Prediction of advanced fibrosis in non-alcoholic fatty liver disease using gut microbiota-based approaches compared with simple non-invasive tools

Abstract

Liver fibrosis is the major determinant of liver related complications in patients with non-alcoholic fatty liver disease (NAFLD). A gut microbiota signature has been explored to predict advanced fibrosis in NAFLD patients. The aim of this study was to validate and compare the diagnostic performance of gut microbiota-based approaches to simple non-invasive tools for the prediction of advanced fibrosis in NAFLD. 16S rRNA gene sequencing was performed in a cohort of 83 biopsy-proven NAFLD patients and 13 patients with non-invasively diagnosed NAFLD-cirrhosis. Random Forest models based on clinical data and sequencing results were compared with transient elastography, the NAFLD fibrosis score (NFS) and FIB-4 index. A Random Forest model containing clinical features and bacterial taxa achieved an area under the curve (AUC) of 0.87 which was only marginally superior to a model without microbiota features (AUC 0.85). The model that aimed to validate a published algorithm achieved an AUC of 0.71. AUC's for NFS and FIB-4 index were 0.86 and 0.85. Transient elastography performed best with an AUC of 0.93. Gut microbiota signatures might help to predict advanced fibrosis in NAFLD. However, transient elastography achieved the best diagnostic performance for the detection of NAFLD patients at risk for disease progression.