Stopping Nucleos(t)ide Analogues in Chronic Hepatitis B Using HBsAg Thresholds: A Meta-Analysis and Meta-Regression

BACKGROUND AND AIMS: Recommendations for stopping nucleoside analogue (NA) therapy in hepatitis B e antigen- negative chronic hepatitis B (CHB) are unclear. End-of-treatment quantitative hepatitis B serum antigen (EOTqHBsAg) thresholds < 100 IU/mL or < 1000 IU/mL have been proposed as stopping criteria, which we assessed by meta-analysis and meta-regression. METHODS: We searched PubMed, EMBASE, and conference abstracts for studies of hepatitis B e antigen- negative CHB NA discontinuation. Extracted studies were analyzed for risk of bias, pooled risk of hepatitis B serum antigen (HBsAg) loss, virological relapse (VR), and biochemical relapse (BR). Significant heterogeneity (I-2) was addressed by subgroup analysis and random- effects meta-regression with known important covariates, including EOTqHBsAg thresholds, ethnicity, duration of therapy, and follow-up. RESULTS: We found 24 articles (3732 subjects); 16 had low and 8 had moderate risk of bias. The pooled risks of HBsAg loss, VR, and BR for stopping therapy at EOTqHBsAg < 100 IU/mL were 41.8%, 33.4%, and 17.3%, respectively, vs 4.6%, 72.1%, and 34.6%, respectively, for EOTqHBsAg >= 100 IU/mL. The pooled risks of HBsAg loss, VR, and BR for stopping therapy at EOTqHBsAg < 1000 IU/mL were 22.0%, 52.7%, and 15.9%, respectively, vs 3.4%, 63.8%, and 26.4%, respectively, for EOTqHBsAg >= 1000 IU/mL. Multivariable analysis for HBsAg loss showed that ethnicity, follow-up duration, and EOTqHBsAg < 100 IU/mL and >= 100 IU/mL explained 85% of the variance in heterogeneity; Asians with EOTqHBsAg < 100 IU/mL had 28.2%, while non-Asians with EOTqHBsAg < 1000 IU/mL had 38.4% HBsAg loss. Multivariable analysis showed EOTqHBsAg < 100 IU/mL and >= 100 IU/mL and other covariates only explained 43% and 63% of the variance in heterogeneity for VR and BR, respectively, suggesting that other factors are also important for relapse. CONCLUSIONS: While EOTqHBsAg thresholds, ethnicity, and follow-up duration strongly predict HBsAg loss, this is not true for VR and BR, hence stopping NA therapy should be considered cautiously.