Genetic variation at the 9p21 locus predicts angiographic coronary artery disease prevalence but not extent and has clinical utility

Article Outline

• Abstract
• Methods
  • Objectives
  • Angiographic study population
  • Population-based controls
  • Assessment of angiographic CAD
  • Clinical information
  • Lipid measurements
  • Inflammatory markers
  • Genotyping
  • Study hypotheses
  • Statistical considerations
• Results
  • Patient characteristics
  • SNP distributions and cross correlations
  • Univariate predictive ability for angiographic CAD in the initial association set
  • Multivariable risk prediction, subset analysis, and population attributable risk
  • Association of haplotypes of the 4 SNPs with CAD
  • Replication and combined sets analyses
  • Association of 9p21 with extent of angiographic disease
  • Inflammatory marker associations
  • Impact on Framingham risk assessment
• Discussion
  • Summary of key results
  • Literature comparisons
  • Mechanistic considerations
  • Clinical implications
  • Limitations
• Conclusions
• Acknowledgment
• Appendix A. Supplementary data
• References
• Copyright

Background

Variants at the 9p21 locus have been associated with coronary heart disease, but their precise disease phenotype and utility for clinical risk assessment are uncertain.

Methods

Consenting patients with early-onset angiographic coronary artery disease (CAD) (n = 1,011) were compared with matched subjects (n = 545) free of angiographic disease and with a random population sample (n = 565). Cases and controls were genotyped for 4 variants, and ORs for angio-CAD were determined. Findings were validated in a separate set of cases and controls (n = 1,452).

Results

Alleles were highly correlated (r² ≥ 0.9), and all predicted angio-CAD compared with both control groups. Genotype at rs2383206 (minor allele frequency 45.9%), the most predictive (P < .0001), was associated with an adjusted odds ratio for angio-CAD of 1.39 (95% CI, 1.05–1.85) for heterozygote and 1.73 (1.26–2.37) for homozygote risk-allele carriers and explained 21% of population attributable risk and was independent of traditional risk factors and myocardial infarction. For the comparison of combined cases versus combined control samples (N = 3,573), CAD was predicted by high-risk allele homozygosity at P = 9 × 10⁻⁸. Despite this, extent of disease was not increased. Applied to patients with intermediate Framingham risk scores, 9p21 genotyping modified risk classification in 24%.

Conclusions

Variants at the 9p21 locus robustly predict angiographic CAD prevalence, independent of standard risk factors, but not CAD extent or myocardial infarction; provide pathophysiological insights; and may be clinically useful in refining coronary heart disease risk classification.

Cardiovascular disease has emerged as the leading cause of morbidity and mortality worldwide.¹, ² Although environmental factors account for much of coronary heart disease (CHD) risk,³ marked interindividual susceptibility suggests a major contribution of genetics.⁴, ⁵, ⁶, ⁷, ⁸ However, the genetic underpinnings remain largely unknown.⁹, ¹⁰ Efforts focusing on variants (polymorphisms) in candidate genes have met with only modest success and frequent failures of replication.⁸, ¹¹, ¹², ¹³, ¹⁴, ¹⁵

The development and recent application of high-density genotyping arrays to genome-wide scanning has opened up a new era in the discovery of the genetic basis for common, complex diseases such as CHD.¹⁶, ¹⁷ Using genome-wide scanning, several groups within the past year have reported associations of CHD with a locus on chromosome 9p21.3.¹⁸, ¹⁹, ²⁰, ²¹, ²², ²³, ²⁴, ²⁵ The locus is located outside of annotated genes.¹⁹, ²⁰, ²³ Moreover, its precise clinical phenotype (e.g., CAD onset vs severity of atherosclerosis¹⁹, ²⁰, ²⁴), its relation to other CHD risk factors,⁹, ¹⁰ and its incremental value in clinical risk assessments, remain to be resolved. Accordingly, we undertook an initial assessment and validation of variants in a large and prospectively collected, angiographically defined cohort of cases and complementary control groups and assessed impact on CHD risk assessment.

Back to Article Outline

Methods 

Objectives 

The primary study objectives were (1) to prospectively assess the association of four 9p21 variants in patients with early-onset, angiographically defined CAD compared with (a) matched angiographically healthy subjects and with (b) a random population sample; (2) to validate the initial findings in a separate set of cases and controls; (3) to assess the association of the 9p21 variants with 8 traditional and 2 novel (inflammatory) risk factors and with CAD distinct from myocardial infarction (MI); (4) to assess impact of 9p21 sequence variation on extent of CAD; and (5) to explore the impact of 9p21 genotyping on individual Framingham risk assessment.

Angiographic study population 

Angiographic cases and controls were selected from among patients ≥18 years undergoing coronary angiography within the Intermountain Healthcare LDS Hospital (Salt Lake City, UT), who gave written, informed consent to participate in the cardiac catheterization registry of the Intermountain Heart Collaborative Study, which included a fasting blood draw for DNA and plasma samples. The study protocols were approved by the LDS Hospital's institutional review board.

The initial association sample comprised years 1994 to 2003, and the replication sample, 2003 to 2007. Early-onset CAD cases and controls were selected 2:1 by matching for sex, age and nearest date of entry into the registry. Early-onset disease was defined as angiographic CAD at ≤55 years of age in men and ≤65 years in women in the initial association set. Because of changing disease demographics, the definition was modified in the replication set to ≤60 years in men and ≤70 years in women.

Population-based controls 

A population-based control sample was assembled by invitation (letters and follow-up telephone calls) to a randomly generated sample of subjects in the Utah Population Database. Responders were scheduled for a clinic appointment where they completed a health-related questionnaire, had vital signs taken, and donated blood for study-related testing.

Assessment of angiographic CAD 

The presence of angiographic CAD was determined by blinded coronary angiographic analysis. Patients were categorized as free of CAD, moderate CAD (ie, most severe lesion 10%-69% stenosis), or severe CAD (ie, at least one lesion of >70% stenosis). Patients with moderate CAD were excluded as indeterminate. Extent of disease was assessed by counting the number of major vessels with severe disease and by calculating the Duke CAD Index.²⁶

Clinical information 

Demographic, health history, and vital signs at entry were obtained from physician and hospital records and stored in a computer database using previously described criteria.²⁷

Lipid measurements 

Serum total cholesterol and triglyceride concentrations were measured on fasting blood samples drawn at study entry using dry-slide technology on a VITROS 950 analyzer (Ortho Clinical Diagnostics, Rochester, NY). High-density lipoprotein cholesterol (HDL-C) was measured using VITROS HDL-C Magnetic Reagent. Low-density lipoprotein cholesterol (LDL-C) was calculated.

Inflammatory markers 

High-sensitivity C-reactive protein (hsCRP) concentration was determined in a sample subset using an immunoturbidimetric assay on the Hitachi 917 analyzer (Roche Diagnostics, Indianapolis, IN) using reagents and calibrators from Denka Seiken (Niigata, Japan). Lipoprotein-associated phospholipase A2 (LpPLA₂) was measured in a sample subset by DiaDexus (San Francisco, CA) using the PLAC test (www.plactest.com).

Genotyping 

Genotyping of rs2383206, rs2383207, rs10757274, and rs10757278 single-nucleotide polymorphisms (SNPs) at 9p21 was performed with 5′ exonuclease (Taqman) chemistry on the ABI Prism 7000 (Applied Biosystems, Foster City, CA).

Study hypotheses 

The primary hypothesis was that ≥1 of the 4 9p21 allelic variants would predict an altered risk of angiographic CAD compared with (1) angiographic controls and (2) population-based controls. A corollary hypothesis was that the initial association(s) would be validated in the replication set of cases and controls. Secondary hypotheses were that the association with CAD would be independent of traditional risk factors and would predict extent of disease. A tertiary hypothesis was that 9p21 would modify individual Framingham risk assessment in the control group at intermediate (10%-20% per 10 years) primary CHD risk.²⁸

Statistical considerations 

The primary statistical comparison was between the initial angiographic case and angiographic control set. Assuming an additive effect model, a minor allele frequency of 0.45, and a per-allele effect size of 20%, a sample consisting of 1000 cases and 500 controls yields a power of >99% at an α of .001.

Evaluation of continuous variables by genotype was performed by Student t test or analysis of variance (triglyceride and hsCRP levels were log-transformed). Differences in CAD prevalence were evaluated for each SNP by the χ² test. Correlations used Pearson's test. Univariable and multivariable logistic regression provided adjusted ORs and 95% CI, with forced entry of covariables for multivariable testing. Covariables were age; sex; and 6 other CHD risk factors, i.e., hyperlipidemia, hypertension, smoking, diabetes, family history, and body mass index (BMI). Adjustment was made in other models for total cholesterol, LDL-C, HDL-C, triglycerides, systolic and diastolic blood pressure, and ethnicity. Finally, adjustment was made for hsCRP and LpPLA2 in subgroups with marker values. Framingham risk assessment was made using published, sex-specific nomograms.²⁸

Population attributable risk was calculated from the formula: 100D / (1 + D), where D is equal to P₁ (RR₁ − 1) = P₂ (RR₂ − 1), where P₁ and P₂ are the frequencies of the at-risk genotypes, and RR₁ and RR₂ are their associated relative risks, as compared with the low-risk genotype.

Statistical analyses were performed with SPSS (v.13.0, SPSS Inc, Chicago, IL). HapMC (personal communication, N.J.C.) was used to estimate haplotypes and perform permutation analysis.²⁹ Two-tailed P values for the primary comparison were considered provisionally significant at P ≤ .04, based on the high correlation (r² ≥ 0.9) of the 4 variant SNPs, and definitive for P ≤ .0125, based on a conservative Bonferroni correction for the comparisons of the 4 SNPs with CAD.

Back to Article Outline

Results 

Patient characteristics 

Characteristics of the initial association set of cases and controls are summarized in Table I. As expected, traditional risk factors were more prevalent in cases than either control group. Lipid profile and blood pressure were more favorable in population controls, whereas they were more similar in angiographic controls compared to cases. Levels of hsCRP and LpPLA2 were higher in cases.

Table I. Characteristics of initial association sets of angiographic cases and controls

H/o, History of. Note that lipid levels were generally on treatment.

SNP distributions and cross correlations 

Allelic frequencies and representative genotypes in the initial case and control sample sets are shown in Table II. Genotype distributions were in Hardy Weinberg equilibrium. All variants were highly prevalent, with control population minor (high-risk) allele frequencies >45%. Distributions differed between cases and each control group but not between the 2 control groups. The 4 SNPs were highly correlated with each other, (r² ≥ 0.9; Supplemental Table A, available online) but did not correlate with standard CHD risk factors (Supplemental Table B, available online). Based on this linkage group analysis, rs2383206 was selected as the tagging SNP for the set of 4 9p21 SNPs in subsequent analyses. ln-LpPLA2 but not ln-hsCRP correlated weakly with rs2383206.

Table II. Allelic and genotypic frequencies by CAD diagnosis

Variable	Angiographic CAD cases	Angiographic healthy subjects	Population healthy subjects
Frequency, rs2383206 allele (n [%])	1114/2022†, ‡ (55.1%)	531/1090 (48.7%)	549/1132 (48.4%)
Frequency, rs2383207 allele (n [%])	1117/2000⁎, ‡ (55.9%)	541/1084 (49.8%)	561/1138 (49.3%)
Frequency, rs10757278 allele (n [%])	1023/1980⁎, ‡ (51.7%)	498/1074 (46.4%)	4898/1078 (45.4%)
Frequency, rs10757274 allele (n [%])	1030/1998⁎, ‡ (51.6%)	509/1088 (46.8%)	511/1132 (45.1%)
rs2383206 genotype (n [%])	1000 (100%)†	542 (100%)	569 (100%)
Low-risk homozygote	201 (20.1%)	141 (26.0%)	141 (24.8%)
Heterozygote	481 (48.1%)	261 (48.2%)	295 (51.8%)
High-risk homozygote	318 (31.8%)	140 (25.8%)	133 (23.4%)
rs2383207 genotype (n [%])	1011 (100%)⁎	545 (100%)	566 (100%)
Low-risk homozygote	209 (20.7%)	148 (27.2%)	144 (25.4%)
Heterozygote	490 (48.5%)	263 (48.3%)	295 (52.1%)
High-risk homozygote	312 (30.9%)	134 (24.6%)	127 (22.4%)

Univariate predictive ability for angiographic CAD in the initial association set 

All 4 alleles predicted CAD when compared to sex-/age-matched angiographic controls, to population controls, and to the combined control group (P = .012-.0008 vs angiographic controls; P = .001-<.0001 vs population and all controls) (Table III). Rs2383206, the tagging SNP, was most highly predictive, with an OR for CAD per high-risk allele compared to the combined control groups of 1.29 (1.15-1.46) (P = .000028). Results were most consistent with an additive model.

Table III. Predictive ability of 9p21 SNPs for angiographic CAD in the initial association sets

SNP	Heterozygote OR (95% CI)	High-risk homozygote OR (95% CI)	Allele OR (95% CI)
Angio CAD vs angio controls
rs10757278 UV	1.32 (1.02-1.69)	1.52 (1.13-2.03)	1.23 (1.07-1.43)
rs10757274 UV	1.28 (0.99-1.64)	1.47 (1.10-1.97)	1.21 (1.05-1.41)
rs2383207 UV	1.30 (1.0-1.69)	1.61 (1.20-2.16)	1.27 (1.10-1.47)
rs2383206 UV	1.33 (1.03-1.72)	1.67 (1.25-2.24)	1.29 (1.12-1.50)
rs2383206 MV adj-1	1.39 (1.05-1.85)	1.73 (1.26-2.37)	1.31 (1.12-1.54)
Angio CAD vs population controls
rs10757278 UV	1.13 (0.89-1.49)	1.69 (1.25-2.28)	1.29 (1.11-1.49)
rs10757274 UV	1.20 (0.94-1.54)	1.70 (1.26-2.29)	1.29 (1.12-1.50)
rs2383207 UV	1.15 (0.89-1.49)	1.67 (1.24-2.25)	1.30 (1.12-1.51)
rs2383206 UV	1.14 (0.89-1.48)	1.69 (1.26-2.28)	1.30 (1.13-1.51)
rs2383206 MV adj	1.17 (0.86-1.58)	1.64 (1.16-2.32)
Angio CAD vs combined controls
rs10757278 UV	1.22 (0.99-1.50)	1.60 (1.25-2.03)	1.26 (1.11-1.42)
rs10757274 UV	1.24 (1.01-1.52)	1.56 (1.23-2.00)	1.25 (1.11-1.41)
rs2383207 UV	1.22 (0.98-1.52)	1.64 (1.28-2.09)	1.28 (1.14-1.45)
rs2383206 UV	1.23 (0.99-1.52)	1.67 (1.31-2.13)	1.29 (1.15-1.46)
rs2383206 MV adj	1.26 (0.99-1.61)	1.74 (1.33-2.28)

Angio, Angiographic; UV, univariable; MV, multivariable; adj, adjusted.

In contrast, allelic frequencies of the 4 SNPs were similar in the 2 control populations (P = .44-.92, linear-by-linear associations), indicating a lack of differential selection bias despite somewhat differing risk profiles (Table I), and CAD risk prediction was similar using either control group.

Multivariable risk prediction, subset analysis, and population attributable risk 

The association of the 9p21 SNPs with CAD was undiminished by adjusting for multiple standard CHD risk factors. Results for rs2383206 are shown in Table III.

Predictive ability for CAD was generally consistent in risk factor subsets of interest (Supplemental Table C, available online). Apparent trends by family history, age, and sex disappeared when retested in the replication set (adjusted P-interaction = .30, .33, .66, respectively, for the combined initial and replication sets).

A multivariable model for CAD combining rs2383206 with standard risk factors for the initial angiographic association cohort is shown in Table IV. A highly significant contribution of rs2383206 was noted (P = .003), placing it ahead of BMI, family history, and hypertension in power and behind hyperlipidemia, smoking, and diabetes for these age- and sex-matched cohorts.

Table IV. Multivariable predictive model for CAD of rs2383206 and standard risk factors in the initial angiographic association set

The population attributable risk of rs2383206, based on frequencies of the at-risk genotypes and their multivariable adjusted relative risks (Table III), was estimated to be 21%.

Association of haplotypes of the 4 SNPs with CAD 

Haplotypes were estimated using our previously generated and validated expectation-maximization algorithm based on 100 000 simulations.²⁹ Results for the initial CAD set compared to the 2 control groups combined are shown in Supplemental Table B, available online. The most common variant (high-risk) haplotype, GGGG, showed a strong association with CAD compared to the common (wild-type, low-risk) comparator, AAAA, with OR = 1.29 (1.14-1.46), P-empirical = .00004.

Replication and combined sets analyses 

An equally sized, independent and temporally consecutive series of cases and controls was selected for validation testing and genotyped for rs2383206. By design, age was slightly older in cases and controls (55.4 years; SD, 7.8 years), whereas sex distribution (62.4% men) was similar to the initial (discovery) set. As with the discovery population, CAD was predicted by the high-risk variant when compared to sex-/age-matched angiographic controls, to population controls, and to the combined control group (Figure 1). When all replication and initial discovery sets of cases and controls were combined (total N = 3573), odds ratios for the rs2383206 heterozygote of 1.28 (1.09-1.51) (P = .003) and for the high-risk homozygote of 1.66 (1.38-2.00) (P = 9 × 10⁻⁸) were obtained.

• 
  • View Large Image
  • Download to PowerPoint

• Figure 1. 

  Predictive ability of rs238206 for angiographic CAD in the replication and combined association sets. A, Heterozygous patients. B, Homozygous patients. *Multivariable analysis could not be completed because of incomplete risk data in the population control set. Angio, Angiographic.

Association of 9p21 with extent of angiographic disease 

Despite the highly significant predictive ability of 9p21 for the presence of angiographic CAD, the extent of CAD was not significantly impacted by 9p21 genotype, either when assessed by number of vessels severely diseased or by the Duke CAD Index²⁶ (Table V).

Table V. Extent of CAD by rs2383206 genotype

Inflammatory marker associations 

9p21 SNPs were not associated with hsCRP, but homozygotes showed a modest association with increased LpPLA2 levels (ie, median 330 ng/mL in rs2383206 minor allele noncarriers vs 397 ng/mL in homozygote carriers, P = .04, ln-transformed analysis) (Supplemental Table B, available online).

Impact on Framingham risk assessment 

Novel risk factors (eg, hsCRP), including genetic factors, are recommended as most appropriately applied in subjects in the Adult Treatment Panel III intermediate risk category,²⁸ ie, with a calculated 10-year Framingham risk of 10% to 20%.³⁰ Among the 565 population-based control subjects (our true primary CHD risk cohort), 21% (n = 116) were at intermediate risk. Of these, 24% (n = 28) were reclassified into high-risk (>20%, n = 15) or low-risk (<10%, n = 13) by adding 9p21 genotype to risk assessment (Figure 2) (we assumed the 9p21 rs2383206 heterozygote to represent the population average or referent subject, the high-risk homozygote to increment risk by 30%, and the low-risk homozygote to decrement risk by 30%).

• 
  • View Large Image
  • Download to PowerPoint

• Figure 2. 

  Reclassification of intermediate risk subjects by addition of 9p21 genotype to Framingham risk assessment. y-Axis represents numbers of subjects in population based sample in low-, intermediate-, and high-risk categories before and after addition of 9p21 rs2383206 genotype to risk assessment. See text for details. FRS, Framingham risk score.

Back to Article Outline

Discussion 

Summary of key results 

This large study, representing a total of 3573 cases and controls, prospectively assessed and validated the value of highly correlated genetic sequence variants at 9p21.3 to predict the specific CHD phenotype of angiographic CAD in a geographically distinct, primarily Caucasian population. The study demonstrated several levels of internal consistency and was statistically robust: a precise phenotype for cases and controls was established angiographically; findings among the 4 highly linked SNPs were consistent; predictive ability was shown in comparison to each of 2 independent and distinctively selected control groups; and results were replicated in a separate, similarly selected set of cases and controls. Associations with CAD were similar whether single SNPs or a combined haplotype was considered. Thorough clinical and biomarker characterization of the sample sets allowed the strikingly independent nature of the risk association (with respect to traditional coronary risk factors) to be demonstrated. Angiographic CAD association was independent of MI history. At a population level, 9p21 variation explained an important proportion (21%) of attributable CAD risk. At an individual level, knowledge of 9p21 genotype reclassified 1 quarter of subjects at intermediate CHD risk. In external comparisons, the study shows major consistency with several other recent reports while addressing outstanding questions and inconsistencies among these reports. Despite this, extent of disease was not increased in high-risk allele carriers, a novel observation with pathophysiological implications.

Literature comparisons 

Using genome-wide scanning, at least 7 reports within the past year, representing at least 20 distinct population samples, have reported associations of CHD with a locus on chromosome 9p21.3.¹⁸, ¹⁹, ²⁰, ²¹, ²², ²³, ²⁴, ²⁵, ³¹ Of multiple tested SNPs within a 13- to 100-kb region, 4 previously well-studied SNPs were prospectively selected for validation and clinical utility testing.¹⁹, ²⁰, ²³, ²⁴, ²⁵ The current study provides complementary and incremental information to these reports.

First, the need for validation of in multiple diverse populations is emphasized by past problems of replication.⁸, ¹¹, ¹², ¹³, ¹⁴ Our results for allelic and haplotypic distributions, risk associations, and attributable risk are strikingly consistent with and complementary to these other reports.

Second, this study refines the CHD phenotype to angiographically defined CAD. Other studies have included a spectrum of phenotypes, including a spectrum of clinical CHD phenotypes,¹⁹, ²¹, ²², ²³ MI,²⁰, ²¹ coronary artery calcification,¹⁹ or angiographic CAD plus history of MI.²⁴, ²⁵ Similarly, there has been a diversity of controls. The present result establishes an association of 9p21 with anatomic CAD per se, which is equally well distinguished from angiographic healthy subjects and from a random population sample.

Previous work has suggested independence from traditional risk factors with varying characterization; here, we demonstrate a striking independence based on both qualitative and quantitative assessment (ie, for lipids/lipoproteins, blood pressure, weight, inflammatory biomarkers). Our data also firmly support utility of the universally common 9p21 variants independent of family history—a controversial issue until now.²¹, ²², ²⁵ In addition, we show that genotyping can provide incremental value to Framingham risk score for clinical CHD risk assessment in a sizable proportion (1 quarter) of subjects at intermediate (ie, ambiguous) pretest risk.

Mechanistic considerations 

The genetic/molecular basis of 9p21 genetic variation on CHD risk is unknown. The 9p21.3 locus is located outside of annotated genes. A role for CDKN2A and CDKN2B, which play an important role in regulation of the cell cycle and lie in relatively close proximity,²¹ appears to be excluded by resequencing studies of McPherson et al.¹⁹ Subsequently, Broadbent et al.²³ reported that the 9p21 high-risk haplotype collocates with a large antisense noncoding RNA gene, which is expressed in tissues and cell types affected by atherosclerosis and which might act as an important growth-regulatory element.

By refining the clinical phenotype to initiation of angiographic CAD, this study provides mechanistic insight, i.e., the risk association appears to involve an initiator/promoter mechanism of atherosclerosis.⁸ A major potentiator role is excluded by the similar angiographic severity scores between risk-allele carriers and noncarriers. It will now be of interest to prospectively test whether it affects incident MI apart from angiographic CAD (ie, prospective comparison of CAD cases which do vs do not suffer incident MI during follow-up). A modest inflammatory element is possible for LpPLA2, but not hsCRP. Finally, our data favor an additive genetic model.³¹

Clinical implications 

For clinical application, a risk marker should be firmly validated; be of sufficient magnitude to be of value; be independent of and incremental to standard risk factors; and provide valuable prognostic, diagnostic, and/or therapeutic information. 9p21 is now established as a CHD risk-associated genetic marker³¹; its risk magnitude appears to be similar to that of other recently approved and applied novel markers, that is, hsCRP³⁰ and LpPLA2,³² contributing importantly to population attributable risk; it shows independence from traditional risk factors; and we show that it may refine the Framingham risk category in intermediate risk subjects. These properties commend it for testing as a novel CHD risk marker.

Limitations 

The problems inherent to all observational studies, e.g. selection bias, population stratification, and statistical issues, should be considered in reviewing these results. However, the prospective hypothesis; the precise matching by age, sex, and time; the relatively large sample size; the replication design; and the internal and external consistency strongly suggest the overall results to be valid. Assessing the impact of risk reclassification on clinical outcomes will require prospective clinical studies.³³ Our results do not address the outstanding question of predictive ability in populations of African descent.¹⁹

Back to Article Outline

Conclusions 

In this large, angiographically defined case-control study, variants at the 9p21 locus were found to robustly predict CAD prevalence independent of standard risk factors, extent of disease, and MI. Taken together, these results suggest a specific role for 9p21 in atherosclerosis initiation/promotion. Finally, 9p21 genotyping appears to be useful in refining risk classification in those at intermediate risk and deserves prospective clinical testing as a novel risk marker.

Back to Article Outline

We thank Lisa Cannon-Albright, PhD, Paul Hopkins, MD, MPH, and Donald L. Lappé, MD, for scientific, clinical, administrative, and laboratory assistance.

Back to Article Outline

Appendix A. Supplementary data 

Supplemental Table A. Correlations among the four 9p21 SNPs

All correlations are P < .001.

Supplemental Table B. Correlations of standard risk factors, hsCRP, and LpPLA2 with rs2383206

H/o, History of.

Supplemental Table C. Predictive ability of rs2383206 for CAD in population risk factor subsets of interest

Table shows ORs and 95% CIs for CAD in CHD risk subsets of rs2383206 allele together with the P value for interaction for the overall combined study population of cases and controls (multivariable adjustment accounts for other factors in table).

Supplemental Table D. Association analysis of haplotypes of the 4 SNPs with CAD

Back to Article Outline

References 

1. Rosamond W, Flegal K, Friday G, et al. Heart disease and stroke statistics–2007 update: a report from the American Heart Association Statistics Committee and the Stroke Statistics Subcommittee. Circulation. 2007;115:69–171
   • View In Article
2. Yusuf S, Reddy S, Ôunpuu S, et al. Global burden of cardiovascular diseases, part I: general considerations, the epidemiologic transition, risk factors, and impact of urbanization. Circulation. 2001;104:2746–2753
   • View In Article
   • CrossRef
3. Yusuf S, Hawken S, Ounpuu S, et al. Effect of potentially modifiable risk factors associated with myocardial infarction in 52 countries (the INTERHEART study): case-control study. Lancet. 2004;364:934–952
   • View In Article
   • Full Text
   • Full-Text PDF (58 KB)
   • CrossRef
4. Williams RR, Hunt SC, Heiss G, et al. Usefulness of cardiovascular family history data for population-based preventive medicine and medical research (the Health Family Tree Study and the NHLBI Family Heart Study). Am J Cardiol. 2001;87:129–135
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (87 KB)
   • CrossRef
5. Marenberg ME, Risch N, Berkman LF, et al. Genetic susceptibility to death from coronary heart disease in a study of twins. N Engl J Med. 1994;330:1041–1046
   • View In Article
   • MEDLINE
   • CrossRef
6. Zdravkovic L, Wienke A, Pedersen NL, et al. Heritability of death from coronary heart disease: a 36-year follow-up of 20 966 Swedish twins. J Intern Med. 2002;252:247–254
   • View In Article
   • MEDLINE
   • CrossRef
7. Wienke A, Holm NV, Skytthe A, et al. The heritability of mortality due to heart diseases: a correlated frailty model applied to Danish twins. Twin Res. 2001;4:266–274
   • View In Article
   • MEDLINE
   • CrossRef
8. Hopkins PN, Hunt SC, Wu LL. Family history and genetic factors. In:  Wong ND,  Black HR,  Gardin JM editor. Preventive cardiology, a practical approach. New York: McGraw-Hill; 2005;p. 92–148
   • View In Article
9. Topol EJ, Smith J, Plow EF, et al. Genetic susceptibility to myocardial infarction and coronary artery disease. Hum Molec Genet. 2006;15:R117–R123
   • View In Article
10. Anderson JL, Carlquist JF, Horne BD, et al. Progress in unraveling the genetics of coronary artery disease and myocardial infarction. Curr Atheroscl Rep. 2007;9:179–186
    • View In Article
11. Ioannidis JPA, Ntzani EE, Trikalinos TA, et al. Replication validity of genetic association studies. Nat Genet. 2001;29:306–309
    • View In Article
    • MEDLINE
    • CrossRef
12. Hegele RA. SNP judgments and freedom of association. Arterioscler Thromb Vasc Biol. 2002;22:1058–1061
    • View In Article
    • CrossRef
13. Morgan TM, Krumholz HM, Lifton RP, et al. Nonvalidation of reported genetic risk factors for acute coronary syndrome in a large-scale replication study. JAMA. 2007;297:1551–1561
    • View In Article
    • CrossRef
14. Yamada YH, Izawa H, Ichihara S, et al. Prediction of the risk of myocardial infarction from polymorphisms in candidate genes. N Engl J Med. 2002;347:1916–1923
    • View In Article
    • CrossRef
15. Chanock SJ, Manolio T, Boehnke M, et al. Replicating genotype-phenotype associations. Nature. 2007;447:655–660
    • View In Article
    • CrossRef
16. Christensen K, Murray JC. What genome-wide association studies can do for medicine. N Engl J Med. 2007;356:1094–1097
    • View In Article
    • CrossRef
17. Topol EJ, Murray SS, Frazer KA. The genomics gold rush. JAMA. 2007;298:218–221
    • View In Article
    • CrossRef
18. Wellcome Trust Case-Control Consortium . Genome-wide association study of 14,000 cases of seven common disease and 3,000 shared controls. Nature. 2007;447:661–678
    • View In Article
    • CrossRef
19. McPherson R, Pertsemlidis A, Kavaslav N, et al. A common allele on chromosome 9 associated with coronary heart disease. Science. 2007;316:1488–1491
    • View In Article
    • CrossRef
20. Helgadottir A, Thorleifsson G, Manolescu A, et al. A common variant on chromosome 9p21 affects the risk of myocardial infarction. Science. 2007;316:91–93
    • View In Article
21. Samani NJ, Erdmann J, Hall AS, et al. Genomewide association analysis of coronary artery disease. N Engl J Med. 2007;357:443–453
    • View In Article
    • CrossRef
22. Larson MG, Atwood LD, Benjamin EJ, et al. Framingham Heart Study 100K project: genome-wide associations for cardiovascular disease outcomes. BMC Med Genet. 2007;8(Suppl I: S5):1–9
    • View In Article
    • MEDLINE
    • CrossRef
23. Broadbent HM, Peden JF, Lorkowski S, et al. Susceptibility to coronary artery disease and diabetes is encoded by distinct, tightly linked SNPs in the ANRIL locus on chromosome 9p. Hum Mol Genet. 2008;17:806–814
    • View In Article
    • CrossRef
24. Shen G-Q, Shen LL, Rao S, et al. Four SNPs on chromosome 9p21 in a South Korean population implicate a genetic locus that confers high cross-race risk for development of coronary artery disease. Arterioscler Thromb Vasc Biol. 2008;28:360–365
    • View In Article
    • CrossRef
25. Shen G-Q, Rao S, Martinelli N, et al. Association between four SNPs on chromosome 9p21 and myocardial infarction is replicated in an Italian population. J Hum Genet. 2008;53:144–150
    • View In Article
    • CrossRef
26. Smith LR, Harrell FEJ, Rankin JS, et al. Determinants of early versus late cardiac death in patients undergoing coronary artery bypass graft surgery. Circulation. 1991;84(Suppl 5):III245–III253
    • View In Article
    • MEDLINE
27. Horne BD, Camp NJ, Carlquist JF, et al. Multiple-polymorphism associations of seven matrix metalloproteinase and tissue inhibitor metalloproteinase genes with myocardial infarction and angiographic coronary artery disease. Am Heart J. 2007;154:751–758
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (232 KB)
    • CrossRef
28. Third Report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III) final report. Circulation. 2002;106:3143–3421
    • View In Article
29. Allen-Brady K, Wong J, Camp NJ. PedGenie: an analysis approach for genetic association testing in extended pedigrees and genealogies of arbitrary size. BMC Bioinformatics. 2006;7:209–220
    • View In Article
    • MEDLINE
    • CrossRef
30. Pearson TA, Mensah G, Alexander W, et al. Markers of inflammation and cardiovascular disease: application to clinical and public health practice. A statement for healthcare professionals from the Centers for Disease Control and Prevention and the American Heart Association. Circulation. 2003;107:499–511
    • View In Article
    • CrossRef
31. Schunkert J, Goetz A, Braund P, et al. Repeated replication and a prospective meta-analysis of the association between chromosome 9p21.3 and coronary artery disease. Circulation. 2008;117:1675–1684
    • View In Article
    • CrossRef
32. Lp-PLA2 Studies Collaboration. Collaborative meta-analysis of individual participant data from observational studies of Lp-PLA2 and cardiovascular diseases. Eur J Cardiovasc Prev Rehabil. 2007;14:3–11
    • View In Article
    • MEDLINE
    • CrossRef
33. Pencina JJ, D'Agostino RBS, D'Agostino RBJ, et al. Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Statist Med. 2008;27:157–172
    • View In Article