Yoon E Shin, Jae Young Kim, Jeong Ju Yoo, Sang Gyune Kim, Young Seok Kim
Adherence to antiviral therapy is essential for viral suppression in chronic hepatitis B (CHB), yet its effect on hepatocellular carcinoma (HCC) remains uncertain. We examined the association between nucleos(t)ide analog (NA) adherence and HCC risk, focusing on tenofovir alafenamide (TAF) versus entecavir (ETV). Treatment-naïve adults with CHB initiating TAF or ETV were identified from the Health Insurance Review and Assessment Service, Republic of Korea. Adherence was measured using the medication possession ratio, with high adherence defined as ≥ 90%, and the primary outcome was HCC incidence. A total of 28 448 patients were included, of whom 19 001 (66.8%) showed high adherence and 9447 (33.2%) showed low adherence. In the overall cohort, HCC incidence was comparable between the high- and low-adherence groups (14.89 vs. 13.76 per 1000 person-years; IRR, 1.08; 95% CI, 0.95-1.24). Among patients with claims-defined cirrhosis, however, high adherence was associated with lower observed HCC incidence (32.92 vs. 49.59 per 1000 person-years; IRR, 0.66; 95% CI, 0.57-0.78), and this association remained significant in multivariable Cox regression (HR, 0.77; 95% CI, 0.66-0.90) as well as in IPTW, landmark, and sensitivity analyses. In exploratory regimen-stratified analyses, high-adherence TAF users showed a lower observed cumulative HCC incidence than high-adherence ETV users, whereas no significant difference was observed in the low-adherence group. Logistic regression identified ETV use as a predictor of low adherence (OR, 2.80; 95% CI, 2.65-2.96). High adherence to antiviral therapy was associated with lower observed HCC incidence among patients with CHB and claims-defined cirrhosis. Although this association remained evident across multiple analyses, residual confounding, treatment-selection effects, time-related bias, and competing-risk issues cannot be fully excluded.