Diagnostic Accuracy of Non-Invasive Tests for MASLD Across Age, Type 2 Diabetes, and Obesity Subgroups: A Multinational Study

Background & aims: Non-invasive tests (NITs) are widely used to risk-stratify patients with metabolic dysfunction-associated steatotic liver disease (MASLD); however, their performance may vary according to patient characteristics. We evaluated the accuracy of NITs in a large, multinational MASLD cohort across select subpopulations. Methods: We analyzed 18,759 adults with biopsy-confirmed MASLD from 41 countries. NITs included FIB-4, liver stiffness measurement (LSM), and Agile-3+. Diagnostic performance for advanced fibrosis (F3-F4) was measured using AUCs across subgroups defined by age, sex, type 2 diabetes (T2D), obesity, and alcohol use. Subgroup-specific cutoffs were derived. Results: Advanced fibrosis was present in 37% of patients. Pooled AUCs were 0.79 for FIB-4, 0.83 for LSM, and 0.86 for Agile-3+. FIB-4 accuracy declined with age (AUC 0.70 in ≥65 years vs. 0.79 in <65 years, p<0.0001) and in middle-aged patients with T2D. The LSM performance remained stable across T2D status but was moderately reduced in patients with obesity and, more profoundly, morbid obesity (BMI >35). Sex and alcohol use had minimal impact on AUCs. Age- and T2D-specific FIB-4 cutoffs varied substantially to maintain predefined accuracy (sensitivity or specificity). The cutoffs for LSM also differed based on patients' BMI, with lower diagnostic cutoffs for advanced fibrosis required in non-obese MASLD (sensitivity 80%: 8.8 kPa in lean, 9.0 kPa overweight, 9.6 kPa in obesity, 11.0 kPa in morbid obesity). Conclusions: Accuracy of non-invasive tests for advanced fibrosis in MASLD is influenced by age, T2D, and obesity. Age-adjusted FIB-4 thresholds may enhance risk stratification. Imaging-based and composite NITs (LSM and Agile-3+) provide more consistent performance across MASLD subpopulations.