TOPLINE

Researchers used nationwide health screening data and developed and validated traditional and machine learning models to predict the risk for early-onset hepatocellular carcinoma (HCC) in young adults. They found that an advanced boosted survival model performed best and that the main risk factors were viral hepatitis, prior non-liver cancer, cirrhosis, and abnormal liver and kidney tests.

METHODOLOGY

• Most existing risk prediction models for HCC focus predominantly on older populations and specific etiologies such as hepatitis B or C virus infection, metabolic dysfunction-associated steatotic liver disease, and alcohol-related liver disease, without specifically addressing the unique characteristics and risk profiles of younger individuals.
• Researchers used nationwide health screening data of 1,756,593 young adults aged 20-39 years who were screened in South Korea during 2013-2014 to develop and validate risk prediction models for early-onset HCC.
• Participants were randomly divided (1:1) into the training (n = 878,296) and validation (n = 878,297) cohorts; the mean age of the study population was 33.8 years, and 64.5% were male.
• The primary outcome was early-onset HCC, defined as a diagnosis using standard codes, confirmed radiologically or pathologically, and accompanied by cancer-specific critical condition codes.
• Follow‑up was conducted until January 2022, the occurrence of HCC, death, or loss to follow-up, whichever came first.

TAKEAWAY

• Key predictors in the generalized boosted survival model included a history of non-HCC; chronic viral hepatitis; and aspartate aminotransferase, gamma-glutamyl transferase, and serum creatinine levels.
• In the validation cohort, the generalized boosted survival model achieved an area under the receiver operating characteristic curves of 0.945 (95% CI, 0.915-0.975) for 1-year prediction and 0.825 (95% CI, 0.804-0.847) for 5-year prediction.
• Across all models, risk stratification by quartiles revealed that individuals in the highest quartile had a 7.87- to 13.73-fold higher risk for early-onset HCC than those in the lowest quartile.
• A simple, easy-to-use version of the generalized boosted survival model demonstrated excellent performance (root mean square error, 0.287; Pearson correlation coefficient, 0.887) and identified a small group at a very high risk, whose 5-year risk of developing early-onset liver cancer was 11.8%.

IN PRACTICE

“Our finding that the models maintained [HCC] predictability even among participants without established major risk factors is particularly valuable for identifying individuals who might otherwise have been overlooked by conventional risk factor-based screening approaches,” the authors of the study wrote.

SOURCE

The study was led by Seogsong Jeong, Korea University College of Medicine, and Gi-Ae Kim, Kyung Hee University Hospital, Seoul, South Korea. It was published online in Alimentary Pharmacology & Therapeutics.

LIMITATIONS

The study population consisted of Korean individuals, potentially limiting generalizability to other populations. Advanced biomarkers such as liver stiffness measurements, genetic markers, and novel serum indicators were unavailable in the dataset. The relatively low incidence of early-onset HCC in young adults may have presented challenges for model precision.

DISCLOSURES

The study received support from grants provided by the National Research Foundation of Korea and the multidisciplinary research grant aid from the Seoul Metropolitan Government-Seoul National University Boramae Medical Center. No conflicts of interest were reported by the authors.

This article was created using several editorial tools, including AI, as part of the process. Human editors reviewed this content before publication.