Genetic architecture of human plasma lipidome and its link to cardiovascular disease

Abstract

Understanding genetic architecture of plasma lipidome could provide better insights into lipid metabolism and its link to cardiovascular diseases (CVDs). Here, we perform genome-wide association analyses of 141 lipid species (n = 2,181 individuals), followed by phenome-wide scans with 25 CVD related phenotypes (n = 511,700 individuals). We identify 35 lipid-species-associated loci (P <5 ×10^-8), 10 of which associate with CVD risk including five new loci-COL5A1, GLTPD2, SPTLC3, MBOAT7 and GALNT16 (false discovery rate<0.05). We identify loci for lipid species that are shown to predict CVD e.g., SPTLC3 for CER(d18:1/24:1). We show that lipoprotein lipase (LPL) may more efficiently hydrolyze medium length triacylglycerides (TAGs) than others. Polyunsaturated lipids have highest heritability and genetic correlations, suggesting considerable genetic regulation at fatty acids levels. We find low genetic correlations between traditional lipids and lipid species. Our results show that lipidomic profiles capture information beyond traditional lipids and identify genetic variants modifying lipid levels and risk of CVD.

Fig. 1

Study design and work flow. The figure illustrates the study design and key findings of the study

Fig. 2

Heritability of lipidomic profiles and genetic correlations among the lipid species. a Histogram and kernel density curve showing the distribution of heritability estimates across all the lipid species. b Boxplot showing the heritability estimates in each lipid class. c Boxplot showing comparison of the median heritability estimates of lipid species containing C20:4, C20:5 and C22:6 acyl chains and all others. The P-values were calculated using the Wilcoxon rank-sum test. d Hierarchical clustering of lipid species based on genetic correlations among lipid species. Lipids containing polyunsaturated fatty acids C20:5, C20:4 and C22:6 are highlighted with black bars. The data presented in the boxplots represent the interquartile range (IQR) defined by the bounds of the box with the median (middle line of the box) and whiskers extending to the largest/smallest values no further than 1.5 times the IQR. CER   ceramide, DAG  diacylglyceride, LPC    lysophosphatidylcholine, LPE    lysophosphatidylethanolamine, PC    phosphatidylcholine, PCO    phosphatidylcholine-ether, PE    phosphatidylethanolamine, PEO    phosphatidylethanolamine-ether, PI    phosphatidylinositol, CE    cholesteryl ester, SM    sphingomyelin, ST    sterol, TAG    triacyglycerol, Trad traditional lipids

Fig. 3

Lipidomic profiles capture information beyond traditional lipids. The genetic and phenotypic correlations between traditional lipids and molecular lipid species are shown in lower panel. The bar plot in the upper panel shows the heritability estimates of each lipid species (red bars) and the variance explained by all the known loci together (green bars). The lipid species are ordered based on the hierarchical clustering showing the correlations between the lipid species and traditional lipids. TC total cholesterol, TG triglycerides

Fig. 4

Genetic architecture of the lipidome. a Manhattan plot showing associations for all 141 lipid species. Only the associations with P < 1.0 × 10⁻⁴ in the meta-analysis and consistent in directions in all three batches are plotted. The y-axis is capped at −log10 P-value = 30 for better representation of the data. The dotted line represents the threshold for genome-wide significant associations at P < 5.0 × 10⁻⁸. b Genome-wide significant associations between the identified lipid species-associated loci and lipid species showing effect of the loci on the lipidome. The plotted P-values were calculated from the meta-analyses using the inverse variance weighted method for fixed effects. New hits with P < 5.0 × 10⁻⁸ are shown as red dots, new independent hits in previously reported loci are presented as blue dots and hits in previously known loci are presented as black dots

Fig. 5

Relationship between lipid species-associated variants and risk of CVDs. The upper panel shows the association of the identified variants with the strongest associated lipid species. Boxplots show the interquartile range (IQR) defined by the bounds of the box with the median (middle line of the box) of plasma levels of the respective lipid species for each genotype of the variants; whiskers extend to the largest/smallest values no further than 1.5 times the IQR. The lower panel depicts the relationship between the identified variants with CVD phenotypes. The effect sizes (odds ratio) with 95% confidence interval are plotted with respect to the alternate alleles. The associations with CVD phenotypes highlighted in red colour are significant at FDR <0.05

Fig. 6

Patterns in associations and proposed mechanisms for the effect of identified variants on lipid metabolism and clinical outcomes. a Associations of LPL rs11570891-T and LPL activity with TAGs. Change (beta and standard errors) in plasma levels of TAGs per increase in standard deviation of LPL activity with their corresponding P-values, as calculated using linear regression model, are plotted in lower panel. The upper panel shows change (beta and standard errors) in plasma levels of TAGs per T allele with their corresponding P-values, as obtained from meta-analyses of genome-wide association analysis. b Association of LPL variant rs11570891 with LPL activity. The effect size (beta in standardised units and standard error in parenthesis) and P-value were calculated using linear mixed model. Boxplot depicts the interquartile range (IQR) defined by the bounds of the box, median (middle line) and whiskers extending to the largest/smallest values no further than 1.5 times the IQR. c Based on the patterns of the association of lipid species-associated loci with different lipid species, we propose that: (1) LPL rs11570891-T and APOA5 rs964184-C might result in more efficient hydrolysis of medium length TAGs which might results in reduced CVD risk, (2) FADS2 rs28456-G may have observed effect on PUFA metabolism through its inverse effect on FADS2 and FADS1 expressions, (3) SYNGR1 rs18680008-C might have a role in the negative regulation of either desaturation of linoleic acid (C18:2,n-6) or elongation of gamma linoleic acid (C18:3,n-6). (4) PTPRN2 rs10281741-G and MIR100HG rs10790495-G, which have very similar patterns of association with reduced level of long polyunsaturated TAGs, might have a role in negative regulation of either elongation and desaturation of fatty acids or incorporation of long chain unsaturated fatty acids in glycerol backbone during TAG biosynthesis. The positive (+) and negative (−) signs indicate increase or decrease, respectively, in level of lipid species or risk of disease as observed in our study, with different colours for different genetic variant

Fig. 7

Association of known variants for traditional lipids with lipid species and traditional lipids. The P-values for the associations of the lead SNPs (557 SNPs available in our data set) identified through different genome-wide or exome-wide studies of traditional lipids (HDL-C, LDL-C, TG and TC) with lipid species (upper panel) and traditional lipids (lower panel) are plotted. The y-axis in the upper panel is capped at −log10 P-value = 30 for better representation of the data. The SNPs on the x-axis are serially arranged based on their chromosomal positions and as listed in the Supplementary Data 8. The points on the plots are colour coded by the lipid classes in the upper panel and traditional lipid in the lower panel. CER    ceramide, DAG    diacylglyceride, LPC    lysophosphatidylcholine, LPE    lysophosphatidylethanolamine, PC    phosphatidylcholine, PCO    phosphatidylcholine-ether, PE    phosphatidylethanolamine, PEO    phosphatidylethanolamine-ether, PI    phosphatidylinositol, CE    cholesteryl ester, SM    sphingomyelin, ST    sterol, TAG    triacyglycerol, TC total cholesterol, TG triglycerides