Patients deriving long-term benefit from immune checkpoint inhibitors demonstrate conserved patterns of site-specific mutations.

Immune checkpoint inhibitors (ICIs) have revolutionized cancer therapy and are now the preferred treatment for several tumor types. Though ICIs have shown remarkable efficacy in several cancer histologies, in many cases providing long-term disease control, not all patients will derive clinical benefit from such approaches. Given the lack of a reliable predictive biomarker for therapeutic responses to ICIs, we conducted a retrospective analysis of publicly available genomic data from a large pan-cancer cohort of patients receiving ICI-based immunotherapy. Consistent with previous results, patients in the combined cohort deriving a long-term survival benefit from ICIs were more likely to have a higher tumor mutational burden (TMB). However, this was not uniform across tumor-types, failing to predict for long-term survivorship in most non-melanoma cancers. Interestingly, long-term survivors in most cancers had conserved patterns of mutations affecting several genes. In melanoma, this included mutations affecting TET1 or PTPRD. In patients with colorectal cancer, mutations affecting TET1, RNF43, NCOA3, LATS1, NOTCH3, or CREBBP were also associated with improved prognosis, as were mutations affecting PTPRD, EPHA7, NTRK3, or ZFHX3 in non-small cell lung cancer, RNF43, LATS1, or CREBBP mutations in bladder cancer, and VHL mutations in renal cell carcinoma patients. Thus, this study identified several genes that may have utility as predictive biomarkers for therapeutic responses in patients receiving ICIs. As many have no known relationship to immunotherapy or ICIs, these genes warrant continued exploration, particularly for cancers in which established biomarkers such as PD-L1 expression or TMB have little predictive value.