Lower FIB-4 threshold in patients with diabetes improves diagnostic accuracy of the test in a Hispanic population

Abstract

Background Non-invasive tests (NITs) can be used to estimate the severity of fibrosis in patients with nonalcoholic fatty liver disease (NAFLD) but their diagnostic accuracy is variable. Hispanic patients are at increased risk of NAFLD and diabetes. We evaluate the diagnostic performance of the fibrosis index based on 4 factors (FIB-4) in a population of Hispanic patients who underwent vibration-controlled transient elastography (VCTE). Methods A total of 1,524 patients underwent VCTE at University of California, Los Angeles from July 18, 2019 to June 7, 2022. Ultimately 110 patients were identified as Hispanic, with confirmed NAFLD. Sensitivity, specificity, positive predictive value and negative predictive value of FIB-4 threshold ≥1.3 were calculated. Logistic regression models were used to determine updated thresholds for patients with and without diabetes based on Youden’s index. Results Of the 110 patients, the majority (65%) were female. Prevalence of diabetes was higher in the group with clinically significant fibrosis (76% vs. 36%, P<0.001). Using a FIB-4 threshold ≥1.3 to predict clinically significant fibrosis (F2–F4 on VCTE), area under the receiver operating characteristic (AUROC) was 0.74. By incorporating diabetes status, AUROC was 0.81 when employing a FIB-4 threshold of ≥1.0 in patients with diabetes and ≥1.5 in patients without diabetes. Conclusions Using a FIB-4 threshold of ≥1.0 in patients with diabetes and ≥1.5 in patients without diabetes improves the diagnostic performance of the test. The new FIB-4 including diabetes status will lead to improved screening in patients who are at risk of clinically significant fibrosis.