Immunotherapy Efficacy-related Risk Classifier Differentiate Prognostic Characteristics of Gastric Cancer-A Large-scale Retrospective Study.

Abstract

Prognostic signatures related to the efficacy of immunotherapy have not been determined in gastric cancer (GC). We identified the differentially expressed genes between the CR/PR and SD/PD groups with the R package "limma" (false discovery rate <0.05) in the IMvigor210 data set. The GSE13861 (n=65), GSE15459 (n=192), GSE26899 (n=93), GSE26901 (n=109), GSE28541 (n=40), GSE34942 (n=56), and GSE62254 (n=300) cohorts were merged into a training cohort (n=855). Univariate Cox regression analysis, LASSO penalized Cox regression analysis, and multivariate Cox regression analysis were jointly applied to construct the prognostic model. The Cancer Genome Atlas (TCGA)-STAD (n=371), GSE84437 (n=433), GSE26253 (n=432), and IMvigor210 (n=348) cohorts were utilized for external validation. The GC patients were divided into 16 subgroups according to clinical features for universal applicability validation. Repeated validation confirmed that the overall survival of the high-risk (HR) group was significantly reduced compared with that of the low-risk (LR) group. The HR group showed a higher infiltration abundance of regulatory T cells, macrophages, T follicular helper cells, and natural killer T cells, whereas the infiltration levels of activated CD4 T cells and monocytes were upregulated in the LR group. The calcium, TGF-β, MAPK, Hedgehog, and KRAS signaling pathways were overactivated in the HR group, while the hallmarks related to DNA damage repair and metabolism were enriched in the LR group. In addition, the LR group had high tumor mutation burden, FLG, and OBSCN mutations. A prognostic risk classifier for GC patients was identified and validated by carrying out a multicenter retrospective study.