SummaryBackgroundBiomarkers such as quantitative HBsAg (qHBsAg), quantitative hepatitis B virus (HBV) core‐related antigen (qHBcrAg) and HBV RNA may be useful in predicting HBsAg loss in patients with chronic hepatitis B (CHB) undergoing antiviral therapy.Aim(s)Our study evaluated qHBsAg, HBV RNA and qHBcrAg as a posthoc analysis of a randomized clinical trial of peginterferon±NA to determine their utility in predicting HBsAg loss.MethodsCHB patients who completed therapy with 48weeks peginterferon alpha2b ± nucleoside analogue therapy (clinicaltrial.gov NCT01928511) were evaluated at week 72 for HBsAg loss. The predictive ability of qHBsAg, qHBcrAg, HBV RNA and other variables were investigated by univariate and multivariate logistic models for HBeAg‐negative patients by odds ratios, area under the curve (AUC), sensitivity, specificity, and positive and negative likelihood ratios (LR).ResultsHBsAg loss occurred in 15/114(13%) HBeAg‐negative CHB patients who completed 48 weeks of peginterferon. At baseline, qHBsAg was superior to HBcrAg and HBV RNA with AUC 0.916, 0.649 and 0.542, respectively. Using multivariate analysis, the model comprising treatmentarm, age, gender, baseline qHBsAg, HBcrAg and HBV RNA, weeks 4 & 8 qHBsAg had the highest AUC(0.98), but the univariate model with week 8 qHBsAg <70 IU/mL had AUC 0.96. Hence, the contributions of variables other than qHBsAg were marginal. HBV RNA and qHBcrAg were weak predictors of HBsAg loss. Kinetics of the novel markers showed only qHBsAg had a good relationship with HBsAg loss while HBV RNA had a marginal relationship and HBcrAg did not change at all, and none had a good relationship with viral rebound.ConclusionsOn‐treatment biomarker predictors were better than baseline ones, and the best predictor of HBsAg loss at 72 weeks was week 8 qHBsAg <70 IU/mL.