A risk prediction model for hepatocellular carcinoma after hepatitis B surface antigen seroclearance

After hepatitis B surface antigen (HBsAg) seroclearance, the risk of hepatocellular carcinoma (HCC) remains, and the optimal surveillance strategy has yet to be determined. Herein, we aimed to evaluate incidence and risk factors for HCC and establish a novel prediction model for HCC development after HBsAg seroclearance.

A total of 1,443 patients with chronic hepatitis B who achieved HBsAg seroclearance between 1991 and 2020 were retrospectively screened for study eligibility. The data from 831 of these patients were included in the final analysis. A prediction model was developed based on multivariable Cox models. Harrell's C-index and a time-dependent AUROC were used for discrimination. Bootstrap analysis was performed for internal validation.

Overall, 40 patients (4.8%) developed HCC after HBsAg seroclearance during a follow-up of 4,644 person-years (0.86%/year). Age at HBsAg seroclearance, presence of cirrhosis, family history of HCC, and more-than-moderate alcohol consumption were independently predictive of HCC, and these 4 independent variables were used to develop the prediction model. The C-index of the model was 0.804. The time-dependent AUROCs of the score for HCC prediction at 5, 10, and 15 years were 0.799, 0.835, and 0.817, respectively. The score also showed good discrimination in the internal validation and sensitivity analysis.

The novel prediction model based on age, cirrhosis, family history of HCC, and alcohol consumption enables reliable risk estimation of HCC after HBsAg seroclearance and may serve as a useful reference for decision-making in HCC surveillance for HBsAg-cleared patients.

After spontaneous hepatitis B surface antigen (HBsAg) seroclearance, the risk of hepatocellular carcinoma (HCC) remains. Age at HBsAg seroclearance, presence of cirrhosis, family history of HCC, and more-than-moderate alcohol consumption were independently associated with HCC development after HBsAg seroclearance. The novel prediction model using these 4 variables enables reliable risk estimation of HCC and serves as a useful reference for decision-making in HCC surveillance and management for HBsAg-cleared patients.