Abstract
BACKGROUND: By the time they are clinically diagnosed, patients with hepatocellular carcinoma (HCC) are often at the advanced stage. DNA methylation has become a useful predictor of prognosis for cancer patients. Research on DNA methylation as a biomarker for assessing the risk of occurrence in HCC patients is limited. The purpose of this study was to develop an efficient methylation site model for predicting survival in patients with HCC. METHODS: DNA methylation and gene expression profile data were extracted from The Cancer Genome Atlas (TCGA) database. Markers of DNA-methylated site in two subsets (the training subset and the test subset) were identified using a random survival forest algorithm and Cox proportional hazards regression. Then, Gene Ontology annotations were applied to investigate the functions of DNA methylation signatures. RESULTS: A total of 37 hub genes containing 713 methylated sites were identified among the differentially methylated genes (DMGs) and differentially expressed genes (DEGs). Finally, seven methylation sites (cg12824782, cg24871714, cg18683774, cg22796509, cg19450025, cg10474350, and cg06511917) were identified. In the training group and the test group, the area under the curve predicting the survival of patients with HCC was 0.750 and 0.742, respectively. The seven methylation sites signature could be used to divide the patients in the training group into high- and low-risk subgroups [overall survival (OS): 2.81 vs. 2.11 years; log-rank test, P<0.05]. Then, the prediction ability of the model was validated in the test dataset through risk stratification (OS: 2.04 vs. 2.88 years; log-rank test, P<0.05). Functional analysis demonstrated that these signature genes were related to the activity of DNA-binding transcription activator, RNA polymerase II distal enhancer sequence-specific DNA binding, and enhancer sequence-specific DNA binding. CONCLUSIONS: The results of this study showed that the signature is useful for predicting the survival of HCC patients and thus, can facilitate treatment-related decision-making.