Background and aims: Several predictive scores have been developed internationally to rationalise hepatocellular carcinoma screening in patients with chronic hepatitis B. This study evaluated the performance of these scores in a large Australian cohort.
Methods: A retrospective analysis was undertaken on patients with chronic hepatitis B attending two tertiary hospitals from 1st January 2017 to 30th June 2023. Predictive scores were calculated at baseline and patients were followed for the development of hepatocellular carcinoma.
Results: A total of 1080 patients were included. The average age was 48 years, 43% were female, 91% were non-cirrhotic, and 51% received antivirals. Most patients (62%) were of Asian ethnicity, 13% African, and 9% Caucasian. Twenty-one patients (1.9%) developed hepatocellular carcinoma during a median follow-up of 5.2 years. The calculated sensitivity was 90.5% for PAGE-B and modified PAGE-B, and 42.9% for REACH-B. Positive predictive values were below 4%, and negative predictive values exceeded 98% for all scores. The time-dependent area under the curve at 5 years for PAGE-B, modified PAGE-B, and REACH-B were 0.74, 0.80, and 0.68 respectively. The 5-year cumulative incidence of hepatocellular carcinoma for patients with low, intermediate, and high PAGE-B scores was 0.33%, 2.1% and 4.9% respectively (log-rank, p < 0.001) and for modified PAGE-B was 0.44%, 1.0% and 7.1% respectively (p < 0.001). REACH-B was unable to accurately stratify hepatocellular carcinoma risk (p = 0.19).
Conclusions: In our Australian cohort of diverse ethnicity, cirrhosis and treatment status, both PAGE-B and modified PAGE-B scores performed well with respect to hepatocellular carcinoma risk stratification and identifying low-risk patients who may safely avoid screening.
Keywords: early detection of cancer; hepatitis B; hepatocellular carcinoma.
© 2025 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd.