Addition of liver stiffness enhances the predictive accuracy of the PAGE-B model for hepatitis B-related hepatocellular carcinoma

Abstract

Background: The modified PAGE-B (mPAGE-B) and PAGE-B models reliably predict the risk of developing chronic hepatitis B (CHB)-related hepatocellular carcinoma (HCC).

Aim(s): To investigate whether the addition of liver stiffness (LS) value, assessed using transient elastography, enhanced the predictive accuracies of these models

Methods: Patients with CHB who started anti-viral therapy (AVT) between 2007 and 2017 were enrolled. The training (Yonsei University Hospital) and validation (seven Korean referral institutes) cohorts contained 1211 and 973 patients, respectively.

Results: Based on multivariate analysis, older age (hazard ratio [HR] = 1.051, 95% confidence interval [CI] = 1.031-1.071), male sex (HR = 2.265, 95% CI = 1.463-3.506), lower platelet count (HR = 0.993, 95% CI = 0.989-0.997) and greater LS values (HR = 1.015, 95% CI = 1.002-1.028) were independently associated with an increased risk of HCC development (all P < 0.05). Thus, we developed a modified PAGELS-B model (maximum score 34) that included age, male sex, platelet count and LS value. The integrated area under the curve of the modified PAGELS model was greater than those of the PAGE-B and mPAGE-B models (0.760 vs 0.714 and 0.716, respectively) in the derivation dataset. The cumulative HCC incidence was significantly higher in the high-risk (modified PAGE-BLS score ≥ 24) group than in the intermediate-risk (modified PAGELS-B score 12-24) or low-risk (modified PAGELS-B score < 12) group (all P < 0.001). Similar results were observed in the validation cohort.

Conclusions: The predictive accuracies of the PAGE-B and mPAGE-B models were validated in Korean patients with CHB receiving AVT. However, the modified PAGELS-B model featuring the addition of LS value showed higher predictability than the PAGE-B and mPAGE-B models.