Performance of Noninvasive Liver Fibrosis Tests in Morbidly Obese Patients with Nonalcoholic Fatty Liver Disease

Background Nonalcoholic fatty liver disease (NAFLD) is highly prevalent in morbidly obese patients, and fibrosis is an independent predictor of mortality. Noninvasive tests (NITs) are being developed for the detection of advanced fibrosis (AF). Purpose To assess the performance of three NITs (NAFLD fibrosis score, NFS, fibrosis-4 index, FIB-4, and aspartate aminotransferase-to-platelet ratio, APRI), in the identification of AF among morbidly obese patients. Materials and Methods Patients, who underwent bariatric surgery between 2004 and 2009 and had liver biopsy, were included. Fibrosis stages >= F2 and >= F3 were defined as significant and AF, respectively. Published and optimal thresholds (Youden index) for NFS, FIB-4 and APRI, sensitivity, specificity, positive and negative predictive values (PPV-NPV), and area under the receiver operator curves (AUROC) were evaluated. Results Among 584 patients (mean age 43.3 +/- 11.3 years, 21.2% male, 75% white, mean BMI 45.5 +/- 8.80), 31.7% had NASH. Stages distributions were F1 = 68.1%, F2 = 16.4%, F3 = 8%, and F4 = 3.2%. At published thresholds, all 3 NITs performed poorly for detection of AF, with AUROC < 0.62. Overall performance at optimal thresholds improved to 0.68, 0.72, and 0.74 for NFS, FIB-4, and APRI, respectively. At optimal thresholds, all tests had good NPV (94.4-95.9%) but low PPV (24.2-32.5%). Combinations of the tests did not improve their performance. Conclusions NFS, FIB-4, and APRI fall short to detect advanced fibrosis but valuable for excluding advanced fibrosis. More research is needed to develop new NITs with high positive predictive value.

Automated summary

In a study of morbidly obese patients, NFS, FIB-4, and APRI were found to be ineffective at detecting advanced fibrosis but were valuable for excluding it. At optimal thresholds, the tests had good NPV but low PPV, and combining them did not improve performance. Additional research is needed to develop new NITs with high positive predictive value.