Diagnostic performance of FIB-6 for identifying FibroScan-defined advanced fibrosis in Metabolic Dysfunction-Associated Steatotic Liver Disease (MASLD): a cross-sectional diagnostic accuracy study

Metabolic dysfunction–associated steatotic liver disease (MASLD) is a major cause of chronic liver disease worldwide. Accurate non-invasive identification of advanced fibrosis is essential for risk stratification and management. The Fibrosis-6 (FIB-6) score is a recently developed serum-based index, but its performance in MASLD remains unclear. In this retrospective cross-sectional study, 397 adults with MASLD who underwent FibroScan were included. FibroScan-derived liver stiffness categories were used as the surrogate comparator for fibrosis assessment. The diagnostic performance of FIB-6 was evaluated using ROC analysis for significant fibrosis (F2–F4), advanced fibrosis (F3–F4), and cirrhosis (F4), and compared with FIB-4, APRI, NAFLD fibrosis score (NFS), and AST-to-ALT ratio (AAR). Among participants (mean age 47.8 ± 13.0 years; 49.4% male), significant fibrosis was present in 18.4%, advanced fibrosis in 7.6%, and cirrhosis in 4.8%. FIB-6 showed AUROC values of 0.643 for significant fibrosis, 0.749 for advanced fibrosis, and 0.760 for cirrhosis. Among 30 patients with FibroScan-defined advanced fibrosis, FIB-6 identified 29 of 30 cases (96.7%) and had an NPV of 99.4%. Among 19 patients with FibroScan-defined cirrhosis, all cases were identified (19/19), resulting in an NPV of 100% (95% CI: 83.2–100.0); however, these estimates should be interpreted cautiously because of the low prevalence of advanced disease. Outcome-specific optimal cut-offs were identified using the Youden index (1.88 for significant fibrosis and 1.84 for advanced fibrosis/cirrhosis). While FIB-4 and NFS showed slightly higher AUROCs, differences were not statistically significant. FIB-6 demonstrates acceptable diagnostic performance, particularly for excluding advanced fibrosis and cirrhosis in MASLD. Although not superior to established scores, its high negative predictive value and simplicity suggest potential as a rule-out tool pending external validation.