Serum metabolites associate with lipid phenotypes among Bogalusa Heart Study participants

Abstract

Background and aims: Dyslipidemia has been identified as a major risk factor for cardiovascular disease. We aimed to identify metabolites and metabolite modules showing novel association with lipids among Bogalusa Heart Study (BHS) participants using untargeted metabolomics.

Methods and results: Untargeted ultrahigh performance liquid chromatography-tandem mass spectroscopy was used to quantify serum metabolites of 1 243 BHS participants (816 whites and 427 African-Americans). The association of single metabolites with lipids was assessed using multiple linear regression models to adjust for covariables. Weighted correlation network analysis was utilized to identify modules of co-abundant metabolites and examine their covariable adjusted correlations with lipids. All analyses were conducted according to race and using Bonferroni-corrected α-thresholds to determine statistical significance. Thirteen metabolites with known biochemical identities showing novel association achieved Bonferroni-significance, p < 1.04 × 10^-5, and showed consistent effect directions in both whites and African-Americans. Twelve were from lipid sub-pathways including fatty acid metabolism (arachidonoylcholine, dihomo-linolenoyl-choline, docosahexaenoylcholine, linoleoylcholine, oleoylcholine, palmitoylcholine, and stearoylcholine), monohydroxy fatty acids (2-hydroxybehenate, 2-hydroxypalmitate, and 2-hydroxystearate), and lysoplasmalogens [1-(1-enyl-oleoyl)-GPE (P-18:1) and 1-(1-enyl-stearoyl)-GPE (P-18:0)]. The gamma-glutamylglutamine, peptide from the gamma-glutamyl amino acid sub-pathway, were also identified. In addition, four metabolite modules achieved Bonferroni-significance, p < 1.39 × 10^-3, in both whites and African-Americans. These four modules were largely comprised of metabolites from lipid sub-pathways, with one module comprised of metabolites which were not identified in the single metabolite analyses.

Conclusion: The current study identified 13 metabolites and 4 metabolite modules showing novel association with lipids, providing new insights into the physiological mechanisms regulating lipid levels.

Keywords: High-density lipoprotein cholesterol; Lipids; Low-density lipoprotein cholesterol; Metabolomics; Total cholesterol; Triglyceride; Weighted correlation network analysis (WGCNA).

Figure 1.

Volcano plots of effect sizes versus –log₁₀ p values for all 1202 metabolite associations with total cholesterol (A), low-density lipoprotein cholesterol (C), high-density lipoprotein cholesterol (E), and triglyceride (G) among whites and total cholesterol (B), low-density lipoprotein cholesterol (D), high-density lipoprotein cholesterol (F), and triglyceride (H) among African-Americans.

Figure 2.

Correlations of metabolite modules with lipid phenotypes among whites (A) and African-Americans (B). HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; TC, total cholesterol; TG, triglyceride. Color should be used for figures in print.