Predictors of Incident High-Risk Varices in Patients with Hepatitis B-related Cirrhosis Receiving Nucleos(t)ide Analogue Therapy

BACKGROUND: In patients receiving nucleos(t)ide analogue (NA) therapy for chronic hepatitis B (CHB)-related cirrhosis, routine endoscopic surveillance may be deferred in the absence of high-risk varices (HRV). We aimed to identify predictors of incident HRV in this population. METHODS: This retrospective cohort study included patients with CHB-related cirrhosis who initiated NA therapy between March 2008 and June 2020 and were followed through May 2024. Eligible patients underwent baseline and on-treatment upper endoscopy. Patients with baseline HRV, prior variceal ligation, use of nonselective beta-blockers, other viral hepatitis, alcohol overuse, or malignancy were excluded. The cumulative incidence of HRV and adjusted subdistribution hazard ratios (aSHRs) were estimated using competing-risk methods. A point-based risk prediction model was derived from multivariable regression coefficients. RESULTS: Among the 110 included patients, 14 (12.7%) developed HRV over a median follow-up of 3.6 years (IQR, 2.0-5.6). In multivariable analysis, FIB-4 ≥4.0 (aSHR 9.77; 95% CI, 1.45-65.7), coexisting metabolic dysfunction-associated steatotic liver disease (MASLD) (aSHR 6.37; 95% CI, 1.84-22.0), and albumin-bilirubin (ALBI) grade >1 (aSHR 3.02; 95% CI, 1.21-7.55) were independently associated with incident HRV and were assigned 3, 2 and 1 point, respectively, in the prediction model. At 3 years, the cumulative incidence of HRV was 1.5% (95% CI, 0-4.5) in the low-risk group (0-3 points) and 21.4% (95% CI, 8.0-34.8) in the high-risk group (4-6 points) (P=0.001). CONCLUSION: MASLD, FIB-4 and ALBI grade predict incident HRV in NA-treated CHB cirrhosis, enabling risk-stratified endoscopic surveillance.