Abstract

Background: The existing hepatocellular carcinoma (HCC) risk scores have modest accuracy and most are specific to chronic hepatitis B. In this study, we developed and validated a liver stiffness-based machine learning algorithm (ML) for prediction and risk stratification of HCC in various chronic liver diseases (CLDs).

Methods: MLs were trained for prediction of HCC in 5155 adult patients with various CLDs in Korea and further tested in two prospective cohorts from Hong Kong (HK, N=2732) and Europe (N=2384). Model performance was assessed according to Harrell’s C-index and time-dependent receiver operating characteristic (ROC) curve.

Results: We developed the LSM-XGB HCC score, a liver stiffness-based ML HCC risk score, with liver stiffness measurement ranked as the most important among 9 clinical features (Figure). The Harrell’s C-index of the LSM-XGB HCC score in HK and Europe validation cohorts were 0.89 (95% confidence interval [CI] 0.85-0.92) and 0.91 (95%CI 0.87-0.95), respectively. The area under ROC curves of the LSM-XGB HCC score for HCC in 5 years were ≥0.89 in both validation cohorts. The performance of LSM-XGB HCC score was significantly better than existing HCC risk scores including aMAP score, Toronto HCC risk index, and seven hepatitis B related risk scores. Using a cutoff of 0.045, 82.7% and 89.0% of patients in HK and Europe validation cohorts were classified as low-risk for a possible exemption from HCC surveillance, respectively; the annual HCC incidence for low-risk group was 0.10%-0.19%. The high-risk group had an annual HCC incidence of 1.91% and 2.63% in the HK and Europe validation cohorts, respectively.

Conclusion: The LSM-XGB HCC score is a useful machine learning-based tool for clinicians to stratify HCC risk in patients with CLDs.