Streptococcus Species Abundance in the Gut Is Linked to Subclinical Coronary Atherosclerosis in 8973 Participants From the SCAPIS Cohort

Abstract

Background: Gut microbiota have been implicated in atherosclerotic disease, but their relation with subclinical coronary atherosclerosis is unclear. This study aimed to identify associations between the gut microbiome and computed tomography-based measures of coronary atherosclerosis and to explore relevant clinical correlates.

Methods: We conducted a cross-sectional study of 8973 participants (50 to 65 years of age) without overt atherosclerotic disease from the population-based SCAPIS (Swedish Cardiopulmonary Bioimage Study). Coronary atherosclerosis was measured using coronary artery calcium score and coronary computed tomography angiography. Gut microbiota species abundance and functional potential were assessed with shotgun metagenomics sequencing of fecal samples, and associations with coronary atherosclerosis were evaluated with multivariable regression models adjusted for cardiovascular risk factors. Associated species were evaluated for association with inflammatory markers, metabolites, and corresponding species in saliva.

Results: The mean age of the study sample was 57.4 years, and 53.7% were female. Coronary artery calcification was detected in 40.3%, and 5.4% had at least 1 stenosis with >50% occlusion. Sixty-four species were associated with coronary artery calcium score independent of cardiovascular risk factors, with the strongest associations observed for Streptococcus anginosus and Streptococcus oralis subsp oralis (P<1×10^-5). Associations were largely similar across coronary computed tomography angiography-based measurements. Out of the 64 species, 19 species, including streptococci and other species commonly found in the oral cavity, were associated with high-sensitivity C-reactive protein plasma concentrations, and 16 with neutrophil counts. Gut microbial species that are commonly found in the oral cavity were negatively associated with plasma indole propionate and positively associated with plasma secondary bile acids and imidazole propionate. Five species, including 3 streptococci, correlated with the same species in saliva and were associated with worse dental health in the Malmö Offspring Dental Study. Microbial functional potential of dissimilatory nitrate reduction, anaerobic fatty acid β-oxidation, and amino acid degradation were associated with coronary artery calcium score.

Conclusions: This study provides evidence of an association of a gut microbiota composition characterized by increased abundance of Streptococcus spp and other species commonly found in the oral cavity with coronary atherosclerosis and systemic inflammation markers. Further longitudinal and experimental studies are warranted to explore the potential implications of a bacterial component in atherogenesis.

Keywords: Streptococcus; atherosclerosis; gastrointestinal microbiome; metagenomics; tomography.

Figure 1.

Flowchart of the overall study design. CACS indicates coronary artery calcium score; CVD, cardiovascular disease; GMM, gut metabolic module; IBD, inflammatory bowel disease; MGS, metagenomics species; MODS, Malmö Offspring Dental Study; PPI, proton-pump inhibitor; SCAPIS, Swedish Cardiopulmonary Bioimage Study; and SIS, segment involvement score.

Figure 2.

Association of alpha and beta diversity with coronary artery calcium score. A, The association of alpha diversity with coronary atherosclerosis was assessed by fitting linear mixed models with coronary artery calcium score (CACS) as the outcome and the alpha diversity indices as the exposure. The basic model (n=8972) was adjusted for age, sex, country of birth, and metagenomics extraction plate as fixed effects and family relatedness as random effect. The main model (n=8118) was adjusted further for smoking; physical activity; energy-adjusted intake of carbohydrate, protein, and fiber; systolic and diastolic blood pressure; total cholesterol; high-density lipoprotein and low-density lipoprotein cholesterol levels; triglycerides; body mass index; diabetes; and self-reported medication for dyslipidemia, hypertension, and diabetes as fixed effects. B, For beta diversity, a distance-based multivariate ANOVA of CACS as the independent variable was applied and the same fixed-effect adjustments were made without the random-effect adjustment. In the basic (n=8757) and the full model (n=7921), one participant for each family relatedness cluster was removed. Box plots represent the Bray-Curtis dissimilarity between participants with CACS=0 and the other CACS groups. For CACS=0, the boxplot represents the within-group dissimilarity.

Figure 3.

Phylum-level taxonomic profiling of the fecal and saliva samples across the 3 data sets. MODS indicates Malmö Offspring Dental Study; MOS, Malmö Offspring Study; and SCAPIS, Swedish Cardiopulmonary Bioimage Study.

Figure 4.

Cladogram of species investigated in the study. Circles represent metagenomic species annotated to the lowest taxonomic level. Dots indicate the species positively (red) and negatively (blue) associated with coronary artery calcium score (CACS) in the main model (false discovery rate q<0.05). The following unclassified non–CACS-associated metagenomics species were collapsed: *99 species with lowest annotation to the order Bacteroidales; †95 species with lowest annotation to phylum Firmicutes; ‡163 species with lowest annotation to the class Clostridia; and §651 species with lowest annotation to the order Eubacteriales. A indicates Actinobacteria; B, Bacteroides; F, Firmicutes; and P, Proteobacteria.

Figure 5.

Associations between coronary artery calcium score–associated gut species and alternate measurements of atherosclerosis and markers of inflammation and infection. Heatmap showing the associations of the 64 coronary artery calcium score (CACS)–associated species (false discovery rate q<0.05) with CACS, other coronary computed tomography angiography atherosclerosis traits, carotid plaque, and inflammation biomarkers. Associations with CACS and with inflammatory markers high-sensitivity C-reactive protein and neutrophil and leukocyte counts are presented as linear regression coefficients after adjustment for the main model covariates (age; sex; country of birth; metagenomics extraction plate; smoking; physical activity; energy-adjusted carbohydrate, protein, and fiber intake; systolic and diastolic blood pressure; high-density lipoprotein, low-density lipoprotein, and total cholesterol levels; triglycerides; body mass index; diabetes; and self-reported medication for dyslipidemia, hypertension, and diabetes as fixed effects and first-degree family relatedness as a random effect). Associations with atherosclerosis traits (ie, modified Duke index, segment involvement score, any atherosclerosis, and ≥50% stenosis) and presence of carotid plaques are presented as odds ratio (OR) after adjustment for main model covariates. Associations with false discovery rate q<0.05 are marked with 2 asterisks (**) and those with P<0.05 are marked with 1 asterisk (*).