Cohort characteristics

Long-chain free fatty acids (LCFAs) dominate the metabolome alterations in plasma of early acute myocardial infarction (eAMI) patients compared with healthy control (HC) individuals.

(a) Principal component (PC) analysis demonstrates the remarkable metabolome alterations in eAMI patients compared with HC individuals. (b) KEGG metabolic pathway enrichment of differential metabolites (FDR-adjusted p < 0.05) between eAMI and healthy controls. Fisher’s exact test (one-side) followed by FDR-adjusted p value was used and only pathways with FDR-adjusted p < 0.05 were shown. (c) Proportional structure of 32 detected LCFAs in plasma across all samples of eAMI and HC individuals. LCFAs are colored by degree of saturation, saturated LCFAs are colored in yellow, monounsaturated LCFAs are colored in red, and polyunsaturated LCFAs are colored in blue. (d) Principal coordinate analysis (PCoA) plot of the difference between eAMI and HC groups based on LCFAs proportion profile. (e) Volcano plot of 952 detected metabolites by untargeted metabolome profiling in eAMI patients and HC individuals. Red dots represent significantly up-regulated (FDR-adjusted p < 0.05 and Cliff Delta > 0.4) metabolites in eAMI plasma. Blue dots represent notably eAMI-depleted plasma metabolites (FDR-adjusted p < 0.05 and Cliff Delta < 0.4). Grey dots marked the unchanged metabolites. Significantly altered LCFAs were labeled. The horizontal line represents FDR cutoff of 0.05 and the vertical lines denote |Cliff Delta| > 0.4.

Machine learning demonstrates long-chain free fatty acids (LCFAs) are potential diagnostic biomarkers for eAMI.

(a) The top 20 most important LCFAs reducing the variance (mean squared error (%IncMSE)) in classification of eAMI by random forest model. %IncMSE denotes the increase in mean squared error of predictions as a result of variable (LCFAs) being permuted. The training model was constructed by proportional profile of 32 LCFAs in 39 training samples. (b) Distribution of cross-validation error in random forest classification of eAMI patients as the number of top ranking LCFAs increased. Red dot line marks the optimal number of LCFAs (n = 14) for the classification of eAMI patients. (c and d) Receiver operating characteristic (ROC) curve using the 14 LCFAs optimized by random forest model for the (c) training and (d) test set (right panel, n = 17). Vertical heatmap in each panel denotes specificity for the ROC curve in c and d. AUC, area under the curve.

Bacterial alterations characterize eAMI gut microbiome and greatly explain the variance in plasma long-chain free fatty acids (LCFAs).

(a) Density plot showing sample-wide distribution of (% of explained variances) intrinsic factors (gender, age, and body mass index (BMI)) associated with the plasma concentrations of LCFAs. (b) Bar plot demonstrates the variation of 32 LCFAs explained by identified bacterial taxa at species level using LASSO regression. (c) Stacked bar plot shows the prevalence of gut enterotypes in eAMI and control groups. (d) Circular heatmap represents the correlation between LCFAs and gut bacterial species. Only association groups with FDR-adjusted p value of < 0.01 are colored in this plot.

Alterations in viral gut microbiota between eAMI and HC groups.

Box plots showing the relative abundance of increased (a) and decreased (b) viral gut microbiota at species level in eAMI gut microbiome compared to that of HC group. Differences in abundance were detected using MaAslin2. (c) Significant variations in the abundance, abundance ratios, and counts of virulent and temperate phages between the HC and eAMI groups. Significance was determined by Wilcoxon ranked sum test. For box plots in all panels, the vertical lines extend 1.5 times the interquartile range (top and bottom borders of the box) and the median depicted by the horizontal line inside the box.

Bacterial structural variations (SVs) links to long-chain free fatty acids profiles in acute myocardial infarction (AMI) patients.

(a) Number of SVs of each bacterial species in 56 study participants. (b) Pie chart showing the total identified SVs numbers. (c) Boxplot of Canberra distance between samples within eAMI and healthy control (HC) groups based on SVs profile, significance is determined by two-sided Mann-Whitney U-test. (d) Chord diagram showing significant associations between LCFAs and bacterial SVs. (e) Heatmap demonstrates the associations between SVs of Alistipes shahii and LCFAs. Only significant correlations with FDR-adjusted p value of < 0.01 are visualized. ++, FDR-adjusted p < 0.01, and +++, FDR-adjusted p < 0.001. (f) The deletion rate of the 3-kbp deletion harboring 3-oxoacyl-[acyl-carrier protein] reductase (FabG) in Alistipes shahii. (g) Stacking bar plot represents the prevalence of 3-kbp deletion SV of Alistipes shahii in eAMI and HC groups. (h) Relative abundance of LCFAs in individuals with and without the 3-kbp deletion SV of Alistipes shahii. **, p < 0.01, ***, p < 0.001 determined by two-sided Mann-Whitney U-test.

Early acute myocardial infarction (eAMI)-associated long-chain free fatty acids (LCFAs) bind to thrombosis-related receptors and induce thrombosis in human platelets.

(a) Heatmap of binding energies between eAMI-associated LCFAs and thrombosis-related receptors through molecular docking. Bing affinities are quantified at kJ/mol. (b) The representative AutoDock predicted binding conformation complex of LCFAs versus thrombosis-related receptors, within which the receptors are depicted as cartoons and LCFAs as sticks. (c) The thrombogenic impact of eAMI-associated LCFAs on human platelets is depicted in a heatmap. Platelet aggregation ratios of > 10% and 10% are indicated by blue and grey, respectively. (d) Representative platelet aggregation curves in a time interval of 300 seconds after the treatment of LCFAs at a concentration of 5 mM. Red dashed line indicates 10% threshold.