Abstract
Colon cancer (CC) stands as one of the most prevalent malignant neoplasms worldwide. Despite extensive investigations on the function of T cells in antitumor immunity and dynamics of tumor microenvironment (TME), their precise molecular contributions to the CC progression remain incompletely characterized. Differential gene expression analysis was implemented leveraging TCGA-COAD transcriptomic data, followed by the identification of T cell signature genes using single-cell RNA sequencing (scRNA-seq) dataset. Through intersectional analysis and subsequent prognosis-related gene screening using least absolute shrinkage and selection operator (LASSO) regression and multivariate Cox proportional hazards models, a prognostic model was established. Moreover, its performance was evaluated via receiver operating characteristic (ROC) curves. Participants were split into high-risk and low-risk cohorts based on risk scores, to explore potential immunological differences between groups. A prognostic model was developed based on seven genes, encompassing UBE2N, TUBA1C, FXR1, CBLB, YTHDC1, GPRIN3, and AGPAT2. The area under the ROC curve (AUC) for the training cohort at 3, 5, and 7 years reached 0.676, 0.715, and 0.721, respectively. External validation using three GEO datasets demonstrated consistent predictive performance of the model. The AUC values at 3, 5, and 7 years were 0.632, 0.617, and 0.582 in GSE39582, 0.689, 0.755, and 0.951 in GSE17537, and 0.667, 0.653, and 0.649 in GSE161158. The identified T cell signature genes may function as potential therapeutic targets, while the developed prognostic model and nomogram may facilitate clinical decision-making for CC management.