Study overview.

(A) The overall study design. (B) Alluvial diagram of patient composition of groupings across liver histological features. (C) Relationship between clinical test results and stages of liver histological features. Node size was determined by –log10(p value) in ANOVA tests. Color indicates the significance degree [–log10(p value)] and the direction of change from early to late stages. (D) Number of metabolites analyzed in the liver and plasma. (E-F) Principal component analysis of liver metabolome, plasma metabolome, and liver transcriptome. (G) Disease spectra covered in this cohort and comparison to two published datasets.

Patient characteristics

Hepatic and circulating metabolome in obese individuals with MASLD.

(A) Number of metabolites in each class that were significantly associated with histological features in the liver (q value < 0.05). Metabolite classes with at least 5 metabolites with significance are shown in the plot. (B) Associations between steatosis grades and lipid species in each lipid class. Dots were colored by double bond numbers (left) and carbon numbers (right) of lipid species, respectively. (C) Partial correlation network of the plasma metabolome and clinical covariates. Node size reflects the degree of connectivity, with larger nodes indicating connections to a greater number of metabolite nodes. (D) Heatmaps for pairwise analysis between plasma lipid classes and clinical variables. Linear regression analysis was performed for numerical variables, while logistic regression was conducted for binary variables. Significance values refer to –log10(p value)*sign(coefficients) from regression models. (E) Comparisons of liver metabolome and plasma metabolome regarding their associations with liver steatosis and fibrosis.

Integrative view of key metabolic pathways implicated in liver metabolism in obese individuals with MASLD.

(A) Heatmap of log2 fold changes from pair-wise analysis of lipid metabolism-related genes. Association with steatosis (S) and fibrosis (F) is indicated by black dots (q value < 0.05, top). Results were cross-referenced with two published cohorts (VA cohort, GSE130970; EU cohort, GSE135251, bottom). Asterisks indicate genes with a q-value < 0.05 and a consistent change in direction within our cohort. (B) KEGG metabolic pathways with at least 4 genes or metabolites significantly associated with steatosis [S] or fibrosis [F]. (C) Integrative map of gene and metabolite alterations associated with steatosis (left half of each box/circle) and fibrosis (right half of each box/circle). (D) Metabolite alterations corresponding to the advancement of steatosis grades (x-axis).

Dysregulated mitochondrial function and autophagy during the progression of steatosis and fibrosis.

Gene expression patterns related to mitochondrial metabolism (A), mitochondria-related biological processes (A), autophagy activation (B), and lysosomal biogenesis (B) with statistical association to steatosis and fibrosis. S, steatosis. F, fibrosis.

Steatosis and fibrosis as independent processes.

(A) Network of genes, metabolites, and histological features (fibrosis and steatosis) using partial correlation analysis as described in the methods. (B) Venn diagrams depicting the number of genes significantly associated with steatosis and fibrosis (q value < 0.05). (C) Gene ontology (GO) enrichment of gene sets specific to steatosis or fibrosis.

Liver fibrosis signatures and potential therapeutic targets of fibrosis initiation.

(A) Top 60 fibrosis markers in the liver. A total of 213 genes were identified, from which the top 30 upregulated and top 30 downregulated genes are visualized. The middle plot (blue bar on top) shows the comparison of individuals with fibrosis and without pathology (baseline - “no MASLD”). For the right plot (red bar), individuals with fibrosis were compared to those with steatosis but no fibrosis (baseline – “steatosis but no fibrosis”). Results were cross-referenced with two published cohorts (VA cohort, GSE130970; EU cohort, GSE135251, right). Asterisks indicate genes with a q-value < 0.05 and a consistent change in direction within our cohort. The average gene expressions of liver cell types were obtained from GSE115469 at the log2CPM level (left). Dots in the single cell map indicate zero expression in the corresponding cell types. (B) Enrichment of pathways in the transition from simple steatosis to the onset of fibrosis. (C) Protein-protein interaction network of GTPases and their regulators. Nodes are colored based on gene type, with borders indicating the direction of gene regulation. Node size corresponds to the significance of the genes in relation to fibrosis grades within this cohort. GEFs: Guanine nucleotide exchange factors. GAPs: GTPase-activating proteins. (D) Comparison of expression level changes in GTPase-related genes between this human cohort and an independent 3D spheroid MASH system. Log2 fold change for the human cohort was calculated by comparing patients with grade 2 or 3 fibrosis to those without fibrosis or steatosis. Genes with the same direction of change are colored in red, while others are colored in purple. (E) Expression of GTPase-related genes in patient-derived 3D liver spheroids: control spheroids, spheroids from patients with MASH, and Elafibranor-treated MASH spheroids (n = 4). Sixty-eight genes with a p-value < 0.05 from the ANOVA test were plotted in the diagram, with blue lines highlighting 24 genes that exhibited increased expression in the MASH group compared to the control group.