Probability of HBsAg loss after nucleo(s)tide analogue withdrawal depends on HBV genotype and viral antigen levels

Abstract

Background & Aims Nucleo(s)tide analogue (NUC) withdrawal may result in HBsAg clearance in a subset of patients. However, predictors of HBsAg loss after NUC withdrawal remain ill-defined. Methods We studied predictors of HBsAg loss in a global cohort of HBeAg-negative patients with undetectable HBV DNA who discontinued long-term NUC therapy. Patients requiring retreatment after treatment cessation were considered non-responders. Results We enrolled 1,216 patients (991 with genotype data); 98 (8.1%) achieved HBsAg loss. The probability of HBsAg loss was higher in non-Asian patients (adjusted hazard ratio [aHR] 8.26, p <0.001), and in patients with lower HBsAg (aHR 0.243, p <0.001) and HBV core-related antigen (HBcrAg) (aHR 0.718, p = 0.001) levels. Combining HBsAg (<10, 10-100 or >100 IU/ml) and HBcrAg (<2log vs. ≥2 log) levels improved prediction of HBsAg loss, with extremely low rates observed in patients with HBsAg >100 IU/ml with detectable HBcrAg. HBsAg loss rates also varied with HBV genotype; the highest rates were observed for genotypes A and D, and none of the patients with HBV genotype E experienced HBsAg loss (p <0.001 for the overall comparison across genotypes; p <0.001 for genotypes A/D vs. genotypes B/C). HBV genotype C was independently associated with a higher probability of HBsAg loss when compared to genotype B among Asian patients (aHR 2.494; 95% CI 1.490–4.174, p = 0.001). Conclusions The probability of HBsAg loss after NUC cessation varies according to patient ethnicity, HBV genotype and end-of-treatment viral antigen levels. Patients with low HBsAg (<100 IU/ml) and/or undetectable HBcrAg levels, particularly if non-Asian or infected with HBV genotype C, appear to be the best candidates for treatment withdrawal.