Diabetes is associated with an elevated risk of coronary heart disease (CHD). Previous studies have suggested that the genetic factors predisposing to excess cardiovascular risk may be different in diabetic and nondiabetic individuals.
To identify genetic determinants of CHD that are specific to patients with diabetes.
Design, Setting, and Participants
We studied 5 independent sets of CHD cases and CHD-negative controls from the Nurses’ Health Study (enrolled in 1976 and followed up through 2008), Health Professionals Follow-up Study (enrolled in 1986 and followed up through 2008), Joslin Heart Study (enrolled in 2001-2008), Gargano Heart Study (enrolled in 2001-2008), and Catanzaro Study (enrolled in 2004-2010). Included were a total of 1517 CHD cases and 2671 CHD-negative controls, all with type 2 diabetes. Results in diabetic patients were compared with those in 737 nondiabetic CHD cases and 1637 nondiabetic CHD-negative controls from the Nurses’ Health Study and Health Professionals Follow-up Study cohorts. Exposures included 2 543 016 common genetic variants occurring throughout the genome.
Main Outcomes and Measures
Coronary heart disease—defined as fatal or nonfatal myocardial infarction, coronary artery bypass grafting, percutaneous transluminal coronary angioplasty, or angiographic evidence of significant stenosis of the coronary arteries.
A variant on chromosome 1q25 (rs10911021) was consistently associated with CHD risk among diabetic participants, with risk allele frequencies of 0.733 in cases vs 0.679 in controls (odds ratio, 1.36 [95% CI, 1.22-1.51]; P = 2 × 10−8). No association between this variant and CHD was detected among nondiabetic participants, with risk allele frequencies of 0.697 in cases vs 0.696 in controls (odds ratio, 0.99 [95% CI, 0.87-1.13]; P = .89), consistent with a significant gene × diabetes interaction on CHD risk (P = 2 × 10−4). Compared with protective allele homozygotes, rs10911021 risk allele homozygotes were characterized by a 32% decrease in the expression of the neighboring glutamate-ammonia ligase (GLUL) gene in human endothelial cells (P = .0048). A decreased ratio between plasma levels of γ-glutamyl cycle intermediates pyroglutamic and glutamic acid was also shown in risk allele homozygotes (P = .029).
Conclusion and Relevance
A single-nucleotide polymorphism (rs10911021) was identified that was significantly associated with CHD among persons with diabetes but not in those without diabetes and was functionally related to glutamic acid metabolism, suggesting a mechanistic link.
The prevalence of type 2 diabetes has been steadily increasing in the United States and other countries, with the total number of affected people reaching more than 370 million globally.1 Long-term cardiovascular complications, and especially coronary heart disease (CHD), are the principal causes of morbidity and mortality among diabetic patients.2 Although mortality caused by CHD has been declining overall during the past few decades in most industrialized countries,3 the increasing prevalence of diabetes has made the number of CHD deaths attributable to this disease escalate.4,5
The role of genetic factors in modulating susceptibility to CHD has been known for many years,6 and more than 40 chromosomal loci associated with CHD have been identified to date in the general population by genome-wide association studies (GWAS).7-12 Earlier analyses have shown considerable heterogeneity in genetic effects between diabetic and nondiabetic individuals,13 probably owing to the distinct mechanisms of atherogenesis in diabetes. This has led us to hypothesize that other, as yet undiscovered loci may exist that affect CHD risk only or mostly in the presence of diabetes. Finding these genes, if they exist, may point to atherogenic pathways that are specifically activated by the diabetic milieu and as such could be the target of new interventions aimed at preventing or treating CHD specifically among diabetic patients.
In this study, we performed a GWAS of CHD targeted to type 2 diabetic participants to identify genetic determinants of CHD that are specific to diabetic patients.
Detailed information on the study populations is provided in the eMethods in the Supplement. Briefly, stage 1 included diabetic patients from the Nurses’ Health Study (NHS)14 and the Health Professionals Follow-up Study (HPFS)15 (eMethods in the Supplement). CHD cases were defined as those with incident cases after the diagnosis of type 2 diabetes to the end of 2008; controls were participants free of CHD events in the specified period. These studies were approved by the Human Research Committee at the Brigham and Women’s Hospital, Boston, and all participants provided written informed consent.
Stage 2 included diabetic CHD cases and CHD-negative controls from the Joslin Heart Study16 (eMethods in the Supplement). The study protocol and informed consent procedures were approved by the Joslin Committee on Human Studies and the BIDMC Committee on Clinical Investigations. All participants gave written informed consent.
Stage 3 included diabetic CHD cases and CHD-negative controls from the Gargano Heart Study (cross-sectional design)17 and the Catanzaro Study18 ( eMethods in the Supplement). The study protocol and informed consent procedures were approved by the local human subject committees. All participants gave written informed consent.
To compare the association between rs10911021 (http://www.ncbi.nlm.nih.gov/SNP/) and CHD risk in nondiabetic vs diabetic participants, we analyzed a separate nondiabetic CHD case-control GWAS, which included incident CHD cases and non-CHD controls from the NHS and HPFS cohorts,19 after excluding individuals affected by diabetes (eMethods in the Supplement). We also obtained data on rs10911021 from the Coronary Artery Disease Genome-wide Replication and Meta-analysis (CARDIoGRAM).11 To explore whether this variant may contribute to CHD through alterations of insulin sensitivity, we interrogated the Meta-analyses of Glucose and Insulin-Related Traits Consortium (MAGIC) database, a meta-analysis of GWAS data for metabolic traits.20,21 We also interrogated the Diabetes Genetics Replication and Meta-analysis (DIAGRAM) database22 to explore whether this variant might have pleiotropic associations with both type 2 diabetes and CHD.
Single-nucleotide polymorphism (SNP) genotyping and imputation for stage 1 have been described in detail elsewhere23 and in the eMethods in the Supplement. Briefly, samples were genotyped with the Affymetrix Genome-Wide Human 6.0 array. A total of 704 409 and 706 040 SNPs passed quality control in the NHS and HPFS sets, respectively, and were used to impute the genotypes of other SNPs by means of MACH software.24 In stages 2 and 3, SNPs were genotyped by the Joslin DERC Genetics Core by means of TaqMan assays implemented on an ABI PRISM 7700 HT Sequence Detection System (Applied Biosystems).
Gene Expression in Endothelial Cell Lines
To investigate whether the association between rs10911021 and CHD risk could be mediated by gene expression changes, we measured messenger RNA levels of 8 neighboring genes (4 on the centromeric and 4 on the telomeric side) (eFigure 1 in the Supplement) in 124 human umbilical vein endothelial cell lines from nondiabetic mothers. Umbilical cords were obtained from randomly selected healthy mothers who delivered at the Pescara Town Hospital, Italy, and gave written consent to this procedure. Primary human umbilical vein endothelial cell lines were established from the umbilical cords and cultured as described by Gorfien et al.25 Cell lines were typed for the rs10911021 SNP with a TaqMan allelic discrimination assay (Applied Biosystems). Gene expression was assayed by means of real-time quantitative polymerase chain reaction–based TaqMan low-density arrays. Eight target genes neighboring rs10911021, along with a housekeeping gene (GUSB; NCBI Entrez Gene NG_016197.1) as endogenous control, were included in the array.
To obtain further insights into the potential functional influence of rs10911021, we measured plasma glutamine and glutamic acid, as well as the ratio between pyroglutamic acid (the immediate precursor of glutamic acid in the γ-glutamyl cycle) and glutamic acid in 100 diabetic patients from the Joslin Heart Study (50 rs10911021 risk allele C homozygotes and 50 T homozygotes). Plasma concentrations of glutamic acid and glutamine were assessed at the University of Trieste, Italy, by gas chromatography–mass spectrometry, using the internal standard technique, as previously described.26 Known amounts of l-[15N]-glutamic acid and l-[15N]-glutamine (Cambridge Isotope Laboratories) were added as internal standards to a known volume of plasma. Silylated derivatives were measured under electron-impact ionization by selective ion monitoring at a nominal mass-to-charge ratio (m/z) of 432/433 for glutamic acid and 431/432 for glutamine. The pyroglutamic acid derivative was also monitored at a nominal m/z of 300. The pyroglutamic/glutamic acid peak area ratio was determined at nominal m/z of 300 and 432, respectively.
Analyses were carried out in 3 stages. In stage 1, 2 separate GWAS for CHD across 2 543 016 genotyped or imputed SNPs (with imputed SNPs expressed as allele dosage) were performed in the NHS and HPFS sets by means of logistic regression under an additive genetic model with the ProbABEL package.27 The genomic inflation factor λ was estimated from the median χ2 statistic. To control for potential confounding by population stratification, we performed further analyses by including the top principal components of genetic variation chosen for each study in the models (top 3 and 4 eigenvectors for NHS and HPFS, respectively).
Meta-analysis of the 2 GWAS scans was conducted by combining study-specific β-estimates from genome-wide associations, using inverse variance weights under a fixed-effect model in METAL software.28 Variants yielding a P < 1 × 10−4 in stage 1 were carried forward to stage 2, and those yielding a P < 1 × 10−4 in stages 1 and 2 combined were carried forward to stage 3. In stages 2 and 3, sex-adjusted odds ratios (ORs) and their 95% CIs were estimated for each SNP and in each study by means of logistic regression according to an additive model. Associations across stages 1 and 2 studies and across all the studies in stages 1, 2, and 3 were summarized by meta-analyses with Stata version 7.0. The presence of heterogeneity among the 3 studies was tested by means of a χ2 statistics. Because this test was not significant for any of the SNPs, we calculated summary ORs according to a fixed-effect model, ie, by averaging the natural logarithms of the ORs from individual studies, weighted by the inverses of their variances.29
The association between rs10911021 and CHD among nondiabetic participants from the NHS and HPFS cohorts was evaluated as described above for diabetic participants. The interaction between rs10911021 and diabetes on CHD risk was evaluated by adding the rs10911021 × diabetes cross-product to a logistic regression analysis of the combined diabetic and nondiabetic NHS/HPFS sets. The same approach was used to evaluate the interaction between rs10911021 and 36 established type 2–predisposing variants considered in the article by Qi et al.30
Power for the main SNP effects was estimated with the software CaTS,31 assuming a risk allele frequency of 0.30. The GWAS of type 2 diabetes participants had 80% power (α = 5 × 10−8) to detect associations with CHD, with summary ORs across the 3 stages as low as 1.35. The study of rs10911021 in nondiabetic participants had greater than 99% power (α = .05) to detect an association with CHD, with an OR similar to that observed among diabetic participants (OR = 1.36), and 80% power to detect an association with an OR as low as 1.19.
Change in threshold cycle (ΔCt) values were derived from the threshold cycle (Ct) data for each target gene with the equation ΔCt = Ct (target gene) – Ct (endogenous control). ΔΔCt values were then calculated for each sample and gene as the difference between the ΔCt and the mean ΔCt among rs10911021 T/T homozygotes. For each target gene, the association between rs10911021 and ΔΔCt was evaluated by linear regression with an additive genetic model.
The association between plasma amino acid levels and rs10911021 genotype or CHD case-control status was evaluated by means of linear regression models, with the amino acid levels as the dependent variables and age, sex, γ-glutamyltransferase levels, rs10911021 genotype, and CHD case-control status as the independent variables. Glutamic acid and the pyroglutamic/glutamic ratio were evaluated after log transformation because of their non-normal distributions.
For GWAS analyses, 2-sided P values smaller than 5 × 10−8 were considered significant; for all other analyses, 2-sided P values smaller than .05 were considered significant.
GWAS and Validations Among Diabetic Participants
A total of 1517 CHD cases and 2671 CHD-negative controls, all with type 2 diabetes, were included in the 3-stage genome-wide analysis: 350 cases and 976 controls from the NHS and 319 cases and 665 controls from the HPFS (stage 1), 420 cases and 431 controls from the Joslin Heart Study (stage 2), 314 cases and 384 controls from the Gargano Heart Study–cross-sectional design, and 114 cases and 215 controls from the Catanzaro Study (stage 3) (eMethods in the Supplement). The clinical characteristics of the case-control sets analyzed at each stage are summarized in Table 1.
Of the 2 543 016 genetic variants that were tested for association with CHD in stage 1, 26 met the criterion for promotion to stage 2 (P < .001 in stage 1) and 3 of these further met the criterion for promotion to stage 3 (P < .001 in stage 1 + stage 2). Detailed data on the variants associated with CHD at each stage can be found in eTable 1 and eFigures 2 and 3 in the Supplement. Of the 3 variants that were promoted to stage 3, 1 (rs10911021) showed an association with CHD that was nominally significant at each stage and exceeded genome-wide significance in the 3 stages combined (P = 2.0 × 10−8) (Table 2 and eTable 1 in the Supplement).
In a meta-analysis of the 5 case-control sets, the summary OR of CHD for each copy of the risk allele was 1.36 (95% CI, 1.22-1.51; risk allele frequencies, 0.697 in cases vs 0.696 in controls), with no evidence of heterogeneity across studies (I2 = 0%; P = .82) (Table 2). The other 2 variants promoted to stage 3 (rs9361923 on chromosome 6 and rs7542837 on chromosome 1) had summary P values across the 5 sets in the 10−4 range (eTable 1 in the Supplement). None of the loci previously associated with CHD in the general population were among the genetic variants promoted to stages 2 and 3, although 3 of them reached nominal significance at stage 1 (eTable 2 in the Supplement).
Interaction With Diabetes Status
No association between rs10911021 and CHD was found among 737 nondiabetic CHD cases and 1637 nondiabetic CHD-negative controls from the NHS and HPFS cohorts (eTable 3 in the Supplement). The OR among these nondiabetic individuals was not significantly different from 1 (OR = 0.99; 95% CI, 0.87-1.13; P = .89; risk allele frequencies, 0.697 in cases vs 0.696 in controls) while being significantly different from the OR in diabetic participants (OR = 1.36; 95% CI, 1.22-1.51; P for diabetes × genetic variant interaction, 2.6 × 10−4). Among the NHS and HPFS diabetic participants, no significant interaction on CHD risk was observed between rs10911021 and established type 2 diabetes–predisposing variants, considered individually or in combination as a genetic predisposition score30 (all P > .05).
In CARDIoGRAM, which comprises 22 233 CHD cases and 64 762 controls from the general population, rs10911021 showed a nominally significant association with CHD in the same direction as among the diabetic participants in our study (OR = 1.04; 95% CI, 1.01-1.07; P = .011) but was significantly weaker (I2 = 96%; P for heterogeneity = 2.2 × 10−6; fixed-effect model). If we assume a 15% average prevalence of diabetes—an estimate based on the CARDIoGRAM studies for which data on the occurrence of diabetes are available32-35—the OR observed in the CARDIoGRAM population corresponded to the weighted average of the ORs observed in our study in diabetic and nondiabetic participants (OR = 1.36 and 0.99, respectively). No other variant neighboring rs10911021 showed associations at genome-wide significance level in this data set (eFigure 4 in the Supplement).
Genotype Association With the Expression of Neighboring Genes
Variant rs10911021 is located between 2 genes, ZNF648 (≈51 kb in centromeric direction; NCBI Entrez Gene 127665) and GLUL (≈270 kb in telomeric direction; NCBI Entrez Gene NG_013347.1), and neighbors several other genes (eFigure 1 in the Supplement). No missense variants in linkage disequilibrium with rs10911021 were identified in the HapMap or the 1000 Genome Projects databases, suggesting an effect on gene regulation as the mechanism underlying the observed association with CHD. In support of this hypothesis, rs10911021 is listed in the Regulome DB as occurring in an E-box binding site for basic helix-loop-helix transcription factors, and ENCODE data indicate that a variant in linkage disequilibrium with this variant (rs7517310; r2 = 0.72 in the HapMap database) is placed in a high DNase I sensitivity cluster binding to the RE1-silencing transcription factor in a variety of cell types. As shown in Table 3, the expression of GLUL, the closest gene in telomeric direction, was significantly associated with rs10911021 in endothelial cells, being 32% lower in risk allele (C/C) homozygotes compared with protective allele (T/T) homozygotes, with heterozygotes having intermediate levels (P for trend = .0048). ZNF648, the closest gene on the 5′ side, was not expressed in endothelial cells and none of the other neighboring genes were significantly associated with rs10911021.
Association With Plasma Markers of Glutamic Acid Metabolism and the γ-Glutamyl Cycle
In a sample of 100 Joslin Heart Study participants, no significant differences in plasma glutamic acid or glutamine (the substrate and the product, respectively, of the enzyme encoded by GLUL) were observed between risk allele C homozygotes and allele T homozygotes (Table 4). However, the ratio between plasma pyroglutamic acid (the immediate precursor of glutamic acid in the γ-glutamyl cycle) and glutamic acid was significantly lower in C/C as compared to T/T carriers (P = .03) (Table 4). In this sample, the pyroglutamic-to-glutamic ratio was also significantly lower in the 44 participants who had developed CHD (median = 0.79; interquartile range 0.62-0.97) than in the 56 who were CHD negative (median, 0.92; interquartile range, 0.78-1.14) (P = .02). The OR of CHD for the rs10911021 C/C genotype in this subsample decreased from 1.83 to 1.39 (≈50% reduction in the log scale) after adjustment for the pyroglutamic-to-glutamic acid ratio, suggesting that the effect of this locus on CHD was at least in part mediated by its effect on this parameter.
Association With Other Cardiovascular Risk Factors
No significant association between rs10911021 and serum fasting insulin, homeostatic assessment model insulin-resistance index, or 2-hour glucose during an oral glucose tolerance test was found in the MAGIC database including data on more than 35 000 nondiabetic individuals. Similarly, no significant association was found with type 2 diabetes in the DIAGRAM database (OR = 1.01; 95% CI, 0.97-1.04; P = .76).
In this study, we have identified a previously unknown genetic locus associated with increased CHD risk among type 2 diabetic patients. The locus is in the region of the GLUL gene on chromosome 1q25 and may affect CHD risk by reducing the expression of this gene and affecting glutamate and glutamine metabolism in endothelial cells. This genetic variant appeared to be specifically associated with CHD in the diabetic population and showed a significant gene-by-diabetes synergism on CHD risk.
Several pieces of evidence suggest that these findings are unlikely to be due to chance. First, the P value for the association between this locus and CHD in type 2 diabetes participants meets genome-wide significance (P < 5 × 10−8), namely, withstands adjustment for the large number of comparisons that are made in a genome-wide analysis. Second, the association was consistent across multiple samples of type 2 diabetic participants of different ethnic and geographic origin, reaching nominal significance in 4 of the 5 sets that were considered. Third, the difference in ORs between diabetic and nondiabetic participants was supported by a robust P value for interaction. Finally, an association between this locus and CHD was also found in a large study of the general population (CARDIoGRAM) with a magnitude similar to what one would expect according to the effects detected in our study in diabetic and nondiabetic participants and the prevalence of diabetes in CARDIoGRAM.
GLUL, the gene whose expression is decreased in risk allele carriers, encodes glutamate-ammonia ligase (also known as glutamine synthase), which catalyzes the conversion of glutamic acid and ammonia into glutamine.36 Both amino acids play important roles in human physiology. Glutamic acid is a key intermediate of several metabolic pathways, most notably of the γ-glutamyl cycle, through which the antioxidant glutathione is generated37; glutamine is involved in the regulation of cell proliferation, inhibition of apoptosis, and cell signaling.38 Evidence from experimental and human studies points to glutamine/glutamic acid metabolism as contributing to the regulation of insulin secretion and glucose metabolism. In islets, glutamine enhances both mitochondrial metabolism and insulin secretion.39 In diabetic patients, it was found that glutamine reduced glucose excursions when administered before oral glucose40 and effectively increased circulating incretin and insulin concentrations.41 Several clinical trials also suggest cardioprotective effects of glutamine used parenterally and enterally.42,43 In epidemiologic studies, abnormal metabolism of these amino acids has been shown to be related to insulin resistance, type 2 diabetes, and cardiovascular disorders.44-46
The mechanisms through which alterations of glutamate and glutamine metabolism, such as those that one would expect from the reduced GLUL expression observed in risk allele carriers, may lead to increased CHD risk are unclear. The newly identified CHD risk variant was not associated with risk of type 2 diabetes in DIAGRAM, suggesting that the pathways underlying the association with CHD are distinct from those involved in the etiology of type 2 diabetes. Similarly, the absence of association between the risk variant and serum fasting insulin, homeostatic assessment model insulin-resistance index, or 2-hour glucose during a glucose tolerance test in the MAGIC database seems to exclude insulin resistance as the underlying mechanism.
Rather, our finding of association between the risk variant and a lower pyroglutamic-to-glutamic acid ratio in plasma and the fact that the association between risk allele and CHD was attenuated after adjustment for this variable suggest an impairment of the γ-glutamyl cycle, of which pyroglutamic acid is an intermediate, as a possible mechanism. Such alteration might increase CHD risk by limiting the availability of the natural antioxidant glutathione, compounding the known negative effect of diabetes on this metabolite47 and potentially explaining the fact that this genetic effect can be observed only among diabetic participants. Consistent with this hypothesis, an association between rs10911021 and pyroglutamine (expressed as the ratio with the fatty acid sebacate) is also found in the KORA/Twins UK metabolomic databases (P = .00096 in KORA; http://metabolomics.helmholtz-muenchen.de/gwa).48 However, additional contributions by pathways that are not directly related to glutamate and glutamine may also be present because other metabolites implicated in vascular biology and atherogenesis, such as the long-chain ω3-polyunsaturated fatty acid eicosapentaenoate (20:5n3) and a variety of lysophospholipids,49,50 are associated with rs10911021 in those same databases.
Further studies are needed to dissect the mechanisms linking this locus to the development and progression of atherosclerosis in diabetes. As part of these efforts, it would be useful to extend the study to type 1 diabetes because this may provide clues about whether the gene × diabetes interaction involves hyperglycemia or instead concerns factors that are specific to type 2 diabetes, such as insulin resistance or some of the genes predisposing to this form of diabetes. However, the lack of interaction in our study between rs10911021 and genetic variants predisposing to type 2 diabetes makes the latter hypothesis unlikely.
Our study has several strengths, namely, the validation design with 5 independent cohorts of diabetic patients, a rigorous definition of CHD, and a sample size that was adequate for the detection of additive genetic effects of the magnitude reported. Nonetheless, some limitations should be acknowledged. First, although our study was powered to detect major genetic effects such as that described in this article, larger studies would be necessary to detect loci having smaller but still relevant effects on CHD risk in diabetes. In this context, the use of analytic methods based on biological pathways such as gene ontology51 might lead to the identification of additional genetic determinants of CHD in diabetic patients and provide further insights on the links between diabetes and atherogenesis. Second, our study was restricted to non-Hispanic whites, and whether these findings can be generalized to other races remains to be determined. Also, according to the known differences in linkage disequilibrium patterns among races, different genetic markers may be more effective in capturing the predisposing effect of the locus described in this article in other racial groups. Third, although the achieved level of statistical significance meets genome-wide significance, it still corresponds to a 5% probability of a false-positive result. Although the likelihood of such an event is reduced by the validation of the association in CARDIoGRAM, further studies are needed before this CHD locus can be considered as fully validated.
In summary, through a 3-stage GWAS in 4188 type 2 diabetic patients, we have identified a novel susceptibility locus for CHD in the region of the GLUL gene. This locus appears to be associated with CHD very weakly or not at all among nondiabetic participants, consistent with a gene × diabetes synergism. Preliminary evidence suggests that this locus may modulate CHD risk by affecting glutamate/glutamine metabolism and the activity of the γ-glutamyl cycle, but further studies are needed to fully understand the biological mechanisms linking it to CHD in diabetes.
Corresponding Authors: Lu Qi, MD, PhD, Department of Nutrition, Harvard School of Public Health, 665 Huntington Ave, Boston, MA 02115 (firstname.lastname@example.org) and Alessandro Doria, MD, PhD, MPH, Research Division, Joslin Diabetes Center, One Joslin Place, Boston, MA 02215 (email@example.com).
Author Contributions: Drs Q. Qi and Doria had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Trischitta, Hu, Doria, L. Qi.
Acquisition of data: Q. Qi, Prudente, Mendonca, Andreozzi, di Pietro, Novelli, Mannino, Formoso, Hauser, Muehlschlegel, Niewczas, Krolewski, Pandolfi, Rimm, Sesti, Trischitta, Doria, L. Qi.
Analysis and interpretation of data: Q. Qi, Prudente, Andreozzi, Sturma, Novelli, Mannino, Gervino, Hauser, Niewczas, Krolewski, Biolo, Sesti, Trischitta, Hu, Doria, L. Qi.
Drafting of the manuscript: Q. Qi, Trischitta, Doria, L. Qi.
Critical revision of the manuscript for important intellectual content: Q. Qi, Prudente, Mendonca, Andreozzi, di Pietro, Sturma, Novelli, Mannino, Formoso, Gervino, Hauser, Muehlschlegel, Niewczas, Krolewski, Biolo, Pandolfi, Rimm, Sesti, Hu, L. Qi.
Statistical analysis: Q. Qi, Niewczas, Doria, L. Qi.
Obtained funding: Rimm, Trischitta, Doria, L. Qi.
Administrative, technical, or material support: Prudente, Mendonca, Andreozzi, Sturma, Novelli, Mannino, Gervino, Hauser, Muehlschlegel, Sesti, Hu, L. Qi.
Study supervision: Krolewski, Pandolfi, Trischitta, Doria, L. Qi.
Drs L. Qi and Doria contributed equally to this work and are co–senior authors.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr L. Qi reported serving as a consultant to Genovive. Dr Hauser reported serving as a consultant to Astellas and the Harvard Clinical Research Institute. Dr Krolewski reported receiving honoraria from Merck, Eli Lilly, and Boehringer Ingelheim. Dr Biolo reported receiving payment for lectures from Baxter and Fresenius Kabi. Dr Sesti reported serving as a consultant to Novo Nordisk, Merck, Servier, Boehringer Ingelheim, and sanofi and serving on the speakers bureau for Novartis, Novo Nordisk, Bristol-Myers Squibb, and Laboratori Guidotti. Dr Hu reported serving as a consultant to Novo Nordisk and his institution received a grant from Merck. Dr Doria reported that his institution received a grant from sanofi-aventis.
Funding/Support: This study was supported by grants HL071981, DK091718, HL073168, DK046200 (Boston Obesity Nutrition Research Center), and DK36836 (Genetics Core of the Diabetes Research Center at the Joslin Diabetes Center) from the National Institutes of Health, an American Heart Association Scientist Development Award (0730094N), a grant from the Italian Ministry of Health (“Ricerca Corrente 2011 e 2012”), and a grant from Fondazione Roma (“Sostegno alla ricerca scientifica biomedica 2008”). A portion of this work was conducted in a facility constructed with support from the National Institutes of Health Research Facilities Improvement Program (RR10600-01, CA62528-01, RR14514-01) from the National Center for Research Resources.
Role of the Sponsor: The funding agencies had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or decision to submit the manuscript for publication.
Additional Contributions: We thank all the participants of the study and CARDIoGRAM for providing data on the association between variants in the rs10911021 region and CHD in the general population. Data on glycemic traits were contributed by MAGIC investigators and downloaded from http://www.magicinvestigators.org. Data on type 2 diabetes were contributed by DIAGRAM investigators and downloaded from http://diagram-consortium.org.
M. Type 2 diabetes across generations: from pathophysiology to prevention and management. Lancet
. 2011;378(9786):169-181.PubMedGoogle ScholarCrossref
S. Proportion of the decline in cardiovascular mortality disease due to prevention versus treatment: public health versus clinical care. Annu Rev Public Health
. 2011;32:5-22.PubMedGoogle ScholarCrossref
S. Explaining the decline in coronary heart disease mortality in England and Wales between 1981 and 2000. Circulation
. 2004;109(9):1101-1107.PubMedGoogle ScholarCrossref
et al; Italian Atherosclerosis, Thrombosis, and Vascular Biology Working Group; Myocardial Infarction Genetics Consortium; Wellcome Trust Case Control Consortium; Cardiogenics Consortium. New susceptibility locus for coronary artery disease on chromosome 3q22.3. Nat Genet
. 2009;41(3):280-282.PubMedGoogle ScholarCrossref
et al. A common variant on chromosome 9p21 affects the risk of myocardial infarction. Science
. 2007;316(5830):1491-1493.PubMedGoogle ScholarCrossref
et al. A common allele on chromosome 9 associated with coronary heart disease. Science
. 2007;316(5830):1488-1491.PubMedGoogle ScholarCrossref
et al; WTCCC and the Cardiogenics Consortium. Genomewide association analysis of coronary artery disease. N Engl J Med
. 2007;357(5):443-453.PubMedGoogle ScholarCrossref
et al; Cardiogenics; CARDIoGRAM Consortium. Large-scale association analysis identifies 13 new susceptibility loci for coronary artery disease. Nat Genet
. 2011;43(4):333-338.PubMedGoogle ScholarCrossref
et al; Wellcome Trust Case Control Consortium; Cardiogenics Consortium. Genome-wide haplotype association study identifies the SLC22A3-LPAL2-LPA gene cluster as a risk locus for coronary artery disease. Nat Genet
. 2009;41(3):283-285.PubMedGoogle ScholarCrossref
et al. Genetic susceptibility to coronary heart disease in type 2 diabetes. J Am Coll Cardiol
. 2011;58(25):2675-2682.PubMedGoogle ScholarCrossref
SE. The Nurses’ Health Study: 20-year contribution to the understanding of health among women. J Womens Health
. 1997;6(1):49-62.PubMedGoogle ScholarCrossref
et al. Prospective study of alcohol consumption and risk of coronary disease in men. Lancet
. 1991;338(8765):464-468.PubMedGoogle ScholarCrossref
et al. Interaction between poor glycemic control and 9p21 locus on risk of coronary artery disease in type 2 diabetes. JAMA
. 2008;300(20):2389-2397.PubMedGoogle ScholarCrossref
et al. The K121Q polymorphism of the ENPP1/PC-1 gene is associated with insulin resistance/atherogenic phenotypes, including earlier onset of type 2 diabetes and myocardial infarction. Diabetes
. 2005;54(10):3021-3025.PubMedGoogle ScholarCrossref
et al. The type 2 diabetes and insulin-resistance locus near IRS1 is a determinant of HDL cholesterol and triglycerides levels among diabetic subjects. Atherosclerosis
. 2011;216(1):157-160.PubMedGoogle ScholarCrossref
et al. Protein interaction-based genome-wide analysis of incident coronary heart disease. Circ Cardiovasc Genet
. 2011;4(5):549-556.PubMedGoogle ScholarCrossref
et al; DIAGRAM Consortium; GIANT Consortium; Global BPgen Consortium; Anders Hamsten on Behalf of Procardis Consortium; MAGIC Investigators. New genetic loci implicated in fasting glucose homeostasis and their impact on type 2 diabetes risk. Nat Genet
. 2010;42(2):105-116.PubMedGoogle ScholarCrossref
et al; GIANT Consortium; MAGIC Investigators. Genetic variation in GIPR influences the glucose and insulin responses to an oral glucose challenge. Nat Genet
. 2010;42(2):142-148.PubMedGoogle ScholarCrossref
et al; Wellcome Trust Case Control Consortium; Meta-Analyses of Glucose and Insulin-Related Traits Consortium (MAGIC) Investigators; Genetic Investigation of Anthropometric Traits (GIANT) Consortium; Asian Genetic Epidemiology Network–Type 2 Diabetes (AGEN-T2D) Consortium; South Asian Type 2 Diabetes (SAT2D) Consortium; Diabetes Genetics Replication and Meta-analysis (DIAGRAM) Consortium. Large-scale association analysis provides insights into the genetic architecture and pathophysiology of type 2 diabetes. Nat Genet
. 2012;44(9):981-990.PubMedGoogle ScholarCrossref
et al; Meta-analysis of Glucose and Insulin-Related Traits Consortium (MAGIC); Diabetes Genetics Replication and Meta-analysis (DIAGRAM) Consortium. Genetic variants at 2q24 are associated with susceptibility to type 2 diabetes. Hum Mol Genet
. 2010;19(13):2706-2715.PubMedGoogle ScholarCrossref
GR. MaCH: using sequence and genotype data to estimate haplotypes and unobserved genotypes. Genet Epidemiol
. 2010;34(8):816-834.PubMedGoogle ScholarCrossref
S. Growth and physiological functions of vascular endothelial cells in a new serum-free medium (SFM). Exp Cell Res
. 1993;206(2):291-301.PubMedGoogle ScholarCrossref
et al. Positive energy balance is associated with accelerated muscle atrophy and increased erythrocyte glutathione turnover during 5 wk of bed rest. Am J Clin Nutr
. 2008;88(4):950-958.PubMedGoogle Scholar
MV, van Duijn
CM. ProbABEL package for genome-wide association analysis of imputed data. BMC Bioinformatics
. 2010;11:134.PubMedGoogle ScholarCrossref
GR. METAL: fast and efficient meta-analysis of genomewide association scans. Bioinformatics
. 2010;26(17):2190-2191.PubMedGoogle ScholarCrossref
L. Diabetes genetic predisposition score and cardiovascular complications among patients with type 2 diabetes. Diabetes Care
. 2013;36(3):737-739.PubMedGoogle ScholarCrossref
M. Joint analysis is more efficient than replication-based analysis for two-stage genome-wide association studies. Nat Genet
. 2006;38(2):209-213.PubMedGoogle ScholarCrossref
et al; Myocardial Infarction Genetics Consortium; Wellcome Trust Case Control Consortium. Genome-wide association of early-onset myocardial infarction with single nucleotide polymorphisms and copy number variants. Nat Genet
. 2009;41(3):334-341.PubMedGoogle ScholarCrossref
et al; Myocardial Infarction Genetics Consortium; Wellcome Trust Case Control Consortium; Cardiogenics. Lack of association between the Trp719Arg polymorphism in kinesin-like protein-6 and coronary artery disease in 19 case-control studies. J Am Coll Cardiol
. 2010;56(19):1552-1563.PubMedGoogle ScholarCrossref
et al; Myocardial Infarction Genetics Consortium; Wellcome Trust Case Control Consortium. Identification of ADAMTS7 as a novel locus for coronary atherosclerosis and association of ABO with myocardial infarction in the presence of coronary atherosclerosis. Lancet
. 2011;377(9763):383-392.PubMedGoogle ScholarCrossref
et al. Association of genome-wide variation with the risk of incident heart failure in adults of European and African ancestry: a prospective meta-analysis from the Cohorts for Heart and Aging Research in Genomic Epidemiology (CHARGE) consortium. Circ Cardiovasc Genet
. 2010;3(3):256-266.PubMedGoogle ScholarCrossref
HA. Metabolism of amino-acids: the synthesis of glutamine from glutamic acid and ammonia, and the enzymic hydrolysis of glutamine in animal tissues. Biochem J
. 1935;29(8):1951-1969.PubMedGoogle Scholar
SS. Glutathione and related gamma-glutamyl compounds: biosynthesis and utilization. Annu Rev Biochem
. 1976;45:559-604.PubMedGoogle ScholarCrossref
T. Q’s next: the diverse functions of glutamine in metabolism, cell biology and cancer. Oncogene
. 2010;29(3):313-324.PubMedGoogle ScholarCrossref
et al. Insulin secretion profiles are modified by overexpression of glutamate dehydrogenase in pancreatic islets. Diabetologia
. 2004;47(2):266-276.PubMedGoogle ScholarCrossref
et al. Glutamine reduces postprandial glycemia and augments the glucagon-like peptide-1 response in type 2 diabetes patients. J Nutr
. 2011;141(7):1233-1238.PubMedGoogle ScholarCrossref
et al. Oral glutamine increases circulating glucagon-like peptide 1, glucagon, and insulin concentrations in lean, obese, and type 2 diabetic subjects. Am J Clin Nutr
. 2009;89(1):106-113.PubMedGoogle ScholarCrossref
et al. Glutamine is cardioprotective in patients with ischemic heart disease following cardiopulmonary bypass. Heart Surg Forum
. 2011;14(6):E384-E388.PubMedGoogle ScholarCrossref
et al. Pharmacologically dosed oral glutamine reduces myocardial injury in patients undergoing cardiac surgery: a randomized pilot feasibility trial. JPEN J Parenter Enteral Nutr
. 2012;36(5):556-561.PubMedGoogle ScholarCrossref
et al. Metabolite profiling identifies pathways associated with metabolic risk in humans. Circulation
. 2012;125(18):2222-2231.PubMedGoogle ScholarCrossref
et al. Metabolic effects of glutamine on the heart: anaplerosis versus the hexosamine biosynthetic pathway. J Mol Cell Cardiol
. 2013;55:92-100.PubMedGoogle ScholarCrossref
M. Nutritional regulation of insulin secretion: implications for diabetes. Clin Biochem Rev
. 2012;33(2):35-47.PubMedGoogle Scholar
et al. Weakened cellular scavenging activity against oxidative stress in diabetes mellitus. Diabetologia
. 1995;38(2):201-210.PubMedGoogle ScholarCrossref
et al; CARDIoGRAM. Human metabolic individuality in biomedical and pharmaceutical research. Nature
. 2011;477(7362):54-60.PubMedGoogle ScholarCrossref
C. n-3 Fatty acids and the prevention of coronary atherosclerosis. Am J Clin Nutr
. 2000;71(1)(suppl):224S-227S.PubMedGoogle Scholar
K. Metabolism and atherogenic disease association of lysophosphatidylcholine. Atherosclerosis
. 2010;208(1):10-18.PubMedGoogle ScholarCrossref
et al; Wellcome Trust Case-Control Consortium. Gene ontology analysis of GWA study data sets provides insights into the biology of bipolar disorder. Am J Hum Genet
. 2009;85(1):13-24.PubMedGoogle ScholarCrossref