Skip to main content
Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Large Scale Association Analysis Identifies Three Susceptibility Loci for Coronary Artery Disease

  • Stephanie Saade ,

    Contributed equally to this work with: Stephanie Saade, Jean-Baptiste Cazier, Michella Ghassibe-Sabbagh

    Affiliation School of Medicine, Lebanese American University, Beirut, Lebanon

  • Jean-Baptiste Cazier ,

    Contributed equally to this work with: Stephanie Saade, Jean-Baptiste Cazier, Michella Ghassibe-Sabbagh

    Affiliation The Wellcome Trust Centre for Human Genetics, University of Oxford, Headington, Oxford, United Kingdom

  • Michella Ghassibe-Sabbagh ,

    Contributed equally to this work with: Stephanie Saade, Jean-Baptiste Cazier, Michella Ghassibe-Sabbagh

    Affiliation School of Medicine, Lebanese American University, Beirut, Lebanon

  • Sonia Youhanna,

    Affiliation School of Medicine, Lebanese American University, Beirut, Lebanon

  • Danielle A. Badro,

    Affiliation School of Medicine, Lebanese American University, Beirut, Lebanon

  • Yoichiro Kamatani,

    Affiliation Centre d'Etude du Polymorphisme Humain (CEPH), Paris, France

  • Jörg Hager,

    Affiliation CEA-Genomics Institute, Centre National de Génotypage, Bât G2, Evry, France

  • Joumana S. Yeretzian,

    Affiliation School of Medicine, Lebanese American University, Beirut, Lebanon

  • Georges El-Khazen,

    Affiliation Brain Mind Institute, Ecole Polytechnique Fédérale de Lausanne, Lausanne, Switzerland

  • Marc Haber,

    Affiliation School of Medicine, Lebanese American University, Beirut, Lebanon

  • Angelique K. Salloum,

    Affiliation School of Medicine, Lebanese American University, Beirut, Lebanon

  • Bouchra Douaihy,

    Affiliation School of Medicine, Lebanese American University, Beirut, Lebanon

  • Raed Othman,

    Affiliation Division of Cardiology, Rafic Hariri University Hospital, Beirut, Lebanon

  • Nabil Shasha,

    Affiliation Division of Cardiology, Rafic Hariri University Hospital, Beirut, Lebanon

  • Samer Kabbani,

    Affiliation Division of Cardiology, Rafic Hariri University Hospital, Beirut, Lebanon

  • Hamid El Bayeh,

    Affiliation School of Medicine, Lebanese American University, Beirut, Lebanon

  • Elie Chammas,

    Affiliation School of Medicine, Lebanese American University, Beirut, Lebanon

  • Martin Farrall,

    Affiliation The Wellcome Trust Centre for Human Genetics, University of Oxford, Headington, Oxford, United Kingdom

  • Dominique Gauguier,

    Affiliations The Wellcome Trust Centre for Human Genetics, University of Oxford, Headington, Oxford, United Kingdom, INSERM UMRS872, Centre de Recherche des Cordeliers, Paris, France

  • Daniel E. Platt,

    Affiliation IBM T. J. Watson Research Center, Yorktown Heights, New York, United States of America

  • Pierre A. Zalloua ,

    Affiliations School of Medicine, Lebanese American University, Beirut, Lebanon, Harvard School of Public Health, Boston, Massachusetts, United States of America

  •  [ ... ],
  • the FGENTCARD consortium

    For a full list of the members of the FGENTCARD consortium please see the Acknowledgments section.

  • [ view all ]
  • [ view less ]


Genome wide association studies (GWAS) and their replications that have associated DNA variants with myocardial infarction (MI) and/or coronary artery disease (CAD) are predominantly based on populations of European or Eastern Asian descent. Replication of the most significantly associated polymorphisms in multiple populations with distinctive genetic backgrounds and lifestyles is crucial to the understanding of the pathophysiology of a multifactorial disease like CAD. We have used our Lebanese cohort to perform a replication study of nine previously identified CAD/MI susceptibility loci (LTA, CDKN2A-CDKN2B, CELSR2-PSRC1-SORT1, CXCL12, MTHFD1L, WDR12, PCSK9, SH2B3, and SLC22A3), and 88 genes in related phenotypes. The study was conducted on 2,002 patients with detailed demographic, clinical characteristics, and cardiac catheterization results. One marker, rs6922269, in MTHFD1L was significantly protective against MI (OR = 0.68, p = 0.0035), while the variant rs4977574 in CDKN2A-CDKN2B was significantly associated with MI (OR = 1.33, p = 0.0086). Associations were detected after adjustment for family history of CAD, gender, hypertension, hyperlipidemia, diabetes, and smoking. The parallel study of 88 previously published genes in related phenotypes encompassed 20,225 markers, three quarters of which with imputed genotypes The study was based on our genome-wide genotype data set, with imputation across the whole genome to HapMap II release 22 using HapMap CEU population as a reference. Analysis was conducted on both the genotyped and imputed variants in the 88 regions covering selected genes. This approach replicated HNRNPA3P1-CXCL12 association with CAD and identified new significant associations of CDKAL1, ST6GAL1, and PTPRD with CAD. Our study provides evidence for the importance of the multifactorial aspect of CAD/MI and describes genes predisposing to their etiology.


Given increased life expectancy, prevention of diseases with severe manifestations and later life complications has become of tremendous importance. Researchers' interest in identifying coronary artery disease (CAD) susceptibility genes reflects the very high mortality associated with both the disease [1] and one of its most serious complications: myocardial infarction (MI). While identifying the primary causes of CAD has been complicated by the multifactorial nature of the disease, genome-wide association studies (GWAS) have been moderately successful in revealing some of the polygenic structure of CAD. However, they only explain a modest fraction of the total heritability of CAD [2], [3].

Linkage disequilibrium is the primary mechanism for generating direct (causal) and indirect (non-causal) genetic associations. Population stratification following population mixing [4], [5] can lead to large numbers of correlated mutations [5], [6], [7], almost all of which are expected to be non-causal [8]. Population effects may also produce strong associations in some populations but not others by means of specific enrichment due to population genetic effects (e.g. drift and founder effect) [4], [5], [9], [10]. Furthermore, in view of the multifactorial nature of CAD, the effects of genes acting as component causes may be discernible in some populations but not in others depending on diet, lifestyle, and genetic background. Recent GWAS and their replications that have established highly significant associations of variants with MI and/or CAD are predominantly based on populations of European [11], [12], [13], Eastern Asian [11], [14], [15], and more recently, South Asian descent [16]. Replication of these results for the most significantly associated polymorphisms in multiple populations with distinguishing genetic and lifestyle backgrounds is crucial in isolating the pathophysiology of a multifactorial disease like CAD.

Here, in complement to a GWAS performed for MI and CAD on the same subjects, we have sought to perform a replication study in a Middle-Eastern population of SNPs identified in GWAS of other populations. We have selected susceptibility loci based on prior studies to confirm the association of 10 recently identified SNPs at nine independent loci (Table 1) associated with increased risk of CAD and/or MI in 4 major GWAS conducted in populations of European [11], [12], [13] and East Asian ancestry [11], [14]. In parallel, we have assessed association with CAD phenotype at 88 genes associated with related phenotype using patterns of correlation among SNPs from the International HapMap project around 4,784 genotyped SNPs from these candidate regions to impute genotype at 15,441 collocated, but unavailable marker.

Materials and Methods

Study Subjects

A total of 2,002 subjects recruited between 2006 and 2009 for inclusion in the FGENTCARD database ( were selected. These subjects were referred to a catheterization care unit for clinical evaluation. The 4 main coronary arteries: the left main artery (LMCA), the left anterior descending artery (LAD), the left circumflex artery (LCx), and the right coronary artery (RCA) were visualized from different angles by angiography. The extent of stenosis in these vessels was assessed and recorded by percentage. Some of these patients were admitted to the hospital for having an MI as diagnosed by electrocardiogram and high troponin levels. Cardiologists performing the coronary angiography collected a 20 mL blood sample from the peripheral femoral artery of patients who provided a written consent for the whole study including the genetic analysis. Blood chemistry metrics were obtained. Trained healthcare professionals collected further data on the socio-demographic background of the patients. Annotations were coded from medical charts according to our study protocol, which included results of any additional data such as laboratory tests, prescribed medications, and presence of other diseases and conditions. Physician classified a patient as dyslipidemic when LDL>130 mg/dl and HDL<40 mg/dl, diabetic (type 2) if fasting glucose level>125 mg/dl, hypertensive when blood pressure was >10/14 mm Hg. A positive family history was defined as a sibling, parent, or second-degree relative with a coronary event.

Genomic DNA was extracted using a standard phenol extraction protocol. Genotyping was performed on Illumina Human610-Quad and Illumina Human660W-Quad BeadChips, at the Centre National de Génotypage, France. Out of the 2,002 genomic DNA samples analyzed, 1,949 non-duplicated samples with quality filtered genotyping data were retained. After the removal of three individuals with genotyping success rate less than 95% and 19 pairs of individuals which were suspected to have been duplicated, 1,949 individuals with quality filtered genotyping data were retained. Genotyping success rate was more than 98%, minor allele frequency more than 1%, and the p value of Hardy-Weinberg equilibrium test was less than 1×10−7. The Institutional Review Board at the Lebanese American University approved the study protocol.

Statistical analyses of the nine candidate loci

Our study is a retrospective cross-sectional design with subjects drawn from catheterized patients. Two clinical outcomes were considered in this phase of the study: coronary artery stenosis and MI. In the coronary artery stenosis outcome, subjects were categorized as those with no sign of stenosis (“super-controls”, CAD category 1, n = 425), whereas cases were patients categorized into two classes [17] CAD category 2 for patients with stenosis of less than 50% in at least one vessel (n = 214), and CAD category 3 for patients with stenosis greater than 50% in at least one vessel diagnosed by catheterization (n = 1,310). MI diagnosis was established based on medical examination upon admission to the hospital. In the MI outcome, cases were patients admitted to the hospital with MI positive (n = 222) whereas controls (n = 1,727) were patients admitted for reasons other than MI, such as unstable angina or cardiac diagnostic workup.

We addressed the effects of conventional risk factors for coronary artery. We determined the Chi-squared significance for categorical variables: gender, presence of diabetes mellitus, hypertension, smoking, and family history of CAD, and Welch t-test for continuous variables: FBG, total cholesterol, HDL, LDL, triglycerides, age, and body mass index (BMI). Categorical variables were summarized as percentages while continuous variables with parametric distributions were summarized as means and standard deviations. Age of onset was expressed as median and interquartile range (IQR) and triglyceride levels were logarithmically scaled before analysis (Table 2). Statistical differences between controls and cases for these risk factors were assessed using the unpaired two-sided t-test for the continuous variables and the chi-square test for categorical variables (Table 2).

Table 2. Distribution of biological and biochemical characteristics of patients by stenosis category (n = 1,949).

The 10 SNPs analyzed (Table 1) were selected based on their association with CAD/MI and were searched for using the NCBI PubMed Entrez database ( The association of each SNP with CAD and/or MI was assessed in genotypic and allelic association models using logistic regression.

Proportional odds logistic regression was performed across CAD categories, and standard logistic regression was performed on MI cases vs. controls. The candidate risk alleles were categorized as absent (allele score = 0), heterozygous (allele score = 1), and homozygous (allele score = 2). Logistic regression analyses treated the allele score linearly, corresponding to treating the genotype heterozygous vs. homozygous states as additive to risk, and in separate analyses, treating each allele score as distinct risk categories. Regressions were performed for unadjusted risk odds ratio estimates, as well as adjusted analyses for family history of CAD, gender, hypertension, hyperlipidemia, diabetes, and smoking.

Imputation-based analysis of candidate genes

Coronary artery stenosis was the only clinical outcome considered in this phase of the study. A total of 20,225 variants located in 88 different genes were tested for association in our Lebanese CAD cohort using imputation-based analysis. SNPs were selected on the basis of MI/CAD phenotype in NCBI Phenotype-Genotype Integrator (PheGenI) ( and on literature review [16], [18], [19], [20]. Imputation utilizes the Lebanese population genotype data from 4,784 SNPs, in combination with haplotype information from the CEU population of the release 22 of HapMap to predict genotypes of 15,441 nearby SNPs not genotyped in our subjects. Although the Principal Component Analysis identified our cohort to be closest to the HapMap CEU (data not shown), some underlying different genetic background with the discovery cohorts may exist. We therefore tested for association across the corresponding 88 genes rather than individual variants (Table S1). After imputing across the whole genome to HapMap CEU reference population using Impute v2 [21], we selected both the genotyped and imputed variants in the 88 regions covering the selected genes, allowing for an extra 50 Kb on either side. Because of the probabilistic nature of the genotypes inferred by imputation, we applied SNPTest2 to perform an additive association test using a missing data likelihood score ("-frequentist 1 -method score"), including age and sex as covariates [22]. Effect and significance were reported for association with p-value<0.01 and marker with info>0.4 (Table S1).


Following QC filtering, the sample consisted of 1,949 subjects, 72.7% of whom were males, with a median ±IQR age of 62±16 years, and a mean BMI of 29.0±4.7 kg/m2. Among these subjects, 54.6% were hypertensive, 48.0% were hyperlipidemic, 61.8% had a family history of CAD (parents or siblings), and 29.7% were diabetic. Moreover, higher prevalence among cases compared to controls of diabetes, hypertension, hyperlipidemia, family history of CAD, raised glucose levels, and reduced HDL levels, were observed (Table 2).

The probability of 2 or fewer successes out of 10 Bernoulli trials at the 0.05 significance threshold is 0.0861. However, minor allele frequencies were sufficiently low among a number of SNPs for cases or controls that information about those specific counts was compromised. Only the MI test for rs4977574 was significant after over conservative Bonferroni correction for a one-tail test, while only the rs6922269 was found significantly protective after Bonferroni correction for CAD.

Logistic regressions testing 600K SNP PCA k-means clusters (k = 4), capturing extremes of the leading Lebanese-specific principal components, did not predict MI or CAD, nor interact with risk factors or identified risk SNPs in the prediction of CAD or MI (data not shown). Population stratification identified in the cohort was shown not to have a significant effect in the association with a genomic control of λ =  1.033 (data not shown).

Risks associated with each allele expressed as odds ratios, odds ratios adjusted for gender, hypertension, hyperlipidemia, smoking, and diabetes, are displayed in Table S2 for CAD, and Table S3 for MI. Eight of the ten SNPs, rs1041981, rs11206510, rs1746048, rs3184504, rs646776, rs653178, rs6725887, and rs2048327, showed no two-tailed significant associations with CAD or MI (Tables S2 and S3).

SNP rs4977574 was not significantly associated with CAD (Table S2), but was significantly associated with MI in a proportional odds logistic regression (OR = 1.86, 95%CI = 1.17−3.07, p = 0.012) for the GG genotype compared to the AA genotype (Table S3). In an adjusted analysis, this association with MI was unaltered (OR = 1.84, 95%CI = 1.14−3.09, p = 0.011) (Table S3). The heterozygous GA genotype vs. AA showed odds ratios of 1.39 (unadjusted) and 1.42 (adjusted), but were insignificant (Table S3). Allele frequency results were also insignificant contrasting G vs. A alleles (OR = 1.38, 95%CI = 1.01−1.90, p = 0.039). All adjustment risk factors made highly significant contributions to the proportional odds logistic regression (Table 3).

Table 3. Association of CAD with rs497754 adjusting for each of gender, hypertension, hyperlipidemia, smoking, and diabetes.

The variant rs6922269, was seen to be robustly significant as protective against MI. The heterozygous odds ratio was highly significant (OR = 0.66, 95%CI = 0.48−0.9, p = 0.0087), and remained significant after adjustment (OR = 0.63, 95%CI = 0.45−0.85, p = 0.0037) (Table S3). The additive model risk showed an OR = 0.72, 95%CI = 0.56−0.91, p = 0.0083, and adjusted OR = 0.69, 95%CI = 0.53−0.88, p = 0.0035. It is interesting to note that the measured OR for the homozygous allele was 0.63, 95%CI = 0.31−1.15, p = 0.163 before adjustment, and OR = 0.60, 95%CI = 0.30−1.10, p = 0.121 after adjustment, though while as strong as the heterozygous risk, the error bars were much wider suggesting the population size of n = 11 for MI subjects carrying the homozygous allele was too small to resolve this risk, also suggestive of a protective power for the allele.

Analysis of our Illumina data allowed us to query much of the known SNP variations in the genome. Using an imputation method as a thorough survey of the variations from 88 genes, a total of 15,441 SNPs were imputed genotypes in addition to the 4,784 genotyped SNPs. Table S1 presents an overview of the most nominally significant SNPs (p<0.01) in the regions that have been reported for CAD. SNPs located in the linked regions of CDKAL12, ST6GAL1, CXCL12, and PTPRD were the most highly associated with p values <0.0009. Rs9295489 in the CDKAL1 locus presented the highest association with CAD (beta coefficient = 0.12, p = 0.0004) (Table S1). The locus increased the risk of CAD of about 1.12. Three imputed SNPs in the 50 Kb block of a second locus, ST6GAL1, had p values ranging from 0.0007 (rs16861460) to 0.0008 (rs16861471). The locus had a protective effect against CAD with an estimated beta coefficient of −0.17, corresponding to an odds ratio of 0.8 (Table S1). A third gene, CXCL12, was found to play a role in CAD etiology. The SNP rs7900896 (beta coefficient = 0.12, p = 0.0008) located in the CXCL12 LD block gave high relative information about the additive model and showed significant association with CAD. A fourth LD region defining the PTPRD locus had one in the top six hits of our analysis (rs10115782; beta coefficient = 0.2, p = 0.0009). It was highly informative and presented significant association with CAD (Table S1).


Replication studies are subject to variations in design adding a significant level of complication to comparative analysis [23]. Variability in study enrollment, as well as the use of different inclusion and exclusion criteria such as age of onset, all have the potential to impact risk estimates [12], [24]. Most importantly, CAD outcomes are classified by varying standards. While some use coronary angiography to firmly verify the disease status and to determine whether a patient should belong to the cases or controls groups, others are considering apparently healthy individuals from the general population as controls. A major strength of our study relies on the unambiguous angiographic visualization of CAD phenotypic characterization. A longitudinal prospective study sampling the population at large would have been ideal to increase the frequency of some genotypes and exclude any risk of bias consequent to low frequency representation. However such invasive measures of disease would be ethically impossible.

We have found that the variant rs4977574 located within the 9p21 locus confers an increased risk of developing MI in the GG state. Moreover, subjects homozygous for the G allele are at a much higher risk of MI. These results are in accordance with a study by Kathiresan where rs4977574 G allele was shown to be associated with MI [12]. This study argued that altered transcript levels of CDKN2B in visceral tissue may increase the risk of atherosclerosis emergence [12]. It is pertinent to note that since our study and the GWAS mentioned are all retrospective studies, the rs4977574 risk was survived by patients in treatment of CAD and/or MI long enough to be enrolled in the study.

It is of importance to note that the 9p21 locus, in particular variant rs4977574, did not show genome wide significance for CAD (p = 0.0107) or MI (p = 0.00448) when a GWAS was conducted on this study population (data not shown). This SNP is within a range consistent with the logistic regression results, though not significant after over-conservative Bonferroni correction.

The rs4977574 variant is located within the ANRIL gene that encodes a large non-coding RNA [25]. While it still needs to be elucidated, implication of ANRIL in the pathophysiology of coronary artery syndrome has been strongly supported by expression studies [26], [27], [28]. The ANRIL promoter region contains binding sites for zinc-finger proteins that are critical for the transcription of CDKN2A and CDKN2B [29]. In the mouse, deletion of the 9p21 orthologous non-coding region including the 3′ region of ANRIL affects proliferation of vascular cells while expression of CDKN2A and CDKN2B is severely reduced [30] thus indicating potential implications in the disease etiology in humans.

The variant rs6922269 has shown a protective effect against MI in our population. Association of this variant was already characterized by others [31] through analysis of the Wellcome Trust Case Control Consortium (WTCCC) [32] population. However this association was not confirmed in the Tunisian population [33], and functional implications of the gene in CAD etiology have not been investigated. It can be hypothesized that the inverse relationship of rs6922269 compared to other studies suggests significant interactions with other exposure, or an inverted linkage with the active alleles in the Lebanese population relative to other populations showing risk.

None of the remaining eight gene variants identified as significant CAD or MI risks in other populations showed a contribution to risk of CAD or MI in the Lebanese population, even though some of them are implicated as candidates through genetic association [11], [12] and molecular mechanisms [34], [35], [36], [37], [38], [39]. The negative results regarding these SNPs are important to consider since the differences in association across populations could indicate differences due to lifestyle or environment, or population-specific mutation interaction factors, or linkage variations mitigating disease development. For a number of these alleles, low power due to small numbers, especially in the reference category, may account for some of the losses of significance.

High-throughput genotyping chips such as the Illumina 660Hap used here provide a very good genome-wide coverage via tagSNPs which best represent the underlying haplotypes. This should allow for the identification of regions associated with given phenotypes. However, the tagSNPs provide statistical rather than functional association. It is therefore important to go beyond the original chip and look for the causative variant. Despite a potentially lower MAF, causative variants should have a stronger effect on the phenotype, and more biological relevance. Furthermore, in the case of replication of association originally observed in Caucasian, it is possible that the reported SNP is not associated in our Lebanese cohort. Since our population is sufficiently closely related to the CEU, we could use HapMap catalogue of haplotypes to impute the genotypes of all the markers in regions of interest to test optimally for association between the gene and the phenotype.

This approach unraveled positive association of CAD with CXCL12, a gene that maps to 10q11. A CXCL12 intergenic SNP, rs1746048, had a C-risk allele associated with MI (OR = 1.17, p = 7×10−9) [12]. The same rs1746048-C allele yielded positive association results with CAD in a recent study (OR = 1.09, p = 3×10−10) [20]. Although rs1746048 was not associated with CAD in the Lebanese cohort, another SNP located in the same LD block, rs7900896, showed positive association results. The two SNPs are located 5′ of CXCL12, 467 Kb away. They are not in the same HapMap, CEU-based, LD block. Our results imply that this genomic region contains several causative variants. Whether the associated SNPs are predisposing by themselves, or whether they are in LD with the real variant needs further investigations.

Another SNP, rs9295489 in CDKAL1, showed positive association with CAD in our cohort. As a part of the Cardiovascular Health Study (CHS) GWAS that aims to identify genetic variants associated with cardiovascular risk factors (, rs2206734 in CDKAL1 was associated with MI occurrence (p = 5.05×10−4). The two SNPs are intragenic, 345 Kb away, defining separate LD blocks. While our study confirms that variants near CDKAL1 are associated with risk for cardiovascular anomalies, no published data has associated CDKAL1 with CAD phenotype. Interestingly, since 2007, numerous studies have associated CDKAL1 SNPs with a cardiovascular disease (CVD) risk factor: Type 2 Diabetes (T2D) [19], [40], [41], [42], [43]. Thus, rs9295489 in CDKAL1 may influence T2D levels and consequently affect the risk for CVD.

Two genes, ST6GAL1 and PTPRD (See Table S1 for positive SNPs), were shown to be associated with CAD for the first time in our study. Both genes are known to carry common variants predisposing to Diabetes Mellitus [18], [19], [44], [45]. Rs16861329 in ST6GAL1 locus was associated with T2D and pancreatic beta-cell function [18]. The fact that in our study ST6GAL1 has a protective effect on CAD occurrence suggests an inverted linkage between the active alleles or different alleles arising in various impacts on gene function, and thus, on health. PTPRD, previously shown to be a susceptibility locus for T2D [19], [44], [45], carries variants predisposing to CAD in the Lebanese cohort, showing again a genetic link between T2D and CAD.

A Bonferroni correction per interrogated marker (20,224) would be very over conservative since imputation identified many markers in perfect LD to be tested. A correction for the number of regions tested (88) would be more appropriate, and would retain the significance of these results even though these regions can encompass independent variants. We therefore consider these candidate genes to be of strong interest, but that they require further validation.

This association study investigates genetic factors underlying CAD/MI in individuals of Lebanese ancestry, a population with increased susceptibility to CAD. We identified common genetic variants at two loci previously identified as CAD/MI genetic risk factors (9p21 and CXCL12), and two loci newly associated with CAD (CDKAL1 and PTPRD). We identified MTHFD1L and ST6GAL1 as having a protective effect against CAD/MI. Cardiovascular diseases coexist with metabolic risk factors including central obesity and diabetes. Our study provides new insights into the genetic link between metabolic and cardiovascular traits. Some of the identified loci have implicated metabolic genes without a previously known connection with CAD. Future studies will build on these successes by identifying additional variants and by determining the functional impact of the underlying genes. Further, our findings show the potential for new discovery from genetic association studies in populations of non-European ancestry.

Supporting Information

Table S1.

Variants' imputed genotype probabilities. Imputed genotype probabilities for variants in 88 candidate genes with +/- 50Kb on each side. The test was considered significant when P<0.01, relative information >0.4 and minor allele frequency >5%.


Table S2.

Proportional Odds Logistic regression predicting CAD in graded categories from 10 SNPs. Proportional Odds Logistic regression predicting CAD in graded categories from 10 SNPs, both with additive and independent homozygous/heterozygous odds, without and with adjustment by family history of CAD, history of smoking, diagnoses of diabetes, hyperlipidemia, hypertension, and gender are represented. Odds ratio tests of disease vs. haplotype frequency are also indicated.


Table S3.

Logistic regression predicting MI from 10 SNPs. Analysis was performed both with additive and independent homozygous/heterozygous odds, without and with adjustment by family history of CAD, history of smoking, diagnoses of diabetes, hyperlipidemia, hypertension, and gender. Odds ratio tests of disease vs. haplotype frequency are also represented.



We thank the patients for agreeing to participate in the study. We thank Nour Moukalled and Bariaa Khalil for their help with subject recruitment and data collection. We thank the Rafic Hariri University Hospital, “Centre Hospitalier du Nord”, and Saint Georges Hospital, Lebanon for their collaboration and support.

The FGENTCARD Consortium: Dominique Gauguier, The Wellcome Trust Centre for Human Genetics, University of Oxford, UK; Mark Lathrop, Jörg Hager, CEA-Genomics Institute, Centre National de Génotypage, Evry, France; Jeremy K. Nicholson, Imperial College London, UK; Pierre Zalloua, Lebanese American University, School of Medicine, Beirut, Lebanon; Ulla Grove Sidelmann, Novo Nordisk, Måløv, Denmark; Frank Bonner, Metabometrix Ltd, London, UK.

Author Contributions

Conceived and designed the experiments: JH MF DG DEP PAZ. Performed the experiments: SS SY. Analyzed the data: SS DAB YK GEK MGS JBC DEP PAZ. Contributed reagents/materials/analysis tools: JSY MH RO NS SK HEB EC AS BD. Wrote the paper: SS DAB SY DEP PAZ.


  1. 1. Lloyd-Jones D, Adams RJ, Brown TM, Carnethon M, et al. WRITING GROUP MEMBERS (2010) Heart Disease and Stroke Statistics--2010 Update: A Report From the American Heart Association. Circulation 121: e46–215.
  2. 2. Manolio TA, Collins FS, Cox NJ, Goldstein DB, Hindorff LA, et al. (2009) Finding the missing heritability of complex diseases. Nature 461: 747–753.
  3. 3. Zhang X, Huang S, Zou F, Wang W (2010) TEAM: efficient two-locus epistasis tests in human genome-wide association study. Bioinformatics 26: i217–227.
  4. 4. Cardon LR, Palmer LJ (2003) Population stratification and spurious allelic association. Lancet 361: 598–604.
  5. 5. Khoury MJ (1998) Genetic Epidemiology. In: Rothman KJ, Greenland S, editors. Modern Epidemiology, 2nd Ed. Philadelphia: Lippincott, Williams, and Wilkins. pp. 609–621.
  6. 6. Slatkin M (1991) Inbreeding coefficients and coalescence times. Genet Res 58: 167–175.
  7. 7. Cavalli-Sforza LL, Menozzi P, Piazza A (1994) The History and Geography of Human Genes. Princeton, NJ: Princeton University Press.
  8. 8. Ioannidis JP, Ntzani EE, Trikalinos TA, Contopoulos-Ioannidis DG (2001) Replication validity of genetic association studies. Nat Genet 29: 306–309.
  9. 9. Lin PI, Vance JM, Pericak-Vance MA, Martin ER (2007) No gene is an island: the flip-flop phenomenon. Am J Hum Genet 80: 531–538.
  10. 10. Colhoun HM, McKeigue PM, Davey Smith G (2003) Problems of reporting genetic associations with complex outcomes. Lancet 361: 865–872.
  11. 11. Gudbjartsson DF, Bjornsdottir US, Halapi E, Helgadottir A, Sulem P, et al. (2009) Sequence variants affecting eosinophil numbers associate with asthma and myocardial infarction. Nat Genet 41: 342–347.
  12. 12. Kathiresan S, Voight BF, Purcell S, Musunuru K, Ardissino D, et al. (2009) Genome-wide association of early-onset myocardial infarction with single nucleotide polymorphisms and copy number variants. Nat Genet 41: 334–341.
  13. 13. Tregouet DA, Konig IR, Erdmann J, Munteanu A, Braund PS, et al. (2009) Genome-wide haplotype association study identifies the SLC22A3-LPAL2-LPA gene cluster as a risk locus for coronary artery disease. Nat Genet 41: 283–285.
  14. 14. Ozaki K, Ohnishi Y, Iida A, Sekine A, Yamada R, et al. (2002) Functional SNPs in the lymphotoxin-alpha gene that are associated with susceptibility to myocardial infarction. Nat Genet 32: 650–654.
  15. 15. Wang F, Xu CQ, He Q, Cai JP, Li XC, et al. (2011) Genome-wide association identifies a susceptibility locus for coronary artery disease in the Chinese Han population. Nature genetics.
  16. 16. Peden JF, Hopewell JC, Saleheen D, Chambers JC, Hager J, et al. (2011) A genome-wide association study in Europeans and South Asians identifies five new loci for coronary artery disease. Nature genetics.
  17. 17. van Velzen JE, Schuijf JD, van Werkhoven JM, Herzog BA, Pazhenkottil AP, et al. (2010) Predictive value of multislice computed tomography variables of atherosclerosis for ischemia on stress-rest single-photon emission computed tomography. Circulation Cardiovascular imaging 3: 718–726.
  18. 18. Kooner JS, Saleheen D, Sim X, Sehmi J, Zhang W, et al. (2011) Genome-wide association study in individuals of South Asian ancestry identifies six new type 2 diabetes susceptibility loci. Nature genetics 43: 984–989.
  19. 19. McCarthy MI (2010) Genomics, type 2 diabetes, and obesity. N Engl J Med 363: 2339–2350.
  20. 20. Schunkert H, Konig IR, Kathiresan S, Reilly MP, Assimes TL, et al. (2011) Large-scale association analysis identifies 13 new susceptibility loci for coronary artery disease. Nat Genet.
  21. 21. Howie BN, Donnelly P, Marchini J (2009) A flexible and accurate genotype imputation method for the next generation of genome-wide association studies. PLoS genetics 5: e1000529.
  22. 22. Marchini J, Howie B, Myers S, McVean G, Donnelly P (2007) A new multipoint method for genome-wide association studies by imputation of genotypes. Nature genetics 39: 906–913.
  23. 23. Risch N (2000) Searching for genes in complex diseases: lessons from systemic lupus erythematosus. J Clin Invest 105: 1503–1506.
  24. 24. Abdullah KG, Li L, Shen GQ, Hu Y, Yang Y, et al. (2008) Four SNPS on chromosome 9p21 confer risk to premature, familial CAD and MI in an American Caucasian population (GeneQuest). Ann Hum Genet 72: 654–657.
  25. 25. Pasmant E, Laurendeau I, Heron D, Vidaud M, Vidaud D, et al. (2007) Characterization of a germ-line deletion, including the entire INK4/ARF locus, in a melanoma-neural system tumor family: identification of ANRIL, an antisense noncoding RNA whose expression coclusters with ARF. Cancer Res 67: 3963–3969.
  26. 26. Liu Y, Sanoff HK, Cho H, Burd CE, Torrice C, et al. (2009) INK4/ARF transcript expression is associated with chromosome 9p21 variants linked to atherosclerosis. PLoS One 4: e5027.
  27. 27. Folkersen L, Kyriakou T, Goel A, Peden J, Malarstig A, et al. (2009) Relationship between CAD risk genotype in the chromosome 9p21 locus and gene expression. Identification of eight new ANRIL splice variants. PLoS One 4: e7677.
  28. 28. Jarinova O, Stewart AF, Roberts R, Wells G, Lau P, et al. (2009) Functional analysis of the chromosome 9p21.3 coronary artery disease risk locus. Arterioscler Thromb Vasc Biol 29: 1671–1677.
  29. 29. Rodriguez C, Borgel J, Court F, Cathala G, Forne T, et al. (2010) CTCF is a DNA methylation-sensitive positive regulator of the INK/ARF locus. Biochem Biophys Res Commun 392: 129–134.
  30. 30. Visel A, Zhu Y, May D, Afzal V, Gong E, et al. (2010) Targeted deletion of the 9p21 non-coding coronary artery disease risk interval in mice. Nature 464: 409–412.
  31. 31. Lluis-Ganella C, Lucas G, Subirana I, Senti M, Jimenez-Conde J, et al. (2010) Additive effect of multiple genetic variants on the risk of coronary artery disease. Revista espanola de cardiologia 63: 925–933.
  32. 32. WTCCC (2007) Genome-wide association study of 14,000 cases of seven common diseases and 3,000 shared controls. Nature 447: 661–678.
  33. 33. Ghazouani L, Khalifa SB, Abboud N, Perret C, Nicaud V, et al. (2010) Association of three polymorphisms selected from a genome-wide association study with coronary heart disease in the Tunisian population. Journal of thrombosis and thrombolysis 29: 114–118.
  34. 34. Kathiresan S, Melander O, Guiducci C, Surti A, Burtt NP, et al. (2008) Six new loci associated with blood low-density lipoprotein cholesterol, high-density lipoprotein cholesterol or triglycerides in humans. Nat Genet 40: 189–197.
  35. 35. McNutt MC, Kwon HJ, Chen C, Chen JR, Horton JD, et al. (2009) Antagonism of secreted PCSK9 increases low density lipoprotein receptor expression in HepG2 cells. J Biol Chem 284: 10561–10570.
  36. 36. Ronald J, Rajagopalan R, Ranchalis JE, Marshall JK, Hatsukami TS, et al. (2009) Analysis of recently identified dyslipidemia alleles reveals two loci that contribute to risk for carotid artery disease. Lipids Health Dis 8: 52.
  37. 37. Mehta NN, William D, Pruscino L, Khera A, Edmondson A, et al. (2009) Abstract 2077: A Novel GWAS Atherosclerosis Locus Regulates Plasma Levels of CXCL12. Circulation. 120. S589-a- p.
  38. 38. Holzel M, Rohrmoser M, Schlee M, Grimm T, Harasim T, et al. (2005) Mammalian WDR12 is a novel member of the Pes1-Bop1 complex and is required for ribosome biogenesis and cell proliferation. J Cell Biol 170: 367–378.
  39. 39. Prasannan P, Pike S, Peng K, Shane B, Appling DR (2003) Human mitochondrial C1-tetrahydrofolate synthase: gene structure, tissue distribution of the mRNA, and immunolocalization in Chinese hamster ovary calls. J Biol Chem 278: 43178–43187.
  40. 40. Saxena R, Voight BF, Lyssenko V, Burtt NP, de Bakker PI, et al. (2007) Genome-wide association analysis identifies loci for type 2 diabetes and triglyceride levels. Science 316: 1331–1336.
  41. 41. Scott LJ, Mohlke KL, Bonnycastle LL, Willer CJ, Li Y, et al. (2007) A genome-wide association study of type 2 diabetes in Finns detects multiple susceptibility variants. Science 316: 1341–1345.
  42. 42. Steinthorsdottir V, Thorleifsson G, Reynisdottir I, Benediktsson R, Jonsdottir T, et al. (2007) A variant in CDKAL1 influences insulin response and risk of type 2 diabetes. Nature genetics 39: 770–775.
  43. 43. Zeggini E, Weedon MN, Lindgren CM, Frayling TM, Elliott KS, et al. (2007) Replication of genome-wide association signals in UK samples reveals risk loci for type 2 diabetes. Science 316: 1336–1341.
  44. 44. Below JE, Gamazon ER, Morrison JV, Konkashbaev A, Pluzhnikov A, et al. (2011) Genome-wide association and meta-analysis in populations from Starr County, Texas, and Mexico City identify type 2 diabetes susceptibility loci and enrichment for expression quantitative trait loci in top signals. Diabetologia 54: 2047–2055.
  45. 45. Tsai FJ, Yang CF, Chen CC, Chuang LM, Lu CH, et al. (2010) A genome-wide association study identifies susceptibility variants for type 2 diabetes in Han Chinese. PLoS genetics 6: e1000847.