Background Genetic variants on chromosome 9p21 confer a robust risk for coronary artery disease but inconsistent risk for stroke. This study investigated whether such genetic variants exert differential risks on myocardial infarction (MI) and ischemic stroke in a Taiwanese population.
Methods The study recruited 425 MI patients, 687 patients with ischemic stroke, and 1377 healthy controls. Four key single nucleotide polymorphisms (SNPs) on chromosome 9p21 were genotyped.
Results Multivariate permutation analyses demonstrated that the risk T allele of rs1333040 and G allele of rs2383207 were associated with MI (P = 0.045 and 0.002, respectively). Subjects with the rs2383207 GG genotype had a 1.85-fold (P = 0.021) risk for MI when compared with the subjects with the AA genotype. Further analysis showed that significant results only exist in the young MI group (<65 years) but not in the old MI group (≥65 years). These SNPs were not statistically significant for stroke (adjusted P ranged from 0.097 to 0.540). Haplotype analysis showed global P values of 0.032 for MI and 0.290 for stroke.
Conclusions Genetic variations in the 9p21 region are associated with MI but not with stroke in a Taiwanese population. Early-onset MI was more likely to carry the risk genotypes of 9p21 SNPs.
Stroke and coronary artery disease (CAD) together are the leading causes of death and disability worldwide. Although conventional risk factors such as hypertension, diabetes, smoking, and abnormal lipid metabolism can account for the diseases susceptibility,1,2genetic risk factors have also been implicated in the pathogenesis of CAD and stroke.3,4
Recent independent genome-wide association analysis has identified genetic variants on chromosome 9p21 associated with CAD and myocardial infarction (MI) in several white cohorts.5-7Within a short period of its identification, the risk genetic variants in this region have been shown to exert a strong association with CAD in several populations,8-10including East Asians.11-13To date, no known protein-coding genes reside within the chromosome 9p21 risk locus, except for a long noncoding transcript, designated antisense noncoding RNA in the INK4 locus.14The expression levels of this noncoding RNA have been associated with atherosclerosis.15The genetic variants on 9p21 also reportedly increase the risk of various atherosclerotic conditions in recent clinical studies.16-18Because both ischemic stroke and CAD can be caused by atherosclerosis and thrombosis, chromosome 9p21 genetic variants may also be associated with ischemic stroke.
However, the chromosome 9p21 has been identified susceptible for ischemic stroke in some19,20but not in all studies.21,22Matarin et al.19have reported evidence that the haplotype on chromosome 9p21 is associated with 1.4- to 1.7-fold risk for ischemic stroke in the United States. In an Icelandic population study, Helgadottir et al.18have found that the risk genetic variants have very mild association with ischemic stroke (odds ratio [OR], 1.15; P = 0.015), but the association is attenuated after excluding CAD cases. Other studies in the US white and Belgian populations fail to show an association between ischemic stroke and polymorphisms on chromosome 9p21.21,22Similar studies in Asian populations, as well as stroke subtype specificity, are warranted.
The present study aimed to examine whether chromosome 9p21 genetic variants confer susceptibility to MI and stroke in a Taiwanese population. Because early-onset MI was attributed to the 9p21 genetic predisposition,6,23the present study also assessed whether age can modify the genetic effect on MI and ischemic stroke.
MATERIALS AND METHODS
Six hundred eighty-seven patients with ischemic stroke were diagnosed based on the World Health Organization criteria24and admitted to the stroke ward of Kaohsiung Municipal Hsiao-Kang Hospital from August 2006 to July 2009 were included. All patients had a standard stroke investigation that included laboratory examination and cranial computed tomography or magnetic resonance imaging. Ischemic stroke was classified when brain imaging revealed acute infarction or showed no evidence of hemorrhage. Stroke subtypes were classified based on the Trial of ORG 10172 in Acute Stroke Treatment.25
Four hundred twenty-five MI patients, including both inpatients and outpatients with documented MI in the Cardiology Department of Kaohsiung Medical University Hospital, were enrolled. The diagnosis of MI was based on typical chest pain more than 30 minutes in duration, characteristic electrocardiographic patterns of acute MI, and significantly elevated cardiac enzymes.
Between January 2006 and December 2008, 1377 stroke- and MI-free subjects were recruited through an advertisement solicitation at the Kaohsiung Medical University Hospital. Each control subject filled a self-administered questionnaire that included sociodemographic information, medical history, and medication data.
For each participant, total cholesterol, triglycerides, and glucose levels were measured from venous blood after fasting for at least 8 hours. Hypertension was defined as systolic or diastolic blood pressure of 140/90 mm Hg or higher or antihypertensive medication use. Diabetes was defined as fasting blood glucose of 126 mg/dL or higher or known treatment of diabetes. Hypercholesterolemia was defined as serum levels of total cholesterol of 240 mg/dL or more or the use of lipid-lowering medication. The local institutional review board approved the study, and every participant provided written informed consent.
Genomic DNA was isolated from whole blood using commercially available kits (Puregene; Gentra, Research Triangle, NC). Four single nucleotide polymorphisms (SNPs; rs10757278, rs2383207, rs1333040, and rs1333049) at chromosome 9p21 were selected based on previous studies that consistently reported these SNPs influencing the risk for CAD and stroke.6,7,19
Genotypes were determined using the TaqMan technology (Applied Biosystems, Foster City, CA). Briefly, polymerase chain reaction primers and TaqMan Minor Groove Binder probes were designed, and reactions were performed in 96-well microplates with ABI 9700 (Applied Biosystems) thermal cyclers. Fluorescence was measured with an ABI 7500 Real-Time PCR System (Applied Biosystems) and analyzed with its System SDS software version 1.2.3. The genotyping success rates of the 4 SNPs ranged from 92.3% to 99.5%.
Allele frequencies at each SNP were estimated by direct genotype counting. Hardy-Weinberg equilibrium was tested using the χ2 test. Continuous variables were presented as mean (SD), and dichotomous variables were presented as n (%).
Assuming the minor allele of each SNP has an additive effect, analysis for an association of the studied SNPs with the phenotypes of interest (MI and stroke as a whole and stroke subtypes) was done using the χ2 test. Multivariate logistic regression analysis was done to estimate ORs and their corresponding 95% confidence intervals (CIs), with adjustments for sex, age, smoking, hypertension, diabetes, and hypercholesterolemia. In addition, the age-specific effect was also assessed by analyzing young (<65 years) and old age (≥65 years) data separately. The SPSS 13.0 version for Windows (SPSS Inc, Chicago, IL) was used for statistical analysis.
Linkage disequilibrium (LD) was assessed using Haploview, with blocks graphically identified from the LD intensity expressed in D'.26For haplotype construction, the package Hap-Clustering of R software (version 2.9.0; free, open-source toolset download from website: http://www4.stat.ncsu.edu/∼jytzeng/Software/Hap-Clustering/R)27was used to infer haplotypes by combining the study SNPs and estimating the haplotypephenotype association. Multiple testing corrections were done using the max (T) permutation procedure 10,000 times to obtain empirical P value by the PLINK software (free, open-source toolset download from: http://pngu.mgh.harvard.edu/purcell/plink).28Two-tailed P < 0.05 was considered statistically significant.
The general characteristics of the study subjects were presented in Table 1. As expected, traditional risk factors such as hypertension, diabetes, smoking, and hypercholesterolemia were significantly different between cases and controls. For stroke subtypes, there were 126 cases (18.3%) of large-artery atherosclerosis, 66 cases (9.6%) of cardioembolism, 300 cases (43.7%) of small vessel occlusion, 18 cases (2.6%) of other determined etiology, and 177 cases (25.8%) of undetermined etiology by the Trial of ORG 10172 in Acute Stroke Treatment classification.25Observed genotype frequencies of the 4 selected SNPs (rs1333040, rs2383207, rs10757278, and rs1333049) did not deviate from the Hardy-Weinberg equilibrium. All 4 SNPs were in the same LD block. Two SNPs (rs10757278 and rs1333049) had high r 2 of 0.95, and thus, only rs1333049 was presented hereafter.
Associations of the studied SNPs with MI are shown in Table 2. Univariate analyses indicated that SNPs rs1333040 and rs2383207 were significantly associated with MI (P = 0.004 and 0.001, respectively). After adjusting for conventional risk factors, the risk T allele of rs1333040 and the risk G allele of rs2383207 remained significantly associated with MI (adjusted P = 0.045 and 0.002, respectively). Subjects with the GG genotype of rs2383207 had a 1.85-fold risk for MI (95% CI, 1.13-3.10; adjusted P = 0.021) when compared with those with the AA genotype. Subjects with the AG genotype of rs2383207 carried a 1.20-fold risk for MI, yet with a borderline significance. Similarly, the TT genotype frequency of rs1333040 in the MI group was higher than the controls, although the difference had borderline significance (adjusted OR, 1.66; 95% CI, 0.94-3.02; P = 0.082).
To examine age-specific effects, the participants were stratified by age. Compared with controls, the frequency of risk T allele of rs1333040 and the risk G allele of rs2383207 were significantly higher in the young MI group (<65 years; adjusted P = 0.032 and 0.005, respectively) but not in the old group (≥65 years; adjusted P = 0.501 and 0.108, respectively; Table 3). A higher genetic risk for homozygous carriers of G allele at rs2383207 was also noted in the young MI group than in the old MI group, with an OR of 2.18 and 1.61, respectively.
Similar to MI, stroke subjects had a higher frequency of the risk allele in SNPs rs1333040 and rs2383207 compared with control subjects. However, the difference was not significant by univariate analysis (P = 0.135 and 0.057, respectively). Even after adjusting for potential confounders, there was still no significant association (Table 2). Further age-specific analysis showed that neither the young nor the old stroke group was associated with the studied SNPs (adjusted P ranged from 0.134 to 0.906; Table 3). Because previous studies reported an association for large-artery stroke subtype,29,30certain stroke subtypes were investigated for association with genetic variants on 9p21. However, there was no significant relationship between any of the studied SNPs and stroke subtypes (data not shown). For the large-artery atherosclerotic stroke subtype, the difference of genotype distributions of the studied SNPs between cases and controls revealed an adjusted P value from 0.368 to 0.924.
The association of haplotypes comprising 3 SNPs (rs1333040, rs2383207, and rs1333049) was further investigated and revealed associations with MI (global P = 0.032) but not with stroke (global P = 0.290) after adjustment for covariates. A global adjusted P = 0.021 for MI was noted for haplotypes comprising the 2 most significant SNPs (rs13330040 and rs2383207). Compared with other haplotypes, the risk TG haplotype had higher risk for MI, with an adjusted P = 0.006.
In this study, the association of chromosome 9p21 genetic variants with MI is reproduced in a Chinese population residing in Taiwan. Our results provide evidence that the 9p21 risk locus confers higher risk for the young-onset MI. Although ischemic stroke shares similar risk factors and pathophysiologic mechanism with MI, there is no evidence to support the association between these genetic variants and stroke in the study population. The 9p21 genetic risk is also not related to any particular stroke subtype.
Although previous studies have tested different SNPs, associations between the 9p21 genetic variants and the risk of CAD or MI have been consistently reported in multiple populations, including East Asians.12,13,31In a Japanese study that enrolled 2475 controls and 589 MI cases, Hiura et al.13noted that SNP rs1333049 on 9p21 was associated with a 1.47-fold risk for MI. In a Korean study of 611 cases and 294 controls, SNPs rs10757274, rs2383206, rs2383207, and rs10757278 on 9p21 were all associated with CAD.12Similar significant results were reported in a couple of studies conducted in China using various phenotypes including MI,11,32CAD,31and early-onset CAD.33These Chinese studies showed that the phenotypes were consistently related to several 9p21 SNPs including rs2383207,31,33rs10757274,11rs2383206,11rs10757278,11,33and rs1333049.32The present study also observes significant association between rs1333040 and rs2383207 and MI in Chinese residing in Taiwan. Furthermore, we have recently shown the association of 9p21 genetic variants with subclinical atherosclerosis, including carotid intima-media thickness and plaque.34Taken together, there is substantial evidence that SNPs on chromosome 9p21 cause genetic predisposition not only to subclinical atherosclerosis but also to solid clinical events.
The impact of genetic factors on MI has been shown to be greater in the early ages.35In an Icelandic population study by Helgadottir et al.,6the corresponding risk at the 9p21 genetic variant for MI is 1.64-fold in all study subjects, but the risk is doubled for the early-onset MI subgroup. A recent New Zealand cohort study shows that participants with the homozygous risk genotype at rs1333049 experienced MI at an earlier age than noncarriers.23The present study successfully demonstrates that the 9p21 genetic variants are associated with the young MI group (<65 years) but not with the old MI group (≥65 years). Furthermore, the homozygous genotype (GG) at rs2383207 has a 2.18-fold MI risk in the young age group that is stronger than the old age group (OR, 1.61). These results confirm that the 9p21 genetic variants may have more prominent influence on early-onset MI than the late-onset MI. Consistent with our results, Chen et al.33have shown similar early-onset genetic specificity results in a Chinese population. However, the current study is different from Chen et al. First, the phenotype they observed is CAD and not MI. Second, their study only enrolled young-onset cases, and it is difficult to compare the age-differential effect in such a homogenous range of ages.
Recently, the relationship between ischemic stroke and the 9p21 genetic variants in various ethnic groups yields conflicting studies. The results are positive in European and Swedish populations,29,30negative in Belgian and US populations,21,22and weakly positive in the Icelandic population.18Similar to a recent Chinese study,36the current study does not find any association between genetic polymorphisms and ischemic stroke. The 9p21 race-specific genetic effect may add to this complexity.
The reasons why the 9p21 locus has different genetic risks for cardiovascular disorders (ie, significant risk in MI but not in stroke) remain unclear. Although most cardiovascular disorders share similar pathogenesis, genetic heterogeneity may complicate the pathophysiology. The pathophysiology of cardiovascular disorders involves aspects of altered coagulation, vascular inflammation, oxidative stress, endothelial dysfunction, plaque stability, and so on. Thus, differential processes may account for ischemic stroke and MI. This may be further supported by findings that biomarkers such as C-reactive protein or lipid levels have different predictive values on ischemic stroke and CAD.37
The present study includes only individuals from the same geographic region in southern Taiwan. Furthermore, no detectable population stratification was found in part of our cohort in our previous study.38Because the 9p21 genetic risk is stronger for MI and weaker for stroke according to previous studies, comprehensive data on concurrent MI (eg, separate MI, stroke, or non-MI, nonstroke) are another advantage of the current study. However, the study has some limitations. First, it is a hospital-based case-control study, and only surviving patients are included. Second, the sample size in the present study is moderate, and type II error for negative association between 9p21 genetic variant and stroke is possible. Nevertheless, based on post hoc power analysis, the current sample size still provides a power of 82% and 93% to detect the risk allelic effect of rs1333040 and rs2383207, respectively, for stroke.
In conclusion, the association of chromosome 9p21 genetic variants with MI is shown in a Taiwanese population. The genetic risks are higher in the young-onset MI group. However, the risk locus is not related to ischemic stroke, either stroke as a whole or in any particular subtype. Further studies are necessary to disentangle the mechanisms of different genotype-phenotype associations.