Modified thresholds for fibrosis risk scores in nonalcoholic fatty liver disease are necessary in the obese

Obesity and its related comorbidities are significant risk factors for nonalcoholic fatty liver disease (NAFLD). Liver fibrosis is the major determinant of long-term outcomes in NAFLD. A non-invasive tool that accurately identifies obese patients at elevated risk of liver fibrosis would be of significant value. Fibrosis risk scores in patients with NAFLD have been proposed but have not been validated in obese populations. We aimed to validate established simple fibrosis scores in bariatric surgical patients. We conducted a prospective study of 107 consecutive high-risk obese patients undergoing primary bariatric surgery. Proposed fibrosis scores (NAFLD fibrosis score; body mass index (BMI), aspartate aminotransferase (AST)/alanine aminotransferase ratio (ALT), and diabetes (BARD); Fibrosis-4 (FIB-4); Forn; and AST to platelet ratio index) were calculated and compared hepatic fibrosis determined by histology of intraoperative liver biopsies. Accuracy was determined, and fibrosis score thresholds were optimized. These modified thresholds were then validated in an independent bariatric surgical population. Liver biopsies were available in 101 patients. Sixty-eight patients had some degree of fibrosis, with 23 patients (23 %) having significant fibrosis (F2-4). The Forn score best predicted significant fibrosis (area under the receiver operator characteristic curve (AUROC) 0.724, p = 0.001). With standard thresholds, the sensitivity for the Forn score for identification of significant fibrosis (F2-4) was 0 %. Using modified thresholds of 3.5, the sensitivity and negative predictive value increased to 85.7 and 94.7 %. This threshold was applied to an independent validation cohort with good accuracy. Fibrosis risk scores using simple markers have moderate success at delineating obese patients with significant NAFLD-related fibrosis. Thresholds, however, need to be lowered to maximize diagnostic accuracy in this cohort.