Genetic Loci Associated With Plasma Concentration of Low-Density Lipoprotein Cholesterol, High-Density Lipoprotein Cholesterol, Triglycerides, Apolipoprotein A1, and Apolipoprotein B Among 6382 White Women in Genome-Wide Analysis With ReplicationCLINICAL PERSPECTIVE
Background— Genome-wide genetic association analysis represents an opportunity for a comprehensive survey of the genes governing lipid metabolism, potentially revealing new insights or even therapeutic strategies for cardiovascular disease and related metabolic disorders.
Methods and Results— We have performed large-scale, genome-wide genetic analysis among 6382 white women with replication in 2 cohorts of 970 additional white men and women for associations between common single-nucleotide polymorphisms and low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, triglycerides, apolipoprotein (Apo) A1, and ApoB. Genome-wide associations (P<5×10−8) were found at the PCSK9 gene, the APOB gene, the LPL gene, the APOA1-APOA5 locus, the LIPC gene, the CETP gene, the LDLR gene, and the APOE locus. In addition, genome-wide associations with triglycerides at the GCKR gene confirm and extend emerging links between glucose and lipid metabolism. Still other genome-wide associations at the 1p13.3 locus are consistent with emerging biological properties for a region of the genome, possibly related to the SORT1 gene. Below genome-wide significance, our study provides confirmatory evidence for associations at 5 novel loci with low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, or triglycerides reported recently in separate genome-wide association studies. The total proportion of variance explained by common variation at the genome-wide candidate loci ranges from 4.3% for triglycerides to 12.6% for ApoB.
Conclusion— Genome-wide associations at the GCKR gene and near the SORT1 gene, as well as confirmatory associations at 5 additional novel loci, suggest emerging biological pathways for lipid metabolism among white women.
Received February 13, 2008; accepted June 23, 2008.
Plasma lipid levels, which are highly variable in the population and heritable, are major determinants of cardiovascular disease. To date, genetic dissection of mendelian dyslipidemias and biochemical analysis of plasma lipid determinants have revealed a limited number of genes and pathways relevant to lipid metabolism.1–4 Nevertheless, identifying these few pathways has been crucial for understanding the pathophysiology of cardiovascular disease and for developing targeted therapeutic strategies. It remains possible that additional genes and pathways may be involved and may lead to further clues about the origins of cardiovascular disease, including potential differences between men and women. This possibility can be explored with genome-wide genetic analysis enabled by recent progress in understanding the human genome.
Editorial p 3
Clinical Perspective p 30
We performed genome-wide genotyping of 341 518 polymorphisms among ≈6382 white women in the Women’s Genome Health Study5,6 to evaluate the extent of common genetic influence on 5 plasma lipid fractions: low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), triglycerides, apolipoprotein (Apo) A1, and ApoB. The associations with genome-wide significance (P<5×10−8) confirmed involvement of known and emerging loci in lipid metabolism. Associations at the emerging loci and others could be replicated in smaller samples of white men and women from the Pravastatin Inflammation/CRP Evaluation (PRINCE)7 and the Cholesterol and Pharmacogenetics (CAP)8 studies when genotype information was available from separate genome-wide scans. In these data, we further confirm multiple associations from very recently reported and independent genome-wide association studies of LDL-C, HDL-C, and triglycerides.9–12
The primary analyses were performed in the large-scale Women’s Genome Health Study (WGHS)5 with replication in the smaller PRINCE and CAP studies.7,8 In brief, the WGHS is an initiative to examine genetic correlates with cardiovascular disease and other clinical outcomes among participants in the Women’s Health Study,6 an ongoing prospective cohort of initially healthy American women ≥45 years of age at baseline. Plasma levels of LDL-C, HDL-C, triglycerides, ApoA1, and ApoB were determined by direct assay and had low coefficients of variation.5,13 The PRINCE7 and CAP8 data used for replication derive from baseline blood samples in 2 statin trials included in the Pharmacogenomics and Risk of Cardiovascular Disease program (http://www.pharmgkb.org/network/members/parc.jsp). Clinical characteristics of all 3 study populations are provided in Table 1.
Genotype Data Sets
Genotyping was performed with the Illumina (San Diego, Calif) Infinium II assay. For the WGHS samples, either Illumina HumanHap300 Duo “+” chips or the combination of the Illumina HumanHap300 Duo and iSelect chips was used. In both cases, the custom content (“+” or iSelect content) was the same and consisted of candidate single-nucleotide polymorphisms (SNPs) chosen without regard to allele frequency for potential functional effects in cardiovascular disease and other clinical conditions. In all, 363 808 SNPs were represented by the combination of tag SNPs in the standard panel14 and custom content before exclusions based on quality control measures. In data reduction, the default genotype cluster assignments for the standard HumanHap300 panel were used in BeadStudio version 3.1 software. For the custom content, cluster definitions were first determined by the automated procedure in BeadStudio and then manually curated when ≥1 of the following conditions were met: Hardy-Weinberg equilibrium deviation P<0.0005, call rate <95%, or GenTrain score <0.65, leading to manual review for 15 479 assays. In all, manually assigned cluster positions were used for 2264 of the 45 882 custom-content SNPs with successful assays. In total, 7635 WGHS samples passed quality control, with 98% of the SNPs successfully genotyped. In addition, these samples were further validated on the basis of matching calls for a set for 44 SNPs that had already been genotyped by an independent method across the entire WGHS cohort. Among these retained samples, principal-component analysis based on identity by state of genotypes in PLINK15 for 1443 ancestry informative SNPs (Fst>0.4 in CEU, YRB, and JPN+CHB HapMap populations16) was used to identify a subgroup with nearly complete correspondence (99.7%) to self-reported European ancestry. The 6382 WGHS participants who made up the final data set for analysis of lipid associations had both inferred and self-reported European ancestry, were free of diabetes at baseline, and were not using lipid-lowering therapy, again by self-report. Within this final collection of subjects, 341 518 SNPs were retained on the basis of minor allele frequency >1% and deviations from Hardy-Weinberg equilibrium with a value of P>10−6 as determined by an exact method in PLINK.15,17 Of the final SNPs, 307 595 derived from the HumanHap300 panel and the remaining 33 923 derived from the custom content. Of the SNPs with minor allele frequency between 1% and 5%, 4868 (68%) derived from the custom content.
Similarly, genotyping of the PRINCE and CAP cohorts was performed with the HumanHap300 platform but did not include the custom SNP content from the WGHS. The final data for analysis included 671 and 299 samples from PRINCE and CAP, respectively, all with self-reported European ancestry. Among 314 621 SNPs, the final PRINCE and CAP samples had at least 96.8% and 97.5% complete genotyping, respectively.
Before testing for genetic associations, lipid fractions from the WGHS were adjusted by linear regression in R18 for age, body mass index, smoking status, menopausal status, and use of hormone replacement therapy. A small number of samples (<9) for each lipid fraction were excluded from further analysis on the basis of outlying values in the residuals. In PRINCE and CAP, the adjustment also included gender (both studies) and enrollment criteria (PRINCE only, primary or secondary prevention) but not menopausal status. Analysis for triglycerides was performed on log-transformed values. SNP association tests were performed by linear regression in PLINK under the assumption of an additive relationship between the number of copies of the minor allele and the mean residual lipid values. With 6382 samples, there was 80% power to detect additive genetic effects explaining 0.61% of the variance in one of the lipid fractions at the 5×10−8 significance level. To confirm that any ancestral diversity remaining in the final data set of whites did not unduly inflate the significance of the reported associations, the regression procedure was repeated on all SNPs with correction by genomic control in PLINK. Assessment of the influence of all common SNPs at the candidate loci on lipid fractions was performed by multiple linear regression on the residuals of the adjusted lipid fractions, again with additive model encoding of genetic effects. For each combination of lipid fraction and candidate locus, initial regression models included all SNPs with association value of P<0.1 within a maximum of 500 kb of a candidate locus SNP with association value of P<10−7. A minimal set of nonredundant SNPs in these models was determined by stepwise backward-forward selection based on the bayesian information criterion.19 Genomic context for all annotations derived from human genome build 36.1 and dbSNP build 126.
The authors had full access to and take full responsibility for the integrity of the data. All authors have read and agree to the manuscript as written.
Among the 6382 white WGHS participants (Table 1), we identified 13 primary SNPs at 10 loci on 7 chromosomes that were associated with at least 1 adjusted lipid fraction and an association probability value of <5×10−8, generally considered sufficient for genome-wide significance among whites (Table 2).20 Many of the SNPs with genome-wide association for 1 lipid fraction are also strongly associated with a second, sometimes correlated, lipid fraction, eg, rs506585 with LDL-C and ApoB. The same 10 primary loci were identified at genome-wide significance without adjustment of the lipid fractions for standard clinical covariates. Essentially identical results were found for the analysis that adjusted association statistics in the white sample by genomic control (parameter range, 1.02 to 1.03), suggesting that residual population stratification was not responsible for the associations (Data Supplement Table 1). A full listing of all associations with P<10−7, including median lipid values by genotype and additional statistics, is provided in the online Data Supplement (Data Supplement Table II).
Eight of the loci include genes known to have allelic variation associated with ≥1 of the lipid fractions (see Data Supplement Figure IA through IH). For example, SNPs at or near the CETP gene were associated with HDL-C (P<10−40), SNPs at or near the APOA5 gene were associated with triglycerides (P<10−14), and SNPs at or near the LDLR gene were associated with LDL-C (P<10−14). For the loci that included recognized lipid metabolic genes, nonsynonymous SNPs with P<10−7 were found in PCSK9 (rs11591147), APOB (rs693), LPL (rs328), APOA5 (rs3135506), and CETP (rs5880, rs5882), whereas a synonymous substitution was found for LDLR (rs2228671), thereby reinforcing the candidacy of these genes in determining the lipid fractions. The genome-wide association for ApoA1 at 11q23.3 encoded a potential nonsynonymous change in a predicted gene from the UCSC catalog (QSK gene, similar to serine/threonine kinases, not in Refseq) adjacent to the APOA1 gene. The most significant SNP for HDL-C at 15q21.3 is somewhat remote (≈41 kb) from the presumed functional gene LIPC, but HDL-C associations at this locus are correlated with associations for ApoA1 and, within 0.5 kB of the LIPC gene, rs1800588 is highly significant for both lipid fractions (P=2.4×10−12 [ApoA1], P=1.2×10−7 [HDL]). Similarly, at 19p13.31, the most significant association with LDL-C is far from the APOE gene (rs4803750; >160 kb; P<10−13), but the second-most-significant SNP at the locus for the related ApoB fraction (rs769449) maps within the APOE gene itself and is strongly associated with LDL-C (P=4.7×10−7; Data Supplement Figure IH and Data Supplement Table I). Genotypes in a subset of WGHS whites from a separate study21 showed moderate linkage disequilibrium (r2=0.25) between rs4803750 and the nonsynonymous SNP (rs7412, R158C) in the APOE coding region that distinguishes the E2 and E3 alleles with known differences in lipid levels. Nevertheless, it is impossible to exclude additional functional contributions to the genetic effects at the APOE locus from neighboring genes, including 3 that encode the other apolipoproteins, APOC1, APOC2, and APOC4.
The highly significant associations at the APOA1-APOA5 locus included not only ApoA1 and triglycerides but also ApoB. The association with ApoB appears related to the triglyceride association inasmuch as adjusting ApoB for (log-transformed) triglyceride level eliminated the ApoB genetic associations but not vice versa. In contrast, neither adjustment of ApoA1 for triglyceride level nor adjustment of triglycerides for ApoA1 level had a large effect on the genetic associations for either of these 2 lipid fractions. The SNPs associated with ApoA1 are nearer the APOA1 gene than the APOA5 gene, known for its influence on triglycerides (Data Supplement Figure ID).
The remaining associations correspond to emerging loci for genetic effects on lipids. The SNPs at 2p23.3 that are associated with triglycerides coincide with the glucokinase regulatory protein gene (GCKR) (the Figure 1A). The most significant SNP here, rs1260326, encodes a nonsynonymous change (P446L) and is also associated with ApoA1 (P=9.9×10−7) and ApoB (P=1.6×10−6), although with less significance than the association with triglycerides (P<10−14). At 1p13.3, the association of rs646776 with ApoB (P<10−23) and LDL-C (P<10−18) maps near the genes for the cadherin EGF LAG 7-pass G-type receptor 2 (CELSR2) and the proline/serine-rich coiled-coil protein 1 (PSRC1), separated from the sortilin 1 (SORT1) gene by a site of elevated recombination frequency (Figure 1B).
Characteristics of the most significant SNP at each candidate locus within the WGHS were examined in greater detail. Among all of these primary SNPs, the minor alleles ranged in frequency from 1.7% to 41.4%. Some of the effects of the minor alleles were consistent with increased cardiovascular risk; others were consistent with decreased risk. To compare the shifts in mean lipid level with genotype across all lipid fractions at once, we scaled the additive model regression β coefficients by the adjusted lipid fraction SDs (SDs are given in the legend to Table 2). Among all primary associations, the absolute magnitude of this standardized shift per allele ranged from a minimum of 0.11 SD corresponding to an increase of 2.6 mg/dL in ApoA1, with each additional copy of the minor allele of rs331 at 8p21.3 (LPL), to a maximum of 0.44 SD corresponding to a decrease of 11.2 mg/dL in ApoB for the minor allele of rs11591147 at 1p32.3 (PCSK9). The most significant primary association (P=1.05×10−41) involved rs3764261 at 16q13 (CETP) with minor allele frequency 37% and was characterized by an increase of 4.0 mg/dL for the mean value of HDL-C for each additional copy of the minor allele. The effects of the single most significant SNP at each locus explained a maximum of 3% of the residual lipid variance (for HDL-C and rs3764261 at CETP) and a minimum of 0.5% (for ApoB and rs3135506 at APOA5-APOA1). Among the primary associations, there was no strong evidence for nonadditive effects of the minor allele as judged by the lack of significance for a likelihood ratio test comparing the additive regression model with an alternative genotype model with an additional degree of freedom.
Associations for LDL-C, HDL-C, and triglycerides could be confirmed at the P<0.05 significance level in at least 1 of 2 independent and much smaller samples of white men and women from the PRINCE (n=671) and CAP (n=299) studies, for which genotypes at 6 of the primary SNPs from the candidate loci were available as a result of parallel genome-wide scans (Table 3). These replications were consistent with the larger WGHS population in terms of minor allele frequency, the estimated magnitude and direction of the influence of the minor allele on lipid levels, and the proportion of variance explained; they included the associations at both the established loci (2p24.1, 15q21.3, 16q13, 19q13.31) and the emerging loci (1p13.3 and 2p23.3). Of the 2 WGHS associations that were not significant in one of the replication cohorts, the rs4803750 association that maps near the APOE gene differed in PRINCE in the magnitude but not the sign of the effect estimate compared with the WGHS sample, whereas the rs506585 (APOB) association with LDL-C in CAP had an effect estimate that was comparable to the effect in the WGHS in both magnitude and sign. In addition, all of the associations with P<10−7 in the whole WGHS sample had consistent effect estimates in 2 internal subsets of 4693 and 2058 individuals corresponding to consecutive and separate genotyping batches within the WGHS. Had these subsets been viewed as conventional discovery and replication cohorts, all of the candidate loci except PCSK9 would have had a strongest association with a value of P<10−4 in the first sample and at least nominal significance in the second sample after Bonferroni correction for all SNPs carried forward for replication from the first sample (Data Supplement Table II).
To estimate the contribution of common variation within the WGHS sample at each of the candidate loci to the lipid fractions, we constructed multiple regression models that initially included the most significant SNP at each candidate locus (P<10−7) and neighboring SNPs with association values of P<0.1. For each locus and initial set of SNPs, a nonredundant final set of SNPs was chosen with backward-forward step selection using the bayesian information criterion (Data Supplement Table III). As with the primary associations, the additional SNPs retained in the final models varied widely in both minor allele frequency and the direction of the effect of the minor alleles. The most explanatory loci within the WGHS for the residuals of the lipid fractions were APOE for LDL-C and ApoB, CETP for HDL-C and ApoA1, and APOA5-APOA1 for triglycerides (Table 4). The proportion of the residual variance explained for the common variation of any single locus ranged from a minimum of 0.50% in ApoA1 for locus 8p21.3 (LPL) to a maximum of 5.87% in ApoB for locus 19q13.31 (APOE).
Similarly, the total proportion of the variance explained for each residual lipid fraction in the WGHS by all of the significant common variation was estimated by summing the individual contributions from each of the candidate loci. As much as 12.56% of the residual variance in ApoB was explained by common variation at the candidate loci, whereas as little as 4.34% was explained for triglycerides (Table 4). At the same time, after further adjustment of the residual lipid fractions for the significant genetic variation at all of the candidate loci at once, essentially none of the remaining genetic variation across the genome met a formal standard for genome-wide significance with any of the lipid fractions (Data Supplement Figure II).
While this work was in progress, 4 reports describing genome-wide association studies of LDL-C, HDL-C, and triglycerides in white men and women also confirmed associations at previously established loci and identified additional novel loci.9–12 One of the novel associations is equivalent to our association with LDL-C and ApoB near the SORT1 gene at 1p13.3. Among the loci in the recent reports that did not reach genome-wide significance in our study, 5 SNPs were represented on our genotyping platform explicitly (Table 5). Associations at the ANGPTL3/DOC7/ATG4C locus (rs12130333) with LDL-C and at the LCAT locus (rs255052) with HDL-C were confirmed in our data. Unexpectedly, in the WGHS sample, the reported association of rs12130333 with triglycerides is less significant than the association of this SNP with LDL-C. At the LIPG/ACAA2 locus, the reported association with HDL-C was marginal in 2-sided testing in the WGHS (P=0.05), but the same variant was highly significant for association with the related lipid fraction ApoA1 (P=1.9×10−5). Neither the reported association involving rs2228603 at the CILP2/PBX4/NCAN/SF4 locus nor that involving rs4149268 at the ABCA1 locus was significant in our data.
However, on a locus-wide basis, alternative SNPs from our data could confirm associations at 7 loci from the recent reports that did not have genome-wide significance in our primary analysis (Table 6). Thus, SNPs with nominally significant association probability values after Bonferroni correction for all locus SNPs were identified at B3GALT4, BCL7B/TBL2/MLXIPL, HMGCR, MVK/MMAB, and TRIB1 loci, as well as at the ANGPTL3/DOC7/ATG4C and LIPC/ACAA2 loci mentioned above. As with associations between rs12130333 and LDL-C or triglycerides (Table 5), the association at 4 of the 7 loci with the lipid fraction specified in the recent reports was weaker than the best locus association with a second fraction from among LDL-C, HDL-C, and triglycerides. The reported lipid fraction represented the strongest locus association in the WGHS only for HMGCR with LDL-C, BCL7B/TBL2/MLXIPL with triglycerides, and MVK/MMAB with HDL-C, suggesting possible population-dependent or environmental interaction effects on lipid metabolism at the other loci. Tables 5 and 6⇓ also provide summaries of the associations at the candidate loci with the ApoA1 and ApoB lipid fractions that were not evaluated in the recent reports.
In this study of a primary cohort with 6382 white women, we found genome-wide associations with plasma lipid fractions at 10 loci, 6 of which could be replicated in smaller cohorts totaling 970 white men and women when genotypes were available. Most of the associations are consistent with known pathways for lipid metabolism, whereas others suggest emerging pathways. Below the genome-wide significance threshold, 9 additional loci also including known and emerging links to lipid metabolism could be confirmed by replicating associations from recent genome-wide association studies that were reported while our analysis was being completed.9–12 Differences between our results and the recent reports, including the inability to confirm some associations in the WGHS, may be related to differences in the populations, including possible effects related to the exclusively female gender in the WGHS.
The genome-wide association of rs1260326 and other SNPs at the GCKR locus with triglycerides replicates recent novel findings.22 At this locus, our study also found associations with ApoA1 and ApoB fractions at the approximate P=10−6 level that extend the connection between glucose and lipid metabolism but were not detected in the previous report. The major function of the GCKR protein is posttranslastional regulation of glucokinase and thus is intimately linked to glucose metabolism. Consistent with the effects of overexpression of glucokinase in animal models,23,24 the levels of triglycerides, ApoA1, and ApoB are all increased by the minor allele of rs1260326 in the WGHS, trends that would suggest decreased cardiovascular disease risk from the ApoA1 effects but increased disease risk from the triglyceride and ApoB effects. Separate results from the WGHS described elsewhere implicate GCKR also in baseline plasma levels of C-reactive protein, a component of innate immunity, with a direction of effects suggesting decreased risk among minor allele carriers24a. Although resolving the clinical consequences of the differences in metabolic profile associated with GCKR variation remains a priority, it is noteworthy that the minor allele of rs1260326 is nearly completely linked to the minor allele of another SNP (rs780094) that may trend toward lower diabetes risk in preliminary analysis.22
Genome-wide associations of rs646776 and others at 1p13.3 with LDL-C and ApoB may be consistent with the reported roles of the SORT1 protein in vesicular transport, including interactions with lipoprotein lipase, specialized storage vesicles associated with insulin-responsive glucose transport, and a receptor-associated protein that binds LDL receptor family members.25–28 Nevertheless, more analysis is required to exclude causal roles for other genes (CELSR2, PSRC1) that map closer to the strongest associations across a neighboring recombination hotspot from SORT1. The association at 1p13.3 is remarkably strong. The primary SNP represents the most significant association for LDL-C and ApoB in the entire genome scan. Similarly, the sampled variation at the locus as a whole is second only to variation near the APOE locus for explaining variance in the lipids caused by genetics. In an unrelated study, the association with ApoB could be confirmed at just below genome-wide significance.22 In yet another separate study, a second SNP from the locus (rs599839) was associated with coronary artery disease at the genome-wide level in analysis combining 2 cohorts.29 Based on HapMap linkage disequilibrium estimates,16 the trend of the rs599839 association could be inferred to be consistent with the effects of rs646776 on LDL-C and ApoB in our study.
The sampled variation at the candidate loci accounts for only part of the heritability in lipid traits estimated in other white populations.30–34 Even after adjustment for clinical and environmental factors, 7.3%, 5.7%, and 5.0% of the variance in LDL-C, HDL-C, and triglycerides, respectively, is explained by the common variation at the candidate loci. Without adjustment, the values are slightly smaller: 6.1%, 5.0%, and 3.4%. It is possible that common variation at loci that were not thoroughly sampled by the SNP panel could account for some of the remaining variance, but it would be surprising if any of the missing SNPs were as strongly associated as the most significant variants reported here. Even tagging with linkage disequilibrium to causal SNPs in the range of r2=0.2 to 0.4, ie, considerably less than the r2=0.7 to 0.8 used to design the genotyping panel, would have been adequate to identify the genome-wide associations typified by the CETP influences on HDL-C or the SORT1 influences on ApoB in our study with at least 80% power.
Identifying the sources of this discrepancy is crucial to the larger goal of elucidating the complete set of genes and pathways contributing to lipid metabolism. In part, the remaining genetic effects may be explained by less common alleles in the candidate loci. For example, the association of LDL-C and ApoB with PCSK9 involved an SNP (rs11591147) that was included in the custom content of our SNP panel and had a minor allele frequency of 1.6%. However, a multiplicity of both common and rare variants at loci with less pronounced associations (because of either sparse coverage of the SNP panel or relatively less existing functional variation) and structural variation that may not be measured by conventional SNP genotyping35 likely contributes as well. Similarly, the present analysis did not consider interaction between genetic variation and itself or the environment, 2 potential components of heritability that remain relatively unexplored on a genome-wide basis. Capturing these alternative contributions with certainty will require the statistical power of large cohorts or combinations of smaller cohorts with precise phenotype information as exemplified by the WGHS.
Sources of Funding
This work was supported by grants from the National Heart, Lung, and Blood Institute (HL 043851) and the National Cancer Institute (CA 047988) (both Bethesda, Md), the Donald W. Reynolds Foundation (Las Vegas, Nev), the Doris Duke Charitable Foundation, the Fondation Leducq (Paris, France), and the Fonds de la Recherche en Santé du Quebec (to Dr Paré). Collaborative scientific support and funding for genotyping were provided by Amgen, Inc.
Drs Parker and Miletich are employees of Amgen, Inc. Drs Chasman, Williams, Rieder, Rotter, Nickerson, and Krauss receive support from the National Heart, Lung, and Blood Institute (HL69757; Dr Krauss, principal investigator) for genetic analysis of lipid lowering with statin therapy. Dr Krauss receives research support from Merck, Inc. Drs Ridker and Zee receive research support for genotyping from Roche Diagnostics, Inc. The other authors report no conflicts.
Lusis AJ, Fogelman AM, Fonarow GC. Genetic basis of atherosclerosis, part II: clinical implications. Circulation. 2004; 110: 2066–2071.
Simon JA, Lin F, Hulley SB, Blanche PJ, Waters D, Shiboski S, Rotter JI, Nickerson DA, Yang H, Saad M, Krauss RM. Phenotypic predictors of response to simvastatin therapy among African-Americans and Caucasians: the Cholesterol and Pharmacogenetics (CAP) Study. Am J Cardiol. 2006; 97: 843–850.
Kathiresan S, Melander O, Guiducci C, Surti A, Burtt NP, Rieder MJ, Cooper GM, Roos C, Voight BF, Havulinna AS, Wahlstrand B, Hedner T, Corella D, Tai ES, Ordovas JM, Berglund G, Vartiainen E, Jousilahti P, Hedblad B, Taskinen MR, Newton-Cheh C, Salomaa V, Peltonen L, Groop L, Altshuler DM, Orho-Melander M. Six new loci associated with blood low-density lipoprotein cholesterol, high-density lipoprotein cholesterol or triglycerides in humans. Nat Genet. 2008; 40: 189–197.
Willer CJ, Sanna S, Jackson AU, Scuteri A, Bonnycastle LL, Clarke R, Heath SC, Timpson NJ, Najjar SS, Stringham HM, Strait J, Duren WL, Maschio A, Busonero F, Mulas A, Albai G, Swift AJ, Morken MA, Narisu N, Bennett D, Parish S, Shen H, Galan P, Meneton P, Hercberg S, Zelenika D, Chen WM, Li Y, Scott LJ, Scheet PA, Sundvall J, Watanabe RM, Nagaraja R, Ebrahim S, Lawlor DA, Ben-Shlomo Y, Davey-Smith G, Shuldiner AR, Collins R, Bergman RN, Uda M, Tuomilehto J, Cao A, Collins FS, Lakatta E, Lathrop GM, Boehnke M, Schlessinger D, Mohlke KL, Abecasis GR. Newly identified loci that influence lipid concentrations and risk of coronary artery disease. Nat Genet. 2008; 40: 161–169.
Sandhu MS, Waterworth DM, Debenham SL, Wheeler E, Papadakis K, Zhao JH, Song K, Yuan X, Johnson T, Ashford S, Inouye M, Luben R, Sims M, Hadley D, McArdle W, Barter P, Kesaniemi YA, Mahley RW, McPherson R, Grundy SM, Bingham SA, Khaw KT, Loos RJ, Waeber G, Barroso I, Strachan DP, Deloukas P, Vollenweider P, Wareham NJ, Mooser V. LDL-cholesterol concentrations: a genome-wide association study. Lancet. 2008; 371: 483–491.
de Bakker PI, Burtt NP, Graham RR, Guiducci C, Yelensky R, Drake JA, Bersaglieri T, Penney KL, Butler J, Young S, Onofrio RC, Lyon HN, Stram DO, Haiman CA, Freedman ML, Zhu X, Cooper R, Groop L, Kolonel LN, Henderson BE, Daly MJ, Hirschhorn JN, Altshuler D. Transferability of tag SNPs in genetic association studies in multiple populations. Nat Genet. 2006; 38: 1298–1303.
R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing; 2007.
Spirin V, Schmidt S, Pertsemlidis A, Cooper RS, Cohen JC, Sunyaev SR. Common single-nucleotide polymorphisms act in concert to affect plasma levels of high-density lipoprotein cholesterol. Am J Hum Genet. 2007; 81.
Saxena R, Voight BF, Lyssenko V, Burtt NP, de Bakker PI, Chen H, Roix JJ, Kathiresan S, Hirschhorn JN, Daly MJ, Hughes TE, Groop L, Altshuler D, Almgren P, Florez JC, Meyer J, Ardlie K, Bengtsson Bostrom K, Isomaa B, Lettre G, Lindblad U, Lyon HN, Melander O, Newton-Cheh C, Nilsson P, Orho-Melander M, Rastam L, Speliotes EK, Taskinen MR, Tuomi T, Guiducci C, Berglund A, Carlson J, Gianniny L, Hackett R, Hall L, Holmkvist J, Laurila E, Sjogren M, Sterner M, Surti A, Svensson M, Svensson M, Tewhey R, Blumenstiel B, Parkin M, Defelice M, Barry R, Brodeur W, Camarata J, Chia N, Fava M, Gibbons J, Handsaker B, Healy C, Nguyen K, Gates C, Sougnez C, Gage D, Nizzari M, Gabriel SB, Chirn GW, Ma Q, Parikh H, Richardson D, Ricke D, Purcell S. Genome-wide association analysis identifies loci for type 2 diabetes and triglyceride levels. Science. 2007; 316: 1331–1336.
Ferre T, Pujol A, Riu E, Bosch F, Valera A. Correction of diabetic alterations by glucokinase. Proc Natl Acad Sci U S A. 1996; 93: 7225–7230.
O'Doherty RM, Lehman DL, Telemaque-Potts S, Newgard CB. Metabolic impact of glucokinase overexpression in liver: lowering of blood glucose in fed rats is accompanied by hyperlipidemia. Diabetes. 1999; 48: 2022–2027.
Ridker PM, Pare G, Parker A, Zee RY, Danik JS, Buring JE, Kwiatkowski D, Cook NR, Miletich JP, Chasman DI. Loc related to metabolic syndrome pathways including LEPR, HNF1A, IL6R, and GCKR associate with plasma c-reactive protein: the Women’s Genome Health Study. Am J Hum Genet. 2008; 82: 1185–1192.
Nielsen MS, Madsen P, Christensen EI, Nykjaer A, Gliemann J, Kasper D, Pohlmann R, Petersen CM. The sortilin cytoplasmic tail conveys Golgi-endosome transport and binds the VHS domain of the GGA2 sorting protein. EMBO J. 2001; 20: 2180–2190.
Petersen CM, Nielsen MS, Nykjaer A, Jacobsen L, Tommerup N, Rasmussen HH, Roigaard H, Gliemann J, Madsen P, Moestrup SK. Molecular identification of a novel candidate sorting receptor purified from human brain by receptor-associated protein affinity chromatography. J Biol Chem. 1997; 272: 3599–3605.
Shi J, Kandror KV. The luminal Vps10p domain of sortilin plays the predominant role in targeting to insulin-responsive Glut4-containing vesicles. J Biol Chem. 2007; 282: 9008–9016.
Samani NJ, Erdmann J, Hall AS, Hengstenberg C, Mangino M, Mayer B, Dixon RJ, Meitinger T, Braund P, Wichmann HE, Barrett JH, Konig IR, Stevens SE, Szymczak S, Tregouet DA, Iles MM, Pahlke F, Pollard H, Lieb W, Cambien F, Fischer M, Ouwehand W, Blankenberg S, Balmforth AJ, Baessler A, Ball SG, Strom TM, Braenne I, Gieger C, Deloukas P, Tobin MD, Ziegler A, Thompson JR, Schunkert H. Genomewide association analysis of coronary artery disease. N Engl J Med. 2007; 357: 443–453.
Kathiresan S, Manning AK, Demissie S, D'Agostino RB, Surti A, Guiducci C, Gianniny L, Burtt NP, Melander O, Orho-Melander M, Arnett DK, Peloso GM, Ordovas JM, Cupples LA. A genome-wide association study for blood lipid phenotypes in the Framingham Heart Study. BMC Med Genet. 2007; 8 (suppl 1): S17.
Lamon-Fava S, Jimenez D, Christian JC, Fabsitz RR, Reed T, Carmelli D, Castelli WP, Ordovas JM, Wilson PW, Schaefer EJ. The NHLBI Twin Study: heritability of apolipoprotein A-I, B, and low density lipoprotein subclasses and concordance for lipoprotein 33(a). Atherosclerosis. 1991; 91: 97–106.
Redon R, Ishikawa S, Fitch KR, Feuk L, Perry GH, Andrews TD, Fiegler H, Shapero MH, Carson AR, Chen W, Cho EK, Dallaire S, Freeman JL, Gonzalez JR, Gratacos M, Huang J, Kalaitzopoulos D, Komura D, MacDonald JR, Marshall CR, Mei R, Montgomery L, Nishimura K, Okamura K, Shen F, Somerville MJ, Tchinda J, Valsesia A, Woodwark C, Yang F, Zhang J, Zerjal T, Zhang J, Armengol L, Conrad DF, Estivill X, Tyler-Smith C, Carter NP, Aburatani H, Lee C, Jones KW, Scherer SW, Hurles ME. Global variation in copy number in the human genome. Nature. 2006; 444: 444–454.
McVean GA, Myers SR, Hunt S, Deloukas P, Bentley DR, Donnelly P. The fine-scale structure of recombination rate variation in the human genome. Science. 2004; 304: 581–584.
Knowledge of the plasma lipid fractions low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, and triglycerides is essential for managing cardiovascular disease. Historically, the genetic determinants of these lipid fractions were identified through familial patterns of inheritance of extreme dyslipidemia. These studies revealed the first genes known to underlie lipid metabolic pathways and provided crucial insights for the development of lipid-targeted therapies. Researchers had postulated the existence of additional important genes with less extreme effects on lipid levels, but detecting these genes was impossible until the recent development of technologies allowing unbiased survey of the entire genome in large populations. Here, we report a genome-wide association study for genetic determinants of low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, and triglycerides, as well as the related apolipoproteins apolipoprotein A1 and apolipoprotein B, in a large population of white women (n=6382). In addition to the historically recognized genes for lipid metabolism, we identified 2 important gene regions influencing plasma lipid levels. Variation at the gene encoding the glucokinase regulatory protein (GCKR) affected plasma triglycerides, apolipoprotein A1, and apolipoprotein B, implicating this gene in the interplay between glucose and lipid metabolism. Similarly, genetic variation near the gene encoding the SORT1 protein was very strongly associated with low-density lipoprotein cholesterol and apolipoprotein B. Although the function of this gene remains uncertain, it may be involved in intracellular transport of cell surface receptors and/or glucose-containing intracellular vesicles. In addition, secondary analysis confirmed roles in lipid metabolism for another 5 genes recently reported elsewhere. Our study thus expands the known biochemical pathways of lipid metabolism by independently identifying 2 genes and confirming others.
↵*Drs Chasman and Paré contributed equally.
Guest Editor for this article was Donna K. Arnett, PhD.
The online Data Supplement can be found with this article at http://circgenetics.ahajournals.org/cgi/content/full/1/1/21/DC1.