Increased Risk of Cutaneous Squamous Cell Carcinoma After Vismodegib Therapy for Basal Cell Carcinoma

Importance  Smoothened inhibitors (SIs) are a new type of targeted therapy for advanced basal cell carcinoma (BCC), and their long-term effects, such as increased risk of subsequent malignancy, are still being explored.

Objective  To evaluate the risk of developing a non-BCC malignancy after SI exposure in patients with BCC.

Design, Setting, and Participants  A case-control study at Stanford Medical Center, an academic hospital. Participants were higher-risk patients with BCC diagnosed from January 1, 1998, to December 31, 2014. The dates of the analysis were January 1 to November 1, 2015.

Exposures  The exposed participants (cases) comprised patients who had confirmed prior vismodegib treatment, and the nonexposed participants (controls) comprised patients who had never received any SI. Because vismodegib was the first approved SI, only patients exposed to this SI were included.

Main Outcomes and Measures  Hazard ratio for non-BCC malignancies after vismodegib exposure, adjusting for covariates.

Results  The study cohort comprised 180 participants. Their mean (SD) age at BCC diagnosis was 56 (16) years, and 68.9% (n = 124) were male. Fifty-five cases were compared with 125 controls, accounting for age, sex, prior radiation therapy or cisplatin treatment, Charlson Comorbidity Index, clinical follow-up time, immunosuppression, and basal cell nevus syndrome status. Patients exposed to vismodegib had a hazard ratio of 6.37 (95% CI, 3.39-11.96; P < .001), indicating increased risk of developing a non-BCC malignancy. Most non-BCC malignancies were cutaneous squamous cell carcinomas, with a hazard ratio of 8.12 (95% CI, 3.89-16.97; P < .001), accounting for age and basal cell nevus syndrome status. There was no significant increase in other cancers.

Conclusions and Relevance  Increased risk for cutaneous squamous cell carcinomas after vismodegib therapy highlights the importance of continued skin surveillance after initiation of this therapy.

Hedgehog pathway inhibitors are a growing class of drugs to treat advanced basal cell carcinomas (BCCs), with 2 drugs in this class (both smoothened inhibitors [SIs]) approved by the US Food and Drug Administration.^1-4 However, long-term adverse effects of this drug class, including the potential for second malignancy, are not well known.^5-8

Case reports in the literature have suggested that cutaneous squamous cell carcinoma (CSCC) may develop after initiation of the SI vismodegib, the first drug of its class to be commercially available.^9-13 However, systematic assessment of this risk has not been reported, to our knowledge. Targeted inhibition by SIs has the potential to select for tumor cells that can circumvent hedgehog pathway dependency, and studies^14,15 show that these mechanisms include the RAS/MAPK pathway. Given the association between the RAS pathway and multiple human cancers,¹⁶ inhibiting the hedgehog signaling pathway to treat BCCs could potentially lead to the development of other cancer types.¹⁴

To explore this possibility, we assessed the development of non-BCC cancers in a case-control study comprising groups of individuals exposed and not exposed to vismodegib, the first SI to be approved by the US Food and Drug Administration for treatment of advanced BCCs.

This study was approved by the Stanford University Human Subjects Panel. Using the Stanford Cancer Research Database, which included records from January 1, 1998, to December 31, 2014, we identified cases of higher-risk BCCs at Stanford Medical Center using a keyword search of the electronic medical records. The keywords used include the following: metastatic basosquamous cell carcinoma; metatypical basal cell carcinoma; metastatic BCC; metastatic basal cell carcinoma; infiltrative basal cell carcinoma; invasive basal cell carcinoma, nodular infiltrating type; recurrent basal cell carcinoma; basal cell carcinoma, metatypical type; metatypical and desmoplastic basal cell carcinoma; metastatic carcinoma consistent with keratotic basal cell carcinoma; keratotic basal cell carcinoma; metastatic keratotic basal cell carcinoma; metastatic basaloid carcinoma; basal cell carcinoma infiltrating; malignant neoplasm, compatible with basal cell carcinoma; locally advanced basal cell carcinoma; locally advanced basal cell skin cancer; metastatic basal cell skin cancer; stage 2 basal cell carcinoma; stage 3 basal cell carcinoma; and stage 4 basal cell carcinoma. After this initial screening with keywords, a detailed manual medical record review was performed to confirm biopsy-proven BCC. The dates of the analysis were January 1 to November 1, 2015.

Cases were defined as vismodegib-exposed patients, and controls were defined as unexposed patients. To qualify for vismodegib exposure, documented use for at least 7 days was required. Non-BCC second malignancy was defined as the new diagnosis of a cancer documented at least 2 weeks after the index BCC diagnosis (based on clinical reports of keratoacanthomas appearing within weeks of drug initiation).¹⁷ This definition allowed for exclusion of concurrent non-BCC malignancies present at the time of BCC diagnosis that had not yet been treated. Non-BCC second malignancies were defined as associated with vismodegib therapy if they were diagnosed at least 2 weeks after the first exposure to vismodegib (based on clinical reports of keratoacanthomas appearing within weeks of drug initiation¹⁷) but excluded concurrent non-BCC malignancies present, but not yet treated, before vismodegib initiation. Latency period, defined as the number of years from BCC diagnosis to the first non-BCC malignancy (as shown in Figure 1), was recorded for the cases and controls. In scenarios in which patients had multiple BCCs, the last BCC documented in the medical record was chosen as the index BCC. Total clinical follow-up duration, including those patients who did not develop any non-BCC malignancy after BCC diagnosis, was recorded to assess for detection bias.

Second CSCC subtype (in situ, keratoacanthoma, adenoid, or unspecified due to lack of subtype reporting in the pathology report) was assessed by manual medical record review. The CSCC disease status was classified according to the TNM staging system developed in 2010 by the American Joint Committee on Cancer.¹⁸

The cases and controls were compared with respect to demographics and clinical characteristics using t test or Wilcoxon rank sum test for continuous variables and χ² test or Fisher exact test for categorical variables. The Charlson Comorbidity Index for each participant was calculated based on medical record review.¹⁹ Other relevant clinical characteristics included in the analysis were prior exposure to known carcinogenic therapies (eg, radiation therapy²⁰ and cisplatin-based chemotherapy), history of immunosuppression (human immunodeficiency virus status or transplantation history), known genetic predisposition to cancers (eg, basal cell nevus syndrome [BCNS]), and years of clinical follow-up (starting from the date of BCC diagnosis to the date of the last clinic visit or death), as documented in the electronic medical record.

A Cox proportional hazards regression model was applied to estimate the hazard ratio (HR) of developing a subsequent malignancy after BCC diagnosis, controlling for covariates. Vismodegib exposure and radiation therapy were treated as time-dependent variables. Incidence rates and 95% exact CIs for both groups were also calculated. All statistical analyses were conducted using a software package (SAS, version 9.4; SAS Institute Inc).

Of the 180 patients with BCC identified, 55 were exposed to vismodegib (cases), and 125 were not exposed (controls). The overall rate of subsequent malignancy in the 180 study patients was 36.7% (66 of 180). The rate of subsequent malignancy was 29.1% (16 of 55) in the cases and 40.0% (50 of 125) in the controls (Table 1).

The cases and controls were not significantly different in demographic and clinical characteristics, including sex, race, and prior exposures or conditions that might increase cancer risk, such as radiation therapy, chemotherapy, or immunosuppression (Table 1). The mean age and BCNS status of the cases and controls were significantly different. The mean age of cases was younger than that of controls at the time of BCC diagnosis (52 vs 59 years, P < .001) (Table 1). There were more cases than controls with BCNS diagnosis (23.6% [13 of 55] vs 0.8% [1 of 125], P < .001). While patients with BCNS were diagnosed as having BCC at an earlier age than nonsyndromic patients, this age difference was still significant after BCNS was controlled for (Table 2). Because age is also significantly associated with general cancer risk,^21,22 this covariate was adjusted for in subsequent analysis.

Cox proportional hazards regression analysis that adjusted for age and BCNS status (Table 2) showed that the cases were at significantly increased risk of developing a non-BCC malignancy compared with the controls (HR, 6.37; 95% CI, 3.39-11.96; P < .001). To examine if the increased risk could be owing to BCNS status, the data were analyzed without patients with BCNS, and the increased risk of subsequent malignancy was still significant (HR, 5.02; 95% CI, 2.43-10.36; P < .001).

Clinical follow-up durations were not significantly different between the cases and controls. These values were a median of 8.5 years (range, 0.3-44.4 years) for cases and a median of 5.5 years (range, 0.1-44.5 years) for controls (P = .33) (Table 1).

Latency period, defined as the number of years from BCC diagnosis to the first non-BCC malignancy, was not significantly different between the cases (median, 6.6 years; range, 0.3-21.3 years) and the controls (median, 2.3 years; range, 0.1-29.0 years) (P = .12). While the ranges for latency period are large, 50.0% (8 of 16) of non-BCC malignancies in the cases were diagnosed within 1 year after vismodegib initiation (the median number of years between vismodegib initiation and the first non-BCC malignancy was 0.9 years [range, 0.1-2.7 years]). The types of non-BCC cancers reported are listed in Table 3.

The most common type of non-BCC second malignancy in both the cases and controls was CSCC (Table 3). The subtypes of second CSCC (in situ, keratoacanthoma, adenoid, and not specified if not reported in the pathology report) were significantly different between the cases and controls who developed second CSCC (P = .04). The CSCC subtype was not specified in most of the cases (58.3% [7 of 12]), while in situ CSCC was most predominant in the controls (54.5% [18 of 33]) (eTable 1 in the Supplement). The TNM stage of second CSCC was not significantly different between the cases and controls (P = .12).

The median number of years between vismodegib initiation and the development of CSCC was 0.6 years (range, 0.1-2.0 years). Figure 2 shows Kaplan-Meier curves demonstrating CSCC-free survival in the cases and controls. After accounting for age and BCNS status, vismodegib exposure increased the risk of developing a CSCC (HR, 8.12; 95% CI, 3.89-16.97; P < .001).

We assessed for the possibility of sampling bias in the index BCCs with the potential to miss metatypical components that are at the base of the lesions by examining the depth of the biopsies and the type of biopsy (shave vs punch technique). The mean depth of the index BCC biopsy was not significantly different between the cases and controls (P = .14) (eTable 1 in the Supplement). Shave biopsy was the most commonly reported biopsy type in both the cases (32.7% [18 of 55]) and controls (49.6% [62 of 125]). However, annotation of shave or punch biopsy was not present in the procedure note or pathology report in 52.7% (29 of 55) of cases and in 30.4% (38 of 125) of controls, and these missing data led to our inability to stratify analysis by shave or punch technique. However, actual depths of the biopsies could be ascertained by pathology report and were not significantly different between the cases and controls.

We also examined whether patients with a metatypical or basosquamous index BCC were more likely to develop second CSCCs. Cox proportional hazards regression analysis showed that the HR of developing second CSCC in patients with metatypical or basosqumaous index BCCs after vismodegib exposure was 2.19 (95% CI, 0.52-9.24). Patients with metatypical or basosquamous index BCC represented 20.0% (11 of 55) of cases and 8.0% (10 of 125) of controls (P = .02). Because of this significant difference, a sensitivity analysis removing patients with an index BCC of metatypical or basosquamous subtype was performed. eTable 2A in the Supplement summarizes the results of this sensitivity analysis (n = 159), demonstrating that the risk of second cancer was still elevated (HR, 7.80; 95% CI, 3.84-15.83; P < .001) with exclusion of patients with metatypical or basosquamous index BCCs. This finding is consistent with the analysis of all patients in the study (N = 180), in whom the effect of metatypical or basosquamous index BCC did not significantly influence the risk of secondary cancer (P = .83) (eTable 2B in the Supplement).

The epidermis is the first level of defense against external mutagenic insults, such as UV radiation, and CSCCs are among the most highly mutated of human cancers.²³ According to the classic model of carcinogenesis, the progression from actinic keratosis to CSCC involves the progressive accumulation of mutations, including those that promote genetic instability (eg, TP53 [OMIM 191170]).²⁴ Additional oncogenic mutations, such as activating mutations of the Ras oncogene,²⁵ found in up to 21% of CSCCs,²⁶ can drive the progression of CSCC from actinic keratosis or de novo. Our study demonstrates a significant risk of second malignancy, specifically CSCC, with vismodegib therapy. We speculate that the high mutational load of the epidermis (compared with other organs) from UV radiation and other insults may lead to its being “spring-loaded” to produce CSCC once the hedgehog pathway is inhibited by vismodegib.

The exact mechanism by which hedgehog inhibition may increase the risk of CSCC is unknown, but recent data offer a glimpse into one possibility. In medulloblastoma models, hedgehog pathway inhibition has been shown to activate the RAS/MAPK pathway, thereby circumventing hedgehog pathway dependency for tumor growth.^14,15 Because of the association between RAS pathway dysregulation and multiple human cancers,¹⁶ hedgehog pathway inhibition, such as through vismodegib therapy to treat BCCs, could promote the development of CSCC and other second cancers.¹⁴

Confirmation of the findings in this case-control study awaits prospective studies across multiple sites. Because the most commonly observed malignancy after exposure to vismodegib was CSCC, these study data suggest that careful surveillance through regular total-body skin checks by a dermatologist may be prudent, particularly in older adults. Furthermore, our data show that patients receiving vismodegib therapy were less likely to have in situ CSCC than the controls, and investigation into whether vismodegib-associated CSCCs are more aggressive merits future study. Finally, because most CSCCs developed within 1 year of vismodegib initiation, surveillance over the course of at least 1 year may be needed.

Our study has some limitations. Because BCC is not routinely staged like other cancers, one limitation of this study was the inability to assess the effect of BCC stage on subsequent cancer development. Another limitation is that, while we recorded the development of a single non-BCC malignancy most proximate in time to the BCC diagnosis, we did not record whether a patient had additional subsequent non-BCC malignancies.

Another potential limitation pertains to the difficulty in quantifying field cancerization of the skin, and differences in lifetime radiation exposure between the cases and controls could have led to differential rates of CSCC. For instance, lifetime UV radiation exposure is difficult to accurately ascertain. However, we controlled for exposure to radiation therapy in our study, and this characteristic was not significantly different in the cases vs controls. An additional type of field cancerization, namely, genetic predisposition to cancer, was examined by comparing percentages of patients with BCNS among the cases vs controls, and this characteristic was found to be significantly different. However, this variable was adjusted for in the Cox proportional hazards regression model.

The potential conversion of BSCs to CSCCs after vismodegib exposure has been a concern for advanced lesions,⁹ and larger studies are needed to assess whether this phenomenon is occurring and leading to nonadvanced CSCCs. Larger multicenter studies are also required to evaluate whether longer duration of vismodegib exposure increases the risk of non-BCC malignancies.

Accepted for Publication: November 18, 2015.

Corresponding Author Anne Lynn S. Chang, MD, Department of Dermatology, Stanford University School of Medicine, 450 Broadway St, Mail Code 5334, Pavilion C, 2nd Floor, Redwood City, CA 94063 (alschang@stanford.edu).

Published Online: February 24, 2016. doi:10.1001/jamadermatol.2015.4330.

Author Contributions: Drs Mohan and Chang contributed equally to this article. Drs Mohan and Chang had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Mohan, Li, A. L. S. Chang.

Acquisition, analysis or interpretation of data: All authors.

Drafting of the manuscript: Mohan, J. Chang, A. L. S. Chang.

Critical revision of the manuscript for important intellectual content: J. Chang, A. L. S. Chang.

Statistical analysis: Li.

Obtained funding: A. L. S. Chang.

Administrative, technical, or material support: Mohan.

Study supervision: A. L. S. Chang.

Conflict of Interest Disclosures: Dr Chang reported receiving research support from Genentech, Novartis, and Eli Lilly and reported being a consultant for Genentech and Novartis. No other disclosures were reported.

Funding/Support: None related to the present work. However, the Stanford Cancer Research Database was supported in part by Cancer Center Support Grant 5P30CA124435 from the National Cancer Center and by Stanford Clinical and Translational Science Award grant UL1 RR025744 from the National Center for Research Resources.

Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.