Identification of a 5-Hydroxymethylation Signature in Circulating Cell-Free DNA for the Noninvasive Detection of Colorectal Cancer

Background. As a crucial epigenetic modification, DNA 5-hydroxymethylcytosine (5-hmC) plays a key role during colorectal cancer (CRC) carcinogenesis. Nevertheless, the levels of 5-hmC-related genes in the circulating DNA of CRC remain largely unknown. Methods and Results. The GSE81314 dataset from the Gene Expression Omnibus (GEO), which was generated by chemical marking-based low-input shotgun sequencing to detect 5-hmC in circulating cell-free DNA (cfDNA) was used in the present study. The GSE81314 dataset includes data for 8 plasma samples from healthy individuals and 4 plasma samples from CRC patients. The difference in the 5-hmC levels in cfDNA between the CRC group and healthy individuals was analyzed by the differentially expressed genes (DEG) package. Weighted gene coexpression network analysis (WGCNA) was conducted to analyze gene coexpression modules associated with sample characteristics. DEG analysis identified 19 upregulated and 9 downregulated 5-hmC-related genes. WGCNA showed that the pink, purple, and brown modules, which contain 531 genes in total, were significantly correlated with CRC (0.66, 0.61, and -0.59, respectively). We used gene set enrichment analysis (GSEA) software to compare 5-hmC-related genes and pathways between CRC patients and healthy controls. We further performed a protein–protein interaction (PPI) analysis and identified 4 nodes (LCN2, LRG1, S100P, and TACSTD2) that played key roles in the network, and we analyzed the expression of these nodes S100P in the GEPIA database. Consistent with the 5-hmC levels in CRC patient plasma, our external validation results from the GEPIA and UALCAN databases showed that LCN2, LRG1, S100P, and TACSTD2 were highly expressed in CRC tissue compared with controls. The DNA promoter methylation levels of LCN2, LRG1, and S100P were lower in CRC tissue than in normal control tissue. Conclusion. The present findings suggest that abnormality in cell-free DNA hydroxylation in plasma may be associated with CRC. In addition, the 5-hmC levels of LCN2, LRG1, S100P, and TACSTD2 in circulating cfDNA may be used as potential noninvasive markers for CRC.