Identification of Methylation Driven Biomarkers for Diagnosis and Prognosis in Colon Cancer by Integrative Analysis of TCGA, GTEx, and GEO Data

Abstract Background: The intention of the present work was to investigate methylation driven biomarkers for diagnosis and prognosis in colorectal cancer (CRC) by integrative analysis of DNA methylation and gene expression data from The Cancer Genome Atlas (TCGA), The Genotype-Tissue Expression (GTEx), and Gene Expression Omnibus (GEO). Methods and Results: The differentially expression genes (DEGs) and differentially methylated genes (DMGs) were screened using mRNA expression and DNA methylation data from TCGA, respectively. And the methylation driven genes (MDGs) of CRC were further identified using MethylMix R package. Subsequently, the MDGs were underwent Random Forest (RF) analyses to establish diagnosis prediction model using mRNA expression data from TCGA and GTEx, which were then validated by GSE39582 from GEO dataset. In addition, prognostic biomarkers that were used to establish the methylation related risk score model was generated by the univariate and multivariate Cox regression analysis. 9 out of 10 DMGs revealed excellent performance as independent diagnostic predictors, including CLDN1, EPHX4, TCN1, ARHGAP20, LY6G6D, FAM150A, KCNJ12, KRT7 and STK33. Furthermore, STK33 and EPHX4 were found to be associated with Overall survival (OS). Conclusions: Our findings suggest that the identified MDGs could be potential biomarkers for diagnosis and prognosis of CRC.