. There have been reported cases of PH due to AL amyloidosis or MM and is a rare clinical presentation.3,4,7 Our case series add to the existing scarce literature on the pulmonary vascular involvement in plasma cell dyscrasias. The revised WHO classification of PH does not include monoclonal gammopathy, AL amyloidosis, or LCDD as a possible etiology for PH.2
Light chain MGUS accounts for 20% of cases of MM and 65% of AL amyloidosis.11 The prevalence of LC-MGUS is 0.8% in Whites 50 years or older, which is much lower than that of MGUS with heavy chain expression.12 Although the prevalence of MGUS in Blacks is 3 times that in Whites, the exact prevalence of LC-MGUS in Blacks is unknown.13 The progression of LC-MGUS to MM is only 0.3% per year. The evolution of LC-MGUS to MM, AL amyloidosis, or LCDD may depend on the genetics of plasma cell clone, amyloidogenesis, and tissue microenvironment.14 The plasma cell clone secreting light chains may have low malignant potential but can be of clinical significance, with damage to kidneys, bone, skin, and peripheral nervous system.15 The most common organs involved in AL amyloidosis is the heart (71%), kidney (58%), and peripheral nervous system (23%).16 Cardiac involvement portends a poor prognosis with a median survival of 6 months but often diagnosis is delayed.17 Ten to fifteen percent of patients with MM and 9% of patients with MGUS develop AL amyloidosis.18 Patients with LCDD may have coexistent MM (10%-30%) and AL amyloidosis (17%).19 Light chain deposition disease with cardiac involvement (33%) has overall poor survival and is related to increased transplant-related mortality. Only 18% of patients had LV ejection fraction less than 50%, and most frequent manifestations are atrial arrhythmias, sinus bradycardia, and diastolic dysfunction. Light chain deposition disease with coexisting amyloid is commonly associated with increased production of lambda light chains (40%), whereas LCDD is predominantly kappa restricted.
The transformation of LC-MGUS to AL amyloidosis or LCDD may occur over a period during which time occult damage to kidneys, heart, lung, and peripheral nervous system may occur. Proposed high-risk criteria for progression include involved/uninvolved FLC ratio less than 0.125 or greater than 8.0 with progression rates of more than 8% per year.20 The risk of progression is 3 times higher with an abnormal FLC ratio irrespective of the type and size of serum monoclonal M protein. Presence of combined abnormal serum FLC ratio, non-IgG MGUS, and elevated M protein level (>15 g/L) predicts the highest risk of progression (58% at 20 years).21
We propose that LC-MGUS screening should include cardiopulmonary testing. We recommend screening for cardiac involvement with routine biomarkers such as N-terminal pro-brain natriuretic peptide, brain natriuretic peptide, cardiac troponin I or T, and electrocardiography followed by TTE if serum biomarkers are abnormal. Detailed TTE evaluation of LV morphology, LV systolic and diastolic function, RV systolic function, and estimated pulmonary artery systolic pressure should be performed. Diagnosis of pulmonary involvement can be challenging because of the nonspecific radiological findings. A transbronchial or open lung biopsy may be required for confirmation of pulmonary amyloidosis. Thoracentesis with pleural fluid FLC assay can be performed for patients with recurrent pleural effusions (Table 2).Table 2 Diagnostic Criteria for LC-MGUS
Serum free light chain ratio (<0.26 or >1.65)
Elevated monoclonal kappa or lambda light chains
Monoclonal bone marrow plasma cells <10%
Urinary monoclonal protein level < 500 mg/24 h
Absence of immunoglobulin heavy chain expression on immunofixation
No evidence of end-organ damage (hypercalcemia, renal insufficiency, anemia, and bone lesions)
Proposed organ inclusion for screening with noninvasive testing:
Heart (restrictive cardiomyopathy, HFpEF, and conduction system abnormalities)—ECG; cardiac biomarkers—NT-proBNP or BNP and cardiac troponin T or I, 2D echocardiography with speckle tracking
Pulmonary (vascular—PH) (parenchymal—nodular, diffuse alveolar septal, intrathoracic lymphadenopathy, laryngeal, and tracheobronchial)
Chest radiography. CT scan of the chest if indicated or suspicion high (dyspnea, cough, or abnormal chest radiography). Bronchoscopy with biopsy. Thoracentesis and pleural fluid free light chain assay
2D = 2-dimensional; BNP = brain natriuretic peptide; CT = omputed tomography; ECG = electrocardiogram; HFpEF = heart failure with preserved ejection fraction; LC-MGUS = light chain monoclonal gammopathy of undetermined significance; NT-proBNP = N-terminal pro–B-type natriuretic peptide; PH = pulmonary hypertension.
Conclusion
We describe our experience of 2 cases of IgG κ MM, 1 case of IgG λ MM, and 1 case of AL amyloidosis as etiologies for PH. Early diagnosis and timely treatment of plasma cell dyscrasia will improve survival, and select patients may be candidates for solid organ transplantation. We recommend serum FLC testing in patients with heart failure with preserved ejection fraction or unexplained PH. Incorporation of cardiopulmonary involvement to the “CRAB (Hypercalcemia, renal insufficiency, Anemia and Lytic Bone lesions)” criteria may detect early disease and timely treatment to prolong survival. We propose the term “monoclonal gammopathy of cardiac or pulmonary significance” be coined to describe the cardiac and pulmonary manifestations for early initiation of treatment before end-organ function ensues.
Acknowledgments
Drs Rajapreyar and Joly contributed equally to this work.
Potential Competing Interests: The authors report no competing interests. | Oral | DrugAdministrationRoute | CC BY-NC-ND | 33718795 | 19,459,856 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Cardiac tamponade'. | Timed Controlled Repeated Rotation of the CAR-170-C NXSTAGE Chronic Cartridge Hemodialysis Filter: A Novel Approach to Enabling Heparin-Free Frequent Daily Home Hemodialysis.
Heparin-free hemodialysis is often warranted in postoperative states, bleeding diathesis, and critically ill patients. Conventionally, this is achieved through normal saline flushes or regional citrate anticoagulation. An 87-year-old white man with end-stage renal disease and atrial fibrillation, who was taking warfarin and using maintenance home hemodialysis (HHD) with an NxStage machine, underwent laparoscopic appendicectomy. The procedure was complicated by intra-abdominal abscess, sepsis, and tamponade from a bloody pericardial effusion. He needed emergent therapeutic pericardiocentesis. Warfarin was promptly discontinued. He was discharged home with heparin-free HHD. Prior heparin anticoagulation for HHD was an initial bolus of 4000 units of heparin. He continued to clot his extracorporeal system with resultant very high venous pressures and compromised HHD. Heparin anticoagulation was still contraindicated. Flushes with 250-500 mL normal saline, delivered in aliquots every 15-30 minutes, failed to prevent the frequent clotting. The first author, our HD Senior Technician, had astutely observed that the horizontally placed hemodialysis filter exhibited early "clot" formation at the 12-o'clock position, despite the saline flushes. Through trial and error, he discovered that rotating the horizontally placed hemodialysis filter along its long axis, 60 degrees clockwise for 15 minutes, return to the neutral position for 15 minutes, rotating the filter another 60 degrees counterclockwise for 15 minutes, with this repeated cycle of rotations "did the trick." It promptly and consistently resolved the clotting problem. The lines stopped clotting, and he has not needed saline flushes for smooth heparin-free HHD for more than 7 months. To our knowledge, this is the first such report. Further study is justified. We have hypothesized a mechanism and have named this the "Locke-Onuigbo Maneuver."
Generally, anticoagulation is necessary for hemodialysis (HD) lasting more than 1-2 hours, because blood contact with the extracorporeal circuit results in activation of the intrinsic coagulation pathway, leading to thrombogenesis.1, 2, 3, 4 Nevertheless, heparin-free hemodialysis is often warranted as in postoperative states, in patients with bleeding diathesis, and in critically ill patients in the intensive care unit.5, 6, 7, 8, 9 Conventionally, this is achieved through normal saline flushes, or with the application of regional citrate anticoagulation.1,6, 7, 8, 9 We recently encountered a set of converging clinical scenarios in which a male patient on home hemodialysis (HHD) using an NxStage machine had to be discharged home from hospital on heparin-free HHD; despite large-volume saline flushes, he had continued to clot his extracorporeal circuits and filters with resulting inadequate HHD. The use of heparin or warfarin was contraindicated, and the patient and the caregiver (wife) were considering stopping HHD. Indeed, there was some discussion about a possible switch to peritoneal dialysis. Thanks to the observations of our Senior HD technician, the first author, we, the last author working closely together with the first author, found a unique solution to this problem. We have mastered a novel approach to performgin sustainable, heparin-free HHD by intermittently rotating the horizontally aligned CAR-170-C NXSTAGE Chronic cartridge clockwise and counterclockwise, as described herein. The CAR-170-C NXSTAGE Chronic cartridges are single use, are gamma sterilized, include an integrated dialyzer, and are manufactured from glycerin-free polyethersulfone membranes (PUREMA H) known for biocompatibility.10 The filter comes with the dialysis lines from the manufacturer as a preattached dialyzer.10
Case Report
An 87-year-old white man with end-stage renal disease and atrial fibrillation, who was also using warfarin and maintenance HHD with an NxStage machine, underwent laparoscopic appendicectomy. The procedure was complicated by intra-abdominal abscess, sepsis, and tamponade from a bloody pericardial effusion. He needed emergent life-saving therapeutic pericardiocentesis in the intensive care unit. Warfarin was discontinued promptly. He was discharged home on heparin-free HHD. Before discharging the patient from the hospital, we were able to perform intermittent heparin-free hemodialysis with a Fresenius machine and Optiflux 160 NR dialyzers, with pump-delivered normal saline flushes. Extracorporeal circuits and filters started to clot, with resultant high venous pressures and compromised HHD. Heparin anticoagulation was still contraindicated. Flushes with 250-500 mL of normal saline, delivered in aliquots every 15-30 minutes, failed to prevent frequent clotting. The use of heparin or warfarin was still contraindicated, and the patient and the caregiver (wife) were considering stopping HHD. Indeed, there was some discussion about a possible switch to peritoneal dialysis. A thorough review of the literature failed to provide guidance to resolve this repeated clotting problem.
Nevertheless, the first author, our HD Senior Technician, observed that the horizontally placed CAR-170-C NXSTAGE Chronic cartridge (hemodialysis filter) exhibited an early “clot” formation at the 12-o’clock position, despite repeated, large-volume, normal saline flushes (Figure 1). He noticed that blood was pooling around the site and that the first signs of clotting was appearing. Through trial and error, multiple and staggered physical manipulations and maneuvers of the HD filter, he confirmed one finding: that rotating the horizontally placed hemodialysis filter along its long axis, 60 degrees clockwise for 15 minutes, returning to the neutral position for 15 minutes, rotating the filter another 60 degrees counterclockwise for 15 minutes, and continuing these repeated cycles of to-and-fro rotations throughout the HHD session “did the trick” (Figure 1). It promptly and consistently resolved the clotting problem. The lines stopped clotting, and he has not needed saline flushes for smooth heparin-free HHD for more than 7 months (Table). The Table demonstrates the clear difference in the need for normal saline flushes since mid-December 2019, when we started this maneuver. The patient did not just have smooth heparin-free and normal-saline flush-free HHD treatments; the treatments were stable and streamlined. In November 2019 and in December 2019, because of multiple treatment interruptions owing to clotting of the extracorporeal circuits including the dialyzers, some treatment sessions lasted almost twice as long (Table). Others lasted more than 1 hour, and treatment was abandoned for that day (Table).Figure 1 Composite showing pictures of the horizontally aligned NxSTAGE CAR-170-C cartridge (A) in the neutral position, (B) rotated clockwise to the 2-o’clock position, (C) rotated back to the neutral position, and (D) rotated counterclockwise to the 10-o’clock position. This “Locke-Onuigbo Maneuver” is repeated throughout the home hemodialysis treatment.
Table Transition to Heparin-free Hemodialysis in Mid-December 2019 With No More Need for Repeated Normal-Saline Flushes
Treatment date Treatment duration, min Dialysate, L (processed/Rx) UF, L Weight, kg (post/dry-Rx) Pre-Tx sitting BP, mm Hg Pre-Tx standing BP, mm Hg Post-Tx sitting BP, mm Hg Post-Tx standing BP, mm Hg VP at 200 BFR, mm Hg Normal saline flushes, mL
1/3/2020 108 20.0 / 20.0 1.3 76.3 / 76.0 109 / 55 113 / 55 136 / 65 135 / 65 76 —
1/2/2020 108 20.0 / 25.0 1.3 76.2 / 76.0 135 / 57 138 / 63 143 / 70 130 / 60 52 —
12/31/2019 108 20.0 / 25.0 1.3 77.0 / 76.0 141 / 60 122 / 54 142 / 64 124 / 58 — —
12/30/2019 104 20.0 / 25.0 1.0 77.6 / 76.0 163 / 69 136 / 67 155 / 72 138 / 68 74 —
12/27/2019 105 20.0 / 25.0 1.0 77.1 / 76.0 132 / 59 119 / 46 141 / 66 122 / 60 100 —
12/26/2019 96 18.4 / 20.0 0.9 77.2 / 76.0 125 / 53 124 / 53 148 / 66 125 / 66 70 —
12/24/2019 116 20.0 / 25.0 1.3 76.7 / 76.0 130 / 55 134 / 60 135 / 57 122 / 62 82 —
12/23/2019 77 14.7 / 25.0 0.8 76.8 / 76.0 163 / 72 131 / 51 147 / 66 124 / 57 96 —
12/20/2019 114 18.8 / 25.0 1.3 76.6 / 76.0 120 / 54 128 / 54 134 / 65 109 / 54 107 —
12/19/2019 90 16.7 / 25.0 1.0 77.1 / 76.0 111 / 43 111 / 51 136 / 68 125 / 58 86 —
12/17/2019 129 20.0 / 25.0 1.5 76.7 / 76.0 126 / 60 123 / 60 122 / 58 119 / 58 113 —
12/16/2019 87 16.2 / 25.0 0.9 77.6 / 76.0 139 / 70 121 / 61 133 / 62 124/ 55 87 450
12/13/2019 76 14.4 / 25.0 0.8 75.7 / 76.0 134 / 58 114 / 58 119 / 55 113 / 59 75 400
12/12/2019 118 22.1 / 30.0 1.2 75.9 / 76.0 128 / 55 108 / 55 — / — 109 / 49 90 350
12/11/2019 157 29.2 / 30.0 1.8 75.6 / 76.0 107 / 48 100 / 49 118 / 52 112 / 54 66 400
12/9/2019 163 30.0 / 30.0 1.6 76.7 / 76.0 137 / 58 117 / 52 140 / 54 120 / 58 99 500
12/6/2019 167 30.0 / 30.0 2.0 77.3 / 76.0 122 / 48 116 / 55 123 / 63 118 / 55 83 450
12/5/2019 146 26.8 / 30.0 1.7 77.8 / 80.0 126 / 61 114 / 51 134 / 65 147 / 60 56 600
12/3/2019 166 29.4 / 30.0 2.0 77.0 / 80.0 149 / 63 133 / 61 127 / 64 135 / 59 60 600
12/2/2019 197 30.0 / 30.0 2.4 78.9 / 80.0 123 / 62 138 / 62 — / — — / — 51 750
12/1/2019 197 29.4 / 30.0 2.4 79.5 / 80.0 143 / 70 133 / 56 159 / 69 122 / 54 72 750
11/30/2019 196 30.0 / 30.0 2.4 79.8 / 80.0 152 / 66 143 / 68 154 / 71 149 / 70 — 750
BFR = blood flow rate; BP = blood pressure; Rx = hemodialysis treatment; Tx = treatment; UF = ultrafiltration; VP = venous pressure.
All through these manipulations and rotational maneuvers of the dialysis filters, the signs of early clot formation continued to be only observed around the air bubble at the 12-o’oclock position (Figure 1). After each hemodialysis treatment, at rinse back, we had consistently observed that all the dialyzer fibers that exhibited increased fibrin formation (characterized by a different color from the rest of the otherwise all-white fibers) were aligned and restricted to the group of fibers around the 12-o’clock position of the filter or cartridge (Figure 1). With the repeated timed-controlled rotational maneuvers, only these 12-o’clock filter fibers showed, if at all, any of these signs of early clotting (Figure 1). We together observed that the longer the cartridge was left at the same position (without rotation), the more likely it was to clot. By rotating the filter repeatedly throughout the HHD treatment, 60 degrees back and forth clockwise and counterclockwise, around the 12-o’clock neutral position, and along the long axis of the filter, every 15 minutes, as described earlier, the NxSTAGE filter is able to run for longer periods without clotting (Figure 1). The last author and the first author have revisited these manipulations and maneuvers with the same patient at least once per month during the last 7 months, during patient visits to the home dialysis unit, with consistently reproducible results, 100% of the time.
To our knowledge, this is the first such report in the English-language literature. We have hypothesized a mechanism and have named it the “Locke-Onuigbo Maneuver.”
Coincidentally and most fortuitously, the patient’s wife maintained a meticulously detailed daily diary of events with a calendar that we routinely provided to all of our patients for record keeping of HHD. She kept these daily diaries since the patient started his HHD treatments. The month of December 2019 was troublesome, and was a difficult month with lots of notations owing to the multiple interruptions during the HHD treatments (Figure 2A). After the new rotating maneuver, the month of January 2020 “was a breeze,” as recalled recently by the patient’s wife during a visit to the home dialysis unit (Figure 2B). Both patient and wife have been highly satisfied with his heparin-free normal saline-flush-free HHD during the last 7 months.Figure 2 Home hemodialysis dialysis daily diary notations by patient’s wife for the months of (A) December 2019 and (B) January 2020, 1 month before and 1 month after the initiation of the “Locke-Onuigbo Maneuver.”
Discussion
Despite considerable technological advances in the design and functionalities of hemodialysis, there still remains space for significant improvement both in the patient outcomes and in the outcomes of circuit survival, as frequent circuit changes increase not only the nursing workload, blood loss, and economic costs, but also compromise achievement of the filtration rate goals.11 We completed a meticulous review of the literature regarding the origins and development of the art of hemodialysis dating back several decades.12 The nearest comparator to our observations and results is the description of the rotating drum dialyzers by Zbylut J. Twardoski in his 2008 treatise on the history of hemodialyzers’ designs—the Kolff rotating drum (1943).13 In another related counterpoint scenario, Ponte et al14 described, in 2007, the use of an extracorporeal membrane oxygenation (ECMO) system aided by gravitational pull through the filter of the ECMO to treat anuric postoperative acute kidney injury. The authors used “gravity dialysis” in this report, with the ECMO system hemofilter as another depurative technique.14 We therefore, after a methodical review of the foregoing historical evidences have hypothesized that our repeated rotational manipulations of the NxSTAGE CAR-170-C Chronic cartridge created a continuing antigravity effect, thus preventing any sustained thrombogenesis in the extracorporeal system. In this way, we were able to achieve reproducible smooth, heparin-free and normal saline-flush–free HHD in our patient. Our hypothesis is that somehow, the repeated helical rotational movements every 15 minutes of the CAR-170-C NXSTAGE Chronic cartridge, created continuously along the long axis of the cartridge, provides some kind of physical defense against progressive clotting of the filter fibers. The result is therefore a sustainable maintenance of fluidity in the extracorporeal circuit to enable adequate hemodialysis to occur. Our observations with one patient demand further study for repeatability and reproducibility. Furthermore, our hypothesis calls for further study by biomedical engineers. We did not consider just having the dialyzer placed vertically, because this is not the manufacturer’s recommendation and the NxStage machine holder for the filter is only so designed. Moreover, unlike ‘standard’ dialyzer filters used with conventional hemodialysis machines like the Optiflux 160 NR filter, the CAR-170-C NxStage Chronic cartridges do not have an air trap; therefore, the visualization of the air bubble in the 12-o’oclock position is possible only with the cartridge aligned in the horizontal position.
Of the various ways and means of anticoagulation available for hemodialysis—including the use of unfractionated heparin (systemic and regional), the use of low-molecular-weight heparin, saline flushing, regional citrate, prostacyclin, heparinoids, hirudoid (recombinant hirudin), protease inhibitors, and extracorporeal device modifications1—the method that we have described in this case report has not been reported previously.
Conclusion
We have described a novel method of achieving heparin-free home hemodialysis using the NxStage machine with the CAR-170-C Chronic Cartridge Hemodialysis Filter. We have hypothesized a mechanism to explain this observation and have named this the “Locke-Onuigbo Maneuver.” If confirmed by subsequent research, we propose that a miniaturized electric motor that is programmed to mimic these repetitive, timed controlled rotations of the horizontally placed filter along the long axis of the filter could translate to a commercially successful method of performing heparin-free dialysis.
Potential Competing Interests: The authors report no competing interests. | WARFARIN SODIUM | DrugsGivenReaction | CC BY-NC-ND | 33718796 | 18,556,293 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Pericardial haemorrhage'. | Timed Controlled Repeated Rotation of the CAR-170-C NXSTAGE Chronic Cartridge Hemodialysis Filter: A Novel Approach to Enabling Heparin-Free Frequent Daily Home Hemodialysis.
Heparin-free hemodialysis is often warranted in postoperative states, bleeding diathesis, and critically ill patients. Conventionally, this is achieved through normal saline flushes or regional citrate anticoagulation. An 87-year-old white man with end-stage renal disease and atrial fibrillation, who was taking warfarin and using maintenance home hemodialysis (HHD) with an NxStage machine, underwent laparoscopic appendicectomy. The procedure was complicated by intra-abdominal abscess, sepsis, and tamponade from a bloody pericardial effusion. He needed emergent therapeutic pericardiocentesis. Warfarin was promptly discontinued. He was discharged home with heparin-free HHD. Prior heparin anticoagulation for HHD was an initial bolus of 4000 units of heparin. He continued to clot his extracorporeal system with resultant very high venous pressures and compromised HHD. Heparin anticoagulation was still contraindicated. Flushes with 250-500 mL normal saline, delivered in aliquots every 15-30 minutes, failed to prevent the frequent clotting. The first author, our HD Senior Technician, had astutely observed that the horizontally placed hemodialysis filter exhibited early "clot" formation at the 12-o'clock position, despite the saline flushes. Through trial and error, he discovered that rotating the horizontally placed hemodialysis filter along its long axis, 60 degrees clockwise for 15 minutes, return to the neutral position for 15 minutes, rotating the filter another 60 degrees counterclockwise for 15 minutes, with this repeated cycle of rotations "did the trick." It promptly and consistently resolved the clotting problem. The lines stopped clotting, and he has not needed saline flushes for smooth heparin-free HHD for more than 7 months. To our knowledge, this is the first such report. Further study is justified. We have hypothesized a mechanism and have named this the "Locke-Onuigbo Maneuver."
Generally, anticoagulation is necessary for hemodialysis (HD) lasting more than 1-2 hours, because blood contact with the extracorporeal circuit results in activation of the intrinsic coagulation pathway, leading to thrombogenesis.1, 2, 3, 4 Nevertheless, heparin-free hemodialysis is often warranted as in postoperative states, in patients with bleeding diathesis, and in critically ill patients in the intensive care unit.5, 6, 7, 8, 9 Conventionally, this is achieved through normal saline flushes, or with the application of regional citrate anticoagulation.1,6, 7, 8, 9 We recently encountered a set of converging clinical scenarios in which a male patient on home hemodialysis (HHD) using an NxStage machine had to be discharged home from hospital on heparin-free HHD; despite large-volume saline flushes, he had continued to clot his extracorporeal circuits and filters with resulting inadequate HHD. The use of heparin or warfarin was contraindicated, and the patient and the caregiver (wife) were considering stopping HHD. Indeed, there was some discussion about a possible switch to peritoneal dialysis. Thanks to the observations of our Senior HD technician, the first author, we, the last author working closely together with the first author, found a unique solution to this problem. We have mastered a novel approach to performgin sustainable, heparin-free HHD by intermittently rotating the horizontally aligned CAR-170-C NXSTAGE Chronic cartridge clockwise and counterclockwise, as described herein. The CAR-170-C NXSTAGE Chronic cartridges are single use, are gamma sterilized, include an integrated dialyzer, and are manufactured from glycerin-free polyethersulfone membranes (PUREMA H) known for biocompatibility.10 The filter comes with the dialysis lines from the manufacturer as a preattached dialyzer.10
Case Report
An 87-year-old white man with end-stage renal disease and atrial fibrillation, who was also using warfarin and maintenance HHD with an NxStage machine, underwent laparoscopic appendicectomy. The procedure was complicated by intra-abdominal abscess, sepsis, and tamponade from a bloody pericardial effusion. He needed emergent life-saving therapeutic pericardiocentesis in the intensive care unit. Warfarin was discontinued promptly. He was discharged home on heparin-free HHD. Before discharging the patient from the hospital, we were able to perform intermittent heparin-free hemodialysis with a Fresenius machine and Optiflux 160 NR dialyzers, with pump-delivered normal saline flushes. Extracorporeal circuits and filters started to clot, with resultant high venous pressures and compromised HHD. Heparin anticoagulation was still contraindicated. Flushes with 250-500 mL of normal saline, delivered in aliquots every 15-30 minutes, failed to prevent frequent clotting. The use of heparin or warfarin was still contraindicated, and the patient and the caregiver (wife) were considering stopping HHD. Indeed, there was some discussion about a possible switch to peritoneal dialysis. A thorough review of the literature failed to provide guidance to resolve this repeated clotting problem.
Nevertheless, the first author, our HD Senior Technician, observed that the horizontally placed CAR-170-C NXSTAGE Chronic cartridge (hemodialysis filter) exhibited an early “clot” formation at the 12-o’clock position, despite repeated, large-volume, normal saline flushes (Figure 1). He noticed that blood was pooling around the site and that the first signs of clotting was appearing. Through trial and error, multiple and staggered physical manipulations and maneuvers of the HD filter, he confirmed one finding: that rotating the horizontally placed hemodialysis filter along its long axis, 60 degrees clockwise for 15 minutes, returning to the neutral position for 15 minutes, rotating the filter another 60 degrees counterclockwise for 15 minutes, and continuing these repeated cycles of to-and-fro rotations throughout the HHD session “did the trick” (Figure 1). It promptly and consistently resolved the clotting problem. The lines stopped clotting, and he has not needed saline flushes for smooth heparin-free HHD for more than 7 months (Table). The Table demonstrates the clear difference in the need for normal saline flushes since mid-December 2019, when we started this maneuver. The patient did not just have smooth heparin-free and normal-saline flush-free HHD treatments; the treatments were stable and streamlined. In November 2019 and in December 2019, because of multiple treatment interruptions owing to clotting of the extracorporeal circuits including the dialyzers, some treatment sessions lasted almost twice as long (Table). Others lasted more than 1 hour, and treatment was abandoned for that day (Table).Figure 1 Composite showing pictures of the horizontally aligned NxSTAGE CAR-170-C cartridge (A) in the neutral position, (B) rotated clockwise to the 2-o’clock position, (C) rotated back to the neutral position, and (D) rotated counterclockwise to the 10-o’clock position. This “Locke-Onuigbo Maneuver” is repeated throughout the home hemodialysis treatment.
Table Transition to Heparin-free Hemodialysis in Mid-December 2019 With No More Need for Repeated Normal-Saline Flushes
Treatment date Treatment duration, min Dialysate, L (processed/Rx) UF, L Weight, kg (post/dry-Rx) Pre-Tx sitting BP, mm Hg Pre-Tx standing BP, mm Hg Post-Tx sitting BP, mm Hg Post-Tx standing BP, mm Hg VP at 200 BFR, mm Hg Normal saline flushes, mL
1/3/2020 108 20.0 / 20.0 1.3 76.3 / 76.0 109 / 55 113 / 55 136 / 65 135 / 65 76 —
1/2/2020 108 20.0 / 25.0 1.3 76.2 / 76.0 135 / 57 138 / 63 143 / 70 130 / 60 52 —
12/31/2019 108 20.0 / 25.0 1.3 77.0 / 76.0 141 / 60 122 / 54 142 / 64 124 / 58 — —
12/30/2019 104 20.0 / 25.0 1.0 77.6 / 76.0 163 / 69 136 / 67 155 / 72 138 / 68 74 —
12/27/2019 105 20.0 / 25.0 1.0 77.1 / 76.0 132 / 59 119 / 46 141 / 66 122 / 60 100 —
12/26/2019 96 18.4 / 20.0 0.9 77.2 / 76.0 125 / 53 124 / 53 148 / 66 125 / 66 70 —
12/24/2019 116 20.0 / 25.0 1.3 76.7 / 76.0 130 / 55 134 / 60 135 / 57 122 / 62 82 —
12/23/2019 77 14.7 / 25.0 0.8 76.8 / 76.0 163 / 72 131 / 51 147 / 66 124 / 57 96 —
12/20/2019 114 18.8 / 25.0 1.3 76.6 / 76.0 120 / 54 128 / 54 134 / 65 109 / 54 107 —
12/19/2019 90 16.7 / 25.0 1.0 77.1 / 76.0 111 / 43 111 / 51 136 / 68 125 / 58 86 —
12/17/2019 129 20.0 / 25.0 1.5 76.7 / 76.0 126 / 60 123 / 60 122 / 58 119 / 58 113 —
12/16/2019 87 16.2 / 25.0 0.9 77.6 / 76.0 139 / 70 121 / 61 133 / 62 124/ 55 87 450
12/13/2019 76 14.4 / 25.0 0.8 75.7 / 76.0 134 / 58 114 / 58 119 / 55 113 / 59 75 400
12/12/2019 118 22.1 / 30.0 1.2 75.9 / 76.0 128 / 55 108 / 55 — / — 109 / 49 90 350
12/11/2019 157 29.2 / 30.0 1.8 75.6 / 76.0 107 / 48 100 / 49 118 / 52 112 / 54 66 400
12/9/2019 163 30.0 / 30.0 1.6 76.7 / 76.0 137 / 58 117 / 52 140 / 54 120 / 58 99 500
12/6/2019 167 30.0 / 30.0 2.0 77.3 / 76.0 122 / 48 116 / 55 123 / 63 118 / 55 83 450
12/5/2019 146 26.8 / 30.0 1.7 77.8 / 80.0 126 / 61 114 / 51 134 / 65 147 / 60 56 600
12/3/2019 166 29.4 / 30.0 2.0 77.0 / 80.0 149 / 63 133 / 61 127 / 64 135 / 59 60 600
12/2/2019 197 30.0 / 30.0 2.4 78.9 / 80.0 123 / 62 138 / 62 — / — — / — 51 750
12/1/2019 197 29.4 / 30.0 2.4 79.5 / 80.0 143 / 70 133 / 56 159 / 69 122 / 54 72 750
11/30/2019 196 30.0 / 30.0 2.4 79.8 / 80.0 152 / 66 143 / 68 154 / 71 149 / 70 — 750
BFR = blood flow rate; BP = blood pressure; Rx = hemodialysis treatment; Tx = treatment; UF = ultrafiltration; VP = venous pressure.
All through these manipulations and rotational maneuvers of the dialysis filters, the signs of early clot formation continued to be only observed around the air bubble at the 12-o’oclock position (Figure 1). After each hemodialysis treatment, at rinse back, we had consistently observed that all the dialyzer fibers that exhibited increased fibrin formation (characterized by a different color from the rest of the otherwise all-white fibers) were aligned and restricted to the group of fibers around the 12-o’clock position of the filter or cartridge (Figure 1). With the repeated timed-controlled rotational maneuvers, only these 12-o’clock filter fibers showed, if at all, any of these signs of early clotting (Figure 1). We together observed that the longer the cartridge was left at the same position (without rotation), the more likely it was to clot. By rotating the filter repeatedly throughout the HHD treatment, 60 degrees back and forth clockwise and counterclockwise, around the 12-o’clock neutral position, and along the long axis of the filter, every 15 minutes, as described earlier, the NxSTAGE filter is able to run for longer periods without clotting (Figure 1). The last author and the first author have revisited these manipulations and maneuvers with the same patient at least once per month during the last 7 months, during patient visits to the home dialysis unit, with consistently reproducible results, 100% of the time.
To our knowledge, this is the first such report in the English-language literature. We have hypothesized a mechanism and have named it the “Locke-Onuigbo Maneuver.”
Coincidentally and most fortuitously, the patient’s wife maintained a meticulously detailed daily diary of events with a calendar that we routinely provided to all of our patients for record keeping of HHD. She kept these daily diaries since the patient started his HHD treatments. The month of December 2019 was troublesome, and was a difficult month with lots of notations owing to the multiple interruptions during the HHD treatments (Figure 2A). After the new rotating maneuver, the month of January 2020 “was a breeze,” as recalled recently by the patient’s wife during a visit to the home dialysis unit (Figure 2B). Both patient and wife have been highly satisfied with his heparin-free normal saline-flush-free HHD during the last 7 months.Figure 2 Home hemodialysis dialysis daily diary notations by patient’s wife for the months of (A) December 2019 and (B) January 2020, 1 month before and 1 month after the initiation of the “Locke-Onuigbo Maneuver.”
Discussion
Despite considerable technological advances in the design and functionalities of hemodialysis, there still remains space for significant improvement both in the patient outcomes and in the outcomes of circuit survival, as frequent circuit changes increase not only the nursing workload, blood loss, and economic costs, but also compromise achievement of the filtration rate goals.11 We completed a meticulous review of the literature regarding the origins and development of the art of hemodialysis dating back several decades.12 The nearest comparator to our observations and results is the description of the rotating drum dialyzers by Zbylut J. Twardoski in his 2008 treatise on the history of hemodialyzers’ designs—the Kolff rotating drum (1943).13 In another related counterpoint scenario, Ponte et al14 described, in 2007, the use of an extracorporeal membrane oxygenation (ECMO) system aided by gravitational pull through the filter of the ECMO to treat anuric postoperative acute kidney injury. The authors used “gravity dialysis” in this report, with the ECMO system hemofilter as another depurative technique.14 We therefore, after a methodical review of the foregoing historical evidences have hypothesized that our repeated rotational manipulations of the NxSTAGE CAR-170-C Chronic cartridge created a continuing antigravity effect, thus preventing any sustained thrombogenesis in the extracorporeal system. In this way, we were able to achieve reproducible smooth, heparin-free and normal saline-flush–free HHD in our patient. Our hypothesis is that somehow, the repeated helical rotational movements every 15 minutes of the CAR-170-C NXSTAGE Chronic cartridge, created continuously along the long axis of the cartridge, provides some kind of physical defense against progressive clotting of the filter fibers. The result is therefore a sustainable maintenance of fluidity in the extracorporeal circuit to enable adequate hemodialysis to occur. Our observations with one patient demand further study for repeatability and reproducibility. Furthermore, our hypothesis calls for further study by biomedical engineers. We did not consider just having the dialyzer placed vertically, because this is not the manufacturer’s recommendation and the NxStage machine holder for the filter is only so designed. Moreover, unlike ‘standard’ dialyzer filters used with conventional hemodialysis machines like the Optiflux 160 NR filter, the CAR-170-C NxStage Chronic cartridges do not have an air trap; therefore, the visualization of the air bubble in the 12-o’oclock position is possible only with the cartridge aligned in the horizontal position.
Of the various ways and means of anticoagulation available for hemodialysis—including the use of unfractionated heparin (systemic and regional), the use of low-molecular-weight heparin, saline flushing, regional citrate, prostacyclin, heparinoids, hirudoid (recombinant hirudin), protease inhibitors, and extracorporeal device modifications1—the method that we have described in this case report has not been reported previously.
Conclusion
We have described a novel method of achieving heparin-free home hemodialysis using the NxStage machine with the CAR-170-C Chronic Cartridge Hemodialysis Filter. We have hypothesized a mechanism to explain this observation and have named this the “Locke-Onuigbo Maneuver.” If confirmed by subsequent research, we propose that a miniaturized electric motor that is programmed to mimic these repetitive, timed controlled rotations of the horizontally placed filter along the long axis of the filter could translate to a commercially successful method of performing heparin-free dialysis.
Potential Competing Interests: The authors report no competing interests. | WARFARIN SODIUM | DrugsGivenReaction | CC BY-NC-ND | 33718796 | 18,556,293 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Endocarditis histoplasma'. | Prosthetic Joint Infections Due to Histoplasma capsulatum: A Report of 3 Cases.
Histoplasma capsulatum causes pneumonia and multisystemic disease in humans. Musculoskeletal involvement in histoplasmosis is most often tenosynovitis and rarely septic arthritis. Even more uncommon is the involvement of prosthetic joints. Here, we report a series of 3 cases of prosthetic joint failures caused by infection due to H capsulatum. Together with a review of 4 previously reported cases, we summarize host characteristics, clinical presentation, surgical approaches, antifungal management, and outcomes of this rare orthopedic joint infection.
Histoplasma capsulatum is a dimorphic fungus that is endemic throughoutthe Ohio and Mississippi river valleys and in certain pockets around the world. Although the fungus primarily causes lung infection, extrapulmonary histoplasmosis may involve the bone marrow, liver, brain, spleen, and the gastrointestinal tract.1,2 Osteomyelitis and septic arthritis are uncommon forms of histoplasmosis, while the involvement of joint prosthesis is even much more uncommon.2,3 In our review of the medical literature, we found only 4 cases of H capsulatum infection involving a prosthetic joint.4, 5, 6 Here, we present the clinical outcomes of 3 additional cases of prosthetic joint infection (PJI) due to H capsulatum.
Case Report
Case 1
A 77-year-old diabetic woman from Wisconsin presented to our clinic in 2019 with a 2-day history of right knee pain. She was a recipient of a deceased donor kidney transplant for end-stage renal disease due to autosomal dominant polycystic kidney disease in 1991, followed by a living-related donor kidney transplant due to allograft failure from calcineurin inhibitor nephrotoxicity in 2009. She was receiving maintenance immunosuppression with tacrolimus, mycophenolate mofetil, and prednisone. She also had bilateral total knee arthroplasties (TKAs) in 1998 (right) and 2012 (left) for degenerative joint disease. She did not have fever, chills, rigors, or other signs and symptoms of systemic illness. On physical examination, her right knee was erythematous, swollen, warm, and tender. Right knee synovial fluid analysis was suggestive of infection (Table). She had an elevated C-reactive protein level (23.1 mg/L; normal range, <8 mg/L). However, the synovial fluid cultures for bacteria were negative at 14 days.Table Clinical Characteristics and Outcomes of 7 Cases of Prosthetic Joint Infection Due to Histoplasma capsulatum
Case number (author; patient location) Age (y)/sex Comorbidity Presenting symptoms Joint involvement Synovial fluid cell count Synovial fluid or tissue culture Blood culture Serology Urine Histoplasma antigen Surgical treatment Antimicrobial treatment Outcome
1 (Berbari; Wisconsin) 77/F Kidney transplant Right knee pain and swelling Right knee joint 2530 cells (88% neutrophils) H capsulatum on surgical culture after 15 d of incubation Negative Negative Negative Debridement and retention Itraconazole (lifelong plan) Good function at last follow-up (12 mo)
2 (Berbari; Iowa) 69/F Rheumatoid arthritis, previous breast cancer Left shoulder pain Left shoulder joint 3512 cells (85% neutrophils) H capsulatum on surgical cultures after 12 d of incubation Negative Negative Negative Resection arthroplasty Itraconazole (ongoing) Reimplanted at 7 mo after resection
3 (Berbari; Nebraska) 75/M Hypertension Right knee wound drainage Right knee joint Not done H capsulatum on surgical cultures Negative Immunodiffusion M band positive Negative Resection arthroplasty Itraconazole (heart failure) → posaconazole × 1 y Reimplanted at 10 mo after the initiation of antifungal treatment
4 (Fowler; North Carolina) 84/F Polymyalgia rheumatica Sinus track over the left hip, pain and swelling Left hip joint 192,500 cells (88% granulocytes) H capsulatum on cultures 6 wk after surgery Not reported Not reported Not reported Debridement and retention Itraconazole (lifelong plan) Good function at the time of the last report (3 y)
5 (Meiyappan; travel to Ethiopia) 57/F Vasculitis Right knee pain and swelling Right knee joint 1800 cells (31% neutrophils) Histoplasma capsulatum on surgical and synovial fluid cultures Not reported Negative Negative Resection arthroplasty Amphotericin B + posaconazole (prolonged) Reimplanted at 11 mo; good function at 3.8 y
6 (Nowbakht; Wisconsin, Guatemala) 77/F Cardiac disease, diabetes mellitus Right knee pain Right knee joint Not reported H capsulatum on synovial fluid culture Negative Not reported Not reported Resection arthroplasty Itraconazole × 9 mo Reimplanted at 9 mo; good function at 2 y
7 (Foo; Singapore) 68/M End-stage renal disease receiving hemodialysis Right knee pain and swelling Right knee joint 9600 cells (56% neutrophils) H capsulatum on tissue culture Not reported Not reported Not reported Revision with implantation of prosthesis Itraconazole (ongoing at the time of the report) No recurrence of symptoms at 8 mo
F = female; M = male.
From references 4-7.
She underwent irrigation and debridement of the right knee joint. Intraoperatively, the right knee joint and prosthesis did not look grossly infected, and because the components were well-seated, they were retained. Fifteen days later, multiple cultures of intraoperative tissue specimens grew H capsulatum. Histoplasma serology and urinary Histoplasma antigen were negative. The chest radiograph was normal. Oral itraconazole was initiated. The dose of mycophenolate mofetil was reduced. Twelve months later, she was tolerating itraconazole and she had a pain-free functional right knee joint. There were no symptoms of immune reconstitution inflammatory syndrome. Because of the retained components and her immunocompromised status, chronic long term itraconazole suppression is planned.
Case 2
A 69-year-old woman from Iowa with rheumatoid arthritis taking etanercept and prednisone had multiple prosthetic joints implanted for painful destructive arthritis, including a left shoulder reverse arthroplasty in 2019. Eight months after her left shoulder arthroplasty, she complained of progressive lingering pain in her left shoulder. The C-reactive protein level was 13.8 mg/L. The radiograph of the shoulder revealed lucency about the glenoid component, which was suggestive of loosening. Left shoulder synovial fluid analysis revealed 3512 cells/μL with predominance of neutrophils. The culture of the synovial fluid was negative at 14 days.
Over concerns of underlying infection and component loosening, she underwent resection of left shoulder prosthesis. On histopathology, there was no evidence of acute inflammation. However, after 12 days of incubation, the intraoperative tissue cultures were positive for H capsulatum. The chest radiograph was normal. Histoplasma serology and urine antigen were negative. Oral itraconazole therapy was initiated. She has completed 7 months of itraconazole treatment before shoulder reimplantation. She continued receiving itraconazole treatment at the time of this report, with reassessment planned at 1 year. Despite not resuming etanercept treatment of her underlying rheumatoid arthritis, there were no symptoms related to immune reconstitution inflammatory syndrome.
Case 3
A 75-year-old man from Nebraska with hypertension underwent a right TKA in 2019 for degenerative joint disease at his local hospital. Two weeks later, he developed redness, swelling, and drainage of the surgical wound. He did not have a fever or other systemic symptoms of disseminated infection. During surgical debridement of the right knee, the wound was found to track deep into the knee joint prosthesis. Thus, resection TKA was performed. Multiple intraoperative tissue cultures grew H capsulatum. A computed tomography scan of the chest revealed mediastinal and hilar adenopathy. Histoplasma M band serology was positive, whereas urinary antigen was negative. Oral itraconazole therapy was initiated but later complicated by congestive heart failure. His antifungal regimen was subsequently changed to posaconazole.
Three months after resection arthroplasty, he presented to our hospital because of recurrence of right knee pain. Radiographic studies revealed a fractured knee spacer. He underwent irrigation, debridement, and exchange of the spacer. Intraoperative cultures were negative. The chest radiograph was normal. Histoplasma serology and urine Histoplasma antigen were negative. He continued receiving posaconazole therapy to complete 1 year of antifungal treatment. Right TKA reimplantation was performed 10 months after initial resection arthroplasty.
Discussion
Our case series of PJI due to H capsulatum add to 4 previously reported cases.4, 5, 6, 7 The clinical details of all 7 cases are summarized in the Table. One common characteristic of these cases is the delayed growth of H capsulatum in cultures, which led to the initial use of broad-spectrum antibacterial therapy for presumed culture-negative PJI.8,9 About 7% to 10% of PJIs are culture-negative, which is most often due to the recent intake of antibiotics.8,9 However, as highlighted in this report, it may be due to an unusual organism such as fungi.10 To diagnose unusual infections, one may need to prolong culture incubation, use special culture medium, obtain serology, or perform molecular testing. In addition, one should consider unusual infection when there is a history of exposures, such as residence or travel to endemic areas. Our 3 cases reside in the Midwest United States and were at risk of histoplasmosis. In all the reported cases, H capsulatum eventually grew in cultures, albeit delayed (among the cases in which time to growth was reported [at least 12 days]).5 Notably, Histoplasma serology and urinary Histoplasma antigen were negative in all except 1 case in which the M band was positive, suggesting that cases were localized infections involving only joint arthroplasty. This finding is consistent with a study reporting that most bone and joint infections due to H capsulatum are localized.3 Our third case, however, could have been a component of a systemic disease because of the reported presence of mediastinal and hilar adenopathy on the chest computed tomography scan. Histoplasma M band serology was also positive, although urine antigen and blood cultures were negative in this case. A detailed review of this third case did not identify unusual exposures that could have predisposed him to acquire a high burden of infection.
Overall, PJI due to fungi is extremely rare.11 All fungi accounted for only 1% of 3525 PJI cases at Mayo Clinic during 1996 to 2014, and most are due to Candida species.11 As mentioned, bone and joint infections due to H capsulatum are rare.3,12 None of the fungal PJIs in our institution from 1996 to 2014 were due to H capsulatum.11 In a review of 222 osteoarticular infections due to dimorphic fungi from 1970 to 2012, only 18 cases of H capsulatum were observed.3 Most cases of osteoarticular infections due to H capsulatum (89%) were described as a solitary bone lesion or joint infection.3 The clinical presentation in our cases is consistent with this report—PJI due to H capsulatum appear to be mostly isolated without other active focus of infection.
Factors that predispose patients to develop PJI due to H capsulatum is unknown. Although many of the cases are residents in endemic regions, there are no unifying host characteristics in the 7 patients. Although the infection has occurred in immunocompetent patients, it is possible that an underlying immunosuppression may predispose a patient to develop PJI due to H capsulatum.5 However, in a case-control study of immunocompromised transplant patients with PJI, no case of histoplasmosis was reported.13 In another report on 152 transplant patients with histoplasmosis, none involved bones or joints.14 In addition, the time to infection in relation to initial joint implantation was highly variable (from weeks to many years), suggesting that there is no unifying factor that predisposes to this rare orthopedic infection.
The antifungal treatment of histoplasmosis depends on the immune status and disease severity.15 Because of the localized nature of joint infection in our patients, oral itraconazole alone was used. In one of the previously reported cases, amphotericin B was also used over concerns of disseminated histoplasmosis, although this was not conclusively proven. The duration of treatment of PJI due to H capsulatum is not clear; however, all cases were treated with more than 7 months of antifungal therapy after resection arthroplasty. Longer (anticipated lifelong) courses were given to 2 immunocompromised individuals with retained joint components.
In general, a 2-stage exchange surgery (in which the infected joint prosthesis is removed during debridement, followed by antimicrobial treatment, and implantation of new prosthesis once the infection has cleared) is the recommended curative therapy for PJI.16 Because H capsulatum is efficient in forming a biofilm,17 a 2-stage exchange approach may also be needed for the complete eradication of PJI due to H capsulatum. Resection arthroplasty was used in 5 cases in this report, but the time to prosthesis reimplantation was highly variable. In addition to microbial and pharmacological factors, the decision on the timing of reimplantation considers host and social considerations, such as the desire to resume activities of daily living and the availability of caretakers during the debilitating periods related to surgical joint explantation. In 2 immunocompromised patients, resection arthroplasties were not performed because prostheses were well fixed and patients were functional without pain. Because of retained prostheses, both patients were receiving maintenance indefinite itraconazole suppression.5
Conclusion
Histoplasma capsulatum is a rare cause of PJI. The diagnosis of histoplasmosis was initially not suspected, and most patients were receiving antibacterial therapy for presumed culture-negative PJI until cultures surprisingly grew H capsulatum. Resection arthroplasty was performed in 5 of 7 cases, whereas debridement and retention strategy was used in 2 immunocompromised cases. All the cases were treated initially or fully with itraconazole for prolonged durations; one was also given amphotericin B over concerns of disseminated infection, while another one was transitioned to posaconazole after he developed heart failure attributed to itraconazole therapy.
Potential Competing Interests: The authors report no competing interests. | ETANERCEPT, PREDNISONE | DrugsGivenReaction | CC BY-NC-ND | 33718797 | 19,466,418 | 2021-02 |
What was the outcome of reaction 'Arthritis infective'? | Prosthetic Joint Infections Due to Histoplasma capsulatum: A Report of 3 Cases.
Histoplasma capsulatum causes pneumonia and multisystemic disease in humans. Musculoskeletal involvement in histoplasmosis is most often tenosynovitis and rarely septic arthritis. Even more uncommon is the involvement of prosthetic joints. Here, we report a series of 3 cases of prosthetic joint failures caused by infection due to H capsulatum. Together with a review of 4 previously reported cases, we summarize host characteristics, clinical presentation, surgical approaches, antifungal management, and outcomes of this rare orthopedic joint infection.
Histoplasma capsulatum is a dimorphic fungus that is endemic throughoutthe Ohio and Mississippi river valleys and in certain pockets around the world. Although the fungus primarily causes lung infection, extrapulmonary histoplasmosis may involve the bone marrow, liver, brain, spleen, and the gastrointestinal tract.1,2 Osteomyelitis and septic arthritis are uncommon forms of histoplasmosis, while the involvement of joint prosthesis is even much more uncommon.2,3 In our review of the medical literature, we found only 4 cases of H capsulatum infection involving a prosthetic joint.4, 5, 6 Here, we present the clinical outcomes of 3 additional cases of prosthetic joint infection (PJI) due to H capsulatum.
Case Report
Case 1
A 77-year-old diabetic woman from Wisconsin presented to our clinic in 2019 with a 2-day history of right knee pain. She was a recipient of a deceased donor kidney transplant for end-stage renal disease due to autosomal dominant polycystic kidney disease in 1991, followed by a living-related donor kidney transplant due to allograft failure from calcineurin inhibitor nephrotoxicity in 2009. She was receiving maintenance immunosuppression with tacrolimus, mycophenolate mofetil, and prednisone. She also had bilateral total knee arthroplasties (TKAs) in 1998 (right) and 2012 (left) for degenerative joint disease. She did not have fever, chills, rigors, or other signs and symptoms of systemic illness. On physical examination, her right knee was erythematous, swollen, warm, and tender. Right knee synovial fluid analysis was suggestive of infection (Table). She had an elevated C-reactive protein level (23.1 mg/L; normal range, <8 mg/L). However, the synovial fluid cultures for bacteria were negative at 14 days.Table Clinical Characteristics and Outcomes of 7 Cases of Prosthetic Joint Infection Due to Histoplasma capsulatum
Case number (author; patient location) Age (y)/sex Comorbidity Presenting symptoms Joint involvement Synovial fluid cell count Synovial fluid or tissue culture Blood culture Serology Urine Histoplasma antigen Surgical treatment Antimicrobial treatment Outcome
1 (Berbari; Wisconsin) 77/F Kidney transplant Right knee pain and swelling Right knee joint 2530 cells (88% neutrophils) H capsulatum on surgical culture after 15 d of incubation Negative Negative Negative Debridement and retention Itraconazole (lifelong plan) Good function at last follow-up (12 mo)
2 (Berbari; Iowa) 69/F Rheumatoid arthritis, previous breast cancer Left shoulder pain Left shoulder joint 3512 cells (85% neutrophils) H capsulatum on surgical cultures after 12 d of incubation Negative Negative Negative Resection arthroplasty Itraconazole (ongoing) Reimplanted at 7 mo after resection
3 (Berbari; Nebraska) 75/M Hypertension Right knee wound drainage Right knee joint Not done H capsulatum on surgical cultures Negative Immunodiffusion M band positive Negative Resection arthroplasty Itraconazole (heart failure) → posaconazole × 1 y Reimplanted at 10 mo after the initiation of antifungal treatment
4 (Fowler; North Carolina) 84/F Polymyalgia rheumatica Sinus track over the left hip, pain and swelling Left hip joint 192,500 cells (88% granulocytes) H capsulatum on cultures 6 wk after surgery Not reported Not reported Not reported Debridement and retention Itraconazole (lifelong plan) Good function at the time of the last report (3 y)
5 (Meiyappan; travel to Ethiopia) 57/F Vasculitis Right knee pain and swelling Right knee joint 1800 cells (31% neutrophils) Histoplasma capsulatum on surgical and synovial fluid cultures Not reported Negative Negative Resection arthroplasty Amphotericin B + posaconazole (prolonged) Reimplanted at 11 mo; good function at 3.8 y
6 (Nowbakht; Wisconsin, Guatemala) 77/F Cardiac disease, diabetes mellitus Right knee pain Right knee joint Not reported H capsulatum on synovial fluid culture Negative Not reported Not reported Resection arthroplasty Itraconazole × 9 mo Reimplanted at 9 mo; good function at 2 y
7 (Foo; Singapore) 68/M End-stage renal disease receiving hemodialysis Right knee pain and swelling Right knee joint 9600 cells (56% neutrophils) H capsulatum on tissue culture Not reported Not reported Not reported Revision with implantation of prosthesis Itraconazole (ongoing at the time of the report) No recurrence of symptoms at 8 mo
F = female; M = male.
From references 4-7.
She underwent irrigation and debridement of the right knee joint. Intraoperatively, the right knee joint and prosthesis did not look grossly infected, and because the components were well-seated, they were retained. Fifteen days later, multiple cultures of intraoperative tissue specimens grew H capsulatum. Histoplasma serology and urinary Histoplasma antigen were negative. The chest radiograph was normal. Oral itraconazole was initiated. The dose of mycophenolate mofetil was reduced. Twelve months later, she was tolerating itraconazole and she had a pain-free functional right knee joint. There were no symptoms of immune reconstitution inflammatory syndrome. Because of the retained components and her immunocompromised status, chronic long term itraconazole suppression is planned.
Case 2
A 69-year-old woman from Iowa with rheumatoid arthritis taking etanercept and prednisone had multiple prosthetic joints implanted for painful destructive arthritis, including a left shoulder reverse arthroplasty in 2019. Eight months after her left shoulder arthroplasty, she complained of progressive lingering pain in her left shoulder. The C-reactive protein level was 13.8 mg/L. The radiograph of the shoulder revealed lucency about the glenoid component, which was suggestive of loosening. Left shoulder synovial fluid analysis revealed 3512 cells/μL with predominance of neutrophils. The culture of the synovial fluid was negative at 14 days.
Over concerns of underlying infection and component loosening, she underwent resection of left shoulder prosthesis. On histopathology, there was no evidence of acute inflammation. However, after 12 days of incubation, the intraoperative tissue cultures were positive for H capsulatum. The chest radiograph was normal. Histoplasma serology and urine antigen were negative. Oral itraconazole therapy was initiated. She has completed 7 months of itraconazole treatment before shoulder reimplantation. She continued receiving itraconazole treatment at the time of this report, with reassessment planned at 1 year. Despite not resuming etanercept treatment of her underlying rheumatoid arthritis, there were no symptoms related to immune reconstitution inflammatory syndrome.
Case 3
A 75-year-old man from Nebraska with hypertension underwent a right TKA in 2019 for degenerative joint disease at his local hospital. Two weeks later, he developed redness, swelling, and drainage of the surgical wound. He did not have a fever or other systemic symptoms of disseminated infection. During surgical debridement of the right knee, the wound was found to track deep into the knee joint prosthesis. Thus, resection TKA was performed. Multiple intraoperative tissue cultures grew H capsulatum. A computed tomography scan of the chest revealed mediastinal and hilar adenopathy. Histoplasma M band serology was positive, whereas urinary antigen was negative. Oral itraconazole therapy was initiated but later complicated by congestive heart failure. His antifungal regimen was subsequently changed to posaconazole.
Three months after resection arthroplasty, he presented to our hospital because of recurrence of right knee pain. Radiographic studies revealed a fractured knee spacer. He underwent irrigation, debridement, and exchange of the spacer. Intraoperative cultures were negative. The chest radiograph was normal. Histoplasma serology and urine Histoplasma antigen were negative. He continued receiving posaconazole therapy to complete 1 year of antifungal treatment. Right TKA reimplantation was performed 10 months after initial resection arthroplasty.
Discussion
Our case series of PJI due to H capsulatum add to 4 previously reported cases.4, 5, 6, 7 The clinical details of all 7 cases are summarized in the Table. One common characteristic of these cases is the delayed growth of H capsulatum in cultures, which led to the initial use of broad-spectrum antibacterial therapy for presumed culture-negative PJI.8,9 About 7% to 10% of PJIs are culture-negative, which is most often due to the recent intake of antibiotics.8,9 However, as highlighted in this report, it may be due to an unusual organism such as fungi.10 To diagnose unusual infections, one may need to prolong culture incubation, use special culture medium, obtain serology, or perform molecular testing. In addition, one should consider unusual infection when there is a history of exposures, such as residence or travel to endemic areas. Our 3 cases reside in the Midwest United States and were at risk of histoplasmosis. In all the reported cases, H capsulatum eventually grew in cultures, albeit delayed (among the cases in which time to growth was reported [at least 12 days]).5 Notably, Histoplasma serology and urinary Histoplasma antigen were negative in all except 1 case in which the M band was positive, suggesting that cases were localized infections involving only joint arthroplasty. This finding is consistent with a study reporting that most bone and joint infections due to H capsulatum are localized.3 Our third case, however, could have been a component of a systemic disease because of the reported presence of mediastinal and hilar adenopathy on the chest computed tomography scan. Histoplasma M band serology was also positive, although urine antigen and blood cultures were negative in this case. A detailed review of this third case did not identify unusual exposures that could have predisposed him to acquire a high burden of infection.
Overall, PJI due to fungi is extremely rare.11 All fungi accounted for only 1% of 3525 PJI cases at Mayo Clinic during 1996 to 2014, and most are due to Candida species.11 As mentioned, bone and joint infections due to H capsulatum are rare.3,12 None of the fungal PJIs in our institution from 1996 to 2014 were due to H capsulatum.11 In a review of 222 osteoarticular infections due to dimorphic fungi from 1970 to 2012, only 18 cases of H capsulatum were observed.3 Most cases of osteoarticular infections due to H capsulatum (89%) were described as a solitary bone lesion or joint infection.3 The clinical presentation in our cases is consistent with this report—PJI due to H capsulatum appear to be mostly isolated without other active focus of infection.
Factors that predispose patients to develop PJI due to H capsulatum is unknown. Although many of the cases are residents in endemic regions, there are no unifying host characteristics in the 7 patients. Although the infection has occurred in immunocompetent patients, it is possible that an underlying immunosuppression may predispose a patient to develop PJI due to H capsulatum.5 However, in a case-control study of immunocompromised transplant patients with PJI, no case of histoplasmosis was reported.13 In another report on 152 transplant patients with histoplasmosis, none involved bones or joints.14 In addition, the time to infection in relation to initial joint implantation was highly variable (from weeks to many years), suggesting that there is no unifying factor that predisposes to this rare orthopedic infection.
The antifungal treatment of histoplasmosis depends on the immune status and disease severity.15 Because of the localized nature of joint infection in our patients, oral itraconazole alone was used. In one of the previously reported cases, amphotericin B was also used over concerns of disseminated histoplasmosis, although this was not conclusively proven. The duration of treatment of PJI due to H capsulatum is not clear; however, all cases were treated with more than 7 months of antifungal therapy after resection arthroplasty. Longer (anticipated lifelong) courses were given to 2 immunocompromised individuals with retained joint components.
In general, a 2-stage exchange surgery (in which the infected joint prosthesis is removed during debridement, followed by antimicrobial treatment, and implantation of new prosthesis once the infection has cleared) is the recommended curative therapy for PJI.16 Because H capsulatum is efficient in forming a biofilm,17 a 2-stage exchange approach may also be needed for the complete eradication of PJI due to H capsulatum. Resection arthroplasty was used in 5 cases in this report, but the time to prosthesis reimplantation was highly variable. In addition to microbial and pharmacological factors, the decision on the timing of reimplantation considers host and social considerations, such as the desire to resume activities of daily living and the availability of caretakers during the debilitating periods related to surgical joint explantation. In 2 immunocompromised patients, resection arthroplasties were not performed because prostheses were well fixed and patients were functional without pain. Because of retained prostheses, both patients were receiving maintenance indefinite itraconazole suppression.5
Conclusion
Histoplasma capsulatum is a rare cause of PJI. The diagnosis of histoplasmosis was initially not suspected, and most patients were receiving antibacterial therapy for presumed culture-negative PJI until cultures surprisingly grew H capsulatum. Resection arthroplasty was performed in 5 of 7 cases, whereas debridement and retention strategy was used in 2 immunocompromised cases. All the cases were treated initially or fully with itraconazole for prolonged durations; one was also given amphotericin B over concerns of disseminated infection, while another one was transitioned to posaconazole after he developed heart failure attributed to itraconazole therapy.
Potential Competing Interests: The authors report no competing interests. | Recovered | ReactionOutcome | CC BY-NC-ND | 33718797 | 19,471,700 | 2021-02 |
What was the outcome of reaction 'Device failure'? | Prosthetic Joint Infections Due to Histoplasma capsulatum: A Report of 3 Cases.
Histoplasma capsulatum causes pneumonia and multisystemic disease in humans. Musculoskeletal involvement in histoplasmosis is most often tenosynovitis and rarely septic arthritis. Even more uncommon is the involvement of prosthetic joints. Here, we report a series of 3 cases of prosthetic joint failures caused by infection due to H capsulatum. Together with a review of 4 previously reported cases, we summarize host characteristics, clinical presentation, surgical approaches, antifungal management, and outcomes of this rare orthopedic joint infection.
Histoplasma capsulatum is a dimorphic fungus that is endemic throughoutthe Ohio and Mississippi river valleys and in certain pockets around the world. Although the fungus primarily causes lung infection, extrapulmonary histoplasmosis may involve the bone marrow, liver, brain, spleen, and the gastrointestinal tract.1,2 Osteomyelitis and septic arthritis are uncommon forms of histoplasmosis, while the involvement of joint prosthesis is even much more uncommon.2,3 In our review of the medical literature, we found only 4 cases of H capsulatum infection involving a prosthetic joint.4, 5, 6 Here, we present the clinical outcomes of 3 additional cases of prosthetic joint infection (PJI) due to H capsulatum.
Case Report
Case 1
A 77-year-old diabetic woman from Wisconsin presented to our clinic in 2019 with a 2-day history of right knee pain. She was a recipient of a deceased donor kidney transplant for end-stage renal disease due to autosomal dominant polycystic kidney disease in 1991, followed by a living-related donor kidney transplant due to allograft failure from calcineurin inhibitor nephrotoxicity in 2009. She was receiving maintenance immunosuppression with tacrolimus, mycophenolate mofetil, and prednisone. She also had bilateral total knee arthroplasties (TKAs) in 1998 (right) and 2012 (left) for degenerative joint disease. She did not have fever, chills, rigors, or other signs and symptoms of systemic illness. On physical examination, her right knee was erythematous, swollen, warm, and tender. Right knee synovial fluid analysis was suggestive of infection (Table). She had an elevated C-reactive protein level (23.1 mg/L; normal range, <8 mg/L). However, the synovial fluid cultures for bacteria were negative at 14 days.Table Clinical Characteristics and Outcomes of 7 Cases of Prosthetic Joint Infection Due to Histoplasma capsulatum
Case number (author; patient location) Age (y)/sex Comorbidity Presenting symptoms Joint involvement Synovial fluid cell count Synovial fluid or tissue culture Blood culture Serology Urine Histoplasma antigen Surgical treatment Antimicrobial treatment Outcome
1 (Berbari; Wisconsin) 77/F Kidney transplant Right knee pain and swelling Right knee joint 2530 cells (88% neutrophils) H capsulatum on surgical culture after 15 d of incubation Negative Negative Negative Debridement and retention Itraconazole (lifelong plan) Good function at last follow-up (12 mo)
2 (Berbari; Iowa) 69/F Rheumatoid arthritis, previous breast cancer Left shoulder pain Left shoulder joint 3512 cells (85% neutrophils) H capsulatum on surgical cultures after 12 d of incubation Negative Negative Negative Resection arthroplasty Itraconazole (ongoing) Reimplanted at 7 mo after resection
3 (Berbari; Nebraska) 75/M Hypertension Right knee wound drainage Right knee joint Not done H capsulatum on surgical cultures Negative Immunodiffusion M band positive Negative Resection arthroplasty Itraconazole (heart failure) → posaconazole × 1 y Reimplanted at 10 mo after the initiation of antifungal treatment
4 (Fowler; North Carolina) 84/F Polymyalgia rheumatica Sinus track over the left hip, pain and swelling Left hip joint 192,500 cells (88% granulocytes) H capsulatum on cultures 6 wk after surgery Not reported Not reported Not reported Debridement and retention Itraconazole (lifelong plan) Good function at the time of the last report (3 y)
5 (Meiyappan; travel to Ethiopia) 57/F Vasculitis Right knee pain and swelling Right knee joint 1800 cells (31% neutrophils) Histoplasma capsulatum on surgical and synovial fluid cultures Not reported Negative Negative Resection arthroplasty Amphotericin B + posaconazole (prolonged) Reimplanted at 11 mo; good function at 3.8 y
6 (Nowbakht; Wisconsin, Guatemala) 77/F Cardiac disease, diabetes mellitus Right knee pain Right knee joint Not reported H capsulatum on synovial fluid culture Negative Not reported Not reported Resection arthroplasty Itraconazole × 9 mo Reimplanted at 9 mo; good function at 2 y
7 (Foo; Singapore) 68/M End-stage renal disease receiving hemodialysis Right knee pain and swelling Right knee joint 9600 cells (56% neutrophils) H capsulatum on tissue culture Not reported Not reported Not reported Revision with implantation of prosthesis Itraconazole (ongoing at the time of the report) No recurrence of symptoms at 8 mo
F = female; M = male.
From references 4-7.
She underwent irrigation and debridement of the right knee joint. Intraoperatively, the right knee joint and prosthesis did not look grossly infected, and because the components were well-seated, they were retained. Fifteen days later, multiple cultures of intraoperative tissue specimens grew H capsulatum. Histoplasma serology and urinary Histoplasma antigen were negative. The chest radiograph was normal. Oral itraconazole was initiated. The dose of mycophenolate mofetil was reduced. Twelve months later, she was tolerating itraconazole and she had a pain-free functional right knee joint. There were no symptoms of immune reconstitution inflammatory syndrome. Because of the retained components and her immunocompromised status, chronic long term itraconazole suppression is planned.
Case 2
A 69-year-old woman from Iowa with rheumatoid arthritis taking etanercept and prednisone had multiple prosthetic joints implanted for painful destructive arthritis, including a left shoulder reverse arthroplasty in 2019. Eight months after her left shoulder arthroplasty, she complained of progressive lingering pain in her left shoulder. The C-reactive protein level was 13.8 mg/L. The radiograph of the shoulder revealed lucency about the glenoid component, which was suggestive of loosening. Left shoulder synovial fluid analysis revealed 3512 cells/μL with predominance of neutrophils. The culture of the synovial fluid was negative at 14 days.
Over concerns of underlying infection and component loosening, she underwent resection of left shoulder prosthesis. On histopathology, there was no evidence of acute inflammation. However, after 12 days of incubation, the intraoperative tissue cultures were positive for H capsulatum. The chest radiograph was normal. Histoplasma serology and urine antigen were negative. Oral itraconazole therapy was initiated. She has completed 7 months of itraconazole treatment before shoulder reimplantation. She continued receiving itraconazole treatment at the time of this report, with reassessment planned at 1 year. Despite not resuming etanercept treatment of her underlying rheumatoid arthritis, there were no symptoms related to immune reconstitution inflammatory syndrome.
Case 3
A 75-year-old man from Nebraska with hypertension underwent a right TKA in 2019 for degenerative joint disease at his local hospital. Two weeks later, he developed redness, swelling, and drainage of the surgical wound. He did not have a fever or other systemic symptoms of disseminated infection. During surgical debridement of the right knee, the wound was found to track deep into the knee joint prosthesis. Thus, resection TKA was performed. Multiple intraoperative tissue cultures grew H capsulatum. A computed tomography scan of the chest revealed mediastinal and hilar adenopathy. Histoplasma M band serology was positive, whereas urinary antigen was negative. Oral itraconazole therapy was initiated but later complicated by congestive heart failure. His antifungal regimen was subsequently changed to posaconazole.
Three months after resection arthroplasty, he presented to our hospital because of recurrence of right knee pain. Radiographic studies revealed a fractured knee spacer. He underwent irrigation, debridement, and exchange of the spacer. Intraoperative cultures were negative. The chest radiograph was normal. Histoplasma serology and urine Histoplasma antigen were negative. He continued receiving posaconazole therapy to complete 1 year of antifungal treatment. Right TKA reimplantation was performed 10 months after initial resection arthroplasty.
Discussion
Our case series of PJI due to H capsulatum add to 4 previously reported cases.4, 5, 6, 7 The clinical details of all 7 cases are summarized in the Table. One common characteristic of these cases is the delayed growth of H capsulatum in cultures, which led to the initial use of broad-spectrum antibacterial therapy for presumed culture-negative PJI.8,9 About 7% to 10% of PJIs are culture-negative, which is most often due to the recent intake of antibiotics.8,9 However, as highlighted in this report, it may be due to an unusual organism such as fungi.10 To diagnose unusual infections, one may need to prolong culture incubation, use special culture medium, obtain serology, or perform molecular testing. In addition, one should consider unusual infection when there is a history of exposures, such as residence or travel to endemic areas. Our 3 cases reside in the Midwest United States and were at risk of histoplasmosis. In all the reported cases, H capsulatum eventually grew in cultures, albeit delayed (among the cases in which time to growth was reported [at least 12 days]).5 Notably, Histoplasma serology and urinary Histoplasma antigen were negative in all except 1 case in which the M band was positive, suggesting that cases were localized infections involving only joint arthroplasty. This finding is consistent with a study reporting that most bone and joint infections due to H capsulatum are localized.3 Our third case, however, could have been a component of a systemic disease because of the reported presence of mediastinal and hilar adenopathy on the chest computed tomography scan. Histoplasma M band serology was also positive, although urine antigen and blood cultures were negative in this case. A detailed review of this third case did not identify unusual exposures that could have predisposed him to acquire a high burden of infection.
Overall, PJI due to fungi is extremely rare.11 All fungi accounted for only 1% of 3525 PJI cases at Mayo Clinic during 1996 to 2014, and most are due to Candida species.11 As mentioned, bone and joint infections due to H capsulatum are rare.3,12 None of the fungal PJIs in our institution from 1996 to 2014 were due to H capsulatum.11 In a review of 222 osteoarticular infections due to dimorphic fungi from 1970 to 2012, only 18 cases of H capsulatum were observed.3 Most cases of osteoarticular infections due to H capsulatum (89%) were described as a solitary bone lesion or joint infection.3 The clinical presentation in our cases is consistent with this report—PJI due to H capsulatum appear to be mostly isolated without other active focus of infection.
Factors that predispose patients to develop PJI due to H capsulatum is unknown. Although many of the cases are residents in endemic regions, there are no unifying host characteristics in the 7 patients. Although the infection has occurred in immunocompetent patients, it is possible that an underlying immunosuppression may predispose a patient to develop PJI due to H capsulatum.5 However, in a case-control study of immunocompromised transplant patients with PJI, no case of histoplasmosis was reported.13 In another report on 152 transplant patients with histoplasmosis, none involved bones or joints.14 In addition, the time to infection in relation to initial joint implantation was highly variable (from weeks to many years), suggesting that there is no unifying factor that predisposes to this rare orthopedic infection.
The antifungal treatment of histoplasmosis depends on the immune status and disease severity.15 Because of the localized nature of joint infection in our patients, oral itraconazole alone was used. In one of the previously reported cases, amphotericin B was also used over concerns of disseminated histoplasmosis, although this was not conclusively proven. The duration of treatment of PJI due to H capsulatum is not clear; however, all cases were treated with more than 7 months of antifungal therapy after resection arthroplasty. Longer (anticipated lifelong) courses were given to 2 immunocompromised individuals with retained joint components.
In general, a 2-stage exchange surgery (in which the infected joint prosthesis is removed during debridement, followed by antimicrobial treatment, and implantation of new prosthesis once the infection has cleared) is the recommended curative therapy for PJI.16 Because H capsulatum is efficient in forming a biofilm,17 a 2-stage exchange approach may also be needed for the complete eradication of PJI due to H capsulatum. Resection arthroplasty was used in 5 cases in this report, but the time to prosthesis reimplantation was highly variable. In addition to microbial and pharmacological factors, the decision on the timing of reimplantation considers host and social considerations, such as the desire to resume activities of daily living and the availability of caretakers during the debilitating periods related to surgical joint explantation. In 2 immunocompromised patients, resection arthroplasties were not performed because prostheses were well fixed and patients were functional without pain. Because of retained prostheses, both patients were receiving maintenance indefinite itraconazole suppression.5
Conclusion
Histoplasma capsulatum is a rare cause of PJI. The diagnosis of histoplasmosis was initially not suspected, and most patients were receiving antibacterial therapy for presumed culture-negative PJI until cultures surprisingly grew H capsulatum. Resection arthroplasty was performed in 5 of 7 cases, whereas debridement and retention strategy was used in 2 immunocompromised cases. All the cases were treated initially or fully with itraconazole for prolonged durations; one was also given amphotericin B over concerns of disseminated infection, while another one was transitioned to posaconazole after he developed heart failure attributed to itraconazole therapy.
Potential Competing Interests: The authors report no competing interests. | Recovered | ReactionOutcome | CC BY-NC-ND | 33718797 | 19,471,700 | 2021-02 |
What was the outcome of reaction 'Device related infection'? | Prosthetic Joint Infections Due to Histoplasma capsulatum: A Report of 3 Cases.
Histoplasma capsulatum causes pneumonia and multisystemic disease in humans. Musculoskeletal involvement in histoplasmosis is most often tenosynovitis and rarely septic arthritis. Even more uncommon is the involvement of prosthetic joints. Here, we report a series of 3 cases of prosthetic joint failures caused by infection due to H capsulatum. Together with a review of 4 previously reported cases, we summarize host characteristics, clinical presentation, surgical approaches, antifungal management, and outcomes of this rare orthopedic joint infection.
Histoplasma capsulatum is a dimorphic fungus that is endemic throughoutthe Ohio and Mississippi river valleys and in certain pockets around the world. Although the fungus primarily causes lung infection, extrapulmonary histoplasmosis may involve the bone marrow, liver, brain, spleen, and the gastrointestinal tract.1,2 Osteomyelitis and septic arthritis are uncommon forms of histoplasmosis, while the involvement of joint prosthesis is even much more uncommon.2,3 In our review of the medical literature, we found only 4 cases of H capsulatum infection involving a prosthetic joint.4, 5, 6 Here, we present the clinical outcomes of 3 additional cases of prosthetic joint infection (PJI) due to H capsulatum.
Case Report
Case 1
A 77-year-old diabetic woman from Wisconsin presented to our clinic in 2019 with a 2-day history of right knee pain. She was a recipient of a deceased donor kidney transplant for end-stage renal disease due to autosomal dominant polycystic kidney disease in 1991, followed by a living-related donor kidney transplant due to allograft failure from calcineurin inhibitor nephrotoxicity in 2009. She was receiving maintenance immunosuppression with tacrolimus, mycophenolate mofetil, and prednisone. She also had bilateral total knee arthroplasties (TKAs) in 1998 (right) and 2012 (left) for degenerative joint disease. She did not have fever, chills, rigors, or other signs and symptoms of systemic illness. On physical examination, her right knee was erythematous, swollen, warm, and tender. Right knee synovial fluid analysis was suggestive of infection (Table). She had an elevated C-reactive protein level (23.1 mg/L; normal range, <8 mg/L). However, the synovial fluid cultures for bacteria were negative at 14 days.Table Clinical Characteristics and Outcomes of 7 Cases of Prosthetic Joint Infection Due to Histoplasma capsulatum
Case number (author; patient location) Age (y)/sex Comorbidity Presenting symptoms Joint involvement Synovial fluid cell count Synovial fluid or tissue culture Blood culture Serology Urine Histoplasma antigen Surgical treatment Antimicrobial treatment Outcome
1 (Berbari; Wisconsin) 77/F Kidney transplant Right knee pain and swelling Right knee joint 2530 cells (88% neutrophils) H capsulatum on surgical culture after 15 d of incubation Negative Negative Negative Debridement and retention Itraconazole (lifelong plan) Good function at last follow-up (12 mo)
2 (Berbari; Iowa) 69/F Rheumatoid arthritis, previous breast cancer Left shoulder pain Left shoulder joint 3512 cells (85% neutrophils) H capsulatum on surgical cultures after 12 d of incubation Negative Negative Negative Resection arthroplasty Itraconazole (ongoing) Reimplanted at 7 mo after resection
3 (Berbari; Nebraska) 75/M Hypertension Right knee wound drainage Right knee joint Not done H capsulatum on surgical cultures Negative Immunodiffusion M band positive Negative Resection arthroplasty Itraconazole (heart failure) → posaconazole × 1 y Reimplanted at 10 mo after the initiation of antifungal treatment
4 (Fowler; North Carolina) 84/F Polymyalgia rheumatica Sinus track over the left hip, pain and swelling Left hip joint 192,500 cells (88% granulocytes) H capsulatum on cultures 6 wk after surgery Not reported Not reported Not reported Debridement and retention Itraconazole (lifelong plan) Good function at the time of the last report (3 y)
5 (Meiyappan; travel to Ethiopia) 57/F Vasculitis Right knee pain and swelling Right knee joint 1800 cells (31% neutrophils) Histoplasma capsulatum on surgical and synovial fluid cultures Not reported Negative Negative Resection arthroplasty Amphotericin B + posaconazole (prolonged) Reimplanted at 11 mo; good function at 3.8 y
6 (Nowbakht; Wisconsin, Guatemala) 77/F Cardiac disease, diabetes mellitus Right knee pain Right knee joint Not reported H capsulatum on synovial fluid culture Negative Not reported Not reported Resection arthroplasty Itraconazole × 9 mo Reimplanted at 9 mo; good function at 2 y
7 (Foo; Singapore) 68/M End-stage renal disease receiving hemodialysis Right knee pain and swelling Right knee joint 9600 cells (56% neutrophils) H capsulatum on tissue culture Not reported Not reported Not reported Revision with implantation of prosthesis Itraconazole (ongoing at the time of the report) No recurrence of symptoms at 8 mo
F = female; M = male.
From references 4-7.
She underwent irrigation and debridement of the right knee joint. Intraoperatively, the right knee joint and prosthesis did not look grossly infected, and because the components were well-seated, they were retained. Fifteen days later, multiple cultures of intraoperative tissue specimens grew H capsulatum. Histoplasma serology and urinary Histoplasma antigen were negative. The chest radiograph was normal. Oral itraconazole was initiated. The dose of mycophenolate mofetil was reduced. Twelve months later, she was tolerating itraconazole and she had a pain-free functional right knee joint. There were no symptoms of immune reconstitution inflammatory syndrome. Because of the retained components and her immunocompromised status, chronic long term itraconazole suppression is planned.
Case 2
A 69-year-old woman from Iowa with rheumatoid arthritis taking etanercept and prednisone had multiple prosthetic joints implanted for painful destructive arthritis, including a left shoulder reverse arthroplasty in 2019. Eight months after her left shoulder arthroplasty, she complained of progressive lingering pain in her left shoulder. The C-reactive protein level was 13.8 mg/L. The radiograph of the shoulder revealed lucency about the glenoid component, which was suggestive of loosening. Left shoulder synovial fluid analysis revealed 3512 cells/μL with predominance of neutrophils. The culture of the synovial fluid was negative at 14 days.
Over concerns of underlying infection and component loosening, she underwent resection of left shoulder prosthesis. On histopathology, there was no evidence of acute inflammation. However, after 12 days of incubation, the intraoperative tissue cultures were positive for H capsulatum. The chest radiograph was normal. Histoplasma serology and urine antigen were negative. Oral itraconazole therapy was initiated. She has completed 7 months of itraconazole treatment before shoulder reimplantation. She continued receiving itraconazole treatment at the time of this report, with reassessment planned at 1 year. Despite not resuming etanercept treatment of her underlying rheumatoid arthritis, there were no symptoms related to immune reconstitution inflammatory syndrome.
Case 3
A 75-year-old man from Nebraska with hypertension underwent a right TKA in 2019 for degenerative joint disease at his local hospital. Two weeks later, he developed redness, swelling, and drainage of the surgical wound. He did not have a fever or other systemic symptoms of disseminated infection. During surgical debridement of the right knee, the wound was found to track deep into the knee joint prosthesis. Thus, resection TKA was performed. Multiple intraoperative tissue cultures grew H capsulatum. A computed tomography scan of the chest revealed mediastinal and hilar adenopathy. Histoplasma M band serology was positive, whereas urinary antigen was negative. Oral itraconazole therapy was initiated but later complicated by congestive heart failure. His antifungal regimen was subsequently changed to posaconazole.
Three months after resection arthroplasty, he presented to our hospital because of recurrence of right knee pain. Radiographic studies revealed a fractured knee spacer. He underwent irrigation, debridement, and exchange of the spacer. Intraoperative cultures were negative. The chest radiograph was normal. Histoplasma serology and urine Histoplasma antigen were negative. He continued receiving posaconazole therapy to complete 1 year of antifungal treatment. Right TKA reimplantation was performed 10 months after initial resection arthroplasty.
Discussion
Our case series of PJI due to H capsulatum add to 4 previously reported cases.4, 5, 6, 7 The clinical details of all 7 cases are summarized in the Table. One common characteristic of these cases is the delayed growth of H capsulatum in cultures, which led to the initial use of broad-spectrum antibacterial therapy for presumed culture-negative PJI.8,9 About 7% to 10% of PJIs are culture-negative, which is most often due to the recent intake of antibiotics.8,9 However, as highlighted in this report, it may be due to an unusual organism such as fungi.10 To diagnose unusual infections, one may need to prolong culture incubation, use special culture medium, obtain serology, or perform molecular testing. In addition, one should consider unusual infection when there is a history of exposures, such as residence or travel to endemic areas. Our 3 cases reside in the Midwest United States and were at risk of histoplasmosis. In all the reported cases, H capsulatum eventually grew in cultures, albeit delayed (among the cases in which time to growth was reported [at least 12 days]).5 Notably, Histoplasma serology and urinary Histoplasma antigen were negative in all except 1 case in which the M band was positive, suggesting that cases were localized infections involving only joint arthroplasty. This finding is consistent with a study reporting that most bone and joint infections due to H capsulatum are localized.3 Our third case, however, could have been a component of a systemic disease because of the reported presence of mediastinal and hilar adenopathy on the chest computed tomography scan. Histoplasma M band serology was also positive, although urine antigen and blood cultures were negative in this case. A detailed review of this third case did not identify unusual exposures that could have predisposed him to acquire a high burden of infection.
Overall, PJI due to fungi is extremely rare.11 All fungi accounted for only 1% of 3525 PJI cases at Mayo Clinic during 1996 to 2014, and most are due to Candida species.11 As mentioned, bone and joint infections due to H capsulatum are rare.3,12 None of the fungal PJIs in our institution from 1996 to 2014 were due to H capsulatum.11 In a review of 222 osteoarticular infections due to dimorphic fungi from 1970 to 2012, only 18 cases of H capsulatum were observed.3 Most cases of osteoarticular infections due to H capsulatum (89%) were described as a solitary bone lesion or joint infection.3 The clinical presentation in our cases is consistent with this report—PJI due to H capsulatum appear to be mostly isolated without other active focus of infection.
Factors that predispose patients to develop PJI due to H capsulatum is unknown. Although many of the cases are residents in endemic regions, there are no unifying host characteristics in the 7 patients. Although the infection has occurred in immunocompetent patients, it is possible that an underlying immunosuppression may predispose a patient to develop PJI due to H capsulatum.5 However, in a case-control study of immunocompromised transplant patients with PJI, no case of histoplasmosis was reported.13 In another report on 152 transplant patients with histoplasmosis, none involved bones or joints.14 In addition, the time to infection in relation to initial joint implantation was highly variable (from weeks to many years), suggesting that there is no unifying factor that predisposes to this rare orthopedic infection.
The antifungal treatment of histoplasmosis depends on the immune status and disease severity.15 Because of the localized nature of joint infection in our patients, oral itraconazole alone was used. In one of the previously reported cases, amphotericin B was also used over concerns of disseminated histoplasmosis, although this was not conclusively proven. The duration of treatment of PJI due to H capsulatum is not clear; however, all cases were treated with more than 7 months of antifungal therapy after resection arthroplasty. Longer (anticipated lifelong) courses were given to 2 immunocompromised individuals with retained joint components.
In general, a 2-stage exchange surgery (in which the infected joint prosthesis is removed during debridement, followed by antimicrobial treatment, and implantation of new prosthesis once the infection has cleared) is the recommended curative therapy for PJI.16 Because H capsulatum is efficient in forming a biofilm,17 a 2-stage exchange approach may also be needed for the complete eradication of PJI due to H capsulatum. Resection arthroplasty was used in 5 cases in this report, but the time to prosthesis reimplantation was highly variable. In addition to microbial and pharmacological factors, the decision on the timing of reimplantation considers host and social considerations, such as the desire to resume activities of daily living and the availability of caretakers during the debilitating periods related to surgical joint explantation. In 2 immunocompromised patients, resection arthroplasties were not performed because prostheses were well fixed and patients were functional without pain. Because of retained prostheses, both patients were receiving maintenance indefinite itraconazole suppression.5
Conclusion
Histoplasma capsulatum is a rare cause of PJI. The diagnosis of histoplasmosis was initially not suspected, and most patients were receiving antibacterial therapy for presumed culture-negative PJI until cultures surprisingly grew H capsulatum. Resection arthroplasty was performed in 5 of 7 cases, whereas debridement and retention strategy was used in 2 immunocompromised cases. All the cases were treated initially or fully with itraconazole for prolonged durations; one was also given amphotericin B over concerns of disseminated infection, while another one was transitioned to posaconazole after he developed heart failure attributed to itraconazole therapy.
Potential Competing Interests: The authors report no competing interests. | Recovered | ReactionOutcome | CC BY-NC-ND | 33718797 | 19,471,700 | 2021-02 |
What was the outcome of reaction 'Histoplasmosis'? | Prosthetic Joint Infections Due to Histoplasma capsulatum: A Report of 3 Cases.
Histoplasma capsulatum causes pneumonia and multisystemic disease in humans. Musculoskeletal involvement in histoplasmosis is most often tenosynovitis and rarely septic arthritis. Even more uncommon is the involvement of prosthetic joints. Here, we report a series of 3 cases of prosthetic joint failures caused by infection due to H capsulatum. Together with a review of 4 previously reported cases, we summarize host characteristics, clinical presentation, surgical approaches, antifungal management, and outcomes of this rare orthopedic joint infection.
Histoplasma capsulatum is a dimorphic fungus that is endemic throughoutthe Ohio and Mississippi river valleys and in certain pockets around the world. Although the fungus primarily causes lung infection, extrapulmonary histoplasmosis may involve the bone marrow, liver, brain, spleen, and the gastrointestinal tract.1,2 Osteomyelitis and septic arthritis are uncommon forms of histoplasmosis, while the involvement of joint prosthesis is even much more uncommon.2,3 In our review of the medical literature, we found only 4 cases of H capsulatum infection involving a prosthetic joint.4, 5, 6 Here, we present the clinical outcomes of 3 additional cases of prosthetic joint infection (PJI) due to H capsulatum.
Case Report
Case 1
A 77-year-old diabetic woman from Wisconsin presented to our clinic in 2019 with a 2-day history of right knee pain. She was a recipient of a deceased donor kidney transplant for end-stage renal disease due to autosomal dominant polycystic kidney disease in 1991, followed by a living-related donor kidney transplant due to allograft failure from calcineurin inhibitor nephrotoxicity in 2009. She was receiving maintenance immunosuppression with tacrolimus, mycophenolate mofetil, and prednisone. She also had bilateral total knee arthroplasties (TKAs) in 1998 (right) and 2012 (left) for degenerative joint disease. She did not have fever, chills, rigors, or other signs and symptoms of systemic illness. On physical examination, her right knee was erythematous, swollen, warm, and tender. Right knee synovial fluid analysis was suggestive of infection (Table). She had an elevated C-reactive protein level (23.1 mg/L; normal range, <8 mg/L). However, the synovial fluid cultures for bacteria were negative at 14 days.Table Clinical Characteristics and Outcomes of 7 Cases of Prosthetic Joint Infection Due to Histoplasma capsulatum
Case number (author; patient location) Age (y)/sex Comorbidity Presenting symptoms Joint involvement Synovial fluid cell count Synovial fluid or tissue culture Blood culture Serology Urine Histoplasma antigen Surgical treatment Antimicrobial treatment Outcome
1 (Berbari; Wisconsin) 77/F Kidney transplant Right knee pain and swelling Right knee joint 2530 cells (88% neutrophils) H capsulatum on surgical culture after 15 d of incubation Negative Negative Negative Debridement and retention Itraconazole (lifelong plan) Good function at last follow-up (12 mo)
2 (Berbari; Iowa) 69/F Rheumatoid arthritis, previous breast cancer Left shoulder pain Left shoulder joint 3512 cells (85% neutrophils) H capsulatum on surgical cultures after 12 d of incubation Negative Negative Negative Resection arthroplasty Itraconazole (ongoing) Reimplanted at 7 mo after resection
3 (Berbari; Nebraska) 75/M Hypertension Right knee wound drainage Right knee joint Not done H capsulatum on surgical cultures Negative Immunodiffusion M band positive Negative Resection arthroplasty Itraconazole (heart failure) → posaconazole × 1 y Reimplanted at 10 mo after the initiation of antifungal treatment
4 (Fowler; North Carolina) 84/F Polymyalgia rheumatica Sinus track over the left hip, pain and swelling Left hip joint 192,500 cells (88% granulocytes) H capsulatum on cultures 6 wk after surgery Not reported Not reported Not reported Debridement and retention Itraconazole (lifelong plan) Good function at the time of the last report (3 y)
5 (Meiyappan; travel to Ethiopia) 57/F Vasculitis Right knee pain and swelling Right knee joint 1800 cells (31% neutrophils) Histoplasma capsulatum on surgical and synovial fluid cultures Not reported Negative Negative Resection arthroplasty Amphotericin B + posaconazole (prolonged) Reimplanted at 11 mo; good function at 3.8 y
6 (Nowbakht; Wisconsin, Guatemala) 77/F Cardiac disease, diabetes mellitus Right knee pain Right knee joint Not reported H capsulatum on synovial fluid culture Negative Not reported Not reported Resection arthroplasty Itraconazole × 9 mo Reimplanted at 9 mo; good function at 2 y
7 (Foo; Singapore) 68/M End-stage renal disease receiving hemodialysis Right knee pain and swelling Right knee joint 9600 cells (56% neutrophils) H capsulatum on tissue culture Not reported Not reported Not reported Revision with implantation of prosthesis Itraconazole (ongoing at the time of the report) No recurrence of symptoms at 8 mo
F = female; M = male.
From references 4-7.
She underwent irrigation and debridement of the right knee joint. Intraoperatively, the right knee joint and prosthesis did not look grossly infected, and because the components were well-seated, they were retained. Fifteen days later, multiple cultures of intraoperative tissue specimens grew H capsulatum. Histoplasma serology and urinary Histoplasma antigen were negative. The chest radiograph was normal. Oral itraconazole was initiated. The dose of mycophenolate mofetil was reduced. Twelve months later, she was tolerating itraconazole and she had a pain-free functional right knee joint. There were no symptoms of immune reconstitution inflammatory syndrome. Because of the retained components and her immunocompromised status, chronic long term itraconazole suppression is planned.
Case 2
A 69-year-old woman from Iowa with rheumatoid arthritis taking etanercept and prednisone had multiple prosthetic joints implanted for painful destructive arthritis, including a left shoulder reverse arthroplasty in 2019. Eight months after her left shoulder arthroplasty, she complained of progressive lingering pain in her left shoulder. The C-reactive protein level was 13.8 mg/L. The radiograph of the shoulder revealed lucency about the glenoid component, which was suggestive of loosening. Left shoulder synovial fluid analysis revealed 3512 cells/μL with predominance of neutrophils. The culture of the synovial fluid was negative at 14 days.
Over concerns of underlying infection and component loosening, she underwent resection of left shoulder prosthesis. On histopathology, there was no evidence of acute inflammation. However, after 12 days of incubation, the intraoperative tissue cultures were positive for H capsulatum. The chest radiograph was normal. Histoplasma serology and urine antigen were negative. Oral itraconazole therapy was initiated. She has completed 7 months of itraconazole treatment before shoulder reimplantation. She continued receiving itraconazole treatment at the time of this report, with reassessment planned at 1 year. Despite not resuming etanercept treatment of her underlying rheumatoid arthritis, there were no symptoms related to immune reconstitution inflammatory syndrome.
Case 3
A 75-year-old man from Nebraska with hypertension underwent a right TKA in 2019 for degenerative joint disease at his local hospital. Two weeks later, he developed redness, swelling, and drainage of the surgical wound. He did not have a fever or other systemic symptoms of disseminated infection. During surgical debridement of the right knee, the wound was found to track deep into the knee joint prosthesis. Thus, resection TKA was performed. Multiple intraoperative tissue cultures grew H capsulatum. A computed tomography scan of the chest revealed mediastinal and hilar adenopathy. Histoplasma M band serology was positive, whereas urinary antigen was negative. Oral itraconazole therapy was initiated but later complicated by congestive heart failure. His antifungal regimen was subsequently changed to posaconazole.
Three months after resection arthroplasty, he presented to our hospital because of recurrence of right knee pain. Radiographic studies revealed a fractured knee spacer. He underwent irrigation, debridement, and exchange of the spacer. Intraoperative cultures were negative. The chest radiograph was normal. Histoplasma serology and urine Histoplasma antigen were negative. He continued receiving posaconazole therapy to complete 1 year of antifungal treatment. Right TKA reimplantation was performed 10 months after initial resection arthroplasty.
Discussion
Our case series of PJI due to H capsulatum add to 4 previously reported cases.4, 5, 6, 7 The clinical details of all 7 cases are summarized in the Table. One common characteristic of these cases is the delayed growth of H capsulatum in cultures, which led to the initial use of broad-spectrum antibacterial therapy for presumed culture-negative PJI.8,9 About 7% to 10% of PJIs are culture-negative, which is most often due to the recent intake of antibiotics.8,9 However, as highlighted in this report, it may be due to an unusual organism such as fungi.10 To diagnose unusual infections, one may need to prolong culture incubation, use special culture medium, obtain serology, or perform molecular testing. In addition, one should consider unusual infection when there is a history of exposures, such as residence or travel to endemic areas. Our 3 cases reside in the Midwest United States and were at risk of histoplasmosis. In all the reported cases, H capsulatum eventually grew in cultures, albeit delayed (among the cases in which time to growth was reported [at least 12 days]).5 Notably, Histoplasma serology and urinary Histoplasma antigen were negative in all except 1 case in which the M band was positive, suggesting that cases were localized infections involving only joint arthroplasty. This finding is consistent with a study reporting that most bone and joint infections due to H capsulatum are localized.3 Our third case, however, could have been a component of a systemic disease because of the reported presence of mediastinal and hilar adenopathy on the chest computed tomography scan. Histoplasma M band serology was also positive, although urine antigen and blood cultures were negative in this case. A detailed review of this third case did not identify unusual exposures that could have predisposed him to acquire a high burden of infection.
Overall, PJI due to fungi is extremely rare.11 All fungi accounted for only 1% of 3525 PJI cases at Mayo Clinic during 1996 to 2014, and most are due to Candida species.11 As mentioned, bone and joint infections due to H capsulatum are rare.3,12 None of the fungal PJIs in our institution from 1996 to 2014 were due to H capsulatum.11 In a review of 222 osteoarticular infections due to dimorphic fungi from 1970 to 2012, only 18 cases of H capsulatum were observed.3 Most cases of osteoarticular infections due to H capsulatum (89%) were described as a solitary bone lesion or joint infection.3 The clinical presentation in our cases is consistent with this report—PJI due to H capsulatum appear to be mostly isolated without other active focus of infection.
Factors that predispose patients to develop PJI due to H capsulatum is unknown. Although many of the cases are residents in endemic regions, there are no unifying host characteristics in the 7 patients. Although the infection has occurred in immunocompetent patients, it is possible that an underlying immunosuppression may predispose a patient to develop PJI due to H capsulatum.5 However, in a case-control study of immunocompromised transplant patients with PJI, no case of histoplasmosis was reported.13 In another report on 152 transplant patients with histoplasmosis, none involved bones or joints.14 In addition, the time to infection in relation to initial joint implantation was highly variable (from weeks to many years), suggesting that there is no unifying factor that predisposes to this rare orthopedic infection.
The antifungal treatment of histoplasmosis depends on the immune status and disease severity.15 Because of the localized nature of joint infection in our patients, oral itraconazole alone was used. In one of the previously reported cases, amphotericin B was also used over concerns of disseminated histoplasmosis, although this was not conclusively proven. The duration of treatment of PJI due to H capsulatum is not clear; however, all cases were treated with more than 7 months of antifungal therapy after resection arthroplasty. Longer (anticipated lifelong) courses were given to 2 immunocompromised individuals with retained joint components.
In general, a 2-stage exchange surgery (in which the infected joint prosthesis is removed during debridement, followed by antimicrobial treatment, and implantation of new prosthesis once the infection has cleared) is the recommended curative therapy for PJI.16 Because H capsulatum is efficient in forming a biofilm,17 a 2-stage exchange approach may also be needed for the complete eradication of PJI due to H capsulatum. Resection arthroplasty was used in 5 cases in this report, but the time to prosthesis reimplantation was highly variable. In addition to microbial and pharmacological factors, the decision on the timing of reimplantation considers host and social considerations, such as the desire to resume activities of daily living and the availability of caretakers during the debilitating periods related to surgical joint explantation. In 2 immunocompromised patients, resection arthroplasties were not performed because prostheses were well fixed and patients were functional without pain. Because of retained prostheses, both patients were receiving maintenance indefinite itraconazole suppression.5
Conclusion
Histoplasma capsulatum is a rare cause of PJI. The diagnosis of histoplasmosis was initially not suspected, and most patients were receiving antibacterial therapy for presumed culture-negative PJI until cultures surprisingly grew H capsulatum. Resection arthroplasty was performed in 5 of 7 cases, whereas debridement and retention strategy was used in 2 immunocompromised cases. All the cases were treated initially or fully with itraconazole for prolonged durations; one was also given amphotericin B over concerns of disseminated infection, while another one was transitioned to posaconazole after he developed heart failure attributed to itraconazole therapy.
Potential Competing Interests: The authors report no competing interests. | Recovered | ReactionOutcome | CC BY-NC-ND | 33718797 | 19,471,700 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Drug ineffective'. | Efficacy of 177Lu-Dotatate Therapy in the Treatment of Recurrent Meningioma.
A 62-year-old man presented with a history of atypical meningioma (World Health Organization grade II) and recurrent as anaplastic meningioma (World Health Organization grade III). His previous treatments included multiple surgical resections, fractionated radiation therapy, stereotactic radiosurgery, everolimus/octreotide long-acting release, bevacizumab, and hydroxyurea. Magnetic resonance imaging revealed rapid volumetric progression over the prior 9 months, with a near tripling in size from 29.9 cm3 to 80.4 cm3. Indium In 111 octreotide scanning confirmed the presence of somatostatin receptors within the tumor. Lutetium Lu 177 dotatate was administered intravenously at a dose of 200 mCi per dose every 8 weeks for 4 cycles. Treatment was tolerated very well, with no notable adverse events. Tumor volume initially increased to 98.3 cm3 after cycle 1 of treatment and subsequently decreased to 91.2 cm3 after cycle 2. Eight months after treatment onset, the tumor volume remained stable (93.4 cm3).
Atypical (World Health Organization [WHO] grade II) and anaplastic (WHO grade III) meningiomas are relatively uncommon subtypes, collectively accounting for approximately 20% of all meningiomas.1,2 Both are associated with poorer prognosis than the more common WHO grade I meningiomas. Initial treatments typically involve surgical resection and radiation therapy but recurrence is common. Systemic therapy is often pursued thereafter, with limited known efficacy. Somatostatin receptor (SSTR) types 2 through 5 are commonly overexpressed in meningiomas and have been found to be present in the vast majority of meningiomas, regardless of tumor grade.3, 4, 5 Lutetium Lu 177 (177Lu) dotatate, which is a beta particle–emitting somatostatin analogue, has been approved recently by the US Food and Drug Administration (FDA) for the treatment of SSTR-positive gastroenteropancreatic neuroendocrine tumors. Given the near-universal presence of SSTRs in meningiomas, 177Lu-dotatate may be a promising strategy for treatment of an anaplastic meningioma.6
For the case reported in this article, informed consent was obtained for treatment with the described therapy. All data are documented in the electronic medical record. Mayo Clinic Institutional Review Board approval was not required to report findings of routine clinical patient care.
Report of Case
Presentation and Treatment
A 62-year-old man presented with recurrent meningioma. A left parietal meningioma had been diagnosed in 2000 after the patient had presented with a seizure. He underwent his first surgical resection in 2001, with pathologic examination documenting WHO grade I meningioma.
Two years after the initial diagnosis, he presented with recurrent generalized tonic-clonic seizures. Subsequent magnet resonance imaging (MRI) revealed multiple sites of new meningiomas along with tumor regrowth at the primary tumor site. Between 2003 and 2017, the patient experienced several episodes of tumor recurrence and underwent multimodality therapy, including a total of 8 surgical interventions and fractionated external beam radiotherapy with a total of 59.4 Gy in 33 fractions in 2002 and 54 Gy in 30 fractions in 2013. Additionally, stereotactic radiosurgery with gamma knife was performed in 2007 and 2016 with maximum radiation doses of 24 Gy and 28 Gy. Findings on pathologic examination of specimens from surgical resection of tumor in 2016 were suspicious for anaplastic progression (WHO grade III), which was confirmed at the next resection in 2017 when pathologic examination revealed over 40 mitotic cells per 10 high-power fields within an area of the tumor. Eventually, continued progression of the treatment-refractory anaplastic meningioma led to referral of the patient for systemic therapeutic measures because standard surgical and radiation options had been exhausted.
An indium In 111 (111In)-octreotide scan performed in 2017 revealed radiotracer uptake within the meningioma, documenting SSTR expression (Figure A). The patient was initially treated with everolimus and octreotide long-acting release based on results of the CEVOREM (Combination of Everolimus and Octreotide in Resistant Meningiomas) trial.7 At progression, he was treated with bevacizumab and hydroxyurea.8,9 However, these systemic therapies were not effective, and the meningioma exhibited rapid volumetric progression, with a near tripling in size from 29.9 cm3 in November 2017 to 80.4 cm3 in August 2018 (Figure B-D).Figure Indium In 111-octreotide scan performed in November 2017 showing somatostatin receptor expression (A; arrow indicates meningioma). Corresponding T1-weighted, gadolinium-enhanced magnetic resonance images (B, coronal; C, axial). Axial T1-weighted, gadolinium-enhanced magnetic resonance images from August 2018 prior to initiation of lutetium Lu 177-dotatate treatment (D) and April 2019 after 4 cycles of lutetium Lu 177-dotatatate (E).
The FDA approval of 177Lu-dotatate for SSTR-positive gastroenteropancreatic neuroendocrine tumors in January 2018 led to consideration of off-label treatment with 177Lu-dotatate for this patient. Consistent with the methods of the NETTER (Neuroendocrine Tumors Therapy) trial,10 this patient was treated with a 4-cycle regimen of 200 mCi intravenously, administered every 8 weeks. The rationale behind this approach was that this dosing schedule was known to be safe and effective in neuroendocrine tumors and there was no prior evidence to suspect that a different dosing regimen would be superior in meningioma. Treatment with
177Lu-dotatate was initiated in August 2018, and serial MRI and volumetric analyses were conducted throughout the treatment to monitor for tumor progression.
Outcomes and Follow-up
The treatment was well tolerated with no notable adverse events. Specifically, there was no evidence of substantial hematologic or renal toxicity and no evidence of cerebral radiation necrosis. Two months after initiation of treatment, the patient experienced 2 breakthrough partial seizures, which were controlled with an increase in levetiracetam dose. Neurologic examination findings remained unremarkable other than a mild gait abnormality that preceded initiation of 177Lu-dotatate therapy.
Magnetic resonance imaging confirmed that the tumor volume had slightly increased to 98 cm3 after 1 cycle of treatment with subsequent decrease to 91 cm3 after cycle 2. The tumor volume remained essentially stable thereafter for the duration of therapy, with a volume of 93 cm3 after completion of cycle 4, 8 months after treatment onset (Figure E). The degree of perilesional T2-weighted hyperintensities and mass effect of the tumor were also essentially stable during therapy.
Two months after completion of 177Lu-dotatate therapy, the patient presented to the clinic with increased seizure frequency, fatigue, and right-sided weakness. Magnetic resonance imaging revealed interval tumor enlargement (tumor volume, 116 cm3) with an increase in vasogenic edema. Meningioma-directed therapy was not considered in the immediate term because of the need for reconstructive surgery for a wound defect. Dexamethasone and levetiracetam doses were adjusted for symptomatic relief.
Five months after the completion of treatment, the patient presented to the emergency department with altered mental status. Magnetic resonance imaging revealed a substantial increase in tumor volume (146 cm3) compared with the previously stable volume of 93 cm3. Ultimately, the decision was made to proceed with symptomatic management only, and the patient entered hospice care, dying shortly thereafter.
Discussion
Meningiomas represent one of the largest subgroups of intracranial neoplasms, accounting for about 34% of all central nervous system tumors. They are usually benign, slow-growing tumors. The WHO subclassifies meningioma as grade I, grade II (atypical), and grade III (anaplastic), with progressively more aggressive behavior and poorer prognosis.2,11 Surgical procedures remain the standard initial treatment for most meningiomas.12 However, it is not unusual for WHO grade II and III meningiomas to recur, thus leading to multiple surgical resections and repeated radiation therapy.13
Treatment options for relapsing and treatment-refractory anaplastic meningiomas are limited. Recent studies have found that tumor growth in meningioma is related to hormonal factors.14 Meningiomas have been found to express estrogen and progesterone receptors, androgens, and nonsteroid hormones, including somatostatin. In vitro and in vivo studies have found that meningioma, regardless of histology and classification, may express SSTRs, and immunohistochemical studies document the presence of 5 SSTR subtypes (SSTR1 through SSTR5) in tumor tissue, with predominance of SSTR2 and SSTR5.3 The presence of SSTRs within meningiomas can be noninvasively documented by 111In-octreotide single-photon emission computed tomography, which historically has been the standard imaging modality for this purpose. More recently, gallium Ga 68-dotatate positron emission tomography has become the test of choice because it offers higher resolution and shorter imaging time and should more closely approximate the distribution of 177Lu-dotatate therapy. The use of SSTR analogues for meningioma treatment is recognized in European studies, with reported long-term stable disease and long overall survival in rapidly progressive multiple recurrent anaplastic meningiomas.6,10
177Lu-dotatate is a radioconjugate consisting of the tyrosine-containing SSTR analogue Tyr3-octreotate (TATE) conjugated with the bifunctional, macrocyclic chelating agent tetra-azacyclododecanetetra-acetic acid (DOTA) and radiolabeled with the beta-emitting radioisotope 177Lu. 177Lu-dotatate binds to SSRTs, with high affinity for SSTR2. On binding and internalization, this radioconjugate specifically delivers a cytotoxic dose of beta radiation to these SSTR-positive cells, with a very short beta particle length limiting radiation exposure of surrounding brain tissue.
In January 2018, 177Lu-dotatate therapy was FDA approved in the United States for adult patients with SSTR-positive gastroenteropancreatic neuroendocrine tumor, including those of the foregut, midgut, and hindgut. This FDA approval was mainly based on the results of the phase III NETTER trial.10 Eligible patients were adults who had inoperable midgut neuroendocrine tumors that had metastasized or were locally advanced and those who had disease progression on computed tomography or MRI. The patients were treated with 7.4 GBq (200 mCi) of 177Lu-dotatate infused intravenously every 8 weeks for 4 cycles. Treated patients had notable improvements in response rate, progression-free survival, and mortality.
A transient increase in tumor volume was observed after the first cycle, which may have represented delayed treatment effect or pseudoprogression secondary to vasogenic edema. Subsequently, the tumor volume decreased by 7% and remained stable with minimal change in volume throughout the remainder of treatment. Over the 4 cycles (8 months) of treatment, the tumor volume increased only 16% during 177Lu-dotatate therapy in comparison to the rapid increase of 169% during the 8 months prior to 177Lu-dotatate initiation. However, the tumor resumed its previous rapid growth after therapy cessation.
There was no evidence of any major biochemical and hematologic toxicities during therapy or posttreatment. The short path length of 177Lu-dotatate in comparison to other agents like yttrium Y 90-dotatoc may have allowed for better results with lower risk of cerebral radiation necrosis given previous external beam radiation and gamma knife radiosurgery.15
A phase II clinical trial has been activated at Mayo Clinic to determine the progression-free survival associated with 177Lu-dotatate in patients with recurrent meningioma.16
Conclusion
177Lu-dotatate is a theranostic radioisotope for targeted radionuclide therapy of SSTR-positive gastroenteropancreatic neuroendocrine tumors that may also be a reasonable consideration in SSTR-expressing meningiomas. In this case, treatment was well tolerated and demonstrated biological effect by temporarily halting growth in a previous rapidly growing WHO grade III meningioma refractory to numerous previous therapies. Although benefit was transient in this patient, this observation nonetheless builds confidence that 177Lu-dotatate may be a safe and effective strategy for the treatment of SSTR-positive recurrent meningiomas. Until additional data are available from larger efficacy studies and cost-benefit analyses, this approach should be considered only as a salvage therapy option once other standard therapies have been exhausted.
Acknowledgments
The content of this work is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Grant Support: This work was supported in part by grant K12CA090628 from the National Cancer Institute, United States (S.H.K).
Potential Competing Interests: The authors report no competing interests. | BEVACIZUMAB, EVEROLIMUS, HYDROXYUREA, OCTREOTIDE | DrugsGivenReaction | CC BY-NC-ND | 33718799 | 19,472,503 | 2021-02 |
What was the outcome of reaction 'Drug ineffective'? | Efficacy of 177Lu-Dotatate Therapy in the Treatment of Recurrent Meningioma.
A 62-year-old man presented with a history of atypical meningioma (World Health Organization grade II) and recurrent as anaplastic meningioma (World Health Organization grade III). His previous treatments included multiple surgical resections, fractionated radiation therapy, stereotactic radiosurgery, everolimus/octreotide long-acting release, bevacizumab, and hydroxyurea. Magnetic resonance imaging revealed rapid volumetric progression over the prior 9 months, with a near tripling in size from 29.9 cm3 to 80.4 cm3. Indium In 111 octreotide scanning confirmed the presence of somatostatin receptors within the tumor. Lutetium Lu 177 dotatate was administered intravenously at a dose of 200 mCi per dose every 8 weeks for 4 cycles. Treatment was tolerated very well, with no notable adverse events. Tumor volume initially increased to 98.3 cm3 after cycle 1 of treatment and subsequently decreased to 91.2 cm3 after cycle 2. Eight months after treatment onset, the tumor volume remained stable (93.4 cm3).
Atypical (World Health Organization [WHO] grade II) and anaplastic (WHO grade III) meningiomas are relatively uncommon subtypes, collectively accounting for approximately 20% of all meningiomas.1,2 Both are associated with poorer prognosis than the more common WHO grade I meningiomas. Initial treatments typically involve surgical resection and radiation therapy but recurrence is common. Systemic therapy is often pursued thereafter, with limited known efficacy. Somatostatin receptor (SSTR) types 2 through 5 are commonly overexpressed in meningiomas and have been found to be present in the vast majority of meningiomas, regardless of tumor grade.3, 4, 5 Lutetium Lu 177 (177Lu) dotatate, which is a beta particle–emitting somatostatin analogue, has been approved recently by the US Food and Drug Administration (FDA) for the treatment of SSTR-positive gastroenteropancreatic neuroendocrine tumors. Given the near-universal presence of SSTRs in meningiomas, 177Lu-dotatate may be a promising strategy for treatment of an anaplastic meningioma.6
For the case reported in this article, informed consent was obtained for treatment with the described therapy. All data are documented in the electronic medical record. Mayo Clinic Institutional Review Board approval was not required to report findings of routine clinical patient care.
Report of Case
Presentation and Treatment
A 62-year-old man presented with recurrent meningioma. A left parietal meningioma had been diagnosed in 2000 after the patient had presented with a seizure. He underwent his first surgical resection in 2001, with pathologic examination documenting WHO grade I meningioma.
Two years after the initial diagnosis, he presented with recurrent generalized tonic-clonic seizures. Subsequent magnet resonance imaging (MRI) revealed multiple sites of new meningiomas along with tumor regrowth at the primary tumor site. Between 2003 and 2017, the patient experienced several episodes of tumor recurrence and underwent multimodality therapy, including a total of 8 surgical interventions and fractionated external beam radiotherapy with a total of 59.4 Gy in 33 fractions in 2002 and 54 Gy in 30 fractions in 2013. Additionally, stereotactic radiosurgery with gamma knife was performed in 2007 and 2016 with maximum radiation doses of 24 Gy and 28 Gy. Findings on pathologic examination of specimens from surgical resection of tumor in 2016 were suspicious for anaplastic progression (WHO grade III), which was confirmed at the next resection in 2017 when pathologic examination revealed over 40 mitotic cells per 10 high-power fields within an area of the tumor. Eventually, continued progression of the treatment-refractory anaplastic meningioma led to referral of the patient for systemic therapeutic measures because standard surgical and radiation options had been exhausted.
An indium In 111 (111In)-octreotide scan performed in 2017 revealed radiotracer uptake within the meningioma, documenting SSTR expression (Figure A). The patient was initially treated with everolimus and octreotide long-acting release based on results of the CEVOREM (Combination of Everolimus and Octreotide in Resistant Meningiomas) trial.7 At progression, he was treated with bevacizumab and hydroxyurea.8,9 However, these systemic therapies were not effective, and the meningioma exhibited rapid volumetric progression, with a near tripling in size from 29.9 cm3 in November 2017 to 80.4 cm3 in August 2018 (Figure B-D).Figure Indium In 111-octreotide scan performed in November 2017 showing somatostatin receptor expression (A; arrow indicates meningioma). Corresponding T1-weighted, gadolinium-enhanced magnetic resonance images (B, coronal; C, axial). Axial T1-weighted, gadolinium-enhanced magnetic resonance images from August 2018 prior to initiation of lutetium Lu 177-dotatate treatment (D) and April 2019 after 4 cycles of lutetium Lu 177-dotatatate (E).
The FDA approval of 177Lu-dotatate for SSTR-positive gastroenteropancreatic neuroendocrine tumors in January 2018 led to consideration of off-label treatment with 177Lu-dotatate for this patient. Consistent with the methods of the NETTER (Neuroendocrine Tumors Therapy) trial,10 this patient was treated with a 4-cycle regimen of 200 mCi intravenously, administered every 8 weeks. The rationale behind this approach was that this dosing schedule was known to be safe and effective in neuroendocrine tumors and there was no prior evidence to suspect that a different dosing regimen would be superior in meningioma. Treatment with
177Lu-dotatate was initiated in August 2018, and serial MRI and volumetric analyses were conducted throughout the treatment to monitor for tumor progression.
Outcomes and Follow-up
The treatment was well tolerated with no notable adverse events. Specifically, there was no evidence of substantial hematologic or renal toxicity and no evidence of cerebral radiation necrosis. Two months after initiation of treatment, the patient experienced 2 breakthrough partial seizures, which were controlled with an increase in levetiracetam dose. Neurologic examination findings remained unremarkable other than a mild gait abnormality that preceded initiation of 177Lu-dotatate therapy.
Magnetic resonance imaging confirmed that the tumor volume had slightly increased to 98 cm3 after 1 cycle of treatment with subsequent decrease to 91 cm3 after cycle 2. The tumor volume remained essentially stable thereafter for the duration of therapy, with a volume of 93 cm3 after completion of cycle 4, 8 months after treatment onset (Figure E). The degree of perilesional T2-weighted hyperintensities and mass effect of the tumor were also essentially stable during therapy.
Two months after completion of 177Lu-dotatate therapy, the patient presented to the clinic with increased seizure frequency, fatigue, and right-sided weakness. Magnetic resonance imaging revealed interval tumor enlargement (tumor volume, 116 cm3) with an increase in vasogenic edema. Meningioma-directed therapy was not considered in the immediate term because of the need for reconstructive surgery for a wound defect. Dexamethasone and levetiracetam doses were adjusted for symptomatic relief.
Five months after the completion of treatment, the patient presented to the emergency department with altered mental status. Magnetic resonance imaging revealed a substantial increase in tumor volume (146 cm3) compared with the previously stable volume of 93 cm3. Ultimately, the decision was made to proceed with symptomatic management only, and the patient entered hospice care, dying shortly thereafter.
Discussion
Meningiomas represent one of the largest subgroups of intracranial neoplasms, accounting for about 34% of all central nervous system tumors. They are usually benign, slow-growing tumors. The WHO subclassifies meningioma as grade I, grade II (atypical), and grade III (anaplastic), with progressively more aggressive behavior and poorer prognosis.2,11 Surgical procedures remain the standard initial treatment for most meningiomas.12 However, it is not unusual for WHO grade II and III meningiomas to recur, thus leading to multiple surgical resections and repeated radiation therapy.13
Treatment options for relapsing and treatment-refractory anaplastic meningiomas are limited. Recent studies have found that tumor growth in meningioma is related to hormonal factors.14 Meningiomas have been found to express estrogen and progesterone receptors, androgens, and nonsteroid hormones, including somatostatin. In vitro and in vivo studies have found that meningioma, regardless of histology and classification, may express SSTRs, and immunohistochemical studies document the presence of 5 SSTR subtypes (SSTR1 through SSTR5) in tumor tissue, with predominance of SSTR2 and SSTR5.3 The presence of SSTRs within meningiomas can be noninvasively documented by 111In-octreotide single-photon emission computed tomography, which historically has been the standard imaging modality for this purpose. More recently, gallium Ga 68-dotatate positron emission tomography has become the test of choice because it offers higher resolution and shorter imaging time and should more closely approximate the distribution of 177Lu-dotatate therapy. The use of SSTR analogues for meningioma treatment is recognized in European studies, with reported long-term stable disease and long overall survival in rapidly progressive multiple recurrent anaplastic meningiomas.6,10
177Lu-dotatate is a radioconjugate consisting of the tyrosine-containing SSTR analogue Tyr3-octreotate (TATE) conjugated with the bifunctional, macrocyclic chelating agent tetra-azacyclododecanetetra-acetic acid (DOTA) and radiolabeled with the beta-emitting radioisotope 177Lu. 177Lu-dotatate binds to SSRTs, with high affinity for SSTR2. On binding and internalization, this radioconjugate specifically delivers a cytotoxic dose of beta radiation to these SSTR-positive cells, with a very short beta particle length limiting radiation exposure of surrounding brain tissue.
In January 2018, 177Lu-dotatate therapy was FDA approved in the United States for adult patients with SSTR-positive gastroenteropancreatic neuroendocrine tumor, including those of the foregut, midgut, and hindgut. This FDA approval was mainly based on the results of the phase III NETTER trial.10 Eligible patients were adults who had inoperable midgut neuroendocrine tumors that had metastasized or were locally advanced and those who had disease progression on computed tomography or MRI. The patients were treated with 7.4 GBq (200 mCi) of 177Lu-dotatate infused intravenously every 8 weeks for 4 cycles. Treated patients had notable improvements in response rate, progression-free survival, and mortality.
A transient increase in tumor volume was observed after the first cycle, which may have represented delayed treatment effect or pseudoprogression secondary to vasogenic edema. Subsequently, the tumor volume decreased by 7% and remained stable with minimal change in volume throughout the remainder of treatment. Over the 4 cycles (8 months) of treatment, the tumor volume increased only 16% during 177Lu-dotatate therapy in comparison to the rapid increase of 169% during the 8 months prior to 177Lu-dotatate initiation. However, the tumor resumed its previous rapid growth after therapy cessation.
There was no evidence of any major biochemical and hematologic toxicities during therapy or posttreatment. The short path length of 177Lu-dotatate in comparison to other agents like yttrium Y 90-dotatoc may have allowed for better results with lower risk of cerebral radiation necrosis given previous external beam radiation and gamma knife radiosurgery.15
A phase II clinical trial has been activated at Mayo Clinic to determine the progression-free survival associated with 177Lu-dotatate in patients with recurrent meningioma.16
Conclusion
177Lu-dotatate is a theranostic radioisotope for targeted radionuclide therapy of SSTR-positive gastroenteropancreatic neuroendocrine tumors that may also be a reasonable consideration in SSTR-expressing meningiomas. In this case, treatment was well tolerated and demonstrated biological effect by temporarily halting growth in a previous rapidly growing WHO grade III meningioma refractory to numerous previous therapies. Although benefit was transient in this patient, this observation nonetheless builds confidence that 177Lu-dotatate may be a safe and effective strategy for the treatment of SSTR-positive recurrent meningiomas. Until additional data are available from larger efficacy studies and cost-benefit analyses, this approach should be considered only as a salvage therapy option once other standard therapies have been exhausted.
Acknowledgments
The content of this work is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Grant Support: This work was supported in part by grant K12CA090628 from the National Cancer Institute, United States (S.H.K).
Potential Competing Interests: The authors report no competing interests. | Fatal | ReactionOutcome | CC BY-NC-ND | 33718799 | 19,472,503 | 2021-02 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Cutaneous sarcoidosis'. | Combined pulsed-dye laser and medical therapy for treatment of cutaneous sarcoidosis lesions: a case report.
Sarcoidosis is a systemic granulomatous disease of unknown etiology that most frequently occurs in the lungs. However, cutaneous lesions are often the primary sign. Cutaneous sarcoidosis is difficult to treat, although different therapies have been applied. We herein report a case in which cutaneous sarcoidosis was treated with pulsed-dye laser (PDL) therapy along with oral administration of acitretin and hydroxychloroquine; no topical medications were applied. All patient details are de-identified. The treatment areas gradually improved after several courses of PDL therapy. This case illustrates that PDL therapy can serve as an auxiliary treatment for cutaneous sarcoidosis.
Introduction
Sarcoidosis is a systemic granulomatous disease of unknown etiology that most frequently occurs in the lungs. However, cutaneous lesions, which are present in approximately 10% to 30% of affected patients, are often the primary sign.1,2 Cutaneous sarcoidosis has various clinical manifestations, such as papules and plaques. Regardless of the clinical presentation, noncaseating granulomas are the most characteristic histologic manifestation. Cutaneous sarcoidosis remains difficult to treat, although different therapies have been applied. We herein report a case in which cutaneous sarcoidosis was improved by pulsed-dye laser (PDL) therapy.
Case report
In December 2018, a 46-year-old Asian woman presented for a consultation regarding left facial lesions (Figure 1(a)), which had been gradually increasing in size for more than 7 years. Symptoms such as pain and pruritus were absent.
Figure 1. Clinical photographs before and after the treatment. (a) Initial findings. The main lesion was a red plaque with an approximate size of 8 × 8 cm anterior to the left ear and involving the auricle. Telangiectasia was visible in the middle of the plaque, surrounded by several papules and nodules. (b) Findings 15 months post-treatment. The plaque was thinned (which was more obvious in the marked yellow rectangular regions) after the use of oral acitretin and hydroxychloroquine and 10 pulsed-dye laser therapy sessions. These changes were more clearly visible in the central lesion, where several tiny islands of normal skin had appeared (green arrows). Telangiectasia was less obvious (yellow arrow). However, some small papules had emerged in the periphery (blue arrows).
The patient had visited several hospitals. Laboratory test results were normal, and a T-SPOT.TB test was negative. Chest computed tomography showed no evidence of organ involvement. Pulmonary function tests and electrocardiograms were normal. Repeated histopathological examinations suggested a noncaseating granulomatous disease, and acid-fast staining was negative. Furthermore, polymerase chain reaction analysis detected no Mycobacterium tuberculosis DNA in the tissues.
The patient had been diagnosed with cutaneous sarcoidosis. She was treated with oral acitretin, compound glycyrrhizin, and thalidomide and topical pimecrolimus. However, no improvement was obtained, and the lesions continued to grow.
In our hospital, another lesion biopsy was taken for further diagnosis. Microscopic examination showed granulomas composed of epithelioid cells (Figure 2). This histologic pattern of sarcoidosis was similar to her previous pathologic findings. Considering the results of all auxiliary examinations to date, we made the same diagnosis of cutaneous sarcoidosis because there was no obvious evidence of infection or systemic involvement. The treatment regimen included oral administration of acitretin (10 mg per day) and hydroxychloroquine (200 mg twice a day) along with PDL therapy. The lesions were treated with PDL therapy at 14 J/cm2 (595 nm, 6 ms, 7 mm) at 1- to 2-month intervals. Fifteen months later, the treated area showed improvement after 10 PDL therapy sessions (Figure 1(b)). The plaque was thinner than at the beginning of the treatment and was more visible in the central area of the lesion. Moreover, several tiny islands of normal skin had appeared. Telangiectasia was less obvious than at baseline. However, small emerging papules continued to develop in the periphery.
Figure 2. Histologic findings (hematoxylin and eosin staining). The epidermis was generally normal. Granulomas composed of epithelioid cells and surrounded by lymphocyte-dominant inflammation were present in the dermis. Multinucleated giant cells were also visible.
Discussion
PDL therapy is mainly applied to treat vascular dermatoses based on the principle of selective photothermolysis to destroy vessels. Nevertheless, several reports have described extensive use of PDL therapy in patients with lupus erythematosus,3–5 hypertrophic scars,6–8 and keloids.6,8,9 The mechanisms likely underlying the action of PDL therapy include stimulation of immunomodulatory processes,8 downregulation of connective tissue growth factor expression,10 and hypoxemia leading to alterations in local collagen production and increases in matrix metalloproteinases. 11 Therefore, these factors may represent targets for PDL therapy in patients with cutaneous sarcoidosis. In addition, various studies have suggested that angiogenic and angiostatic factors contribute to the pathogenesis of sarcoidosis,12–14 further emphasizing the value and significance of PDL therapy in the treatment of cutaneous sarcoidosis.
A case of cutaneous sarcoidosis successfully treated with PDL therapy alone was reported by Roos et al.15 Although the patient subsequently developed disease involvement in the right calf and eyes that was eventually eliminated by systemic steroids, she remained free of skin lesions after the steroids were discontinued. Our case is similar in that improvements were achieved only in the areas treated with PDL therapy. Although oral anti-inflammatory medications were simultaneously administered, lesions continued to develop in the periphery. We considered that the patient’s disease condition was in a progressive stage despite the lack of evidence of systemic involvement. In fact, glucocorticoid therapy was preferred for better control but was refused by the patient. Therefore, we could only provide relatively conservative treatment involving oral acitretin and hydroxychloroquine along with PDL therapy. Meanwhile, regular reevaluation by chest computed tomography and related laboratory tests was necessary to monitor for systemic involvement.
As in the previous report by Roos et al.,15 our case reveals the potential clinical value of PDL therapy in the treatment of cutaneous sarcoidosis. Large-sample randomized controlled trials are required to investigate the application of this therapy and elucidate the precise underlying mechanisms of action.
Ethics approval and consent to participate: The patient provided written informed consent for biopsy, treatment, and publication of this report. This case report was written according to the CARE Guidelines. Ethics committee/review board approval was not needed because of the nature of this study (case report).
Declaration of conflicting interest: The authors declare that there is no conflict of interest.
Funding: This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.
Authors’ contributions: WD: manuscript drafting and laser operation.
CL: manuscript revision.
YS: laser operation.
WZ: study design and critical revision of the manuscript.
All authors have read and approved the manuscript.
Availability of data and materials: The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.
ORCID iD: Wei Zhang https://orcid.org/0000-0002-5497-9926 | ACITRETIN, HYDROXYCHLOROQUINE | DrugsGivenReaction | CC BY-NC | 33719667 | 19,146,020 | 2021-03 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Drug ineffective for unapproved indication'. | Combined pulsed-dye laser and medical therapy for treatment of cutaneous sarcoidosis lesions: a case report.
Sarcoidosis is a systemic granulomatous disease of unknown etiology that most frequently occurs in the lungs. However, cutaneous lesions are often the primary sign. Cutaneous sarcoidosis is difficult to treat, although different therapies have been applied. We herein report a case in which cutaneous sarcoidosis was treated with pulsed-dye laser (PDL) therapy along with oral administration of acitretin and hydroxychloroquine; no topical medications were applied. All patient details are de-identified. The treatment areas gradually improved after several courses of PDL therapy. This case illustrates that PDL therapy can serve as an auxiliary treatment for cutaneous sarcoidosis.
Introduction
Sarcoidosis is a systemic granulomatous disease of unknown etiology that most frequently occurs in the lungs. However, cutaneous lesions, which are present in approximately 10% to 30% of affected patients, are often the primary sign.1,2 Cutaneous sarcoidosis has various clinical manifestations, such as papules and plaques. Regardless of the clinical presentation, noncaseating granulomas are the most characteristic histologic manifestation. Cutaneous sarcoidosis remains difficult to treat, although different therapies have been applied. We herein report a case in which cutaneous sarcoidosis was improved by pulsed-dye laser (PDL) therapy.
Case report
In December 2018, a 46-year-old Asian woman presented for a consultation regarding left facial lesions (Figure 1(a)), which had been gradually increasing in size for more than 7 years. Symptoms such as pain and pruritus were absent.
Figure 1. Clinical photographs before and after the treatment. (a) Initial findings. The main lesion was a red plaque with an approximate size of 8 × 8 cm anterior to the left ear and involving the auricle. Telangiectasia was visible in the middle of the plaque, surrounded by several papules and nodules. (b) Findings 15 months post-treatment. The plaque was thinned (which was more obvious in the marked yellow rectangular regions) after the use of oral acitretin and hydroxychloroquine and 10 pulsed-dye laser therapy sessions. These changes were more clearly visible in the central lesion, where several tiny islands of normal skin had appeared (green arrows). Telangiectasia was less obvious (yellow arrow). However, some small papules had emerged in the periphery (blue arrows).
The patient had visited several hospitals. Laboratory test results were normal, and a T-SPOT.TB test was negative. Chest computed tomography showed no evidence of organ involvement. Pulmonary function tests and electrocardiograms were normal. Repeated histopathological examinations suggested a noncaseating granulomatous disease, and acid-fast staining was negative. Furthermore, polymerase chain reaction analysis detected no Mycobacterium tuberculosis DNA in the tissues.
The patient had been diagnosed with cutaneous sarcoidosis. She was treated with oral acitretin, compound glycyrrhizin, and thalidomide and topical pimecrolimus. However, no improvement was obtained, and the lesions continued to grow.
In our hospital, another lesion biopsy was taken for further diagnosis. Microscopic examination showed granulomas composed of epithelioid cells (Figure 2). This histologic pattern of sarcoidosis was similar to her previous pathologic findings. Considering the results of all auxiliary examinations to date, we made the same diagnosis of cutaneous sarcoidosis because there was no obvious evidence of infection or systemic involvement. The treatment regimen included oral administration of acitretin (10 mg per day) and hydroxychloroquine (200 mg twice a day) along with PDL therapy. The lesions were treated with PDL therapy at 14 J/cm2 (595 nm, 6 ms, 7 mm) at 1- to 2-month intervals. Fifteen months later, the treated area showed improvement after 10 PDL therapy sessions (Figure 1(b)). The plaque was thinner than at the beginning of the treatment and was more visible in the central area of the lesion. Moreover, several tiny islands of normal skin had appeared. Telangiectasia was less obvious than at baseline. However, small emerging papules continued to develop in the periphery.
Figure 2. Histologic findings (hematoxylin and eosin staining). The epidermis was generally normal. Granulomas composed of epithelioid cells and surrounded by lymphocyte-dominant inflammation were present in the dermis. Multinucleated giant cells were also visible.
Discussion
PDL therapy is mainly applied to treat vascular dermatoses based on the principle of selective photothermolysis to destroy vessels. Nevertheless, several reports have described extensive use of PDL therapy in patients with lupus erythematosus,3–5 hypertrophic scars,6–8 and keloids.6,8,9 The mechanisms likely underlying the action of PDL therapy include stimulation of immunomodulatory processes,8 downregulation of connective tissue growth factor expression,10 and hypoxemia leading to alterations in local collagen production and increases in matrix metalloproteinases. 11 Therefore, these factors may represent targets for PDL therapy in patients with cutaneous sarcoidosis. In addition, various studies have suggested that angiogenic and angiostatic factors contribute to the pathogenesis of sarcoidosis,12–14 further emphasizing the value and significance of PDL therapy in the treatment of cutaneous sarcoidosis.
A case of cutaneous sarcoidosis successfully treated with PDL therapy alone was reported by Roos et al.15 Although the patient subsequently developed disease involvement in the right calf and eyes that was eventually eliminated by systemic steroids, she remained free of skin lesions after the steroids were discontinued. Our case is similar in that improvements were achieved only in the areas treated with PDL therapy. Although oral anti-inflammatory medications were simultaneously administered, lesions continued to develop in the periphery. We considered that the patient’s disease condition was in a progressive stage despite the lack of evidence of systemic involvement. In fact, glucocorticoid therapy was preferred for better control but was refused by the patient. Therefore, we could only provide relatively conservative treatment involving oral acitretin and hydroxychloroquine along with PDL therapy. Meanwhile, regular reevaluation by chest computed tomography and related laboratory tests was necessary to monitor for systemic involvement.
As in the previous report by Roos et al.,15 our case reveals the potential clinical value of PDL therapy in the treatment of cutaneous sarcoidosis. Large-sample randomized controlled trials are required to investigate the application of this therapy and elucidate the precise underlying mechanisms of action.
Ethics approval and consent to participate: The patient provided written informed consent for biopsy, treatment, and publication of this report. This case report was written according to the CARE Guidelines. Ethics committee/review board approval was not needed because of the nature of this study (case report).
Declaration of conflicting interest: The authors declare that there is no conflict of interest.
Funding: This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.
Authors’ contributions: WD: manuscript drafting and laser operation.
CL: manuscript revision.
YS: laser operation.
WZ: study design and critical revision of the manuscript.
All authors have read and approved the manuscript.
Availability of data and materials: The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.
ORCID iD: Wei Zhang https://orcid.org/0000-0002-5497-9926 | ACITRETIN, HYDROXYCHLOROQUINE | DrugsGivenReaction | CC BY-NC | 33719667 | 19,146,020 | 2021-03 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Off label use'. | Combined pulsed-dye laser and medical therapy for treatment of cutaneous sarcoidosis lesions: a case report.
Sarcoidosis is a systemic granulomatous disease of unknown etiology that most frequently occurs in the lungs. However, cutaneous lesions are often the primary sign. Cutaneous sarcoidosis is difficult to treat, although different therapies have been applied. We herein report a case in which cutaneous sarcoidosis was treated with pulsed-dye laser (PDL) therapy along with oral administration of acitretin and hydroxychloroquine; no topical medications were applied. All patient details are de-identified. The treatment areas gradually improved after several courses of PDL therapy. This case illustrates that PDL therapy can serve as an auxiliary treatment for cutaneous sarcoidosis.
Introduction
Sarcoidosis is a systemic granulomatous disease of unknown etiology that most frequently occurs in the lungs. However, cutaneous lesions, which are present in approximately 10% to 30% of affected patients, are often the primary sign.1,2 Cutaneous sarcoidosis has various clinical manifestations, such as papules and plaques. Regardless of the clinical presentation, noncaseating granulomas are the most characteristic histologic manifestation. Cutaneous sarcoidosis remains difficult to treat, although different therapies have been applied. We herein report a case in which cutaneous sarcoidosis was improved by pulsed-dye laser (PDL) therapy.
Case report
In December 2018, a 46-year-old Asian woman presented for a consultation regarding left facial lesions (Figure 1(a)), which had been gradually increasing in size for more than 7 years. Symptoms such as pain and pruritus were absent.
Figure 1. Clinical photographs before and after the treatment. (a) Initial findings. The main lesion was a red plaque with an approximate size of 8 × 8 cm anterior to the left ear and involving the auricle. Telangiectasia was visible in the middle of the plaque, surrounded by several papules and nodules. (b) Findings 15 months post-treatment. The plaque was thinned (which was more obvious in the marked yellow rectangular regions) after the use of oral acitretin and hydroxychloroquine and 10 pulsed-dye laser therapy sessions. These changes were more clearly visible in the central lesion, where several tiny islands of normal skin had appeared (green arrows). Telangiectasia was less obvious (yellow arrow). However, some small papules had emerged in the periphery (blue arrows).
The patient had visited several hospitals. Laboratory test results were normal, and a T-SPOT.TB test was negative. Chest computed tomography showed no evidence of organ involvement. Pulmonary function tests and electrocardiograms were normal. Repeated histopathological examinations suggested a noncaseating granulomatous disease, and acid-fast staining was negative. Furthermore, polymerase chain reaction analysis detected no Mycobacterium tuberculosis DNA in the tissues.
The patient had been diagnosed with cutaneous sarcoidosis. She was treated with oral acitretin, compound glycyrrhizin, and thalidomide and topical pimecrolimus. However, no improvement was obtained, and the lesions continued to grow.
In our hospital, another lesion biopsy was taken for further diagnosis. Microscopic examination showed granulomas composed of epithelioid cells (Figure 2). This histologic pattern of sarcoidosis was similar to her previous pathologic findings. Considering the results of all auxiliary examinations to date, we made the same diagnosis of cutaneous sarcoidosis because there was no obvious evidence of infection or systemic involvement. The treatment regimen included oral administration of acitretin (10 mg per day) and hydroxychloroquine (200 mg twice a day) along with PDL therapy. The lesions were treated with PDL therapy at 14 J/cm2 (595 nm, 6 ms, 7 mm) at 1- to 2-month intervals. Fifteen months later, the treated area showed improvement after 10 PDL therapy sessions (Figure 1(b)). The plaque was thinner than at the beginning of the treatment and was more visible in the central area of the lesion. Moreover, several tiny islands of normal skin had appeared. Telangiectasia was less obvious than at baseline. However, small emerging papules continued to develop in the periphery.
Figure 2. Histologic findings (hematoxylin and eosin staining). The epidermis was generally normal. Granulomas composed of epithelioid cells and surrounded by lymphocyte-dominant inflammation were present in the dermis. Multinucleated giant cells were also visible.
Discussion
PDL therapy is mainly applied to treat vascular dermatoses based on the principle of selective photothermolysis to destroy vessels. Nevertheless, several reports have described extensive use of PDL therapy in patients with lupus erythematosus,3–5 hypertrophic scars,6–8 and keloids.6,8,9 The mechanisms likely underlying the action of PDL therapy include stimulation of immunomodulatory processes,8 downregulation of connective tissue growth factor expression,10 and hypoxemia leading to alterations in local collagen production and increases in matrix metalloproteinases. 11 Therefore, these factors may represent targets for PDL therapy in patients with cutaneous sarcoidosis. In addition, various studies have suggested that angiogenic and angiostatic factors contribute to the pathogenesis of sarcoidosis,12–14 further emphasizing the value and significance of PDL therapy in the treatment of cutaneous sarcoidosis.
A case of cutaneous sarcoidosis successfully treated with PDL therapy alone was reported by Roos et al.15 Although the patient subsequently developed disease involvement in the right calf and eyes that was eventually eliminated by systemic steroids, she remained free of skin lesions after the steroids were discontinued. Our case is similar in that improvements were achieved only in the areas treated with PDL therapy. Although oral anti-inflammatory medications were simultaneously administered, lesions continued to develop in the periphery. We considered that the patient’s disease condition was in a progressive stage despite the lack of evidence of systemic involvement. In fact, glucocorticoid therapy was preferred for better control but was refused by the patient. Therefore, we could only provide relatively conservative treatment involving oral acitretin and hydroxychloroquine along with PDL therapy. Meanwhile, regular reevaluation by chest computed tomography and related laboratory tests was necessary to monitor for systemic involvement.
As in the previous report by Roos et al.,15 our case reveals the potential clinical value of PDL therapy in the treatment of cutaneous sarcoidosis. Large-sample randomized controlled trials are required to investigate the application of this therapy and elucidate the precise underlying mechanisms of action.
Ethics approval and consent to participate: The patient provided written informed consent for biopsy, treatment, and publication of this report. This case report was written according to the CARE Guidelines. Ethics committee/review board approval was not needed because of the nature of this study (case report).
Declaration of conflicting interest: The authors declare that there is no conflict of interest.
Funding: This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.
Authors’ contributions: WD: manuscript drafting and laser operation.
CL: manuscript revision.
YS: laser operation.
WZ: study design and critical revision of the manuscript.
All authors have read and approved the manuscript.
Availability of data and materials: The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.
ORCID iD: Wei Zhang https://orcid.org/0000-0002-5497-9926 | ACITRETIN, HYDROXYCHLOROQUINE | DrugsGivenReaction | CC BY-NC | 33719667 | 19,146,020 | 2021-03 |
What was the administration route of drug 'ACITRETIN'? | Combined pulsed-dye laser and medical therapy for treatment of cutaneous sarcoidosis lesions: a case report.
Sarcoidosis is a systemic granulomatous disease of unknown etiology that most frequently occurs in the lungs. However, cutaneous lesions are often the primary sign. Cutaneous sarcoidosis is difficult to treat, although different therapies have been applied. We herein report a case in which cutaneous sarcoidosis was treated with pulsed-dye laser (PDL) therapy along with oral administration of acitretin and hydroxychloroquine; no topical medications were applied. All patient details are de-identified. The treatment areas gradually improved after several courses of PDL therapy. This case illustrates that PDL therapy can serve as an auxiliary treatment for cutaneous sarcoidosis.
Introduction
Sarcoidosis is a systemic granulomatous disease of unknown etiology that most frequently occurs in the lungs. However, cutaneous lesions, which are present in approximately 10% to 30% of affected patients, are often the primary sign.1,2 Cutaneous sarcoidosis has various clinical manifestations, such as papules and plaques. Regardless of the clinical presentation, noncaseating granulomas are the most characteristic histologic manifestation. Cutaneous sarcoidosis remains difficult to treat, although different therapies have been applied. We herein report a case in which cutaneous sarcoidosis was improved by pulsed-dye laser (PDL) therapy.
Case report
In December 2018, a 46-year-old Asian woman presented for a consultation regarding left facial lesions (Figure 1(a)), which had been gradually increasing in size for more than 7 years. Symptoms such as pain and pruritus were absent.
Figure 1. Clinical photographs before and after the treatment. (a) Initial findings. The main lesion was a red plaque with an approximate size of 8 × 8 cm anterior to the left ear and involving the auricle. Telangiectasia was visible in the middle of the plaque, surrounded by several papules and nodules. (b) Findings 15 months post-treatment. The plaque was thinned (which was more obvious in the marked yellow rectangular regions) after the use of oral acitretin and hydroxychloroquine and 10 pulsed-dye laser therapy sessions. These changes were more clearly visible in the central lesion, where several tiny islands of normal skin had appeared (green arrows). Telangiectasia was less obvious (yellow arrow). However, some small papules had emerged in the periphery (blue arrows).
The patient had visited several hospitals. Laboratory test results were normal, and a T-SPOT.TB test was negative. Chest computed tomography showed no evidence of organ involvement. Pulmonary function tests and electrocardiograms were normal. Repeated histopathological examinations suggested a noncaseating granulomatous disease, and acid-fast staining was negative. Furthermore, polymerase chain reaction analysis detected no Mycobacterium tuberculosis DNA in the tissues.
The patient had been diagnosed with cutaneous sarcoidosis. She was treated with oral acitretin, compound glycyrrhizin, and thalidomide and topical pimecrolimus. However, no improvement was obtained, and the lesions continued to grow.
In our hospital, another lesion biopsy was taken for further diagnosis. Microscopic examination showed granulomas composed of epithelioid cells (Figure 2). This histologic pattern of sarcoidosis was similar to her previous pathologic findings. Considering the results of all auxiliary examinations to date, we made the same diagnosis of cutaneous sarcoidosis because there was no obvious evidence of infection or systemic involvement. The treatment regimen included oral administration of acitretin (10 mg per day) and hydroxychloroquine (200 mg twice a day) along with PDL therapy. The lesions were treated with PDL therapy at 14 J/cm2 (595 nm, 6 ms, 7 mm) at 1- to 2-month intervals. Fifteen months later, the treated area showed improvement after 10 PDL therapy sessions (Figure 1(b)). The plaque was thinner than at the beginning of the treatment and was more visible in the central area of the lesion. Moreover, several tiny islands of normal skin had appeared. Telangiectasia was less obvious than at baseline. However, small emerging papules continued to develop in the periphery.
Figure 2. Histologic findings (hematoxylin and eosin staining). The epidermis was generally normal. Granulomas composed of epithelioid cells and surrounded by lymphocyte-dominant inflammation were present in the dermis. Multinucleated giant cells were also visible.
Discussion
PDL therapy is mainly applied to treat vascular dermatoses based on the principle of selective photothermolysis to destroy vessels. Nevertheless, several reports have described extensive use of PDL therapy in patients with lupus erythematosus,3–5 hypertrophic scars,6–8 and keloids.6,8,9 The mechanisms likely underlying the action of PDL therapy include stimulation of immunomodulatory processes,8 downregulation of connective tissue growth factor expression,10 and hypoxemia leading to alterations in local collagen production and increases in matrix metalloproteinases. 11 Therefore, these factors may represent targets for PDL therapy in patients with cutaneous sarcoidosis. In addition, various studies have suggested that angiogenic and angiostatic factors contribute to the pathogenesis of sarcoidosis,12–14 further emphasizing the value and significance of PDL therapy in the treatment of cutaneous sarcoidosis.
A case of cutaneous sarcoidosis successfully treated with PDL therapy alone was reported by Roos et al.15 Although the patient subsequently developed disease involvement in the right calf and eyes that was eventually eliminated by systemic steroids, she remained free of skin lesions after the steroids were discontinued. Our case is similar in that improvements were achieved only in the areas treated with PDL therapy. Although oral anti-inflammatory medications were simultaneously administered, lesions continued to develop in the periphery. We considered that the patient’s disease condition was in a progressive stage despite the lack of evidence of systemic involvement. In fact, glucocorticoid therapy was preferred for better control but was refused by the patient. Therefore, we could only provide relatively conservative treatment involving oral acitretin and hydroxychloroquine along with PDL therapy. Meanwhile, regular reevaluation by chest computed tomography and related laboratory tests was necessary to monitor for systemic involvement.
As in the previous report by Roos et al.,15 our case reveals the potential clinical value of PDL therapy in the treatment of cutaneous sarcoidosis. Large-sample randomized controlled trials are required to investigate the application of this therapy and elucidate the precise underlying mechanisms of action.
Ethics approval and consent to participate: The patient provided written informed consent for biopsy, treatment, and publication of this report. This case report was written according to the CARE Guidelines. Ethics committee/review board approval was not needed because of the nature of this study (case report).
Declaration of conflicting interest: The authors declare that there is no conflict of interest.
Funding: This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.
Authors’ contributions: WD: manuscript drafting and laser operation.
CL: manuscript revision.
YS: laser operation.
WZ: study design and critical revision of the manuscript.
All authors have read and approved the manuscript.
Availability of data and materials: The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.
ORCID iD: Wei Zhang https://orcid.org/0000-0002-5497-9926 | Oral | DrugAdministrationRoute | CC BY-NC | 33719667 | 19,146,020 | 2021-03 |
What was the outcome of reaction 'Cutaneous sarcoidosis'? | Combined pulsed-dye laser and medical therapy for treatment of cutaneous sarcoidosis lesions: a case report.
Sarcoidosis is a systemic granulomatous disease of unknown etiology that most frequently occurs in the lungs. However, cutaneous lesions are often the primary sign. Cutaneous sarcoidosis is difficult to treat, although different therapies have been applied. We herein report a case in which cutaneous sarcoidosis was treated with pulsed-dye laser (PDL) therapy along with oral administration of acitretin and hydroxychloroquine; no topical medications were applied. All patient details are de-identified. The treatment areas gradually improved after several courses of PDL therapy. This case illustrates that PDL therapy can serve as an auxiliary treatment for cutaneous sarcoidosis.
Introduction
Sarcoidosis is a systemic granulomatous disease of unknown etiology that most frequently occurs in the lungs. However, cutaneous lesions, which are present in approximately 10% to 30% of affected patients, are often the primary sign.1,2 Cutaneous sarcoidosis has various clinical manifestations, such as papules and plaques. Regardless of the clinical presentation, noncaseating granulomas are the most characteristic histologic manifestation. Cutaneous sarcoidosis remains difficult to treat, although different therapies have been applied. We herein report a case in which cutaneous sarcoidosis was improved by pulsed-dye laser (PDL) therapy.
Case report
In December 2018, a 46-year-old Asian woman presented for a consultation regarding left facial lesions (Figure 1(a)), which had been gradually increasing in size for more than 7 years. Symptoms such as pain and pruritus were absent.
Figure 1. Clinical photographs before and after the treatment. (a) Initial findings. The main lesion was a red plaque with an approximate size of 8 × 8 cm anterior to the left ear and involving the auricle. Telangiectasia was visible in the middle of the plaque, surrounded by several papules and nodules. (b) Findings 15 months post-treatment. The plaque was thinned (which was more obvious in the marked yellow rectangular regions) after the use of oral acitretin and hydroxychloroquine and 10 pulsed-dye laser therapy sessions. These changes were more clearly visible in the central lesion, where several tiny islands of normal skin had appeared (green arrows). Telangiectasia was less obvious (yellow arrow). However, some small papules had emerged in the periphery (blue arrows).
The patient had visited several hospitals. Laboratory test results were normal, and a T-SPOT.TB test was negative. Chest computed tomography showed no evidence of organ involvement. Pulmonary function tests and electrocardiograms were normal. Repeated histopathological examinations suggested a noncaseating granulomatous disease, and acid-fast staining was negative. Furthermore, polymerase chain reaction analysis detected no Mycobacterium tuberculosis DNA in the tissues.
The patient had been diagnosed with cutaneous sarcoidosis. She was treated with oral acitretin, compound glycyrrhizin, and thalidomide and topical pimecrolimus. However, no improvement was obtained, and the lesions continued to grow.
In our hospital, another lesion biopsy was taken for further diagnosis. Microscopic examination showed granulomas composed of epithelioid cells (Figure 2). This histologic pattern of sarcoidosis was similar to her previous pathologic findings. Considering the results of all auxiliary examinations to date, we made the same diagnosis of cutaneous sarcoidosis because there was no obvious evidence of infection or systemic involvement. The treatment regimen included oral administration of acitretin (10 mg per day) and hydroxychloroquine (200 mg twice a day) along with PDL therapy. The lesions were treated with PDL therapy at 14 J/cm2 (595 nm, 6 ms, 7 mm) at 1- to 2-month intervals. Fifteen months later, the treated area showed improvement after 10 PDL therapy sessions (Figure 1(b)). The plaque was thinner than at the beginning of the treatment and was more visible in the central area of the lesion. Moreover, several tiny islands of normal skin had appeared. Telangiectasia was less obvious than at baseline. However, small emerging papules continued to develop in the periphery.
Figure 2. Histologic findings (hematoxylin and eosin staining). The epidermis was generally normal. Granulomas composed of epithelioid cells and surrounded by lymphocyte-dominant inflammation were present in the dermis. Multinucleated giant cells were also visible.
Discussion
PDL therapy is mainly applied to treat vascular dermatoses based on the principle of selective photothermolysis to destroy vessels. Nevertheless, several reports have described extensive use of PDL therapy in patients with lupus erythematosus,3–5 hypertrophic scars,6–8 and keloids.6,8,9 The mechanisms likely underlying the action of PDL therapy include stimulation of immunomodulatory processes,8 downregulation of connective tissue growth factor expression,10 and hypoxemia leading to alterations in local collagen production and increases in matrix metalloproteinases. 11 Therefore, these factors may represent targets for PDL therapy in patients with cutaneous sarcoidosis. In addition, various studies have suggested that angiogenic and angiostatic factors contribute to the pathogenesis of sarcoidosis,12–14 further emphasizing the value and significance of PDL therapy in the treatment of cutaneous sarcoidosis.
A case of cutaneous sarcoidosis successfully treated with PDL therapy alone was reported by Roos et al.15 Although the patient subsequently developed disease involvement in the right calf and eyes that was eventually eliminated by systemic steroids, she remained free of skin lesions after the steroids were discontinued. Our case is similar in that improvements were achieved only in the areas treated with PDL therapy. Although oral anti-inflammatory medications were simultaneously administered, lesions continued to develop in the periphery. We considered that the patient’s disease condition was in a progressive stage despite the lack of evidence of systemic involvement. In fact, glucocorticoid therapy was preferred for better control but was refused by the patient. Therefore, we could only provide relatively conservative treatment involving oral acitretin and hydroxychloroquine along with PDL therapy. Meanwhile, regular reevaluation by chest computed tomography and related laboratory tests was necessary to monitor for systemic involvement.
As in the previous report by Roos et al.,15 our case reveals the potential clinical value of PDL therapy in the treatment of cutaneous sarcoidosis. Large-sample randomized controlled trials are required to investigate the application of this therapy and elucidate the precise underlying mechanisms of action.
Ethics approval and consent to participate: The patient provided written informed consent for biopsy, treatment, and publication of this report. This case report was written according to the CARE Guidelines. Ethics committee/review board approval was not needed because of the nature of this study (case report).
Declaration of conflicting interest: The authors declare that there is no conflict of interest.
Funding: This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.
Authors’ contributions: WD: manuscript drafting and laser operation.
CL: manuscript revision.
YS: laser operation.
WZ: study design and critical revision of the manuscript.
All authors have read and approved the manuscript.
Availability of data and materials: The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.
ORCID iD: Wei Zhang https://orcid.org/0000-0002-5497-9926 | Not recovered | ReactionOutcome | CC BY-NC | 33719667 | 19,146,020 | 2021-03 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Acute myocardial infarction'. | A retrospective comparison of respiratory events with JAK inhibitors or rituximab for rheumatoid arthritis in patients with pulmonary disease.
Janus kinase inhibitors (JAKi) are an exciting option for the treatment of rheumatoid arthritis (RA) but little is known about their safety and tolerability in patients with existing respiratory disorders. The objective was to compare pulmonary safety of JAKi versus rituximab in patients with concurrent interstitial lung disease (ILD) or bronchiectasis. We performed a retrospective electronic patient record review of patients with known ILD or bronchiectasis commencing JAKi or rituximab for the treatment of RA. Patients initiating treatment from January 2016 to February 2020 were included. Respiratory events (hospitalization or death from a respiratory cause) were compared using Kaplan-Meier survival analysis. We analysed patients who received JAKi (n = 28) and rituximab (n = 19) for a mean (SD) of 1.1 (0.62) and 2.14 (1) years respectively. Patients were predominantly female (68%), anti-CCP antibody positive (94%) and non-smoking (89%) with a median (IQR) percentage predicted FVC at baseline of 100% (82-115%) and percentage predicted TLCO of 62% (54.5-68%). Respiratory events occurred in five patients treated with JAKi (18%; 5 hospitalizations, 2 deaths) and in four patients treated with rituximab (21%; 3 hospitalizations, 1 death). Respiratory event rates did not differ between groups (Cox-regression proportional hazard ratio = 1.38, 95% CI 0.36-5.28; p = 0.64). In this retrospective study, JAKi for the treatment of RA with existing ILD or bronchiectasis did not increase the rate of hospitalization or death due to respiratory causes compared to those treated with rituximab. JAK inhibition may provide a relatively safe option for RA in such patients.
Introduction
Therapeutic decision-making in the treatment of rheumatoid arthritis (RA) for patients with concurrent pulmonary disease can be challenging. Fear regarding increased infection risk in bronchiectasis and progression of interstitial lung disease (ILD) secondary to drug administration restricts treatment options. Controversy exists as to whether medications, such as methotrexate, leflunomide and tumour necrosis factor-α inhibitors (TNFi), exacerbate pulmonary disease in patients with concurrent ILD and bronchiectasis [1]. This has led to hesitancy in utilizing these medications although systematic review of the evidence is beginning to allay fears [2, 3]. Recently, data have begun to support the safe and effective use of B and T cell therapies, rituximab [4] and abatacept [5], in the management of rheumatoid arthritis-related ILD and bronchiectasis [6, 7]. However, therapeutic options remain limited.
The approval of Janus kinase inhibitors (JAKi) (i.e., baricitinib, tofacitinib, upadacitinib) in the management of moderate to severe RA offers a new and exciting treatment paradigm [8–10]. However, inclusion of patients with concurrent ILD or bronchiectasis in relevant clinical trials is extremely limited. While the use of JAKi for the treatment of RA with co-existent lung disease is tempting due to the oral route of administration and relatively short half-life, their safety in relation to pulmonary toxicity in patients with established ILD or bronchiectasis is unknown.
To investigate this evidence gap, the objective of the current study is to compare the pulmonary safety of Janus kinase inhibition for the treatment of rheumatoid arthritis in patients with existing ILD or bronchiectasis, with rituximab. The primary outcome of interest was the incidence of severe respiratory events (hospitalization or death due to respiratory cause) in those receiving JAKi versus those receiving rituximab.
Materials and methods
Study design and patient inclusion
The study comprised a retrospective observational study. The electronic patient health record system (TrakCare®) was used to collect relevant data. Adult patients (aged 18 or over) undergoing treatment for rheumatoid arthritis at the Rheumatic Diseases Unit, Western General Hospital, Edinburgh, U.K., were included in the study. Only patients with an established diagnosis of ILD or bronchiectasis before the commencement of the drug of interest were eligible for inclusion. Patients without these respiratory conditions were excluded. Dedicated pharmacy and biologic clinic records were used to identify all adult patients (16 years or older) commencing any JAKi or rituximab for the treatment of RA from January 2016 to February 2020 (Fig. 1). From these records, all patients’ health records were screened by the investigators (OC, LK) for co-existing diagnoses of interstitial lung disease and/or bronchiectasis. For relevant patients, data were collected in detail from electronic health records by the investigators (OC, OM). Data were collected until discontinuation of the drug of interest (as documented in the clinical record) or to the end of the follow-up period (1 August 2020). They study was conducted in accordance with the Declaration of Helsinki. As all data were collected as part of routine clinical care and all treatment decisions were made prior to conduction of the study, ethical approval was not required.Fig. 1 Study outline, selection of patients and follow-up
Clinical features and demographics
Identical clinical and demographic data were collected for both treatment groups. Standard clinical variables, such as age, gender, disease duration (rheumatoid arthritis and respiratory diseases), smoking history, past and current medications, anti-CCP antibody and rheumatoid factor (IgM) positivity, were collected along with the dates of commencement and cessation for the drugs of interest. Rheumatoid arthritis disease activity (DAS28-ESR or DAS28-CRP) before the commencement of the treatment of interest was recorded along with activity scores at 3–4 months post initiation (where available).
Baseline pulmonary function tests prior to the commencement of the medications of interest were also assessed. As only a small proportion of patients had follow-up pulmonary function assessment within a reasonable time frame, this was not analysed. The pattern of ILD involvement was described as recommended by the American Thoracic Society/European Respiratory Society following review of high-resolution (HR) CT-thorax imaging by a multi-disciplinary team including chest radiologists and an expert ILD lead physician (NH) who additionally attributed a modified semi-quantitative CT-fibrosis score by visual inspection [11, 12].
Outcome measures
The primary outcome of interest was time to first respiratory event (days). Respiratory events were defined as either admission to hospital with a respiratory illness (e.g., infection, ILD exacerbation), or death from a respiratory cause while taking the medication of interest.
Additional outcomes of interest included drug continuation. Drug continuation was defined for JAKi as from the date of first prescription to the date of drug cessation, death, end of follow-up period, or whichever end-point came first. Drug continuation for rituximab was defined as from the date of first infusion to the date of cessation, death, end of follow-up period or whichever end-point came first. The number of cycles, frequency and dose of rituximab received was recorded for each patient.
Statistical analysis
Statistical analysis was performed using IBM SPSS statistics v.25.0 (IBM, Armonk, New York, USA). Distributions of clinical variables and demographics were illustrated using histograms. Depending on the data distribution and type (i.e., categorical, continuous), variables were then described using means (with standard deviations) or medians (with inter-quartile ranges). Continuous variables were compared for differences between groups using either student’s t tests or Mann–Whitney U testing subject to normality or non-normality of distribution. Proportions of categorical variables were compared using Chi-squared testing. Kaplan–Meier survival analysis (log-rank) was used to compare respiratory event survival and drug continuation survival between treatment groups. Cox regression proportional hazard was used to calculate unadjusted hazard ratios with 95% confidence intervals (CI 95%). Statistical significance was set at p < 0.05 and analysis was two -tailed.
Results
Demographic and clinical characteristics
A total of 291 patients were commenced on JAKi to February 2020 at the Rheumatic Diseases Unit, Edinburgh (Fig. 1). Twenty-eight patients commenced JAK inhibition for the treatment of rheumatoid arthritis who also had concurrent interstitial lung disease (67.9%), bronchiectasis (25%) or both (7.1%). Baricitinib was commenced in 26 patients and tofacitinib in 2 patients. A total of 318 patients were commenced on rituximab from January 2016 to February 2020 (Fig. 1). Nineteen patients commenced rituximab for the treatment of RA and had concurrent ILD (68.4%), bronchiectasis (26.3%) or both (5.3%).
Mean duration of follow-up (to discontinuation of drug, death or end of follow-up period) for patients receiving JAKi was 1.1 years (SD = 0.62) and for patients receiving rituximab infusions was 2.14 years (SD = 1). Demographic and clinical characteristics were similar in both treatment groups (Table 1); however, those commencing JAKi had a significantly longer disease duration (median 11 years) compared to those starting rituximab (median 3 years) (Table 1; Mann–Whitney U test, p = 0.003).Table 1 Demographic and clinical details of included patients
JAKi (n = 28) Rituximab (n = 19) No. of patients with available data (n =)
ILD/Bronchiectasis/both
n, (%)
ILD n = 19 (67.9%)
Bronchiectasis n = 7 (25%)
Both n = 2 (7.1%)
ILD n = 13 (68.4%)
Bronchiectasis n = 5 (26.3%)
Both n = 1 (5.3%)
47
Age (median, IQR) 69 (62.3–75) 70 (59–76) 47
Female n, (%) 18 (64.3%) 14 (73.7%) 47
Smoking status
n, (%)
Smoker:
n = 5 (17.9%)
Ex-smoker:
n = 15 (53.6%)
Never-smoker:
n = 8 (28.6%)
Smoker:
n = 0 (0%),
Ex-smoker:
n = 12 (63.2%)
Never-smoker:
n = 7 (36.8%)
47
Anti-CCP positivity
n, (%)
25 (96.2%) 19 (100%) 45
Rh-Factor IgM positivity
n, (%)
14 (77.8%) 9 (90%) 28
RA diagnosis age median, (IQR) 53 (44–61) 54 (48–67) 42
Respiratory diagnosis age
median, (IQR)
64 (55–73) 64 (53–71) 46
Years from RA diagnosis to drug of interest
median, (IQR)
11 (3–20) 3 (1–6) 42
DAS-28 Baseline
median, (IQR)
5.85 (4.98–6.18) 6.2 (5.53–6.86) 43
DAS-28 3–4 months
median, (IQR)
3.4 (2.38–4.11) 4.2 (3.61–5.44) 32
JAKi Janus kinase inhibitor, ILD interstitial lung disease, Rh-factor rheumatoid factor, RA rheumatoid arthritis, DAS-28 disease activity score 28 joints
Pre-treatment pulmonary function tests indicated mild severity of pulmonary disease in patients commencing JAKi and rituximab (Table 2). Review of patients’ baseline high-resolution CT-thorax revealed similar radiological patterns and severity of ILD in both treatment groups (Table 2). However, a greater degree of ground glass was evident on CT-imaging before treatment in those receiving JAKi compared to those receiving rituximab (p = 0.036).Table 2 Baseline pulmonary function (all patients); and radiological pattern and severity of ILD on CT-thorax imaging (patients with ILD)
JAKi (n = 28) Rituximab (n = 19) No. of patients with available data (n =)
Baseline predicted FEV1 median %, (IQR) 93.5 (74.5–105) 92 (76–105) 43
Baseline predicted FVC median %, (IQR) 103 (78–112) 99 (86–119) 44
Baseline predicted TLCO median %, (IQR) 62 (51.25–69) 62 (55–67) 41
Radiological ILD-pattern JAKi (n = 20) Rituximab (n = 14)
UIP
n, (%)
0 (0%) 2 (14%)
Probable UIP
n, (%)
4 (20%) 5 (36%)
Fibrotic NSIP
n, (%)
6 (30%) 1 (7%)
Chronic HP
n, (%)
3 (15%) 1 (7%)
Respiratory bronchiolitis-ILD
n, (%)
1 (5%) 0 (0%)
Unclassifiable
n, (%)
6 (30%) 5 (36%)
% of lung with ILD
median, (IQR)
20 (10–45) 12.5 (10–30)
% of lung with ground glass appearance
median, (IQR)
20 (5–45) 7.5 (5–10)
% of lung with honeycombing
median, (IQR)
0 (0–1.75) 0 (0–2)
% of lung with emphysema
median, (IQR)
0.5 (0–8.75) 0 (0–3.75)
JAKi Janus kinase inhibitor, ILD interstitial lung disease, FEV1 forced expiratory volume in 1 second, FVC forced vital capacity, TLco lung transfer capacity for carbon monoxide, IQR inter-quartile range, UIP usual interstitial pneumonia, NSIP non-specific interstitial pneumonia, HP hypersensitivity pneumonitis
Treatment details
Concurrent treatments including conventional synthetic DMARDs and corticosteroids were similar in both groups (Table 3). Almost 30% of patients were biologic naïve before commencing the drug of interest. Prior treatment history indicated that those commencing JAKi had previously been treated with a greater number of biologics including TNFi, tocilizumab and abatacept, compared to those commencing rituximab.Table 3 Previous and concurrent DMARD treatments
JAKi (n = 28) Rituximab (n = 19)
Previous treatments
No. of previous csDMARDs
median, (IQR)
3 (2–4) 3 (2–4)
Methotrexate
n, (%)
22 (78.6%) 13 (68.4%)
Biologic Naïve
n, (%)
8 (28.6%) 6 (31.6%)
No. of previous biologics
median, (IQR)
1 (0–2) 0 (0–1)
Rituximab/JAKi
n, (%)
12 (42.9%) 1 (5.3%)
TNF-α inhibitor
n, (%)
15 (53.6%) 5 (26.3%)
Tocilizumab
n, (%)
6 (21.4%) 1 (5.3%)
Abatacept
n, (%)
6 (21.4%) 1 (5.3%)
Concurrent treatments
Prednisolone; n, (%) 13 (46.4%) 6 (31.6%)
Prednisolone dose (mg) median, (IQR) 7 (5–10) 10 (4.5–12.5)
Methotrexate
n, (%)
4 (14.3%) 2 (10.5%)
Azathioprine
n, (%)
2 (7.1%) 2 (10.5%)
Leflunomide
n, (%)
2 (7.1%) 2 (10.5%)
Hydroxychloroquine
n, (%)
6 (21.4%) 6 (31.6%)
Sulfasalazine
n, (%)
3 (10.7%) 4 (21.1%)
DMARD disease-modifying anti-rheumatic drugs, csDMARDs conventional synthetic DMARDs, JAKi Janus kinase inhibitor, IQR inter-quartile range, TNF Tumour necrosis factor
All except for two patients received standard licensed dosing regimens for JAK inhibition for the treatment of RA (i.e. baricitinib 4 mg once daily; tofacitinib 5 mg twice daily). One patient took 4 mg/2 mg of baricitinib on alternate days and one patient took 5 mg of tofacitinib once daily before increasing to 5 mg twice daily.
All patients who received rituximab received the standard dosing of treatment (2 × intravenous infusions of 1000 mg rituximab with 100 mg intravenous methylprednisolone 14 days apart). Only 2/19 patients were on ‘fixed’ cycles of rituximab therapy repeated every 6 months. All others received rituximab as required depending on clinical status but not within 6 months of the last infusion. 5/19 patients received only 1 cycle of rituximab during their study follow-up period. All others (74%) received between 2 and 5 cycles of rituximab.
Incidence of respiratory events
Respiratory events (i.e. hospitalization or death due to a respiratory cause) occurred in five patients (18%) during JAKi treatment (7 hospitalizations, two of which led to death) and in four patients (21%) (4 hospitalizations, one of which led to death) during treatment with rituximab. In the JAKi group, two patients (7.1%) died of respiratory issues (one new lung cancer with concomitant community acquired pneumonia and one of community acquired pneumonia). In the rituximab-treated patients, two patients (10.5%) died during follow-up; one of an acute myocardial infarction and one of pneumonitis and respiratory failure. No episodes of acute pneumonitis occurred in the JAKi group. There was no difference in respiratory event survival in the JAKi-treated group compared with the rituximab-treated group (Fig. 2a); unadjusted hazard ratio: 1.38 (95% CI 0.36, 5.28); p = 0.64.Fig. 2 a Comparison of time to respiratory event and b drug continuation between JAKi and rituximab for the treatment of rheumatoid arthritis in patients with concurrent bronchiectasis and/or ILD
Drug continuation
Seventy-one percent (20/28) of patients with ILD and/or bronchiectasis who commenced JAKi for the treatment of rheumatoid arthritis remained on the drug at the end of the follow-up period. Reasons for discontinuation included inefficacy (n = 3), death (n = 2, as outlined above), recurrent infections (n = 2) and reduced renal function (n = 1).
Sixty-three percent (12/19) of patients who commenced rituximab continued the drug to the end of the follow-up period. Reasons for cessation included inefficacy (n = 5) or death (n = 2, as outlined above). The rate of discontinuation of JAKi in patients with ILD and/or bronchiectasis did not significantly differ from that of those receiving rituximab (Fig. 2b); unadjusted hazard ratio: 1.9 (95% CI 0.634, 5.73); p = 0.251.
Discussion
There is an absence of patients with interstitial lung disease and bronchiectasis in randomised controlled trials that examine the effectiveness of JAK inhibitors for the treatment of rheumatoid arthritis. Therefore, we have limited knowledge of the pulmonary safety in these cohorts. Prevalence of interstitial lung disease and bronchiectasis is relatively high in patients with rheumatoid arthritis, estimated to affect between 5 and 10% of patients [1, 13]. Interstitial lung disease itself contributes significantly to the excess mortality evident in patients with rheumatoid arthritis [14]. Despite this, the optimal treatment strategy for such patients remains unknown, reflecting a paucity of randomized controlled clinical trials for this patient sub-group. Recently, observational data have suggested treatment with rituximab [6, 15, 16], and laterally abatacept [5, 17] is the most effective and safest biologic agents for the treatment of RA with concurrent ILD or bronchiectasis. Data suggest that treatment with rituximab and abatacept may even infer a benefit to pulmonary function [5, 6]. Nonetheless, concern about lung toxicity restricts the use of many DMARDs and biologic agents, be these concerns soundly proven or not.
Novel, small-molecule treatments, such as Janus kinase inhibitors, represent an exciting option for patients with rheumatoid arthritis. Their safety and potential for effectiveness in patients with concurrent pulmonary disease should be duly considered. Currently, our understanding of inhibition of JAK/STAT signalling pathways in patients with ILD and bronchiectasis is extremely limited. JAKi-use has led to successful treatment of dermatomyositis-associated ILD but evidence is restricted to a small number of case reports and case series [18–20]. Even less is understood about Janus kinase inhibition in patients with bronchiectasis. While post hoc analysis of phase 2, 3 and 3b/4 clinical trials indicates that the de novo incidence of interstitial lung disease is low in patients with rheumatoid arthritis treated with JAKi [21]; to our knowledge, no study has yet examined the pulmonary tolerance and safety of JAKi in those with existing ILD or bronchiectasis. The current study sought to address this important clinical scenario, by comparing time to respiratory events in patients receiving JAKi with patients receiving the current, best-practice medication for this clinical cohort, rituximab.
The majority of our patients commenced JAK1/2 inhibition with baricitinib due to the earlier licensing of this medication for the treatment of rheumatoid arthritis but several patients commenced JAK1/3 inhibition with tofacitinib. Furthermore, the proportion of patients with ‘difficult-to-treat’ rheumatoid disease, manifest by a longer duration of illness and a greater number of previous biologic therapies (e.g. tocilizumab, abatacept), was higher in those subsequently prescribed JAKi. We believe this relates to the fact that a novel medication provides an attractive option for patients who have had difficult-to-treat disease for many years and in such cases patients often chose to switch therapy in an attempt to achieve remission.
Encouragingly, serious respiratory events (i.e., hospitalization or death due to a respiratory cause) were no different between those treated with JAKi and those treated with rituximab suggesting that JAKi may offer a realistic treatment option for patients with RA and concurrent ILD or bronchiectasis. Our cohort of patients represents a group with mild to moderate pulmonary disease, all of whom commenced Janus kinase inhibition or rituximab with a known, pre-existing diagnosis of ILD or bronchiectasis (present on cross sectional CT-imaging and/or attending a tertiary ILD clinic). Very few patients with ILD had a HR-CT pattern consistent with definite UIP and most had a pattern that was consistent with an alternative diagnosis and were highly unlikely to be UIP. Therefore, our findings should not be extrapolated to those with severe ILD or bronchiectasis or a definite UIP pattern of disease and caution is urged in this setting. Furthermore, there was no significant difference in time to drug discontinuation between those prescribed JAKi and those prescribed rituximab. Proportionally, more patients treated with rituximab (almost 40%) ceased treatment during follow-up compared with those treated with JAKi (almost 30%). Reasons for discontinuation of rituximab predominantly related to clinical inefficacy, but comparison of clinical effectiveness was not the aim of this study.
There are several strengths to this study. Both cohorts of patients (i.e., JAKi- and rituximab-receiving) were similar in terms of clinical traits (e.g. gender, concurrent medications, corticosteroid use, anti-CCP and rheumatoid factor seropositivity) minimizing the effect of confounders on respiratory event outcomes. In addition, these data represent real-world experience of patients with RA and concurrent ILD/bronchiectasis with patient–clinician decisions made without influence of a pre-determined research goal. The retrospective nature of this observational study and associated risk of information bias represents the major limitation of this study. The relatively small sample size and short follow-up period also represent limitations that may have affected overall results. Ideally, follow-up pulmonary function testing would have been available for all patients after commencing treatment but as this was only the case for a minority of patients, this was not analysed. It was not possible to collect primary care records of respiratory exacerbations or infections that did not require hospital admission that were treated in the community. Furthermore, the electronic patient health record system used did not permit recording of admissions to hospital abroad or to hospitals outside of NHS Lothian. The authors recommend that the findings of this study and analysis should be investigated further in larger prospective cohorts in the future.
To conclude, within the limitations of this retrospective analysis, JAK inhibition with baricitinib or tofacitinib for the treatment of rheumatoid arthritis in patients with concurrent interstitial lung disease or bronchiectasis did not increase the rate of hospitalization or death due to a respiratory cause in comparison to those treated with rituximab. JAK inhibition may provide a favourable and safe therapeutic option for such patients but larger, prospective studies are required to confirm these findings to inform much-needed therapeutic guidance in this sub-category of patients with RA and ILD/bronchiectasis.
Funding
No specific funding was received from any bodies in the public, commercial or not-for-profit sectors to carry out the work described in this article.
Declarations
Conflict of interest
OC has received sponsorship from Lilly and Pfizer to attend educational conferences. The other authors have no conflicts of interest to declare.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. | METHYLPREDNISOLONE, RITUXIMAB | DrugsGivenReaction | CC BY | 33721094 | 19,059,238 | 2021-05 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Death'. | A retrospective comparison of respiratory events with JAK inhibitors or rituximab for rheumatoid arthritis in patients with pulmonary disease.
Janus kinase inhibitors (JAKi) are an exciting option for the treatment of rheumatoid arthritis (RA) but little is known about their safety and tolerability in patients with existing respiratory disorders. The objective was to compare pulmonary safety of JAKi versus rituximab in patients with concurrent interstitial lung disease (ILD) or bronchiectasis. We performed a retrospective electronic patient record review of patients with known ILD or bronchiectasis commencing JAKi or rituximab for the treatment of RA. Patients initiating treatment from January 2016 to February 2020 were included. Respiratory events (hospitalization or death from a respiratory cause) were compared using Kaplan-Meier survival analysis. We analysed patients who received JAKi (n = 28) and rituximab (n = 19) for a mean (SD) of 1.1 (0.62) and 2.14 (1) years respectively. Patients were predominantly female (68%), anti-CCP antibody positive (94%) and non-smoking (89%) with a median (IQR) percentage predicted FVC at baseline of 100% (82-115%) and percentage predicted TLCO of 62% (54.5-68%). Respiratory events occurred in five patients treated with JAKi (18%; 5 hospitalizations, 2 deaths) and in four patients treated with rituximab (21%; 3 hospitalizations, 1 death). Respiratory event rates did not differ between groups (Cox-regression proportional hazard ratio = 1.38, 95% CI 0.36-5.28; p = 0.64). In this retrospective study, JAKi for the treatment of RA with existing ILD or bronchiectasis did not increase the rate of hospitalization or death due to respiratory causes compared to those treated with rituximab. JAK inhibition may provide a relatively safe option for RA in such patients.
Introduction
Therapeutic decision-making in the treatment of rheumatoid arthritis (RA) for patients with concurrent pulmonary disease can be challenging. Fear regarding increased infection risk in bronchiectasis and progression of interstitial lung disease (ILD) secondary to drug administration restricts treatment options. Controversy exists as to whether medications, such as methotrexate, leflunomide and tumour necrosis factor-α inhibitors (TNFi), exacerbate pulmonary disease in patients with concurrent ILD and bronchiectasis [1]. This has led to hesitancy in utilizing these medications although systematic review of the evidence is beginning to allay fears [2, 3]. Recently, data have begun to support the safe and effective use of B and T cell therapies, rituximab [4] and abatacept [5], in the management of rheumatoid arthritis-related ILD and bronchiectasis [6, 7]. However, therapeutic options remain limited.
The approval of Janus kinase inhibitors (JAKi) (i.e., baricitinib, tofacitinib, upadacitinib) in the management of moderate to severe RA offers a new and exciting treatment paradigm [8–10]. However, inclusion of patients with concurrent ILD or bronchiectasis in relevant clinical trials is extremely limited. While the use of JAKi for the treatment of RA with co-existent lung disease is tempting due to the oral route of administration and relatively short half-life, their safety in relation to pulmonary toxicity in patients with established ILD or bronchiectasis is unknown.
To investigate this evidence gap, the objective of the current study is to compare the pulmonary safety of Janus kinase inhibition for the treatment of rheumatoid arthritis in patients with existing ILD or bronchiectasis, with rituximab. The primary outcome of interest was the incidence of severe respiratory events (hospitalization or death due to respiratory cause) in those receiving JAKi versus those receiving rituximab.
Materials and methods
Study design and patient inclusion
The study comprised a retrospective observational study. The electronic patient health record system (TrakCare®) was used to collect relevant data. Adult patients (aged 18 or over) undergoing treatment for rheumatoid arthritis at the Rheumatic Diseases Unit, Western General Hospital, Edinburgh, U.K., were included in the study. Only patients with an established diagnosis of ILD or bronchiectasis before the commencement of the drug of interest were eligible for inclusion. Patients without these respiratory conditions were excluded. Dedicated pharmacy and biologic clinic records were used to identify all adult patients (16 years or older) commencing any JAKi or rituximab for the treatment of RA from January 2016 to February 2020 (Fig. 1). From these records, all patients’ health records were screened by the investigators (OC, LK) for co-existing diagnoses of interstitial lung disease and/or bronchiectasis. For relevant patients, data were collected in detail from electronic health records by the investigators (OC, OM). Data were collected until discontinuation of the drug of interest (as documented in the clinical record) or to the end of the follow-up period (1 August 2020). They study was conducted in accordance with the Declaration of Helsinki. As all data were collected as part of routine clinical care and all treatment decisions were made prior to conduction of the study, ethical approval was not required.Fig. 1 Study outline, selection of patients and follow-up
Clinical features and demographics
Identical clinical and demographic data were collected for both treatment groups. Standard clinical variables, such as age, gender, disease duration (rheumatoid arthritis and respiratory diseases), smoking history, past and current medications, anti-CCP antibody and rheumatoid factor (IgM) positivity, were collected along with the dates of commencement and cessation for the drugs of interest. Rheumatoid arthritis disease activity (DAS28-ESR or DAS28-CRP) before the commencement of the treatment of interest was recorded along with activity scores at 3–4 months post initiation (where available).
Baseline pulmonary function tests prior to the commencement of the medications of interest were also assessed. As only a small proportion of patients had follow-up pulmonary function assessment within a reasonable time frame, this was not analysed. The pattern of ILD involvement was described as recommended by the American Thoracic Society/European Respiratory Society following review of high-resolution (HR) CT-thorax imaging by a multi-disciplinary team including chest radiologists and an expert ILD lead physician (NH) who additionally attributed a modified semi-quantitative CT-fibrosis score by visual inspection [11, 12].
Outcome measures
The primary outcome of interest was time to first respiratory event (days). Respiratory events were defined as either admission to hospital with a respiratory illness (e.g., infection, ILD exacerbation), or death from a respiratory cause while taking the medication of interest.
Additional outcomes of interest included drug continuation. Drug continuation was defined for JAKi as from the date of first prescription to the date of drug cessation, death, end of follow-up period, or whichever end-point came first. Drug continuation for rituximab was defined as from the date of first infusion to the date of cessation, death, end of follow-up period or whichever end-point came first. The number of cycles, frequency and dose of rituximab received was recorded for each patient.
Statistical analysis
Statistical analysis was performed using IBM SPSS statistics v.25.0 (IBM, Armonk, New York, USA). Distributions of clinical variables and demographics were illustrated using histograms. Depending on the data distribution and type (i.e., categorical, continuous), variables were then described using means (with standard deviations) or medians (with inter-quartile ranges). Continuous variables were compared for differences between groups using either student’s t tests or Mann–Whitney U testing subject to normality or non-normality of distribution. Proportions of categorical variables were compared using Chi-squared testing. Kaplan–Meier survival analysis (log-rank) was used to compare respiratory event survival and drug continuation survival between treatment groups. Cox regression proportional hazard was used to calculate unadjusted hazard ratios with 95% confidence intervals (CI 95%). Statistical significance was set at p < 0.05 and analysis was two -tailed.
Results
Demographic and clinical characteristics
A total of 291 patients were commenced on JAKi to February 2020 at the Rheumatic Diseases Unit, Edinburgh (Fig. 1). Twenty-eight patients commenced JAK inhibition for the treatment of rheumatoid arthritis who also had concurrent interstitial lung disease (67.9%), bronchiectasis (25%) or both (7.1%). Baricitinib was commenced in 26 patients and tofacitinib in 2 patients. A total of 318 patients were commenced on rituximab from January 2016 to February 2020 (Fig. 1). Nineteen patients commenced rituximab for the treatment of RA and had concurrent ILD (68.4%), bronchiectasis (26.3%) or both (5.3%).
Mean duration of follow-up (to discontinuation of drug, death or end of follow-up period) for patients receiving JAKi was 1.1 years (SD = 0.62) and for patients receiving rituximab infusions was 2.14 years (SD = 1). Demographic and clinical characteristics were similar in both treatment groups (Table 1); however, those commencing JAKi had a significantly longer disease duration (median 11 years) compared to those starting rituximab (median 3 years) (Table 1; Mann–Whitney U test, p = 0.003).Table 1 Demographic and clinical details of included patients
JAKi (n = 28) Rituximab (n = 19) No. of patients with available data (n =)
ILD/Bronchiectasis/both
n, (%)
ILD n = 19 (67.9%)
Bronchiectasis n = 7 (25%)
Both n = 2 (7.1%)
ILD n = 13 (68.4%)
Bronchiectasis n = 5 (26.3%)
Both n = 1 (5.3%)
47
Age (median, IQR) 69 (62.3–75) 70 (59–76) 47
Female n, (%) 18 (64.3%) 14 (73.7%) 47
Smoking status
n, (%)
Smoker:
n = 5 (17.9%)
Ex-smoker:
n = 15 (53.6%)
Never-smoker:
n = 8 (28.6%)
Smoker:
n = 0 (0%),
Ex-smoker:
n = 12 (63.2%)
Never-smoker:
n = 7 (36.8%)
47
Anti-CCP positivity
n, (%)
25 (96.2%) 19 (100%) 45
Rh-Factor IgM positivity
n, (%)
14 (77.8%) 9 (90%) 28
RA diagnosis age median, (IQR) 53 (44–61) 54 (48–67) 42
Respiratory diagnosis age
median, (IQR)
64 (55–73) 64 (53–71) 46
Years from RA diagnosis to drug of interest
median, (IQR)
11 (3–20) 3 (1–6) 42
DAS-28 Baseline
median, (IQR)
5.85 (4.98–6.18) 6.2 (5.53–6.86) 43
DAS-28 3–4 months
median, (IQR)
3.4 (2.38–4.11) 4.2 (3.61–5.44) 32
JAKi Janus kinase inhibitor, ILD interstitial lung disease, Rh-factor rheumatoid factor, RA rheumatoid arthritis, DAS-28 disease activity score 28 joints
Pre-treatment pulmonary function tests indicated mild severity of pulmonary disease in patients commencing JAKi and rituximab (Table 2). Review of patients’ baseline high-resolution CT-thorax revealed similar radiological patterns and severity of ILD in both treatment groups (Table 2). However, a greater degree of ground glass was evident on CT-imaging before treatment in those receiving JAKi compared to those receiving rituximab (p = 0.036).Table 2 Baseline pulmonary function (all patients); and radiological pattern and severity of ILD on CT-thorax imaging (patients with ILD)
JAKi (n = 28) Rituximab (n = 19) No. of patients with available data (n =)
Baseline predicted FEV1 median %, (IQR) 93.5 (74.5–105) 92 (76–105) 43
Baseline predicted FVC median %, (IQR) 103 (78–112) 99 (86–119) 44
Baseline predicted TLCO median %, (IQR) 62 (51.25–69) 62 (55–67) 41
Radiological ILD-pattern JAKi (n = 20) Rituximab (n = 14)
UIP
n, (%)
0 (0%) 2 (14%)
Probable UIP
n, (%)
4 (20%) 5 (36%)
Fibrotic NSIP
n, (%)
6 (30%) 1 (7%)
Chronic HP
n, (%)
3 (15%) 1 (7%)
Respiratory bronchiolitis-ILD
n, (%)
1 (5%) 0 (0%)
Unclassifiable
n, (%)
6 (30%) 5 (36%)
% of lung with ILD
median, (IQR)
20 (10–45) 12.5 (10–30)
% of lung with ground glass appearance
median, (IQR)
20 (5–45) 7.5 (5–10)
% of lung with honeycombing
median, (IQR)
0 (0–1.75) 0 (0–2)
% of lung with emphysema
median, (IQR)
0.5 (0–8.75) 0 (0–3.75)
JAKi Janus kinase inhibitor, ILD interstitial lung disease, FEV1 forced expiratory volume in 1 second, FVC forced vital capacity, TLco lung transfer capacity for carbon monoxide, IQR inter-quartile range, UIP usual interstitial pneumonia, NSIP non-specific interstitial pneumonia, HP hypersensitivity pneumonitis
Treatment details
Concurrent treatments including conventional synthetic DMARDs and corticosteroids were similar in both groups (Table 3). Almost 30% of patients were biologic naïve before commencing the drug of interest. Prior treatment history indicated that those commencing JAKi had previously been treated with a greater number of biologics including TNFi, tocilizumab and abatacept, compared to those commencing rituximab.Table 3 Previous and concurrent DMARD treatments
JAKi (n = 28) Rituximab (n = 19)
Previous treatments
No. of previous csDMARDs
median, (IQR)
3 (2–4) 3 (2–4)
Methotrexate
n, (%)
22 (78.6%) 13 (68.4%)
Biologic Naïve
n, (%)
8 (28.6%) 6 (31.6%)
No. of previous biologics
median, (IQR)
1 (0–2) 0 (0–1)
Rituximab/JAKi
n, (%)
12 (42.9%) 1 (5.3%)
TNF-α inhibitor
n, (%)
15 (53.6%) 5 (26.3%)
Tocilizumab
n, (%)
6 (21.4%) 1 (5.3%)
Abatacept
n, (%)
6 (21.4%) 1 (5.3%)
Concurrent treatments
Prednisolone; n, (%) 13 (46.4%) 6 (31.6%)
Prednisolone dose (mg) median, (IQR) 7 (5–10) 10 (4.5–12.5)
Methotrexate
n, (%)
4 (14.3%) 2 (10.5%)
Azathioprine
n, (%)
2 (7.1%) 2 (10.5%)
Leflunomide
n, (%)
2 (7.1%) 2 (10.5%)
Hydroxychloroquine
n, (%)
6 (21.4%) 6 (31.6%)
Sulfasalazine
n, (%)
3 (10.7%) 4 (21.1%)
DMARD disease-modifying anti-rheumatic drugs, csDMARDs conventional synthetic DMARDs, JAKi Janus kinase inhibitor, IQR inter-quartile range, TNF Tumour necrosis factor
All except for two patients received standard licensed dosing regimens for JAK inhibition for the treatment of RA (i.e. baricitinib 4 mg once daily; tofacitinib 5 mg twice daily). One patient took 4 mg/2 mg of baricitinib on alternate days and one patient took 5 mg of tofacitinib once daily before increasing to 5 mg twice daily.
All patients who received rituximab received the standard dosing of treatment (2 × intravenous infusions of 1000 mg rituximab with 100 mg intravenous methylprednisolone 14 days apart). Only 2/19 patients were on ‘fixed’ cycles of rituximab therapy repeated every 6 months. All others received rituximab as required depending on clinical status but not within 6 months of the last infusion. 5/19 patients received only 1 cycle of rituximab during their study follow-up period. All others (74%) received between 2 and 5 cycles of rituximab.
Incidence of respiratory events
Respiratory events (i.e. hospitalization or death due to a respiratory cause) occurred in five patients (18%) during JAKi treatment (7 hospitalizations, two of which led to death) and in four patients (21%) (4 hospitalizations, one of which led to death) during treatment with rituximab. In the JAKi group, two patients (7.1%) died of respiratory issues (one new lung cancer with concomitant community acquired pneumonia and one of community acquired pneumonia). In the rituximab-treated patients, two patients (10.5%) died during follow-up; one of an acute myocardial infarction and one of pneumonitis and respiratory failure. No episodes of acute pneumonitis occurred in the JAKi group. There was no difference in respiratory event survival in the JAKi-treated group compared with the rituximab-treated group (Fig. 2a); unadjusted hazard ratio: 1.38 (95% CI 0.36, 5.28); p = 0.64.Fig. 2 a Comparison of time to respiratory event and b drug continuation between JAKi and rituximab for the treatment of rheumatoid arthritis in patients with concurrent bronchiectasis and/or ILD
Drug continuation
Seventy-one percent (20/28) of patients with ILD and/or bronchiectasis who commenced JAKi for the treatment of rheumatoid arthritis remained on the drug at the end of the follow-up period. Reasons for discontinuation included inefficacy (n = 3), death (n = 2, as outlined above), recurrent infections (n = 2) and reduced renal function (n = 1).
Sixty-three percent (12/19) of patients who commenced rituximab continued the drug to the end of the follow-up period. Reasons for cessation included inefficacy (n = 5) or death (n = 2, as outlined above). The rate of discontinuation of JAKi in patients with ILD and/or bronchiectasis did not significantly differ from that of those receiving rituximab (Fig. 2b); unadjusted hazard ratio: 1.9 (95% CI 0.634, 5.73); p = 0.251.
Discussion
There is an absence of patients with interstitial lung disease and bronchiectasis in randomised controlled trials that examine the effectiveness of JAK inhibitors for the treatment of rheumatoid arthritis. Therefore, we have limited knowledge of the pulmonary safety in these cohorts. Prevalence of interstitial lung disease and bronchiectasis is relatively high in patients with rheumatoid arthritis, estimated to affect between 5 and 10% of patients [1, 13]. Interstitial lung disease itself contributes significantly to the excess mortality evident in patients with rheumatoid arthritis [14]. Despite this, the optimal treatment strategy for such patients remains unknown, reflecting a paucity of randomized controlled clinical trials for this patient sub-group. Recently, observational data have suggested treatment with rituximab [6, 15, 16], and laterally abatacept [5, 17] is the most effective and safest biologic agents for the treatment of RA with concurrent ILD or bronchiectasis. Data suggest that treatment with rituximab and abatacept may even infer a benefit to pulmonary function [5, 6]. Nonetheless, concern about lung toxicity restricts the use of many DMARDs and biologic agents, be these concerns soundly proven or not.
Novel, small-molecule treatments, such as Janus kinase inhibitors, represent an exciting option for patients with rheumatoid arthritis. Their safety and potential for effectiveness in patients with concurrent pulmonary disease should be duly considered. Currently, our understanding of inhibition of JAK/STAT signalling pathways in patients with ILD and bronchiectasis is extremely limited. JAKi-use has led to successful treatment of dermatomyositis-associated ILD but evidence is restricted to a small number of case reports and case series [18–20]. Even less is understood about Janus kinase inhibition in patients with bronchiectasis. While post hoc analysis of phase 2, 3 and 3b/4 clinical trials indicates that the de novo incidence of interstitial lung disease is low in patients with rheumatoid arthritis treated with JAKi [21]; to our knowledge, no study has yet examined the pulmonary tolerance and safety of JAKi in those with existing ILD or bronchiectasis. The current study sought to address this important clinical scenario, by comparing time to respiratory events in patients receiving JAKi with patients receiving the current, best-practice medication for this clinical cohort, rituximab.
The majority of our patients commenced JAK1/2 inhibition with baricitinib due to the earlier licensing of this medication for the treatment of rheumatoid arthritis but several patients commenced JAK1/3 inhibition with tofacitinib. Furthermore, the proportion of patients with ‘difficult-to-treat’ rheumatoid disease, manifest by a longer duration of illness and a greater number of previous biologic therapies (e.g. tocilizumab, abatacept), was higher in those subsequently prescribed JAKi. We believe this relates to the fact that a novel medication provides an attractive option for patients who have had difficult-to-treat disease for many years and in such cases patients often chose to switch therapy in an attempt to achieve remission.
Encouragingly, serious respiratory events (i.e., hospitalization or death due to a respiratory cause) were no different between those treated with JAKi and those treated with rituximab suggesting that JAKi may offer a realistic treatment option for patients with RA and concurrent ILD or bronchiectasis. Our cohort of patients represents a group with mild to moderate pulmonary disease, all of whom commenced Janus kinase inhibition or rituximab with a known, pre-existing diagnosis of ILD or bronchiectasis (present on cross sectional CT-imaging and/or attending a tertiary ILD clinic). Very few patients with ILD had a HR-CT pattern consistent with definite UIP and most had a pattern that was consistent with an alternative diagnosis and were highly unlikely to be UIP. Therefore, our findings should not be extrapolated to those with severe ILD or bronchiectasis or a definite UIP pattern of disease and caution is urged in this setting. Furthermore, there was no significant difference in time to drug discontinuation between those prescribed JAKi and those prescribed rituximab. Proportionally, more patients treated with rituximab (almost 40%) ceased treatment during follow-up compared with those treated with JAKi (almost 30%). Reasons for discontinuation of rituximab predominantly related to clinical inefficacy, but comparison of clinical effectiveness was not the aim of this study.
There are several strengths to this study. Both cohorts of patients (i.e., JAKi- and rituximab-receiving) were similar in terms of clinical traits (e.g. gender, concurrent medications, corticosteroid use, anti-CCP and rheumatoid factor seropositivity) minimizing the effect of confounders on respiratory event outcomes. In addition, these data represent real-world experience of patients with RA and concurrent ILD/bronchiectasis with patient–clinician decisions made without influence of a pre-determined research goal. The retrospective nature of this observational study and associated risk of information bias represents the major limitation of this study. The relatively small sample size and short follow-up period also represent limitations that may have affected overall results. Ideally, follow-up pulmonary function testing would have been available for all patients after commencing treatment but as this was only the case for a minority of patients, this was not analysed. It was not possible to collect primary care records of respiratory exacerbations or infections that did not require hospital admission that were treated in the community. Furthermore, the electronic patient health record system used did not permit recording of admissions to hospital abroad or to hospitals outside of NHS Lothian. The authors recommend that the findings of this study and analysis should be investigated further in larger prospective cohorts in the future.
To conclude, within the limitations of this retrospective analysis, JAK inhibition with baricitinib or tofacitinib for the treatment of rheumatoid arthritis in patients with concurrent interstitial lung disease or bronchiectasis did not increase the rate of hospitalization or death due to a respiratory cause in comparison to those treated with rituximab. JAK inhibition may provide a favourable and safe therapeutic option for such patients but larger, prospective studies are required to confirm these findings to inform much-needed therapeutic guidance in this sub-category of patients with RA and ILD/bronchiectasis.
Funding
No specific funding was received from any bodies in the public, commercial or not-for-profit sectors to carry out the work described in this article.
Declarations
Conflict of interest
OC has received sponsorship from Lilly and Pfizer to attend educational conferences. The other authors have no conflicts of interest to declare.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. | METHYLPREDNISOLONE SODIUM SUCCINATE, RITUXIMAB | DrugsGivenReaction | CC BY | 33721094 | 19,082,833 | 2021-05 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Pneumonitis'. | A retrospective comparison of respiratory events with JAK inhibitors or rituximab for rheumatoid arthritis in patients with pulmonary disease.
Janus kinase inhibitors (JAKi) are an exciting option for the treatment of rheumatoid arthritis (RA) but little is known about their safety and tolerability in patients with existing respiratory disorders. The objective was to compare pulmonary safety of JAKi versus rituximab in patients with concurrent interstitial lung disease (ILD) or bronchiectasis. We performed a retrospective electronic patient record review of patients with known ILD or bronchiectasis commencing JAKi or rituximab for the treatment of RA. Patients initiating treatment from January 2016 to February 2020 were included. Respiratory events (hospitalization or death from a respiratory cause) were compared using Kaplan-Meier survival analysis. We analysed patients who received JAKi (n = 28) and rituximab (n = 19) for a mean (SD) of 1.1 (0.62) and 2.14 (1) years respectively. Patients were predominantly female (68%), anti-CCP antibody positive (94%) and non-smoking (89%) with a median (IQR) percentage predicted FVC at baseline of 100% (82-115%) and percentage predicted TLCO of 62% (54.5-68%). Respiratory events occurred in five patients treated with JAKi (18%; 5 hospitalizations, 2 deaths) and in four patients treated with rituximab (21%; 3 hospitalizations, 1 death). Respiratory event rates did not differ between groups (Cox-regression proportional hazard ratio = 1.38, 95% CI 0.36-5.28; p = 0.64). In this retrospective study, JAKi for the treatment of RA with existing ILD or bronchiectasis did not increase the rate of hospitalization or death due to respiratory causes compared to those treated with rituximab. JAK inhibition may provide a relatively safe option for RA in such patients.
Introduction
Therapeutic decision-making in the treatment of rheumatoid arthritis (RA) for patients with concurrent pulmonary disease can be challenging. Fear regarding increased infection risk in bronchiectasis and progression of interstitial lung disease (ILD) secondary to drug administration restricts treatment options. Controversy exists as to whether medications, such as methotrexate, leflunomide and tumour necrosis factor-α inhibitors (TNFi), exacerbate pulmonary disease in patients with concurrent ILD and bronchiectasis [1]. This has led to hesitancy in utilizing these medications although systematic review of the evidence is beginning to allay fears [2, 3]. Recently, data have begun to support the safe and effective use of B and T cell therapies, rituximab [4] and abatacept [5], in the management of rheumatoid arthritis-related ILD and bronchiectasis [6, 7]. However, therapeutic options remain limited.
The approval of Janus kinase inhibitors (JAKi) (i.e., baricitinib, tofacitinib, upadacitinib) in the management of moderate to severe RA offers a new and exciting treatment paradigm [8–10]. However, inclusion of patients with concurrent ILD or bronchiectasis in relevant clinical trials is extremely limited. While the use of JAKi for the treatment of RA with co-existent lung disease is tempting due to the oral route of administration and relatively short half-life, their safety in relation to pulmonary toxicity in patients with established ILD or bronchiectasis is unknown.
To investigate this evidence gap, the objective of the current study is to compare the pulmonary safety of Janus kinase inhibition for the treatment of rheumatoid arthritis in patients with existing ILD or bronchiectasis, with rituximab. The primary outcome of interest was the incidence of severe respiratory events (hospitalization or death due to respiratory cause) in those receiving JAKi versus those receiving rituximab.
Materials and methods
Study design and patient inclusion
The study comprised a retrospective observational study. The electronic patient health record system (TrakCare®) was used to collect relevant data. Adult patients (aged 18 or over) undergoing treatment for rheumatoid arthritis at the Rheumatic Diseases Unit, Western General Hospital, Edinburgh, U.K., were included in the study. Only patients with an established diagnosis of ILD or bronchiectasis before the commencement of the drug of interest were eligible for inclusion. Patients without these respiratory conditions were excluded. Dedicated pharmacy and biologic clinic records were used to identify all adult patients (16 years or older) commencing any JAKi or rituximab for the treatment of RA from January 2016 to February 2020 (Fig. 1). From these records, all patients’ health records were screened by the investigators (OC, LK) for co-existing diagnoses of interstitial lung disease and/or bronchiectasis. For relevant patients, data were collected in detail from electronic health records by the investigators (OC, OM). Data were collected until discontinuation of the drug of interest (as documented in the clinical record) or to the end of the follow-up period (1 August 2020). They study was conducted in accordance with the Declaration of Helsinki. As all data were collected as part of routine clinical care and all treatment decisions were made prior to conduction of the study, ethical approval was not required.Fig. 1 Study outline, selection of patients and follow-up
Clinical features and demographics
Identical clinical and demographic data were collected for both treatment groups. Standard clinical variables, such as age, gender, disease duration (rheumatoid arthritis and respiratory diseases), smoking history, past and current medications, anti-CCP antibody and rheumatoid factor (IgM) positivity, were collected along with the dates of commencement and cessation for the drugs of interest. Rheumatoid arthritis disease activity (DAS28-ESR or DAS28-CRP) before the commencement of the treatment of interest was recorded along with activity scores at 3–4 months post initiation (where available).
Baseline pulmonary function tests prior to the commencement of the medications of interest were also assessed. As only a small proportion of patients had follow-up pulmonary function assessment within a reasonable time frame, this was not analysed. The pattern of ILD involvement was described as recommended by the American Thoracic Society/European Respiratory Society following review of high-resolution (HR) CT-thorax imaging by a multi-disciplinary team including chest radiologists and an expert ILD lead physician (NH) who additionally attributed a modified semi-quantitative CT-fibrosis score by visual inspection [11, 12].
Outcome measures
The primary outcome of interest was time to first respiratory event (days). Respiratory events were defined as either admission to hospital with a respiratory illness (e.g., infection, ILD exacerbation), or death from a respiratory cause while taking the medication of interest.
Additional outcomes of interest included drug continuation. Drug continuation was defined for JAKi as from the date of first prescription to the date of drug cessation, death, end of follow-up period, or whichever end-point came first. Drug continuation for rituximab was defined as from the date of first infusion to the date of cessation, death, end of follow-up period or whichever end-point came first. The number of cycles, frequency and dose of rituximab received was recorded for each patient.
Statistical analysis
Statistical analysis was performed using IBM SPSS statistics v.25.0 (IBM, Armonk, New York, USA). Distributions of clinical variables and demographics were illustrated using histograms. Depending on the data distribution and type (i.e., categorical, continuous), variables were then described using means (with standard deviations) or medians (with inter-quartile ranges). Continuous variables were compared for differences between groups using either student’s t tests or Mann–Whitney U testing subject to normality or non-normality of distribution. Proportions of categorical variables were compared using Chi-squared testing. Kaplan–Meier survival analysis (log-rank) was used to compare respiratory event survival and drug continuation survival between treatment groups. Cox regression proportional hazard was used to calculate unadjusted hazard ratios with 95% confidence intervals (CI 95%). Statistical significance was set at p < 0.05 and analysis was two -tailed.
Results
Demographic and clinical characteristics
A total of 291 patients were commenced on JAKi to February 2020 at the Rheumatic Diseases Unit, Edinburgh (Fig. 1). Twenty-eight patients commenced JAK inhibition for the treatment of rheumatoid arthritis who also had concurrent interstitial lung disease (67.9%), bronchiectasis (25%) or both (7.1%). Baricitinib was commenced in 26 patients and tofacitinib in 2 patients. A total of 318 patients were commenced on rituximab from January 2016 to February 2020 (Fig. 1). Nineteen patients commenced rituximab for the treatment of RA and had concurrent ILD (68.4%), bronchiectasis (26.3%) or both (5.3%).
Mean duration of follow-up (to discontinuation of drug, death or end of follow-up period) for patients receiving JAKi was 1.1 years (SD = 0.62) and for patients receiving rituximab infusions was 2.14 years (SD = 1). Demographic and clinical characteristics were similar in both treatment groups (Table 1); however, those commencing JAKi had a significantly longer disease duration (median 11 years) compared to those starting rituximab (median 3 years) (Table 1; Mann–Whitney U test, p = 0.003).Table 1 Demographic and clinical details of included patients
JAKi (n = 28) Rituximab (n = 19) No. of patients with available data (n =)
ILD/Bronchiectasis/both
n, (%)
ILD n = 19 (67.9%)
Bronchiectasis n = 7 (25%)
Both n = 2 (7.1%)
ILD n = 13 (68.4%)
Bronchiectasis n = 5 (26.3%)
Both n = 1 (5.3%)
47
Age (median, IQR) 69 (62.3–75) 70 (59–76) 47
Female n, (%) 18 (64.3%) 14 (73.7%) 47
Smoking status
n, (%)
Smoker:
n = 5 (17.9%)
Ex-smoker:
n = 15 (53.6%)
Never-smoker:
n = 8 (28.6%)
Smoker:
n = 0 (0%),
Ex-smoker:
n = 12 (63.2%)
Never-smoker:
n = 7 (36.8%)
47
Anti-CCP positivity
n, (%)
25 (96.2%) 19 (100%) 45
Rh-Factor IgM positivity
n, (%)
14 (77.8%) 9 (90%) 28
RA diagnosis age median, (IQR) 53 (44–61) 54 (48–67) 42
Respiratory diagnosis age
median, (IQR)
64 (55–73) 64 (53–71) 46
Years from RA diagnosis to drug of interest
median, (IQR)
11 (3–20) 3 (1–6) 42
DAS-28 Baseline
median, (IQR)
5.85 (4.98–6.18) 6.2 (5.53–6.86) 43
DAS-28 3–4 months
median, (IQR)
3.4 (2.38–4.11) 4.2 (3.61–5.44) 32
JAKi Janus kinase inhibitor, ILD interstitial lung disease, Rh-factor rheumatoid factor, RA rheumatoid arthritis, DAS-28 disease activity score 28 joints
Pre-treatment pulmonary function tests indicated mild severity of pulmonary disease in patients commencing JAKi and rituximab (Table 2). Review of patients’ baseline high-resolution CT-thorax revealed similar radiological patterns and severity of ILD in both treatment groups (Table 2). However, a greater degree of ground glass was evident on CT-imaging before treatment in those receiving JAKi compared to those receiving rituximab (p = 0.036).Table 2 Baseline pulmonary function (all patients); and radiological pattern and severity of ILD on CT-thorax imaging (patients with ILD)
JAKi (n = 28) Rituximab (n = 19) No. of patients with available data (n =)
Baseline predicted FEV1 median %, (IQR) 93.5 (74.5–105) 92 (76–105) 43
Baseline predicted FVC median %, (IQR) 103 (78–112) 99 (86–119) 44
Baseline predicted TLCO median %, (IQR) 62 (51.25–69) 62 (55–67) 41
Radiological ILD-pattern JAKi (n = 20) Rituximab (n = 14)
UIP
n, (%)
0 (0%) 2 (14%)
Probable UIP
n, (%)
4 (20%) 5 (36%)
Fibrotic NSIP
n, (%)
6 (30%) 1 (7%)
Chronic HP
n, (%)
3 (15%) 1 (7%)
Respiratory bronchiolitis-ILD
n, (%)
1 (5%) 0 (0%)
Unclassifiable
n, (%)
6 (30%) 5 (36%)
% of lung with ILD
median, (IQR)
20 (10–45) 12.5 (10–30)
% of lung with ground glass appearance
median, (IQR)
20 (5–45) 7.5 (5–10)
% of lung with honeycombing
median, (IQR)
0 (0–1.75) 0 (0–2)
% of lung with emphysema
median, (IQR)
0.5 (0–8.75) 0 (0–3.75)
JAKi Janus kinase inhibitor, ILD interstitial lung disease, FEV1 forced expiratory volume in 1 second, FVC forced vital capacity, TLco lung transfer capacity for carbon monoxide, IQR inter-quartile range, UIP usual interstitial pneumonia, NSIP non-specific interstitial pneumonia, HP hypersensitivity pneumonitis
Treatment details
Concurrent treatments including conventional synthetic DMARDs and corticosteroids were similar in both groups (Table 3). Almost 30% of patients were biologic naïve before commencing the drug of interest. Prior treatment history indicated that those commencing JAKi had previously been treated with a greater number of biologics including TNFi, tocilizumab and abatacept, compared to those commencing rituximab.Table 3 Previous and concurrent DMARD treatments
JAKi (n = 28) Rituximab (n = 19)
Previous treatments
No. of previous csDMARDs
median, (IQR)
3 (2–4) 3 (2–4)
Methotrexate
n, (%)
22 (78.6%) 13 (68.4%)
Biologic Naïve
n, (%)
8 (28.6%) 6 (31.6%)
No. of previous biologics
median, (IQR)
1 (0–2) 0 (0–1)
Rituximab/JAKi
n, (%)
12 (42.9%) 1 (5.3%)
TNF-α inhibitor
n, (%)
15 (53.6%) 5 (26.3%)
Tocilizumab
n, (%)
6 (21.4%) 1 (5.3%)
Abatacept
n, (%)
6 (21.4%) 1 (5.3%)
Concurrent treatments
Prednisolone; n, (%) 13 (46.4%) 6 (31.6%)
Prednisolone dose (mg) median, (IQR) 7 (5–10) 10 (4.5–12.5)
Methotrexate
n, (%)
4 (14.3%) 2 (10.5%)
Azathioprine
n, (%)
2 (7.1%) 2 (10.5%)
Leflunomide
n, (%)
2 (7.1%) 2 (10.5%)
Hydroxychloroquine
n, (%)
6 (21.4%) 6 (31.6%)
Sulfasalazine
n, (%)
3 (10.7%) 4 (21.1%)
DMARD disease-modifying anti-rheumatic drugs, csDMARDs conventional synthetic DMARDs, JAKi Janus kinase inhibitor, IQR inter-quartile range, TNF Tumour necrosis factor
All except for two patients received standard licensed dosing regimens for JAK inhibition for the treatment of RA (i.e. baricitinib 4 mg once daily; tofacitinib 5 mg twice daily). One patient took 4 mg/2 mg of baricitinib on alternate days and one patient took 5 mg of tofacitinib once daily before increasing to 5 mg twice daily.
All patients who received rituximab received the standard dosing of treatment (2 × intravenous infusions of 1000 mg rituximab with 100 mg intravenous methylprednisolone 14 days apart). Only 2/19 patients were on ‘fixed’ cycles of rituximab therapy repeated every 6 months. All others received rituximab as required depending on clinical status but not within 6 months of the last infusion. 5/19 patients received only 1 cycle of rituximab during their study follow-up period. All others (74%) received between 2 and 5 cycles of rituximab.
Incidence of respiratory events
Respiratory events (i.e. hospitalization or death due to a respiratory cause) occurred in five patients (18%) during JAKi treatment (7 hospitalizations, two of which led to death) and in four patients (21%) (4 hospitalizations, one of which led to death) during treatment with rituximab. In the JAKi group, two patients (7.1%) died of respiratory issues (one new lung cancer with concomitant community acquired pneumonia and one of community acquired pneumonia). In the rituximab-treated patients, two patients (10.5%) died during follow-up; one of an acute myocardial infarction and one of pneumonitis and respiratory failure. No episodes of acute pneumonitis occurred in the JAKi group. There was no difference in respiratory event survival in the JAKi-treated group compared with the rituximab-treated group (Fig. 2a); unadjusted hazard ratio: 1.38 (95% CI 0.36, 5.28); p = 0.64.Fig. 2 a Comparison of time to respiratory event and b drug continuation between JAKi and rituximab for the treatment of rheumatoid arthritis in patients with concurrent bronchiectasis and/or ILD
Drug continuation
Seventy-one percent (20/28) of patients with ILD and/or bronchiectasis who commenced JAKi for the treatment of rheumatoid arthritis remained on the drug at the end of the follow-up period. Reasons for discontinuation included inefficacy (n = 3), death (n = 2, as outlined above), recurrent infections (n = 2) and reduced renal function (n = 1).
Sixty-three percent (12/19) of patients who commenced rituximab continued the drug to the end of the follow-up period. Reasons for cessation included inefficacy (n = 5) or death (n = 2, as outlined above). The rate of discontinuation of JAKi in patients with ILD and/or bronchiectasis did not significantly differ from that of those receiving rituximab (Fig. 2b); unadjusted hazard ratio: 1.9 (95% CI 0.634, 5.73); p = 0.251.
Discussion
There is an absence of patients with interstitial lung disease and bronchiectasis in randomised controlled trials that examine the effectiveness of JAK inhibitors for the treatment of rheumatoid arthritis. Therefore, we have limited knowledge of the pulmonary safety in these cohorts. Prevalence of interstitial lung disease and bronchiectasis is relatively high in patients with rheumatoid arthritis, estimated to affect between 5 and 10% of patients [1, 13]. Interstitial lung disease itself contributes significantly to the excess mortality evident in patients with rheumatoid arthritis [14]. Despite this, the optimal treatment strategy for such patients remains unknown, reflecting a paucity of randomized controlled clinical trials for this patient sub-group. Recently, observational data have suggested treatment with rituximab [6, 15, 16], and laterally abatacept [5, 17] is the most effective and safest biologic agents for the treatment of RA with concurrent ILD or bronchiectasis. Data suggest that treatment with rituximab and abatacept may even infer a benefit to pulmonary function [5, 6]. Nonetheless, concern about lung toxicity restricts the use of many DMARDs and biologic agents, be these concerns soundly proven or not.
Novel, small-molecule treatments, such as Janus kinase inhibitors, represent an exciting option for patients with rheumatoid arthritis. Their safety and potential for effectiveness in patients with concurrent pulmonary disease should be duly considered. Currently, our understanding of inhibition of JAK/STAT signalling pathways in patients with ILD and bronchiectasis is extremely limited. JAKi-use has led to successful treatment of dermatomyositis-associated ILD but evidence is restricted to a small number of case reports and case series [18–20]. Even less is understood about Janus kinase inhibition in patients with bronchiectasis. While post hoc analysis of phase 2, 3 and 3b/4 clinical trials indicates that the de novo incidence of interstitial lung disease is low in patients with rheumatoid arthritis treated with JAKi [21]; to our knowledge, no study has yet examined the pulmonary tolerance and safety of JAKi in those with existing ILD or bronchiectasis. The current study sought to address this important clinical scenario, by comparing time to respiratory events in patients receiving JAKi with patients receiving the current, best-practice medication for this clinical cohort, rituximab.
The majority of our patients commenced JAK1/2 inhibition with baricitinib due to the earlier licensing of this medication for the treatment of rheumatoid arthritis but several patients commenced JAK1/3 inhibition with tofacitinib. Furthermore, the proportion of patients with ‘difficult-to-treat’ rheumatoid disease, manifest by a longer duration of illness and a greater number of previous biologic therapies (e.g. tocilizumab, abatacept), was higher in those subsequently prescribed JAKi. We believe this relates to the fact that a novel medication provides an attractive option for patients who have had difficult-to-treat disease for many years and in such cases patients often chose to switch therapy in an attempt to achieve remission.
Encouragingly, serious respiratory events (i.e., hospitalization or death due to a respiratory cause) were no different between those treated with JAKi and those treated with rituximab suggesting that JAKi may offer a realistic treatment option for patients with RA and concurrent ILD or bronchiectasis. Our cohort of patients represents a group with mild to moderate pulmonary disease, all of whom commenced Janus kinase inhibition or rituximab with a known, pre-existing diagnosis of ILD or bronchiectasis (present on cross sectional CT-imaging and/or attending a tertiary ILD clinic). Very few patients with ILD had a HR-CT pattern consistent with definite UIP and most had a pattern that was consistent with an alternative diagnosis and were highly unlikely to be UIP. Therefore, our findings should not be extrapolated to those with severe ILD or bronchiectasis or a definite UIP pattern of disease and caution is urged in this setting. Furthermore, there was no significant difference in time to drug discontinuation between those prescribed JAKi and those prescribed rituximab. Proportionally, more patients treated with rituximab (almost 40%) ceased treatment during follow-up compared with those treated with JAKi (almost 30%). Reasons for discontinuation of rituximab predominantly related to clinical inefficacy, but comparison of clinical effectiveness was not the aim of this study.
There are several strengths to this study. Both cohorts of patients (i.e., JAKi- and rituximab-receiving) were similar in terms of clinical traits (e.g. gender, concurrent medications, corticosteroid use, anti-CCP and rheumatoid factor seropositivity) minimizing the effect of confounders on respiratory event outcomes. In addition, these data represent real-world experience of patients with RA and concurrent ILD/bronchiectasis with patient–clinician decisions made without influence of a pre-determined research goal. The retrospective nature of this observational study and associated risk of information bias represents the major limitation of this study. The relatively small sample size and short follow-up period also represent limitations that may have affected overall results. Ideally, follow-up pulmonary function testing would have been available for all patients after commencing treatment but as this was only the case for a minority of patients, this was not analysed. It was not possible to collect primary care records of respiratory exacerbations or infections that did not require hospital admission that were treated in the community. Furthermore, the electronic patient health record system used did not permit recording of admissions to hospital abroad or to hospitals outside of NHS Lothian. The authors recommend that the findings of this study and analysis should be investigated further in larger prospective cohorts in the future.
To conclude, within the limitations of this retrospective analysis, JAK inhibition with baricitinib or tofacitinib for the treatment of rheumatoid arthritis in patients with concurrent interstitial lung disease or bronchiectasis did not increase the rate of hospitalization or death due to a respiratory cause in comparison to those treated with rituximab. JAK inhibition may provide a favourable and safe therapeutic option for such patients but larger, prospective studies are required to confirm these findings to inform much-needed therapeutic guidance in this sub-category of patients with RA and ILD/bronchiectasis.
Funding
No specific funding was received from any bodies in the public, commercial or not-for-profit sectors to carry out the work described in this article.
Declarations
Conflict of interest
OC has received sponsorship from Lilly and Pfizer to attend educational conferences. The other authors have no conflicts of interest to declare.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. | METHYLPREDNISOLONE, RITUXIMAB | DrugsGivenReaction | CC BY | 33721094 | 19,059,614 | 2021-05 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Respiratory disorder'. | A retrospective comparison of respiratory events with JAK inhibitors or rituximab for rheumatoid arthritis in patients with pulmonary disease.
Janus kinase inhibitors (JAKi) are an exciting option for the treatment of rheumatoid arthritis (RA) but little is known about their safety and tolerability in patients with existing respiratory disorders. The objective was to compare pulmonary safety of JAKi versus rituximab in patients with concurrent interstitial lung disease (ILD) or bronchiectasis. We performed a retrospective electronic patient record review of patients with known ILD or bronchiectasis commencing JAKi or rituximab for the treatment of RA. Patients initiating treatment from January 2016 to February 2020 were included. Respiratory events (hospitalization or death from a respiratory cause) were compared using Kaplan-Meier survival analysis. We analysed patients who received JAKi (n = 28) and rituximab (n = 19) for a mean (SD) of 1.1 (0.62) and 2.14 (1) years respectively. Patients were predominantly female (68%), anti-CCP antibody positive (94%) and non-smoking (89%) with a median (IQR) percentage predicted FVC at baseline of 100% (82-115%) and percentage predicted TLCO of 62% (54.5-68%). Respiratory events occurred in five patients treated with JAKi (18%; 5 hospitalizations, 2 deaths) and in four patients treated with rituximab (21%; 3 hospitalizations, 1 death). Respiratory event rates did not differ between groups (Cox-regression proportional hazard ratio = 1.38, 95% CI 0.36-5.28; p = 0.64). In this retrospective study, JAKi for the treatment of RA with existing ILD or bronchiectasis did not increase the rate of hospitalization or death due to respiratory causes compared to those treated with rituximab. JAK inhibition may provide a relatively safe option for RA in such patients.
Introduction
Therapeutic decision-making in the treatment of rheumatoid arthritis (RA) for patients with concurrent pulmonary disease can be challenging. Fear regarding increased infection risk in bronchiectasis and progression of interstitial lung disease (ILD) secondary to drug administration restricts treatment options. Controversy exists as to whether medications, such as methotrexate, leflunomide and tumour necrosis factor-α inhibitors (TNFi), exacerbate pulmonary disease in patients with concurrent ILD and bronchiectasis [1]. This has led to hesitancy in utilizing these medications although systematic review of the evidence is beginning to allay fears [2, 3]. Recently, data have begun to support the safe and effective use of B and T cell therapies, rituximab [4] and abatacept [5], in the management of rheumatoid arthritis-related ILD and bronchiectasis [6, 7]. However, therapeutic options remain limited.
The approval of Janus kinase inhibitors (JAKi) (i.e., baricitinib, tofacitinib, upadacitinib) in the management of moderate to severe RA offers a new and exciting treatment paradigm [8–10]. However, inclusion of patients with concurrent ILD or bronchiectasis in relevant clinical trials is extremely limited. While the use of JAKi for the treatment of RA with co-existent lung disease is tempting due to the oral route of administration and relatively short half-life, their safety in relation to pulmonary toxicity in patients with established ILD or bronchiectasis is unknown.
To investigate this evidence gap, the objective of the current study is to compare the pulmonary safety of Janus kinase inhibition for the treatment of rheumatoid arthritis in patients with existing ILD or bronchiectasis, with rituximab. The primary outcome of interest was the incidence of severe respiratory events (hospitalization or death due to respiratory cause) in those receiving JAKi versus those receiving rituximab.
Materials and methods
Study design and patient inclusion
The study comprised a retrospective observational study. The electronic patient health record system (TrakCare®) was used to collect relevant data. Adult patients (aged 18 or over) undergoing treatment for rheumatoid arthritis at the Rheumatic Diseases Unit, Western General Hospital, Edinburgh, U.K., were included in the study. Only patients with an established diagnosis of ILD or bronchiectasis before the commencement of the drug of interest were eligible for inclusion. Patients without these respiratory conditions were excluded. Dedicated pharmacy and biologic clinic records were used to identify all adult patients (16 years or older) commencing any JAKi or rituximab for the treatment of RA from January 2016 to February 2020 (Fig. 1). From these records, all patients’ health records were screened by the investigators (OC, LK) for co-existing diagnoses of interstitial lung disease and/or bronchiectasis. For relevant patients, data were collected in detail from electronic health records by the investigators (OC, OM). Data were collected until discontinuation of the drug of interest (as documented in the clinical record) or to the end of the follow-up period (1 August 2020). They study was conducted in accordance with the Declaration of Helsinki. As all data were collected as part of routine clinical care and all treatment decisions were made prior to conduction of the study, ethical approval was not required.Fig. 1 Study outline, selection of patients and follow-up
Clinical features and demographics
Identical clinical and demographic data were collected for both treatment groups. Standard clinical variables, such as age, gender, disease duration (rheumatoid arthritis and respiratory diseases), smoking history, past and current medications, anti-CCP antibody and rheumatoid factor (IgM) positivity, were collected along with the dates of commencement and cessation for the drugs of interest. Rheumatoid arthritis disease activity (DAS28-ESR or DAS28-CRP) before the commencement of the treatment of interest was recorded along with activity scores at 3–4 months post initiation (where available).
Baseline pulmonary function tests prior to the commencement of the medications of interest were also assessed. As only a small proportion of patients had follow-up pulmonary function assessment within a reasonable time frame, this was not analysed. The pattern of ILD involvement was described as recommended by the American Thoracic Society/European Respiratory Society following review of high-resolution (HR) CT-thorax imaging by a multi-disciplinary team including chest radiologists and an expert ILD lead physician (NH) who additionally attributed a modified semi-quantitative CT-fibrosis score by visual inspection [11, 12].
Outcome measures
The primary outcome of interest was time to first respiratory event (days). Respiratory events were defined as either admission to hospital with a respiratory illness (e.g., infection, ILD exacerbation), or death from a respiratory cause while taking the medication of interest.
Additional outcomes of interest included drug continuation. Drug continuation was defined for JAKi as from the date of first prescription to the date of drug cessation, death, end of follow-up period, or whichever end-point came first. Drug continuation for rituximab was defined as from the date of first infusion to the date of cessation, death, end of follow-up period or whichever end-point came first. The number of cycles, frequency and dose of rituximab received was recorded for each patient.
Statistical analysis
Statistical analysis was performed using IBM SPSS statistics v.25.0 (IBM, Armonk, New York, USA). Distributions of clinical variables and demographics were illustrated using histograms. Depending on the data distribution and type (i.e., categorical, continuous), variables were then described using means (with standard deviations) or medians (with inter-quartile ranges). Continuous variables were compared for differences between groups using either student’s t tests or Mann–Whitney U testing subject to normality or non-normality of distribution. Proportions of categorical variables were compared using Chi-squared testing. Kaplan–Meier survival analysis (log-rank) was used to compare respiratory event survival and drug continuation survival between treatment groups. Cox regression proportional hazard was used to calculate unadjusted hazard ratios with 95% confidence intervals (CI 95%). Statistical significance was set at p < 0.05 and analysis was two -tailed.
Results
Demographic and clinical characteristics
A total of 291 patients were commenced on JAKi to February 2020 at the Rheumatic Diseases Unit, Edinburgh (Fig. 1). Twenty-eight patients commenced JAK inhibition for the treatment of rheumatoid arthritis who also had concurrent interstitial lung disease (67.9%), bronchiectasis (25%) or both (7.1%). Baricitinib was commenced in 26 patients and tofacitinib in 2 patients. A total of 318 patients were commenced on rituximab from January 2016 to February 2020 (Fig. 1). Nineteen patients commenced rituximab for the treatment of RA and had concurrent ILD (68.4%), bronchiectasis (26.3%) or both (5.3%).
Mean duration of follow-up (to discontinuation of drug, death or end of follow-up period) for patients receiving JAKi was 1.1 years (SD = 0.62) and for patients receiving rituximab infusions was 2.14 years (SD = 1). Demographic and clinical characteristics were similar in both treatment groups (Table 1); however, those commencing JAKi had a significantly longer disease duration (median 11 years) compared to those starting rituximab (median 3 years) (Table 1; Mann–Whitney U test, p = 0.003).Table 1 Demographic and clinical details of included patients
JAKi (n = 28) Rituximab (n = 19) No. of patients with available data (n =)
ILD/Bronchiectasis/both
n, (%)
ILD n = 19 (67.9%)
Bronchiectasis n = 7 (25%)
Both n = 2 (7.1%)
ILD n = 13 (68.4%)
Bronchiectasis n = 5 (26.3%)
Both n = 1 (5.3%)
47
Age (median, IQR) 69 (62.3–75) 70 (59–76) 47
Female n, (%) 18 (64.3%) 14 (73.7%) 47
Smoking status
n, (%)
Smoker:
n = 5 (17.9%)
Ex-smoker:
n = 15 (53.6%)
Never-smoker:
n = 8 (28.6%)
Smoker:
n = 0 (0%),
Ex-smoker:
n = 12 (63.2%)
Never-smoker:
n = 7 (36.8%)
47
Anti-CCP positivity
n, (%)
25 (96.2%) 19 (100%) 45
Rh-Factor IgM positivity
n, (%)
14 (77.8%) 9 (90%) 28
RA diagnosis age median, (IQR) 53 (44–61) 54 (48–67) 42
Respiratory diagnosis age
median, (IQR)
64 (55–73) 64 (53–71) 46
Years from RA diagnosis to drug of interest
median, (IQR)
11 (3–20) 3 (1–6) 42
DAS-28 Baseline
median, (IQR)
5.85 (4.98–6.18) 6.2 (5.53–6.86) 43
DAS-28 3–4 months
median, (IQR)
3.4 (2.38–4.11) 4.2 (3.61–5.44) 32
JAKi Janus kinase inhibitor, ILD interstitial lung disease, Rh-factor rheumatoid factor, RA rheumatoid arthritis, DAS-28 disease activity score 28 joints
Pre-treatment pulmonary function tests indicated mild severity of pulmonary disease in patients commencing JAKi and rituximab (Table 2). Review of patients’ baseline high-resolution CT-thorax revealed similar radiological patterns and severity of ILD in both treatment groups (Table 2). However, a greater degree of ground glass was evident on CT-imaging before treatment in those receiving JAKi compared to those receiving rituximab (p = 0.036).Table 2 Baseline pulmonary function (all patients); and radiological pattern and severity of ILD on CT-thorax imaging (patients with ILD)
JAKi (n = 28) Rituximab (n = 19) No. of patients with available data (n =)
Baseline predicted FEV1 median %, (IQR) 93.5 (74.5–105) 92 (76–105) 43
Baseline predicted FVC median %, (IQR) 103 (78–112) 99 (86–119) 44
Baseline predicted TLCO median %, (IQR) 62 (51.25–69) 62 (55–67) 41
Radiological ILD-pattern JAKi (n = 20) Rituximab (n = 14)
UIP
n, (%)
0 (0%) 2 (14%)
Probable UIP
n, (%)
4 (20%) 5 (36%)
Fibrotic NSIP
n, (%)
6 (30%) 1 (7%)
Chronic HP
n, (%)
3 (15%) 1 (7%)
Respiratory bronchiolitis-ILD
n, (%)
1 (5%) 0 (0%)
Unclassifiable
n, (%)
6 (30%) 5 (36%)
% of lung with ILD
median, (IQR)
20 (10–45) 12.5 (10–30)
% of lung with ground glass appearance
median, (IQR)
20 (5–45) 7.5 (5–10)
% of lung with honeycombing
median, (IQR)
0 (0–1.75) 0 (0–2)
% of lung with emphysema
median, (IQR)
0.5 (0–8.75) 0 (0–3.75)
JAKi Janus kinase inhibitor, ILD interstitial lung disease, FEV1 forced expiratory volume in 1 second, FVC forced vital capacity, TLco lung transfer capacity for carbon monoxide, IQR inter-quartile range, UIP usual interstitial pneumonia, NSIP non-specific interstitial pneumonia, HP hypersensitivity pneumonitis
Treatment details
Concurrent treatments including conventional synthetic DMARDs and corticosteroids were similar in both groups (Table 3). Almost 30% of patients were biologic naïve before commencing the drug of interest. Prior treatment history indicated that those commencing JAKi had previously been treated with a greater number of biologics including TNFi, tocilizumab and abatacept, compared to those commencing rituximab.Table 3 Previous and concurrent DMARD treatments
JAKi (n = 28) Rituximab (n = 19)
Previous treatments
No. of previous csDMARDs
median, (IQR)
3 (2–4) 3 (2–4)
Methotrexate
n, (%)
22 (78.6%) 13 (68.4%)
Biologic Naïve
n, (%)
8 (28.6%) 6 (31.6%)
No. of previous biologics
median, (IQR)
1 (0–2) 0 (0–1)
Rituximab/JAKi
n, (%)
12 (42.9%) 1 (5.3%)
TNF-α inhibitor
n, (%)
15 (53.6%) 5 (26.3%)
Tocilizumab
n, (%)
6 (21.4%) 1 (5.3%)
Abatacept
n, (%)
6 (21.4%) 1 (5.3%)
Concurrent treatments
Prednisolone; n, (%) 13 (46.4%) 6 (31.6%)
Prednisolone dose (mg) median, (IQR) 7 (5–10) 10 (4.5–12.5)
Methotrexate
n, (%)
4 (14.3%) 2 (10.5%)
Azathioprine
n, (%)
2 (7.1%) 2 (10.5%)
Leflunomide
n, (%)
2 (7.1%) 2 (10.5%)
Hydroxychloroquine
n, (%)
6 (21.4%) 6 (31.6%)
Sulfasalazine
n, (%)
3 (10.7%) 4 (21.1%)
DMARD disease-modifying anti-rheumatic drugs, csDMARDs conventional synthetic DMARDs, JAKi Janus kinase inhibitor, IQR inter-quartile range, TNF Tumour necrosis factor
All except for two patients received standard licensed dosing regimens for JAK inhibition for the treatment of RA (i.e. baricitinib 4 mg once daily; tofacitinib 5 mg twice daily). One patient took 4 mg/2 mg of baricitinib on alternate days and one patient took 5 mg of tofacitinib once daily before increasing to 5 mg twice daily.
All patients who received rituximab received the standard dosing of treatment (2 × intravenous infusions of 1000 mg rituximab with 100 mg intravenous methylprednisolone 14 days apart). Only 2/19 patients were on ‘fixed’ cycles of rituximab therapy repeated every 6 months. All others received rituximab as required depending on clinical status but not within 6 months of the last infusion. 5/19 patients received only 1 cycle of rituximab during their study follow-up period. All others (74%) received between 2 and 5 cycles of rituximab.
Incidence of respiratory events
Respiratory events (i.e. hospitalization or death due to a respiratory cause) occurred in five patients (18%) during JAKi treatment (7 hospitalizations, two of which led to death) and in four patients (21%) (4 hospitalizations, one of which led to death) during treatment with rituximab. In the JAKi group, two patients (7.1%) died of respiratory issues (one new lung cancer with concomitant community acquired pneumonia and one of community acquired pneumonia). In the rituximab-treated patients, two patients (10.5%) died during follow-up; one of an acute myocardial infarction and one of pneumonitis and respiratory failure. No episodes of acute pneumonitis occurred in the JAKi group. There was no difference in respiratory event survival in the JAKi-treated group compared with the rituximab-treated group (Fig. 2a); unadjusted hazard ratio: 1.38 (95% CI 0.36, 5.28); p = 0.64.Fig. 2 a Comparison of time to respiratory event and b drug continuation between JAKi and rituximab for the treatment of rheumatoid arthritis in patients with concurrent bronchiectasis and/or ILD
Drug continuation
Seventy-one percent (20/28) of patients with ILD and/or bronchiectasis who commenced JAKi for the treatment of rheumatoid arthritis remained on the drug at the end of the follow-up period. Reasons for discontinuation included inefficacy (n = 3), death (n = 2, as outlined above), recurrent infections (n = 2) and reduced renal function (n = 1).
Sixty-three percent (12/19) of patients who commenced rituximab continued the drug to the end of the follow-up period. Reasons for cessation included inefficacy (n = 5) or death (n = 2, as outlined above). The rate of discontinuation of JAKi in patients with ILD and/or bronchiectasis did not significantly differ from that of those receiving rituximab (Fig. 2b); unadjusted hazard ratio: 1.9 (95% CI 0.634, 5.73); p = 0.251.
Discussion
There is an absence of patients with interstitial lung disease and bronchiectasis in randomised controlled trials that examine the effectiveness of JAK inhibitors for the treatment of rheumatoid arthritis. Therefore, we have limited knowledge of the pulmonary safety in these cohorts. Prevalence of interstitial lung disease and bronchiectasis is relatively high in patients with rheumatoid arthritis, estimated to affect between 5 and 10% of patients [1, 13]. Interstitial lung disease itself contributes significantly to the excess mortality evident in patients with rheumatoid arthritis [14]. Despite this, the optimal treatment strategy for such patients remains unknown, reflecting a paucity of randomized controlled clinical trials for this patient sub-group. Recently, observational data have suggested treatment with rituximab [6, 15, 16], and laterally abatacept [5, 17] is the most effective and safest biologic agents for the treatment of RA with concurrent ILD or bronchiectasis. Data suggest that treatment with rituximab and abatacept may even infer a benefit to pulmonary function [5, 6]. Nonetheless, concern about lung toxicity restricts the use of many DMARDs and biologic agents, be these concerns soundly proven or not.
Novel, small-molecule treatments, such as Janus kinase inhibitors, represent an exciting option for patients with rheumatoid arthritis. Their safety and potential for effectiveness in patients with concurrent pulmonary disease should be duly considered. Currently, our understanding of inhibition of JAK/STAT signalling pathways in patients with ILD and bronchiectasis is extremely limited. JAKi-use has led to successful treatment of dermatomyositis-associated ILD but evidence is restricted to a small number of case reports and case series [18–20]. Even less is understood about Janus kinase inhibition in patients with bronchiectasis. While post hoc analysis of phase 2, 3 and 3b/4 clinical trials indicates that the de novo incidence of interstitial lung disease is low in patients with rheumatoid arthritis treated with JAKi [21]; to our knowledge, no study has yet examined the pulmonary tolerance and safety of JAKi in those with existing ILD or bronchiectasis. The current study sought to address this important clinical scenario, by comparing time to respiratory events in patients receiving JAKi with patients receiving the current, best-practice medication for this clinical cohort, rituximab.
The majority of our patients commenced JAK1/2 inhibition with baricitinib due to the earlier licensing of this medication for the treatment of rheumatoid arthritis but several patients commenced JAK1/3 inhibition with tofacitinib. Furthermore, the proportion of patients with ‘difficult-to-treat’ rheumatoid disease, manifest by a longer duration of illness and a greater number of previous biologic therapies (e.g. tocilizumab, abatacept), was higher in those subsequently prescribed JAKi. We believe this relates to the fact that a novel medication provides an attractive option for patients who have had difficult-to-treat disease for many years and in such cases patients often chose to switch therapy in an attempt to achieve remission.
Encouragingly, serious respiratory events (i.e., hospitalization or death due to a respiratory cause) were no different between those treated with JAKi and those treated with rituximab suggesting that JAKi may offer a realistic treatment option for patients with RA and concurrent ILD or bronchiectasis. Our cohort of patients represents a group with mild to moderate pulmonary disease, all of whom commenced Janus kinase inhibition or rituximab with a known, pre-existing diagnosis of ILD or bronchiectasis (present on cross sectional CT-imaging and/or attending a tertiary ILD clinic). Very few patients with ILD had a HR-CT pattern consistent with definite UIP and most had a pattern that was consistent with an alternative diagnosis and were highly unlikely to be UIP. Therefore, our findings should not be extrapolated to those with severe ILD or bronchiectasis or a definite UIP pattern of disease and caution is urged in this setting. Furthermore, there was no significant difference in time to drug discontinuation between those prescribed JAKi and those prescribed rituximab. Proportionally, more patients treated with rituximab (almost 40%) ceased treatment during follow-up compared with those treated with JAKi (almost 30%). Reasons for discontinuation of rituximab predominantly related to clinical inefficacy, but comparison of clinical effectiveness was not the aim of this study.
There are several strengths to this study. Both cohorts of patients (i.e., JAKi- and rituximab-receiving) were similar in terms of clinical traits (e.g. gender, concurrent medications, corticosteroid use, anti-CCP and rheumatoid factor seropositivity) minimizing the effect of confounders on respiratory event outcomes. In addition, these data represent real-world experience of patients with RA and concurrent ILD/bronchiectasis with patient–clinician decisions made without influence of a pre-determined research goal. The retrospective nature of this observational study and associated risk of information bias represents the major limitation of this study. The relatively small sample size and short follow-up period also represent limitations that may have affected overall results. Ideally, follow-up pulmonary function testing would have been available for all patients after commencing treatment but as this was only the case for a minority of patients, this was not analysed. It was not possible to collect primary care records of respiratory exacerbations or infections that did not require hospital admission that were treated in the community. Furthermore, the electronic patient health record system used did not permit recording of admissions to hospital abroad or to hospitals outside of NHS Lothian. The authors recommend that the findings of this study and analysis should be investigated further in larger prospective cohorts in the future.
To conclude, within the limitations of this retrospective analysis, JAK inhibition with baricitinib or tofacitinib for the treatment of rheumatoid arthritis in patients with concurrent interstitial lung disease or bronchiectasis did not increase the rate of hospitalization or death due to a respiratory cause in comparison to those treated with rituximab. JAK inhibition may provide a favourable and safe therapeutic option for such patients but larger, prospective studies are required to confirm these findings to inform much-needed therapeutic guidance in this sub-category of patients with RA and ILD/bronchiectasis.
Funding
No specific funding was received from any bodies in the public, commercial or not-for-profit sectors to carry out the work described in this article.
Declarations
Conflict of interest
OC has received sponsorship from Lilly and Pfizer to attend educational conferences. The other authors have no conflicts of interest to declare.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. | METHYLPREDNISOLONE, RITUXIMAB | DrugsGivenReaction | CC BY | 33721094 | 19,064,815 | 2021-05 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Respiratory failure'. | A retrospective comparison of respiratory events with JAK inhibitors or rituximab for rheumatoid arthritis in patients with pulmonary disease.
Janus kinase inhibitors (JAKi) are an exciting option for the treatment of rheumatoid arthritis (RA) but little is known about their safety and tolerability in patients with existing respiratory disorders. The objective was to compare pulmonary safety of JAKi versus rituximab in patients with concurrent interstitial lung disease (ILD) or bronchiectasis. We performed a retrospective electronic patient record review of patients with known ILD or bronchiectasis commencing JAKi or rituximab for the treatment of RA. Patients initiating treatment from January 2016 to February 2020 were included. Respiratory events (hospitalization or death from a respiratory cause) were compared using Kaplan-Meier survival analysis. We analysed patients who received JAKi (n = 28) and rituximab (n = 19) for a mean (SD) of 1.1 (0.62) and 2.14 (1) years respectively. Patients were predominantly female (68%), anti-CCP antibody positive (94%) and non-smoking (89%) with a median (IQR) percentage predicted FVC at baseline of 100% (82-115%) and percentage predicted TLCO of 62% (54.5-68%). Respiratory events occurred in five patients treated with JAKi (18%; 5 hospitalizations, 2 deaths) and in four patients treated with rituximab (21%; 3 hospitalizations, 1 death). Respiratory event rates did not differ between groups (Cox-regression proportional hazard ratio = 1.38, 95% CI 0.36-5.28; p = 0.64). In this retrospective study, JAKi for the treatment of RA with existing ILD or bronchiectasis did not increase the rate of hospitalization or death due to respiratory causes compared to those treated with rituximab. JAK inhibition may provide a relatively safe option for RA in such patients.
Introduction
Therapeutic decision-making in the treatment of rheumatoid arthritis (RA) for patients with concurrent pulmonary disease can be challenging. Fear regarding increased infection risk in bronchiectasis and progression of interstitial lung disease (ILD) secondary to drug administration restricts treatment options. Controversy exists as to whether medications, such as methotrexate, leflunomide and tumour necrosis factor-α inhibitors (TNFi), exacerbate pulmonary disease in patients with concurrent ILD and bronchiectasis [1]. This has led to hesitancy in utilizing these medications although systematic review of the evidence is beginning to allay fears [2, 3]. Recently, data have begun to support the safe and effective use of B and T cell therapies, rituximab [4] and abatacept [5], in the management of rheumatoid arthritis-related ILD and bronchiectasis [6, 7]. However, therapeutic options remain limited.
The approval of Janus kinase inhibitors (JAKi) (i.e., baricitinib, tofacitinib, upadacitinib) in the management of moderate to severe RA offers a new and exciting treatment paradigm [8–10]. However, inclusion of patients with concurrent ILD or bronchiectasis in relevant clinical trials is extremely limited. While the use of JAKi for the treatment of RA with co-existent lung disease is tempting due to the oral route of administration and relatively short half-life, their safety in relation to pulmonary toxicity in patients with established ILD or bronchiectasis is unknown.
To investigate this evidence gap, the objective of the current study is to compare the pulmonary safety of Janus kinase inhibition for the treatment of rheumatoid arthritis in patients with existing ILD or bronchiectasis, with rituximab. The primary outcome of interest was the incidence of severe respiratory events (hospitalization or death due to respiratory cause) in those receiving JAKi versus those receiving rituximab.
Materials and methods
Study design and patient inclusion
The study comprised a retrospective observational study. The electronic patient health record system (TrakCare®) was used to collect relevant data. Adult patients (aged 18 or over) undergoing treatment for rheumatoid arthritis at the Rheumatic Diseases Unit, Western General Hospital, Edinburgh, U.K., were included in the study. Only patients with an established diagnosis of ILD or bronchiectasis before the commencement of the drug of interest were eligible for inclusion. Patients without these respiratory conditions were excluded. Dedicated pharmacy and biologic clinic records were used to identify all adult patients (16 years or older) commencing any JAKi or rituximab for the treatment of RA from January 2016 to February 2020 (Fig. 1). From these records, all patients’ health records were screened by the investigators (OC, LK) for co-existing diagnoses of interstitial lung disease and/or bronchiectasis. For relevant patients, data were collected in detail from electronic health records by the investigators (OC, OM). Data were collected until discontinuation of the drug of interest (as documented in the clinical record) or to the end of the follow-up period (1 August 2020). They study was conducted in accordance with the Declaration of Helsinki. As all data were collected as part of routine clinical care and all treatment decisions were made prior to conduction of the study, ethical approval was not required.Fig. 1 Study outline, selection of patients and follow-up
Clinical features and demographics
Identical clinical and demographic data were collected for both treatment groups. Standard clinical variables, such as age, gender, disease duration (rheumatoid arthritis and respiratory diseases), smoking history, past and current medications, anti-CCP antibody and rheumatoid factor (IgM) positivity, were collected along with the dates of commencement and cessation for the drugs of interest. Rheumatoid arthritis disease activity (DAS28-ESR or DAS28-CRP) before the commencement of the treatment of interest was recorded along with activity scores at 3–4 months post initiation (where available).
Baseline pulmonary function tests prior to the commencement of the medications of interest were also assessed. As only a small proportion of patients had follow-up pulmonary function assessment within a reasonable time frame, this was not analysed. The pattern of ILD involvement was described as recommended by the American Thoracic Society/European Respiratory Society following review of high-resolution (HR) CT-thorax imaging by a multi-disciplinary team including chest radiologists and an expert ILD lead physician (NH) who additionally attributed a modified semi-quantitative CT-fibrosis score by visual inspection [11, 12].
Outcome measures
The primary outcome of interest was time to first respiratory event (days). Respiratory events were defined as either admission to hospital with a respiratory illness (e.g., infection, ILD exacerbation), or death from a respiratory cause while taking the medication of interest.
Additional outcomes of interest included drug continuation. Drug continuation was defined for JAKi as from the date of first prescription to the date of drug cessation, death, end of follow-up period, or whichever end-point came first. Drug continuation for rituximab was defined as from the date of first infusion to the date of cessation, death, end of follow-up period or whichever end-point came first. The number of cycles, frequency and dose of rituximab received was recorded for each patient.
Statistical analysis
Statistical analysis was performed using IBM SPSS statistics v.25.0 (IBM, Armonk, New York, USA). Distributions of clinical variables and demographics were illustrated using histograms. Depending on the data distribution and type (i.e., categorical, continuous), variables were then described using means (with standard deviations) or medians (with inter-quartile ranges). Continuous variables were compared for differences between groups using either student’s t tests or Mann–Whitney U testing subject to normality or non-normality of distribution. Proportions of categorical variables were compared using Chi-squared testing. Kaplan–Meier survival analysis (log-rank) was used to compare respiratory event survival and drug continuation survival between treatment groups. Cox regression proportional hazard was used to calculate unadjusted hazard ratios with 95% confidence intervals (CI 95%). Statistical significance was set at p < 0.05 and analysis was two -tailed.
Results
Demographic and clinical characteristics
A total of 291 patients were commenced on JAKi to February 2020 at the Rheumatic Diseases Unit, Edinburgh (Fig. 1). Twenty-eight patients commenced JAK inhibition for the treatment of rheumatoid arthritis who also had concurrent interstitial lung disease (67.9%), bronchiectasis (25%) or both (7.1%). Baricitinib was commenced in 26 patients and tofacitinib in 2 patients. A total of 318 patients were commenced on rituximab from January 2016 to February 2020 (Fig. 1). Nineteen patients commenced rituximab for the treatment of RA and had concurrent ILD (68.4%), bronchiectasis (26.3%) or both (5.3%).
Mean duration of follow-up (to discontinuation of drug, death or end of follow-up period) for patients receiving JAKi was 1.1 years (SD = 0.62) and for patients receiving rituximab infusions was 2.14 years (SD = 1). Demographic and clinical characteristics were similar in both treatment groups (Table 1); however, those commencing JAKi had a significantly longer disease duration (median 11 years) compared to those starting rituximab (median 3 years) (Table 1; Mann–Whitney U test, p = 0.003).Table 1 Demographic and clinical details of included patients
JAKi (n = 28) Rituximab (n = 19) No. of patients with available data (n =)
ILD/Bronchiectasis/both
n, (%)
ILD n = 19 (67.9%)
Bronchiectasis n = 7 (25%)
Both n = 2 (7.1%)
ILD n = 13 (68.4%)
Bronchiectasis n = 5 (26.3%)
Both n = 1 (5.3%)
47
Age (median, IQR) 69 (62.3–75) 70 (59–76) 47
Female n, (%) 18 (64.3%) 14 (73.7%) 47
Smoking status
n, (%)
Smoker:
n = 5 (17.9%)
Ex-smoker:
n = 15 (53.6%)
Never-smoker:
n = 8 (28.6%)
Smoker:
n = 0 (0%),
Ex-smoker:
n = 12 (63.2%)
Never-smoker:
n = 7 (36.8%)
47
Anti-CCP positivity
n, (%)
25 (96.2%) 19 (100%) 45
Rh-Factor IgM positivity
n, (%)
14 (77.8%) 9 (90%) 28
RA diagnosis age median, (IQR) 53 (44–61) 54 (48–67) 42
Respiratory diagnosis age
median, (IQR)
64 (55–73) 64 (53–71) 46
Years from RA diagnosis to drug of interest
median, (IQR)
11 (3–20) 3 (1–6) 42
DAS-28 Baseline
median, (IQR)
5.85 (4.98–6.18) 6.2 (5.53–6.86) 43
DAS-28 3–4 months
median, (IQR)
3.4 (2.38–4.11) 4.2 (3.61–5.44) 32
JAKi Janus kinase inhibitor, ILD interstitial lung disease, Rh-factor rheumatoid factor, RA rheumatoid arthritis, DAS-28 disease activity score 28 joints
Pre-treatment pulmonary function tests indicated mild severity of pulmonary disease in patients commencing JAKi and rituximab (Table 2). Review of patients’ baseline high-resolution CT-thorax revealed similar radiological patterns and severity of ILD in both treatment groups (Table 2). However, a greater degree of ground glass was evident on CT-imaging before treatment in those receiving JAKi compared to those receiving rituximab (p = 0.036).Table 2 Baseline pulmonary function (all patients); and radiological pattern and severity of ILD on CT-thorax imaging (patients with ILD)
JAKi (n = 28) Rituximab (n = 19) No. of patients with available data (n =)
Baseline predicted FEV1 median %, (IQR) 93.5 (74.5–105) 92 (76–105) 43
Baseline predicted FVC median %, (IQR) 103 (78–112) 99 (86–119) 44
Baseline predicted TLCO median %, (IQR) 62 (51.25–69) 62 (55–67) 41
Radiological ILD-pattern JAKi (n = 20) Rituximab (n = 14)
UIP
n, (%)
0 (0%) 2 (14%)
Probable UIP
n, (%)
4 (20%) 5 (36%)
Fibrotic NSIP
n, (%)
6 (30%) 1 (7%)
Chronic HP
n, (%)
3 (15%) 1 (7%)
Respiratory bronchiolitis-ILD
n, (%)
1 (5%) 0 (0%)
Unclassifiable
n, (%)
6 (30%) 5 (36%)
% of lung with ILD
median, (IQR)
20 (10–45) 12.5 (10–30)
% of lung with ground glass appearance
median, (IQR)
20 (5–45) 7.5 (5–10)
% of lung with honeycombing
median, (IQR)
0 (0–1.75) 0 (0–2)
% of lung with emphysema
median, (IQR)
0.5 (0–8.75) 0 (0–3.75)
JAKi Janus kinase inhibitor, ILD interstitial lung disease, FEV1 forced expiratory volume in 1 second, FVC forced vital capacity, TLco lung transfer capacity for carbon monoxide, IQR inter-quartile range, UIP usual interstitial pneumonia, NSIP non-specific interstitial pneumonia, HP hypersensitivity pneumonitis
Treatment details
Concurrent treatments including conventional synthetic DMARDs and corticosteroids were similar in both groups (Table 3). Almost 30% of patients were biologic naïve before commencing the drug of interest. Prior treatment history indicated that those commencing JAKi had previously been treated with a greater number of biologics including TNFi, tocilizumab and abatacept, compared to those commencing rituximab.Table 3 Previous and concurrent DMARD treatments
JAKi (n = 28) Rituximab (n = 19)
Previous treatments
No. of previous csDMARDs
median, (IQR)
3 (2–4) 3 (2–4)
Methotrexate
n, (%)
22 (78.6%) 13 (68.4%)
Biologic Naïve
n, (%)
8 (28.6%) 6 (31.6%)
No. of previous biologics
median, (IQR)
1 (0–2) 0 (0–1)
Rituximab/JAKi
n, (%)
12 (42.9%) 1 (5.3%)
TNF-α inhibitor
n, (%)
15 (53.6%) 5 (26.3%)
Tocilizumab
n, (%)
6 (21.4%) 1 (5.3%)
Abatacept
n, (%)
6 (21.4%) 1 (5.3%)
Concurrent treatments
Prednisolone; n, (%) 13 (46.4%) 6 (31.6%)
Prednisolone dose (mg) median, (IQR) 7 (5–10) 10 (4.5–12.5)
Methotrexate
n, (%)
4 (14.3%) 2 (10.5%)
Azathioprine
n, (%)
2 (7.1%) 2 (10.5%)
Leflunomide
n, (%)
2 (7.1%) 2 (10.5%)
Hydroxychloroquine
n, (%)
6 (21.4%) 6 (31.6%)
Sulfasalazine
n, (%)
3 (10.7%) 4 (21.1%)
DMARD disease-modifying anti-rheumatic drugs, csDMARDs conventional synthetic DMARDs, JAKi Janus kinase inhibitor, IQR inter-quartile range, TNF Tumour necrosis factor
All except for two patients received standard licensed dosing regimens for JAK inhibition for the treatment of RA (i.e. baricitinib 4 mg once daily; tofacitinib 5 mg twice daily). One patient took 4 mg/2 mg of baricitinib on alternate days and one patient took 5 mg of tofacitinib once daily before increasing to 5 mg twice daily.
All patients who received rituximab received the standard dosing of treatment (2 × intravenous infusions of 1000 mg rituximab with 100 mg intravenous methylprednisolone 14 days apart). Only 2/19 patients were on ‘fixed’ cycles of rituximab therapy repeated every 6 months. All others received rituximab as required depending on clinical status but not within 6 months of the last infusion. 5/19 patients received only 1 cycle of rituximab during their study follow-up period. All others (74%) received between 2 and 5 cycles of rituximab.
Incidence of respiratory events
Respiratory events (i.e. hospitalization or death due to a respiratory cause) occurred in five patients (18%) during JAKi treatment (7 hospitalizations, two of which led to death) and in four patients (21%) (4 hospitalizations, one of which led to death) during treatment with rituximab. In the JAKi group, two patients (7.1%) died of respiratory issues (one new lung cancer with concomitant community acquired pneumonia and one of community acquired pneumonia). In the rituximab-treated patients, two patients (10.5%) died during follow-up; one of an acute myocardial infarction and one of pneumonitis and respiratory failure. No episodes of acute pneumonitis occurred in the JAKi group. There was no difference in respiratory event survival in the JAKi-treated group compared with the rituximab-treated group (Fig. 2a); unadjusted hazard ratio: 1.38 (95% CI 0.36, 5.28); p = 0.64.Fig. 2 a Comparison of time to respiratory event and b drug continuation between JAKi and rituximab for the treatment of rheumatoid arthritis in patients with concurrent bronchiectasis and/or ILD
Drug continuation
Seventy-one percent (20/28) of patients with ILD and/or bronchiectasis who commenced JAKi for the treatment of rheumatoid arthritis remained on the drug at the end of the follow-up period. Reasons for discontinuation included inefficacy (n = 3), death (n = 2, as outlined above), recurrent infections (n = 2) and reduced renal function (n = 1).
Sixty-three percent (12/19) of patients who commenced rituximab continued the drug to the end of the follow-up period. Reasons for cessation included inefficacy (n = 5) or death (n = 2, as outlined above). The rate of discontinuation of JAKi in patients with ILD and/or bronchiectasis did not significantly differ from that of those receiving rituximab (Fig. 2b); unadjusted hazard ratio: 1.9 (95% CI 0.634, 5.73); p = 0.251.
Discussion
There is an absence of patients with interstitial lung disease and bronchiectasis in randomised controlled trials that examine the effectiveness of JAK inhibitors for the treatment of rheumatoid arthritis. Therefore, we have limited knowledge of the pulmonary safety in these cohorts. Prevalence of interstitial lung disease and bronchiectasis is relatively high in patients with rheumatoid arthritis, estimated to affect between 5 and 10% of patients [1, 13]. Interstitial lung disease itself contributes significantly to the excess mortality evident in patients with rheumatoid arthritis [14]. Despite this, the optimal treatment strategy for such patients remains unknown, reflecting a paucity of randomized controlled clinical trials for this patient sub-group. Recently, observational data have suggested treatment with rituximab [6, 15, 16], and laterally abatacept [5, 17] is the most effective and safest biologic agents for the treatment of RA with concurrent ILD or bronchiectasis. Data suggest that treatment with rituximab and abatacept may even infer a benefit to pulmonary function [5, 6]. Nonetheless, concern about lung toxicity restricts the use of many DMARDs and biologic agents, be these concerns soundly proven or not.
Novel, small-molecule treatments, such as Janus kinase inhibitors, represent an exciting option for patients with rheumatoid arthritis. Their safety and potential for effectiveness in patients with concurrent pulmonary disease should be duly considered. Currently, our understanding of inhibition of JAK/STAT signalling pathways in patients with ILD and bronchiectasis is extremely limited. JAKi-use has led to successful treatment of dermatomyositis-associated ILD but evidence is restricted to a small number of case reports and case series [18–20]. Even less is understood about Janus kinase inhibition in patients with bronchiectasis. While post hoc analysis of phase 2, 3 and 3b/4 clinical trials indicates that the de novo incidence of interstitial lung disease is low in patients with rheumatoid arthritis treated with JAKi [21]; to our knowledge, no study has yet examined the pulmonary tolerance and safety of JAKi in those with existing ILD or bronchiectasis. The current study sought to address this important clinical scenario, by comparing time to respiratory events in patients receiving JAKi with patients receiving the current, best-practice medication for this clinical cohort, rituximab.
The majority of our patients commenced JAK1/2 inhibition with baricitinib due to the earlier licensing of this medication for the treatment of rheumatoid arthritis but several patients commenced JAK1/3 inhibition with tofacitinib. Furthermore, the proportion of patients with ‘difficult-to-treat’ rheumatoid disease, manifest by a longer duration of illness and a greater number of previous biologic therapies (e.g. tocilizumab, abatacept), was higher in those subsequently prescribed JAKi. We believe this relates to the fact that a novel medication provides an attractive option for patients who have had difficult-to-treat disease for many years and in such cases patients often chose to switch therapy in an attempt to achieve remission.
Encouragingly, serious respiratory events (i.e., hospitalization or death due to a respiratory cause) were no different between those treated with JAKi and those treated with rituximab suggesting that JAKi may offer a realistic treatment option for patients with RA and concurrent ILD or bronchiectasis. Our cohort of patients represents a group with mild to moderate pulmonary disease, all of whom commenced Janus kinase inhibition or rituximab with a known, pre-existing diagnosis of ILD or bronchiectasis (present on cross sectional CT-imaging and/or attending a tertiary ILD clinic). Very few patients with ILD had a HR-CT pattern consistent with definite UIP and most had a pattern that was consistent with an alternative diagnosis and were highly unlikely to be UIP. Therefore, our findings should not be extrapolated to those with severe ILD or bronchiectasis or a definite UIP pattern of disease and caution is urged in this setting. Furthermore, there was no significant difference in time to drug discontinuation between those prescribed JAKi and those prescribed rituximab. Proportionally, more patients treated with rituximab (almost 40%) ceased treatment during follow-up compared with those treated with JAKi (almost 30%). Reasons for discontinuation of rituximab predominantly related to clinical inefficacy, but comparison of clinical effectiveness was not the aim of this study.
There are several strengths to this study. Both cohorts of patients (i.e., JAKi- and rituximab-receiving) were similar in terms of clinical traits (e.g. gender, concurrent medications, corticosteroid use, anti-CCP and rheumatoid factor seropositivity) minimizing the effect of confounders on respiratory event outcomes. In addition, these data represent real-world experience of patients with RA and concurrent ILD/bronchiectasis with patient–clinician decisions made without influence of a pre-determined research goal. The retrospective nature of this observational study and associated risk of information bias represents the major limitation of this study. The relatively small sample size and short follow-up period also represent limitations that may have affected overall results. Ideally, follow-up pulmonary function testing would have been available for all patients after commencing treatment but as this was only the case for a minority of patients, this was not analysed. It was not possible to collect primary care records of respiratory exacerbations or infections that did not require hospital admission that were treated in the community. Furthermore, the electronic patient health record system used did not permit recording of admissions to hospital abroad or to hospitals outside of NHS Lothian. The authors recommend that the findings of this study and analysis should be investigated further in larger prospective cohorts in the future.
To conclude, within the limitations of this retrospective analysis, JAK inhibition with baricitinib or tofacitinib for the treatment of rheumatoid arthritis in patients with concurrent interstitial lung disease or bronchiectasis did not increase the rate of hospitalization or death due to a respiratory cause in comparison to those treated with rituximab. JAK inhibition may provide a favourable and safe therapeutic option for such patients but larger, prospective studies are required to confirm these findings to inform much-needed therapeutic guidance in this sub-category of patients with RA and ILD/bronchiectasis.
Funding
No specific funding was received from any bodies in the public, commercial or not-for-profit sectors to carry out the work described in this article.
Declarations
Conflict of interest
OC has received sponsorship from Lilly and Pfizer to attend educational conferences. The other authors have no conflicts of interest to declare.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. | METHYLPREDNISOLONE, RITUXIMAB | DrugsGivenReaction | CC BY | 33721094 | 19,059,614 | 2021-05 |
What was the administration route of drug 'METHYLPREDNISOLONE SODIUM SUCCINATE'? | A retrospective comparison of respiratory events with JAK inhibitors or rituximab for rheumatoid arthritis in patients with pulmonary disease.
Janus kinase inhibitors (JAKi) are an exciting option for the treatment of rheumatoid arthritis (RA) but little is known about their safety and tolerability in patients with existing respiratory disorders. The objective was to compare pulmonary safety of JAKi versus rituximab in patients with concurrent interstitial lung disease (ILD) or bronchiectasis. We performed a retrospective electronic patient record review of patients with known ILD or bronchiectasis commencing JAKi or rituximab for the treatment of RA. Patients initiating treatment from January 2016 to February 2020 were included. Respiratory events (hospitalization or death from a respiratory cause) were compared using Kaplan-Meier survival analysis. We analysed patients who received JAKi (n = 28) and rituximab (n = 19) for a mean (SD) of 1.1 (0.62) and 2.14 (1) years respectively. Patients were predominantly female (68%), anti-CCP antibody positive (94%) and non-smoking (89%) with a median (IQR) percentage predicted FVC at baseline of 100% (82-115%) and percentage predicted TLCO of 62% (54.5-68%). Respiratory events occurred in five patients treated with JAKi (18%; 5 hospitalizations, 2 deaths) and in four patients treated with rituximab (21%; 3 hospitalizations, 1 death). Respiratory event rates did not differ between groups (Cox-regression proportional hazard ratio = 1.38, 95% CI 0.36-5.28; p = 0.64). In this retrospective study, JAKi for the treatment of RA with existing ILD or bronchiectasis did not increase the rate of hospitalization or death due to respiratory causes compared to those treated with rituximab. JAK inhibition may provide a relatively safe option for RA in such patients.
Introduction
Therapeutic decision-making in the treatment of rheumatoid arthritis (RA) for patients with concurrent pulmonary disease can be challenging. Fear regarding increased infection risk in bronchiectasis and progression of interstitial lung disease (ILD) secondary to drug administration restricts treatment options. Controversy exists as to whether medications, such as methotrexate, leflunomide and tumour necrosis factor-α inhibitors (TNFi), exacerbate pulmonary disease in patients with concurrent ILD and bronchiectasis [1]. This has led to hesitancy in utilizing these medications although systematic review of the evidence is beginning to allay fears [2, 3]. Recently, data have begun to support the safe and effective use of B and T cell therapies, rituximab [4] and abatacept [5], in the management of rheumatoid arthritis-related ILD and bronchiectasis [6, 7]. However, therapeutic options remain limited.
The approval of Janus kinase inhibitors (JAKi) (i.e., baricitinib, tofacitinib, upadacitinib) in the management of moderate to severe RA offers a new and exciting treatment paradigm [8–10]. However, inclusion of patients with concurrent ILD or bronchiectasis in relevant clinical trials is extremely limited. While the use of JAKi for the treatment of RA with co-existent lung disease is tempting due to the oral route of administration and relatively short half-life, their safety in relation to pulmonary toxicity in patients with established ILD or bronchiectasis is unknown.
To investigate this evidence gap, the objective of the current study is to compare the pulmonary safety of Janus kinase inhibition for the treatment of rheumatoid arthritis in patients with existing ILD or bronchiectasis, with rituximab. The primary outcome of interest was the incidence of severe respiratory events (hospitalization or death due to respiratory cause) in those receiving JAKi versus those receiving rituximab.
Materials and methods
Study design and patient inclusion
The study comprised a retrospective observational study. The electronic patient health record system (TrakCare®) was used to collect relevant data. Adult patients (aged 18 or over) undergoing treatment for rheumatoid arthritis at the Rheumatic Diseases Unit, Western General Hospital, Edinburgh, U.K., were included in the study. Only patients with an established diagnosis of ILD or bronchiectasis before the commencement of the drug of interest were eligible for inclusion. Patients without these respiratory conditions were excluded. Dedicated pharmacy and biologic clinic records were used to identify all adult patients (16 years or older) commencing any JAKi or rituximab for the treatment of RA from January 2016 to February 2020 (Fig. 1). From these records, all patients’ health records were screened by the investigators (OC, LK) for co-existing diagnoses of interstitial lung disease and/or bronchiectasis. For relevant patients, data were collected in detail from electronic health records by the investigators (OC, OM). Data were collected until discontinuation of the drug of interest (as documented in the clinical record) or to the end of the follow-up period (1 August 2020). They study was conducted in accordance with the Declaration of Helsinki. As all data were collected as part of routine clinical care and all treatment decisions were made prior to conduction of the study, ethical approval was not required.Fig. 1 Study outline, selection of patients and follow-up
Clinical features and demographics
Identical clinical and demographic data were collected for both treatment groups. Standard clinical variables, such as age, gender, disease duration (rheumatoid arthritis and respiratory diseases), smoking history, past and current medications, anti-CCP antibody and rheumatoid factor (IgM) positivity, were collected along with the dates of commencement and cessation for the drugs of interest. Rheumatoid arthritis disease activity (DAS28-ESR or DAS28-CRP) before the commencement of the treatment of interest was recorded along with activity scores at 3–4 months post initiation (where available).
Baseline pulmonary function tests prior to the commencement of the medications of interest were also assessed. As only a small proportion of patients had follow-up pulmonary function assessment within a reasonable time frame, this was not analysed. The pattern of ILD involvement was described as recommended by the American Thoracic Society/European Respiratory Society following review of high-resolution (HR) CT-thorax imaging by a multi-disciplinary team including chest radiologists and an expert ILD lead physician (NH) who additionally attributed a modified semi-quantitative CT-fibrosis score by visual inspection [11, 12].
Outcome measures
The primary outcome of interest was time to first respiratory event (days). Respiratory events were defined as either admission to hospital with a respiratory illness (e.g., infection, ILD exacerbation), or death from a respiratory cause while taking the medication of interest.
Additional outcomes of interest included drug continuation. Drug continuation was defined for JAKi as from the date of first prescription to the date of drug cessation, death, end of follow-up period, or whichever end-point came first. Drug continuation for rituximab was defined as from the date of first infusion to the date of cessation, death, end of follow-up period or whichever end-point came first. The number of cycles, frequency and dose of rituximab received was recorded for each patient.
Statistical analysis
Statistical analysis was performed using IBM SPSS statistics v.25.0 (IBM, Armonk, New York, USA). Distributions of clinical variables and demographics were illustrated using histograms. Depending on the data distribution and type (i.e., categorical, continuous), variables were then described using means (with standard deviations) or medians (with inter-quartile ranges). Continuous variables were compared for differences between groups using either student’s t tests or Mann–Whitney U testing subject to normality or non-normality of distribution. Proportions of categorical variables were compared using Chi-squared testing. Kaplan–Meier survival analysis (log-rank) was used to compare respiratory event survival and drug continuation survival between treatment groups. Cox regression proportional hazard was used to calculate unadjusted hazard ratios with 95% confidence intervals (CI 95%). Statistical significance was set at p < 0.05 and analysis was two -tailed.
Results
Demographic and clinical characteristics
A total of 291 patients were commenced on JAKi to February 2020 at the Rheumatic Diseases Unit, Edinburgh (Fig. 1). Twenty-eight patients commenced JAK inhibition for the treatment of rheumatoid arthritis who also had concurrent interstitial lung disease (67.9%), bronchiectasis (25%) or both (7.1%). Baricitinib was commenced in 26 patients and tofacitinib in 2 patients. A total of 318 patients were commenced on rituximab from January 2016 to February 2020 (Fig. 1). Nineteen patients commenced rituximab for the treatment of RA and had concurrent ILD (68.4%), bronchiectasis (26.3%) or both (5.3%).
Mean duration of follow-up (to discontinuation of drug, death or end of follow-up period) for patients receiving JAKi was 1.1 years (SD = 0.62) and for patients receiving rituximab infusions was 2.14 years (SD = 1). Demographic and clinical characteristics were similar in both treatment groups (Table 1); however, those commencing JAKi had a significantly longer disease duration (median 11 years) compared to those starting rituximab (median 3 years) (Table 1; Mann–Whitney U test, p = 0.003).Table 1 Demographic and clinical details of included patients
JAKi (n = 28) Rituximab (n = 19) No. of patients with available data (n =)
ILD/Bronchiectasis/both
n, (%)
ILD n = 19 (67.9%)
Bronchiectasis n = 7 (25%)
Both n = 2 (7.1%)
ILD n = 13 (68.4%)
Bronchiectasis n = 5 (26.3%)
Both n = 1 (5.3%)
47
Age (median, IQR) 69 (62.3–75) 70 (59–76) 47
Female n, (%) 18 (64.3%) 14 (73.7%) 47
Smoking status
n, (%)
Smoker:
n = 5 (17.9%)
Ex-smoker:
n = 15 (53.6%)
Never-smoker:
n = 8 (28.6%)
Smoker:
n = 0 (0%),
Ex-smoker:
n = 12 (63.2%)
Never-smoker:
n = 7 (36.8%)
47
Anti-CCP positivity
n, (%)
25 (96.2%) 19 (100%) 45
Rh-Factor IgM positivity
n, (%)
14 (77.8%) 9 (90%) 28
RA diagnosis age median, (IQR) 53 (44–61) 54 (48–67) 42
Respiratory diagnosis age
median, (IQR)
64 (55–73) 64 (53–71) 46
Years from RA diagnosis to drug of interest
median, (IQR)
11 (3–20) 3 (1–6) 42
DAS-28 Baseline
median, (IQR)
5.85 (4.98–6.18) 6.2 (5.53–6.86) 43
DAS-28 3–4 months
median, (IQR)
3.4 (2.38–4.11) 4.2 (3.61–5.44) 32
JAKi Janus kinase inhibitor, ILD interstitial lung disease, Rh-factor rheumatoid factor, RA rheumatoid arthritis, DAS-28 disease activity score 28 joints
Pre-treatment pulmonary function tests indicated mild severity of pulmonary disease in patients commencing JAKi and rituximab (Table 2). Review of patients’ baseline high-resolution CT-thorax revealed similar radiological patterns and severity of ILD in both treatment groups (Table 2). However, a greater degree of ground glass was evident on CT-imaging before treatment in those receiving JAKi compared to those receiving rituximab (p = 0.036).Table 2 Baseline pulmonary function (all patients); and radiological pattern and severity of ILD on CT-thorax imaging (patients with ILD)
JAKi (n = 28) Rituximab (n = 19) No. of patients with available data (n =)
Baseline predicted FEV1 median %, (IQR) 93.5 (74.5–105) 92 (76–105) 43
Baseline predicted FVC median %, (IQR) 103 (78–112) 99 (86–119) 44
Baseline predicted TLCO median %, (IQR) 62 (51.25–69) 62 (55–67) 41
Radiological ILD-pattern JAKi (n = 20) Rituximab (n = 14)
UIP
n, (%)
0 (0%) 2 (14%)
Probable UIP
n, (%)
4 (20%) 5 (36%)
Fibrotic NSIP
n, (%)
6 (30%) 1 (7%)
Chronic HP
n, (%)
3 (15%) 1 (7%)
Respiratory bronchiolitis-ILD
n, (%)
1 (5%) 0 (0%)
Unclassifiable
n, (%)
6 (30%) 5 (36%)
% of lung with ILD
median, (IQR)
20 (10–45) 12.5 (10–30)
% of lung with ground glass appearance
median, (IQR)
20 (5–45) 7.5 (5–10)
% of lung with honeycombing
median, (IQR)
0 (0–1.75) 0 (0–2)
% of lung with emphysema
median, (IQR)
0.5 (0–8.75) 0 (0–3.75)
JAKi Janus kinase inhibitor, ILD interstitial lung disease, FEV1 forced expiratory volume in 1 second, FVC forced vital capacity, TLco lung transfer capacity for carbon monoxide, IQR inter-quartile range, UIP usual interstitial pneumonia, NSIP non-specific interstitial pneumonia, HP hypersensitivity pneumonitis
Treatment details
Concurrent treatments including conventional synthetic DMARDs and corticosteroids were similar in both groups (Table 3). Almost 30% of patients were biologic naïve before commencing the drug of interest. Prior treatment history indicated that those commencing JAKi had previously been treated with a greater number of biologics including TNFi, tocilizumab and abatacept, compared to those commencing rituximab.Table 3 Previous and concurrent DMARD treatments
JAKi (n = 28) Rituximab (n = 19)
Previous treatments
No. of previous csDMARDs
median, (IQR)
3 (2–4) 3 (2–4)
Methotrexate
n, (%)
22 (78.6%) 13 (68.4%)
Biologic Naïve
n, (%)
8 (28.6%) 6 (31.6%)
No. of previous biologics
median, (IQR)
1 (0–2) 0 (0–1)
Rituximab/JAKi
n, (%)
12 (42.9%) 1 (5.3%)
TNF-α inhibitor
n, (%)
15 (53.6%) 5 (26.3%)
Tocilizumab
n, (%)
6 (21.4%) 1 (5.3%)
Abatacept
n, (%)
6 (21.4%) 1 (5.3%)
Concurrent treatments
Prednisolone; n, (%) 13 (46.4%) 6 (31.6%)
Prednisolone dose (mg) median, (IQR) 7 (5–10) 10 (4.5–12.5)
Methotrexate
n, (%)
4 (14.3%) 2 (10.5%)
Azathioprine
n, (%)
2 (7.1%) 2 (10.5%)
Leflunomide
n, (%)
2 (7.1%) 2 (10.5%)
Hydroxychloroquine
n, (%)
6 (21.4%) 6 (31.6%)
Sulfasalazine
n, (%)
3 (10.7%) 4 (21.1%)
DMARD disease-modifying anti-rheumatic drugs, csDMARDs conventional synthetic DMARDs, JAKi Janus kinase inhibitor, IQR inter-quartile range, TNF Tumour necrosis factor
All except for two patients received standard licensed dosing regimens for JAK inhibition for the treatment of RA (i.e. baricitinib 4 mg once daily; tofacitinib 5 mg twice daily). One patient took 4 mg/2 mg of baricitinib on alternate days and one patient took 5 mg of tofacitinib once daily before increasing to 5 mg twice daily.
All patients who received rituximab received the standard dosing of treatment (2 × intravenous infusions of 1000 mg rituximab with 100 mg intravenous methylprednisolone 14 days apart). Only 2/19 patients were on ‘fixed’ cycles of rituximab therapy repeated every 6 months. All others received rituximab as required depending on clinical status but not within 6 months of the last infusion. 5/19 patients received only 1 cycle of rituximab during their study follow-up period. All others (74%) received between 2 and 5 cycles of rituximab.
Incidence of respiratory events
Respiratory events (i.e. hospitalization or death due to a respiratory cause) occurred in five patients (18%) during JAKi treatment (7 hospitalizations, two of which led to death) and in four patients (21%) (4 hospitalizations, one of which led to death) during treatment with rituximab. In the JAKi group, two patients (7.1%) died of respiratory issues (one new lung cancer with concomitant community acquired pneumonia and one of community acquired pneumonia). In the rituximab-treated patients, two patients (10.5%) died during follow-up; one of an acute myocardial infarction and one of pneumonitis and respiratory failure. No episodes of acute pneumonitis occurred in the JAKi group. There was no difference in respiratory event survival in the JAKi-treated group compared with the rituximab-treated group (Fig. 2a); unadjusted hazard ratio: 1.38 (95% CI 0.36, 5.28); p = 0.64.Fig. 2 a Comparison of time to respiratory event and b drug continuation between JAKi and rituximab for the treatment of rheumatoid arthritis in patients with concurrent bronchiectasis and/or ILD
Drug continuation
Seventy-one percent (20/28) of patients with ILD and/or bronchiectasis who commenced JAKi for the treatment of rheumatoid arthritis remained on the drug at the end of the follow-up period. Reasons for discontinuation included inefficacy (n = 3), death (n = 2, as outlined above), recurrent infections (n = 2) and reduced renal function (n = 1).
Sixty-three percent (12/19) of patients who commenced rituximab continued the drug to the end of the follow-up period. Reasons for cessation included inefficacy (n = 5) or death (n = 2, as outlined above). The rate of discontinuation of JAKi in patients with ILD and/or bronchiectasis did not significantly differ from that of those receiving rituximab (Fig. 2b); unadjusted hazard ratio: 1.9 (95% CI 0.634, 5.73); p = 0.251.
Discussion
There is an absence of patients with interstitial lung disease and bronchiectasis in randomised controlled trials that examine the effectiveness of JAK inhibitors for the treatment of rheumatoid arthritis. Therefore, we have limited knowledge of the pulmonary safety in these cohorts. Prevalence of interstitial lung disease and bronchiectasis is relatively high in patients with rheumatoid arthritis, estimated to affect between 5 and 10% of patients [1, 13]. Interstitial lung disease itself contributes significantly to the excess mortality evident in patients with rheumatoid arthritis [14]. Despite this, the optimal treatment strategy for such patients remains unknown, reflecting a paucity of randomized controlled clinical trials for this patient sub-group. Recently, observational data have suggested treatment with rituximab [6, 15, 16], and laterally abatacept [5, 17] is the most effective and safest biologic agents for the treatment of RA with concurrent ILD or bronchiectasis. Data suggest that treatment with rituximab and abatacept may even infer a benefit to pulmonary function [5, 6]. Nonetheless, concern about lung toxicity restricts the use of many DMARDs and biologic agents, be these concerns soundly proven or not.
Novel, small-molecule treatments, such as Janus kinase inhibitors, represent an exciting option for patients with rheumatoid arthritis. Their safety and potential for effectiveness in patients with concurrent pulmonary disease should be duly considered. Currently, our understanding of inhibition of JAK/STAT signalling pathways in patients with ILD and bronchiectasis is extremely limited. JAKi-use has led to successful treatment of dermatomyositis-associated ILD but evidence is restricted to a small number of case reports and case series [18–20]. Even less is understood about Janus kinase inhibition in patients with bronchiectasis. While post hoc analysis of phase 2, 3 and 3b/4 clinical trials indicates that the de novo incidence of interstitial lung disease is low in patients with rheumatoid arthritis treated with JAKi [21]; to our knowledge, no study has yet examined the pulmonary tolerance and safety of JAKi in those with existing ILD or bronchiectasis. The current study sought to address this important clinical scenario, by comparing time to respiratory events in patients receiving JAKi with patients receiving the current, best-practice medication for this clinical cohort, rituximab.
The majority of our patients commenced JAK1/2 inhibition with baricitinib due to the earlier licensing of this medication for the treatment of rheumatoid arthritis but several patients commenced JAK1/3 inhibition with tofacitinib. Furthermore, the proportion of patients with ‘difficult-to-treat’ rheumatoid disease, manifest by a longer duration of illness and a greater number of previous biologic therapies (e.g. tocilizumab, abatacept), was higher in those subsequently prescribed JAKi. We believe this relates to the fact that a novel medication provides an attractive option for patients who have had difficult-to-treat disease for many years and in such cases patients often chose to switch therapy in an attempt to achieve remission.
Encouragingly, serious respiratory events (i.e., hospitalization or death due to a respiratory cause) were no different between those treated with JAKi and those treated with rituximab suggesting that JAKi may offer a realistic treatment option for patients with RA and concurrent ILD or bronchiectasis. Our cohort of patients represents a group with mild to moderate pulmonary disease, all of whom commenced Janus kinase inhibition or rituximab with a known, pre-existing diagnosis of ILD or bronchiectasis (present on cross sectional CT-imaging and/or attending a tertiary ILD clinic). Very few patients with ILD had a HR-CT pattern consistent with definite UIP and most had a pattern that was consistent with an alternative diagnosis and were highly unlikely to be UIP. Therefore, our findings should not be extrapolated to those with severe ILD or bronchiectasis or a definite UIP pattern of disease and caution is urged in this setting. Furthermore, there was no significant difference in time to drug discontinuation between those prescribed JAKi and those prescribed rituximab. Proportionally, more patients treated with rituximab (almost 40%) ceased treatment during follow-up compared with those treated with JAKi (almost 30%). Reasons for discontinuation of rituximab predominantly related to clinical inefficacy, but comparison of clinical effectiveness was not the aim of this study.
There are several strengths to this study. Both cohorts of patients (i.e., JAKi- and rituximab-receiving) were similar in terms of clinical traits (e.g. gender, concurrent medications, corticosteroid use, anti-CCP and rheumatoid factor seropositivity) minimizing the effect of confounders on respiratory event outcomes. In addition, these data represent real-world experience of patients with RA and concurrent ILD/bronchiectasis with patient–clinician decisions made without influence of a pre-determined research goal. The retrospective nature of this observational study and associated risk of information bias represents the major limitation of this study. The relatively small sample size and short follow-up period also represent limitations that may have affected overall results. Ideally, follow-up pulmonary function testing would have been available for all patients after commencing treatment but as this was only the case for a minority of patients, this was not analysed. It was not possible to collect primary care records of respiratory exacerbations or infections that did not require hospital admission that were treated in the community. Furthermore, the electronic patient health record system used did not permit recording of admissions to hospital abroad or to hospitals outside of NHS Lothian. The authors recommend that the findings of this study and analysis should be investigated further in larger prospective cohorts in the future.
To conclude, within the limitations of this retrospective analysis, JAK inhibition with baricitinib or tofacitinib for the treatment of rheumatoid arthritis in patients with concurrent interstitial lung disease or bronchiectasis did not increase the rate of hospitalization or death due to a respiratory cause in comparison to those treated with rituximab. JAK inhibition may provide a favourable and safe therapeutic option for such patients but larger, prospective studies are required to confirm these findings to inform much-needed therapeutic guidance in this sub-category of patients with RA and ILD/bronchiectasis.
Funding
No specific funding was received from any bodies in the public, commercial or not-for-profit sectors to carry out the work described in this article.
Declarations
Conflict of interest
OC has received sponsorship from Lilly and Pfizer to attend educational conferences. The other authors have no conflicts of interest to declare.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. | Intravenous (not otherwise specified) | DrugAdministrationRoute | CC BY | 33721094 | 19,082,833 | 2021-05 |
What was the administration route of drug 'METHYLPREDNISOLONE'? | A retrospective comparison of respiratory events with JAK inhibitors or rituximab for rheumatoid arthritis in patients with pulmonary disease.
Janus kinase inhibitors (JAKi) are an exciting option for the treatment of rheumatoid arthritis (RA) but little is known about their safety and tolerability in patients with existing respiratory disorders. The objective was to compare pulmonary safety of JAKi versus rituximab in patients with concurrent interstitial lung disease (ILD) or bronchiectasis. We performed a retrospective electronic patient record review of patients with known ILD or bronchiectasis commencing JAKi or rituximab for the treatment of RA. Patients initiating treatment from January 2016 to February 2020 were included. Respiratory events (hospitalization or death from a respiratory cause) were compared using Kaplan-Meier survival analysis. We analysed patients who received JAKi (n = 28) and rituximab (n = 19) for a mean (SD) of 1.1 (0.62) and 2.14 (1) years respectively. Patients were predominantly female (68%), anti-CCP antibody positive (94%) and non-smoking (89%) with a median (IQR) percentage predicted FVC at baseline of 100% (82-115%) and percentage predicted TLCO of 62% (54.5-68%). Respiratory events occurred in five patients treated with JAKi (18%; 5 hospitalizations, 2 deaths) and in four patients treated with rituximab (21%; 3 hospitalizations, 1 death). Respiratory event rates did not differ between groups (Cox-regression proportional hazard ratio = 1.38, 95% CI 0.36-5.28; p = 0.64). In this retrospective study, JAKi for the treatment of RA with existing ILD or bronchiectasis did not increase the rate of hospitalization or death due to respiratory causes compared to those treated with rituximab. JAK inhibition may provide a relatively safe option for RA in such patients.
Introduction
Therapeutic decision-making in the treatment of rheumatoid arthritis (RA) for patients with concurrent pulmonary disease can be challenging. Fear regarding increased infection risk in bronchiectasis and progression of interstitial lung disease (ILD) secondary to drug administration restricts treatment options. Controversy exists as to whether medications, such as methotrexate, leflunomide and tumour necrosis factor-α inhibitors (TNFi), exacerbate pulmonary disease in patients with concurrent ILD and bronchiectasis [1]. This has led to hesitancy in utilizing these medications although systematic review of the evidence is beginning to allay fears [2, 3]. Recently, data have begun to support the safe and effective use of B and T cell therapies, rituximab [4] and abatacept [5], in the management of rheumatoid arthritis-related ILD and bronchiectasis [6, 7]. However, therapeutic options remain limited.
The approval of Janus kinase inhibitors (JAKi) (i.e., baricitinib, tofacitinib, upadacitinib) in the management of moderate to severe RA offers a new and exciting treatment paradigm [8–10]. However, inclusion of patients with concurrent ILD or bronchiectasis in relevant clinical trials is extremely limited. While the use of JAKi for the treatment of RA with co-existent lung disease is tempting due to the oral route of administration and relatively short half-life, their safety in relation to pulmonary toxicity in patients with established ILD or bronchiectasis is unknown.
To investigate this evidence gap, the objective of the current study is to compare the pulmonary safety of Janus kinase inhibition for the treatment of rheumatoid arthritis in patients with existing ILD or bronchiectasis, with rituximab. The primary outcome of interest was the incidence of severe respiratory events (hospitalization or death due to respiratory cause) in those receiving JAKi versus those receiving rituximab.
Materials and methods
Study design and patient inclusion
The study comprised a retrospective observational study. The electronic patient health record system (TrakCare®) was used to collect relevant data. Adult patients (aged 18 or over) undergoing treatment for rheumatoid arthritis at the Rheumatic Diseases Unit, Western General Hospital, Edinburgh, U.K., were included in the study. Only patients with an established diagnosis of ILD or bronchiectasis before the commencement of the drug of interest were eligible for inclusion. Patients without these respiratory conditions were excluded. Dedicated pharmacy and biologic clinic records were used to identify all adult patients (16 years or older) commencing any JAKi or rituximab for the treatment of RA from January 2016 to February 2020 (Fig. 1). From these records, all patients’ health records were screened by the investigators (OC, LK) for co-existing diagnoses of interstitial lung disease and/or bronchiectasis. For relevant patients, data were collected in detail from electronic health records by the investigators (OC, OM). Data were collected until discontinuation of the drug of interest (as documented in the clinical record) or to the end of the follow-up period (1 August 2020). They study was conducted in accordance with the Declaration of Helsinki. As all data were collected as part of routine clinical care and all treatment decisions were made prior to conduction of the study, ethical approval was not required.Fig. 1 Study outline, selection of patients and follow-up
Clinical features and demographics
Identical clinical and demographic data were collected for both treatment groups. Standard clinical variables, such as age, gender, disease duration (rheumatoid arthritis and respiratory diseases), smoking history, past and current medications, anti-CCP antibody and rheumatoid factor (IgM) positivity, were collected along with the dates of commencement and cessation for the drugs of interest. Rheumatoid arthritis disease activity (DAS28-ESR or DAS28-CRP) before the commencement of the treatment of interest was recorded along with activity scores at 3–4 months post initiation (where available).
Baseline pulmonary function tests prior to the commencement of the medications of interest were also assessed. As only a small proportion of patients had follow-up pulmonary function assessment within a reasonable time frame, this was not analysed. The pattern of ILD involvement was described as recommended by the American Thoracic Society/European Respiratory Society following review of high-resolution (HR) CT-thorax imaging by a multi-disciplinary team including chest radiologists and an expert ILD lead physician (NH) who additionally attributed a modified semi-quantitative CT-fibrosis score by visual inspection [11, 12].
Outcome measures
The primary outcome of interest was time to first respiratory event (days). Respiratory events were defined as either admission to hospital with a respiratory illness (e.g., infection, ILD exacerbation), or death from a respiratory cause while taking the medication of interest.
Additional outcomes of interest included drug continuation. Drug continuation was defined for JAKi as from the date of first prescription to the date of drug cessation, death, end of follow-up period, or whichever end-point came first. Drug continuation for rituximab was defined as from the date of first infusion to the date of cessation, death, end of follow-up period or whichever end-point came first. The number of cycles, frequency and dose of rituximab received was recorded for each patient.
Statistical analysis
Statistical analysis was performed using IBM SPSS statistics v.25.0 (IBM, Armonk, New York, USA). Distributions of clinical variables and demographics were illustrated using histograms. Depending on the data distribution and type (i.e., categorical, continuous), variables were then described using means (with standard deviations) or medians (with inter-quartile ranges). Continuous variables were compared for differences between groups using either student’s t tests or Mann–Whitney U testing subject to normality or non-normality of distribution. Proportions of categorical variables were compared using Chi-squared testing. Kaplan–Meier survival analysis (log-rank) was used to compare respiratory event survival and drug continuation survival between treatment groups. Cox regression proportional hazard was used to calculate unadjusted hazard ratios with 95% confidence intervals (CI 95%). Statistical significance was set at p < 0.05 and analysis was two -tailed.
Results
Demographic and clinical characteristics
A total of 291 patients were commenced on JAKi to February 2020 at the Rheumatic Diseases Unit, Edinburgh (Fig. 1). Twenty-eight patients commenced JAK inhibition for the treatment of rheumatoid arthritis who also had concurrent interstitial lung disease (67.9%), bronchiectasis (25%) or both (7.1%). Baricitinib was commenced in 26 patients and tofacitinib in 2 patients. A total of 318 patients were commenced on rituximab from January 2016 to February 2020 (Fig. 1). Nineteen patients commenced rituximab for the treatment of RA and had concurrent ILD (68.4%), bronchiectasis (26.3%) or both (5.3%).
Mean duration of follow-up (to discontinuation of drug, death or end of follow-up period) for patients receiving JAKi was 1.1 years (SD = 0.62) and for patients receiving rituximab infusions was 2.14 years (SD = 1). Demographic and clinical characteristics were similar in both treatment groups (Table 1); however, those commencing JAKi had a significantly longer disease duration (median 11 years) compared to those starting rituximab (median 3 years) (Table 1; Mann–Whitney U test, p = 0.003).Table 1 Demographic and clinical details of included patients
JAKi (n = 28) Rituximab (n = 19) No. of patients with available data (n =)
ILD/Bronchiectasis/both
n, (%)
ILD n = 19 (67.9%)
Bronchiectasis n = 7 (25%)
Both n = 2 (7.1%)
ILD n = 13 (68.4%)
Bronchiectasis n = 5 (26.3%)
Both n = 1 (5.3%)
47
Age (median, IQR) 69 (62.3–75) 70 (59–76) 47
Female n, (%) 18 (64.3%) 14 (73.7%) 47
Smoking status
n, (%)
Smoker:
n = 5 (17.9%)
Ex-smoker:
n = 15 (53.6%)
Never-smoker:
n = 8 (28.6%)
Smoker:
n = 0 (0%),
Ex-smoker:
n = 12 (63.2%)
Never-smoker:
n = 7 (36.8%)
47
Anti-CCP positivity
n, (%)
25 (96.2%) 19 (100%) 45
Rh-Factor IgM positivity
n, (%)
14 (77.8%) 9 (90%) 28
RA diagnosis age median, (IQR) 53 (44–61) 54 (48–67) 42
Respiratory diagnosis age
median, (IQR)
64 (55–73) 64 (53–71) 46
Years from RA diagnosis to drug of interest
median, (IQR)
11 (3–20) 3 (1–6) 42
DAS-28 Baseline
median, (IQR)
5.85 (4.98–6.18) 6.2 (5.53–6.86) 43
DAS-28 3–4 months
median, (IQR)
3.4 (2.38–4.11) 4.2 (3.61–5.44) 32
JAKi Janus kinase inhibitor, ILD interstitial lung disease, Rh-factor rheumatoid factor, RA rheumatoid arthritis, DAS-28 disease activity score 28 joints
Pre-treatment pulmonary function tests indicated mild severity of pulmonary disease in patients commencing JAKi and rituximab (Table 2). Review of patients’ baseline high-resolution CT-thorax revealed similar radiological patterns and severity of ILD in both treatment groups (Table 2). However, a greater degree of ground glass was evident on CT-imaging before treatment in those receiving JAKi compared to those receiving rituximab (p = 0.036).Table 2 Baseline pulmonary function (all patients); and radiological pattern and severity of ILD on CT-thorax imaging (patients with ILD)
JAKi (n = 28) Rituximab (n = 19) No. of patients with available data (n =)
Baseline predicted FEV1 median %, (IQR) 93.5 (74.5–105) 92 (76–105) 43
Baseline predicted FVC median %, (IQR) 103 (78–112) 99 (86–119) 44
Baseline predicted TLCO median %, (IQR) 62 (51.25–69) 62 (55–67) 41
Radiological ILD-pattern JAKi (n = 20) Rituximab (n = 14)
UIP
n, (%)
0 (0%) 2 (14%)
Probable UIP
n, (%)
4 (20%) 5 (36%)
Fibrotic NSIP
n, (%)
6 (30%) 1 (7%)
Chronic HP
n, (%)
3 (15%) 1 (7%)
Respiratory bronchiolitis-ILD
n, (%)
1 (5%) 0 (0%)
Unclassifiable
n, (%)
6 (30%) 5 (36%)
% of lung with ILD
median, (IQR)
20 (10–45) 12.5 (10–30)
% of lung with ground glass appearance
median, (IQR)
20 (5–45) 7.5 (5–10)
% of lung with honeycombing
median, (IQR)
0 (0–1.75) 0 (0–2)
% of lung with emphysema
median, (IQR)
0.5 (0–8.75) 0 (0–3.75)
JAKi Janus kinase inhibitor, ILD interstitial lung disease, FEV1 forced expiratory volume in 1 second, FVC forced vital capacity, TLco lung transfer capacity for carbon monoxide, IQR inter-quartile range, UIP usual interstitial pneumonia, NSIP non-specific interstitial pneumonia, HP hypersensitivity pneumonitis
Treatment details
Concurrent treatments including conventional synthetic DMARDs and corticosteroids were similar in both groups (Table 3). Almost 30% of patients were biologic naïve before commencing the drug of interest. Prior treatment history indicated that those commencing JAKi had previously been treated with a greater number of biologics including TNFi, tocilizumab and abatacept, compared to those commencing rituximab.Table 3 Previous and concurrent DMARD treatments
JAKi (n = 28) Rituximab (n = 19)
Previous treatments
No. of previous csDMARDs
median, (IQR)
3 (2–4) 3 (2–4)
Methotrexate
n, (%)
22 (78.6%) 13 (68.4%)
Biologic Naïve
n, (%)
8 (28.6%) 6 (31.6%)
No. of previous biologics
median, (IQR)
1 (0–2) 0 (0–1)
Rituximab/JAKi
n, (%)
12 (42.9%) 1 (5.3%)
TNF-α inhibitor
n, (%)
15 (53.6%) 5 (26.3%)
Tocilizumab
n, (%)
6 (21.4%) 1 (5.3%)
Abatacept
n, (%)
6 (21.4%) 1 (5.3%)
Concurrent treatments
Prednisolone; n, (%) 13 (46.4%) 6 (31.6%)
Prednisolone dose (mg) median, (IQR) 7 (5–10) 10 (4.5–12.5)
Methotrexate
n, (%)
4 (14.3%) 2 (10.5%)
Azathioprine
n, (%)
2 (7.1%) 2 (10.5%)
Leflunomide
n, (%)
2 (7.1%) 2 (10.5%)
Hydroxychloroquine
n, (%)
6 (21.4%) 6 (31.6%)
Sulfasalazine
n, (%)
3 (10.7%) 4 (21.1%)
DMARD disease-modifying anti-rheumatic drugs, csDMARDs conventional synthetic DMARDs, JAKi Janus kinase inhibitor, IQR inter-quartile range, TNF Tumour necrosis factor
All except for two patients received standard licensed dosing regimens for JAK inhibition for the treatment of RA (i.e. baricitinib 4 mg once daily; tofacitinib 5 mg twice daily). One patient took 4 mg/2 mg of baricitinib on alternate days and one patient took 5 mg of tofacitinib once daily before increasing to 5 mg twice daily.
All patients who received rituximab received the standard dosing of treatment (2 × intravenous infusions of 1000 mg rituximab with 100 mg intravenous methylprednisolone 14 days apart). Only 2/19 patients were on ‘fixed’ cycles of rituximab therapy repeated every 6 months. All others received rituximab as required depending on clinical status but not within 6 months of the last infusion. 5/19 patients received only 1 cycle of rituximab during their study follow-up period. All others (74%) received between 2 and 5 cycles of rituximab.
Incidence of respiratory events
Respiratory events (i.e. hospitalization or death due to a respiratory cause) occurred in five patients (18%) during JAKi treatment (7 hospitalizations, two of which led to death) and in four patients (21%) (4 hospitalizations, one of which led to death) during treatment with rituximab. In the JAKi group, two patients (7.1%) died of respiratory issues (one new lung cancer with concomitant community acquired pneumonia and one of community acquired pneumonia). In the rituximab-treated patients, two patients (10.5%) died during follow-up; one of an acute myocardial infarction and one of pneumonitis and respiratory failure. No episodes of acute pneumonitis occurred in the JAKi group. There was no difference in respiratory event survival in the JAKi-treated group compared with the rituximab-treated group (Fig. 2a); unadjusted hazard ratio: 1.38 (95% CI 0.36, 5.28); p = 0.64.Fig. 2 a Comparison of time to respiratory event and b drug continuation between JAKi and rituximab for the treatment of rheumatoid arthritis in patients with concurrent bronchiectasis and/or ILD
Drug continuation
Seventy-one percent (20/28) of patients with ILD and/or bronchiectasis who commenced JAKi for the treatment of rheumatoid arthritis remained on the drug at the end of the follow-up period. Reasons for discontinuation included inefficacy (n = 3), death (n = 2, as outlined above), recurrent infections (n = 2) and reduced renal function (n = 1).
Sixty-three percent (12/19) of patients who commenced rituximab continued the drug to the end of the follow-up period. Reasons for cessation included inefficacy (n = 5) or death (n = 2, as outlined above). The rate of discontinuation of JAKi in patients with ILD and/or bronchiectasis did not significantly differ from that of those receiving rituximab (Fig. 2b); unadjusted hazard ratio: 1.9 (95% CI 0.634, 5.73); p = 0.251.
Discussion
There is an absence of patients with interstitial lung disease and bronchiectasis in randomised controlled trials that examine the effectiveness of JAK inhibitors for the treatment of rheumatoid arthritis. Therefore, we have limited knowledge of the pulmonary safety in these cohorts. Prevalence of interstitial lung disease and bronchiectasis is relatively high in patients with rheumatoid arthritis, estimated to affect between 5 and 10% of patients [1, 13]. Interstitial lung disease itself contributes significantly to the excess mortality evident in patients with rheumatoid arthritis [14]. Despite this, the optimal treatment strategy for such patients remains unknown, reflecting a paucity of randomized controlled clinical trials for this patient sub-group. Recently, observational data have suggested treatment with rituximab [6, 15, 16], and laterally abatacept [5, 17] is the most effective and safest biologic agents for the treatment of RA with concurrent ILD or bronchiectasis. Data suggest that treatment with rituximab and abatacept may even infer a benefit to pulmonary function [5, 6]. Nonetheless, concern about lung toxicity restricts the use of many DMARDs and biologic agents, be these concerns soundly proven or not.
Novel, small-molecule treatments, such as Janus kinase inhibitors, represent an exciting option for patients with rheumatoid arthritis. Their safety and potential for effectiveness in patients with concurrent pulmonary disease should be duly considered. Currently, our understanding of inhibition of JAK/STAT signalling pathways in patients with ILD and bronchiectasis is extremely limited. JAKi-use has led to successful treatment of dermatomyositis-associated ILD but evidence is restricted to a small number of case reports and case series [18–20]. Even less is understood about Janus kinase inhibition in patients with bronchiectasis. While post hoc analysis of phase 2, 3 and 3b/4 clinical trials indicates that the de novo incidence of interstitial lung disease is low in patients with rheumatoid arthritis treated with JAKi [21]; to our knowledge, no study has yet examined the pulmonary tolerance and safety of JAKi in those with existing ILD or bronchiectasis. The current study sought to address this important clinical scenario, by comparing time to respiratory events in patients receiving JAKi with patients receiving the current, best-practice medication for this clinical cohort, rituximab.
The majority of our patients commenced JAK1/2 inhibition with baricitinib due to the earlier licensing of this medication for the treatment of rheumatoid arthritis but several patients commenced JAK1/3 inhibition with tofacitinib. Furthermore, the proportion of patients with ‘difficult-to-treat’ rheumatoid disease, manifest by a longer duration of illness and a greater number of previous biologic therapies (e.g. tocilizumab, abatacept), was higher in those subsequently prescribed JAKi. We believe this relates to the fact that a novel medication provides an attractive option for patients who have had difficult-to-treat disease for many years and in such cases patients often chose to switch therapy in an attempt to achieve remission.
Encouragingly, serious respiratory events (i.e., hospitalization or death due to a respiratory cause) were no different between those treated with JAKi and those treated with rituximab suggesting that JAKi may offer a realistic treatment option for patients with RA and concurrent ILD or bronchiectasis. Our cohort of patients represents a group with mild to moderate pulmonary disease, all of whom commenced Janus kinase inhibition or rituximab with a known, pre-existing diagnosis of ILD or bronchiectasis (present on cross sectional CT-imaging and/or attending a tertiary ILD clinic). Very few patients with ILD had a HR-CT pattern consistent with definite UIP and most had a pattern that was consistent with an alternative diagnosis and were highly unlikely to be UIP. Therefore, our findings should not be extrapolated to those with severe ILD or bronchiectasis or a definite UIP pattern of disease and caution is urged in this setting. Furthermore, there was no significant difference in time to drug discontinuation between those prescribed JAKi and those prescribed rituximab. Proportionally, more patients treated with rituximab (almost 40%) ceased treatment during follow-up compared with those treated with JAKi (almost 30%). Reasons for discontinuation of rituximab predominantly related to clinical inefficacy, but comparison of clinical effectiveness was not the aim of this study.
There are several strengths to this study. Both cohorts of patients (i.e., JAKi- and rituximab-receiving) were similar in terms of clinical traits (e.g. gender, concurrent medications, corticosteroid use, anti-CCP and rheumatoid factor seropositivity) minimizing the effect of confounders on respiratory event outcomes. In addition, these data represent real-world experience of patients with RA and concurrent ILD/bronchiectasis with patient–clinician decisions made without influence of a pre-determined research goal. The retrospective nature of this observational study and associated risk of information bias represents the major limitation of this study. The relatively small sample size and short follow-up period also represent limitations that may have affected overall results. Ideally, follow-up pulmonary function testing would have been available for all patients after commencing treatment but as this was only the case for a minority of patients, this was not analysed. It was not possible to collect primary care records of respiratory exacerbations or infections that did not require hospital admission that were treated in the community. Furthermore, the electronic patient health record system used did not permit recording of admissions to hospital abroad or to hospitals outside of NHS Lothian. The authors recommend that the findings of this study and analysis should be investigated further in larger prospective cohorts in the future.
To conclude, within the limitations of this retrospective analysis, JAK inhibition with baricitinib or tofacitinib for the treatment of rheumatoid arthritis in patients with concurrent interstitial lung disease or bronchiectasis did not increase the rate of hospitalization or death due to a respiratory cause in comparison to those treated with rituximab. JAK inhibition may provide a favourable and safe therapeutic option for such patients but larger, prospective studies are required to confirm these findings to inform much-needed therapeutic guidance in this sub-category of patients with RA and ILD/bronchiectasis.
Funding
No specific funding was received from any bodies in the public, commercial or not-for-profit sectors to carry out the work described in this article.
Declarations
Conflict of interest
OC has received sponsorship from Lilly and Pfizer to attend educational conferences. The other authors have no conflicts of interest to declare.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. | Intravenous (not otherwise specified) | DrugAdministrationRoute | CC BY | 33721094 | 19,059,614 | 2021-05 |
What was the administration route of drug 'RITUXIMAB'? | A retrospective comparison of respiratory events with JAK inhibitors or rituximab for rheumatoid arthritis in patients with pulmonary disease.
Janus kinase inhibitors (JAKi) are an exciting option for the treatment of rheumatoid arthritis (RA) but little is known about their safety and tolerability in patients with existing respiratory disorders. The objective was to compare pulmonary safety of JAKi versus rituximab in patients with concurrent interstitial lung disease (ILD) or bronchiectasis. We performed a retrospective electronic patient record review of patients with known ILD or bronchiectasis commencing JAKi or rituximab for the treatment of RA. Patients initiating treatment from January 2016 to February 2020 were included. Respiratory events (hospitalization or death from a respiratory cause) were compared using Kaplan-Meier survival analysis. We analysed patients who received JAKi (n = 28) and rituximab (n = 19) for a mean (SD) of 1.1 (0.62) and 2.14 (1) years respectively. Patients were predominantly female (68%), anti-CCP antibody positive (94%) and non-smoking (89%) with a median (IQR) percentage predicted FVC at baseline of 100% (82-115%) and percentage predicted TLCO of 62% (54.5-68%). Respiratory events occurred in five patients treated with JAKi (18%; 5 hospitalizations, 2 deaths) and in four patients treated with rituximab (21%; 3 hospitalizations, 1 death). Respiratory event rates did not differ between groups (Cox-regression proportional hazard ratio = 1.38, 95% CI 0.36-5.28; p = 0.64). In this retrospective study, JAKi for the treatment of RA with existing ILD or bronchiectasis did not increase the rate of hospitalization or death due to respiratory causes compared to those treated with rituximab. JAK inhibition may provide a relatively safe option for RA in such patients.
Introduction
Therapeutic decision-making in the treatment of rheumatoid arthritis (RA) for patients with concurrent pulmonary disease can be challenging. Fear regarding increased infection risk in bronchiectasis and progression of interstitial lung disease (ILD) secondary to drug administration restricts treatment options. Controversy exists as to whether medications, such as methotrexate, leflunomide and tumour necrosis factor-α inhibitors (TNFi), exacerbate pulmonary disease in patients with concurrent ILD and bronchiectasis [1]. This has led to hesitancy in utilizing these medications although systematic review of the evidence is beginning to allay fears [2, 3]. Recently, data have begun to support the safe and effective use of B and T cell therapies, rituximab [4] and abatacept [5], in the management of rheumatoid arthritis-related ILD and bronchiectasis [6, 7]. However, therapeutic options remain limited.
The approval of Janus kinase inhibitors (JAKi) (i.e., baricitinib, tofacitinib, upadacitinib) in the management of moderate to severe RA offers a new and exciting treatment paradigm [8–10]. However, inclusion of patients with concurrent ILD or bronchiectasis in relevant clinical trials is extremely limited. While the use of JAKi for the treatment of RA with co-existent lung disease is tempting due to the oral route of administration and relatively short half-life, their safety in relation to pulmonary toxicity in patients with established ILD or bronchiectasis is unknown.
To investigate this evidence gap, the objective of the current study is to compare the pulmonary safety of Janus kinase inhibition for the treatment of rheumatoid arthritis in patients with existing ILD or bronchiectasis, with rituximab. The primary outcome of interest was the incidence of severe respiratory events (hospitalization or death due to respiratory cause) in those receiving JAKi versus those receiving rituximab.
Materials and methods
Study design and patient inclusion
The study comprised a retrospective observational study. The electronic patient health record system (TrakCare®) was used to collect relevant data. Adult patients (aged 18 or over) undergoing treatment for rheumatoid arthritis at the Rheumatic Diseases Unit, Western General Hospital, Edinburgh, U.K., were included in the study. Only patients with an established diagnosis of ILD or bronchiectasis before the commencement of the drug of interest were eligible for inclusion. Patients without these respiratory conditions were excluded. Dedicated pharmacy and biologic clinic records were used to identify all adult patients (16 years or older) commencing any JAKi or rituximab for the treatment of RA from January 2016 to February 2020 (Fig. 1). From these records, all patients’ health records were screened by the investigators (OC, LK) for co-existing diagnoses of interstitial lung disease and/or bronchiectasis. For relevant patients, data were collected in detail from electronic health records by the investigators (OC, OM). Data were collected until discontinuation of the drug of interest (as documented in the clinical record) or to the end of the follow-up period (1 August 2020). They study was conducted in accordance with the Declaration of Helsinki. As all data were collected as part of routine clinical care and all treatment decisions were made prior to conduction of the study, ethical approval was not required.Fig. 1 Study outline, selection of patients and follow-up
Clinical features and demographics
Identical clinical and demographic data were collected for both treatment groups. Standard clinical variables, such as age, gender, disease duration (rheumatoid arthritis and respiratory diseases), smoking history, past and current medications, anti-CCP antibody and rheumatoid factor (IgM) positivity, were collected along with the dates of commencement and cessation for the drugs of interest. Rheumatoid arthritis disease activity (DAS28-ESR or DAS28-CRP) before the commencement of the treatment of interest was recorded along with activity scores at 3–4 months post initiation (where available).
Baseline pulmonary function tests prior to the commencement of the medications of interest were also assessed. As only a small proportion of patients had follow-up pulmonary function assessment within a reasonable time frame, this was not analysed. The pattern of ILD involvement was described as recommended by the American Thoracic Society/European Respiratory Society following review of high-resolution (HR) CT-thorax imaging by a multi-disciplinary team including chest radiologists and an expert ILD lead physician (NH) who additionally attributed a modified semi-quantitative CT-fibrosis score by visual inspection [11, 12].
Outcome measures
The primary outcome of interest was time to first respiratory event (days). Respiratory events were defined as either admission to hospital with a respiratory illness (e.g., infection, ILD exacerbation), or death from a respiratory cause while taking the medication of interest.
Additional outcomes of interest included drug continuation. Drug continuation was defined for JAKi as from the date of first prescription to the date of drug cessation, death, end of follow-up period, or whichever end-point came first. Drug continuation for rituximab was defined as from the date of first infusion to the date of cessation, death, end of follow-up period or whichever end-point came first. The number of cycles, frequency and dose of rituximab received was recorded for each patient.
Statistical analysis
Statistical analysis was performed using IBM SPSS statistics v.25.0 (IBM, Armonk, New York, USA). Distributions of clinical variables and demographics were illustrated using histograms. Depending on the data distribution and type (i.e., categorical, continuous), variables were then described using means (with standard deviations) or medians (with inter-quartile ranges). Continuous variables were compared for differences between groups using either student’s t tests or Mann–Whitney U testing subject to normality or non-normality of distribution. Proportions of categorical variables were compared using Chi-squared testing. Kaplan–Meier survival analysis (log-rank) was used to compare respiratory event survival and drug continuation survival between treatment groups. Cox regression proportional hazard was used to calculate unadjusted hazard ratios with 95% confidence intervals (CI 95%). Statistical significance was set at p < 0.05 and analysis was two -tailed.
Results
Demographic and clinical characteristics
A total of 291 patients were commenced on JAKi to February 2020 at the Rheumatic Diseases Unit, Edinburgh (Fig. 1). Twenty-eight patients commenced JAK inhibition for the treatment of rheumatoid arthritis who also had concurrent interstitial lung disease (67.9%), bronchiectasis (25%) or both (7.1%). Baricitinib was commenced in 26 patients and tofacitinib in 2 patients. A total of 318 patients were commenced on rituximab from January 2016 to February 2020 (Fig. 1). Nineteen patients commenced rituximab for the treatment of RA and had concurrent ILD (68.4%), bronchiectasis (26.3%) or both (5.3%).
Mean duration of follow-up (to discontinuation of drug, death or end of follow-up period) for patients receiving JAKi was 1.1 years (SD = 0.62) and for patients receiving rituximab infusions was 2.14 years (SD = 1). Demographic and clinical characteristics were similar in both treatment groups (Table 1); however, those commencing JAKi had a significantly longer disease duration (median 11 years) compared to those starting rituximab (median 3 years) (Table 1; Mann–Whitney U test, p = 0.003).Table 1 Demographic and clinical details of included patients
JAKi (n = 28) Rituximab (n = 19) No. of patients with available data (n =)
ILD/Bronchiectasis/both
n, (%)
ILD n = 19 (67.9%)
Bronchiectasis n = 7 (25%)
Both n = 2 (7.1%)
ILD n = 13 (68.4%)
Bronchiectasis n = 5 (26.3%)
Both n = 1 (5.3%)
47
Age (median, IQR) 69 (62.3–75) 70 (59–76) 47
Female n, (%) 18 (64.3%) 14 (73.7%) 47
Smoking status
n, (%)
Smoker:
n = 5 (17.9%)
Ex-smoker:
n = 15 (53.6%)
Never-smoker:
n = 8 (28.6%)
Smoker:
n = 0 (0%),
Ex-smoker:
n = 12 (63.2%)
Never-smoker:
n = 7 (36.8%)
47
Anti-CCP positivity
n, (%)
25 (96.2%) 19 (100%) 45
Rh-Factor IgM positivity
n, (%)
14 (77.8%) 9 (90%) 28
RA diagnosis age median, (IQR) 53 (44–61) 54 (48–67) 42
Respiratory diagnosis age
median, (IQR)
64 (55–73) 64 (53–71) 46
Years from RA diagnosis to drug of interest
median, (IQR)
11 (3–20) 3 (1–6) 42
DAS-28 Baseline
median, (IQR)
5.85 (4.98–6.18) 6.2 (5.53–6.86) 43
DAS-28 3–4 months
median, (IQR)
3.4 (2.38–4.11) 4.2 (3.61–5.44) 32
JAKi Janus kinase inhibitor, ILD interstitial lung disease, Rh-factor rheumatoid factor, RA rheumatoid arthritis, DAS-28 disease activity score 28 joints
Pre-treatment pulmonary function tests indicated mild severity of pulmonary disease in patients commencing JAKi and rituximab (Table 2). Review of patients’ baseline high-resolution CT-thorax revealed similar radiological patterns and severity of ILD in both treatment groups (Table 2). However, a greater degree of ground glass was evident on CT-imaging before treatment in those receiving JAKi compared to those receiving rituximab (p = 0.036).Table 2 Baseline pulmonary function (all patients); and radiological pattern and severity of ILD on CT-thorax imaging (patients with ILD)
JAKi (n = 28) Rituximab (n = 19) No. of patients with available data (n =)
Baseline predicted FEV1 median %, (IQR) 93.5 (74.5–105) 92 (76–105) 43
Baseline predicted FVC median %, (IQR) 103 (78–112) 99 (86–119) 44
Baseline predicted TLCO median %, (IQR) 62 (51.25–69) 62 (55–67) 41
Radiological ILD-pattern JAKi (n = 20) Rituximab (n = 14)
UIP
n, (%)
0 (0%) 2 (14%)
Probable UIP
n, (%)
4 (20%) 5 (36%)
Fibrotic NSIP
n, (%)
6 (30%) 1 (7%)
Chronic HP
n, (%)
3 (15%) 1 (7%)
Respiratory bronchiolitis-ILD
n, (%)
1 (5%) 0 (0%)
Unclassifiable
n, (%)
6 (30%) 5 (36%)
% of lung with ILD
median, (IQR)
20 (10–45) 12.5 (10–30)
% of lung with ground glass appearance
median, (IQR)
20 (5–45) 7.5 (5–10)
% of lung with honeycombing
median, (IQR)
0 (0–1.75) 0 (0–2)
% of lung with emphysema
median, (IQR)
0.5 (0–8.75) 0 (0–3.75)
JAKi Janus kinase inhibitor, ILD interstitial lung disease, FEV1 forced expiratory volume in 1 second, FVC forced vital capacity, TLco lung transfer capacity for carbon monoxide, IQR inter-quartile range, UIP usual interstitial pneumonia, NSIP non-specific interstitial pneumonia, HP hypersensitivity pneumonitis
Treatment details
Concurrent treatments including conventional synthetic DMARDs and corticosteroids were similar in both groups (Table 3). Almost 30% of patients were biologic naïve before commencing the drug of interest. Prior treatment history indicated that those commencing JAKi had previously been treated with a greater number of biologics including TNFi, tocilizumab and abatacept, compared to those commencing rituximab.Table 3 Previous and concurrent DMARD treatments
JAKi (n = 28) Rituximab (n = 19)
Previous treatments
No. of previous csDMARDs
median, (IQR)
3 (2–4) 3 (2–4)
Methotrexate
n, (%)
22 (78.6%) 13 (68.4%)
Biologic Naïve
n, (%)
8 (28.6%) 6 (31.6%)
No. of previous biologics
median, (IQR)
1 (0–2) 0 (0–1)
Rituximab/JAKi
n, (%)
12 (42.9%) 1 (5.3%)
TNF-α inhibitor
n, (%)
15 (53.6%) 5 (26.3%)
Tocilizumab
n, (%)
6 (21.4%) 1 (5.3%)
Abatacept
n, (%)
6 (21.4%) 1 (5.3%)
Concurrent treatments
Prednisolone; n, (%) 13 (46.4%) 6 (31.6%)
Prednisolone dose (mg) median, (IQR) 7 (5–10) 10 (4.5–12.5)
Methotrexate
n, (%)
4 (14.3%) 2 (10.5%)
Azathioprine
n, (%)
2 (7.1%) 2 (10.5%)
Leflunomide
n, (%)
2 (7.1%) 2 (10.5%)
Hydroxychloroquine
n, (%)
6 (21.4%) 6 (31.6%)
Sulfasalazine
n, (%)
3 (10.7%) 4 (21.1%)
DMARD disease-modifying anti-rheumatic drugs, csDMARDs conventional synthetic DMARDs, JAKi Janus kinase inhibitor, IQR inter-quartile range, TNF Tumour necrosis factor
All except for two patients received standard licensed dosing regimens for JAK inhibition for the treatment of RA (i.e. baricitinib 4 mg once daily; tofacitinib 5 mg twice daily). One patient took 4 mg/2 mg of baricitinib on alternate days and one patient took 5 mg of tofacitinib once daily before increasing to 5 mg twice daily.
All patients who received rituximab received the standard dosing of treatment (2 × intravenous infusions of 1000 mg rituximab with 100 mg intravenous methylprednisolone 14 days apart). Only 2/19 patients were on ‘fixed’ cycles of rituximab therapy repeated every 6 months. All others received rituximab as required depending on clinical status but not within 6 months of the last infusion. 5/19 patients received only 1 cycle of rituximab during their study follow-up period. All others (74%) received between 2 and 5 cycles of rituximab.
Incidence of respiratory events
Respiratory events (i.e. hospitalization or death due to a respiratory cause) occurred in five patients (18%) during JAKi treatment (7 hospitalizations, two of which led to death) and in four patients (21%) (4 hospitalizations, one of which led to death) during treatment with rituximab. In the JAKi group, two patients (7.1%) died of respiratory issues (one new lung cancer with concomitant community acquired pneumonia and one of community acquired pneumonia). In the rituximab-treated patients, two patients (10.5%) died during follow-up; one of an acute myocardial infarction and one of pneumonitis and respiratory failure. No episodes of acute pneumonitis occurred in the JAKi group. There was no difference in respiratory event survival in the JAKi-treated group compared with the rituximab-treated group (Fig. 2a); unadjusted hazard ratio: 1.38 (95% CI 0.36, 5.28); p = 0.64.Fig. 2 a Comparison of time to respiratory event and b drug continuation between JAKi and rituximab for the treatment of rheumatoid arthritis in patients with concurrent bronchiectasis and/or ILD
Drug continuation
Seventy-one percent (20/28) of patients with ILD and/or bronchiectasis who commenced JAKi for the treatment of rheumatoid arthritis remained on the drug at the end of the follow-up period. Reasons for discontinuation included inefficacy (n = 3), death (n = 2, as outlined above), recurrent infections (n = 2) and reduced renal function (n = 1).
Sixty-three percent (12/19) of patients who commenced rituximab continued the drug to the end of the follow-up period. Reasons for cessation included inefficacy (n = 5) or death (n = 2, as outlined above). The rate of discontinuation of JAKi in patients with ILD and/or bronchiectasis did not significantly differ from that of those receiving rituximab (Fig. 2b); unadjusted hazard ratio: 1.9 (95% CI 0.634, 5.73); p = 0.251.
Discussion
There is an absence of patients with interstitial lung disease and bronchiectasis in randomised controlled trials that examine the effectiveness of JAK inhibitors for the treatment of rheumatoid arthritis. Therefore, we have limited knowledge of the pulmonary safety in these cohorts. Prevalence of interstitial lung disease and bronchiectasis is relatively high in patients with rheumatoid arthritis, estimated to affect between 5 and 10% of patients [1, 13]. Interstitial lung disease itself contributes significantly to the excess mortality evident in patients with rheumatoid arthritis [14]. Despite this, the optimal treatment strategy for such patients remains unknown, reflecting a paucity of randomized controlled clinical trials for this patient sub-group. Recently, observational data have suggested treatment with rituximab [6, 15, 16], and laterally abatacept [5, 17] is the most effective and safest biologic agents for the treatment of RA with concurrent ILD or bronchiectasis. Data suggest that treatment with rituximab and abatacept may even infer a benefit to pulmonary function [5, 6]. Nonetheless, concern about lung toxicity restricts the use of many DMARDs and biologic agents, be these concerns soundly proven or not.
Novel, small-molecule treatments, such as Janus kinase inhibitors, represent an exciting option for patients with rheumatoid arthritis. Their safety and potential for effectiveness in patients with concurrent pulmonary disease should be duly considered. Currently, our understanding of inhibition of JAK/STAT signalling pathways in patients with ILD and bronchiectasis is extremely limited. JAKi-use has led to successful treatment of dermatomyositis-associated ILD but evidence is restricted to a small number of case reports and case series [18–20]. Even less is understood about Janus kinase inhibition in patients with bronchiectasis. While post hoc analysis of phase 2, 3 and 3b/4 clinical trials indicates that the de novo incidence of interstitial lung disease is low in patients with rheumatoid arthritis treated with JAKi [21]; to our knowledge, no study has yet examined the pulmonary tolerance and safety of JAKi in those with existing ILD or bronchiectasis. The current study sought to address this important clinical scenario, by comparing time to respiratory events in patients receiving JAKi with patients receiving the current, best-practice medication for this clinical cohort, rituximab.
The majority of our patients commenced JAK1/2 inhibition with baricitinib due to the earlier licensing of this medication for the treatment of rheumatoid arthritis but several patients commenced JAK1/3 inhibition with tofacitinib. Furthermore, the proportion of patients with ‘difficult-to-treat’ rheumatoid disease, manifest by a longer duration of illness and a greater number of previous biologic therapies (e.g. tocilizumab, abatacept), was higher in those subsequently prescribed JAKi. We believe this relates to the fact that a novel medication provides an attractive option for patients who have had difficult-to-treat disease for many years and in such cases patients often chose to switch therapy in an attempt to achieve remission.
Encouragingly, serious respiratory events (i.e., hospitalization or death due to a respiratory cause) were no different between those treated with JAKi and those treated with rituximab suggesting that JAKi may offer a realistic treatment option for patients with RA and concurrent ILD or bronchiectasis. Our cohort of patients represents a group with mild to moderate pulmonary disease, all of whom commenced Janus kinase inhibition or rituximab with a known, pre-existing diagnosis of ILD or bronchiectasis (present on cross sectional CT-imaging and/or attending a tertiary ILD clinic). Very few patients with ILD had a HR-CT pattern consistent with definite UIP and most had a pattern that was consistent with an alternative diagnosis and were highly unlikely to be UIP. Therefore, our findings should not be extrapolated to those with severe ILD or bronchiectasis or a definite UIP pattern of disease and caution is urged in this setting. Furthermore, there was no significant difference in time to drug discontinuation between those prescribed JAKi and those prescribed rituximab. Proportionally, more patients treated with rituximab (almost 40%) ceased treatment during follow-up compared with those treated with JAKi (almost 30%). Reasons for discontinuation of rituximab predominantly related to clinical inefficacy, but comparison of clinical effectiveness was not the aim of this study.
There are several strengths to this study. Both cohorts of patients (i.e., JAKi- and rituximab-receiving) were similar in terms of clinical traits (e.g. gender, concurrent medications, corticosteroid use, anti-CCP and rheumatoid factor seropositivity) minimizing the effect of confounders on respiratory event outcomes. In addition, these data represent real-world experience of patients with RA and concurrent ILD/bronchiectasis with patient–clinician decisions made without influence of a pre-determined research goal. The retrospective nature of this observational study and associated risk of information bias represents the major limitation of this study. The relatively small sample size and short follow-up period also represent limitations that may have affected overall results. Ideally, follow-up pulmonary function testing would have been available for all patients after commencing treatment but as this was only the case for a minority of patients, this was not analysed. It was not possible to collect primary care records of respiratory exacerbations or infections that did not require hospital admission that were treated in the community. Furthermore, the electronic patient health record system used did not permit recording of admissions to hospital abroad or to hospitals outside of NHS Lothian. The authors recommend that the findings of this study and analysis should be investigated further in larger prospective cohorts in the future.
To conclude, within the limitations of this retrospective analysis, JAK inhibition with baricitinib or tofacitinib for the treatment of rheumatoid arthritis in patients with concurrent interstitial lung disease or bronchiectasis did not increase the rate of hospitalization or death due to a respiratory cause in comparison to those treated with rituximab. JAK inhibition may provide a favourable and safe therapeutic option for such patients but larger, prospective studies are required to confirm these findings to inform much-needed therapeutic guidance in this sub-category of patients with RA and ILD/bronchiectasis.
Funding
No specific funding was received from any bodies in the public, commercial or not-for-profit sectors to carry out the work described in this article.
Declarations
Conflict of interest
OC has received sponsorship from Lilly and Pfizer to attend educational conferences. The other authors have no conflicts of interest to declare.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. | Intravenous (not otherwise specified) | DrugAdministrationRoute | CC BY | 33721094 | 19,059,614 | 2021-05 |
What was the outcome of reaction 'Acute myocardial infarction'? | A retrospective comparison of respiratory events with JAK inhibitors or rituximab for rheumatoid arthritis in patients with pulmonary disease.
Janus kinase inhibitors (JAKi) are an exciting option for the treatment of rheumatoid arthritis (RA) but little is known about their safety and tolerability in patients with existing respiratory disorders. The objective was to compare pulmonary safety of JAKi versus rituximab in patients with concurrent interstitial lung disease (ILD) or bronchiectasis. We performed a retrospective electronic patient record review of patients with known ILD or bronchiectasis commencing JAKi or rituximab for the treatment of RA. Patients initiating treatment from January 2016 to February 2020 were included. Respiratory events (hospitalization or death from a respiratory cause) were compared using Kaplan-Meier survival analysis. We analysed patients who received JAKi (n = 28) and rituximab (n = 19) for a mean (SD) of 1.1 (0.62) and 2.14 (1) years respectively. Patients were predominantly female (68%), anti-CCP antibody positive (94%) and non-smoking (89%) with a median (IQR) percentage predicted FVC at baseline of 100% (82-115%) and percentage predicted TLCO of 62% (54.5-68%). Respiratory events occurred in five patients treated with JAKi (18%; 5 hospitalizations, 2 deaths) and in four patients treated with rituximab (21%; 3 hospitalizations, 1 death). Respiratory event rates did not differ between groups (Cox-regression proportional hazard ratio = 1.38, 95% CI 0.36-5.28; p = 0.64). In this retrospective study, JAKi for the treatment of RA with existing ILD or bronchiectasis did not increase the rate of hospitalization or death due to respiratory causes compared to those treated with rituximab. JAK inhibition may provide a relatively safe option for RA in such patients.
Introduction
Therapeutic decision-making in the treatment of rheumatoid arthritis (RA) for patients with concurrent pulmonary disease can be challenging. Fear regarding increased infection risk in bronchiectasis and progression of interstitial lung disease (ILD) secondary to drug administration restricts treatment options. Controversy exists as to whether medications, such as methotrexate, leflunomide and tumour necrosis factor-α inhibitors (TNFi), exacerbate pulmonary disease in patients with concurrent ILD and bronchiectasis [1]. This has led to hesitancy in utilizing these medications although systematic review of the evidence is beginning to allay fears [2, 3]. Recently, data have begun to support the safe and effective use of B and T cell therapies, rituximab [4] and abatacept [5], in the management of rheumatoid arthritis-related ILD and bronchiectasis [6, 7]. However, therapeutic options remain limited.
The approval of Janus kinase inhibitors (JAKi) (i.e., baricitinib, tofacitinib, upadacitinib) in the management of moderate to severe RA offers a new and exciting treatment paradigm [8–10]. However, inclusion of patients with concurrent ILD or bronchiectasis in relevant clinical trials is extremely limited. While the use of JAKi for the treatment of RA with co-existent lung disease is tempting due to the oral route of administration and relatively short half-life, their safety in relation to pulmonary toxicity in patients with established ILD or bronchiectasis is unknown.
To investigate this evidence gap, the objective of the current study is to compare the pulmonary safety of Janus kinase inhibition for the treatment of rheumatoid arthritis in patients with existing ILD or bronchiectasis, with rituximab. The primary outcome of interest was the incidence of severe respiratory events (hospitalization or death due to respiratory cause) in those receiving JAKi versus those receiving rituximab.
Materials and methods
Study design and patient inclusion
The study comprised a retrospective observational study. The electronic patient health record system (TrakCare®) was used to collect relevant data. Adult patients (aged 18 or over) undergoing treatment for rheumatoid arthritis at the Rheumatic Diseases Unit, Western General Hospital, Edinburgh, U.K., were included in the study. Only patients with an established diagnosis of ILD or bronchiectasis before the commencement of the drug of interest were eligible for inclusion. Patients without these respiratory conditions were excluded. Dedicated pharmacy and biologic clinic records were used to identify all adult patients (16 years or older) commencing any JAKi or rituximab for the treatment of RA from January 2016 to February 2020 (Fig. 1). From these records, all patients’ health records were screened by the investigators (OC, LK) for co-existing diagnoses of interstitial lung disease and/or bronchiectasis. For relevant patients, data were collected in detail from electronic health records by the investigators (OC, OM). Data were collected until discontinuation of the drug of interest (as documented in the clinical record) or to the end of the follow-up period (1 August 2020). They study was conducted in accordance with the Declaration of Helsinki. As all data were collected as part of routine clinical care and all treatment decisions were made prior to conduction of the study, ethical approval was not required.Fig. 1 Study outline, selection of patients and follow-up
Clinical features and demographics
Identical clinical and demographic data were collected for both treatment groups. Standard clinical variables, such as age, gender, disease duration (rheumatoid arthritis and respiratory diseases), smoking history, past and current medications, anti-CCP antibody and rheumatoid factor (IgM) positivity, were collected along with the dates of commencement and cessation for the drugs of interest. Rheumatoid arthritis disease activity (DAS28-ESR or DAS28-CRP) before the commencement of the treatment of interest was recorded along with activity scores at 3–4 months post initiation (where available).
Baseline pulmonary function tests prior to the commencement of the medications of interest were also assessed. As only a small proportion of patients had follow-up pulmonary function assessment within a reasonable time frame, this was not analysed. The pattern of ILD involvement was described as recommended by the American Thoracic Society/European Respiratory Society following review of high-resolution (HR) CT-thorax imaging by a multi-disciplinary team including chest radiologists and an expert ILD lead physician (NH) who additionally attributed a modified semi-quantitative CT-fibrosis score by visual inspection [11, 12].
Outcome measures
The primary outcome of interest was time to first respiratory event (days). Respiratory events were defined as either admission to hospital with a respiratory illness (e.g., infection, ILD exacerbation), or death from a respiratory cause while taking the medication of interest.
Additional outcomes of interest included drug continuation. Drug continuation was defined for JAKi as from the date of first prescription to the date of drug cessation, death, end of follow-up period, or whichever end-point came first. Drug continuation for rituximab was defined as from the date of first infusion to the date of cessation, death, end of follow-up period or whichever end-point came first. The number of cycles, frequency and dose of rituximab received was recorded for each patient.
Statistical analysis
Statistical analysis was performed using IBM SPSS statistics v.25.0 (IBM, Armonk, New York, USA). Distributions of clinical variables and demographics were illustrated using histograms. Depending on the data distribution and type (i.e., categorical, continuous), variables were then described using means (with standard deviations) or medians (with inter-quartile ranges). Continuous variables were compared for differences between groups using either student’s t tests or Mann–Whitney U testing subject to normality or non-normality of distribution. Proportions of categorical variables were compared using Chi-squared testing. Kaplan–Meier survival analysis (log-rank) was used to compare respiratory event survival and drug continuation survival between treatment groups. Cox regression proportional hazard was used to calculate unadjusted hazard ratios with 95% confidence intervals (CI 95%). Statistical significance was set at p < 0.05 and analysis was two -tailed.
Results
Demographic and clinical characteristics
A total of 291 patients were commenced on JAKi to February 2020 at the Rheumatic Diseases Unit, Edinburgh (Fig. 1). Twenty-eight patients commenced JAK inhibition for the treatment of rheumatoid arthritis who also had concurrent interstitial lung disease (67.9%), bronchiectasis (25%) or both (7.1%). Baricitinib was commenced in 26 patients and tofacitinib in 2 patients. A total of 318 patients were commenced on rituximab from January 2016 to February 2020 (Fig. 1). Nineteen patients commenced rituximab for the treatment of RA and had concurrent ILD (68.4%), bronchiectasis (26.3%) or both (5.3%).
Mean duration of follow-up (to discontinuation of drug, death or end of follow-up period) for patients receiving JAKi was 1.1 years (SD = 0.62) and for patients receiving rituximab infusions was 2.14 years (SD = 1). Demographic and clinical characteristics were similar in both treatment groups (Table 1); however, those commencing JAKi had a significantly longer disease duration (median 11 years) compared to those starting rituximab (median 3 years) (Table 1; Mann–Whitney U test, p = 0.003).Table 1 Demographic and clinical details of included patients
JAKi (n = 28) Rituximab (n = 19) No. of patients with available data (n =)
ILD/Bronchiectasis/both
n, (%)
ILD n = 19 (67.9%)
Bronchiectasis n = 7 (25%)
Both n = 2 (7.1%)
ILD n = 13 (68.4%)
Bronchiectasis n = 5 (26.3%)
Both n = 1 (5.3%)
47
Age (median, IQR) 69 (62.3–75) 70 (59–76) 47
Female n, (%) 18 (64.3%) 14 (73.7%) 47
Smoking status
n, (%)
Smoker:
n = 5 (17.9%)
Ex-smoker:
n = 15 (53.6%)
Never-smoker:
n = 8 (28.6%)
Smoker:
n = 0 (0%),
Ex-smoker:
n = 12 (63.2%)
Never-smoker:
n = 7 (36.8%)
47
Anti-CCP positivity
n, (%)
25 (96.2%) 19 (100%) 45
Rh-Factor IgM positivity
n, (%)
14 (77.8%) 9 (90%) 28
RA diagnosis age median, (IQR) 53 (44–61) 54 (48–67) 42
Respiratory diagnosis age
median, (IQR)
64 (55–73) 64 (53–71) 46
Years from RA diagnosis to drug of interest
median, (IQR)
11 (3–20) 3 (1–6) 42
DAS-28 Baseline
median, (IQR)
5.85 (4.98–6.18) 6.2 (5.53–6.86) 43
DAS-28 3–4 months
median, (IQR)
3.4 (2.38–4.11) 4.2 (3.61–5.44) 32
JAKi Janus kinase inhibitor, ILD interstitial lung disease, Rh-factor rheumatoid factor, RA rheumatoid arthritis, DAS-28 disease activity score 28 joints
Pre-treatment pulmonary function tests indicated mild severity of pulmonary disease in patients commencing JAKi and rituximab (Table 2). Review of patients’ baseline high-resolution CT-thorax revealed similar radiological patterns and severity of ILD in both treatment groups (Table 2). However, a greater degree of ground glass was evident on CT-imaging before treatment in those receiving JAKi compared to those receiving rituximab (p = 0.036).Table 2 Baseline pulmonary function (all patients); and radiological pattern and severity of ILD on CT-thorax imaging (patients with ILD)
JAKi (n = 28) Rituximab (n = 19) No. of patients with available data (n =)
Baseline predicted FEV1 median %, (IQR) 93.5 (74.5–105) 92 (76–105) 43
Baseline predicted FVC median %, (IQR) 103 (78–112) 99 (86–119) 44
Baseline predicted TLCO median %, (IQR) 62 (51.25–69) 62 (55–67) 41
Radiological ILD-pattern JAKi (n = 20) Rituximab (n = 14)
UIP
n, (%)
0 (0%) 2 (14%)
Probable UIP
n, (%)
4 (20%) 5 (36%)
Fibrotic NSIP
n, (%)
6 (30%) 1 (7%)
Chronic HP
n, (%)
3 (15%) 1 (7%)
Respiratory bronchiolitis-ILD
n, (%)
1 (5%) 0 (0%)
Unclassifiable
n, (%)
6 (30%) 5 (36%)
% of lung with ILD
median, (IQR)
20 (10–45) 12.5 (10–30)
% of lung with ground glass appearance
median, (IQR)
20 (5–45) 7.5 (5–10)
% of lung with honeycombing
median, (IQR)
0 (0–1.75) 0 (0–2)
% of lung with emphysema
median, (IQR)
0.5 (0–8.75) 0 (0–3.75)
JAKi Janus kinase inhibitor, ILD interstitial lung disease, FEV1 forced expiratory volume in 1 second, FVC forced vital capacity, TLco lung transfer capacity for carbon monoxide, IQR inter-quartile range, UIP usual interstitial pneumonia, NSIP non-specific interstitial pneumonia, HP hypersensitivity pneumonitis
Treatment details
Concurrent treatments including conventional synthetic DMARDs and corticosteroids were similar in both groups (Table 3). Almost 30% of patients were biologic naïve before commencing the drug of interest. Prior treatment history indicated that those commencing JAKi had previously been treated with a greater number of biologics including TNFi, tocilizumab and abatacept, compared to those commencing rituximab.Table 3 Previous and concurrent DMARD treatments
JAKi (n = 28) Rituximab (n = 19)
Previous treatments
No. of previous csDMARDs
median, (IQR)
3 (2–4) 3 (2–4)
Methotrexate
n, (%)
22 (78.6%) 13 (68.4%)
Biologic Naïve
n, (%)
8 (28.6%) 6 (31.6%)
No. of previous biologics
median, (IQR)
1 (0–2) 0 (0–1)
Rituximab/JAKi
n, (%)
12 (42.9%) 1 (5.3%)
TNF-α inhibitor
n, (%)
15 (53.6%) 5 (26.3%)
Tocilizumab
n, (%)
6 (21.4%) 1 (5.3%)
Abatacept
n, (%)
6 (21.4%) 1 (5.3%)
Concurrent treatments
Prednisolone; n, (%) 13 (46.4%) 6 (31.6%)
Prednisolone dose (mg) median, (IQR) 7 (5–10) 10 (4.5–12.5)
Methotrexate
n, (%)
4 (14.3%) 2 (10.5%)
Azathioprine
n, (%)
2 (7.1%) 2 (10.5%)
Leflunomide
n, (%)
2 (7.1%) 2 (10.5%)
Hydroxychloroquine
n, (%)
6 (21.4%) 6 (31.6%)
Sulfasalazine
n, (%)
3 (10.7%) 4 (21.1%)
DMARD disease-modifying anti-rheumatic drugs, csDMARDs conventional synthetic DMARDs, JAKi Janus kinase inhibitor, IQR inter-quartile range, TNF Tumour necrosis factor
All except for two patients received standard licensed dosing regimens for JAK inhibition for the treatment of RA (i.e. baricitinib 4 mg once daily; tofacitinib 5 mg twice daily). One patient took 4 mg/2 mg of baricitinib on alternate days and one patient took 5 mg of tofacitinib once daily before increasing to 5 mg twice daily.
All patients who received rituximab received the standard dosing of treatment (2 × intravenous infusions of 1000 mg rituximab with 100 mg intravenous methylprednisolone 14 days apart). Only 2/19 patients were on ‘fixed’ cycles of rituximab therapy repeated every 6 months. All others received rituximab as required depending on clinical status but not within 6 months of the last infusion. 5/19 patients received only 1 cycle of rituximab during their study follow-up period. All others (74%) received between 2 and 5 cycles of rituximab.
Incidence of respiratory events
Respiratory events (i.e. hospitalization or death due to a respiratory cause) occurred in five patients (18%) during JAKi treatment (7 hospitalizations, two of which led to death) and in four patients (21%) (4 hospitalizations, one of which led to death) during treatment with rituximab. In the JAKi group, two patients (7.1%) died of respiratory issues (one new lung cancer with concomitant community acquired pneumonia and one of community acquired pneumonia). In the rituximab-treated patients, two patients (10.5%) died during follow-up; one of an acute myocardial infarction and one of pneumonitis and respiratory failure. No episodes of acute pneumonitis occurred in the JAKi group. There was no difference in respiratory event survival in the JAKi-treated group compared with the rituximab-treated group (Fig. 2a); unadjusted hazard ratio: 1.38 (95% CI 0.36, 5.28); p = 0.64.Fig. 2 a Comparison of time to respiratory event and b drug continuation between JAKi and rituximab for the treatment of rheumatoid arthritis in patients with concurrent bronchiectasis and/or ILD
Drug continuation
Seventy-one percent (20/28) of patients with ILD and/or bronchiectasis who commenced JAKi for the treatment of rheumatoid arthritis remained on the drug at the end of the follow-up period. Reasons for discontinuation included inefficacy (n = 3), death (n = 2, as outlined above), recurrent infections (n = 2) and reduced renal function (n = 1).
Sixty-three percent (12/19) of patients who commenced rituximab continued the drug to the end of the follow-up period. Reasons for cessation included inefficacy (n = 5) or death (n = 2, as outlined above). The rate of discontinuation of JAKi in patients with ILD and/or bronchiectasis did not significantly differ from that of those receiving rituximab (Fig. 2b); unadjusted hazard ratio: 1.9 (95% CI 0.634, 5.73); p = 0.251.
Discussion
There is an absence of patients with interstitial lung disease and bronchiectasis in randomised controlled trials that examine the effectiveness of JAK inhibitors for the treatment of rheumatoid arthritis. Therefore, we have limited knowledge of the pulmonary safety in these cohorts. Prevalence of interstitial lung disease and bronchiectasis is relatively high in patients with rheumatoid arthritis, estimated to affect between 5 and 10% of patients [1, 13]. Interstitial lung disease itself contributes significantly to the excess mortality evident in patients with rheumatoid arthritis [14]. Despite this, the optimal treatment strategy for such patients remains unknown, reflecting a paucity of randomized controlled clinical trials for this patient sub-group. Recently, observational data have suggested treatment with rituximab [6, 15, 16], and laterally abatacept [5, 17] is the most effective and safest biologic agents for the treatment of RA with concurrent ILD or bronchiectasis. Data suggest that treatment with rituximab and abatacept may even infer a benefit to pulmonary function [5, 6]. Nonetheless, concern about lung toxicity restricts the use of many DMARDs and biologic agents, be these concerns soundly proven or not.
Novel, small-molecule treatments, such as Janus kinase inhibitors, represent an exciting option for patients with rheumatoid arthritis. Their safety and potential for effectiveness in patients with concurrent pulmonary disease should be duly considered. Currently, our understanding of inhibition of JAK/STAT signalling pathways in patients with ILD and bronchiectasis is extremely limited. JAKi-use has led to successful treatment of dermatomyositis-associated ILD but evidence is restricted to a small number of case reports and case series [18–20]. Even less is understood about Janus kinase inhibition in patients with bronchiectasis. While post hoc analysis of phase 2, 3 and 3b/4 clinical trials indicates that the de novo incidence of interstitial lung disease is low in patients with rheumatoid arthritis treated with JAKi [21]; to our knowledge, no study has yet examined the pulmonary tolerance and safety of JAKi in those with existing ILD or bronchiectasis. The current study sought to address this important clinical scenario, by comparing time to respiratory events in patients receiving JAKi with patients receiving the current, best-practice medication for this clinical cohort, rituximab.
The majority of our patients commenced JAK1/2 inhibition with baricitinib due to the earlier licensing of this medication for the treatment of rheumatoid arthritis but several patients commenced JAK1/3 inhibition with tofacitinib. Furthermore, the proportion of patients with ‘difficult-to-treat’ rheumatoid disease, manifest by a longer duration of illness and a greater number of previous biologic therapies (e.g. tocilizumab, abatacept), was higher in those subsequently prescribed JAKi. We believe this relates to the fact that a novel medication provides an attractive option for patients who have had difficult-to-treat disease for many years and in such cases patients often chose to switch therapy in an attempt to achieve remission.
Encouragingly, serious respiratory events (i.e., hospitalization or death due to a respiratory cause) were no different between those treated with JAKi and those treated with rituximab suggesting that JAKi may offer a realistic treatment option for patients with RA and concurrent ILD or bronchiectasis. Our cohort of patients represents a group with mild to moderate pulmonary disease, all of whom commenced Janus kinase inhibition or rituximab with a known, pre-existing diagnosis of ILD or bronchiectasis (present on cross sectional CT-imaging and/or attending a tertiary ILD clinic). Very few patients with ILD had a HR-CT pattern consistent with definite UIP and most had a pattern that was consistent with an alternative diagnosis and were highly unlikely to be UIP. Therefore, our findings should not be extrapolated to those with severe ILD or bronchiectasis or a definite UIP pattern of disease and caution is urged in this setting. Furthermore, there was no significant difference in time to drug discontinuation between those prescribed JAKi and those prescribed rituximab. Proportionally, more patients treated with rituximab (almost 40%) ceased treatment during follow-up compared with those treated with JAKi (almost 30%). Reasons for discontinuation of rituximab predominantly related to clinical inefficacy, but comparison of clinical effectiveness was not the aim of this study.
There are several strengths to this study. Both cohorts of patients (i.e., JAKi- and rituximab-receiving) were similar in terms of clinical traits (e.g. gender, concurrent medications, corticosteroid use, anti-CCP and rheumatoid factor seropositivity) minimizing the effect of confounders on respiratory event outcomes. In addition, these data represent real-world experience of patients with RA and concurrent ILD/bronchiectasis with patient–clinician decisions made without influence of a pre-determined research goal. The retrospective nature of this observational study and associated risk of information bias represents the major limitation of this study. The relatively small sample size and short follow-up period also represent limitations that may have affected overall results. Ideally, follow-up pulmonary function testing would have been available for all patients after commencing treatment but as this was only the case for a minority of patients, this was not analysed. It was not possible to collect primary care records of respiratory exacerbations or infections that did not require hospital admission that were treated in the community. Furthermore, the electronic patient health record system used did not permit recording of admissions to hospital abroad or to hospitals outside of NHS Lothian. The authors recommend that the findings of this study and analysis should be investigated further in larger prospective cohorts in the future.
To conclude, within the limitations of this retrospective analysis, JAK inhibition with baricitinib or tofacitinib for the treatment of rheumatoid arthritis in patients with concurrent interstitial lung disease or bronchiectasis did not increase the rate of hospitalization or death due to a respiratory cause in comparison to those treated with rituximab. JAK inhibition may provide a favourable and safe therapeutic option for such patients but larger, prospective studies are required to confirm these findings to inform much-needed therapeutic guidance in this sub-category of patients with RA and ILD/bronchiectasis.
Funding
No specific funding was received from any bodies in the public, commercial or not-for-profit sectors to carry out the work described in this article.
Declarations
Conflict of interest
OC has received sponsorship from Lilly and Pfizer to attend educational conferences. The other authors have no conflicts of interest to declare.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. | Fatal | ReactionOutcome | CC BY | 33721094 | 19,059,238 | 2021-05 |
What was the outcome of reaction 'Death'? | A retrospective comparison of respiratory events with JAK inhibitors or rituximab for rheumatoid arthritis in patients with pulmonary disease.
Janus kinase inhibitors (JAKi) are an exciting option for the treatment of rheumatoid arthritis (RA) but little is known about their safety and tolerability in patients with existing respiratory disorders. The objective was to compare pulmonary safety of JAKi versus rituximab in patients with concurrent interstitial lung disease (ILD) or bronchiectasis. We performed a retrospective electronic patient record review of patients with known ILD or bronchiectasis commencing JAKi or rituximab for the treatment of RA. Patients initiating treatment from January 2016 to February 2020 were included. Respiratory events (hospitalization or death from a respiratory cause) were compared using Kaplan-Meier survival analysis. We analysed patients who received JAKi (n = 28) and rituximab (n = 19) for a mean (SD) of 1.1 (0.62) and 2.14 (1) years respectively. Patients were predominantly female (68%), anti-CCP antibody positive (94%) and non-smoking (89%) with a median (IQR) percentage predicted FVC at baseline of 100% (82-115%) and percentage predicted TLCO of 62% (54.5-68%). Respiratory events occurred in five patients treated with JAKi (18%; 5 hospitalizations, 2 deaths) and in four patients treated with rituximab (21%; 3 hospitalizations, 1 death). Respiratory event rates did not differ between groups (Cox-regression proportional hazard ratio = 1.38, 95% CI 0.36-5.28; p = 0.64). In this retrospective study, JAKi for the treatment of RA with existing ILD or bronchiectasis did not increase the rate of hospitalization or death due to respiratory causes compared to those treated with rituximab. JAK inhibition may provide a relatively safe option for RA in such patients.
Introduction
Therapeutic decision-making in the treatment of rheumatoid arthritis (RA) for patients with concurrent pulmonary disease can be challenging. Fear regarding increased infection risk in bronchiectasis and progression of interstitial lung disease (ILD) secondary to drug administration restricts treatment options. Controversy exists as to whether medications, such as methotrexate, leflunomide and tumour necrosis factor-α inhibitors (TNFi), exacerbate pulmonary disease in patients with concurrent ILD and bronchiectasis [1]. This has led to hesitancy in utilizing these medications although systematic review of the evidence is beginning to allay fears [2, 3]. Recently, data have begun to support the safe and effective use of B and T cell therapies, rituximab [4] and abatacept [5], in the management of rheumatoid arthritis-related ILD and bronchiectasis [6, 7]. However, therapeutic options remain limited.
The approval of Janus kinase inhibitors (JAKi) (i.e., baricitinib, tofacitinib, upadacitinib) in the management of moderate to severe RA offers a new and exciting treatment paradigm [8–10]. However, inclusion of patients with concurrent ILD or bronchiectasis in relevant clinical trials is extremely limited. While the use of JAKi for the treatment of RA with co-existent lung disease is tempting due to the oral route of administration and relatively short half-life, their safety in relation to pulmonary toxicity in patients with established ILD or bronchiectasis is unknown.
To investigate this evidence gap, the objective of the current study is to compare the pulmonary safety of Janus kinase inhibition for the treatment of rheumatoid arthritis in patients with existing ILD or bronchiectasis, with rituximab. The primary outcome of interest was the incidence of severe respiratory events (hospitalization or death due to respiratory cause) in those receiving JAKi versus those receiving rituximab.
Materials and methods
Study design and patient inclusion
The study comprised a retrospective observational study. The electronic patient health record system (TrakCare®) was used to collect relevant data. Adult patients (aged 18 or over) undergoing treatment for rheumatoid arthritis at the Rheumatic Diseases Unit, Western General Hospital, Edinburgh, U.K., were included in the study. Only patients with an established diagnosis of ILD or bronchiectasis before the commencement of the drug of interest were eligible for inclusion. Patients without these respiratory conditions were excluded. Dedicated pharmacy and biologic clinic records were used to identify all adult patients (16 years or older) commencing any JAKi or rituximab for the treatment of RA from January 2016 to February 2020 (Fig. 1). From these records, all patients’ health records were screened by the investigators (OC, LK) for co-existing diagnoses of interstitial lung disease and/or bronchiectasis. For relevant patients, data were collected in detail from electronic health records by the investigators (OC, OM). Data were collected until discontinuation of the drug of interest (as documented in the clinical record) or to the end of the follow-up period (1 August 2020). They study was conducted in accordance with the Declaration of Helsinki. As all data were collected as part of routine clinical care and all treatment decisions were made prior to conduction of the study, ethical approval was not required.Fig. 1 Study outline, selection of patients and follow-up
Clinical features and demographics
Identical clinical and demographic data were collected for both treatment groups. Standard clinical variables, such as age, gender, disease duration (rheumatoid arthritis and respiratory diseases), smoking history, past and current medications, anti-CCP antibody and rheumatoid factor (IgM) positivity, were collected along with the dates of commencement and cessation for the drugs of interest. Rheumatoid arthritis disease activity (DAS28-ESR or DAS28-CRP) before the commencement of the treatment of interest was recorded along with activity scores at 3–4 months post initiation (where available).
Baseline pulmonary function tests prior to the commencement of the medications of interest were also assessed. As only a small proportion of patients had follow-up pulmonary function assessment within a reasonable time frame, this was not analysed. The pattern of ILD involvement was described as recommended by the American Thoracic Society/European Respiratory Society following review of high-resolution (HR) CT-thorax imaging by a multi-disciplinary team including chest radiologists and an expert ILD lead physician (NH) who additionally attributed a modified semi-quantitative CT-fibrosis score by visual inspection [11, 12].
Outcome measures
The primary outcome of interest was time to first respiratory event (days). Respiratory events were defined as either admission to hospital with a respiratory illness (e.g., infection, ILD exacerbation), or death from a respiratory cause while taking the medication of interest.
Additional outcomes of interest included drug continuation. Drug continuation was defined for JAKi as from the date of first prescription to the date of drug cessation, death, end of follow-up period, or whichever end-point came first. Drug continuation for rituximab was defined as from the date of first infusion to the date of cessation, death, end of follow-up period or whichever end-point came first. The number of cycles, frequency and dose of rituximab received was recorded for each patient.
Statistical analysis
Statistical analysis was performed using IBM SPSS statistics v.25.0 (IBM, Armonk, New York, USA). Distributions of clinical variables and demographics were illustrated using histograms. Depending on the data distribution and type (i.e., categorical, continuous), variables were then described using means (with standard deviations) or medians (with inter-quartile ranges). Continuous variables were compared for differences between groups using either student’s t tests or Mann–Whitney U testing subject to normality or non-normality of distribution. Proportions of categorical variables were compared using Chi-squared testing. Kaplan–Meier survival analysis (log-rank) was used to compare respiratory event survival and drug continuation survival between treatment groups. Cox regression proportional hazard was used to calculate unadjusted hazard ratios with 95% confidence intervals (CI 95%). Statistical significance was set at p < 0.05 and analysis was two -tailed.
Results
Demographic and clinical characteristics
A total of 291 patients were commenced on JAKi to February 2020 at the Rheumatic Diseases Unit, Edinburgh (Fig. 1). Twenty-eight patients commenced JAK inhibition for the treatment of rheumatoid arthritis who also had concurrent interstitial lung disease (67.9%), bronchiectasis (25%) or both (7.1%). Baricitinib was commenced in 26 patients and tofacitinib in 2 patients. A total of 318 patients were commenced on rituximab from January 2016 to February 2020 (Fig. 1). Nineteen patients commenced rituximab for the treatment of RA and had concurrent ILD (68.4%), bronchiectasis (26.3%) or both (5.3%).
Mean duration of follow-up (to discontinuation of drug, death or end of follow-up period) for patients receiving JAKi was 1.1 years (SD = 0.62) and for patients receiving rituximab infusions was 2.14 years (SD = 1). Demographic and clinical characteristics were similar in both treatment groups (Table 1); however, those commencing JAKi had a significantly longer disease duration (median 11 years) compared to those starting rituximab (median 3 years) (Table 1; Mann–Whitney U test, p = 0.003).Table 1 Demographic and clinical details of included patients
JAKi (n = 28) Rituximab (n = 19) No. of patients with available data (n =)
ILD/Bronchiectasis/both
n, (%)
ILD n = 19 (67.9%)
Bronchiectasis n = 7 (25%)
Both n = 2 (7.1%)
ILD n = 13 (68.4%)
Bronchiectasis n = 5 (26.3%)
Both n = 1 (5.3%)
47
Age (median, IQR) 69 (62.3–75) 70 (59–76) 47
Female n, (%) 18 (64.3%) 14 (73.7%) 47
Smoking status
n, (%)
Smoker:
n = 5 (17.9%)
Ex-smoker:
n = 15 (53.6%)
Never-smoker:
n = 8 (28.6%)
Smoker:
n = 0 (0%),
Ex-smoker:
n = 12 (63.2%)
Never-smoker:
n = 7 (36.8%)
47
Anti-CCP positivity
n, (%)
25 (96.2%) 19 (100%) 45
Rh-Factor IgM positivity
n, (%)
14 (77.8%) 9 (90%) 28
RA diagnosis age median, (IQR) 53 (44–61) 54 (48–67) 42
Respiratory diagnosis age
median, (IQR)
64 (55–73) 64 (53–71) 46
Years from RA diagnosis to drug of interest
median, (IQR)
11 (3–20) 3 (1–6) 42
DAS-28 Baseline
median, (IQR)
5.85 (4.98–6.18) 6.2 (5.53–6.86) 43
DAS-28 3–4 months
median, (IQR)
3.4 (2.38–4.11) 4.2 (3.61–5.44) 32
JAKi Janus kinase inhibitor, ILD interstitial lung disease, Rh-factor rheumatoid factor, RA rheumatoid arthritis, DAS-28 disease activity score 28 joints
Pre-treatment pulmonary function tests indicated mild severity of pulmonary disease in patients commencing JAKi and rituximab (Table 2). Review of patients’ baseline high-resolution CT-thorax revealed similar radiological patterns and severity of ILD in both treatment groups (Table 2). However, a greater degree of ground glass was evident on CT-imaging before treatment in those receiving JAKi compared to those receiving rituximab (p = 0.036).Table 2 Baseline pulmonary function (all patients); and radiological pattern and severity of ILD on CT-thorax imaging (patients with ILD)
JAKi (n = 28) Rituximab (n = 19) No. of patients with available data (n =)
Baseline predicted FEV1 median %, (IQR) 93.5 (74.5–105) 92 (76–105) 43
Baseline predicted FVC median %, (IQR) 103 (78–112) 99 (86–119) 44
Baseline predicted TLCO median %, (IQR) 62 (51.25–69) 62 (55–67) 41
Radiological ILD-pattern JAKi (n = 20) Rituximab (n = 14)
UIP
n, (%)
0 (0%) 2 (14%)
Probable UIP
n, (%)
4 (20%) 5 (36%)
Fibrotic NSIP
n, (%)
6 (30%) 1 (7%)
Chronic HP
n, (%)
3 (15%) 1 (7%)
Respiratory bronchiolitis-ILD
n, (%)
1 (5%) 0 (0%)
Unclassifiable
n, (%)
6 (30%) 5 (36%)
% of lung with ILD
median, (IQR)
20 (10–45) 12.5 (10–30)
% of lung with ground glass appearance
median, (IQR)
20 (5–45) 7.5 (5–10)
% of lung with honeycombing
median, (IQR)
0 (0–1.75) 0 (0–2)
% of lung with emphysema
median, (IQR)
0.5 (0–8.75) 0 (0–3.75)
JAKi Janus kinase inhibitor, ILD interstitial lung disease, FEV1 forced expiratory volume in 1 second, FVC forced vital capacity, TLco lung transfer capacity for carbon monoxide, IQR inter-quartile range, UIP usual interstitial pneumonia, NSIP non-specific interstitial pneumonia, HP hypersensitivity pneumonitis
Treatment details
Concurrent treatments including conventional synthetic DMARDs and corticosteroids were similar in both groups (Table 3). Almost 30% of patients were biologic naïve before commencing the drug of interest. Prior treatment history indicated that those commencing JAKi had previously been treated with a greater number of biologics including TNFi, tocilizumab and abatacept, compared to those commencing rituximab.Table 3 Previous and concurrent DMARD treatments
JAKi (n = 28) Rituximab (n = 19)
Previous treatments
No. of previous csDMARDs
median, (IQR)
3 (2–4) 3 (2–4)
Methotrexate
n, (%)
22 (78.6%) 13 (68.4%)
Biologic Naïve
n, (%)
8 (28.6%) 6 (31.6%)
No. of previous biologics
median, (IQR)
1 (0–2) 0 (0–1)
Rituximab/JAKi
n, (%)
12 (42.9%) 1 (5.3%)
TNF-α inhibitor
n, (%)
15 (53.6%) 5 (26.3%)
Tocilizumab
n, (%)
6 (21.4%) 1 (5.3%)
Abatacept
n, (%)
6 (21.4%) 1 (5.3%)
Concurrent treatments
Prednisolone; n, (%) 13 (46.4%) 6 (31.6%)
Prednisolone dose (mg) median, (IQR) 7 (5–10) 10 (4.5–12.5)
Methotrexate
n, (%)
4 (14.3%) 2 (10.5%)
Azathioprine
n, (%)
2 (7.1%) 2 (10.5%)
Leflunomide
n, (%)
2 (7.1%) 2 (10.5%)
Hydroxychloroquine
n, (%)
6 (21.4%) 6 (31.6%)
Sulfasalazine
n, (%)
3 (10.7%) 4 (21.1%)
DMARD disease-modifying anti-rheumatic drugs, csDMARDs conventional synthetic DMARDs, JAKi Janus kinase inhibitor, IQR inter-quartile range, TNF Tumour necrosis factor
All except for two patients received standard licensed dosing regimens for JAK inhibition for the treatment of RA (i.e. baricitinib 4 mg once daily; tofacitinib 5 mg twice daily). One patient took 4 mg/2 mg of baricitinib on alternate days and one patient took 5 mg of tofacitinib once daily before increasing to 5 mg twice daily.
All patients who received rituximab received the standard dosing of treatment (2 × intravenous infusions of 1000 mg rituximab with 100 mg intravenous methylprednisolone 14 days apart). Only 2/19 patients were on ‘fixed’ cycles of rituximab therapy repeated every 6 months. All others received rituximab as required depending on clinical status but not within 6 months of the last infusion. 5/19 patients received only 1 cycle of rituximab during their study follow-up period. All others (74%) received between 2 and 5 cycles of rituximab.
Incidence of respiratory events
Respiratory events (i.e. hospitalization or death due to a respiratory cause) occurred in five patients (18%) during JAKi treatment (7 hospitalizations, two of which led to death) and in four patients (21%) (4 hospitalizations, one of which led to death) during treatment with rituximab. In the JAKi group, two patients (7.1%) died of respiratory issues (one new lung cancer with concomitant community acquired pneumonia and one of community acquired pneumonia). In the rituximab-treated patients, two patients (10.5%) died during follow-up; one of an acute myocardial infarction and one of pneumonitis and respiratory failure. No episodes of acute pneumonitis occurred in the JAKi group. There was no difference in respiratory event survival in the JAKi-treated group compared with the rituximab-treated group (Fig. 2a); unadjusted hazard ratio: 1.38 (95% CI 0.36, 5.28); p = 0.64.Fig. 2 a Comparison of time to respiratory event and b drug continuation between JAKi and rituximab for the treatment of rheumatoid arthritis in patients with concurrent bronchiectasis and/or ILD
Drug continuation
Seventy-one percent (20/28) of patients with ILD and/or bronchiectasis who commenced JAKi for the treatment of rheumatoid arthritis remained on the drug at the end of the follow-up period. Reasons for discontinuation included inefficacy (n = 3), death (n = 2, as outlined above), recurrent infections (n = 2) and reduced renal function (n = 1).
Sixty-three percent (12/19) of patients who commenced rituximab continued the drug to the end of the follow-up period. Reasons for cessation included inefficacy (n = 5) or death (n = 2, as outlined above). The rate of discontinuation of JAKi in patients with ILD and/or bronchiectasis did not significantly differ from that of those receiving rituximab (Fig. 2b); unadjusted hazard ratio: 1.9 (95% CI 0.634, 5.73); p = 0.251.
Discussion
There is an absence of patients with interstitial lung disease and bronchiectasis in randomised controlled trials that examine the effectiveness of JAK inhibitors for the treatment of rheumatoid arthritis. Therefore, we have limited knowledge of the pulmonary safety in these cohorts. Prevalence of interstitial lung disease and bronchiectasis is relatively high in patients with rheumatoid arthritis, estimated to affect between 5 and 10% of patients [1, 13]. Interstitial lung disease itself contributes significantly to the excess mortality evident in patients with rheumatoid arthritis [14]. Despite this, the optimal treatment strategy for such patients remains unknown, reflecting a paucity of randomized controlled clinical trials for this patient sub-group. Recently, observational data have suggested treatment with rituximab [6, 15, 16], and laterally abatacept [5, 17] is the most effective and safest biologic agents for the treatment of RA with concurrent ILD or bronchiectasis. Data suggest that treatment with rituximab and abatacept may even infer a benefit to pulmonary function [5, 6]. Nonetheless, concern about lung toxicity restricts the use of many DMARDs and biologic agents, be these concerns soundly proven or not.
Novel, small-molecule treatments, such as Janus kinase inhibitors, represent an exciting option for patients with rheumatoid arthritis. Their safety and potential for effectiveness in patients with concurrent pulmonary disease should be duly considered. Currently, our understanding of inhibition of JAK/STAT signalling pathways in patients with ILD and bronchiectasis is extremely limited. JAKi-use has led to successful treatment of dermatomyositis-associated ILD but evidence is restricted to a small number of case reports and case series [18–20]. Even less is understood about Janus kinase inhibition in patients with bronchiectasis. While post hoc analysis of phase 2, 3 and 3b/4 clinical trials indicates that the de novo incidence of interstitial lung disease is low in patients with rheumatoid arthritis treated with JAKi [21]; to our knowledge, no study has yet examined the pulmonary tolerance and safety of JAKi in those with existing ILD or bronchiectasis. The current study sought to address this important clinical scenario, by comparing time to respiratory events in patients receiving JAKi with patients receiving the current, best-practice medication for this clinical cohort, rituximab.
The majority of our patients commenced JAK1/2 inhibition with baricitinib due to the earlier licensing of this medication for the treatment of rheumatoid arthritis but several patients commenced JAK1/3 inhibition with tofacitinib. Furthermore, the proportion of patients with ‘difficult-to-treat’ rheumatoid disease, manifest by a longer duration of illness and a greater number of previous biologic therapies (e.g. tocilizumab, abatacept), was higher in those subsequently prescribed JAKi. We believe this relates to the fact that a novel medication provides an attractive option for patients who have had difficult-to-treat disease for many years and in such cases patients often chose to switch therapy in an attempt to achieve remission.
Encouragingly, serious respiratory events (i.e., hospitalization or death due to a respiratory cause) were no different between those treated with JAKi and those treated with rituximab suggesting that JAKi may offer a realistic treatment option for patients with RA and concurrent ILD or bronchiectasis. Our cohort of patients represents a group with mild to moderate pulmonary disease, all of whom commenced Janus kinase inhibition or rituximab with a known, pre-existing diagnosis of ILD or bronchiectasis (present on cross sectional CT-imaging and/or attending a tertiary ILD clinic). Very few patients with ILD had a HR-CT pattern consistent with definite UIP and most had a pattern that was consistent with an alternative diagnosis and were highly unlikely to be UIP. Therefore, our findings should not be extrapolated to those with severe ILD or bronchiectasis or a definite UIP pattern of disease and caution is urged in this setting. Furthermore, there was no significant difference in time to drug discontinuation between those prescribed JAKi and those prescribed rituximab. Proportionally, more patients treated with rituximab (almost 40%) ceased treatment during follow-up compared with those treated with JAKi (almost 30%). Reasons for discontinuation of rituximab predominantly related to clinical inefficacy, but comparison of clinical effectiveness was not the aim of this study.
There are several strengths to this study. Both cohorts of patients (i.e., JAKi- and rituximab-receiving) were similar in terms of clinical traits (e.g. gender, concurrent medications, corticosteroid use, anti-CCP and rheumatoid factor seropositivity) minimizing the effect of confounders on respiratory event outcomes. In addition, these data represent real-world experience of patients with RA and concurrent ILD/bronchiectasis with patient–clinician decisions made without influence of a pre-determined research goal. The retrospective nature of this observational study and associated risk of information bias represents the major limitation of this study. The relatively small sample size and short follow-up period also represent limitations that may have affected overall results. Ideally, follow-up pulmonary function testing would have been available for all patients after commencing treatment but as this was only the case for a minority of patients, this was not analysed. It was not possible to collect primary care records of respiratory exacerbations or infections that did not require hospital admission that were treated in the community. Furthermore, the electronic patient health record system used did not permit recording of admissions to hospital abroad or to hospitals outside of NHS Lothian. The authors recommend that the findings of this study and analysis should be investigated further in larger prospective cohorts in the future.
To conclude, within the limitations of this retrospective analysis, JAK inhibition with baricitinib or tofacitinib for the treatment of rheumatoid arthritis in patients with concurrent interstitial lung disease or bronchiectasis did not increase the rate of hospitalization or death due to a respiratory cause in comparison to those treated with rituximab. JAK inhibition may provide a favourable and safe therapeutic option for such patients but larger, prospective studies are required to confirm these findings to inform much-needed therapeutic guidance in this sub-category of patients with RA and ILD/bronchiectasis.
Funding
No specific funding was received from any bodies in the public, commercial or not-for-profit sectors to carry out the work described in this article.
Declarations
Conflict of interest
OC has received sponsorship from Lilly and Pfizer to attend educational conferences. The other authors have no conflicts of interest to declare.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. | Fatal | ReactionOutcome | CC BY | 33721094 | 19,082,833 | 2021-05 |
What was the outcome of reaction 'Pneumonitis'? | A retrospective comparison of respiratory events with JAK inhibitors or rituximab for rheumatoid arthritis in patients with pulmonary disease.
Janus kinase inhibitors (JAKi) are an exciting option for the treatment of rheumatoid arthritis (RA) but little is known about their safety and tolerability in patients with existing respiratory disorders. The objective was to compare pulmonary safety of JAKi versus rituximab in patients with concurrent interstitial lung disease (ILD) or bronchiectasis. We performed a retrospective electronic patient record review of patients with known ILD or bronchiectasis commencing JAKi or rituximab for the treatment of RA. Patients initiating treatment from January 2016 to February 2020 were included. Respiratory events (hospitalization or death from a respiratory cause) were compared using Kaplan-Meier survival analysis. We analysed patients who received JAKi (n = 28) and rituximab (n = 19) for a mean (SD) of 1.1 (0.62) and 2.14 (1) years respectively. Patients were predominantly female (68%), anti-CCP antibody positive (94%) and non-smoking (89%) with a median (IQR) percentage predicted FVC at baseline of 100% (82-115%) and percentage predicted TLCO of 62% (54.5-68%). Respiratory events occurred in five patients treated with JAKi (18%; 5 hospitalizations, 2 deaths) and in four patients treated with rituximab (21%; 3 hospitalizations, 1 death). Respiratory event rates did not differ between groups (Cox-regression proportional hazard ratio = 1.38, 95% CI 0.36-5.28; p = 0.64). In this retrospective study, JAKi for the treatment of RA with existing ILD or bronchiectasis did not increase the rate of hospitalization or death due to respiratory causes compared to those treated with rituximab. JAK inhibition may provide a relatively safe option for RA in such patients.
Introduction
Therapeutic decision-making in the treatment of rheumatoid arthritis (RA) for patients with concurrent pulmonary disease can be challenging. Fear regarding increased infection risk in bronchiectasis and progression of interstitial lung disease (ILD) secondary to drug administration restricts treatment options. Controversy exists as to whether medications, such as methotrexate, leflunomide and tumour necrosis factor-α inhibitors (TNFi), exacerbate pulmonary disease in patients with concurrent ILD and bronchiectasis [1]. This has led to hesitancy in utilizing these medications although systematic review of the evidence is beginning to allay fears [2, 3]. Recently, data have begun to support the safe and effective use of B and T cell therapies, rituximab [4] and abatacept [5], in the management of rheumatoid arthritis-related ILD and bronchiectasis [6, 7]. However, therapeutic options remain limited.
The approval of Janus kinase inhibitors (JAKi) (i.e., baricitinib, tofacitinib, upadacitinib) in the management of moderate to severe RA offers a new and exciting treatment paradigm [8–10]. However, inclusion of patients with concurrent ILD or bronchiectasis in relevant clinical trials is extremely limited. While the use of JAKi for the treatment of RA with co-existent lung disease is tempting due to the oral route of administration and relatively short half-life, their safety in relation to pulmonary toxicity in patients with established ILD or bronchiectasis is unknown.
To investigate this evidence gap, the objective of the current study is to compare the pulmonary safety of Janus kinase inhibition for the treatment of rheumatoid arthritis in patients with existing ILD or bronchiectasis, with rituximab. The primary outcome of interest was the incidence of severe respiratory events (hospitalization or death due to respiratory cause) in those receiving JAKi versus those receiving rituximab.
Materials and methods
Study design and patient inclusion
The study comprised a retrospective observational study. The electronic patient health record system (TrakCare®) was used to collect relevant data. Adult patients (aged 18 or over) undergoing treatment for rheumatoid arthritis at the Rheumatic Diseases Unit, Western General Hospital, Edinburgh, U.K., were included in the study. Only patients with an established diagnosis of ILD or bronchiectasis before the commencement of the drug of interest were eligible for inclusion. Patients without these respiratory conditions were excluded. Dedicated pharmacy and biologic clinic records were used to identify all adult patients (16 years or older) commencing any JAKi or rituximab for the treatment of RA from January 2016 to February 2020 (Fig. 1). From these records, all patients’ health records were screened by the investigators (OC, LK) for co-existing diagnoses of interstitial lung disease and/or bronchiectasis. For relevant patients, data were collected in detail from electronic health records by the investigators (OC, OM). Data were collected until discontinuation of the drug of interest (as documented in the clinical record) or to the end of the follow-up period (1 August 2020). They study was conducted in accordance with the Declaration of Helsinki. As all data were collected as part of routine clinical care and all treatment decisions were made prior to conduction of the study, ethical approval was not required.Fig. 1 Study outline, selection of patients and follow-up
Clinical features and demographics
Identical clinical and demographic data were collected for both treatment groups. Standard clinical variables, such as age, gender, disease duration (rheumatoid arthritis and respiratory diseases), smoking history, past and current medications, anti-CCP antibody and rheumatoid factor (IgM) positivity, were collected along with the dates of commencement and cessation for the drugs of interest. Rheumatoid arthritis disease activity (DAS28-ESR or DAS28-CRP) before the commencement of the treatment of interest was recorded along with activity scores at 3–4 months post initiation (where available).
Baseline pulmonary function tests prior to the commencement of the medications of interest were also assessed. As only a small proportion of patients had follow-up pulmonary function assessment within a reasonable time frame, this was not analysed. The pattern of ILD involvement was described as recommended by the American Thoracic Society/European Respiratory Society following review of high-resolution (HR) CT-thorax imaging by a multi-disciplinary team including chest radiologists and an expert ILD lead physician (NH) who additionally attributed a modified semi-quantitative CT-fibrosis score by visual inspection [11, 12].
Outcome measures
The primary outcome of interest was time to first respiratory event (days). Respiratory events were defined as either admission to hospital with a respiratory illness (e.g., infection, ILD exacerbation), or death from a respiratory cause while taking the medication of interest.
Additional outcomes of interest included drug continuation. Drug continuation was defined for JAKi as from the date of first prescription to the date of drug cessation, death, end of follow-up period, or whichever end-point came first. Drug continuation for rituximab was defined as from the date of first infusion to the date of cessation, death, end of follow-up period or whichever end-point came first. The number of cycles, frequency and dose of rituximab received was recorded for each patient.
Statistical analysis
Statistical analysis was performed using IBM SPSS statistics v.25.0 (IBM, Armonk, New York, USA). Distributions of clinical variables and demographics were illustrated using histograms. Depending on the data distribution and type (i.e., categorical, continuous), variables were then described using means (with standard deviations) or medians (with inter-quartile ranges). Continuous variables were compared for differences between groups using either student’s t tests or Mann–Whitney U testing subject to normality or non-normality of distribution. Proportions of categorical variables were compared using Chi-squared testing. Kaplan–Meier survival analysis (log-rank) was used to compare respiratory event survival and drug continuation survival between treatment groups. Cox regression proportional hazard was used to calculate unadjusted hazard ratios with 95% confidence intervals (CI 95%). Statistical significance was set at p < 0.05 and analysis was two -tailed.
Results
Demographic and clinical characteristics
A total of 291 patients were commenced on JAKi to February 2020 at the Rheumatic Diseases Unit, Edinburgh (Fig. 1). Twenty-eight patients commenced JAK inhibition for the treatment of rheumatoid arthritis who also had concurrent interstitial lung disease (67.9%), bronchiectasis (25%) or both (7.1%). Baricitinib was commenced in 26 patients and tofacitinib in 2 patients. A total of 318 patients were commenced on rituximab from January 2016 to February 2020 (Fig. 1). Nineteen patients commenced rituximab for the treatment of RA and had concurrent ILD (68.4%), bronchiectasis (26.3%) or both (5.3%).
Mean duration of follow-up (to discontinuation of drug, death or end of follow-up period) for patients receiving JAKi was 1.1 years (SD = 0.62) and for patients receiving rituximab infusions was 2.14 years (SD = 1). Demographic and clinical characteristics were similar in both treatment groups (Table 1); however, those commencing JAKi had a significantly longer disease duration (median 11 years) compared to those starting rituximab (median 3 years) (Table 1; Mann–Whitney U test, p = 0.003).Table 1 Demographic and clinical details of included patients
JAKi (n = 28) Rituximab (n = 19) No. of patients with available data (n =)
ILD/Bronchiectasis/both
n, (%)
ILD n = 19 (67.9%)
Bronchiectasis n = 7 (25%)
Both n = 2 (7.1%)
ILD n = 13 (68.4%)
Bronchiectasis n = 5 (26.3%)
Both n = 1 (5.3%)
47
Age (median, IQR) 69 (62.3–75) 70 (59–76) 47
Female n, (%) 18 (64.3%) 14 (73.7%) 47
Smoking status
n, (%)
Smoker:
n = 5 (17.9%)
Ex-smoker:
n = 15 (53.6%)
Never-smoker:
n = 8 (28.6%)
Smoker:
n = 0 (0%),
Ex-smoker:
n = 12 (63.2%)
Never-smoker:
n = 7 (36.8%)
47
Anti-CCP positivity
n, (%)
25 (96.2%) 19 (100%) 45
Rh-Factor IgM positivity
n, (%)
14 (77.8%) 9 (90%) 28
RA diagnosis age median, (IQR) 53 (44–61) 54 (48–67) 42
Respiratory diagnosis age
median, (IQR)
64 (55–73) 64 (53–71) 46
Years from RA diagnosis to drug of interest
median, (IQR)
11 (3–20) 3 (1–6) 42
DAS-28 Baseline
median, (IQR)
5.85 (4.98–6.18) 6.2 (5.53–6.86) 43
DAS-28 3–4 months
median, (IQR)
3.4 (2.38–4.11) 4.2 (3.61–5.44) 32
JAKi Janus kinase inhibitor, ILD interstitial lung disease, Rh-factor rheumatoid factor, RA rheumatoid arthritis, DAS-28 disease activity score 28 joints
Pre-treatment pulmonary function tests indicated mild severity of pulmonary disease in patients commencing JAKi and rituximab (Table 2). Review of patients’ baseline high-resolution CT-thorax revealed similar radiological patterns and severity of ILD in both treatment groups (Table 2). However, a greater degree of ground glass was evident on CT-imaging before treatment in those receiving JAKi compared to those receiving rituximab (p = 0.036).Table 2 Baseline pulmonary function (all patients); and radiological pattern and severity of ILD on CT-thorax imaging (patients with ILD)
JAKi (n = 28) Rituximab (n = 19) No. of patients with available data (n =)
Baseline predicted FEV1 median %, (IQR) 93.5 (74.5–105) 92 (76–105) 43
Baseline predicted FVC median %, (IQR) 103 (78–112) 99 (86–119) 44
Baseline predicted TLCO median %, (IQR) 62 (51.25–69) 62 (55–67) 41
Radiological ILD-pattern JAKi (n = 20) Rituximab (n = 14)
UIP
n, (%)
0 (0%) 2 (14%)
Probable UIP
n, (%)
4 (20%) 5 (36%)
Fibrotic NSIP
n, (%)
6 (30%) 1 (7%)
Chronic HP
n, (%)
3 (15%) 1 (7%)
Respiratory bronchiolitis-ILD
n, (%)
1 (5%) 0 (0%)
Unclassifiable
n, (%)
6 (30%) 5 (36%)
% of lung with ILD
median, (IQR)
20 (10–45) 12.5 (10–30)
% of lung with ground glass appearance
median, (IQR)
20 (5–45) 7.5 (5–10)
% of lung with honeycombing
median, (IQR)
0 (0–1.75) 0 (0–2)
% of lung with emphysema
median, (IQR)
0.5 (0–8.75) 0 (0–3.75)
JAKi Janus kinase inhibitor, ILD interstitial lung disease, FEV1 forced expiratory volume in 1 second, FVC forced vital capacity, TLco lung transfer capacity for carbon monoxide, IQR inter-quartile range, UIP usual interstitial pneumonia, NSIP non-specific interstitial pneumonia, HP hypersensitivity pneumonitis
Treatment details
Concurrent treatments including conventional synthetic DMARDs and corticosteroids were similar in both groups (Table 3). Almost 30% of patients were biologic naïve before commencing the drug of interest. Prior treatment history indicated that those commencing JAKi had previously been treated with a greater number of biologics including TNFi, tocilizumab and abatacept, compared to those commencing rituximab.Table 3 Previous and concurrent DMARD treatments
JAKi (n = 28) Rituximab (n = 19)
Previous treatments
No. of previous csDMARDs
median, (IQR)
3 (2–4) 3 (2–4)
Methotrexate
n, (%)
22 (78.6%) 13 (68.4%)
Biologic Naïve
n, (%)
8 (28.6%) 6 (31.6%)
No. of previous biologics
median, (IQR)
1 (0–2) 0 (0–1)
Rituximab/JAKi
n, (%)
12 (42.9%) 1 (5.3%)
TNF-α inhibitor
n, (%)
15 (53.6%) 5 (26.3%)
Tocilizumab
n, (%)
6 (21.4%) 1 (5.3%)
Abatacept
n, (%)
6 (21.4%) 1 (5.3%)
Concurrent treatments
Prednisolone; n, (%) 13 (46.4%) 6 (31.6%)
Prednisolone dose (mg) median, (IQR) 7 (5–10) 10 (4.5–12.5)
Methotrexate
n, (%)
4 (14.3%) 2 (10.5%)
Azathioprine
n, (%)
2 (7.1%) 2 (10.5%)
Leflunomide
n, (%)
2 (7.1%) 2 (10.5%)
Hydroxychloroquine
n, (%)
6 (21.4%) 6 (31.6%)
Sulfasalazine
n, (%)
3 (10.7%) 4 (21.1%)
DMARD disease-modifying anti-rheumatic drugs, csDMARDs conventional synthetic DMARDs, JAKi Janus kinase inhibitor, IQR inter-quartile range, TNF Tumour necrosis factor
All except for two patients received standard licensed dosing regimens for JAK inhibition for the treatment of RA (i.e. baricitinib 4 mg once daily; tofacitinib 5 mg twice daily). One patient took 4 mg/2 mg of baricitinib on alternate days and one patient took 5 mg of tofacitinib once daily before increasing to 5 mg twice daily.
All patients who received rituximab received the standard dosing of treatment (2 × intravenous infusions of 1000 mg rituximab with 100 mg intravenous methylprednisolone 14 days apart). Only 2/19 patients were on ‘fixed’ cycles of rituximab therapy repeated every 6 months. All others received rituximab as required depending on clinical status but not within 6 months of the last infusion. 5/19 patients received only 1 cycle of rituximab during their study follow-up period. All others (74%) received between 2 and 5 cycles of rituximab.
Incidence of respiratory events
Respiratory events (i.e. hospitalization or death due to a respiratory cause) occurred in five patients (18%) during JAKi treatment (7 hospitalizations, two of which led to death) and in four patients (21%) (4 hospitalizations, one of which led to death) during treatment with rituximab. In the JAKi group, two patients (7.1%) died of respiratory issues (one new lung cancer with concomitant community acquired pneumonia and one of community acquired pneumonia). In the rituximab-treated patients, two patients (10.5%) died during follow-up; one of an acute myocardial infarction and one of pneumonitis and respiratory failure. No episodes of acute pneumonitis occurred in the JAKi group. There was no difference in respiratory event survival in the JAKi-treated group compared with the rituximab-treated group (Fig. 2a); unadjusted hazard ratio: 1.38 (95% CI 0.36, 5.28); p = 0.64.Fig. 2 a Comparison of time to respiratory event and b drug continuation between JAKi and rituximab for the treatment of rheumatoid arthritis in patients with concurrent bronchiectasis and/or ILD
Drug continuation
Seventy-one percent (20/28) of patients with ILD and/or bronchiectasis who commenced JAKi for the treatment of rheumatoid arthritis remained on the drug at the end of the follow-up period. Reasons for discontinuation included inefficacy (n = 3), death (n = 2, as outlined above), recurrent infections (n = 2) and reduced renal function (n = 1).
Sixty-three percent (12/19) of patients who commenced rituximab continued the drug to the end of the follow-up period. Reasons for cessation included inefficacy (n = 5) or death (n = 2, as outlined above). The rate of discontinuation of JAKi in patients with ILD and/or bronchiectasis did not significantly differ from that of those receiving rituximab (Fig. 2b); unadjusted hazard ratio: 1.9 (95% CI 0.634, 5.73); p = 0.251.
Discussion
There is an absence of patients with interstitial lung disease and bronchiectasis in randomised controlled trials that examine the effectiveness of JAK inhibitors for the treatment of rheumatoid arthritis. Therefore, we have limited knowledge of the pulmonary safety in these cohorts. Prevalence of interstitial lung disease and bronchiectasis is relatively high in patients with rheumatoid arthritis, estimated to affect between 5 and 10% of patients [1, 13]. Interstitial lung disease itself contributes significantly to the excess mortality evident in patients with rheumatoid arthritis [14]. Despite this, the optimal treatment strategy for such patients remains unknown, reflecting a paucity of randomized controlled clinical trials for this patient sub-group. Recently, observational data have suggested treatment with rituximab [6, 15, 16], and laterally abatacept [5, 17] is the most effective and safest biologic agents for the treatment of RA with concurrent ILD or bronchiectasis. Data suggest that treatment with rituximab and abatacept may even infer a benefit to pulmonary function [5, 6]. Nonetheless, concern about lung toxicity restricts the use of many DMARDs and biologic agents, be these concerns soundly proven or not.
Novel, small-molecule treatments, such as Janus kinase inhibitors, represent an exciting option for patients with rheumatoid arthritis. Their safety and potential for effectiveness in patients with concurrent pulmonary disease should be duly considered. Currently, our understanding of inhibition of JAK/STAT signalling pathways in patients with ILD and bronchiectasis is extremely limited. JAKi-use has led to successful treatment of dermatomyositis-associated ILD but evidence is restricted to a small number of case reports and case series [18–20]. Even less is understood about Janus kinase inhibition in patients with bronchiectasis. While post hoc analysis of phase 2, 3 and 3b/4 clinical trials indicates that the de novo incidence of interstitial lung disease is low in patients with rheumatoid arthritis treated with JAKi [21]; to our knowledge, no study has yet examined the pulmonary tolerance and safety of JAKi in those with existing ILD or bronchiectasis. The current study sought to address this important clinical scenario, by comparing time to respiratory events in patients receiving JAKi with patients receiving the current, best-practice medication for this clinical cohort, rituximab.
The majority of our patients commenced JAK1/2 inhibition with baricitinib due to the earlier licensing of this medication for the treatment of rheumatoid arthritis but several patients commenced JAK1/3 inhibition with tofacitinib. Furthermore, the proportion of patients with ‘difficult-to-treat’ rheumatoid disease, manifest by a longer duration of illness and a greater number of previous biologic therapies (e.g. tocilizumab, abatacept), was higher in those subsequently prescribed JAKi. We believe this relates to the fact that a novel medication provides an attractive option for patients who have had difficult-to-treat disease for many years and in such cases patients often chose to switch therapy in an attempt to achieve remission.
Encouragingly, serious respiratory events (i.e., hospitalization or death due to a respiratory cause) were no different between those treated with JAKi and those treated with rituximab suggesting that JAKi may offer a realistic treatment option for patients with RA and concurrent ILD or bronchiectasis. Our cohort of patients represents a group with mild to moderate pulmonary disease, all of whom commenced Janus kinase inhibition or rituximab with a known, pre-existing diagnosis of ILD or bronchiectasis (present on cross sectional CT-imaging and/or attending a tertiary ILD clinic). Very few patients with ILD had a HR-CT pattern consistent with definite UIP and most had a pattern that was consistent with an alternative diagnosis and were highly unlikely to be UIP. Therefore, our findings should not be extrapolated to those with severe ILD or bronchiectasis or a definite UIP pattern of disease and caution is urged in this setting. Furthermore, there was no significant difference in time to drug discontinuation between those prescribed JAKi and those prescribed rituximab. Proportionally, more patients treated with rituximab (almost 40%) ceased treatment during follow-up compared with those treated with JAKi (almost 30%). Reasons for discontinuation of rituximab predominantly related to clinical inefficacy, but comparison of clinical effectiveness was not the aim of this study.
There are several strengths to this study. Both cohorts of patients (i.e., JAKi- and rituximab-receiving) were similar in terms of clinical traits (e.g. gender, concurrent medications, corticosteroid use, anti-CCP and rheumatoid factor seropositivity) minimizing the effect of confounders on respiratory event outcomes. In addition, these data represent real-world experience of patients with RA and concurrent ILD/bronchiectasis with patient–clinician decisions made without influence of a pre-determined research goal. The retrospective nature of this observational study and associated risk of information bias represents the major limitation of this study. The relatively small sample size and short follow-up period also represent limitations that may have affected overall results. Ideally, follow-up pulmonary function testing would have been available for all patients after commencing treatment but as this was only the case for a minority of patients, this was not analysed. It was not possible to collect primary care records of respiratory exacerbations or infections that did not require hospital admission that were treated in the community. Furthermore, the electronic patient health record system used did not permit recording of admissions to hospital abroad or to hospitals outside of NHS Lothian. The authors recommend that the findings of this study and analysis should be investigated further in larger prospective cohorts in the future.
To conclude, within the limitations of this retrospective analysis, JAK inhibition with baricitinib or tofacitinib for the treatment of rheumatoid arthritis in patients with concurrent interstitial lung disease or bronchiectasis did not increase the rate of hospitalization or death due to a respiratory cause in comparison to those treated with rituximab. JAK inhibition may provide a favourable and safe therapeutic option for such patients but larger, prospective studies are required to confirm these findings to inform much-needed therapeutic guidance in this sub-category of patients with RA and ILD/bronchiectasis.
Funding
No specific funding was received from any bodies in the public, commercial or not-for-profit sectors to carry out the work described in this article.
Declarations
Conflict of interest
OC has received sponsorship from Lilly and Pfizer to attend educational conferences. The other authors have no conflicts of interest to declare.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. | Fatal | ReactionOutcome | CC BY | 33721094 | 19,059,614 | 2021-05 |
What was the outcome of reaction 'Respiratory failure'? | A retrospective comparison of respiratory events with JAK inhibitors or rituximab for rheumatoid arthritis in patients with pulmonary disease.
Janus kinase inhibitors (JAKi) are an exciting option for the treatment of rheumatoid arthritis (RA) but little is known about their safety and tolerability in patients with existing respiratory disorders. The objective was to compare pulmonary safety of JAKi versus rituximab in patients with concurrent interstitial lung disease (ILD) or bronchiectasis. We performed a retrospective electronic patient record review of patients with known ILD or bronchiectasis commencing JAKi or rituximab for the treatment of RA. Patients initiating treatment from January 2016 to February 2020 were included. Respiratory events (hospitalization or death from a respiratory cause) were compared using Kaplan-Meier survival analysis. We analysed patients who received JAKi (n = 28) and rituximab (n = 19) for a mean (SD) of 1.1 (0.62) and 2.14 (1) years respectively. Patients were predominantly female (68%), anti-CCP antibody positive (94%) and non-smoking (89%) with a median (IQR) percentage predicted FVC at baseline of 100% (82-115%) and percentage predicted TLCO of 62% (54.5-68%). Respiratory events occurred in five patients treated with JAKi (18%; 5 hospitalizations, 2 deaths) and in four patients treated with rituximab (21%; 3 hospitalizations, 1 death). Respiratory event rates did not differ between groups (Cox-regression proportional hazard ratio = 1.38, 95% CI 0.36-5.28; p = 0.64). In this retrospective study, JAKi for the treatment of RA with existing ILD or bronchiectasis did not increase the rate of hospitalization or death due to respiratory causes compared to those treated with rituximab. JAK inhibition may provide a relatively safe option for RA in such patients.
Introduction
Therapeutic decision-making in the treatment of rheumatoid arthritis (RA) for patients with concurrent pulmonary disease can be challenging. Fear regarding increased infection risk in bronchiectasis and progression of interstitial lung disease (ILD) secondary to drug administration restricts treatment options. Controversy exists as to whether medications, such as methotrexate, leflunomide and tumour necrosis factor-α inhibitors (TNFi), exacerbate pulmonary disease in patients with concurrent ILD and bronchiectasis [1]. This has led to hesitancy in utilizing these medications although systematic review of the evidence is beginning to allay fears [2, 3]. Recently, data have begun to support the safe and effective use of B and T cell therapies, rituximab [4] and abatacept [5], in the management of rheumatoid arthritis-related ILD and bronchiectasis [6, 7]. However, therapeutic options remain limited.
The approval of Janus kinase inhibitors (JAKi) (i.e., baricitinib, tofacitinib, upadacitinib) in the management of moderate to severe RA offers a new and exciting treatment paradigm [8–10]. However, inclusion of patients with concurrent ILD or bronchiectasis in relevant clinical trials is extremely limited. While the use of JAKi for the treatment of RA with co-existent lung disease is tempting due to the oral route of administration and relatively short half-life, their safety in relation to pulmonary toxicity in patients with established ILD or bronchiectasis is unknown.
To investigate this evidence gap, the objective of the current study is to compare the pulmonary safety of Janus kinase inhibition for the treatment of rheumatoid arthritis in patients with existing ILD or bronchiectasis, with rituximab. The primary outcome of interest was the incidence of severe respiratory events (hospitalization or death due to respiratory cause) in those receiving JAKi versus those receiving rituximab.
Materials and methods
Study design and patient inclusion
The study comprised a retrospective observational study. The electronic patient health record system (TrakCare®) was used to collect relevant data. Adult patients (aged 18 or over) undergoing treatment for rheumatoid arthritis at the Rheumatic Diseases Unit, Western General Hospital, Edinburgh, U.K., were included in the study. Only patients with an established diagnosis of ILD or bronchiectasis before the commencement of the drug of interest were eligible for inclusion. Patients without these respiratory conditions were excluded. Dedicated pharmacy and biologic clinic records were used to identify all adult patients (16 years or older) commencing any JAKi or rituximab for the treatment of RA from January 2016 to February 2020 (Fig. 1). From these records, all patients’ health records were screened by the investigators (OC, LK) for co-existing diagnoses of interstitial lung disease and/or bronchiectasis. For relevant patients, data were collected in detail from electronic health records by the investigators (OC, OM). Data were collected until discontinuation of the drug of interest (as documented in the clinical record) or to the end of the follow-up period (1 August 2020). They study was conducted in accordance with the Declaration of Helsinki. As all data were collected as part of routine clinical care and all treatment decisions were made prior to conduction of the study, ethical approval was not required.Fig. 1 Study outline, selection of patients and follow-up
Clinical features and demographics
Identical clinical and demographic data were collected for both treatment groups. Standard clinical variables, such as age, gender, disease duration (rheumatoid arthritis and respiratory diseases), smoking history, past and current medications, anti-CCP antibody and rheumatoid factor (IgM) positivity, were collected along with the dates of commencement and cessation for the drugs of interest. Rheumatoid arthritis disease activity (DAS28-ESR or DAS28-CRP) before the commencement of the treatment of interest was recorded along with activity scores at 3–4 months post initiation (where available).
Baseline pulmonary function tests prior to the commencement of the medications of interest were also assessed. As only a small proportion of patients had follow-up pulmonary function assessment within a reasonable time frame, this was not analysed. The pattern of ILD involvement was described as recommended by the American Thoracic Society/European Respiratory Society following review of high-resolution (HR) CT-thorax imaging by a multi-disciplinary team including chest radiologists and an expert ILD lead physician (NH) who additionally attributed a modified semi-quantitative CT-fibrosis score by visual inspection [11, 12].
Outcome measures
The primary outcome of interest was time to first respiratory event (days). Respiratory events were defined as either admission to hospital with a respiratory illness (e.g., infection, ILD exacerbation), or death from a respiratory cause while taking the medication of interest.
Additional outcomes of interest included drug continuation. Drug continuation was defined for JAKi as from the date of first prescription to the date of drug cessation, death, end of follow-up period, or whichever end-point came first. Drug continuation for rituximab was defined as from the date of first infusion to the date of cessation, death, end of follow-up period or whichever end-point came first. The number of cycles, frequency and dose of rituximab received was recorded for each patient.
Statistical analysis
Statistical analysis was performed using IBM SPSS statistics v.25.0 (IBM, Armonk, New York, USA). Distributions of clinical variables and demographics were illustrated using histograms. Depending on the data distribution and type (i.e., categorical, continuous), variables were then described using means (with standard deviations) or medians (with inter-quartile ranges). Continuous variables were compared for differences between groups using either student’s t tests or Mann–Whitney U testing subject to normality or non-normality of distribution. Proportions of categorical variables were compared using Chi-squared testing. Kaplan–Meier survival analysis (log-rank) was used to compare respiratory event survival and drug continuation survival between treatment groups. Cox regression proportional hazard was used to calculate unadjusted hazard ratios with 95% confidence intervals (CI 95%). Statistical significance was set at p < 0.05 and analysis was two -tailed.
Results
Demographic and clinical characteristics
A total of 291 patients were commenced on JAKi to February 2020 at the Rheumatic Diseases Unit, Edinburgh (Fig. 1). Twenty-eight patients commenced JAK inhibition for the treatment of rheumatoid arthritis who also had concurrent interstitial lung disease (67.9%), bronchiectasis (25%) or both (7.1%). Baricitinib was commenced in 26 patients and tofacitinib in 2 patients. A total of 318 patients were commenced on rituximab from January 2016 to February 2020 (Fig. 1). Nineteen patients commenced rituximab for the treatment of RA and had concurrent ILD (68.4%), bronchiectasis (26.3%) or both (5.3%).
Mean duration of follow-up (to discontinuation of drug, death or end of follow-up period) for patients receiving JAKi was 1.1 years (SD = 0.62) and for patients receiving rituximab infusions was 2.14 years (SD = 1). Demographic and clinical characteristics were similar in both treatment groups (Table 1); however, those commencing JAKi had a significantly longer disease duration (median 11 years) compared to those starting rituximab (median 3 years) (Table 1; Mann–Whitney U test, p = 0.003).Table 1 Demographic and clinical details of included patients
JAKi (n = 28) Rituximab (n = 19) No. of patients with available data (n =)
ILD/Bronchiectasis/both
n, (%)
ILD n = 19 (67.9%)
Bronchiectasis n = 7 (25%)
Both n = 2 (7.1%)
ILD n = 13 (68.4%)
Bronchiectasis n = 5 (26.3%)
Both n = 1 (5.3%)
47
Age (median, IQR) 69 (62.3–75) 70 (59–76) 47
Female n, (%) 18 (64.3%) 14 (73.7%) 47
Smoking status
n, (%)
Smoker:
n = 5 (17.9%)
Ex-smoker:
n = 15 (53.6%)
Never-smoker:
n = 8 (28.6%)
Smoker:
n = 0 (0%),
Ex-smoker:
n = 12 (63.2%)
Never-smoker:
n = 7 (36.8%)
47
Anti-CCP positivity
n, (%)
25 (96.2%) 19 (100%) 45
Rh-Factor IgM positivity
n, (%)
14 (77.8%) 9 (90%) 28
RA diagnosis age median, (IQR) 53 (44–61) 54 (48–67) 42
Respiratory diagnosis age
median, (IQR)
64 (55–73) 64 (53–71) 46
Years from RA diagnosis to drug of interest
median, (IQR)
11 (3–20) 3 (1–6) 42
DAS-28 Baseline
median, (IQR)
5.85 (4.98–6.18) 6.2 (5.53–6.86) 43
DAS-28 3–4 months
median, (IQR)
3.4 (2.38–4.11) 4.2 (3.61–5.44) 32
JAKi Janus kinase inhibitor, ILD interstitial lung disease, Rh-factor rheumatoid factor, RA rheumatoid arthritis, DAS-28 disease activity score 28 joints
Pre-treatment pulmonary function tests indicated mild severity of pulmonary disease in patients commencing JAKi and rituximab (Table 2). Review of patients’ baseline high-resolution CT-thorax revealed similar radiological patterns and severity of ILD in both treatment groups (Table 2). However, a greater degree of ground glass was evident on CT-imaging before treatment in those receiving JAKi compared to those receiving rituximab (p = 0.036).Table 2 Baseline pulmonary function (all patients); and radiological pattern and severity of ILD on CT-thorax imaging (patients with ILD)
JAKi (n = 28) Rituximab (n = 19) No. of patients with available data (n =)
Baseline predicted FEV1 median %, (IQR) 93.5 (74.5–105) 92 (76–105) 43
Baseline predicted FVC median %, (IQR) 103 (78–112) 99 (86–119) 44
Baseline predicted TLCO median %, (IQR) 62 (51.25–69) 62 (55–67) 41
Radiological ILD-pattern JAKi (n = 20) Rituximab (n = 14)
UIP
n, (%)
0 (0%) 2 (14%)
Probable UIP
n, (%)
4 (20%) 5 (36%)
Fibrotic NSIP
n, (%)
6 (30%) 1 (7%)
Chronic HP
n, (%)
3 (15%) 1 (7%)
Respiratory bronchiolitis-ILD
n, (%)
1 (5%) 0 (0%)
Unclassifiable
n, (%)
6 (30%) 5 (36%)
% of lung with ILD
median, (IQR)
20 (10–45) 12.5 (10–30)
% of lung with ground glass appearance
median, (IQR)
20 (5–45) 7.5 (5–10)
% of lung with honeycombing
median, (IQR)
0 (0–1.75) 0 (0–2)
% of lung with emphysema
median, (IQR)
0.5 (0–8.75) 0 (0–3.75)
JAKi Janus kinase inhibitor, ILD interstitial lung disease, FEV1 forced expiratory volume in 1 second, FVC forced vital capacity, TLco lung transfer capacity for carbon monoxide, IQR inter-quartile range, UIP usual interstitial pneumonia, NSIP non-specific interstitial pneumonia, HP hypersensitivity pneumonitis
Treatment details
Concurrent treatments including conventional synthetic DMARDs and corticosteroids were similar in both groups (Table 3). Almost 30% of patients were biologic naïve before commencing the drug of interest. Prior treatment history indicated that those commencing JAKi had previously been treated with a greater number of biologics including TNFi, tocilizumab and abatacept, compared to those commencing rituximab.Table 3 Previous and concurrent DMARD treatments
JAKi (n = 28) Rituximab (n = 19)
Previous treatments
No. of previous csDMARDs
median, (IQR)
3 (2–4) 3 (2–4)
Methotrexate
n, (%)
22 (78.6%) 13 (68.4%)
Biologic Naïve
n, (%)
8 (28.6%) 6 (31.6%)
No. of previous biologics
median, (IQR)
1 (0–2) 0 (0–1)
Rituximab/JAKi
n, (%)
12 (42.9%) 1 (5.3%)
TNF-α inhibitor
n, (%)
15 (53.6%) 5 (26.3%)
Tocilizumab
n, (%)
6 (21.4%) 1 (5.3%)
Abatacept
n, (%)
6 (21.4%) 1 (5.3%)
Concurrent treatments
Prednisolone; n, (%) 13 (46.4%) 6 (31.6%)
Prednisolone dose (mg) median, (IQR) 7 (5–10) 10 (4.5–12.5)
Methotrexate
n, (%)
4 (14.3%) 2 (10.5%)
Azathioprine
n, (%)
2 (7.1%) 2 (10.5%)
Leflunomide
n, (%)
2 (7.1%) 2 (10.5%)
Hydroxychloroquine
n, (%)
6 (21.4%) 6 (31.6%)
Sulfasalazine
n, (%)
3 (10.7%) 4 (21.1%)
DMARD disease-modifying anti-rheumatic drugs, csDMARDs conventional synthetic DMARDs, JAKi Janus kinase inhibitor, IQR inter-quartile range, TNF Tumour necrosis factor
All except for two patients received standard licensed dosing regimens for JAK inhibition for the treatment of RA (i.e. baricitinib 4 mg once daily; tofacitinib 5 mg twice daily). One patient took 4 mg/2 mg of baricitinib on alternate days and one patient took 5 mg of tofacitinib once daily before increasing to 5 mg twice daily.
All patients who received rituximab received the standard dosing of treatment (2 × intravenous infusions of 1000 mg rituximab with 100 mg intravenous methylprednisolone 14 days apart). Only 2/19 patients were on ‘fixed’ cycles of rituximab therapy repeated every 6 months. All others received rituximab as required depending on clinical status but not within 6 months of the last infusion. 5/19 patients received only 1 cycle of rituximab during their study follow-up period. All others (74%) received between 2 and 5 cycles of rituximab.
Incidence of respiratory events
Respiratory events (i.e. hospitalization or death due to a respiratory cause) occurred in five patients (18%) during JAKi treatment (7 hospitalizations, two of which led to death) and in four patients (21%) (4 hospitalizations, one of which led to death) during treatment with rituximab. In the JAKi group, two patients (7.1%) died of respiratory issues (one new lung cancer with concomitant community acquired pneumonia and one of community acquired pneumonia). In the rituximab-treated patients, two patients (10.5%) died during follow-up; one of an acute myocardial infarction and one of pneumonitis and respiratory failure. No episodes of acute pneumonitis occurred in the JAKi group. There was no difference in respiratory event survival in the JAKi-treated group compared with the rituximab-treated group (Fig. 2a); unadjusted hazard ratio: 1.38 (95% CI 0.36, 5.28); p = 0.64.Fig. 2 a Comparison of time to respiratory event and b drug continuation between JAKi and rituximab for the treatment of rheumatoid arthritis in patients with concurrent bronchiectasis and/or ILD
Drug continuation
Seventy-one percent (20/28) of patients with ILD and/or bronchiectasis who commenced JAKi for the treatment of rheumatoid arthritis remained on the drug at the end of the follow-up period. Reasons for discontinuation included inefficacy (n = 3), death (n = 2, as outlined above), recurrent infections (n = 2) and reduced renal function (n = 1).
Sixty-three percent (12/19) of patients who commenced rituximab continued the drug to the end of the follow-up period. Reasons for cessation included inefficacy (n = 5) or death (n = 2, as outlined above). The rate of discontinuation of JAKi in patients with ILD and/or bronchiectasis did not significantly differ from that of those receiving rituximab (Fig. 2b); unadjusted hazard ratio: 1.9 (95% CI 0.634, 5.73); p = 0.251.
Discussion
There is an absence of patients with interstitial lung disease and bronchiectasis in randomised controlled trials that examine the effectiveness of JAK inhibitors for the treatment of rheumatoid arthritis. Therefore, we have limited knowledge of the pulmonary safety in these cohorts. Prevalence of interstitial lung disease and bronchiectasis is relatively high in patients with rheumatoid arthritis, estimated to affect between 5 and 10% of patients [1, 13]. Interstitial lung disease itself contributes significantly to the excess mortality evident in patients with rheumatoid arthritis [14]. Despite this, the optimal treatment strategy for such patients remains unknown, reflecting a paucity of randomized controlled clinical trials for this patient sub-group. Recently, observational data have suggested treatment with rituximab [6, 15, 16], and laterally abatacept [5, 17] is the most effective and safest biologic agents for the treatment of RA with concurrent ILD or bronchiectasis. Data suggest that treatment with rituximab and abatacept may even infer a benefit to pulmonary function [5, 6]. Nonetheless, concern about lung toxicity restricts the use of many DMARDs and biologic agents, be these concerns soundly proven or not.
Novel, small-molecule treatments, such as Janus kinase inhibitors, represent an exciting option for patients with rheumatoid arthritis. Their safety and potential for effectiveness in patients with concurrent pulmonary disease should be duly considered. Currently, our understanding of inhibition of JAK/STAT signalling pathways in patients with ILD and bronchiectasis is extremely limited. JAKi-use has led to successful treatment of dermatomyositis-associated ILD but evidence is restricted to a small number of case reports and case series [18–20]. Even less is understood about Janus kinase inhibition in patients with bronchiectasis. While post hoc analysis of phase 2, 3 and 3b/4 clinical trials indicates that the de novo incidence of interstitial lung disease is low in patients with rheumatoid arthritis treated with JAKi [21]; to our knowledge, no study has yet examined the pulmonary tolerance and safety of JAKi in those with existing ILD or bronchiectasis. The current study sought to address this important clinical scenario, by comparing time to respiratory events in patients receiving JAKi with patients receiving the current, best-practice medication for this clinical cohort, rituximab.
The majority of our patients commenced JAK1/2 inhibition with baricitinib due to the earlier licensing of this medication for the treatment of rheumatoid arthritis but several patients commenced JAK1/3 inhibition with tofacitinib. Furthermore, the proportion of patients with ‘difficult-to-treat’ rheumatoid disease, manifest by a longer duration of illness and a greater number of previous biologic therapies (e.g. tocilizumab, abatacept), was higher in those subsequently prescribed JAKi. We believe this relates to the fact that a novel medication provides an attractive option for patients who have had difficult-to-treat disease for many years and in such cases patients often chose to switch therapy in an attempt to achieve remission.
Encouragingly, serious respiratory events (i.e., hospitalization or death due to a respiratory cause) were no different between those treated with JAKi and those treated with rituximab suggesting that JAKi may offer a realistic treatment option for patients with RA and concurrent ILD or bronchiectasis. Our cohort of patients represents a group with mild to moderate pulmonary disease, all of whom commenced Janus kinase inhibition or rituximab with a known, pre-existing diagnosis of ILD or bronchiectasis (present on cross sectional CT-imaging and/or attending a tertiary ILD clinic). Very few patients with ILD had a HR-CT pattern consistent with definite UIP and most had a pattern that was consistent with an alternative diagnosis and were highly unlikely to be UIP. Therefore, our findings should not be extrapolated to those with severe ILD or bronchiectasis or a definite UIP pattern of disease and caution is urged in this setting. Furthermore, there was no significant difference in time to drug discontinuation between those prescribed JAKi and those prescribed rituximab. Proportionally, more patients treated with rituximab (almost 40%) ceased treatment during follow-up compared with those treated with JAKi (almost 30%). Reasons for discontinuation of rituximab predominantly related to clinical inefficacy, but comparison of clinical effectiveness was not the aim of this study.
There are several strengths to this study. Both cohorts of patients (i.e., JAKi- and rituximab-receiving) were similar in terms of clinical traits (e.g. gender, concurrent medications, corticosteroid use, anti-CCP and rheumatoid factor seropositivity) minimizing the effect of confounders on respiratory event outcomes. In addition, these data represent real-world experience of patients with RA and concurrent ILD/bronchiectasis with patient–clinician decisions made without influence of a pre-determined research goal. The retrospective nature of this observational study and associated risk of information bias represents the major limitation of this study. The relatively small sample size and short follow-up period also represent limitations that may have affected overall results. Ideally, follow-up pulmonary function testing would have been available for all patients after commencing treatment but as this was only the case for a minority of patients, this was not analysed. It was not possible to collect primary care records of respiratory exacerbations or infections that did not require hospital admission that were treated in the community. Furthermore, the electronic patient health record system used did not permit recording of admissions to hospital abroad or to hospitals outside of NHS Lothian. The authors recommend that the findings of this study and analysis should be investigated further in larger prospective cohorts in the future.
To conclude, within the limitations of this retrospective analysis, JAK inhibition with baricitinib or tofacitinib for the treatment of rheumatoid arthritis in patients with concurrent interstitial lung disease or bronchiectasis did not increase the rate of hospitalization or death due to a respiratory cause in comparison to those treated with rituximab. JAK inhibition may provide a favourable and safe therapeutic option for such patients but larger, prospective studies are required to confirm these findings to inform much-needed therapeutic guidance in this sub-category of patients with RA and ILD/bronchiectasis.
Funding
No specific funding was received from any bodies in the public, commercial or not-for-profit sectors to carry out the work described in this article.
Declarations
Conflict of interest
OC has received sponsorship from Lilly and Pfizer to attend educational conferences. The other authors have no conflicts of interest to declare.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. | Fatal | ReactionOutcome | CC BY | 33721094 | 19,059,614 | 2021-05 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Bradypnoea'. | Extracorporeal cardiopulmonary resuscitation for severe chloroquine intoxication in a child - a case report.
BACKGROUND
Chloroquine use has increased worldwide recently in the setting of experimental treatment for the novel coronavirus disease (Covid-19). Nevertheless, in case of chloroquine intoxication, it can be life threatening, with cardiac arrest, due to its cardiac toxicity.
METHODS
This case study reports on a 14-years-old girl who presented in cardiac arrest after an uncommon suicide attempt by ingesting 3 g of chloroquine. After 66 min of cardio-pulmonary resuscitation (CPR), extracorporeal cardiopulmonary resuscitation (ECPR) was initiated, allowing cardiac function to recover.
CONCLUSIONS
Chloroquine intoxication is a rare but serious condition due to its cardiac toxicity. Use of ECPR in this case of transient toxicity allowed a favorable evolution with little neurological impairment.
Background
Chloroquine is a worldwide recognized treatment for varied conditions such as malaria, rheumatoid arthritis and lupus erythematosus. Its use has recently been significantly expanded as an experimental treatment for COVID-19. Overdose however, can lead to life threatening complications [1], and with the lack of a specific antidote, treatment remains supportive, extracorporeal membrane oxygenation (ECMO) being the last resort. We report the case of a teenager who suffered cardiac arrest due to chloroquine, and who survived to ECMO decannulation. To our knowledge, this is the first published pediatric case of chloroquine intoxication requiring extracorporeal cardiopulmonary resuscitation (ECPR).
Case presentation
A 14-years-old girl, with a weight of 68 kg and a history of suicidal thoughts presented in cardiopulmonary arrest following a suicide attempt with ingestion of 3 gram of chloroquine without history of any other substance ingestion. Upon the arrival of the ambulance, she was unconscious with a Glasgow coma scale score (GCS) of 6 (1–1–4), bradypneic and with a trismus. The electrocardiogram (ECG) showed ST-segment changes, with pulseless ventricular tachycardia immediately following. Cardio-pulmonary resuscitation was initiated but despite chest compressions, defibrillation and adrenaline, there was no return to spontaneous circulation. Twelve minutes later, upon arrival at the hospital, she had a pulseless idioventricular rhythm and cardiopulmonary resuscitation was continued. The patient was intubated and external massage was taken over by a Lucas® chest compression system (Stryker Medical, Portage, MI49002 US).
Blood gas revealed severe hypokalemia (potassium 1.8 mmol/l) and a severe mixed acidosis (pH 6.97, pCO2 95 mmHg, glucose 14.5 mmol/l, lactate 7.4 mmol/l, base excess − 9.7 mmol/l, bicarbonate 12.7 mmol/l). Resuscitation was continued with three additional doses of adrenaline followed by continuous adrenaline infusion as well as correction of hypokalemia, administration of an amiodarone bolus, a bicarbonate bolus, fluid resuscitation and intravenous lipid emulsion infusion. Despite these measures, resuscitation was unsuccessful.
Peripheral right femoral veno-arterial extracorporeal membrane oxygenation (ECMO)ʼ was therefore initiated after 66 minutes of resuscitation to provide full cardiac support. During transfer to our pediatric intensive care unit (PICU), she was sedated and paralyzed, ventilated, hemodynamically stable on ECMO with a blood flow of 1.7 L/min/m2 and an adrenaline infusion of 0.1mcg/kg/min, a temperature of 34.9 °C. Cerebral computed tomography scan performed on admission was normal. On arrival in PICU, cardiac ultrasound on ECMO showed left ventricular dysfunction with an estimated ejection fraction of 35 %, no left or right ventricular dilatation and without mitral valve regurgitation. Plasma hydroxychloroquine level taken a few hours after PICU admission (equal to ten hours post ingestion) was 0.06mcmol/L. Urinary toxic screening was positive for THC/cannabinoid, benzodiazepine and opioid (benzodiazepine and opioid being administered during initial medical care) and negative for acetaminophen, amphetamine/metamphetamine, barbiturate, cocaine, methadone, phencyclidine, tricyclic antidepressor.
ECMO blood flow was increased to 2L/min/m2 and adrenaline infusion weaned shortly after her admission, and heparin infusion started with ACT target range of 180–220. Targeted temperature management in the range of 34–35 °C was done for 48 h. ECMO course was uneventful. Cardiac ultrasound 46 hours post event showed recovery with systolic ejection fraction of 52 % and mild right diastolic dysfunction, allowing weaning from ECMO. A few hours later, a poor perfusion of the right lower limb was observed with a vascular doppler ultrasound revealing a significant reduction of arterial flow of the right common femoral artery. Immediate wound exploration revealed right common femoral artery stenosis without thrombosis at the site of the cannula insertion. Consequently, an arterial vascular surgical reconstruction was done with a venous patch. She developed a compartment syndrome of the right leg in the hours following, requiring fasciotomy.
When sedation was discontinued on day 4, the patient showed minimal interaction and no intentional movement. Striatal lesions were described on cerebral magnetic resonance imaging (MRI). Electroencephalogram (EEG) on day 6 revealed moderate reactive encephalopathy. On day 7, significant neurological improvement was observed, and the patient was extubated. Neurological exam revealed full consciousness, good spatiotemporal orientation, some memory deficit, and no focal neurologic deficit except hypoesthesia L5-S1 of the right foot and a right elevator muscle deficit secondary to right leg compartment syndrome. She was discharged from PICU at day 11 and then transferred from our tertiary center to her local rehabilitation hospital to continue intensive neuro-muscular physiotherapy.
Discussion
Chloroquine intoxication is a rare condition, associated with severe cardiotoxicity due to its quinidine-like properties. It is a strong membrane stabilizer acting like a class Ia antiarrhythmic agent (action on voltage-dependent sodium channel). Symptoms appear from two to three hours post ingestion and usually resolve within 24 hours, despite a long half-life (14 days). Cardiac toxicity is the result of the rapid rise in chloroquine plasma level during the first two hours, but it can extend to the first twenty-four hours. Cardiac toxicity includes negative inotropism, inhibition of spontaneous depolarization, slowing of atrioventricular conduction, increasing of the refractory period, prolongation of the QT segment and QRS interval, Torsades de pointes and multiple ventricular arrhythmias [2]. An ingestion of more than 20 mg/kg is considered a toxic dose with a lethal dose if it’s over 30 mg/kg. More than 4 grams of chloroquine ingested, chloroquine plasma levels > 25 mcmol/L and hypokalemia have been linked to poor prognosis [3], the severity of hypokalemia being related to the severity of the intoxication. Rebound hyperkalemia can be observed after aggressive correction so hypokalemia treatment should be cautious. Chloroquine also affects the respiratory, neurological (irritability, drowsiness, dystonia and seizures) and digestive systems and metabolic acidosis is common.
In our patient, severe intoxication had to be considered, with potentially more than 40 mg/kg of chloroquine ingested. The clinical presentation, similar to the above literature, confirmed the overdose: respiratory depression and neurological symptoms (drowsiness and dystonia), followed by pulseless ventricular tachycardia and cardiovascular collapse, profound metabolic acidosis and severe hypokalemia. Surprisingly, hydroxychloroquine plasma level was much lower (0,06mcmol/L) than the toxic levels found in literature (usually around 10–30 mcmol/L). We hypothesize that three reasons may explain this result. The first is possible adsorption of chloroquine by the ECMO system (tubing and oxygenator) or binding by of the intravenous lipid emulsion treatment. The second is the hemodilution by both the ECMO circuit and the fluid administration during CPR and on ECMO support. The last is the quality of the sample itself, possibly altered by dilution or sampling procedure error. In our patient, the chloroquine plasma level was not clinically relevant as management was driven by the patient’s condition.
Overdose cases remain rare, so there are no strong recommendations for management. However, from the existing literature, specific treatment combines assisted ventilation and administration of diazepam, adrenaline and intravenous lipid emulsion [2, 4]. Diazepam administration is controversial. It is part of the supportive treatment: used for sedation, in case of seizures and for its presumed antiarrhythmic properties [4]. However, there is no evidence that this treatment alone, as a potential antidote, significantly improves the outcome of moderately intoxicated patients. Adrenaline counteracts vasodilation and myocardial depression, playing a key role in resuscitation of the severely intoxicated patients [2]. Our patient received both treatments (adrenaline and diazepam), before stabilization on ECMO.
Intravenous lipid emulsion has been used in systemic anesthetic toxicity and in poisoning with other lipophilic drugs. As chloroquine is highly lipophilic, the early use of intravenous lipid emulsion in chloroquine intoxication could possibly reduce its plasma peak level of chloroquine and therefore reduce its toxicity. Our patient received a bolus followed by a continuous, but it was rapidly stopped when on ECMO support; indeed, ECMO is a relative contraindication due to a potential obstructive effect on oxygen filter, fat emulsion agglutination and increased blood clot formation in the circuit [5].
When given early enough after ingestion, implying the time of ingestion is known, activated charcoal could prevent absorption of any chloroquine remaining in the stomach. The use of intravenous bicarbonate is mentioned in case of widening of QRS complex. Hemodialysis and hemoperfusion on the other hand are not effective due to the high volume of distribution of chloroquine, therefore these modalities were not considered in our patient [6]
As chloroquine intoxication is a reversible phenomenon, mainly causing symptoms of direct cardiotoxicity, rapid efficient advanced cardiac life support (ACLS) is key to its management, including ECPR. ECMO is described as an option for selected poisoned patients, as it provides organ support during the acute phase of intoxication [7]. Available data show that the use of ECPR offers the possibility of survival with good neurologic recovery in adult out-of-hospital cardiac arrest (OHCA) of varying causes [8]. A shockable rhythm, female gender, short no flow time or witnessed cardiac arrest, short low-flow time and good quality CPR seem to play a positive role on outcome despite ongoing discrepancy about these prognostic factors in the literature [8]. Furthermore, the outcome of ECPR is improved when provided by experienced and trained centers. ECPR is currently provided on a case-by-case basis, where it can be quickly implemented and in patients for whom the etiology of the cardiac arrest is potentially reversible within a limited period of mechanical cardiorespiratory support [7]. For the pediatric population, ECPR use is described for in-hospital cardiac arrest (IHCA) [9–14] and mainly related to children with underlying cardiac disease or after cardiac surgery. Pediatric ECPR has a high mortality, with survival to decannulation and to hospital discharge of 58 % and 42 % respectively in the last ELSO registry report [15]. For OHCA and intoxication cases, data on children are, on the contrary, very sparse. Despite high mortality in pediatric ECPR, selected intoxication cases might, in our opinion, benefit from ECPR support because of their reversibility, as illustrated in this case.
In summary, outcome of our patient depended on the patient’s favorable prognostic factors, the quality of initial resuscitation and the experience of all the staff involved in the ECMO support. Despite morbidity linked to sustained muscle weakness of her right lower limb secondary to arterial ischemia, the overall neurological outcome was favorable, considering the severity of the insult and the prolonged resuscitation.
Conclusions
Chloroquine intoxication can be life threatening, with cardiac arrest, due to cardiotoxicity. Its management is mainly supportive as no antidote is available. This patient fulfilled criteria for optimal use of ECPR, despite the lack of strong evidence for this procedure in intoxication and OHCA in children. Patient criteria and specific protocols regarding use of ECPR are still under study, aiming to improve outcome after pediatric OHCA. Severe reversible intoxication could be one of its indications and should be considered on a case-by-case basis.
Abbreviations
Covid-19 Novel coronavirus disease
CPR Cardio-pulmonary resuscitation
ECPR Extracorporeal cardiopulmonary resuscitation
ECMO Extracorporeal membrane oxygenation
GCS Glasgow coma scale score
ECG Electrocardiogram
PICU Pediatric intensive care unit
MRI Magnetic resonance imaging
EEG Electroencephalogram
ACLS Advanced cardiac life support
OHCA Out-of-hospital cardiac arrest
IHCA In-hospital cardiac arrest
Acknowledgements
none.
Authors’ contributions
TF and DL conceptualized the case report, analyzed the data and wrote the manuscript. VA, JN, MHP, RP, SC read and completed with modifications the manuscript. The author(s) read and approved the final manuscript.
Funding
The authors did not receive any funds related to this case report.
Availability of data and materials
The data used for this case report are part of the personal clinical electronic file and are not pubicly available for confidentiality reasons, but anonymous data are available from the corresponding author on reasonable request.
Ethics approval and consent to participate
Not applicable.
Consent for publication
the authors declare that the patient and its legal representative have given consent for publication.
Competing interests
The authors declare that they have no competing interests.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. | CHLOROQUINE | DrugsGivenReaction | CC BY | 33722251 | 19,131,661 | 2021-03-15 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Cardiac arrest'. | Extracorporeal cardiopulmonary resuscitation for severe chloroquine intoxication in a child - a case report.
BACKGROUND
Chloroquine use has increased worldwide recently in the setting of experimental treatment for the novel coronavirus disease (Covid-19). Nevertheless, in case of chloroquine intoxication, it can be life threatening, with cardiac arrest, due to its cardiac toxicity.
METHODS
This case study reports on a 14-years-old girl who presented in cardiac arrest after an uncommon suicide attempt by ingesting 3 g of chloroquine. After 66 min of cardio-pulmonary resuscitation (CPR), extracorporeal cardiopulmonary resuscitation (ECPR) was initiated, allowing cardiac function to recover.
CONCLUSIONS
Chloroquine intoxication is a rare but serious condition due to its cardiac toxicity. Use of ECPR in this case of transient toxicity allowed a favorable evolution with little neurological impairment.
Background
Chloroquine is a worldwide recognized treatment for varied conditions such as malaria, rheumatoid arthritis and lupus erythematosus. Its use has recently been significantly expanded as an experimental treatment for COVID-19. Overdose however, can lead to life threatening complications [1], and with the lack of a specific antidote, treatment remains supportive, extracorporeal membrane oxygenation (ECMO) being the last resort. We report the case of a teenager who suffered cardiac arrest due to chloroquine, and who survived to ECMO decannulation. To our knowledge, this is the first published pediatric case of chloroquine intoxication requiring extracorporeal cardiopulmonary resuscitation (ECPR).
Case presentation
A 14-years-old girl, with a weight of 68 kg and a history of suicidal thoughts presented in cardiopulmonary arrest following a suicide attempt with ingestion of 3 gram of chloroquine without history of any other substance ingestion. Upon the arrival of the ambulance, she was unconscious with a Glasgow coma scale score (GCS) of 6 (1–1–4), bradypneic and with a trismus. The electrocardiogram (ECG) showed ST-segment changes, with pulseless ventricular tachycardia immediately following. Cardio-pulmonary resuscitation was initiated but despite chest compressions, defibrillation and adrenaline, there was no return to spontaneous circulation. Twelve minutes later, upon arrival at the hospital, she had a pulseless idioventricular rhythm and cardiopulmonary resuscitation was continued. The patient was intubated and external massage was taken over by a Lucas® chest compression system (Stryker Medical, Portage, MI49002 US).
Blood gas revealed severe hypokalemia (potassium 1.8 mmol/l) and a severe mixed acidosis (pH 6.97, pCO2 95 mmHg, glucose 14.5 mmol/l, lactate 7.4 mmol/l, base excess − 9.7 mmol/l, bicarbonate 12.7 mmol/l). Resuscitation was continued with three additional doses of adrenaline followed by continuous adrenaline infusion as well as correction of hypokalemia, administration of an amiodarone bolus, a bicarbonate bolus, fluid resuscitation and intravenous lipid emulsion infusion. Despite these measures, resuscitation was unsuccessful.
Peripheral right femoral veno-arterial extracorporeal membrane oxygenation (ECMO)ʼ was therefore initiated after 66 minutes of resuscitation to provide full cardiac support. During transfer to our pediatric intensive care unit (PICU), she was sedated and paralyzed, ventilated, hemodynamically stable on ECMO with a blood flow of 1.7 L/min/m2 and an adrenaline infusion of 0.1mcg/kg/min, a temperature of 34.9 °C. Cerebral computed tomography scan performed on admission was normal. On arrival in PICU, cardiac ultrasound on ECMO showed left ventricular dysfunction with an estimated ejection fraction of 35 %, no left or right ventricular dilatation and without mitral valve regurgitation. Plasma hydroxychloroquine level taken a few hours after PICU admission (equal to ten hours post ingestion) was 0.06mcmol/L. Urinary toxic screening was positive for THC/cannabinoid, benzodiazepine and opioid (benzodiazepine and opioid being administered during initial medical care) and negative for acetaminophen, amphetamine/metamphetamine, barbiturate, cocaine, methadone, phencyclidine, tricyclic antidepressor.
ECMO blood flow was increased to 2L/min/m2 and adrenaline infusion weaned shortly after her admission, and heparin infusion started with ACT target range of 180–220. Targeted temperature management in the range of 34–35 °C was done for 48 h. ECMO course was uneventful. Cardiac ultrasound 46 hours post event showed recovery with systolic ejection fraction of 52 % and mild right diastolic dysfunction, allowing weaning from ECMO. A few hours later, a poor perfusion of the right lower limb was observed with a vascular doppler ultrasound revealing a significant reduction of arterial flow of the right common femoral artery. Immediate wound exploration revealed right common femoral artery stenosis without thrombosis at the site of the cannula insertion. Consequently, an arterial vascular surgical reconstruction was done with a venous patch. She developed a compartment syndrome of the right leg in the hours following, requiring fasciotomy.
When sedation was discontinued on day 4, the patient showed minimal interaction and no intentional movement. Striatal lesions were described on cerebral magnetic resonance imaging (MRI). Electroencephalogram (EEG) on day 6 revealed moderate reactive encephalopathy. On day 7, significant neurological improvement was observed, and the patient was extubated. Neurological exam revealed full consciousness, good spatiotemporal orientation, some memory deficit, and no focal neurologic deficit except hypoesthesia L5-S1 of the right foot and a right elevator muscle deficit secondary to right leg compartment syndrome. She was discharged from PICU at day 11 and then transferred from our tertiary center to her local rehabilitation hospital to continue intensive neuro-muscular physiotherapy.
Discussion
Chloroquine intoxication is a rare condition, associated with severe cardiotoxicity due to its quinidine-like properties. It is a strong membrane stabilizer acting like a class Ia antiarrhythmic agent (action on voltage-dependent sodium channel). Symptoms appear from two to three hours post ingestion and usually resolve within 24 hours, despite a long half-life (14 days). Cardiac toxicity is the result of the rapid rise in chloroquine plasma level during the first two hours, but it can extend to the first twenty-four hours. Cardiac toxicity includes negative inotropism, inhibition of spontaneous depolarization, slowing of atrioventricular conduction, increasing of the refractory period, prolongation of the QT segment and QRS interval, Torsades de pointes and multiple ventricular arrhythmias [2]. An ingestion of more than 20 mg/kg is considered a toxic dose with a lethal dose if it’s over 30 mg/kg. More than 4 grams of chloroquine ingested, chloroquine plasma levels > 25 mcmol/L and hypokalemia have been linked to poor prognosis [3], the severity of hypokalemia being related to the severity of the intoxication. Rebound hyperkalemia can be observed after aggressive correction so hypokalemia treatment should be cautious. Chloroquine also affects the respiratory, neurological (irritability, drowsiness, dystonia and seizures) and digestive systems and metabolic acidosis is common.
In our patient, severe intoxication had to be considered, with potentially more than 40 mg/kg of chloroquine ingested. The clinical presentation, similar to the above literature, confirmed the overdose: respiratory depression and neurological symptoms (drowsiness and dystonia), followed by pulseless ventricular tachycardia and cardiovascular collapse, profound metabolic acidosis and severe hypokalemia. Surprisingly, hydroxychloroquine plasma level was much lower (0,06mcmol/L) than the toxic levels found in literature (usually around 10–30 mcmol/L). We hypothesize that three reasons may explain this result. The first is possible adsorption of chloroquine by the ECMO system (tubing and oxygenator) or binding by of the intravenous lipid emulsion treatment. The second is the hemodilution by both the ECMO circuit and the fluid administration during CPR and on ECMO support. The last is the quality of the sample itself, possibly altered by dilution or sampling procedure error. In our patient, the chloroquine plasma level was not clinically relevant as management was driven by the patient’s condition.
Overdose cases remain rare, so there are no strong recommendations for management. However, from the existing literature, specific treatment combines assisted ventilation and administration of diazepam, adrenaline and intravenous lipid emulsion [2, 4]. Diazepam administration is controversial. It is part of the supportive treatment: used for sedation, in case of seizures and for its presumed antiarrhythmic properties [4]. However, there is no evidence that this treatment alone, as a potential antidote, significantly improves the outcome of moderately intoxicated patients. Adrenaline counteracts vasodilation and myocardial depression, playing a key role in resuscitation of the severely intoxicated patients [2]. Our patient received both treatments (adrenaline and diazepam), before stabilization on ECMO.
Intravenous lipid emulsion has been used in systemic anesthetic toxicity and in poisoning with other lipophilic drugs. As chloroquine is highly lipophilic, the early use of intravenous lipid emulsion in chloroquine intoxication could possibly reduce its plasma peak level of chloroquine and therefore reduce its toxicity. Our patient received a bolus followed by a continuous, but it was rapidly stopped when on ECMO support; indeed, ECMO is a relative contraindication due to a potential obstructive effect on oxygen filter, fat emulsion agglutination and increased blood clot formation in the circuit [5].
When given early enough after ingestion, implying the time of ingestion is known, activated charcoal could prevent absorption of any chloroquine remaining in the stomach. The use of intravenous bicarbonate is mentioned in case of widening of QRS complex. Hemodialysis and hemoperfusion on the other hand are not effective due to the high volume of distribution of chloroquine, therefore these modalities were not considered in our patient [6]
As chloroquine intoxication is a reversible phenomenon, mainly causing symptoms of direct cardiotoxicity, rapid efficient advanced cardiac life support (ACLS) is key to its management, including ECPR. ECMO is described as an option for selected poisoned patients, as it provides organ support during the acute phase of intoxication [7]. Available data show that the use of ECPR offers the possibility of survival with good neurologic recovery in adult out-of-hospital cardiac arrest (OHCA) of varying causes [8]. A shockable rhythm, female gender, short no flow time or witnessed cardiac arrest, short low-flow time and good quality CPR seem to play a positive role on outcome despite ongoing discrepancy about these prognostic factors in the literature [8]. Furthermore, the outcome of ECPR is improved when provided by experienced and trained centers. ECPR is currently provided on a case-by-case basis, where it can be quickly implemented and in patients for whom the etiology of the cardiac arrest is potentially reversible within a limited period of mechanical cardiorespiratory support [7]. For the pediatric population, ECPR use is described for in-hospital cardiac arrest (IHCA) [9–14] and mainly related to children with underlying cardiac disease or after cardiac surgery. Pediatric ECPR has a high mortality, with survival to decannulation and to hospital discharge of 58 % and 42 % respectively in the last ELSO registry report [15]. For OHCA and intoxication cases, data on children are, on the contrary, very sparse. Despite high mortality in pediatric ECPR, selected intoxication cases might, in our opinion, benefit from ECPR support because of their reversibility, as illustrated in this case.
In summary, outcome of our patient depended on the patient’s favorable prognostic factors, the quality of initial resuscitation and the experience of all the staff involved in the ECMO support. Despite morbidity linked to sustained muscle weakness of her right lower limb secondary to arterial ischemia, the overall neurological outcome was favorable, considering the severity of the insult and the prolonged resuscitation.
Conclusions
Chloroquine intoxication can be life threatening, with cardiac arrest, due to cardiotoxicity. Its management is mainly supportive as no antidote is available. This patient fulfilled criteria for optimal use of ECPR, despite the lack of strong evidence for this procedure in intoxication and OHCA in children. Patient criteria and specific protocols regarding use of ECPR are still under study, aiming to improve outcome after pediatric OHCA. Severe reversible intoxication could be one of its indications and should be considered on a case-by-case basis.
Abbreviations
Covid-19 Novel coronavirus disease
CPR Cardio-pulmonary resuscitation
ECPR Extracorporeal cardiopulmonary resuscitation
ECMO Extracorporeal membrane oxygenation
GCS Glasgow coma scale score
ECG Electrocardiogram
PICU Pediatric intensive care unit
MRI Magnetic resonance imaging
EEG Electroencephalogram
ACLS Advanced cardiac life support
OHCA Out-of-hospital cardiac arrest
IHCA In-hospital cardiac arrest
Acknowledgements
none.
Authors’ contributions
TF and DL conceptualized the case report, analyzed the data and wrote the manuscript. VA, JN, MHP, RP, SC read and completed with modifications the manuscript. The author(s) read and approved the final manuscript.
Funding
The authors did not receive any funds related to this case report.
Availability of data and materials
The data used for this case report are part of the personal clinical electronic file and are not pubicly available for confidentiality reasons, but anonymous data are available from the corresponding author on reasonable request.
Ethics approval and consent to participate
Not applicable.
Consent for publication
the authors declare that the patient and its legal representative have given consent for publication.
Competing interests
The authors declare that they have no competing interests.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. | CHLOROQUINE | DrugsGivenReaction | CC BY | 33722251 | 19,131,661 | 2021-03-15 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Cardiotoxicity'. | Extracorporeal cardiopulmonary resuscitation for severe chloroquine intoxication in a child - a case report.
BACKGROUND
Chloroquine use has increased worldwide recently in the setting of experimental treatment for the novel coronavirus disease (Covid-19). Nevertheless, in case of chloroquine intoxication, it can be life threatening, with cardiac arrest, due to its cardiac toxicity.
METHODS
This case study reports on a 14-years-old girl who presented in cardiac arrest after an uncommon suicide attempt by ingesting 3 g of chloroquine. After 66 min of cardio-pulmonary resuscitation (CPR), extracorporeal cardiopulmonary resuscitation (ECPR) was initiated, allowing cardiac function to recover.
CONCLUSIONS
Chloroquine intoxication is a rare but serious condition due to its cardiac toxicity. Use of ECPR in this case of transient toxicity allowed a favorable evolution with little neurological impairment.
Background
Chloroquine is a worldwide recognized treatment for varied conditions such as malaria, rheumatoid arthritis and lupus erythematosus. Its use has recently been significantly expanded as an experimental treatment for COVID-19. Overdose however, can lead to life threatening complications [1], and with the lack of a specific antidote, treatment remains supportive, extracorporeal membrane oxygenation (ECMO) being the last resort. We report the case of a teenager who suffered cardiac arrest due to chloroquine, and who survived to ECMO decannulation. To our knowledge, this is the first published pediatric case of chloroquine intoxication requiring extracorporeal cardiopulmonary resuscitation (ECPR).
Case presentation
A 14-years-old girl, with a weight of 68 kg and a history of suicidal thoughts presented in cardiopulmonary arrest following a suicide attempt with ingestion of 3 gram of chloroquine without history of any other substance ingestion. Upon the arrival of the ambulance, she was unconscious with a Glasgow coma scale score (GCS) of 6 (1–1–4), bradypneic and with a trismus. The electrocardiogram (ECG) showed ST-segment changes, with pulseless ventricular tachycardia immediately following. Cardio-pulmonary resuscitation was initiated but despite chest compressions, defibrillation and adrenaline, there was no return to spontaneous circulation. Twelve minutes later, upon arrival at the hospital, she had a pulseless idioventricular rhythm and cardiopulmonary resuscitation was continued. The patient was intubated and external massage was taken over by a Lucas® chest compression system (Stryker Medical, Portage, MI49002 US).
Blood gas revealed severe hypokalemia (potassium 1.8 mmol/l) and a severe mixed acidosis (pH 6.97, pCO2 95 mmHg, glucose 14.5 mmol/l, lactate 7.4 mmol/l, base excess − 9.7 mmol/l, bicarbonate 12.7 mmol/l). Resuscitation was continued with three additional doses of adrenaline followed by continuous adrenaline infusion as well as correction of hypokalemia, administration of an amiodarone bolus, a bicarbonate bolus, fluid resuscitation and intravenous lipid emulsion infusion. Despite these measures, resuscitation was unsuccessful.
Peripheral right femoral veno-arterial extracorporeal membrane oxygenation (ECMO)ʼ was therefore initiated after 66 minutes of resuscitation to provide full cardiac support. During transfer to our pediatric intensive care unit (PICU), she was sedated and paralyzed, ventilated, hemodynamically stable on ECMO with a blood flow of 1.7 L/min/m2 and an adrenaline infusion of 0.1mcg/kg/min, a temperature of 34.9 °C. Cerebral computed tomography scan performed on admission was normal. On arrival in PICU, cardiac ultrasound on ECMO showed left ventricular dysfunction with an estimated ejection fraction of 35 %, no left or right ventricular dilatation and without mitral valve regurgitation. Plasma hydroxychloroquine level taken a few hours after PICU admission (equal to ten hours post ingestion) was 0.06mcmol/L. Urinary toxic screening was positive for THC/cannabinoid, benzodiazepine and opioid (benzodiazepine and opioid being administered during initial medical care) and negative for acetaminophen, amphetamine/metamphetamine, barbiturate, cocaine, methadone, phencyclidine, tricyclic antidepressor.
ECMO blood flow was increased to 2L/min/m2 and adrenaline infusion weaned shortly after her admission, and heparin infusion started with ACT target range of 180–220. Targeted temperature management in the range of 34–35 °C was done for 48 h. ECMO course was uneventful. Cardiac ultrasound 46 hours post event showed recovery with systolic ejection fraction of 52 % and mild right diastolic dysfunction, allowing weaning from ECMO. A few hours later, a poor perfusion of the right lower limb was observed with a vascular doppler ultrasound revealing a significant reduction of arterial flow of the right common femoral artery. Immediate wound exploration revealed right common femoral artery stenosis without thrombosis at the site of the cannula insertion. Consequently, an arterial vascular surgical reconstruction was done with a venous patch. She developed a compartment syndrome of the right leg in the hours following, requiring fasciotomy.
When sedation was discontinued on day 4, the patient showed minimal interaction and no intentional movement. Striatal lesions were described on cerebral magnetic resonance imaging (MRI). Electroencephalogram (EEG) on day 6 revealed moderate reactive encephalopathy. On day 7, significant neurological improvement was observed, and the patient was extubated. Neurological exam revealed full consciousness, good spatiotemporal orientation, some memory deficit, and no focal neurologic deficit except hypoesthesia L5-S1 of the right foot and a right elevator muscle deficit secondary to right leg compartment syndrome. She was discharged from PICU at day 11 and then transferred from our tertiary center to her local rehabilitation hospital to continue intensive neuro-muscular physiotherapy.
Discussion
Chloroquine intoxication is a rare condition, associated with severe cardiotoxicity due to its quinidine-like properties. It is a strong membrane stabilizer acting like a class Ia antiarrhythmic agent (action on voltage-dependent sodium channel). Symptoms appear from two to three hours post ingestion and usually resolve within 24 hours, despite a long half-life (14 days). Cardiac toxicity is the result of the rapid rise in chloroquine plasma level during the first two hours, but it can extend to the first twenty-four hours. Cardiac toxicity includes negative inotropism, inhibition of spontaneous depolarization, slowing of atrioventricular conduction, increasing of the refractory period, prolongation of the QT segment and QRS interval, Torsades de pointes and multiple ventricular arrhythmias [2]. An ingestion of more than 20 mg/kg is considered a toxic dose with a lethal dose if it’s over 30 mg/kg. More than 4 grams of chloroquine ingested, chloroquine plasma levels > 25 mcmol/L and hypokalemia have been linked to poor prognosis [3], the severity of hypokalemia being related to the severity of the intoxication. Rebound hyperkalemia can be observed after aggressive correction so hypokalemia treatment should be cautious. Chloroquine also affects the respiratory, neurological (irritability, drowsiness, dystonia and seizures) and digestive systems and metabolic acidosis is common.
In our patient, severe intoxication had to be considered, with potentially more than 40 mg/kg of chloroquine ingested. The clinical presentation, similar to the above literature, confirmed the overdose: respiratory depression and neurological symptoms (drowsiness and dystonia), followed by pulseless ventricular tachycardia and cardiovascular collapse, profound metabolic acidosis and severe hypokalemia. Surprisingly, hydroxychloroquine plasma level was much lower (0,06mcmol/L) than the toxic levels found in literature (usually around 10–30 mcmol/L). We hypothesize that three reasons may explain this result. The first is possible adsorption of chloroquine by the ECMO system (tubing and oxygenator) or binding by of the intravenous lipid emulsion treatment. The second is the hemodilution by both the ECMO circuit and the fluid administration during CPR and on ECMO support. The last is the quality of the sample itself, possibly altered by dilution or sampling procedure error. In our patient, the chloroquine plasma level was not clinically relevant as management was driven by the patient’s condition.
Overdose cases remain rare, so there are no strong recommendations for management. However, from the existing literature, specific treatment combines assisted ventilation and administration of diazepam, adrenaline and intravenous lipid emulsion [2, 4]. Diazepam administration is controversial. It is part of the supportive treatment: used for sedation, in case of seizures and for its presumed antiarrhythmic properties [4]. However, there is no evidence that this treatment alone, as a potential antidote, significantly improves the outcome of moderately intoxicated patients. Adrenaline counteracts vasodilation and myocardial depression, playing a key role in resuscitation of the severely intoxicated patients [2]. Our patient received both treatments (adrenaline and diazepam), before stabilization on ECMO.
Intravenous lipid emulsion has been used in systemic anesthetic toxicity and in poisoning with other lipophilic drugs. As chloroquine is highly lipophilic, the early use of intravenous lipid emulsion in chloroquine intoxication could possibly reduce its plasma peak level of chloroquine and therefore reduce its toxicity. Our patient received a bolus followed by a continuous, but it was rapidly stopped when on ECMO support; indeed, ECMO is a relative contraindication due to a potential obstructive effect on oxygen filter, fat emulsion agglutination and increased blood clot formation in the circuit [5].
When given early enough after ingestion, implying the time of ingestion is known, activated charcoal could prevent absorption of any chloroquine remaining in the stomach. The use of intravenous bicarbonate is mentioned in case of widening of QRS complex. Hemodialysis and hemoperfusion on the other hand are not effective due to the high volume of distribution of chloroquine, therefore these modalities were not considered in our patient [6]
As chloroquine intoxication is a reversible phenomenon, mainly causing symptoms of direct cardiotoxicity, rapid efficient advanced cardiac life support (ACLS) is key to its management, including ECPR. ECMO is described as an option for selected poisoned patients, as it provides organ support during the acute phase of intoxication [7]. Available data show that the use of ECPR offers the possibility of survival with good neurologic recovery in adult out-of-hospital cardiac arrest (OHCA) of varying causes [8]. A shockable rhythm, female gender, short no flow time or witnessed cardiac arrest, short low-flow time and good quality CPR seem to play a positive role on outcome despite ongoing discrepancy about these prognostic factors in the literature [8]. Furthermore, the outcome of ECPR is improved when provided by experienced and trained centers. ECPR is currently provided on a case-by-case basis, where it can be quickly implemented and in patients for whom the etiology of the cardiac arrest is potentially reversible within a limited period of mechanical cardiorespiratory support [7]. For the pediatric population, ECPR use is described for in-hospital cardiac arrest (IHCA) [9–14] and mainly related to children with underlying cardiac disease or after cardiac surgery. Pediatric ECPR has a high mortality, with survival to decannulation and to hospital discharge of 58 % and 42 % respectively in the last ELSO registry report [15]. For OHCA and intoxication cases, data on children are, on the contrary, very sparse. Despite high mortality in pediatric ECPR, selected intoxication cases might, in our opinion, benefit from ECPR support because of their reversibility, as illustrated in this case.
In summary, outcome of our patient depended on the patient’s favorable prognostic factors, the quality of initial resuscitation and the experience of all the staff involved in the ECMO support. Despite morbidity linked to sustained muscle weakness of her right lower limb secondary to arterial ischemia, the overall neurological outcome was favorable, considering the severity of the insult and the prolonged resuscitation.
Conclusions
Chloroquine intoxication can be life threatening, with cardiac arrest, due to cardiotoxicity. Its management is mainly supportive as no antidote is available. This patient fulfilled criteria for optimal use of ECPR, despite the lack of strong evidence for this procedure in intoxication and OHCA in children. Patient criteria and specific protocols regarding use of ECPR are still under study, aiming to improve outcome after pediatric OHCA. Severe reversible intoxication could be one of its indications and should be considered on a case-by-case basis.
Abbreviations
Covid-19 Novel coronavirus disease
CPR Cardio-pulmonary resuscitation
ECPR Extracorporeal cardiopulmonary resuscitation
ECMO Extracorporeal membrane oxygenation
GCS Glasgow coma scale score
ECG Electrocardiogram
PICU Pediatric intensive care unit
MRI Magnetic resonance imaging
EEG Electroencephalogram
ACLS Advanced cardiac life support
OHCA Out-of-hospital cardiac arrest
IHCA In-hospital cardiac arrest
Acknowledgements
none.
Authors’ contributions
TF and DL conceptualized the case report, analyzed the data and wrote the manuscript. VA, JN, MHP, RP, SC read and completed with modifications the manuscript. The author(s) read and approved the final manuscript.
Funding
The authors did not receive any funds related to this case report.
Availability of data and materials
The data used for this case report are part of the personal clinical electronic file and are not pubicly available for confidentiality reasons, but anonymous data are available from the corresponding author on reasonable request.
Ethics approval and consent to participate
Not applicable.
Consent for publication
the authors declare that the patient and its legal representative have given consent for publication.
Competing interests
The authors declare that they have no competing interests.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. | CHLOROQUINE | DrugsGivenReaction | CC BY | 33722251 | 19,131,661 | 2021-03-15 |
Give an alphabetized list of all active substances of drugs taken by the patient who experienced 'Circulatory collapse'. | Extracorporeal cardiopulmonary resuscitation for severe chloroquine intoxication in a child - a case report.
BACKGROUND
Chloroquine use has increased worldwide recently in the setting of experimental treatment for the novel coronavirus disease (Covid-19). Nevertheless, in case of chloroquine intoxication, it can be life threatening, with cardiac arrest, due to its cardiac toxicity.
METHODS
This case study reports on a 14-years-old girl who presented in cardiac arrest after an uncommon suicide attempt by ingesting 3 g of chloroquine. After 66 min of cardio-pulmonary resuscitation (CPR), extracorporeal cardiopulmonary resuscitation (ECPR) was initiated, allowing cardiac function to recover.
CONCLUSIONS
Chloroquine intoxication is a rare but serious condition due to its cardiac toxicity. Use of ECPR in this case of transient toxicity allowed a favorable evolution with little neurological impairment.
Background
Chloroquine is a worldwide recognized treatment for varied conditions such as malaria, rheumatoid arthritis and lupus erythematosus. Its use has recently been significantly expanded as an experimental treatment for COVID-19. Overdose however, can lead to life threatening complications [1], and with the lack of a specific antidote, treatment remains supportive, extracorporeal membrane oxygenation (ECMO) being the last resort. We report the case of a teenager who suffered cardiac arrest due to chloroquine, and who survived to ECMO decannulation. To our knowledge, this is the first published pediatric case of chloroquine intoxication requiring extracorporeal cardiopulmonary resuscitation (ECPR).
Case presentation
A 14-years-old girl, with a weight of 68 kg and a history of suicidal thoughts presented in cardiopulmonary arrest following a suicide attempt with ingestion of 3 gram of chloroquine without history of any other substance ingestion. Upon the arrival of the ambulance, she was unconscious with a Glasgow coma scale score (GCS) of 6 (1–1–4), bradypneic and with a trismus. The electrocardiogram (ECG) showed ST-segment changes, with pulseless ventricular tachycardia immediately following. Cardio-pulmonary resuscitation was initiated but despite chest compressions, defibrillation and adrenaline, there was no return to spontaneous circulation. Twelve minutes later, upon arrival at the hospital, she had a pulseless idioventricular rhythm and cardiopulmonary resuscitation was continued. The patient was intubated and external massage was taken over by a Lucas® chest compression system (Stryker Medical, Portage, MI49002 US).
Blood gas revealed severe hypokalemia (potassium 1.8 mmol/l) and a severe mixed acidosis (pH 6.97, pCO2 95 mmHg, glucose 14.5 mmol/l, lactate 7.4 mmol/l, base excess − 9.7 mmol/l, bicarbonate 12.7 mmol/l). Resuscitation was continued with three additional doses of adrenaline followed by continuous adrenaline infusion as well as correction of hypokalemia, administration of an amiodarone bolus, a bicarbonate bolus, fluid resuscitation and intravenous lipid emulsion infusion. Despite these measures, resuscitation was unsuccessful.
Peripheral right femoral veno-arterial extracorporeal membrane oxygenation (ECMO)ʼ was therefore initiated after 66 minutes of resuscitation to provide full cardiac support. During transfer to our pediatric intensive care unit (PICU), she was sedated and paralyzed, ventilated, hemodynamically stable on ECMO with a blood flow of 1.7 L/min/m2 and an adrenaline infusion of 0.1mcg/kg/min, a temperature of 34.9 °C. Cerebral computed tomography scan performed on admission was normal. On arrival in PICU, cardiac ultrasound on ECMO showed left ventricular dysfunction with an estimated ejection fraction of 35 %, no left or right ventricular dilatation and without mitral valve regurgitation. Plasma hydroxychloroquine level taken a few hours after PICU admission (equal to ten hours post ingestion) was 0.06mcmol/L. Urinary toxic screening was positive for THC/cannabinoid, benzodiazepine and opioid (benzodiazepine and opioid being administered during initial medical care) and negative for acetaminophen, amphetamine/metamphetamine, barbiturate, cocaine, methadone, phencyclidine, tricyclic antidepressor.
ECMO blood flow was increased to 2L/min/m2 and adrenaline infusion weaned shortly after her admission, and heparin infusion started with ACT target range of 180–220. Targeted temperature management in the range of 34–35 °C was done for 48 h. ECMO course was uneventful. Cardiac ultrasound 46 hours post event showed recovery with systolic ejection fraction of 52 % and mild right diastolic dysfunction, allowing weaning from ECMO. A few hours later, a poor perfusion of the right lower limb was observed with a vascular doppler ultrasound revealing a significant reduction of arterial flow of the right common femoral artery. Immediate wound exploration revealed right common femoral artery stenosis without thrombosis at the site of the cannula insertion. Consequently, an arterial vascular surgical reconstruction was done with a venous patch. She developed a compartment syndrome of the right leg in the hours following, requiring fasciotomy.
When sedation was discontinued on day 4, the patient showed minimal interaction and no intentional movement. Striatal lesions were described on cerebral magnetic resonance imaging (MRI). Electroencephalogram (EEG) on day 6 revealed moderate reactive encephalopathy. On day 7, significant neurological improvement was observed, and the patient was extubated. Neurological exam revealed full consciousness, good spatiotemporal orientation, some memory deficit, and no focal neurologic deficit except hypoesthesia L5-S1 of the right foot and a right elevator muscle deficit secondary to right leg compartment syndrome. She was discharged from PICU at day 11 and then transferred from our tertiary center to her local rehabilitation hospital to continue intensive neuro-muscular physiotherapy.
Discussion
Chloroquine intoxication is a rare condition, associated with severe cardiotoxicity due to its quinidine-like properties. It is a strong membrane stabilizer acting like a class Ia antiarrhythmic agent (action on voltage-dependent sodium channel). Symptoms appear from two to three hours post ingestion and usually resolve within 24 hours, despite a long half-life (14 days). Cardiac toxicity is the result of the rapid rise in chloroquine plasma level during the first two hours, but it can extend to the first twenty-four hours. Cardiac toxicity includes negative inotropism, inhibition of spontaneous depolarization, slowing of atrioventricular conduction, increasing of the refractory period, prolongation of the QT segment and QRS interval, Torsades de pointes and multiple ventricular arrhythmias [2]. An ingestion of more than 20 mg/kg is considered a toxic dose with a lethal dose if it’s over 30 mg/kg. More than 4 grams of chloroquine ingested, chloroquine plasma levels > 25 mcmol/L and hypokalemia have been linked to poor prognosis [3], the severity of hypokalemia being related to the severity of the intoxication. Rebound hyperkalemia can be observed after aggressive correction so hypokalemia treatment should be cautious. Chloroquine also affects the respiratory, neurological (irritability, drowsiness, dystonia and seizures) and digestive systems and metabolic acidosis is common.
In our patient, severe intoxication had to be considered, with potentially more than 40 mg/kg of chloroquine ingested. The clinical presentation, similar to the above literature, confirmed the overdose: respiratory depression and neurological symptoms (drowsiness and dystonia), followed by pulseless ventricular tachycardia and cardiovascular collapse, profound metabolic acidosis and severe hypokalemia. Surprisingly, hydroxychloroquine plasma level was much lower (0,06mcmol/L) than the toxic levels found in literature (usually around 10–30 mcmol/L). We hypothesize that three reasons may explain this result. The first is possible adsorption of chloroquine by the ECMO system (tubing and oxygenator) or binding by of the intravenous lipid emulsion treatment. The second is the hemodilution by both the ECMO circuit and the fluid administration during CPR and on ECMO support. The last is the quality of the sample itself, possibly altered by dilution or sampling procedure error. In our patient, the chloroquine plasma level was not clinically relevant as management was driven by the patient’s condition.
Overdose cases remain rare, so there are no strong recommendations for management. However, from the existing literature, specific treatment combines assisted ventilation and administration of diazepam, adrenaline and intravenous lipid emulsion [2, 4]. Diazepam administration is controversial. It is part of the supportive treatment: used for sedation, in case of seizures and for its presumed antiarrhythmic properties [4]. However, there is no evidence that this treatment alone, as a potential antidote, significantly improves the outcome of moderately intoxicated patients. Adrenaline counteracts vasodilation and myocardial depression, playing a key role in resuscitation of the severely intoxicated patients [2]. Our patient received both treatments (adrenaline and diazepam), before stabilization on ECMO.
Intravenous lipid emulsion has been used in systemic anesthetic toxicity and in poisoning with other lipophilic drugs. As chloroquine is highly lipophilic, the early use of intravenous lipid emulsion in chloroquine intoxication could possibly reduce its plasma peak level of chloroquine and therefore reduce its toxicity. Our patient received a bolus followed by a continuous, but it was rapidly stopped when on ECMO support; indeed, ECMO is a relative contraindication due to a potential obstructive effect on oxygen filter, fat emulsion agglutination and increased blood clot formation in the circuit [5].
When given early enough after ingestion, implying the time of ingestion is known, activated charcoal could prevent absorption of any chloroquine remaining in the stomach. The use of intravenous bicarbonate is mentioned in case of widening of QRS complex. Hemodialysis and hemoperfusion on the other hand are not effective due to the high volume of distribution of chloroquine, therefore these modalities were not considered in our patient [6]
As chloroquine intoxication is a reversible phenomenon, mainly causing symptoms of direct cardiotoxicity, rapid efficient advanced cardiac life support (ACLS) is key to its management, including ECPR. ECMO is described as an option for selected poisoned patients, as it provides organ support during the acute phase of intoxication [7]. Available data show that the use of ECPR offers the possibility of survival with good neurologic recovery in adult out-of-hospital cardiac arrest (OHCA) of varying causes [8]. A shockable rhythm, female gender, short no flow time or witnessed cardiac arrest, short low-flow time and good quality CPR seem to play a positive role on outcome despite ongoing discrepancy about these prognostic factors in the literature [8]. Furthermore, the outcome of ECPR is improved when provided by experienced and trained centers. ECPR is currently provided on a case-by-case basis, where it can be quickly implemented and in patients for whom the etiology of the cardiac arrest is potentially reversible within a limited period of mechanical cardiorespiratory support [7]. For the pediatric population, ECPR use is described for in-hospital cardiac arrest (IHCA) [9–14] and mainly related to children with underlying cardiac disease or after cardiac surgery. Pediatric ECPR has a high mortality, with survival to decannulation and to hospital discharge of 58 % and 42 % respectively in the last ELSO registry report [15]. For OHCA and intoxication cases, data on children are, on the contrary, very sparse. Despite high mortality in pediatric ECPR, selected intoxication cases might, in our opinion, benefit from ECPR support because of their reversibility, as illustrated in this case.
In summary, outcome of our patient depended on the patient’s favorable prognostic factors, the quality of initial resuscitation and the experience of all the staff involved in the ECMO support. Despite morbidity linked to sustained muscle weakness of her right lower limb secondary to arterial ischemia, the overall neurological outcome was favorable, considering the severity of the insult and the prolonged resuscitation.
Conclusions
Chloroquine intoxication can be life threatening, with cardiac arrest, due to cardiotoxicity. Its management is mainly supportive as no antidote is available. This patient fulfilled criteria for optimal use of ECPR, despite the lack of strong evidence for this procedure in intoxication and OHCA in children. Patient criteria and specific protocols regarding use of ECPR are still under study, aiming to improve outcome after pediatric OHCA. Severe reversible intoxication could be one of its indications and should be considered on a case-by-case basis.
Abbreviations
Covid-19 Novel coronavirus disease
CPR Cardio-pulmonary resuscitation
ECPR Extracorporeal cardiopulmonary resuscitation
ECMO Extracorporeal membrane oxygenation
GCS Glasgow coma scale score
ECG Electrocardiogram
PICU Pediatric intensive care unit
MRI Magnetic resonance imaging
EEG Electroencephalogram
ACLS Advanced cardiac life support
OHCA Out-of-hospital cardiac arrest
IHCA In-hospital cardiac arrest
Acknowledgements
none.
Authors’ contributions
TF and DL conceptualized the case report, analyzed the data and wrote the manuscript. VA, JN, MHP, RP, SC read and completed with modifications the manuscript. The author(s) read and approved the final manuscript.
Funding
The authors did not receive any funds related to this case report.
Availability of data and materials
The data used for this case report are part of the personal clinical electronic file and are not pubicly available for confidentiality reasons, but anonymous data are available from the corresponding author on reasonable request.
Ethics approval and consent to participate
Not applicable.
Consent for publication
the authors declare that the patient and its legal representative have given consent for publication.
Competing interests
The authors declare that they have no competing interests.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. | CHLOROQUINE | DrugsGivenReaction | CC BY | 33722251 | 19,131,661 | 2021-03-15 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.