Genomic and immune landscape of metastatic melanoma (MM) treated with pembrolizumab (PEM).

Abstract

184 Background: Predictive biomarkers for PEM in MM is an active area of research. Methods: INSPIRE (NCT02644369) is a biomarker-driven Princess Margaret Cancer Centre initiative to evaluate genomic and immunologic changes in tumor and blood of patient (pts) treated with PEM at 200 mg IV Q3W. Baseline (BL) and on treatment (post 2-3 doses of PEM) fresh tumor biopsies were collected for DNA/RNA sequencing, immune-profiling, and tumor PD-L1 expression by immunohistochemistry (QualTek), with longitudinal blood collection for immunophenotyping. Results: As of 9/2017, 11 MM pts (9 cutaneous, 1 choroidal, 1 mucosal) were enrolled. Median treatment duration was 6 months (0.75 – 14.25); 7 pts remain on study. 7/11 (64%) pts had partial response (PR), 3/11 (27%) progressive disease (PD) and 1 was not evaluable (NE) per RECIST 1.1. Based on data for 8 pts (6 PR, 2 PD), higher percentage (%) PD-1 (mean 70 vs 12%), TIGIT (82 vs 43%) and 4-1BB (23 vs 4%) positivity on CD8 T cells in tumor vs blood was seen (all p < 0.05). In BL blood, lower % TIGIT positive CD8 T cells was detected in pts with PR vs PD (38 vs 58%, p < 0.05). T cell profile of 4 pts with paired BL and post PEM samples is shown in Table 1. Mean tumor PD-L1 score was 24% (0 - 90%), with no correlation to PEM response. Genomic analysis on 8 pts (5 PR, 2 PD, 1 NE) found mean tumor mutation burden (TMB) of 538. B2M deletion was seen in 2/2 PD and 0/6 PR pts. Among pts with PRs, 4/6 (67%) had TAP1 and TAP2 amplification post PEM therapy. HLA-A and HLA-B amplifications were each found in 3/6 (50%) PR pts, with only 1 present at BL. TMB was not associated with response to PEM. Conclusions: In this exploratory analysis of MM pts treated with PEM, there was significantly discordant T cell phenotype in tumor vs blood at BL. Our findings suggest B2M deletion to be associated with PEM resistance. The observed development of HLA-A/B and TAP1/2 amplification in 50-67% pts with PR post PEM suggests potential predictive significance, although further validation is warranted. Clinical trial information: NCT02644369. [Table: see text]