Development and validation of prognostic model based on the analysis of autophagy-related genes in colon cancer

Background: Autophagy, a process of self-digestion, is closely related to multiple biological processes of colon cancer. This study aimed to construct and evaluate prognostic signature of autophagy-related genes (ARGs) to predict overall survival (OS) in colon cancer patients. Materials and Methods: First, a total of 234 ARGs were downloaded via The Cancer Genome Atlas (TCGA) database. Based on the TCGA dataset, differentially expressed ARGs were identified in colon cancer. The univariate and multivariate Cox regression analysis was performed to screen prognostic ARGs to construct the prognostic model. The feasibility of the prognostic model was evaluated using receiver operating characteristic curves and Kaplan-Meier curves. A prognostic model integrating the gene signature with clinical parameters was established with a nomogram. Results: We developed an autophagy risk signature based on the 6 ARGs (ULK3, ATG101, MAP1LC3C, TSC1, DAPK1, and SERPINA1). The risk score was positively correlated with poor outcome and could independently predict prognosis. Furthermore, the autophagy-related signature could effectively reflect the levels of immune cell type fractions and indicate an immunosuppressive microenvironment. Conclusion: We innovatively identified and validated 6 autophagy-related gene signature that can independently predict prognosis and reflect overall immune response intensity in the colon cancer microenvironment.