Background: A comprehensive, replicated atlas of circulating metabolites for incident coronary heart disease (CHD) across race diverse populations is lacking and metabolite signatures of early-onset CHD remain largely unidentified. Methods: We conducted a two-stage metabolome wide-association analysis using Cox regression model for incident CHD, with discovery analyses in 22,742 CHD-free individuals with 1245 blood metabolites profiled from 7 multi-ethnic cohorts (1,124 incident cases over 7.5~17.0 yrs of follow-up) in TOPMed, and replication analyses in 32,615 CHD-free individuals from 7 multi-ethnic cohorts (3,365 incident cases over 7.5~19.3 years) (Fig1.a). Random-effect meta-analysis was used to pool results from each cohort in these two stages. We further evaluated the associations of identified metabolites with incident CHD diagnosed at different ages. Results: We identified 189 metabolites (FDR<0.05) associated with incident CHD, with 127 metabolites (p<0.05) replicated (Fig1.b). Over 90% of these replicated metabolites showed positive associations, with the majority belonging to glycerolipids, phosphatidylethanolamine, fatty acids, lactoyl amino acid, histidine, aromatic amino acids, branched amino acids (Fig1.b). In the Study of Latinos (SOL, n=13,322), 14 out of these 127 metabolites were associated with incident CHD diagnosed before age 50 yrs (FDR<0.05; Fig1.c), including the known atherogenic metabolites (e.g., cholesterol, fibrinopeptide A), harmful microbial derived trimethylamine N−oxide, sugar sweeteners (e.g., mannitol/sorbitol, erythritol), markers of insulin resistance, inflammation and oxidative stress (e.g., mannose, erythronate, gluconate, suberoylcarnitine), and novel metabolites not previously linked to CHD (e.g., C−glycosyltryptophan, hydroxymalonate, and methyl glucopyranoside). Further, associations of these metabolites with CHD diagnosed at younger age tend to be stronger than those with late-onset cases (e.g., the hazard ratio per SD increase in mannose decreased from 4.4 for CHD diagnosed at age 45 to 1.5 for case diagnosed at age 65; Fig1.d). Adding metabolites to conventional risk factors improved AUC of CHD risk prediction from 0.78 to 0.84 (p<0.001) (Fig1.e). Conclusion: We provide the most comprehensive, replicated, multi-ethnic atlas of circulating metabolites for incident CHD, identify a set of early-onset CHD metabolite markers, and demonstrate significant gains in CHD risk prediction with identified metabolites.
Luo et al. (Tue,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: