A methylation-based prognostic model predicts survival in patients with colorectal cancer

Background: To construct a model that could effectively predict the prognosis of colorectal cancer (CRC) by searching for methylated-differentially expressed genes (MDEGs). Methods: We identified MDEGs through four databases from Gene Expression Omnibus (GEO) and annotated their functions via bioinformatics analysis. Subsequently, after adjusting for gender, age, and grading, multivariate Cox hazard analysis was utilized to select MDEGs interrelated with the prognosis of CRC, and LASSO analysis was utilized to fit the prediction model in the training set. Furthermore, another independent dataset was harnessed to verify the effectiveness of the model in predicting prognosis. Results: In total, 252 hypomethylated and up-regulated genes and 132 hypermethylated and downregulated genes were identified, 27 of which were correlated with the prognosis of CRC, and a 10-gene prognostic model was established after LASSO analysis. The overall survival rate could be effectively grouped into different risks by the median score of this model in the training set [risk ratio (HR) =2.27, confidence interval (95% CI), 1.69-3.13, P=8.15x10-8], and the validity of its effect in predicting prognosis in CRC was verified in the validation dataset (HR =1.75, 95% CI, 1.15-2.70, P=9.32x10-3). Conclusions: Our model could effectively predict the overall survival rate of patients with CRC and provides potential application guidelines for its clinically personalized treatment.