Tumor-infiltrating lymphocytes (TILs) in gastric cancer are closely related to clinical prognosis; however, little is known regarding the immune microenvironment in this disease. Thus, RNA-sequencing data from gastric cancer patients were downloaded from the Gene Expression Omnibus (GEO). The proportion of immune cells was determined based on a deconvolution algorithm (CIBERSORT), and gene expression profiles were analyzed in the context of clinical outcomes to construct an immune risk score. Data were analyzed using least absolute shrinkage and selection operator (LASSO) and multivariable Cox regression, to identify prognostic markers of gastric cancer survival. The model included four immune cell types: neutrophils, plasma cells, activated CD4+ memory T cells, and T follicular helper cells. Patients were classified into two subgroups based on risk score, and a significant difference in overall survival (OS) was seen between the subgroups in both the training and testing cohorts, particularly in patients with tumor stages ≥T3. Multivariable analysis revealed that both T-stage and risk score were independent prognostic factors for gastric cancer survival [hazard ratio (HR) 1.505; 95% confidence interval (CI) 1.043-2.173, HR 1.686; 95% CI 1.367-2.080]. Risk scores and clinical factors were then integrated into a nomogram to build a model with both good discriminatory power and accuracy in predicting clinical outcomes. Further analysis using gene set enrichment analysis (GSEA) identified strong associations of immune risk with TGF-β and tumor metastasis-related pathways, which could inform research on the molecular mechanisms of gastric cancer. Collectively, the data presented here suggest that an immune risk model can make an important contribution to predictions prognosis in gastric cancer patients.
Keywords: GEO; TIL; gastric cancer; immune risk score model; prognosis.
Copyright © 2020 Xie, Mo, Li, Cai, Ruan, Ye and Mao.