Background: Hepatitis B virus (HBV) variants in the preS region have been associated with hepatocellular carcinoma (HCC). However, the effect of the preS variants on HCC prognosis remains largely unknown. We aimed to identify the preS variants that reliably predict postoperative prognosis in HCC.

Methods: RNA-seq data of 203 HCC patients retrieved from public database were screened for the preS variants related to HCC prognosis. The variants in the sera and tumors were then validated in our prospective cohort enrolling 103 HBV-associated HCC patients.

Results: By analyzing prognosis-related gene sets in the RNA-seq data, 12 HBV preS variants were associated with HCC recurrence. Of those, G40C and C147T in the sera predicted an unfavorable recurrence-free survival in our cohort (hazard ratio [HR]=2.18, 95% confidence interval [CI]=1.37-3.47, p=0.001 for G40C; HR=1.84, 95% CI=1.15-2.92, p=0.012 for C147T). G40C and C147T were significantly associated with microscopic vascular invasion, larger tumor size, and abnormal liver function. Multivariate Cox regression analysis showed that G40C significantly increased the risk of HCC recurrence in patients with postoperative antiviral treatment. The HCC prognosis-prediction model consisting of α-fetoprotein and G40C in the sera achieved the best performance (sensitivity=0.80, specificity=0.70, and area under the curve=0.79). Functional analysis indicated that these two variants were associated with cell proliferation, chromosome instability, carcinogenesis, metastasis, and anticancer drug resistance.

Conclusions: G40C and C147T are serological biomarkers for HCC prognosis and the prognostic model consisting of serological α-fetoprotein and G40C achieved the best performance in predicting postoperative prognosis.