Molecular Subtypes of Merkel Cell Carcinoma Identified in Large Genomic Study

In the largest genomic study of Merkel cell carcinoma (MCC) tumors to date, investigators identified 2 distinct molecular subgroups, a ultraviolet light (UV)—driven subtype and a viral-driven subtype, which matched with tumor mutational burden (TMB) classifications.

Additionally, the investigators from the H. Lee Moffitt Cancer Center and Research Institute analyzed potential correlates of response to treatment with immunotherapy.

“These results provide a comprehensive genomic characterization of MCC, show a novel means of detecting MCPyV using targeted sequencing, place molecular subgroups in the context of other human cancers, and demonstrate the importance of using immunotherapy as early as possible,” the investigators wrote in their study that was recently published inClinical Cancer Research.

In the genomic study, 317 tumors from patients with MCC underwent comprehensive genomic profiling by Foundation Medicine between May 2013 and April 2018. The patients were profiled for 322 cancer-related genes as well as evaluated for TMB and mutational signatures. The presence of Merkel Cell Polyomavirus (MCPyV) was also detected through DNA sequencing.

Additionally, 57 of these patients with advanced or metastatic MCC who were treated at Moffitt Cancer Center were analyzed in the clinical cohort, and 38 of these patients received an immune checkpoint inhibitor. Thirty-seven patients were also tested for expression of both PD-1 and PD-L1 with dual immunohistochemistry testing, including 27 of the 36 patients with evaluable responses to immunotherapy.

Among the full genomic cohort, the median age was 71 years (range, 63-78) at the time of diagnosis and the male:female ratio was approximately 4:1, which was consistent with previous reports of the disease.

In the clinical cohort, patients were mostly non-Hispanic Caucasians (98%) and had been diagnosed with stage IIIB or IV MCC (78.9%). Almost half of the immunotherapy-treated patients (47%) received immune checkpoint inhibition with pembrolizumab (Keytruda) monotherapy and 26% had received avelumab (Bavencio) monotherapy. Radiation therapy was given prior to immunotherapy for 79% of patients, and 16% of patients did not receive any radiation. Most patients were treated with immunotherapy in the first-line (32%) or second-line (40%) setting; 2 patients were treated in the fifth-line setting.

Genomic testing revealed that there was a bimodal distribution based on TMB in the MCC samples; specimens were mostly either TMB-low (n = 175, 55%), which was defined as <6 mutations per megabase, or TMB-high (n = 117, 37%), defined as &ge;20 mutations/megabase, with only 25 samples (8%) having intermediate TMB. This trend was echoed in the clinical cohort as well with 56% of patients characterized as TMB-low and 39% as TMB-high.

In the TMB-high group, short variant mutations were noted in eitherTP53, RB1, or both genes. The most commonly altered genes in the TMB-high group wereTP53(97%),RB1(80%), and the NOTCH family of genes (50%). Whereas in the TMB-low group, no genomic alterations were found that were likely to be oncogenic drivers, with the most commonly altered genes beingTP53(13%),RB1(9%), andPTEN(7%).

Copy number alterations were noted infrequently, amounting to 24% of the TMB-high specimens, with copy number gains found in 2 genes (MYCLandMYC)and deletions found in 5 genes (RB1, PTEN, TP53, LRP1B, CDKN2A,andNF1).

MCPyV genomic DNA was detected in 36% of the overall samples by next-generation sequencing (NGS) and in 37% of the clinical cohort and was strongly associated with TMB classification. MCPyV DNA was observed in 63% of the TMB-low specimens but not in any of the TMB-high cases (P&le;.00001).

On the other hand, the UV damage mutational signature was correlated with a high TMB classification, with 94% of the TMB-high samples expressing a UV signature. This signature was found to be mutually exclusive with MCPyV integration.

In the TMB-intermediate group, 36% had a UV signature and 16% were viral-positive by NGS, and 44% could not be classified with either of the 2 molecular subgroups.

A clustering analysis of the top 10 most commonly altered genes in a variety of cancer types demonstrated that TMB-high MCC was most similar to neuroendocrine tumors such as prostate neuroendocrine carcinoma, bladder neuroendocrine carcinoma, and small cell lung cancer.

TMB-low MCC, conversely, was most similar to carcinoid tumors and virally-driven cancers such as HPV-positive head and neck squamous cell carcinoma, cervical squamous cell carcinoma, anal squamous cell carcinoma, and Kaposi sarcoma.

Thirty-six of the patients in the clinical cohort who were treatment with immune checkpoint inhibition were evaluable for response at the time of the analysis. The overall response rate (ORR) was 44%, consisting of 5 complete responses and 11 partial responses. At a median follow-up of 16.9 months (range, 14.0-27.3) from the time of diagnosis of advanced or metastatic disease, all 16 responding patients were still alive compared with only 4 patients who did not respond favorably (P<.0001).

Among the evaluable patients, 14 were categorized as having the UV signature and being TMB-high and the other 22 as viral-positive and TMB-low. The ORR was 50% in the UV signature group compared with 41% in the MCPyV-positive group (P= .63).

The response rates changed more significantly according to the line of therapy in which patients received immune checkpoint inhibition. Patients treated in the first-line setting (n = 12) had an ORR of 75%, those treated in the second line (n = 13) had an ORR of 39%, and those treated in the third line or later (n = 11) had an ORR of 18% (P= .0066). This analysis only counted the first immune checkpoint inhibitor received, as several patients received more than 1 in the course of their treatment. Line of therapy was statistically significantly associated with treatment response by multivariate analysis (P<.01).

The study authors noted that &ldquo;while both subsets respond similarly well to immunotherapy, clinical response is associated with a long-term impact on survival. Importantly, the use of other therapies prior to immunotherapy negatively influences the ultimate response to immunotherapy, suggesting that immune checkpoint blockade should be used initially when possible.&rdquo;

Twenty-seven evaluable patients were tested for both PD-1 and PD-L1 expression in the clinical cohort. There was a more significant association found between PD-1 expression and response than between PD-L1 expression and response. The ORR was 54% in PD-L1—negative patients and was 43% in PD-L1&ndash;positive patients (P= .606), whereas the ORR was 21% in PD-1—negative patients and 77% in PD-1&ndash;positive patients (P= .00598).

Reference:

Knepper TC, Montesion M, Russell JS, et al. The genomic landscape of Merkel cell carcinoma and clinicogenomic biomarkers of response to immune checkpoint inhibitor therapy [published online August 9, 2019].Clin Cancer Res. doi: 10.1158/1078-0432.CCR-18-4159.