Four methylation-driven genes detected by linear discriminant analysis model from early-stage colorectal cancer and their methylation levels in cell-free DNA

Abstract

The process of colorectal cancer (CRC) formation is considered a typical model of multistage carcinogenesis in which aberrant DNA methylation plays an important role. In this study, 752 methylation-driven genes (MDGs) were identified by the MethylMix package based on methylation and gene expression data of CRC in The Cancer Genome Atlas (TCGA). Iterative recursive feature elimination (iRFE) based on linear discriminant analysis (LDA) was used to determine the minimum MDGs (iRFE MDGs), which could distinguish between cancer and cancer-adjacent tissues. Further analysis indicated that the changes in methylation levels of the four iRFE MDGs, ADHFE1-Cluster1, CNRIP1-Cluster1, MAFB, and TNS4, occurred in adenoma tissues, while changes did not occur until stage IV in cell-free DNA. Furthermore, the methylation levels of iRFE MDGs were correlated with the genes involved in the reprogramming process of somatic cells to pluripotent stem cells, which is considered the common signature of cancer cells and embryonic stem cells. The above results indicated that the four iRFE MDGs may play roles in the early stage of colorectal carcinogenesis and highlighted the complicated relationship between tissue DNA and cell-free DNA (cfDNA).

Keywords: DNA Methylation; carcinogenesis; cell-free DNA - cfDNA; colorectal cancer; linear discriminant analysis.