Gut microbiome compositional clusters in association with cardiovascular risk: An observational cohort study.

Unsupervised hierarchical clustering of gut microbiome data identified two distinct GM profiles among 211 participants with cardiovascular comorbidities.

• Total cohort: 211 participants, median age 60 [IQR: 50-70] years, 57.3% male
• Cluster H comprised 104 participants and Cluster L comprised 107 participants
• Clusters were statistically distinct (P < 0.001)
• Patients were enrolled from Mayo Clinic between 2013 and 2018 and had coronary artery disease, hypertension, hyperlipidemia, or diabetes mellitus
• Bacterial DNA was analyzed in the V3-V5 region of 16S rDNA

Cluster L participants had a significantly lower incidence of Major Adverse Cardiac Events compared to Cluster H over median follow-up of 7.4 years.

• HR = 0.48, 95% CI: 0.26-0.91, P = 0.024
• MACE was defined as a composite of cardiac events, heart failure, and all-cause mortality
• Association was evaluated using Cox regression
• Median follow-up was 7.4 years

Cluster L participants had a healthier cardiovascular risk profile compared to Cluster H participants.

• Cluster L participants were younger (P < 0.001) and more likely female (P = 0.009)
• Cluster L had lower BMI (P = 0.007)
• Cluster L had lower prevalence of hypertension (P = 0.010) and hyperlipidemia (P = 0.005)
• Cluster L had lower coronary artery disease prevalence (P = 0.003)

Cluster L had higher microbial diversity and a lower Bacillota-to-Bacteroidetes ratio compared to Cluster H.

• Cluster L had higher operational taxonomic units (P < 0.001)
• Cluster L had lower Bacillota-to-Bacteroidetes ratio (P < 0.001) compared to Cluster H
• Beta-diversity was plotted using Principal Coordinates Analysis
• These findings suggest greater microbial diversity and a compositionally distinct microbiome in the lower-risk cluster

The two clusters differed in the predominant bacterial taxa identified by Linear Discriminant Analysis.

• Predominant taxa in Cluster L included Bacteroides, Alistipes, and Parabacteroides
• Predominant taxa in Cluster H included Blautia, Agathobacter, and Clostridium sensu stricto-1
• Linear Discriminant Analysis was used to identify GM taxa with differential abundance among clusters and their effect sizes

Clinical factors contributing to cluster assignment were identified using Permutational Multivariate Analysis of Variance.

• PERMANOVA was used to identify clinical factors contributing to cluster assignment
• The study design was a prospective observational cohort
• Unsupervised hierarchical clustering was used to classify participants without a priori group definitions