Gut microbiota, blood metabolites, and left ventricular diastolic dysfunction in US Hispanics/Latinos.

BACKGROUND: Left ventricular diastolic dysfunction (LVDD) is an important precursor of heart failure (HF), but little is known about its relationship with gut dysbiosis and microbial-related metabolites. By leveraging the multi-omics data from the Hispanic Community Health Study/Study of Latinos (HCHS/SOL), a study with population at high burden of LVDD, we aimed to characterize gut microbiota associated with LVDD and identify metabolite signatures of gut dysbiosis and incident LVDD.RESULTS: We included up to 1996 Hispanic/Latino adults (mean age: 59.4 years; 67.1% female) with comprehensive echocardiography assessments, gut microbiome, and blood metabolome data. LVDD was defined through a composite criterion involving tissue Doppler assessment and left atrial volume index measurements. Among 1996 participants, 916 (45.9%) had prevalent LVDD, and 212 out of 594 participants without LVDD at baseline developed incident LVDD over a median 4.3 years of follow-up. Using multivariable-adjusted analysis of compositions of microbiomes (ANCOM-II) method, we identified 7 out of 512 dominant gut bacterial species (prevalence > 20%) associated with prevalent LVDD (FDR-q < 0.1), with inverse associations being found for Intestinimonas_massiliensis, Clostridium_phoceensis, and Bacteroide_coprocola and positive associations for Gardnerella_vaginali, Acidaminococcus_fermentans, Pseudomonas_aeruginosa, and Necropsobacter_massiliensis. Using multivariable adjusted linear regression, 220 out of 669 circulating metabolites with detection rate > 75% were associated with the identified LVDD-related bacterial species (FDR-q < 0.1), with the majority being linked to Intestinimonas_massiliensis, Clostridium_phoceensis, and Acidaminococcus_fermentans. Furthermore, 46 of these bacteria-associated metabolites, mostly glycerophospholipids, secondary bile acids, and amino acids, were associated with prevalent LVDD (FDR-q < 0.1), 21 of which were associated with incident LVDD (relative risk ranging from 0.81 [p = 0.001, for guanidinoacetate] to 1.25 [p = 9 × 10, for 1-stearoyl-2-arachidonoyl-GPE (18:0/20:4)]). The inclusion of these 21 bacterial-related metabolites significantly improved the prediction of incident LVDD compared with a traditional risk factor model (the area under the receiver operating characteristic curve [AUC] = 0.73 vs 0.70, p = 0.001). Metabolite-based proxy association analyses revealed the inverse associations of Intestinimonas_massilliensis and Clostridium_phoceensis and the positive association of Acidaminococcus_fermentans with incident LVDD.CONCLUSION: In this study of US Hispanics/Latinos, we identified multiple gut bacteria and related metabolites linked to LVDD, suggesting their potential roles in this preclinical HF entity. Video Abstract.