Construction of a Novel Gene-Based Model for Survival Prediction of Hepatitis B Virus Carriers With HCC Development

Despite the effectiveness of hepatitis B virus (HBV) vaccination in reducing the prevalence of chronic HBV infection as well as the incidence of acute hepatitis B, fulminant hepatitis, liver cirrhosis and hepatocellular carcinoma (HCC), there was still a large crowd of chronically infected populations at risk of developing cirrhosis or HCC. In this study, we established a comprehensive prognostic system covering multiple signatures to elevate the predictive accuracy for overall survival (OS) of hepatitis B virus carriers with HCC development. Weighted Gene Co-Expression Network Analysis (WGCNA), Least Absolute Shrinkage and Selection Operator (LASSO), Support Vector Machine Recursive Feature Elimination (SVM-RFE), and multivariate COX analysis, along with a suite of other online analyses were successfully applied to filtrate a three-gene signature model (TP53, CFL1, and UBA1). Afterward, the gene-based risk score was calculated based on the Cox coefficient of the individual gene, and the prognostic power was assessed by time-dependent receiver operating characteristic (tROC) and Kaplan–Meier (KM) survival analysis. Furthermore, the predictive power of the nomogram, integrated with the risk score and clinical parameters (age at diagnosis and TNM stage), was revealed by the calibration plot and tROC curves, which was verified in the validation set. Taken together, our study may be more effective in guiding the clinical decision-making of personalized treatment for HBV carriers.