Introduction

Cerebral infarction (CI) is the most common kind of stroke and is a leading cause of severe disability worldwide. It is a complex disease stemming from environmental or genetic risk factors, or their interactions. Genetic factors have been defined as important risk factors for this disease (Hassan and Markus 2000). Increasing data have shown that not any single common genetic variant could independently cause CI. However, the main responsible molecular and genetic determinants were not well established in the previous studies (Hassan and Markus 2000).

Many candidate gene studies about the association between the variants of a targeted gene and CI have been performed, but the results are inconsistent and nonreplicable. Recently, interest has been shifted to the common variants on chromosome 9p21.3 region, which has been confirmed to associate with multiple arterial phenotypes, such as coronary diseases, intracranial aneurysm, abdominal aortic aneurysm, diabetes mellitus, and tumor (Diabetes Genetics Initiative of Broad Institute of H., Mit, L.U. et al. 2007; Helgadottir et al. 2008; Helgadottir et al. 2007; Nakaoka et al. 2014; Ng et al. 2014), and the results have been replicated. However, the association of 9p21.3 and CI is controversial and has not been fully verified. The positive results are found in European and Swedish populations (Olsson et al. 2011), weakly positive in Icelandic population (Helgadottir et al. 2008), and negative in Belgian and US populations (Lemmens et al. 2009; Zee and Ridker 2007). In China, the results are also conflicting. In the study of Ding et al., no single-nucleotide polymorphism (SNP) was found to be associated with stroke (Ding et al. 2009), whereas positive results were found in other studies (Hu et al. 2009; Zhang et al. 2012). Thus, verification of the association of 9p21.3 and CI in China is warranted. In this context, we performed the case-control association study of the common variants in 9p21.3 region and cerebral infarction in a Chinese Han population.

Subjects and Methods

Study Population

In the present study, the participants of the case group were the patients who presented with first-onset cerebral infarction and admitted in Jinling Hospital between January 2009 and June 2010. The demographic and clinical data of all patients were registered prospectively in Nanjing Stroke Program. The patients with history of intra- or extra-cranial aneurysm, cerebral hemorrhage, subarachnoid hemorrhage, cerebral sinus and venous thrombosis, tumors, coronary artery disease (CAD), and uncompleted clinical data were excluded. Cerebral infarction was defined as local neurologic deficit lasting more than 24 h and confirmed by computer tomography or magnetic resonance imaging. The participants of the control group were selected from the health examination center of Jinling Hospital, and they were confirmed to be neurologically normal. The data of all participants, including age, sex, body mass index (BMI, calculated as weight in kilograms divided by height in meters squared), and history of hypertension, diabetes mellitus, hyperlipemia, smoking, alcohol abuse, and atrial fibrillation (AF), were collected. Each participant was measured for total cholesterol, triglyceride, and glucose levels from venous blood after fasting for at least 8 h. The study was approved by the local Institutional Review Board and written consent was obtained from all the participants.

Single-Nucleotide Polymorphism Selection

We prioritized SNPs taking into account a replication study of East Asians, with a minor allele frequency (MAF) of >30 % in Chinese (www.hapmap.org) and associations with cerebral infarction. Although the MAFs of rs3731245 and rs1537378 are less than 30 %, we included these SNPs in our study, as rs3731245 was initially found in a study of a Chinese population and rs1537378 was found in a meta-analysis study (Anderson et al. 2010; Hu et al. 2009). A total of eight SNPs were included in order to investigate the association between the following genotypes in chromosome 9p21.3 region and the incidence of cerebral infarction in a Chinese population: rs10757274, rs10757278, rs2383206, rs2383207, rs1333049, rs1004638, rs3731245, and rs1537378. Genomic DNA was isolated from the peripheral blood collected in an EDTA-K2 tube using a commercially available kit (Bioteke Corporation, Beijing, China). DNA was quantified and diluted to a final concentration of 10 ng/μL.

Genotype Analysis and Quality Control

All samples were genotyped using the TaqMan 7900HT Sequence Detection System (Applied Biosystems) according to the manufacturer’s instructions, performed in the Department of Epidemiology and Biostatistics School of Public Health, Nanjing Medical University. Each assay was carried out using 10 ng DNA in a 10-μL reaction consisting of a TaqMan universal polymerase chain reaction master mix (Applied Biosystems, Foster City, CA), forward and reverse primers, and 6-carboxyfluorescein (FAM)- and 4,7,2-trichloro-7-phenyl-6-carboxyfluorescein (VIC)-labeled probes designed by Applied Biosystems (ABI Assays-on-Demand). Allelic discrimination was measured automatically using the Sequence Detection Systems 2.1 software (autocaller confidence level 95 %). For quality control of genotyping, all the DNA samples of case and control subjects were run in the same batches, which included two case and two control replicated samples and two negative target control (NTC) samples. No mismatches were identified. In addition, an assay with a call rate below 95 % was repeated on a fresh DNA aliquot. If the call rate persisted below 95 %, then the sample was excluded from further analysis.

Statistical Analysis

A statistical power analysis was performed for power and sample size computations for case-control design (Dupont and Plummer 1990). The level of linkage disequilibrium was indicated, in our article, by D’. The presence of Hardy-Weinberg equilibrium of the SNPs was examined using Haploview 4.2, which is based on χ2 goodness-of-fit test. Haplotype frequencies for various SNP combinations were first estimated by haplo.stats (Gonzalez et al. 2007) (version 1.6.11) for R statistical package and then verified using Haploview 4.2. Both of the aforementioned software use the expectation-maximization algorithm when constructing the haplotypes. The haplo.stats program helps to compute global score and haplotype-specific score P values while allowing for adjusting covariates under an additive model using default settings.

Continuous data were expressed as mean ± SD and were analyzed using an unpaired t test or ANOVA. Differences between proportions were assessed with the χ2 test or Fisher’s exact test. In this study, we first applied the χ2 or Fisher’s exact test for an initial screening, and then followed by the multiple unconditional logistic regression model for verification with more rigorous evaluations. All association analyses were conducted in three genetic models: dominant, recessive, and additive. Gender, age, BMI, hypertension, diabetes, hyperlipidemia, smoking status, and AF were entered into the multivariable logistic regression model. Statistical analysis was performed with SPSS 13.0 (SPSS Inc, Chicago, IL). In our haplotype analysis, global score tests were applied to evaluate overall haplotype frequency differences between cases and controls, whereas the haplotype-specific score tests were performed to test individual haplotype difference between cases and controls, allowing for adjustment of covariates.

To minimize the false positive results generated from multiple statistical testing in our analysis, we adopted a method proposed by Story and Tibshirani to estimate the false discovery rate–based Q value using QVALUE software (setting [lambda] = 0, false discovery rate level = 0.05) (Storey and Tibshirani 2003).

Results

Power Analysis

Before implementation of this study, we performed a statistical power analysis using the PS program to verify whether the recruited samples could provide adequate power in identifying the association between modest-effect-size SNPs and CI, provided that the chromosome 9p21 locus confers the same size of risk for development of CI in a Chinese population. Under the population parameter settings of the effect size of odds ratios of 1.25 and the allelic frequency of 0.45 derived from the previous studies, our samples with 769 well-characterized CI cases and 682 healthy controls with evidenced lack of CI can provide a statistical power of 87.1 % at the nominal type I error rate of 0.05. The power analysis indicates that our sample size is sufficient for identifying the chromosome 9p21 CI locus.

Characteristics of the Study Population

During the study, 838 patients with first-onset cerebral infarction were admitted in our institute. Sixty-nine patients were excluded, which included 28 patients with the history of intracranial hemorrhage, 8 with subarachnoid hemorrhage, 11 with tumors, 12 with unruptured intra- or extra-cranial aneurysms, 4 with venous stroke, and 6 for uncompleted clinical data. In this study, all of the 1451 participants involving 769 patients and 682 controls were analyzed.

The characteristics of baselines and traditional risk factors between case and control groups are listed in Table 1. There were no significant differences in age or sex between the CI patients and controls. The mean ages of the CI and control population were 59.91 years (±13.11 years) and 59.37 years (±11.53 years). In the CI patients, the body mass index and the frequency of hypertension, diabetes mellitus, hyperlipemia, and arterial fibrillation tended to be higher than those observed in the controls. All the SNPs did not deviate significantly (P > 0.3) from the Hardy-Weinberg equilibrium (HWE) (Supplementary Table S1).

Table 1 Characteristics of the baseline and traditional risk factors for cerebral diseases between the case and control groups
Full size table

Association of Studied Polymorphisms with Stroke

All SNPs did not deviate significantly from the Hardy-Weinberg equilibrium in cases and control subjects (all P > 0.05) (not shown). In an allelic association analysis, rs2383207, rs3731245, and rs1537378 were significantly associated with CI; the odds ratios were 1.18 (95 % confidence interval (CI) = 1.01–1.37, P = 0.04), 1.29 (95 % CI = 1.06–1.56, P = 0.01), and 1.30 (95 % CI = 1.05–1.60, P = 0.02), respectively. In univariable regression model, rs3731245 and rs1537378 were significantly different in an additive model, and rs3731245 was significantly different in a recessive model. In addition, our unconditional multivariate logistical analyses further demonstrated that rs1537378 remains significantly associated with CI independent of traditional cerebrovascular risk factors (age, sex, BMI, hypertension, DM, hyperlipemia, smoking, alcohol abuse, and AF) in a recessive model (odds ratio (OR) = 1.35, 95 % CI = 1.06–1.71, P = 0.013, Q = 0.03) and in an additive model (OR = 1.38, 95 % CI = 1.11–1.71, P = 0.004, Q = 0.02), and rs2383207 (OR = 1.28, 95 % CI = 1.03–1.59, P = 0.02, Q = 0.03) and rs3731245 (OR = 1.31, 95 % CI = 1.05–1.65, P = 0.02, Q = 0.03) were significantly different in a recessive model (Table 2).

Table 2 Association of eight SNPs on Chr9p21.3 with cerebral infarction
Full size table

Haplotype Analysis for Genetic Association Between the Eight SNPs on 9p21 and CI

An extended SNP haplotype analysis was conducted to provide some insights into the relationship between the SNP patterns and CI that is beyond what single-point SNP analysis can reveal. Using the genotypes of 682 controls, we defined the haploblock structure of SNPs within the region of 9p21 in a Chinese Han population. By defining a solid spine of linkage disequilibrium (LD) as D’ > 0.90, we identified hyploblocks (Fig. 1). Each haplotype was treated as a single variant to test its association with the disease trait. The results are shown in Table 3. Cerebral infarction was restricted to SNPs in the block as manifested by the global P values (P = 5.85 × 103). When GGAGG haplotype was chosen as the baseline for adjusting conventional risk factors, the protective effect for haplotype AATAA remained significant (AATAA vs GGAGG; OR = 0.87, 95 % CI = 0.73–1.00, P = 2.99 × 103, Q = 2.15 × 103) (Table 3).

Fig. 1
figure 1

Annotated genes and linkage disequilibrium (LD) plot of eight SNPs (rs10757274, rs10757278, rs2383206, rs2383207, rs1333049, rs1004638, rs3731245, and rs1537378) on the chromosome 9p21. LD structure and haplotype blocks between eight SNPs were derived from the genotypes of 682 healthy Chinese Han patients. Linkage disequilibrium structure of the locus (D’) was generated using the solid spine of LD method by Haploview 4.2. Haplotype blocks derived from these genotypes using the solid spine LD setting are outlined in black in the D’ chart

Full size image
Table 3 Assessment of association between haplotypes with IS
Full size table

Discussion

This study demonstrated a significant association between SNPs on chromosome 9p21.3 and the risk for cerebral infarction in a Chinese population, which is independent of traditional risk factors of cerebrovascular diseases. The risk alleles are located at CDKN2A and ANRIL genes. Among the eight SNPs, five SNPs were in a strong LD block. Haplotype profiles on chromosome 9.21 were significantly different between control and case of CI.

CI shares similar risk factors and pathophysiologic mechanism with myocardial infarction. Although gene variants on 9p21.3 associated with CAD have been established, the association of 9P21.3 and CI was not consistent. Several studies have examined the role of genetic variants on 9p21.3 in CI. Most previous studies suggested that variants on chromosome 9p21.3 were associated with CI (Dichgans et al. 2014; Heckman et al. 2013; Hu et al. 2009; Luke et al. 2009; Smith et al. 2009; Zhang et al. 2012). Two studies found no association (Lemmens et al. 2009; Zee and Ridker 2007). Hu et al. showed that 9p21 is a shared susceptibility locus, strongly for CAD and weakly for cerebral infarction in a Chinese Han population (Ding et al. 2009). These findings were confirmed by meta-analysis (Anderson et al. 2010; Ni and Zhang 2014). Our findings verified the association of three variants rs2383207, rs3731245, and rs1537378 with susceptibility. The study in a Chinese Han sample by Hu et al. showed that rs2383206 and rs3731245 have a significant association with stroke (Hu et al. 2009). Zhang et al. study showed that rs10757278 and rs10757274 increased risk for stroke, but rs2383206 did not reach statistical significance related to the risk of stroke (Zhang et al. 2012). The specific variants within Chr9p21.3 may represent either the same signal or different signals that give similar effects.

The reasons why the 9p21 locus has different genetic risks for CI remain unclear. The mechanisms of cerebral infarction are complicated and heterogeneous, but atherosclerotic disease is the major and initial cause of the noncardioembolic CI. There is evidence that carotid intima-media thickness and plaque are largely controlled by genetics (Fox et al. 2003). A population-based prospective study shows that the sequence variation on chromosome 9p21 influences atherosclerosis development and progression (Ye et al. 2008). In a recent study recruiting subjects from the Atherosclerosis Risk in Communities (ARIC) study, the sequence variation on chromosome 9p21 is associated with carotid atherosclerosis (Yamagishi et al. 2009).

In the region of chromosome 9p21, three candidate tumor suppressor genes have been identified: cyclin-dependent kinase inhibitor 2A (CDKN2A) gene, cyclin-dependent kinase inhibitor 2B (CDKN2B) gene, and alternate reading frame (ARF). A large antisense noncoding RNA in the INK4 locus (ANRIL) gene has also been identified within the CDKN2B/CDKN2A/ARF gene cluster (Pasmant et al. 2011). These genes play an important role in the regulation of the cell cycle. Disease-associated SNPs at the 9p21 locus predominantly affect the expression of ANRIL (Congrains et al. 2012). Cell proliferation is an important event in atherosclerosis, and these SNPs may alter expression of one of the numerous ANRIL spliced transcripts, which, in turn, might affect cellular proliferation pathways (Jarinova et al. 2009). Moreover, expression of ANRIL transcripts was directly correlated with severity of atherosclerosis (Holdt et al. 2010). These genes also play a central role in the regulation of the cell cycle and may be implicated in the pathogenesis of atherosclerosis through their role in transforming growth factor (TGF)-β–induced growth inhibition (Hannon and Beach 1994; Kalinina et al. 2004).

The present study includes only individuals from the same geographic region in Eastern China. Some limits of the present study should be considered. First, this is a hospital-based case-control study, and only surviving patients were involved. Second, the controls were assessed for stroke history and neurologic function, but imaging screens for excluding silent cerebral infarction were not performed. Third, the sample size in the present study is moderate, and type II error for negative association between 9p21 genetic variant and stroke is possible. Some associated loci may not be found.

In conclusion, the study replicates the association of SNPs on chromosome 9p21.3 with CI in a Chinese Han population. Further studies are necessary to confirm the relation of different genotype-phenotype associations.