Abstract
BACKGROUND: Metabolic dysfunction-associated fatty liver disease (MAFLD) has emerged as the leading cause of liver dysfunction and poses a significant risk for progression to cirrhosis. In this study, we aimed to compare the ability of various metabolic indices and identify those most effective for the prediction of MAFLD. METHODS: This cross-sectional study included 1471 patients (69.6% MAFLD) between 2011 and 2024. Specific metabolic indices were calculated. The associations between indices and MAFLD were analyzed via logistic regression analysis and restricted cubic splines. The prediction ability of the indices was evaluated via receiver operating characteristic (ROC) analysis in different subgroups. RESULTS: Triglyceride glucose (TyG) index, homeostatic model assessment for insulin resistance (HOMA-IR), and estimated glucose disposal rate (eGDR) had the strongest associations with MAFLD (final adjusted ORs for TyG = 3.60, HOMA-IR = 3.35, eGDR = 2.97 per unit change). According to the ROC curve analysis, TyG index and hepatic steatosis index (HSI) had the highest area under the curve (AUC) for the prediction of MAFLD in females (AUC = 0.832) and males (AUC = 0.848), respectively. TyG also performed the best in lean/normal-weight and overweight/obesity individuals, with AUCs of 0.890 and 0.787, respectively. TyG in those ≥ 65 years old (AUC = 0.844) and eGDR in those < 65 years old (AUC = 0.843) had the highest prediction performance for MAFLD. CONCLUSION: Insulin resistance markers, including TyG, HOMA-IR, and eGDR, had the strongest associations with MAFLD. For predicting MAFLD, TyG had the highest performance across all subgroups, except for males and those < 65 years, where HSI and eGDR performed better, respectively.