An artificial intelligence (AI) model has been developed in South Korea to predict the risk of liver function deterioration before treatment for patients with hepatocellular carcinoma. This model is expected to aid in formulating treatment strategies by considering not only liver function but also tumor characteristics.
A research team led by Professor Han Ji-won of the Gastroenterology Department at Catholic University’s Seoul St. Mary’s Hospital analyzed data from 2,026 patients treated for hepatocellular carcinoma at eight hospitals under the Catholic Central Medical Center from 2010 to 2024 to establish a machine learning-based Liver Safety Score (MHSS) model.
The model integrates various factors, including blood test results, liver function indicators, platelet counts, tumor size and number, vascular invasion, and tumor markers, to assist in selecting safe and effective personalized treatment options.
Previously, evaluation metrics primarily focused on liver function, such as the Child-Pugh score, Albumin-Bilirubin (ALBI) score, and the Model for End-Stage Liver Disease (MELD) score for prioritizing liver transplants. However, these metrics did not account for tumor size or vascular invasion, which are critical cancer characteristics.
The research team explained that the MHSS model more accurately predicts the risk of variceal bleeding and liver function deterioration after treatment compared to existing evaluation tools, and its stable performance was confirmed in an independent validation cohort consisting of patients from other institutions.
Patients classified as high-risk in the model showed a 3.25 times higher risk of liver function deterioration, a 4.90 times higher risk of variceal bleeding, and a 2.21 times higher risk of mortality compared to those in the low-risk group.
In a simulation analyzing outcomes based on treatment selection, the low-risk group demonstrated significant survival benefits with the immunotherapy combination of atezolizumab and bevacizumab, while the high-risk group did not show clear survival advantages due to increased risks of variceal bleeding.
Based on these findings, the research team conducted a personalized treatment simulation that indicated a predicted reduction of 24% in the risk of liver function deterioration, 40% in the risk of variceal bleeding, and 26% in overall mortality risk.
Professor Han Ji-won stated, "We have established an objective basis for presenting rational treatment pathways for patients by comprehensively evaluating tumor characteristics, liver function, and portal hypertension risk within a single AI framework. We aim to develop this into a personalized precision medicine tool that can be utilized in actual clinical practice through prospective studies and various data validations."
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.