The identification of a potential prognostic model of colon cancer using integrated bioinformatics analysis

Abstract Background This study aimed to construct prognostic model by screening prognostic gene signature of colon cancer. Methods The gene expression profile data of colon cancer were obtained from The Cancer Genome Atlas (TCGA) and gene expression omnibus (GEO) and differently expressed genes (DEGs) between tumor and control samples were identified. Prognosis-associated genes were then identified and used for the construction of prognostic model. The independent factors that associated with the prognosis of colon in the TCGA cohort was identified. Results Totally, 1153 consistent DEGs were screened out between tumor and normal tissues in the TCGA cohort, GSE44861 and GSE44076 datasets. Among these genes, 12 DEGs were related to the prognosis of colon cancer and were used for constructing the prognostic model. This model presented a high predictive power for the prognosis of colon cancer both in the training dataset and in the validation datasets (AUC > 0.8). Statistical analysis showed that age, pathological T, tumor recurrence, and model status were the independent factors for prognosis of patients with colon cancer in TCGA. Conclusions The 12-gene signature prognostic model had a high predictive power for colon cancer prognosis.