Hepatocellular carcinoma (HCC) often occurs following chronic hepatitis B virus (HBV) infection, leading to high recurrence and a low 5-year survival rate. We developed an overall survival (OS) prediction model based on protein expression profiles in HBV-infected nontumor liver tissues. We aimed to demonstrate the feasibility of using protein expression profiles in nontumor liver tissues for survival prediction. A univariate Cox and differential expression analysis were performed to identify candidate prognostic factors. A multivariate Cox analysis was performed to develop the liver gene prognostic index (LGPI). The survival differences between the different risk groups in the training and validation cohorts were also estimated. A total of 363 patients, 159 in the training cohort, and 204 in the validation cohort were included. Of the 6478 proteins extracted from nontumor liver tissues, we identified 1275 proteins altered between HCC and nontumor liver tissues. A total of 1090 out of 6478 proteins were significantly related to OS. The prognostic values of the proteins in nontumor tissues were mostly positively related to those in the tumor tissues. Protective proteins were mainly enriched in the metabolism-related pathways. From the differentially expressed proteins, the top 10 most significant prognosis-related proteins were submitted for LGPI construction. In the training and validation cohorts, this LGPI showed a great ability for distinguishing patients' OS risk stratifications. After adjusting for clinicopathological features, the LGPI was an independent prognostic factor in the training and validation cohorts. We demonstrated the prognostic value of protein expression profiling in nontumor liver tissues. The proposed LGPI was a promising predictive model for estimating OS in HBV-related HCC.
Keywords: biostatistics & bioinformatics; hepatitis B virus; oncogenes; survival analysis; virus classification.
© 2022 Wiley Periodicals LLC.