Abstract 1283: Integrative genome-scale analysis of TCGA dataset for features of tumor immunity

Abstract

Abstract Background: Only a small portion of patients respond to immune checkpoint inhibitors (ICI), and the response rate varies among different cancer types. While an inflammatory tumor microenvironment (TME) is often believed to be critical for positive ICI response, the inflammatory status of different cancers and the molecular differences between the inflamed and non-inflamed tumors are yet to be elucidated. Methods: We performed an integrative genome-scale analysis of TCGA dataset, using 9 emerging predictive biomarkers of ICI treatment to evaluate cancer immunity. The biomarkers include tumor mutation burden (TMB), innate anti-PD1 resistance (IPRES), immune infiltrated score (IIS), T-cell infiltrated score (TIS), cytolytic activity index (CYT), antigen presentation machine (APM), angiogenesis, and IFN-γ 18-gene (IFN18). We first analyzed the correlation between the biomarkers, then combined the representative biomarkers to classify tumors into ‘inflamed' and ‘non-inflamed' subtypes for comparison to identify prognostic significance of ‘inflamed' tumor and associated genomic features. Results: A total of 9338 tumors from 24 types of cancer were analyzed. Of all analyzed cancer types, non-small cell lung carcinoma and melanoma have the highest TMB, which was consistent with previous report. Renal cell carcinoma, despite the moderate TMB, has the highest cytolytic activity as determined by CYT, which is consistent with the descent response rate to ICI (25~46%). TMB was not strongly correlated with immunity aspects reflected by other biomarkers in most cancers, and IPRES was actually positively associated with T-cell inflammatory status in some cancers including bladder cancer. The IFN-18 signature is highly associated with the antigen presentation activity and T cell cytolytic activity as reflected by the APM and CYT signature respectively. The IIS and TIS are largely overlapping. Taking together, an integrated signature (IFN18+IIS) might have more broad predictive power and was therefore used to identify the “inflamed” (IFN18highIIShigh) and “non-inflamed” (IFN18lowIISlow) tumors in 5 cancer types. We found that patients with more inflamed TME have significantly longer overall survival in lung adenocarcinoma, liver hepatocellular carcinoma, colon adenocarcinoma and head and neck squamous cell carcinoma. Genes involved in cell cycle are generally downregulated in “inflamed” tumors. Genes involved in DNA repair are mostly down-regulated in lung squamous cell carcinoma, and genes involved in chromatin organization are down-regulated in liver hepatocellular carcinoma. Conclusions: Our findings suggest that a high TMB does not necessarily associate with inflammation, and IPRES has limited value as a pan-cancer predictive biomarker. These findings will contribute to better understanding of the most up-to-date predictive biomarkers and the application of ICI in different cancers. Citation Format: Binchen Mao, Sheng Guo, Henry Li, Xinzhong Wang, Ying Pan. Integrative genome-scale analysis of TCGA dataset for features of tumor immunity [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 1283.