Machine learning-based construction of an oxidative stress-related model reveals molecular subtypes and tumor microenvironment infiltration signatures in colon cancer

Colon cancer is a common malignant tumor of digestive system. More and more research has shown that oxidative stress plays an important role in the development and progression of colon cancer. Oxidative stress occurs when redox homeostasis in cells is broken which is usually accompanied by overproduction of reactive oxygen species (ROS). In this study, we constructed a colon cancer prognostic model based on oxidative stress-related (OSR) genes. Lasso regression and multivariable cox proportional hazards were performed to identify the best gene signature. We established the prognostic model through OSR genes. To further explore the function of OSR genes, Gene Ontology (GO), Kyoto Encyclopaedia of Genes and Genomes (KEGG) database and Gene set enrichment analysis (GSEA) were applied to reveal relative mechanisms. External datasets were employed for validation, and molecular biology experiments were conducted to verify the functional roles of key genes in colon cancer. By screening out oxidative stress-related genes associated with prognosis, we constructed a prognosis model based on PXDNL, PTGS2, DLAT, GLRX, P4HA1 and P4HA3. Through validation with external datasets, the model accurately predicted the prognosis of colon cancer patients. Moreover, we also utilized single sample GSEA (ssGSEA) to investigate the degree of immune infiltration in the tumor environment. In addition, we conducted cell experiments to confirm siGLRX can induce tumor cell apoptosis by promoting the level of oxidative stress in cells. These results suggest that oxidative stress-related genes may serve as potential biomarkers for the prognosis of colon cancer, providing guidance for clinical treatment.