Extracellular Vesicle‐Based miRNA Profiling in Metabolic Dysfunction‐Associated Steatotic Liver Disease: A Secondary Analysis of the MULTISITE Clinical Trial

Figures

• 

  FIGURE 1. Participant flowchart. Participant flowchart showing the recruitment, initial group allocation, reclassification resulting from the change in nomenclature from NAFLD to MASLD, sample loss, and the final sample size included in data analysis.

• 

  FIGURE 2. Method overview and EV characterization. (A) Method overview of 148 plasma samples (1), (red arrows) pooled in the five study groups for EV characterization, and (2), (green arrows) individually analysed for miRNA sequencing. Created in BioRender.com. EV characterization using (B) Nanoparticle tracking analysis (NTA) of EV‐enriched plasma pools (Study pools) and PBS control (PBS background) presented as mean ± SD (n = 5), (C) Western blot analysis (WB) showing the five plasma pools in each well: (1) Lean control group, (2) Obesity control group, (3) MASLD group at baseline, (4) 1 month, and (5) 5 months targeting Flot‐1 (MW ∼49 kDa), CD63 (MW ∼25‐60 kDa). (D) Representative immuno electron microscopy (IEM) image of CD9‐positive gold particles, scale bar 200 nm.

• 

  FIGURE 3. Volcano plot and group comparison of miRNA expression levels. (A) Volcano plot analysis illustrating differentially expressed miRNAs between MASLD and the Obesity control group at baseline, miRNAs with a p value < 0.05 and log2foldchange ≥0.29 (equivalent to FC ≥ 1.22) are considered significantly differentially expressed. miRNA VST‐normalized expression between the baseline groups is illustrated for (B) miR‐122, (C) miR‐185, (D) miR‐146b, (E) miR‐10a, whereas the effect of weight loss intervention for the MASLD group is illustrated with downregulated miRNAs (F) miR‐191, (G) miR‐335, (H) miR‐26a, and (I) miR‐151a, and upregulated miRNAs (J) miR‐451a, and (K) miR‐337 (n = 28) at Baseline, 1 month, and 5 months. A box plot illustrating each group and the individual participants as dots (Connected with a line for the intervention group). Data presented as median, interquartile range, whiskers and significant level: p < 0.05 (*), p < 0.01 (**), p < 0.001 (***), p < 0.0001 (****).

• 

  FIGURE 4. miRNA functional analysis. Functional analysis for the four significantly differential expressed miRNAs between the MASLD group and obesity control group: miR‐122, miR‐146b, miR‐10a, and miR‐185 are visualized as (A) miRNA target gene network using miRTarBase, miRecords, and TarBase, miRNA shown as nodes (green) and gene node (red) connected with regulatory interaction lines (grey). (B) GO and C) KEGG analysis visualized as dot plots with colour and size reflecting statistical significance and gene count, respectively.

• 

  FIGURE 5. Linear regression between miRNA expression and clinical parameters. Linear regression analysis with BH correction showcasing VST‐normalized expression (X‐axis) of miR‐122 and clinical parameters (Y‐axis) for (A) liver fat (PDFF, %), (B) ALT levels units per litre (U/L), (C) AST levels (U/L), (D) GGT levels (U/L), (E) HDL cholesterol (mmol/L), and F) miR‐10a and HDL levels (mmol/L) among all individuals with obesity (n = 59). Data are presented as R 2 value and p value with significance level at p < 0.05.

• 

  FIGURE 6. miRNA prediction of MASLD and high liver fat content. Logistic regression analysis with a ROC curve, and leave‐one‐out cross‐validation model with a SHAP analysis was created to (A+B) predict MASLD among all individuals with obesity (n = 59) using panel 1, (C+D) predict high liver fat (≥9.5%) among all individuals with obesity (n = 59) using panel 1, (E) using panel 3, and (F) using panel 2.