Does Incorporating Change in APRI or FIB-4 Indices Over Time Improve the Accuracy of a Single Index for Identifying Liver Fibrosis in Persons With Chronic Hepatitis C Virus Infection?

Abstract

The aspartate aminotransferase-to-platelet ratio index (APRI) and a fibrosis index calculated using platelets (FIB-4) have been proposed as noninvasive markers of liver fibrosis. To determine APRI/FIB-4 accuracy for predicting histologic liver fibrosis and evaluate whether incorporating change in index improves test accuracy in hepatitis C virus (HCV)-infected Alaska Native persons. Using liver histology as the gold standard, we determined the test characteristics of APRI to predict Metavir ≥F2 fibrosis and FIB-4 to predict Metavir ≥F3 fibrosis. Index discrimination was measured as the area under the receiver operator characteristic curve. We fit a logistic regression model to determine whether incorporating change in APRI/FIB-4 over time improved index discrimination. Among 283 participants, 46% were female, 48% had a body mass index >30, 11% had diabetes mellitus, 8% reported current heavy alcohol use. Participants were infected with HCV genotypes 1 (68%), 2 (17%), or 3 (15%). On liver histology, 30% of study participants had ≥F2 fibrosis and 15% had ≥F3 fibrosis. The positive predictive value of an APRI>1.5/FIB-4>3.25 for identifying fibrosis was 77%/78%. The negative predictive value of an APRI<0.5/FIB-4<1.45 was 91%/87%. The area under the receiver operator characteristic curve of an APRI/FIB-4 for identifying fibrosis was 0.82/0.84. Incorporating change in APRI/FIB-4 did not improve index discrimination. The accuracy of APRI/FIB-4 for identifying liver fibrosis in HCV-infected Alaska Native persons is similar to that reported in other populations and could help prioritize patients for treatment living in areas without access to liver biopsy. Change in APRI/FIB-4 was not predictive of degree of fibrosis.