Translational strategy in NASH diversity: learning from mouse models

Abstract

Objectives

Nonalcoholic steatohepatitis (NASH) is a histological definition that groups together defects in diverse biochemical processes causing hepatic fat accumulation, inflammation, necrosis and fibrosis. Increasing evidence points to different subtypes of nonalcoholic fatty liver disease (NAFLD) which progress to NASH and fibrosis at different rates and may respond differently to treatment. The identification of the types of mechanisms leading to NASH and the discovery of non-invasive biomarkers of NASH subtypes are central for the development of effective treatments and precise diagnosis. This study aims to capture the metabolic signature of different NASH subtypes through a translational research.

Method

We undertook metabolomic serum analysis in a mouse model that spontaneously develops NASH, methionine adenosyltransferase 1a knockout (Mat/a-KO), and compared with WT mice. Top fifty metabolites that more significantly differentiated between genotypes were selected and translated to a human cohort: 535 biopsied patients (353 steatosis, 182 NASH). For that, we performed a Silhouette cluster analysis and validate the process in 1000-fold repetition of a random partition (50/50) of samples into two cohorts with equal proportional representation of steatosis/NASH. The frequency distribution of NAFLD patients into subtypes and of metabolites that significantly differentiated between NASH and steatosis per subtype was calculated.

Results

Silhouette cluster analysis revealed that Mat/a-KO signature sub-classified the patients into two clusters, one showing a serum metabolic profile similar to that observed in Mat/a-KO mice (M-subtype) and other showing a different profile (non-M-subtype). Following the criteria based on 2:70% reproducibility of the frequency distribution, 262 patients were classified as M-subtype and 171 as non-M-subtype. The remaining 102 patients showed a reproducibility of less than 70%. A NASH biomarkers list per subtype was generated based on the frequency distribution (2:70% reproducibility) of the metabolites that significantly differentiated between NASH and steatosis: 54 and 6 metabolites for M- and non-M-subtypes, respectively.

Conclusions

We identified a serum specific metabolomic signature characteristic of Mat1a-KO mice and found that about half of NAFLD patients share it, suggesting that in these patients SAMe synthesis may be impaired. Interestingly, this phenotype was observed in patients with steatosis and NASH, which suggests that impaired SAMe synthesis may occur early in the development of NAFLD in a subgroup of patients. This translational strategy can be applied to different mouse models with diverse mechanisms leading to NASH. These results also indicate that the traditional, mainly pathology-driven classification of NAFLD/NASH, can be refined and perhaps represented by metabolomics classification.