Abstract
MicroRNAs as biomarkers play an important role in the tumorigenesis process, including hepatocellular carcinomas (HCCs). In this paper, we used The Cancer Genome Atlas (TCGA) database to mine hepatitis B-related liver cancer microRNAs that could predict survival in patients with hepatitis B-related liver cancer. There were 93 cases of HBV-HCC and 49 cases of adjacent normal controls included in the study. Kaplan-Meier survival analysis of a liver cancer group versus a normal control group of differentially expressed genes identified eight genes with statistical significance. Compared with the normal liver cell line, hepatocellular carcinoma cell lines had high expression of 8 microRNAs, albeit at different levels. A Cox proportional hazards regression model for multivariate analysis showed that four genes had a significant difference. We established classification models to distinguish short survival time and long survival time of liver cancers. Eight genes (mir9-3, mir10b, mir31, mir519c, mir522, mir3660, mir4784, and mir6883) were identified could predict survival in patients with HBV-HCC. There was a significant correlation between mir10b and mir31 and clinical stages (p < 0.05). A random forests model effectively estimated patient survival times.