Abstract
BACKGROUND: Colon cancer is a malignant tumor with high malignancy and a low survival rate whose heterogeneity limits systemic immunotherapy. Transforming growth factor-β (TGF-β) signaling pathway-related genes are associated with multiple tumors, but their role in prognosis prediction and tumor microenvironment (TME) regulation in colon cancer is poorly understood. Using bioinformatics, this study aimed to construct a risk prediction signature for colon cancer, which may provide a means for developing new effective treatment strategies. METHODS: Using consensus clustering, patients in The Cancer Genome Atlas (TCGA) with colon adenocarcinoma were classified into several subtypes based on the expression of TGF-β signaling pathway-related genes, and differences in survival, molecular, and immunological TME characteristics and drug sensitivity were examined in each subtype. Ten genes that make up a TGF-β-related predictive signature were found by least absolute shrinkage and selector operation (LASSO) regression using colon cancer data from the TCGA database and confirmed using a Gene Expression Omnibus (GEO) dataset. A nomogram incorporating risk scores and clinicopathologic factors was developed to stratify the prognosis of patients with colon cancer for accurate clinical diagnosis and therapy. RESULTS: Two TGF-β subtypes were identified, with the TGF-β-high subtype being associated with a poorer prognosis and superior sensitivity to immunotherapy. Mutation analyses showed a high incidence of gene mutations in the TGF-β-high subtype. After completing signature construction, patients with colon cancer were categorized into high- and low-risk subgroups based on the median risk score of the TGF-β-related predictive signature. The risk score exhibited superior predictive performance relative to age, gender, and stage, as evidenced by its AUC of 0.686. Patients in the high-risk subgroup had higher levels of immunosuppressive cell infiltration and immune checkpoints in the TME, suggesting that these patients had better responses to immunotherapy. CONCLUSIONS: Patients with colon cancer were divided into two subtypes with different survival and immune characteristics using consensus clustering analysis based on TGF-β signaling pathway-related genes. The constructed risk prediction signature may show promise as a biomarker for evaluating the prognosis of colon cancer, with potential utility for screening individuals for immunotherapy.