This case-control study comprised 32 male individuals without coronary heart disease (referred to as the Control group) and 66 male patients diagnosed with STEMI (referred to as the STEMI group) ( Figure 1 ). A range of individual characteristics, including age, height, weight, risk factors, blood biochemical parameters, medications administered during hospitalization, and coronary angiography findings, were collected. The body mass index (BMI) was calculated by dividing weight (in kilograms) by the square of height (in meters). Inclusion criteria for the STEMI group included: (i) cardiac troponin (cTn) I/T levels exceeding the upper normal reference value, or creatine kinase isoenzymes exceeding the normal reference limit; (ii) ST-segment elevation observed in two or more contiguous leads on the electrocardiogram (ECG); (iii) and one or more of the following: persistent ischemic chest pain, abnormal wall motion on the ECG, and abnormal coronary angiography findings. Exclusion criteria for participants were: (i) other identifiable causes of coronary thrombosis (e.g., coronary vasospasm or systemic thromboembolism); (ii) evident active infection during hospitalization; (iii) prior history of organic digestive system or gastrointestinal surgery; (iv) history of kidney or respiratory diseases; (v) infection within one month before the study or use of probiotics, prebiotics, postbiotics, antacids, antibiotics, or related preparations. Control subjects were recruited voluntarily from individuals undergoing routine health check-ups at the same institution, all of whom had normal results. The study was approved by the Ethics Committee of Nanjing Jiangbei Hospital (#2022031). All participants provided written informed consent in accordance with the principles outlined in the Declaration of Helsinki. Additionally, this study is purely observational, and no clinical trial registration was conducted.

Fecal samples, approximately the size of two soybean grains, were collected from each participant, either by themselves or by their family members, within 3 minutes of defecation while in the hospital. After collection, the samples were placed in fecal samplers (Biotecan, Shanghai, China), sealed, labeled, and transported to Biotecan Laboratories within 2 days, ensuring the temperature remained below 18°C. Upon arrival, the samples were promptly stored at -80°C.

A total of 98 fecal samples (32 from the control group and 66 from the STEMI group) were collected in fecal samplers and stored at -80°C until they were processed for high-throughput sequencing. Genomic DNA was extracted from the fecal samples using the QIAamp PowerFecal Pro DNA Kit (QIAGEN, Germany). The concentration and integrity of the extracted DNA were evaluated using a NanoDrop ND-1000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA) and agarose gel electrophoresis, respectively. The V3-V4 hypervariable region of the bacterial 16S rRNA gene was amplified by PCR using the primer pair 341F (5′-CCTACGGGNGGCWGCAG-3′) and 805R (5′-GACTACHVGGGTATCTAATCC-3′). Sample-specific paired-end 6-bp barcodes were integrated into TruSeq adapters for multiplexed sequencing. The PCR reaction mixture included 25 μL of Phusion High-Fidelity PCR Master Mix (New England Biolabs, MA, USA), 3 μL of each 10 μM forward and reverse primer, 10 μL of template DNA, 3 μL of DMSO, and 6 μL of nuclease-free water. The thermal cycling protocol comprised an initial denaturation at 98 °C for 30 seconds, followed by 25 cycles of 98 °C for 15 seconds, 58 °C for 15 seconds, and 72 °C for 15 seconds, with a final extension at 72 °C for 1 minute. PCR products were purified using Agencourt AMPure XP Beads (Beckman Coulter, Indianapolis, IN) and quantified with the PicoGreen dsDNA Assay Kit (Invitrogen, Carlsbad, CA, USA). Following individual quantification, the amplicons were pooled in equal amounts for paired-end sequencing (2 × 250 bp) on the Illumina NovaSeq 6000 platform, conducted by Shanghai Biotecan Pharmaceuticals Co., Ltd. (Shanghai, China).

To analyze the sequencing data, the Quantitative Insights Into Microbial Ecology 2 (QIIME 2, v2017.6.0) pipeline and predefined criteria were employed (31, 32). Paired-end reads were assembled using Vsearch V2.4.4 (33). Operational taxonomic units (OTUs) were defined based on 16S rDNA gene sequences, applying a 97% similarity threshold and referencing the Greengenes database via Vsearch V2.4.4. OTUs representing less than 0.001% of the total sequences were excluded from the analysis. The final OTU table was generated by averaging, rounding, and rarefying, using 100 evenly resampled OTU subsets at 90% of the minimum sequencing depth. Abundance curves were plotted at the OTU level, and sequencing depth was evaluated and confirmed through rarefaction analysis. Alpha diversity was assessed using the Chao1, Simpson, and Shannon indices within the QIIME 2 platform. Statistical comparisons were conducted using the Pairwise Wilcox test. Beta diversity was evaluated using Weighted UniFrac principal component analysis (PCoA). To compare the intestinal bacterial composition and structure between the control and STEMI groups, Permutational Multivariate Analysis of Variance (PERMANOVA) was applied. The comparability between groups was assessed using One-way Analysis of Similarities (ANOSIM). Linear Discriminant Analysis Effect Size (LEfSe) was used to identify taxa with significantly different abundances between the groups, based on default parameters (logarithmic LDA score ≥ 2) (34, 35). The phylogenetic tree was visualized using GraPhlAn (http://huttenhower.sph.harvard.edu/GraPhlAn). To predict gut microbial functions, the Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt, PICRUSt2 v2.3.0-b) was employed (36, 37), with the Univariate Test used to assess significant differences.

Blood samples were obtained from all participants after an overnight fast, prior to undergoing coronary angiography and receiving any concomitant medications. The samples were then analyzed in the Department of Clinical Laboratory for routine biomarkers, including total cholesterol (TC), triglycerides (TG), high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), serum creatinine, and other relevant indices. Plasma was immediately separated and subjected to centrifugation at 800×g for 10 minutes at 4°C. The LDL-C subfractions were classified and quantified using the Quantimetrix Lipoprint LDL System (Quantimetrix Corporation, Redondo Beach, CA, USA), following the manufacturer’s protocol (38). Briefly, plasma mixed with liquid loading gel was applied to the top of pre-cast 3% polyacrylamide gel tubes. After 30 minutes of photopolymerization at room temperature, the samples were electrophoresed for 1 hour, and densitometric analysis was performed at 610 nm.

The primary endpoint of the follow-up was the occurrence of the first MACEs, which encompassed cardiovascular death, non-fatal ischemic stroke, recurrent myocardial infarction, the need for emergency or repeated revascularization, and rehospitalization due to heart failure, as previously defined (39–41).

Data analysis was performed using SPSS version 23.0 (IBM, Chicago, IL, USA) and GraphPad Prism version 7.0 (GraphPad, San Diego, CA, USA). Categorical variables were compared between groups using the chi-square test or Fisher’s exact test. For continuous variables, the Shapiro-Wilk test was applied to assess the normality of the data distribution. Comparisons between groups were made using a two-tailed Student’s t-test for normally distributed data and a Mann-Whitney U-test for non-normally distributed data. The exact number of patients and the values of continuous variables for each group are provided in the figure legends. Venn diagrams, heatmaps, and Spearman’s rank correlation were generated using R software (v3.6.3). A p-value < 0.05 was considered statistically significant in all analyses.

The baseline characteristics of the Control and STEMI groups were presented in Table 1 . When compared to the Control group, the STEMI group exhibited significantly elevated fasting plasma glucose (FPG) levels (P=0.0068) and significantly reduced high-density lipoprotein cholesterol (HDL-C) levels (P=0.0065). Moreover, the prevalence of smoking was notably higher in the STEMI group than in the Control group (P<0.0001).

Over a median follow-up period of 13 months (range: 12–22 months), a total of 18 patients experienced MACEs, comprising 0 cardiovascular deaths, 3 non-fatal ischemic strokes, 1 recurrent myocardial infarction, 3 cases requiring either emergency or elective repeat revascularization, 11 hospital readmissions due to heart failure, and 8 readmissions for unstable angina ( Figure 1 ). When comparing patients with STEMI who developed MACEs (MACEs group) with those who remained event-free (Non-MACEs group), the MACEs group exhibited significantly lower BMI (P=0.0066) and a markedly higher CK-MB level (P=0.0349) ( Table 2 ). Additionally, NT-proBNP levels were noticeably higher in the MACEs group than in the Non-MACEs group (P=0.0731).

In the Venn diagram, a total of 15,040 OTUs were shared between the control and STEMI groups, with the STEMI group harboring a greater number of unique OTUs (27,194) compared to the control group (4,294) ( Figure 2A ). The alpha diversity of the stool microbiome, reflecting the number of species within each sample, varied significantly between the two groups (P = 0.0019 for the Shannon index, P = 0.0769 for the Simpson index, and P = 0.4464 for the Chao1 index) ( Figures 2B-D ). Regarding β-diversity, intergroup differences were more distinct than intragroup variations (R = 0.034, P = 0.202) ( Figure 2E ). Principal component analysis (PCoA) based on Bray-Curtis distance further revealed significant differences in gut microbiota composition between the control and STEMI groups (P = 0.034 for PC1 vs. PC2, P = 0.034 for PC1 vs. PC3, and P = 0.017 for PC2 vs. PC3) ( Figures 2F-H ).

Given the limitations of 16S rDNA amplicon pyrosequencing, our analysis primarily focused on genus-level microbial composition. Figure 3A and Supplementary Figure 1 illustrated the distribution of gut microbiota at this taxonomic level. Both groups exhibited a predominance of Bacteroides, Prevotella, Faecalibacterium, Megamonas, Phascolarctobacterium, Bifidobacterium, Roseburia, Agathobacter, Escherichia−Shigella, and Streptococcus. However, Bacteroides, Phascolarctobacterium, Bifidobacterium, and Streptococcus were more enriched in the STEMI group, whereas Prevotella, Faecalibacterium, Megamonas, Roseburia, Agathobacter, and Escherichia−Shigella were relatively more abundant in the control group. The Wilcoxon rank sum test (LEfSe) (P < 0.05, LDA > 2) identified key genera distinguishing the two groups, with the STEMI group exhibiting a greater number of biomarker bacteria than the control group ( Figures 3B, C ). Specifically, Bifidobacterium, Streptococcus, Collinsella, Alistipes, Megasphaera, Odoribacter, and Lactobacillus were characteristic genera of the STEMI group, whereas Romboutsia and Agathobacter were indicative of the control group ( Figure 3C ). Additionally, statistical significance was confirmed for five differentially abundant genera, with P values of 0.0499, 0.012, 0.0043, 0.0035, and 0.0007 for Bifidobacterium, Megasphaera, Alistipes, Collinsella, and Streptococcus, respectively ( Figure 3D ).

To further explore functional differences between the two groups, we performed Picrust2 analysis, which predicted variations in KEGG, METACYC, CAZY, and GMM modules. The top 10 most significantly altered pathways were presented in Supplementary Figures 2A-D . Notably, several KEGG pathways associated with metabolic regulation were disrupted in the STEMI group, including the insulin signaling pathway (P = 0.0006), thiamine metabolism (P = 0.0017), terpenoid backbone biosynthesis (P = 0.0039), and porphyrin and chlorophyll metabolism (P = 0.0056) ( Supplementary Figure 2A ). Specifically, all these metabolic pathways were computational and should be interpreted as hypothesis-generating rather than confirmatory.

In the Venn diagram, 10,163 OTUs were shared between the Non-MACEs and MACEs groups, with the MACEs group possessing a greater number of unique OTUs (10,015) in comparison to the Non-MACEs group (9,850) ( Figure 4A ). The alpha diversity of the gut microbiome, reflecting the species richness within each sample, showed no statistically significant differences between the two groups (P = 0.0549 for the Shannon index, P = 0.1182 for the Simpson index, and P = 0.7193 for the Chao1 index) ( Figures 4B-D ). In terms of β-diversity, intergroup differences were more pronounced than intragroup variations (R = 0.139, P = 0.004) ( Figure 4E ). However, Principal Component Analysis (PCoA) based on Bray-Curtis distance did not reveal significant compositional differences in gut microbiota between the Non-MACEs and MACEs groups (P = 0.312 for PC1 vs. PC2, P = 0.300 for PC1 vs. PC3, and P = 0.312 for PC2 vs. PC3) ( Figures 4F-H ).

Due to the inherent limitations of 16S rDNA amplicon pyrosequencing, our analysis predominantly focused on microbial composition at the genus level. Figure 5A and Supplementary Figure 3 illustrated the taxonomic distribution of gut microbiota at this resolution. Both the Non-MACEs and MACEs groups were characterized by the presence of Bacteroides, Prevotella, Faecalibacterium, Bifidobacterium, Megamonas, Streptococcus, Collinsella, Roseburia, Phascolarctobacterium, and [Eubacterium]_coprostanoligenes_group. However, Bacteroides, Megamonas, Collinsella, and Phascolarctobacterium were more prevalent in the MACEs group, whereas Prevotella, Faecalibacterium, Bifidobacterium, Streptococcus, Roseburia, and [Eubacterium]_coprostanoligenes_group were found in higher abundance in the Non-MACEs group. The Wilcoxon rank sum test (LEfSe) (P < 0.05, LDA > 2) identified distinct marker genera differentiating the two groups, with the Non-MACEs group exhibiting a greater number of characteristic bacterial taxa compared to the MACEs group ( Figures 5B, C ). Specifically, Coprococcus, Christensenellaceae_R_7_group, UCG_002, Eubacterium_coprostanoligenes_group, UCG_005, Dorea, Family_XIII_UCG_001, Ruminococcus_gauvreauii_group, NK4A214_group, Family_XIII_AD3011_group, Marvinbryantia, Fusicatenibacter, Senegalimassilia, Negativibacillus, Olsenella, UCG_010, and CAG_56 were signature genera for the Non-MACEs group, while Hungatella, Lachnoclostridium, Ruminococcus_gnavus_group, and Bacteroides were representative of the MACEs group ( Figure 5C ). Moreover, statistical analysis confirmed significant differences in 10 genera, with P values of 0.0486, 0.0435, 0.0358, 0.0339, 0.0236, 0.0232, 0.0224, 0.0075, 0.005, and 0.002 for [Ruminococcus]_gnavus_group, Bacteroides, Christensenellaceae_R-7_group, Coprococcus, [Eubacterium]_coprostanoligenes_group, UCG-005, Lachnoclostridium, Dorea, NK4A214_group, and UCG-002, respectively ( Figure 5D ).

To further investigate functional discrepancies between the groups, Picrust2 analysis was conducted to predict variations in KEGG, CAZY, METACYC, and GMM functional modules, with 5, 5, 9, and 5 significantly different pathways identified, respectively, as presented in Supplementary Figure 4A-D . Notably, metabolic pathways such as chloroalkane and chloroalkene degradation (P = 0.0499), carotenoid biosynthesis (P = 0.0435), and steroid biosynthesis (P = 0.0046) were implicated in organic compound metabolism, while the mRNA surveillance pathway (P = 0.0142) was associated with signal transduction and regulatory processes ( Supplementary Figure 4A ). It is important to note that all of these metabolic pathways were inferred through computational analysis and should be regarded as exploratory hypotheses rather than definitive findings.

Table 2 presented the blood lipid profiles of patients with STEMI, stratified by the presence or absence of MACEs. No statistically significant differences were observed in TC, TG, HDL-C, and LDL-C levels between the MACEs and Non-MACEs groups. Furthermore, none of the LDL-C subfractions exhibited significant variation between the two groups ( Figure 6 ).

As illustrated in Figure 7 , both significant positive and negative correlations were detected among the top 10 marker genera. Notable correlations included a negative association between Bacteroides and UCG-002 (ρ=-0.4816, P<0.05), as well as a positive correlation between Bacteroides and Lachnoclostridium (ρ=0.5294, P<0.05). Additionally, [Eubacterium]_coprostanoligenes_group demonstrated a positive correlation with Coprococcus (ρ=0.5384, P<0.05), UCG-002 (ρ=0.5921, P<0.01), Christensenellaceae_R-7_group (ρ=0.5776, P<0.05), and UCG-005 (ρ=0.5806, P<0.05), while exhibiting a negative association with Lachnoclostridium (ρ=-0.6463, P<0.01) and [Ruminococcus]_gnavus_group (ρ=-0.5280, P<0.05). Furthermore, Coprococcus displayed a positive correlation with UCG-002 (ρ=0.5794, P<0.05), whereas UCG-002 showed significant positive associations with Christensenellaceae_R-7_group (ρ=0.7595, P<0.001), UCG-005 (ρ=0.6173, P<0.01), Dorea (ρ=0.5012, P<0.01), and NK4A214_group (ρ=0.5554, P<0.05), but was negatively correlated with Lachnoclostridium (ρ=-0.9111, P<0.001) and [Ruminococcus]_gnavus_group (ρ=-0.5114, P<0.05). Similarly, Christensenellaceae_R-7_group exhibited a negative correlation with Lachnoclostridium (ρ=-0.6914, P<0.01) and a positive correlation with UCG-005 (ρ=0.5311, P<0.05). Additional significant associations included a positive correlation between Lachnoclostridium and [Ruminococcus]_gnavus_group (ρ=0.6388, P<0.01) and a negative correlation with UCG-005 (ρ=-0.6727, P<0.01), while [Ruminococcus]_gnavus_group also displayed a negative correlation with UCG-005 (ρ=-0.6296, P<0.01). Moreover, we further examined the relationship between gut microbiota and LDL-C subfractions. Interestingly, significant negative correlations were identified between Coprococcus and LDLC-4 (ρ=-0.5488, P<0.05), LDLC-5 (ρ=-0.6418, P<0.01), and LDLC-6 (ρ=-0.4988, P<0.05). Similarly, UCG-002 (ρ=-0.4948, P<0.05) and Christensenellaceae_R-7_group (ρ=-0.5032, P<0.05) were negatively associated with LDLC-4. Notably, no significant positive correlations were observed ( Figure 7 ).

As shown in Figure 8 , both significantly positive and negative correlations were observed among the top 10 marker genera, including between Bacteroides and [Ruminococcus]_gnavus_group (ρ=0.6666, P<0.01), between [Eubacterium]_coprostanoligenes_group and UCG-002 (ρ=0.6987, P<0.01), between [Eubacterium]_coprostanoligenes_group and Christensenellaceae_R-7_group (ρ=0.5533, P<0.05), between [Eubacterium]_coprostanoligenes_group and UCG-005 (ρ=0.5253, P<0.05), between [Eubacterium]_coprostanoligenes_group and NK4A214_group (ρ=0.8014, P<0.001), between Coprococcus and [Ruminococcus]_gnavus_group (ρ=-0.6936, P<0.01), between UCG-002 and Christensenellaceae_R-7_group (ρ=0.8449, P<0.001), between UCG-002 and [Ruminococcus]_gnavus_group (ρ=-0.4965, P<0.05), between UCG-002 and UCG-005 (ρ=0.7977, P<0.001), between UCG-002 and NK4A214_group (ρ=0.7353, P<0.001), between Christensenellaceae_R-7_group and UCG-005 (ρ=0.940, P<0.001), between Christensenellaceae_R-7_group and NK4A214_group (ρ=0.5668, P<0.05), between [Ruminococcus]_gnavus_group and NK4A214_group (ρ=-0.5146, P<0.05), and between UCG-005 and NK4A214_group (ρ=0.5512, P<0.05). However, we did not observe any significant correlations between gut microbiota and LDL-C subfractions, which was inconsistent with the MACEs group ( Figure 8 ).

The ROC curve analysis demonstrated the predictive potential of various gut microbiota genera, with AUC values as follows: Bacteroides (AUC = 0.698, 95% CI: 0.524-0.871), [Eubacterium]_coprostanoligenes_group (AUC = 0.722, 95% CI: 0.548-0.897), Coprococcus (AUC = 0.727, 95% CI: 0.554-0.900), UCG-002 (AUC = 0.759, 95% CI: 0.595-0.924), Christensenellaceae_R-7_group (AUC = 0.752, 95% CI: 0.593-0.910), Lachnoclostridium (AUC = 0.722, 95% CI: 0.545-0.900), [Ruminococcus]_gnavus_group (AUC = 0.677, 95% CI: 0.498-0.857), UCG-005 (AUC = 0.685, 95% CI: 0.503-0.868), Dorea (AUC = 0.750, 95% CI: 0.581-0.919), and NK4A214_group (AUC = 0.779, 95% CI: 0.625-0.934) ( Table 3 ). Among these key marker genera, UCG-002, Christensenellaceae_R-7_group, and NK4A214_group exhibited relatively superior predictive performance, with AUC values exceeding 0.75. Additionally, the predictive capacities of six LDL-C subfractions were evaluated: LDLC-1 (AUC = 0.571, 95% CI: 0.380–0.762), LDLC-2 (AUC = 0.509, 95% CI: 0.315–0.704), LDLC-3 (AUC = 0.552, 95% CI: 0.361–0.744), LDLC-4 (AUC = 0.560, 95% CI: 0.368–0.753), LDLC-5 (AUC = 0.515, 95% CI: 0.324–0.707), and LDLC-6 (AUC = 0.551, 95% CI: 0.360–0.741) ( Table 3 ). Notably, in comparison to gut microbiota markers, LDL-C subfractions exhibited lower predictive powers for MACEs.