Abstract
OBJECTIVE: Hepatocellular carcinoma (HCC) is a common malignancy associated with high morbidity and mortality rates worldwide. Early diagnosis plays an important role in the improvement of HCC prognosis. METHODS: In this study, we conducted a comprehensive analysis of HCC DNA methylation and gene expression datasets in The Cancer Genome Atlas (TCGA), to identify a prognostic signature for HCC diagnosis and survival prediction. First, we identified differential methylation CpG (dmCpG) sites in HCC samples and compared them with those in adjacent normal liver tissues; this was followed by univariate analysis and Sure Independence Screening (SIS) in the training set. The robustness of the identified prognostic signature was evaluated using the testing set. To explore the biological processes involved in HCC progression, we also performed functional enrichment analysis for overlapping genes between genes containing dmCpG sites (DMGs) and differential expression genes (DEGs) in HCC patients, using data from the Database for Annotation, Visualization, and Integrated Discovery (DAVID). RESULTS: As a result, we identified five CpG sites that were significantly associated with HCC survival through univariate analysis and SIS. Univariate analysis of clinical characteristics identified age and risk factors (including alcohol consumption and smoking) as independent factors that indicated HCC survival. Multivariate analysis indicated that the integrated prognostic signature (weighted combination of the five CpG sites) that took age and risk factors into consideration resulted in more accurate survival prediction. CONCLUSIONS: This study provides a novel signature for predicting HCC survival, and should be helpful for early HCC diagnosis and personalized treatment.