Comprehensive analysis reveals a CTHRC1-based fourteen-gene signature for prognosis prediction in colon cancer

Abstract Background: Colon cancer is a malignant gastrointestinal tumor and is still a worldwide leading cause of cancer-related mortality. Owing to the development of high-throughput sequencing technologies, large-scale data have become available in public databases. We devoted to construct and validate a gene signature that can predict patients’ outcomes for colon cancer using bioinformatics methods. Methods: We integrated GSE44076 and GSE39582 to identify differentially expressed genes (DEGs). Timer 2.0 was used to analysis CTHRC1 expression in various types of cancer. The relationship of clinicopathological features and CTHRC1 expression was explored based on TCGA database. Hub genes were identified by String database and Cytoscape software. The gene prognostic signature was established based on the Cox regression analysis.Results: We obtained 170 common DEGs and confirmed CTHRC1 was significantly overexpressed in colon cancer samples (P < 0.001). High CTHRC1 level was associated with advanced stage and shorter overall survival (OS) time and relapse-free survival (RFS) time in colon cancer patients. We identified thirteen hub genes among 516 genes co-expressed with CTHRC1 and subsequently constructed a 14-gene risk signature, which associated advanced stage and poor OS (log-rank test p < 0.01). Multivariate Cox regression indicated that high risk score was an independent risk factor for poor prognosis (p < 0.01). Finally, Receiver operating characteristic curve analysis was performed to further confirm the validity of the signature. Conclusions: We confirmed the important role of CTHRC1 in colon cancer and further constructed a 14-gene risk signature with promising prognostic accuracy for colon cancer patients.