Abstract
INTRODUCTION: Hepatitis B surface antigen (HBsAg) clearance leads to favorable outcomes in patients with chronic hepatitis B. HBsAg levels <200 IU/mL with HBsAg decline >0.5 log 10 IU/mL in 1 year have been reportedly predictive of HBsAg loss. This study aimed to use the REVEAL-hepatitis B virus cohort to validate and simplify this prediction rule and verify whether the simplified algorithm can be used among various clinical subgroups. METHOD: We analyzed 707 patients with untreated chronic hepatitis B who had 3 or more HBsAg measurements within 5 years before HBsAg seroclearance or last visit, greater than 1 year apart from one another. Rapid HBsAg decline was defined as HBsAg decline >0.5 log 10 IU/mL in 1 year or >1 log 10 IU/mL in 2 years. Sensitivity, specificity, positive predictive values, and negative predictive values were compared to assess the predictability of HBsAg seroclearance. RESULTS: During a median follow-up of 10.7 years, 41 of the 707 patients cleared serum HBsAg. HBsAg levels at all measurements were lower ( P < 0.0001) and HBsAg decline was greater ( P < 0.0001) in patients with seroclearance compared with non-seroclearance patients. The predictive accuracy of predicting 1-year HBsAg loss using only the rapid decline algorithm (sensitivity = 0.4412, specificity = 0.9792, positive predictive value = 0.5172, negative predictive value = 0.972) was the same as the model combining rapid HBsAg decline and HBsAg levels <200 IU/mL. The simplified algorithm including only the rapid decline performed similarly among various levels of HBsAg, hepatitis B virus DNA, and alanine aminotransferase and was independent of inactive carrier state. DISCUSSION: HBsAg decline >0.5 log 10 IU/mL/yr was a practical predictor of HBsAg seroclearance within 1 year in our community-based untreated cohort.