Gongping Sun1, He Duan1, Yuanhao Xing2, Dewei Zhang1. 1. Department of General Surgery, The Fourth Affiliated Hospital of the China Medical University, Shenyang, 110032, People's Republic of China. 2. China Medical University, Shenyang, 110000, People's Republic of China.
Abstract
PURPOSE: We aimed to screen novel genetic biomarkers for use in a prognostic score (PS) model for the accurate prediction of survival outcomes for patients with colon adenocarcinoma (COAD). METHODS: Gene expression and methylation data were downloaded from The Cancer Genome Atlas database, and the samples were randomly divided into training and validation sets for the screening of differentially methylated genes (DMGs) and differentially expressed genes (DEGs). Co-methylated genes were screened using weighted gene co-expression network analysis. Functional enrichment analysis was performed using the Database for Annotation, Visualization, and Integrated Discovery. Univariate and multivariate Cox regression analyses were performed to identify prognosis-related genes and clinical factors. Receiver operating characteristic curve analysis was carried out to evaluate the predictive performance of the PS model. RESULTS: In total, 1434 DEGs and 1038 DMGs were screened in the training set, among which 284 were found to be overlapping genes. For 127 of these overlapping genes, the methylation and expression levels were significantly negatively correlated. An optimal signature from 10 DMGs was identified to construct the PS model. Patients with a high PS seemed to have worse outcomes than those with a low PS. Moreover, cancer recurrence and the PS model status were independent prognostic factors. CONCLUSION: This PS model based on an optimal 10-gene signature would help in the stratification of patients with COAD and improve the assessment of their clinical outcomes.
PURPOSE: We aimed to screen novel genetic biomarkers for use in a prognostic score (PS) model for the accurate prediction of survival outcomes for patients with colon adenocarcinoma (COAD). METHODS: Gene expression and methylation data were downloaded from The Cancer Genome Atlas database, and the samples were randomly divided into training and validation sets for the screening of differentially methylated genes (DMGs) and differentially expressed genes (DEGs). Co-methylated genes were screened using weighted gene co-expression network analysis. Functional enrichment analysis was performed using the Database for Annotation, Visualization, and Integrated Discovery. Univariate and multivariate Cox regression analyses were performed to identify prognosis-related genes and clinical factors. Receiver operating characteristic curve analysis was carried out to evaluate the predictive performance of the PS model. RESULTS: In total, 1434 DEGs and 1038 DMGs were screened in the training set, among which 284 were found to be overlapping genes. For 127 of these overlapping genes, the methylation and expression levels were significantly negatively correlated. An optimal signature from 10 DMGs was identified to construct the PS model. Patients with a high PS seemed to have worse outcomes than those with a low PS. Moreover, cancer recurrence and the PS model status were independent prognostic factors. CONCLUSION: This PS model based on an optimal 10-gene signature would help in the stratification of patients with COAD and improve the assessment of their clinical outcomes.