| 10BackgroundColon carcinoma is the second leading cause of death in the world,and the new incidence rate ranks third among all cancers.Abnormal DNA methylation is related to the occurrence,development and prognosis of tumors,but the mechanism of action still remains to be explored.In this study,we aimed to identify genes associated with abnormal methylation in Colon carcinoma to provide a theoretical basis for the diagnosis and treatment of colon cancer.MethodsColon carcinoma transcriptome data,methylation data and clinical information were download from the TCGA database and GEO database.The differentially expressed genes(DEGs)and methylated genes(DMGs)we analyzed and identified in Colon carcinoma.PCA analysis was applied to divide Colon carcinoma into subtypes,and the survival and immune cell infiltration of each subtype were evaluated.Cox and LASSO analysis were performed to construct Colon carcinoma risk model.GSEA was used to evaluate the enrichment pathways.The Kaplan–Meier was used to analyze the difference in survival.ROC curve was plotted to evaluate the accuracy of the model,and GSE17536 was used to verify the accuracy of the risk model.The risk model is combined with the clinicopathological characteristics of Colon carcinoma patients to perform multivariate Cox regression analysis to obtain independent risk factors and draw nomograms.ResultsWe screened a total of 4564 DEGs and 1093 DMGs,of which 298 genes were found to have intersecting sets.Methylation in 220 of these genes was significantly negatively correlated with expression levels.Based on the Lasso model,we finally selected 4 methylated genes for risk scoring for prognostic risk stratification.Therefore,we divided the training set samples into risk groups according to the median risk score,including high-risk scoring group(high-risk group)and low-risk scoring group(low-risk group).By analyzing the correlation between clinical features and risk scores in the high-and low-risk groups,we found that the severity of clinical traits was positively correlated with risk scores.In addition,we analyzed GSE17536(validation set)by using Kaplan-Meier and also showed significant differences in Disease Specific Survival(DSS)and Overall Survival(OS)in both high-and low-risk groups.Therefore,we obtained definite results through internal and external validation,that is,the prognosis of patients in the low-risk group was significantly better than that of patients in the high-risk group.Multivariate regression analysis showed that the above risk subgroup was an independent factor affecting the prognosis of patients,and based on the results of multivariate regression analysis,we included common clinicopathological features to construct a nomogram model to predict the survival probability of patients.In the training group of the predictive model,a robust prediction performance(AUC under the ROC curve: 0.8)was obtained.Based on calibration curves for 1-,3-,and 5-year colon cancer patients,we found that the survival rates predicted by this model were very close to the actual survival rates.Later,we further verified the high accuracy of this model in predicting the survival rate of colon cancer patients through ROC curves,with AUCs of 0.778,0.81 and 0.836 at 1-year,3-year and 5-year AUCs,respectively.ConclusionOur study identified 4 methylated biomarkers in the Colon carcinoma.Then we constructed the risk model to provide a theoretical basis and reference value for the research and treatment of Colon carcinoma. |
PDF Full Text Request