Improving the Prediction of Relapse After Nucleos(t)ide Analogue Discontinuation in Patients With Chronic Hepatitis B

Background. Current guidelines recommend rules for stopping nucleos(t)ide analogues (NAs) in patients with chronic hepatitis B (CHB), but off-therapy relapse is still high. This study aimed to identify predictors of off-therapy relapse and improve existing stopping rules. Methods. This retrospective study included 488 patients with CHB (262 hepatitis B e antigen [HBeAg]-positive and 226 HBeAg-negative) who discontinued NAs. Posttreatment relapse was investigated. Results. During the median follow-up period of 73.3 months, the cumulative 5-year and 10-year virologic relapse (VR) rates were 73.5% and 76.1%, respectively. In HBeAg- positive patients, end-of-therapy hepatitis B surface antigen (HBsAg) levels (hazard ratio [HR], 1.93 [95% confidence interval {CI}, 1.42-2.61]) and consolidation duration >= 2 years (HR, 0.31 [95% CI:.17-. 58]) were independent predictors of VR. Consolidation >= 2 years and low HBsAg levels (<= 560 IU/mL) significantly lowered VR rates. In HBeAg-negative patients, only the HBsAg level (HR, 1.61 [ 95% CI: 1.24-2.11]) was independently predictive of VR. Cirrhosis was significantly associated with higher VR rates in HBeAg-negative patients with low HBsAg levels (<= 800 IU/mL). Combining end- of-therapy HBsAg levels with current stopping rules or consolidation duration further reduced off-therapy relapse, with 2-year VR rates of approximately 15%-25% in HBeAg-positive patients and 35% in HBeAg-negative patients. Conclusions. End-of-therapy HBsAg levels, consolidation duration, and cirrhosis are key determinants of off-therapy relapse. Together with low HBsAg levels, extended consolidation therapy for >= 2 years should be ensured, and cirrhotic patients should continue NAs even if low HBsAg levels are achieved. A combination of these parameters will help identify individuals at low risk of relapse and significantly improve the predictive ability of the existing stopping rules.