Voriconazole exposure and risk of cutaneous squamous cell carcinoma among lung or hematopoietic cell transplant patients: A systematic review and meta-analysis Short title: Voriconazole and SCC risk

Background: Current evidence about the association between voriconazole and risk of cutaneous squamous cell carcinoma (SCC) remains inconsistent. Objective: To assess the association between voriconazole use and risk of SCC. Methods: We systematically searched PubMed and Embase and performed a random effects model meta-analysis to calculate the pooled relative risk (RR) with 95% confidence interval (CI). Results: Of the 8 studies involving 3,710 individuals with lung transplant (LT) or hematopoietic cell transplant (HCT) included in qualitative analysis, five studies were included in the meta-analysis. Use of voriconazole was significantly associated with increased risk of SCC (RR, 1.86; 95% CI, 1.36 – 2.55). The increased risk did not differ according to type of transplantation or adjustment for sun exposure. Longer duration of voriconazole was found to be positively associated with risk of SCC (RR, 1.72; 95% CI, 1.09 – 2.72). Voriconazole use was not associated with increased risk of basal cell carcinoma (RR, 0.84; 95% CI, 0.41 – 1.71). Limitations: There were some heterogeneities in retrospective observational studies. Conclusions: Our findings support an increased risk of SCC associated with voriconazole in individuals with LT or HCT. Routine dermatologic surveillance should be performed, especially among individuals at high risk of developing SCC.


INTRODUCTION
Non-melanoma skin cancer (NMSC) is the most common malignancy among individuals with solid organ transplantation 1 or hematopoietic cell transplantation (HCT). 2,3 The most common NMSC among this population is cutaneous squamous cell carcinoma (SCC), followed by basal cell carcinoma (BCC), 1 which together account for 95% of skin cancers in organ transplant recipients. 4 Individuals after solid organ transplantation had a higher risk for NMSC compared to the general population, 5,6 and this risk increased with time after transplantation. 7 Moreover, NMSC appears to be more aggressive among solid organ transplant recipients than in the general populations, which increases mortality among solid organ recipients. 8,9 Several risk factors, including history of frequent sun exposure, male sex, Fitzpatrick skin type I to III, older age at transplantation, immunotherapies, and underlying disease were found to be associated with increased risk of NMSC post-transplantation. [10][11][12][13][14] Since 2002, voriconazole has been used to manage or prevent fungal infections, which are important complications after LT or HCT and can result in significant morbidity and mortality. 15,16 However, voriconazole can cause significant toxicity and side effects, including hepatotoxicity, visual disturbances, and photosensitivity. 17 Recently, the increased risk of NMSC (primarily SCC) associated with voriconazole use attracted our attention. Voriconazole and its major hepatic metabolite, voriconazole N-oxide (VNO) may generate reactive oxygen species and induce DNA damage by sensitizing keratinocytes to ultraviolet (UV) A light. 18 However, current evidence regarding the association between use of voriconazole and risk of SCC among the patients with LT and HCT remains controversial. 7,[19][20][21][22][23][24][25][26] The conflicting results might be due to small sample size in individual studies, heterogeneity in populations, duration or dose of voriconazole, or the use of combination treatments. We, therefore, conducted this systematic review and meta-analysis of available observational studies to critically analyze and synthesize the evidence regarding the association between use of voriconazole and risk of SCC or BCC following LT and HCT.

MATERIALS AND METHODS
The study was performed in accordance with the Meta-analysis Of Observational Studies in Epidemiology (MOOSE) guidelines for reviews of observational studies. 27

Search strategy and study selection
PubMed and Embase were searched from inception to September 2017 to identify eligible observational studies (Supplemental Table 1). Additionally, we searched the reference lists of relevant review and included studies. Two reviewers (HT and WS) selected the studies that met the following criteria: 1) observational (both prospective and retrospective) studies; 2) evaluating the association between voriconazole and risk of SCC or BCC; and 3) reporting the outcome of SCC or BCC. We included the latest study only in case of multiple reports using the same database. Conference abstracts were excluded because they offered limited information on study quality, population, and outcomes.

Data extraction and quality assessment
We collected information on study design, data source, number of participants, age (years), selection criteria, exposure definition, adjusted covariates, and outcomes of interest. Estimates on risk of SCC or BCC were extracted if appropriate. The quality of the observational study was assessed using a 9-star scoring system as described by Newcastle-Ottawa quality-assessment scale (NOS), with total stars of 7 -9 and 5 -6 indicating high and moderate quality, respectively. 28 Two reviewers (HT and WS) independently extracted the data and assessed the quality of each study. We contacted the original author for more information if any missing information. Any disagreement was resolved by consensus or referral to a third reviewer (JH).

Statistical analysis
To account for heterogeneity between studies, a random-effects model was used to calculate the pooled relative risk (RR) and 95% confidence interval (CI) for the association between voriconazole exposure and risk of SCC or BCC. Statistical heterogeneity was quantified using the I² statistic (low heterogeneity: 25%, moderate heterogeneity: 50%, and high heterogeneity: 75%). 29 Subgroup analysis by type of transplantation or adjustment for sun/SUV exposure was performed to assess the consistency of the association between voriconazole and SCC risk. A sensitivity analysis was performed by removing one study at a time from the pooled analysis to evaluate its influence on the pooled estimate. The development of the evidence on the association between voriconazole and risk of SCC was tested by using a cumulative meta-analysis based on the date of publication. A visual inspection of the funnel plots and the Begg's and Egger's tests were applied to examine potential publication bias. All statistical analyses were performed with STATA (Version 14; Stata Corp., College Station, TX).

Study selection and study characteristics
Of 294 citations retrieved from electronic databases, eight observational studies involving 3,710 individuals met the eligibility criteria and were included in our systematic review (Fig 1). Two studies used the same database, 21,22 thus we included the latest study only. 22 Of the eight studies included, seven were retrospective cohort studies and one was a retrospective case-control study. The characteristics and main results of the included studies are presented in Table 1 and Supplemental Table 2, respectively.
Four studies were assessed as high quality, 20,22,24,25 and the remaining four studies were determined to be of moderate quality. 7, 19, 23, 26 (Supplemental Table 3) Six studies providing adequate data on the risk of SCC or BCC associated with voriconazole were included in the meta-analysis.
The significant association between voriconazole and increased risk of SCC remained robust in the sensitivity analysis when each study was removed from meta-analysis at a time (Supplemental Fig 3). Our cumulative meta-analysis ordered by publication year indicated that the association became significant since 2017 (Supplemental Fig 4). There was no evidence of substantial publication bias based on the Egger's test (P = 0.98), Begg's test (P = 0.81), or visual inspection of the funnel plot (Supplemental Fig 5).

Dose-and duration-response analyses
Six studies evaluated the relationship between duration of voriconazole therapy and risk of SCC. 7, 19-21, 25, 26 (Supplemental Table 2) Five studies found that voriconazole duration was significantly associated with the development of SCC 19-21, 25, 26 , while one study found no such association. 7 Meta-analysis of four studies found that longer duration of voriconazole was significantly associated with increased risk of SCC (RR, 1.72; 95% CI, 1.09 -2.72), 7, 19, 25, 26 while a non-significant positive association was observed in cumulative days of voriconazole use (RR, 1.74 per 180 days; 95% CI, 0.95 - Fig 6). In addition, two studies evaluating cumulative dose and risk of SCC reported a statistically significant dose-response relationship (Supplemental Table 2). 22,25

DISCUSSION
Our meta-analysis of observational studies found that voriconazole use was significantly associated with increased risk of SCC in both LT recipients and HCT recipients. Our sensitivity analysis omitting each study, one at a time, confirmed the robustness of our results. Cumulative meta-analysis indicated that the significant increase in the risk of SCC associated with voriconazole became robust beginning in 2017. Furthermore, longer duration or higher dose of voriconazole was associated with increased risk of SCC. However, there was no significant association between voriconazole exposure and risk of BCC.
Consistent with most previous studies, 19,20,[22][23][24][25][26] our results found that voriconazole use was significantly associated with increased risk of SCC. Although the potential carcinogenic mechanisms by which voriconazole cause SCC have not been fully elucidated, it has been hypothesized that either voriconazole or VNO may facilitate UV-induced DNA damage and inhibit DNA repair. 18,30 Furthermore, VNO may cause phototoxicity through non-radiation related mechanisms after exposure to UVB. [31][32][33] Since cytochrome P450 enzymes are expressed not only in the liver but also in human keratinocytes, 34 an accumulation of VNO in the skin may explain our findings that longer duration and higher dose of voriconazole were independent risks for SCC. 18 Sun exposure is more strongly related to the risk of SCC than BCC. 35 Photosensitizing medications (e.g., diuretics) were found to be more strongly associated with SCC than BCC. 36 Therefore, it was not surprising that our study found no association between voriconazole and BCC, though this might also be falsely negative due to the inclusion of only two studies. In addition, it should be noted that we observed a significant increase in the risk of SCC, regardless of whether a study adjusted for sun exposure. Thus, phototoxicity may not be the sole carcinogenic pathway involved. Some studies have found that voriconazole may promote tumor development by upregulating aryl hydrocarbon receptor-dependent COX pathway, 33 and induce SCC by regulating distinct cell cycle and terminal differentiation pathways in human keratinocytes. 32 Our study systematically searched all available cohort or case-control studies on voriconazole use and risk of SCC or BCC without any restriction (e.g., language). Moreover, we fully assessed the methodological quality of the included studies and provided separate outcomes for SCC and BCC. Finally, subgroup analysis, sensitivity analysis, cumulative meta-analysis, and dose-response analysis were performed to test the robustness of our findings.
Our meta-analysis had several limitations. First, the definition of voriconazole treatment varied considerably among studies, ranging from ever-exposure to at least 3 consecutive months of voriconazole therapy, which was undefined in many studies. Additionally, voriconazole was commonly used in combination with immunosuppressants in both LT recipients and HCT recipients.
Immunosuppressants, especially azathioprine, were considered a strong risk factor for SCC. 37,38 However, several studies did not provide details of treatment combinations so that we could not address this issue in our study.
One study adjusted for immunosuppression regimen, mean cyclosporine level, and mean tacrolimus level, found a significant increase in the risk of SCC. 25 One included study found no association between any particular immunosuppressive medication and risk of skin cancer. 7 Further studies are clearly warranted to explore potential interaction between voriconazole and immunosuppressive therapies on skin cancer among those patients with LT or HCT. Finally, some clinical factors, such as time since transplantation, 39