Tumor mutational burden (TMB) may be a promising predictive biomarker of response to PD-1/PD-L1 targeting in MSI-H colorectal cancer.

Abstract

43 Background: PD-1 targeting with pembrolizumab or nivolumab leads to durable clinical benefits in patients (pts) with microsatellite instability-high (MSI-H) tumors. However, 30-35% of mCRC pts with MSI-H tumors will experience progressive disease (PD) as a best response when treated with anti-PD1 agents, highlighting the need of additional predictive biomarkers. Methods: We performed a retrospective multi-center clinical investigation to evaluate the impact of TMB, age, gender, stage at initial presentation, pattern of metastatic disease, tumor grade, and RAS/RAF status on response to anti-PD1/PD-L1 in MSI-H mCRC. TMB and MSI status were determined by hybrid capture-based next-generation sequencing (Foundation Medicine [FM]). The TMB distribution in MSI-H CRC was estimated from a large data set from FM. Results: 22 eligible MSI-H mCRC pts were identified across 5 cancer centers: 19 pts received pembrolizumab, 1 pt received nivolumab, 1 pt received nivolumab + ipilimumab, and 1 pt received durvalumab + tremelimumab. Among tested variables, TMB (as a continuous variable) showed the strongest association with an objective response (OR; p < 0.001). Also, both univariate and multivariate analyses supported that TMB serves as an independent prognostic variable in predicting progression-free survival (PFS; p < 0.001 and p < 0.01, respectively). Using log-rank statistics, the optimal predictive cut-point for TMB was estimated between 37-41 mutations/Mb to dichotomize pts into TMBhigh and TMBlow groups. All 13 pts (100%) with TMBhigh had an OR, while only 2/9 (22%) pts with TMBlow had an OR and 6/9 had PD. The median PFS for TMBhigh pts has not been reached (no progressors, median follow-up > 18 mos), while the median PFS for TMBlow pts was 2 mos. Amongst 821 MSI-H CRC cases tested at FM, the 25th, 35th, 50th and 75th percentile TMB cutoffs were 33.1, 37.4, 46.1, and 61.8 mutations/Mb, respectively. Our optimal TMB cut-point range suggests that MSI-H mCRC with the lowest 35th percentile of TMB have a low likelihood of benefit from anti-PD1. Conclusions: These TMB findings require validation in prospective trials and may guide the sequencing of PD-1 inhibitor monotherapy in MSI-H mCRC.