Dynamic evaluation of hepatocellular carcinoma prediction models in patients with chronic hepatitis B receiving nucleotide/nucleoside analogue treatment.

Carcinogenesis risk scores for chronic hepatitis B have been proposed, but it remains unclear whether these scores during nucleoside/nucleotide analogue (NA) therapy are useful for risk assessment. In this study, we examined changes of these scores and the predictability during NA treatment. 432 patients with no history of hepatocellular carcinoma (HCC) treated with NA were enrolled. PAGE-B, modified PAGE-B (mPAGE-B), and REACH-B scores were calculated at NA administration, 1 and 2 years after administration. The median follow-up duration was 5.1 years, during which 37 patients (8.6%) developed HCC. Cumulative incidence HCC development in patients with high risk of PAGE at NA administration, and 1 and 2 years after NA administration was significantly higher than those with intermediate and low-risk groups (p < .05 for all time points), whereas HCC incidence in patients with high risk of mPAGE-B and REACH-B at 2 years after NA administration were equivalent to those with intermediate and low-risk groups (p = .2 for mPAGE-B, and p = .1 for REACH-B). The area under the receiver operating characteristic (AUROC) for HCC development of PAGE-B at NA administration, and 1 and 2 years after administration were 0.773, 0.803 and 0.737, respectively. The AUROCs of PAGE-B at each point were continuously higher than those of REACH-B (0.646, 0.725, and 0.653, respectively) and mPAGE-B (0.754, 0.734, and 0.678, respectively).PAGE-B score has a high diagnostic accuracy for HCC development at any time point during NA treatment, indicating its potential use as a real-time monitor of HCC development.