Overall, 10 studies using vismodegib met search criteria for inclusion in our systematic review and pooled analysis; 1 additional article that is in press was added for completeness. Eight of 11 vismodegib articles were able to be pooled for analysis. Sonidegib did not yield enough results to merit a pooled analysis.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Jacobsen AA, Aldahan AS, Hughes OB, Shah VV, Strasswimmer J. Hedgehog Pathway Inhibitor Therapy for Locally Advanced and Metastatic Basal Cell Carcinoma: A Systematic Review and Pooled Analysis of Interventional Studies. JAMA Dermatol. 2016;152(7):816–824. doi:10.1001/jamadermatol.2016.0780
Hedgehog pathway inhibitors (HPIs) were made available by US Food and Drug Administration approval in 2012 for vismodegib and 2015 for sonidegib. Both target the Smoothened molecule and are indicated for locally advanced basal cell carcinoma (laBCC) and metastatic basal cell carcinoma (mBCC).
To evaluate clinical experience with HPIs, including efficacy and adverse effects.
We conducted a systematic review in concordance with the PRISMA guidelines of PubMed, the Cochrane Central Register of Controlled Trials, ClinicalTrials.gov, and EMBASE, using search terms “vismodegib,” “sonidegib,” “Erivedge,” “Odomza,” “basal cell carcinoma,” and “BCC.”
We included clinical trials, retrospective medical record reviews, and prospective case series that used HPIs for the treatment of laBCC or mBCC in human subjects. Individual case reports and limited, retrospective case series were excluded from our review.
Data Extraction and Synthesis
Data were extracted independently by 2 reviewers on a predesigned, standardized form.
Main Outcomes and Measures
The following data were recorded: number of patients with laBCC or mBCC, dose and frequency of drug administration, median duration of treatment, clearance and recurrence rates, and adverse effects.
Eleven vismodegib articles (published between 2009 and 2015) met criteria for inclusion, and 8 articles were able to be pooled for analysis. The 8 pooled articles included 744 total patients with 704 patients clinically evaluable. Sonidegib did not yield enough publications for a formal analysis. Objective response to vismodegib for laBCC had a weighted average of 64.7% (95% CI, 63.7%-65.6%); complete response averaged 31.1% (95% CI, 30.4%-31.8%). Objective response for mBCC was 33.6% (95% CI, 33.1%-34.2%); complete response averaged 3.9% (95% CI, 3.3%-4.4%). Median duration of therapy was 35.8 weeks (95% CI, 35.1-36.5 weeks).
Conclusions and Relevance
In a systematic review of HPIs for laBCC and mBCC, vismodegib, but not sonidegib, had enough studies to warrant a pooled analysis. Vismodegib was identified to have a significant, consistent effect on the median duration of therapy of laBCC and mBCC. While mBCC responses are superior to any traditional approach, the response rate for laBCC might be considered in the context of other standard treatment options including surgery and radiation therapy.
Basal cell carcinoma (BCC) is the most common skin cancer in the United States with approximately 2 million Americans affected every year,1 producing an incidence of 0.1% to 0.5% per year.2,3 Most BCCs can be cured by standard surgical or radiation interventions; however, there are a small percentage of people who develop locally advanced BCC (laBCC) or metastatic BCC (mBCC), presenting a challenge to clinicians.
In mBCC, the median survival is 8 months,2 but more recent studies suggest that survival may range up to 54 months or more.4,5 There have been reports6 of cisplatin improving tumor response, but progression-free survival and overall survival are not affected. There is no standard therapy for patients with mBCC, and until recently, treatment options were limited.2,7,8 Both hedgehog pathway inhibitors (HPIs) vismodegib and sonedigib have received US Food and Drug Administration (FDA) approval for this use and there is no reasonable alternative therapy at present.
Locally advanced BCC is a clinical situation whereby surgery or radiation may have previously failed or may be considered inappropriate by the clinician.2,6,8 However, the FDA labeling of HPIs for laBCC therapy is not salvage therapy or palliative therapy, leaving its use to the discretion of the prescriber. Rather, the FDA labeling has 2 parts: first, HPIs are indicated for “locally advanced basal cell carcinoma that has recurred following surgery,”9 and second, HPI therapy is indicated for patients “who are not candidates for surgery, and who are not candidates for radiation.”9 While the second situation provides an option for laBCC where there may be no other, there are other well-established options in the first indication. Additionally, until recently, there was no consensus on the definition of laBCC.10 The advent of HPIs has prompted the need for an expert consensus on the defining features of laBCC to aid physicians who are determining whether to initiate HPI treatment.
Two HPI medications were approved by the FDA: vismodegib in 2012 and sonidegib in 2015. Both are small-molecule inhibitors of the Smoothened receptor (SMO) of the Hedgehog signalling pathway. Neither has been subjected to a phase 3 randomized clinical trial, the standard for most medications that undergo FDA approval.
Vismodegib received priority FDA approval based on the promising results of ERIVANCE,11 a nonrandomized phase 2 parallel cohort study of 104 patients with laBCC or mBCC. In this study, there was a 43% response rate in patients with laBCC and a 30% response rate in patients with mBCC.12 Sonidegib more recently obtained approved after the BOLT trial,20 a randomized double-blind phase 2 trial that showed a 36% objective response rate for the dose that was subsequently approved by the FDA.
In the 4 years since HPI therapy became available, more studies have been completed. In an effort to evaluate the safety and efficacy of HPI therapy, this report summarizes the available data on vismodegib and sonidegib including response rates, adverse events, and adverse effect profiles in patients with advanced BCC.
Question What is the clinical experience to date of the safety and efficacy of hedgehog pathway inhibitors (HPIs) for locally advanced and metastatic basal cell carcinoma (BCC)?
Findings In this systematic review, safety and efficacy of the hedgehog pathway inhibitors vismodegib and sonidegib were evaluated with a pooled statistical analysis of interventional studies. Objective response for locally advanced BCC (laBCC) was nearly twice as high as for metastatic BCC (mBCC) and is higher than previously reported.
Meaning Hedgehog pathway inhibitors are an effective and well-tolerated therapy for laBCC and mBCC.
The review was conducted according to the PRIMSA guidelines for systematic reviews.13 Details of the protocol are registered on PROSPERO (http://www.crd.york.ac.uk/PROSPERO/display_record.asp?ID=CRD42015025475).
We sought to obtain available articles, both published and in press, that studied vismodegib for the treatment of laBCC or mBCC. A broad literature search in PubMed, the Cochrane Central Register of Controlled Trials (CENTRAL), ClinicalTrials.gov, and EMBASE was performed in August 2015 using the following terms: vismodegib, Erivedge, basal cell carcinoma, and BCC. The search was then narrowed to include only clinical trials, retrospective medical record reviews, and prospective case series that used vismodegib for the treatment of laBCC or mBCC. Individual case reports and limited case series were excluded from our review owing to high risk of publication bias. Only articles written in English involving human subjects with laBCC or mBCC were included. After this initial filtering process, the remaining titles and abstracts were screened for relevance independently by 2 authors. Studies that analyzed vismodegib in conjunction with other interventions, such as an adjuvant to surgery or radiation, were excluded. Full-text articles were reviewed in studies that appeared to meet the inclusion criteria or when there was disagreement between the 2 authors about eligibility. For completeness, the references of each included article were also reviewed.
Because sonidegib was recently approved, we modified our search criteria to include studies of sonidegib for laBCC and mBCC using the following search terms: sonidegib, Odomza, basal cell carcinoma, and BCC. Inclusion and exclusion criteria remained the same as that of the vismodegib studies.
Data were extracted independently by 2 reviewers and entered into a standardized, predesigned Microsoft Excel form. The following data were recorded: number of total patients and those with laBCC or mBCC, dose and frequency of drug administration, median duration of drug exposure, clinical or histological clearance rates, recurrence rates, and adverse effects. When available, we used central or independent faculty review results. Each reviewer also assessed the quality of reporting and risk of bias for the included studies, qualitatively noting whether each study had a good, intermediate, or poor quality of reporting. We also evaluated whether studies included a declaration of sponsorship by a pharmaceutical manufacturer. The search was rerun in September, October, and November 2015 to check for recently published articles.
We calculated raw proportions of events divided by the total number of clinically evaluable patients. Owing to the heterogeneity in study size and thus large variations in proportions, we calculated weighted averages of events by the number of clinically evaluable patients. As laBCC and mBCC represent distinct clinical entities, we also conducted a subgroup analysis of response rates for each condition separately. Analyses were conducted using IBM SPSS Statistics, version 22 (IBM Corporation).
The preliminary search for vismodegib interventions returned 207 published articles, and after the initial search reseults were filtered, 103 articles remained. Further screening of abstracts for relevance revealed 38 full-text articles that were reviewed independently by 2 authors. Overall, 10 articles with vismodegib as the therapeutic agent were selected, with an additional article included that is currently in press,14 for a total of 11 articles. The Figure depicts the article selection process in detail. The selected vismodegib articles include 1 randomized placebo-controlled trial,3 2 phase 1 clinical trials,2,7 2 open-label trials,6,8 1 prospective cohort study,12 3 prospective case series,14-16 and 2 retrospective medical record reviews.17,18 The dates of publication ranged from 2009 to 2015. A detailed description of individual study designs and treatment outcomes can be found in Table 1. Common adverse effects reported in each study are described in Table 2. Three studies had limited data available regarding vismodegib efficacy and adverse effects, but we felt it was important to include them as a reference. Our search for sonidegib interventions did not yield enough results to merit a systematic review or pooled analysis. However, we will comment on the phase 119 and phase 2 (BOLT)20 trial for completeness.
Two studies (18%) collected data retrospectively while 9 (82%) collected data prospectively. Based on methods of patient recruitment, overall study design, consistency and guidelines used in reporting outcomes, number of patients included, and duration of follow-up, 4 studies (36%) were assessed to be of good reporting value; 5 (45%), intermediate; and 2 (18%), poor.
Of the 11 studies included in the review, we were able to pool the results of 8 for statistical analysis (Table 3). The 8 pooled articles included 744 total patients with 704 patients clinically evaluable. The remaining 3 studies2,3,7 were excluded owing to heterogeneity in vismodegib dose and lack of appropriate subgroup data available. Of the 8 pooled studies, there were 744 patients with 704 patients clinically evaluable. All patients in this statistical analysis received 150 mg of vismodegib daily. The range of response to vismodegib, either complete or partial, was 43.2% to 100.0%, with a weighted average of 62.1% (95% CI, 61.2%-63.0%). Complete response was observed to range from 0.0% to 54.2% between the studies, with a weighted average of 28.0% (95% CI, 27.2%-28.8%). Partial response ranged from 29.3% to 66.7%, with a weighted average of 34.1% (95% CI, 33.7%-34.5%). Patients with stable disease in response to vismodegib treatment ranged from 0.0% to 53.4%, with a weighted average of 31.4% (95% CI, 30.5%-32.3%). The range of progressive disease was 0.0% to 50.0%, with a weighted average of 4.1% (95% CI, 3.8%-4.5%). Median duration of drug exposure ranged from 11.0 to 51.6 weeks, with a weighted duration of 35.8 weeks (95% CI, 35.1-36.5 weeks).
We conducted a subgroup analysis of vismodegib efficacy in laBCC vs mBCC. The response range (partial or complete) in patients with laBCC was 28.0% to 100%, with a weighted average of 64.7% (95% CI, 63.7%-65.6%); in patients with mBCC, the response range was 30.8% to 38.0%, with a weighted average of 33.6% (95% CI, 33.1%-34.2%). Patients with laBCC exhibited a complete response in 0.0% to 54.1% of cases, with a weighted mean of 31.1% (95% CI, 30.4%-31.8%); patients with mBCC had a complete response in 0.0% to 7.0% of cases, with a mean of 3.9% (95% CI 3.3%-4.4%). Partial response in patients with laBCC was seen in 25.5% to 66.7%, with a mean of 33.6% (95% CI, 33.1%-34.0%); partial response in patients with mBCC was 25.6% to 33.3%, with a mean of 29.8% (95% CI, 29.1%-30.4%). Disease remained stable in 0.0% to 48.2% of patients with laBCC, with a weighted average of 27.2% (95% CI, 26.4%-27.9%); in patients with mBCC, disease remained stable in 34.0% to 66.7% with a weighted average of 49.9% (95% CI, 47.8%-52.0%). Disease progressed in 0.0% to 28.6% of laBCC cases, with a mean of 3.1% (95% CI, 2.7%-3.4%); in mBCC cases, disease progressed in 0.0% to 14.0%, with a weighted average of 7.7% (95% CI, 6.9%-8.6%).
Nine of 11 vismodegib studies were included in the pooled analysis of adverse effects resulting from vismodegib (Table 4); the 2 remaining studies had insufficient data to be included. A total of 803 patients were included in the adverse effect profile analysis. The range of serious adverse effects was 0.0% to 38.5%, with a weighted average of 21.4% (95% CI, 20.9%-21.9%). The proportion of patients discontinuing vismodegib due to adverse effects ranged from 0.0% to 53.8%, with a weighted average of 28.2% (95% CI, 27.3%-29.1%). Muscle spasms ranged from 20.0% to 100%, the weighted average being 66.4% (95% CI, 65.8%-67.0%). The range of dysguesia was 20.0% to 75.0%, with a weighted average of 57.3% (95% CI, 56.7%-58.0%). Weight loss affected a range of 16.0% to 83.3% of patients, with a weighted average of 33.4% (95% CI, 32.7% to 34.2%). Fatigue affected 7.7% to 40.4% of patients, with a weighted average of 20.1% (95% CI, 19.5%-20.7%). Alopecia ranged from 20.0% to 75.0%, with a weighted average of 61.1% (95% CI, 60.8%-61.4%). New-onset squamous cell carcinoma was reported in 0.8% to 20.0% of cases, with a weighted average of 1.3% (95% CI, 1.2%-1.5%). Finally, amenorrhea ranged from 27.6% to 100%, with a weighted average of 32.9% (95% CI, 32.1%-33.6%).
We were unable to pool the results of sonidegib trials but comment on the 2 available trials to date. In the phase 1, open-label trial,1916 patients with BCC were treated with sonidegib, and 6 patients (37.5%) with BCC achieved an objective response at varying doses of sonidegib. The BOLT trial20 is a multicenter, randomized, double-blind phase 2 trial to assess the efficacy and tolerance of sonidegib. We report the results of patients assigned to the 200 mg daily group (n = 79) because this is the FDA-approved dosage. Overall, 20 patients (36%) receiving this dose achieved an objective tumor response when assessed by central review, 2 patients with complete response and 18 with a partial response. When assessed by the investigators, 31 patients (56%) were considered to have achieved an objective response.
In the 200-mg group (FDA-approved dose) of the phase 2 BOLT trial,20 39 patients (49%) experienced muscle spasms; 34 (43%), alopecia; 30 (38%), dysgeusia; 23 (29%), fatigue; and 21 (27%), weight loss. Serious adverse events were reported in 11 patients (14%) of patients. New-onset squamous cell carcinoma was observed in 3 patients (3.8%) in the 200-mg dose group.
In our pooled analysis, response rates varied greatly between studies, and the weighted average of complete or partial response with vismodegib (62.1% [95% CI, 61.2%-63.0%]) may be misleading for patients who are looking for a cure. Thus, it is important to distinguish between the response rates for laBCC and mBCC—64.7% and 33.6%, respectively. Particularly in cases of mBCC, complete response rate is low at 3.9% compared with 31.1% in laBCC. However, patients with both mBCC and laBCC have encouraging rates of partial response (33.6% and 29.8%, respectively) and stable disease (27.2% and 49.9%, respectively). This supports previously reported opinions that vismodegib may be more useful as a means to control laBCC or mBCC rather than provide a definitive cure.15 Additionally, in a photo analysis of patients from the ERIVANCE study,34 investigators found a clinical benefit in 76.2% of evaluable patients, which is higher than the objective response rates in clinical trials. Recommendations based on the quality of evidence are listed in Table 4.
There are several recently published articles on the possible use of vismodegib as an adjuvant to surgery or radiation,24,35-38 but more experience is needed before establishing the role of vismodegib to this extent. However, FDA indications for vismodegib, as stated in the introduction, include patients who are not candidates for radiation and/or surgery. This use of vismodegib has the potential to cause further confusion for physicians who are deciding on treatment plans for their patients. Again, the definition of laBCC and specific indications for HPI treatment will need to be further outlined.
The most common adverse effects from HPI therapy are well reported. In our pooled analysis, the 3 most common adverse effects were muscle spasms (66.4%), alopecia (61.1%) and dysguesia (57.3%). More than a quarter of patients in our weighted analysis (28.2% [95% CI, 27.3-29.1]) discontinued therapy due to adverse effects. We also aimed to analyze 2 emerging concerns that at present do not carry a high enough reported incidence to warrant labeling requirement by the FDA as a common adverse reaction (incidence >10%).9 The first is amenorrhea. We found that 32.9% (95% CI, 32.1%-33.6%) of women of childbearing age experienced amenorrhea when exposed to vismodegib. However, the majority of studies did not comment on this effect. Future studies should include exploration of this common adverse effect and other potential adverse effects relating to disruptions in follicle-stimulating hormone actions.27 While the FDA does not yet indicate HPIs for basal cell nevus syndrome, clinical trials are being conducted to formally evaluate their role in this syndrome.3,39 Amenorrhea is a particular concern, because it is possible that patients with basal cell nevus (Gorlin) syndrome, who develop tumors at an early age, might receive this medication for prolonged periods of time while still premenopausal. As such, we anticipate the true incidence of this adverse event could increase, raising management questions.
We also sought to identify the risk of developing a new squamous cell carcinoma while exposed to vismodegib. We found the percentage of patients developing a new squamous cell carcinoma was relatively low compared with other adverse effects, weighted at 1.3% (95% CI, 1.2%-1.5%), ranging from 0.8% to 20.0%, but still a concern that should be monitored more consistently as patients receive prolonged HPI treatment.
Adverse effects of vismodegib are known to be a significant limiting factor to treatment duration, with more than a quarter of patients in our pooled analysis discontinuing treatment due to adverse effects. This brings up the possibility of vismodegib being used a pulse treatment, an opinion echoed by other clinicians.15 A formal analysis of the effect of drug holidays or multiple rounds of shorter-term treatment was not possible due to a lack of data. Further, the range of follow-up time in each of the studies is relatively short, ranging from 7.3 to 22.4 months. Therefore, the proportion of patients experiencing long-term adverse effects even after drug discontinuation is not known. Some adverse effects like amenorrhea are thought to be reversible, but more data are needed to determine an accurate time frame of reversibility.27 Alopecia, a common adverse effect of cancer therapy that is one of the most distressing for patients,40 is usually reversible but persistent cases have been reported,41 again strengthening the argument for longer follow-up periods.
Another adverse effect that merits comment is elevated creatine phosphokinase (CPK). Monitoring CPK is required with sonidegib33 but not vismodegib9 because CPK levels were not monitored in vismodegib trials. However, elevated CPK during vismodegib therapy has been reported has a treatment-limiting adverse event in 1 case series.24 It is reasonable that it may be monitored on a case-by-case basis in patients taking vismodegib who have additional risk factors for elevated CPK, such as concurrent mediations known to elevate CPK levels (ie, statins).
Despite a high proportion of patients experiencing adverse events, very few deaths are directly attributable to HPI therapy. In the ERIVANCE trial,12 none of the 24 deaths were considered a result of HPI therapy. Similarly, in the STEVIE trial,8 2 of 21 deaths were investigated as possibly drug-related, but both patients had confounding medical histories.8
The issue of resistance or regrowth during and/or after HPI therapy is an emerging issue. Danial et al5 showed that patients can demonstrate resistance to both vismodegib and sonidegib, which both target the Smoothened molecule but are distinct in molecular structure.42 Several Smoothened mutations were identified by Danial et al,5 but even patients without an identifiable mutation exhibited resistance. One potential issue is the possibility of tumor resistance or regrowth occurring during drug holidays. As the optimum time frame for pulse dosing is not known, this possibility needs to be monitored carefully.
We were unable to make comparisons of vismodegib to sonidegib in this analysis. While both the ERIVANCE12 and BOLT20 trials used an independent review to assess tumor response, the Response Evaluation Criteria In Solid Tumors (RECIST)44 criteria used to assess response in the BOLT trial were more stringent than that in the ERIVANCE and STEVIE trials.8,11,20,43 However, we suspect that response rates and adverse effect profiles are similar owing to class effect. The study by Danial el al5 tested the effect of sonidegib in patients with advanced BCC who become resistant to vismodegib; patients subsequently demonstrated resistance to sonidegib, supporting our hypothesis that sonidegib and vismodegib exhibit comparable class effects.
Heterogeneity in study methods was a significant limitation to our study. While the majority of studies, particularly the larger ones, used RECIST44 to define tumor response to treatment, others used a more subjective approach. However, we opted to include studies that did not use RECIST but still reported tumor clearance in terms of “partial, complete, stable, or progressive.” Additionally, not every study used histologic clearance to evaluate tumor response. Other studies were simply not able to be included in the pooled analysis because they did not report tumor clearance in a method consistent with the majority of previously published studies. This heterogeneity has the potential to affect statistical analysis and outcomes. Additionally, while we attempted to identify patients who were included across multiple publications, it is still possible that patients were duplicated in our review.
As is inherent in observational studies, we acknowledge the possibility of publication bias. We tried to decrease this possibility by only including retrospective medical record reviews, prospective case series, and clinical trials, excluding case series and case reports that were not conducted in an all-inclusive manner. Furthermore, while some of the studies used an independent faculty review,12 the majority of studies were investigator or clinician assessed, lending to a possible bias for reporting response rates.
With the advent of novel modalities to treat laBCC and mBCC, there needs to be a consensus on the definition of advanced BCC. Before HPIs, a variety of descriptions, including “advanced,” “extreme,” “complex,” and “inoperable” were used to label what might be called laBCC.45 Indeed, few of the studies included in our review have attempted to define the features of what they considered to be laBCC as an indication for therapy. Defining advanced BCC will become increasingly important as HPI use becomes more widespread and third-party payers request proof of need for HPI therapy.
Our review summarizes the efficacy and incidence of adverse events of HPI therapy for laBCC and mBCC and should serve to guide physicians in treatment management. Hedgehog pathway inhibitors are an effective, well-tolerated therapy for laBCC and mBCC and have great potential to improve quality of life for many patients who have limited treatment options. The clinical experiences of these newly FDA-approved medications should be critically evaluated to ensure best practices for future use.
Corresponding Author: John Strasswimmer, MD, PhD, 2605 W Atlantic Ave #204, Delray Beach, FL 33445 (email@example.com).
Accepted for Publication: February 27, 2016
Published Online: April 20, 2016. doi:10.1001/jamadermatol.2016.0780.
Author Contributions: Dr Strasswimmer and Ms Jacobsen had full access to all of 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: Jacobsen, Strasswimmer.
Acquisition, analysis, or interpretation of data: Jacobsen, Aldahan, Hughes, Shah.
Drafting of the manuscript: Jacobsen, Aldahan, Hughes, Shah.
Critical revision of the manuscript for important intellectual content: Hughes, Strasswimmer.
Statistical analysis: Jacobsen.
Administrative, technical, or material support: Strasswimmer.
Study supervision: Strasswimmer.
Conflict of Interest Disclosures: Dr Strasswimmer has served as a paid consultant to Novartis, the manufacturer of sonidegib, and to Genentech-Roche, the manufacturer of vismodegib.
Create a personal account or sign in to: