Abstract
BACKGROUND AND AIMS: At-risk metabolic dysfunction-associated steatohepatitis (MASH) elevates risks of liver-related and all-cause morbidity and mortality. We developed and validated a noninvasive score using routine clinical indicators to identify at-risk MASH in obesity.
APPROACH AND RESULTS: Using data from 1961 individuals across 5 independent bariatric cohorts with liver biopsy, we developed the predictive score in 1 derivation cohort (n=1095), performed internal validation (bootstrapping), and conducted external validation using the remaining 4 biopsy-confirmed cohorts (n=866). The score was also validated in the international overweight/obese cohorts from the UK Biobank (n=15,745) and NHANES database (n=1573). Predictive value for severe liver-related outcomes (SLROs, including cirrhosis, hepatocellular carcinoma, etc) was assessed in a UK Biobank subcohort (n=1955; median 13.7-year follow-up). Head-to-head comparisons with existing indices were performed. The predictive model, designated as FMO (Fibrotic/at-risk MASH in Obesity), incorporated aspartate aminotransferase, alanine aminotransferase, triglyceride, and high-density lipoprotein cholesterol. The FMO model demonstrated robust discrimination in derivation (AUROC=0.874, 95% CI 0.844-0.905) and nationwide external validation cohorts (AUROCs=0.803-0.874), and in global validation in both NHANES and UK Biobank (AUROCs=0.866 and 0.753, respectively). Longitudinal analysis confirmed SLRO’s prediction (the Harrell C-index=0.703). In the derivation cohort, the FMO model demonstrated optimal rule-out [cutoff 0.05, sensitivity ≥0.90, negative predictive value (NPV) 0.976] and rule-in [cutoff 0.22, specificity ≥0.90, positive predictive value (PPV) 0.481] performance. External validation showed NPVs of 0.907-1.00 and PPVs of 0.333-0.630. Comparative analyses revealed superior diagnostic performance of the FMO model versus some existing models.
CONCLUSIONS: The FMO is an accurate and cost-effective noninvasive score for at-risk MASH identification in populations with obesity.