| Abstract: | PURPOSE: Noninvasive tools for the detection of hepatic steatosis are needed. The Fatty Liver Index (FLI), which includes body mass index (BMI), waist circumference, triglycerides, and γ-glutamyl-transferase, has been proposed as a screening tool for fatty liver. Our objective was to validate the FLI for the detection and quantification of hepatic steatosis in an obese population. METHODS: Patients with chronic liver disease and BMI ≥ 28 kg/m(2) underwent liver biopsy and FLI determination. FLI performance for diagnosing steatosis compared with biopsy was assessed using areas under receiver operating characteristic curves (AUROCs), and a novel model for the prediction of significant steatosis (≥5 %) was derived. RESULTS: Among 250 included patients, 65 % were male, and the median BMI was 33 kg/m(2); 48 % had nonalcoholic fatty liver disease, and 77 % had significant (≥5 %) steatosis. The FLI was weakly correlated with the percentage (ρ = 0.25, p = 0.0001) and grade of steatosis (ρ = 0.28, p < 0.00005). The median FLI was higher among patients with significant steatosis (91 vs. 80 with <5 % steatosis; p = 0.0001) and the AUROC for this outcome was 0.67 (95 % CI 0.59-0.76). At an optimal FLI cut-off of 79, the FLI was 81 % sensitive and 49 % specific, and had positive and negative predictive values of 84 and 43 %, respectively. A novel index including triglycerides, glucose, alkaline phosphatase, and BMI outperformed the FLI for predicting significant steatosis [AUROCs 0.78 vs. 0.68; p = 0.009 (n = 247)]. CONCLUSIONS: In obese patients, the FLI is a poor predictor of significant steatosis and has limited utility for steatosis quantification compared with liver histology. A novel index including triglycerides, glucose, alkaline phosphatase, and BMI may be useful, but requires validation. |
