The prediction of hepatocellular carcinoma (HCC) development during nucleotide/nucleoside analog (NA) therapy is clinically important in patients with chronic hepatitis B. Although several useful models for HCC prediction have been previously reported, their usefulness in the Japanese population is unclear. Therefore, this study examines the applicability of these models in Japanese patients. Four hundred forty-three patients with no history of HCC who were treated with entecavir, tenofovir alafenamide, or tenofovir disoproxil fumarate were enrolled. PAGE-B, modified-PAGE-B, and REACH-B scores were calculated, and subsequent HCC development was investigated. The mean follow-up duration was 5.1years, and a total of 33 patients (7.4%) developed HCC during the follow-up period. Multivariate analysis revealed that old age (hazard ratio [HR] 1.05, 95% confidence interval [CI] 1.01-1.09, P=0.011), male gender (HR 2.62, 95% CI 1.06-6.49, P=0.037), and low platelet count (HR 0.83, 95% CI 0.77-0.91, P<0.001) were independent predictors of HCC development. These factors are the same as the factors identified in the PAGE-B model. Time-dependent area under the receiver operating characteristic (AUROC) curve revealed that the AUROCs for 3 and 7years of PAGE-B (AUROC: 0.786 and 0.744 at 3 and 7years, respectively) were continuously higher than those of REACH-B (0.658 and 0.543) and modified PAGE-B AUROC (0.772 and 0.731). PAGE-B, which can easily identify high-risk cases, can be useful for predicting HCC development in Japanese patients treated with NA therapy.
Read full abstract