Multi-omics profiling reveals altered mitochondrial metabolism in adipose tissue from patients with metabolic dysfunction-associated steatohepatitis

Abstract

Background: Metabolic dysfunction-associated steatotic liver disease (MASLD) and its more severe form steatohepatitis (MASH) contribute to rising morbidity and mortality rates. The storage of fat in humans is closely associated with these diseases' progression. Thus, adipose tissue metabolic homeostasis could be key in both the onset and progression of MASH.

Methods: We conducted a case-control observational research using a systems biology-based approach to analyse liver, abdominal subcutaneous adipose tissue (SAT), omental visceral adipose tissue (VAT), and blood of n = 100 patients undergoing bariatric surgery (NCT05554224). MASH was diagnosed through histologic assessment. Whole-slide image analysis, lipidomics, proteomics, and transcriptomics were performed on tissue samples. Lipidomics and proteomics profiles were determined on plasma samples.

Findings: Liver transcriptomics, proteomics, and lipidomics revealed interconnected pathways associated with inflammation, mitochondrial dysfunction, and lipotoxicity in MASH. Paired adipose tissue biopsies had larger adipocyte areas in both fat depots in MASH. Enrichment analyses of proteomics and lipidomics data confirmed the association of liver lesions with mitochondrial dysfunction in VAT. Plasma lipidomics identified candidates with high diagnostic accuracy (AUC = 0.919, 95% CI 0.840-0.979) for screening MASH.

Interpretation: Mitochondrial dysfunction is also present in VAT in patients with obesity-associated MASH. This may cause a disruption in the metabolic equilibrium of lipid processing and storage, which impacts the liver and accelerates detrimental adaptative responses.

Funding: The project leading to these results has received funding from 'la Caixa' Foundation (HR21-00430), and from the Instituto de Salud Carlos III (ISCIII) (PI21/00510) and co-funded by the European Union.

Keywords: Interorgan crosstalk; Lipidomics; MASLD; Multi-omics; Multi-tissue.

Fig. 1

Patient selection and study design. Hepatic and adipose tissues and blood were obtained from patients with obesity and with or without MASH for multi-omics analyses, created with BioRender.com (a). Digital pathology involves scanning standard biopsies to generate whole-slide images, breaking them into individual features, and assessing them on continuous or categorical scales. The total score obtained from these features helps in making a diagnosis. Ballooned hepatocytes and lobular inflammation distinguish between patients with and without MASH (b).

Fig. 2

Understanding the mechanisms underlying the development of MASH requires a multi-omics approach, which involves investigating genetic susceptibility. Genome-wide association studies have identified genetic variants in specific genes and facilitate calculation of polygenic scores. The panel shows a chord diagram representing the gene’s association with their affected biological processes (a). Transcriptome data and pathway enrichment analysis have revealed differentially expressed genes in livers with and without MASH (b, c). Furthermore, mass spectrometry-based proteomics (d, e) and lipidomics (f–h) have provided distinct signatures that can differentiate between livers with and without MASH. Genotypes (n = 100), transcriptomics (n = 16), proteomics (n = 18), and lipidomics (n = 100) data were analysed using non-parametric statistical tests (Mann–Whitney U test) for univariate comparisons, and multivariate analyses were applied for multivariable data interpretation.

Fig. 3

White adipose tissue shows anatomical and functional diversity in patients with severe obesity. Using digital pathology, we examined subcutaneous (SAT) and omentum visceral (VAT) adipose tissue. We found structural differences in adipocyte size and fibrosis at the specific collection sites in all patients (a–c). Our proteomics data showed that immune response and cellular repair-related proteins were expressed differently in VAT compared to SAT (d, e). Additionally, our lipidomics analysis revealed distinct compositions for both fat depots (f) and highlighted significant differences in enrichment of oxylipins derived from various fatty acid sources and lysophoshpatidylethanolamines (LPE) (g, h), which provided the most distinctive value. Proteomics (n = 18, per tissue) and lipidomics (n = 100, per tissue) data were analysed using non-parametric statistical tests (Mann–Whitney U test) for univariate comparisons, and multivariate analyses were applied for multivariable data interpretation.

Fig. 4

Patients with and without MASH exhibit distinct differences in the structure and lipid composition in both fat depots, with the most significant disparities in visceral adipose tissue. Subcutaneous adipose tissue (SAT) from patients with MASH had increased adipocyte area and a similar amount of lipids (a, b). Lipidomics studies revealed differences in composition that included significant changes in the distribution of fatty acids, bile acids, and carnitines, which contributed to the classification of patients with and without MASH (c–f). These effects were more apparent in visceral adipose tissue (VAT) (g–i) with differences in the concentration of lipid species that could predict MASH. In VAT, decreased glycerolipids and glycerophospholipids accompanied a significant decrease in carnitines and increased bile acids, also observed in SAT (j–l). Histological determinations and lipidomics (n = 100, per tissue) data were analysed using non-parametric statistical test (Mann–Whitney U test) for univariate comparisons, and multivariate analyses were applied for multivariable data interpretation.

Fig. 5

Proteomics of visceral adipose tissue in patients with MASH unveils mitochondrial dysfunction and an increased extracellular matrix response. In subcutaneous adipose tissue (SAT), specific proteomic signatures were poorly clustered and did not differentiate patients with and without MASH (a, b). However, in visceral adipose tissue (VAT) there were significant MASH-related responses (c–d). In patients with MASH, we observed a significant decrease in proteins associated with mitochondrial beta-oxidation and tricarboxylate transport and an increase in the expression of extracellular matrix peptides in VAT (e). Proteomics (n = 18, per tissue) data were analysed by multivariate statistics.

Fig. 6

Plasma biomarkers for MASH screening. Plasma proteomics do not provide distinct patterns to distinguish patients with and without MASH (a). There were MASH-related differences in circulating VLDL, adiponectin and FGF-21, without practical implementation in clinical practice (b). However, plasma lipidomics delivered robust models for classifying MASH (c, d). Mathematical models indicate that circulating TG 50:1, TG 52:1 and CE 18:1 used in combination yield sufficient diagnostic accuracy to propose a candidate biomarker (e). Proteomics (n = 40), lipidomics (n = 100), and other laboratory assessments (n = 100) data were analysed using non-parametric statistical test (Mann–Whitney U test) for univariate comparisons, and multivariate analyses were applied for multivariable data interpretation.