New Model Sharply Improves Liver Cancer Risk Prediction in Hepatitis B Patients

Healthcare | 2025-10-29 17:24:48
[medi K / HEALTH IN NEWS] A research team led by Seung Up Kim, a professor of gastroenterology at Severance Hospital in Seoul, and Hye Won Lee, a professor at Yongin Severance Hospital, has developed a predictive model that combines an established scoring system with a noninvasive measure of liver stiffness, achieving 82 percent accuracy in forecasting hepatocellular carcinoma risk among patients with chronic hepatitis B.

The study, published in Clinical Gastroenterology and Hepatology, involved five tertiary medical centers, including institutions in South Korea and the Chinese University of Hong Kong. Investigators evaluated two models: aMLaf, which defines advanced fibrosis, and aMLc, which uses cirrhosis as the threshold. Both integrate the aMAP score—based on age, sex, albumin, bilirubin, and platelet count—with liver stiffness measurements obtained via vibration-controlled transient elastography (VCTE).

From May 2005 to July 2021, the team followed 944 antiviral-treated chronic hepatitis B patients for an average of more than five years. The new models yielded an area under the receiver operating characteristic curve (AUROC) of 0.82, significantly outperforming the PAGE-B (AUROC 0.74) and modified PAGE-B (AUROC 0.75) scores. In the aMLaf low-risk group, no cases of liver cancer occurred during follow-up.

(From left) Seung Up Kim, professor of gastroenterology at Severance Hospital, and Hye Won Lee, professor at Yongin Severance Hospital (Photo provided by Severance Hospital)
(From left) Seung Up Kim, professor of gastroenterology at Severance Hospital, and Hye Won Lee, professor at Yongin Severance Hospital (Photo provided by Severance Hospital)


Validation in a separate cohort of 61 hepatitis B patients from the Chinese University of Hong Kong confirmed predictive accuracy exceeding 80 percent.
Chronic hepatitis B remains a leading cause of liver cancer in South Korea, where vertical transmission predominates and residual fibrosis persists as a key oncogenic driver even after viral suppression. Traditional fibrosis assessment relied on biopsy, limited by cost, complications, and interobserver variability. VCTE has emerged as a reliable alternative.

“This integration of aMAP with liver stiffness enables precise risk stratification,” Dr. Kim said. Dr. Lee added that the models clearly separate low- and high-risk groups, reducing unnecessary surveillance in the former while guiding intensified monitoring in the latter.

Kim Kuk Ju / press@themedik.kr
Copyright © 2020 mediK. All rights reserved.