Whole metabolome profiling identifies the combination of an eicosanoid, a bile acid and an androgen as a highly accurate marker of liver fibrosis in patients with non-alcoholic fatty liver disease

Background and aims:

Non-alcoholic fatty liver disease (NAFLD) is among the most frequent causes of liver disease and estimates based on imaging and biopsy studies suggest that about 20 – 30% of adults in the United States and Europe have excess hepatic fat accumulation, and are at risk of developing progressive fibrosis, end-stage liver disease or hepatocellular carcinoma. Liver biopsy is considered the gold standard in the diagnostic evaluation of patients with suspected NAFLD, and fibrosis at index biopsy identifies those at risk of progression. However, liver biopsy is uncomfortable, more expensive than most non-invasive tools and carries a procedure-related risk which together renders it unsuitable for regular screening. This study aims to identify novel non-invasive markers by metabolite profiling that correlate with histological key lesions of progressive NAFLD.

Methods:

In this multicenter cohort study, 74 patients with biopsy-proven NAFLD (NAFL, n = 42; NASH, n = 32) and 62 healthy blood donors were enrolled. Among the NAFLD patients, 13 had advanced fibrosis (≥F2) while the remaining 61 showed absent or mild fibrosis (F0-F1). Obtained EDTA plasma samples were subjected to metabolite analysis by the GC-MS- and LC-MS/MS-based MxP® Broad Profiling approach, and by the targeted platforms MxP® Steroids, MxP® Lipids and MxP® Eicosanoids. The metabolic data were statistically analyzed by applying univariate (analysis of variance) as well as multivariate (random forest, elastic net, and linear discriminant analysis) algorithms, and peak signals were compared to key histology lesions of NAFLD, i.e. fibrosis, activity, and steatosis.

Results:

A multi-marker panel, consisting of an eicosanoid, a bile acid and an androgen, successfully differentiated the EDTA plasma samples into two main clusters: fibrosis stage ≥2 and fibrosis stage < 2 (AUC = 0.95, Sens = 0.92, Spec = 0.90, PPV = 0.225, NPV = 0.997). Thus, our biomarker outperforms the FIB4 index (AUC = 0.80), the ELF score (AUC = 0.82), and plasma levels of caspase-cleaved keratin 18 fragments (AUC = 0.68).

Conclusions:

The identified metabolite panel was significantly associated with liver fibrosis stage in NAFLD patients and reveals an excellent test performance to detect early prognostic histological liver lesions. In sum, this biomarker holds high promise to accurately detect fibrosis in NAFLD patients non-invasively in plasma, and thus may serve as a novel non-invasive screening tool in NAFLD patients.