Constructing a novel gene signature derived from oxidative stress specific subtypes for predicting survival in stomach adenocarcinoma

Oxidative stress (OS) response is crucial in oncogenesis and progression of tumor. But the potential prognostic importance of OS-related genes (OSRGs) in stomach adenocarcinoma (STAD) lacked comprehensive study. STAD clinical information and transcriptome data were retrieved from the Gene Expression Omnibus and The Cancer Genome Atlas databases. The prognostic OSRGs were filtered via the univariate Cox analysis and OSRG-based molecular subtypes of STAD were developed using consensus clustering. Weighted gene co-expression network analysis (WGCNA) was subsequently conducted to filter molecular subtype-associated gene modules. The prognosis-related genes were screened via univariate and least absolute shrinkage and selection operator Cox regression analysis were used to construct a prognostic risk signature. Finally, a decision tree model and nomogram were developed by integrating risk signature and clinicopathological characteristics to analyze individual STAD patient’s survival. Four OSRG-based molecular subtypes with significant diversity were developed based on 36 prognostic OSRGs for STAD, and an OSRGs-based subtype-specific risk signature with eight genes for prognostic prediction of STAD was built. Survival analysis revealed a strong prognostic performance of the risk signature exhibited in predicting STAD survival. There were significant differences in mutation patterns, chemotherapy sensitivity, clinicopathological characteristics, response to immunotherapy, biological functions, immune microenvironment, immune cell infiltration among different molecular subtypes and risk groups. The risk score and age were verified as independent risk factors for STAD, and a nomogram integrating risk score and age was established, which showed superior predictive performance for STAD prognosis. We developed an OSRG-based molecular subtype and identified a novel risk signature for prognosis prediction, providing a useful tool to facilitate individual treatment for patients with STAD.