Voriconazole and squamous cell carcinoma after lung transplantation: A multicenter study

This study evaluated the independent contribution of voriconazole to the development of squamous cell carcinoma (SCC) in lung transplant recipients, by attempting to account for important confounding factors, particularly immunosuppression. This international, multicenter, retrospective, cohort study included adult patients who underwent lung transplantation during 2005‐2008. Cox regression analysis was used to assess the effects of voriconazole and other azoles, analyzed as time‐dependent variables, on the risk of developing biopsy‐confirmed SCC. Nine hundred lung transplant recipients were included. Median follow‐up time from transplantation to end of follow‐up was 3.51 years. In a Cox regression model, exposure to voriconazole alone (adjusted hazard ratio 2.39, 95% confidence interval 1.31‐4.37) and exposure to voriconazole and other azole(s) (adjusted hazard ratio 3.45, 95% confidence interval 1.07‐11.06) were associated with SCC compared with those unexposed after controlling for important confounders including immunosuppressants. Exposure to voriconazole was associated with increased risk of SCC of the skin in lung transplant recipients. Residual confounding could not be ruled out because of the use of proxy variables to control for some confounders. Benefits of voriconazole use when prescribed to lung transplant recipients should be carefully weighed versus the potential risk of SCC. EU PAS registration number: EUPAS5269.


| INTRODUCTION
Skin cancer is the most common malignancy reported in recipients of solid organ transplants, with squamous cell carcinoma (SCC) of the skin being most frequently diagnosed. [1][2][3] The incidence of SCC in recipients of solid organ transplants is >65 times that of the general population but varies by organ transplant. 1,4 Risk factors for the development of SCC include prolonged exposure to sunlight, long duration of immunosuppressive therapy, infection with human papillomavirus, lower CD4 cell counts, and certain host factors, such as male sex, older age, white race, and Fitzpatrick skin types I to III. [4][5][6] The rates of all skin cancers among adult lung transplant (LT) recipients surviving 1, 5, and 10 years after transplantation have been reported to be 1.3%, 11.9%, and 20.8%, respectively. 7 LT recipients may be particularly vulnerable for developing SCC as a result of concurrent intensity of immunosuppression. 3 Fungal infection is also an important complication for LT recipients, with 15% to 35% of patients being diagnosed with fungal infections, such as invasive aspergillosis (IA). 8 The treatment of IA in LT recipients generally involves minimizing immunosuppression, followed by early initiation of a suitable antifungal agent until resolution of the pulmonary lesions. 9 Voriconazole (VFEND; Roerig, Pfizer, New York, NY) was approved in 2002 for the treatment of IA and other invasive fungal infections.
Voriconazole was also recently approved for prophylaxis of invasive fungal infections in high-risk recipients of hematopoietic stem cell transplantations in the European Union and other countries. In addition to its approved indications, many transplant programs have implemented universal antifungal prophylaxis using voriconazole to prevent IA in recipients of LTs or heart-lung transplants. 10,11 Data from a worldwide survey showed that voriconazole is the most widely prescribed antifungal agent for prophylaxis in LT recipients. 12,13 Single cases [14][15][16] and small case series 17,18 of SCC in voriconazoletreated immunocompromised patients have been reported, including in recipients of solid organ transplants and those with hematologic malignancies or human immunodeficiency virus infection. The risk of SCC or nonmelanoma skin cancer with voriconazole exposure has been investigated in single-center, retrospective, observational casecontrol or cohort studies, primarily among recipients of LTs [19][20][21][22][23][24] but also in recipients of hematopoietic stem cell transplants. 25 Although most studies reported that exposure to voriconazole increased the risk of skin cancer, findings were not generally consistent, perhaps due to the differences in the exposure/endpoint assessments and/or analytical methods used. Also, some of the studies did not control for the presence of comorbidities or underlying conditions, sun exposure, or the use of potentially phototoxic concomitant medications, nor did they adequately account for immunosuppressives or antifungal agents, other than voriconazole, received by transplant recipients.
Given the limitations of the currently available data, the objective of this study was to assess the independent contribution of voriconazole to the development of SCC in LT or heart-lung transplant recipients, by accounting for important confounding factors, particularly comorbidities and the use of concomitant immunosuppressants and other azoles.

| Study design and patients
This multicenter, retrospective cohort study used patient-level data collected from 14 LT centers across 9 countries: United States, Canada, France, Germany, Italy, the Netherlands, Spain, Switzerland, and Australia. Consecutive patients aged ≥18 years who underwent single LT, double LT, or heart-lung transplantation at the participating study centers between January 1, 2005, and December 31, 2008, were eligible for inclusion ( Figure 1). Patients with simultaneous or sequential abdominal organ transplant and those with a previous history of biopsy-confirmed SCC were excluded. The study protocol was approved by the University Health Network Research Ethics Board (REB No. 10-0622-AE) as the coordinating site and by institutional review boards and/or independent ethics committees at each site.

| Exposure
Cumulative voriconazole exposure of ≥30 days, not necessarily consecutive, was considered clinically meaningful for the risk of SCC; the same criterion was applied to other azoles. Sensitivity analysis using a cumulative exposure of 1 day was also conducted. LT  1. Unexposed to any azole or exposed to some azole for <30 days (referred to as "unexposed") 2. Exposed to voriconazole for ≥30 days but not to any other azole for ≥30 days (referred to as "exposed to voriconazole alone") 3. Exposed to other azole(s) for ≥30 days but not to voriconazole for ≥30 days (referred to as "exposed to other azoles alone") 4. Exposed to voriconazole for ≥30 days as well as exposed to some other azole for ≥30 days [referred to as "exposed to voriconazole and other azole(s)"].
Finally, to assess the possibility that the risk of SCC may be dose dependent, we modelled exposure to voriconazole based on the mean cumulative daily dose, measured as "defined daily doses." 26

| Covariates
Comprehensive data were collected on demographic and clinical characteristics including potential confounders and effect modifiers for SCC. In addition to the use of immunosuppressive agents post-LT and potentially phototoxic agents, data on whole blood concentrations of calcineurin inhibitors and number of episodes of elevated calcineurin inhibitor levels were collected (Appendix S2 presents all covariates and their transformation).

| Outcome assessment
The primary outcome was the first occurrence of biopsy-confirmed SCC during the follow-up period from the index date of LT. Recipients of LTs were followed from the index date to whichever of the following occurred first: SCC, death, last patient visit (based on documentation in the medical records), or December 31, 2012.

| Statistical analysis
Assuming an incidence of SCC in LT recipients unexposed to voriconazole of 5% and a voriconazole exposed:unexposed ratio of 1:2, at least 157 LT recipients in the voriconazole-exposed cohort and 314 LT F I G U R E 1 Eligibility in this study that evaluated the association between voriconazole exposure and squamous cell carcinoma in patients receiving a lung or heart-lung transplant and categorization of treatment exposure  Most LT recipients received a concomitant immunosuppressive regimen that included tacrolimus (70.7%) or a mycophenolate derivative (77.3%), and almost all (98.7%) had been exposed to corticosteroids. Forty-five percent (401 of 900) of LT recipients were exposed to voriconazole for ≥30 days, and approximately one-third were exposed to an azole other than voriconazole, including itraconazole (21.4%), fluconazole (5.8%), and posaconazole (4.6%). Table 1  LT recipients overall, all of whom were exposed to voriconazole.

| Voriconazole exposure and other risk factors for SCC
At the univariate level (   melanoma. e Absolute neutrophil counts <500 cells/mm 3 . f Elevated CNI levels were defined as cyclosporine trough >350 μg/L or tacrolimus trough >20 μg/L. g Patients receiving at least 1 dose were classified as being exposed.
T A B L E 1 (Continued) compared with those unexposed, after controlling for important confounding variables. The study findings corroborate previous studies examining the association between voriconazole exposure and SCC or nonmelanoma skin cancer and suggest a dose-response relationship (ie, increasing dose or duration of voriconazole increases the risk of SCC). This risk increased to 3.5-fold when LT recipients were exposed to voriconazole for a cumulative period of >180 days compared with those unexposed and to 2.7-fold for every 400-mg increase in the mean daily dose. LT recipients receiving prophylaxis had a longer duration of exposure compared with those receiving treatment only and were at an increased hazard for SCC.
The impact of voriconazole exposure on the risk of SCC in LT recipients has been suggested in several epidemiologic studies 19-24 ; however, this is the first multicenter study, and largest study overall, that has attempted to control for the potential confounding effect of immunosuppression and exposure to sunlight. Incidence of SCC varies across studies depending on the study methodology used, including follow-up time, but also likely as a result of varying patient demographic and clinical characteristics, including the use of different immunosuppression protocols during distinct transplantation eras. A retrospective cohort study by Singer and coworkers evaluated 327 LT recipients during 20 years and reported a crude incidence of 16.5% in patients exposed to voriconazole compared with 11.8% among those unexposed, with an overall median time from LT to SCC diagnosis of 3.6 years. 22 This study also suggested that any exposure to voriconazole was associated with a 2.6-fold increase in the risk of cutaneous SCC adjusted for age, sex, and race (HR 2.62, 95% CI 1.21-5.65), consistent with our finding of increased risk for SCC 22 ; however, importantly, this did not account for the type or the duration or intensity of immunosuppression.
A single-center retrospective cohort study of 455 LT recipients showed that voriconazole exposure was associated with risk of SCC, adjusted for age, sex, and race (adjusted HR 1.73, 95% CI 1.04-2.88). 20 In contrast to our study, they were unable to account for voriconazole exposure during hospitalization, nor did they control for the type and intensity of immunosuppression, all of which may have biased their estimate. 20 Although exposure to voriconazole was associated with an increased risk of SCC, our study also suggests a potential dose-response relationship. Evidence from a retrospective cohort study suggests that longer duration of voriconazole therapy, but not cumulative dose, is an independent risk factor for SCC (odds ratio 1.8, 95% CI 1.3-2.6). 19 Rather than cumulative dose, the present study modeled the mean cumulative dose as the summary exposure measure in an effort to better specify the intensity of voriconazole therapy. 27 Two other studies provide opposing evidence, with both reporting that neither increasing voriconazole duration nor increasing mean cumulative dose was associated with increased risk of skin cancer. 21,27 Although both included measures of immunosuppression use, one of the studies failed to account for duration of use 27 and the other used a large prescription claims database, which did not contain data on voriconazole use during hospitalization, to inform both voriconazole exposure and the outcome of nonmelanoma skin cancer. 21 The final model in the present study not only accounted for both inpatient/outpatient exposure to voriconazole and individual immunosuppressive agents, including dose and duration, but also controlled for other patient-specific indicators of immunosuppression, including episodes of rejection and neutropenia and calcineurin inhibitor whole blood concentrations, which tend to demonstrate a high degree of both interpatient and intrapatient variability that may not correlate with dose administered. 28 Although the means by which voriconazole may lead to the development of SCC has not been fully elucidated, it has been proposed that voriconazole itself, or one of its metabolites, may cause ultravioletinduced DNA damage or disrupt mechanisms used to repair damaged DNA. 20 This process may initially present clinically as photosensitivity or phototoxicity, followed by the development of a cutaneous malignancy in some patients, perhaps accelerated by intense immunosuppression or other immune impairment. 29,30 Thus far, the association T A B L E 2 Incidence rate (per 1000 person-years) of squamous cell carcinoma by 4 treatment exposure categories: overall and by time since lung transplantation