Comparative performance of HCC risk scores in identifying low-risk patients with cirrhosis: A multicenter cohort study.
BACKGROUND & AIMS The rising prevalence of cirrhosis, driven by non-viral etiologies, strains healthcare systems and necessitates personalized hepatocellular carcinoma (HCC) surveillance. We conducted a head-to-head comparison of the Toronto HCC risk index (THRI) and aMAP score within an etiology-diverse Western population to evaluate their performance in identifying low-risk patients. METHODS We included adults with cirrhosis from 3 sites in the Netherlands. The exclusion criteria were follow-up <6 months, missing risk score data, or prior HCC. THRI and aMAP were calculated as previously reported. Performance was assessed using time-dependent AUC analysis and cumulative HCC incidence using Kaplan-Meier analysis. RESULTS In a cohort of 1531 patients, the median age was 53 years (IQR: 44-61), 63% were male, the median CTP was 5 (IQR: 5-7). The etiology of liver disease was non-viral in 60% of patients, with the majority having steatotic liver disease (SLD: 453, 30%). Over a median follow-up of 5.4 years (IQR: 3.0-9.1), 196 patients developed HCC. Overall 5-year cumulative HCC incidence was 7.2% (95%CI: 5.7-8.7). THRI offered superior discrimination over aMAP at 3 years (AUC 0.74 vs 0.64) and at 5 years (0.73 versus 0.70). Patients identified by THRI as low-risk (n = 269, 18%) had a lower 5-year risk of HCC (0.5%) than those identified as low risk by aMAP (2.9%). Among the patients considered low-risk by aMAP (n = 437), two-thirds were assigned a higher risk category by THRI and had a substantially higher 5-year HCC incidence than low-risk THRI and low-risk aMAP (4.3%vs. 0%, p = 0.009). CONCLUSION THRI appears more effective than aMAP in identifying patients with negligible 5-year HCC risk and may inform personalized HCC surveillance strategies, particularly in supporting deferral decisions in low-risk patients.