Prediction of long-term HBsAg seroclearance in patients with HBeAg-negative chronic hepatitis B.

Background & Aims: Predicting the long-term HBsAg seroclearance, an ideal endpoint, is relevant for decision-making regarding antiviral therapy for patients with chronic hepatitis B (CHB). This study aimed to identify predictors and develop a prediction model for HBsAg seroclearance in patients with HBeAg-negative CHB.

Methods: A total of 2,032 untreated HBeAg-negative patients who underwent a 2-year baseline observation period were enrolled. Prediction models were developed using independent predictors of seroclearance, and their performance was evaluated through internal and external validation using an independent cohort of 753 patients, along with sensitivity analyses.

Results: The estimated annual incidence of HBsAg seroclearance was 2.22% (15,508 person-years). atitis virus DNA evel (Low-to-intermittently high-level viremia), ld age, male ex, and hepatitis B urface antigen level <250 IU/ml independently predicted seroclearance. Subsequently, two prediction models were developed: HepBLOSS-1 and a simplified version, HepBLOSS-2. These models demonstrated excellent performance in predicting seroclearance at 5, 10, and 15 years, with C-indices and time-dependent area under the receiver operating characteristics curve (AUROC) values of 0.81-0.89. The 10-year cumulative incidence rate in patients with scores of ≥13 in HepBLOSS-1 and those with scores of 8 in HepBLOSS-2 was over 50%. Both models underwent rigorous internal and external validation, demonstrating good predictability with time-dependent AUROCs exceeding 0.80. The predicted seroclearance rate closely aligned with the observed rate in both models.

Conclusions: The HepBLOSS models for HBsAg seroclearance exhibited an outstanding ability to stratify the probability of seroclearance over a 15-year period. These models hold promising potential to guide treatment decisions, aiming to achieve a functional cure in patients with CHB.

Impact And Implications: Achieving a functional cure for chronic hepatitis B, defined as HBsAg seroclearance, is a realistic treatment endpoint, especially given the virus's lifelong persistence in hepatocytes and its significant association with improved prognosis. This study identified independent predictors of seroclearance, including HBV DNA levels, age, sex, and HBsAg levels, and developed a robust prediction model based on these factors. The models demonstrated strong predictive accuracy over a 15-year period, offering valuable guidance for clinicians in establishing treatment strategies and predicting patient prognosis.