Introduction

A large number of candidate gene association studies have attempted to identify genes involved in stroke.1 The etiology of stroke is heterogeneous, and familial predisposition contributes only moderately (OR=1.3–1.76) to the risk of all stroke.2 One way to increase the power of such studies would be to select a predefined phenotype. Genetic influence has been found to be the strongest in ischemic stroke (IS) due to small-vessel occlusion (SVO) or large-vessel atherosclerosis (LVA), as defined by the Trial of Org 10172 in Acute Stroke Treatment (TOAST) classification. This influence was especially strong in younger (<60 years) patients.3, 4, 5 Thus, middle-aged SVO and LVA stroke patients seem to be a suitable study population for studies attempting to identify genes involved in the pathogenesis of IS.

Despite the association of the APOE ɛ4 allele with elevated LDL cholesterol levels and slightly elevated cardiovascular risk,6 a recent meta-analysis found no clear link between the APOE ɛ4 allele and IS.7 On the other hand, elevated plasma concentration of apoE is a risk factor for stroke.8, 9 Furthermore, it has been shown that apoE concentration is strongly associated with cardiovascular mortality in old age, independently of APOE genotype and plasma lipids.10 Plasma level of apoE is dependent on total protein, albumin level, body mass index (BMI) and alcohol consumption. Moreover, plasma apoE is largely liver-derived and could be regulated by hepatic factors. Besides the APOE ɛ2/ɛ3/ɛ4 genotype affecting the structure of apoE protein, the biological activity of apoE can also be influenced by genetic factors that modify its synthesis and quantity. It has been shown that genetic variation in the APOE gene promoter, the −219G/T single nucleotide polymorphism (SNP) located in the regulatory region of the APOE gene and the closely linked +113G/C SNP located in the intron 1 enhancer region, may affect the transcription of APOE gene.11 Carriers of the −219G/G-genotype had 10–20% higher apoE plasma concentrations compared to −219T/T genotype.12 Moreover, the −219G/T polymorphism has been reported to associate with atherogenic lipid and lipoprotein profile,13, 14 as well as with coronary atherosclerosis.15 However, the role of these SNPs in IS and in cerebral atherosclerosis has not been studied before.

In this study, we tested the association between the APOE ɛ4, −219G/T and +113G/C SNPs and IS in a setting of a case–control study of 237 middle-aged patients and 326 ethnicity- and gender-matched controls. To evaluate whether any positive result may be related to predisposition to intracranial atherosclerosis, we also tested whether these genetic variants are associated with atherosclerosis of the circle of Willis in an independent Finnish autopsy series.

Material and methods

The Belgian stroke study (BSS)

BSS is a case–control study. The 237 cases had SVO or LVA stroke (TOAST classification), all occurring between 45 and 60 years of age. The patients were selected from the databases of seven stroke units in Belgium. All patients were of central European origin (>90% were Belgians). Cardiovascular risk factors (hypertension, diabetes mellitus, hyperlipidemia (hypercholesterolemia or hypertriglyceridemia), alcohol consumption (>20 g/day), smoking (former, current and never), obesity (BMI >30)) were recorded.

The control group was composed of 326 gender- and ethnicity-matched healthy volunteers without a history of stroke and living in the same area. As the mean age of stroke occurrence is typically around 70 years of age, and because we selected only younger (<60 years) stroke patients, we explicitly selected older controls than cases to decrease the likelihood that they would later in their life get an IS. Stroke patients had higher prevalence of conventional cardiovascular risk factors than controls. Therefore, to avoid a potential selection bias due to the different cardiovascular risk factors and differences in the mean age, all our genetic results were adjusted for cardiovascular risk factors and age.

Optimal methods to identify and control for population stratification in genetic association studies are not established. A recent study showed that grandparental country of origin provided better control for stratification than the SNP-based approach.16 In this study, ethnicity was checked up to the four grandparents. The ethical committees of all participating hospitals approved the study protocol. Written informed consent was obtained from all patients before study entry.

The Finnish autopsy series

The Finnish autopsy series comprised of two cross-sectional population-based autopsy studies. In total, the two series included 1004 medico-legal Caucasians autopsy cases that had died suddenly out of hospital, or were found dead. The atherosclerosis of each of the nine branches of the circle of Willis was scored semi-quantitatively as follows: 0=normal, 1=slight (streaks with or without elevated fibrous lesions), 2=moderate (fibrous lesions that cause <50% stenosis), 3=severe (>50% stenosis with extensive atherosclerosis (fatty, fibrous and calcified lesions)), giving a range of scores from 0 to 27. The cases were from two studies: the Tampere Autopsy Study (TASTY) (n=604) and the Helsinki Sudden Death Study (HSDS)17 (n=400). The TASTY comprised both men (64.3%, mean age 59.7 years) and women (35.7%, mean age 68.2 years), whereas the HSDS series included only men (mean age 53.7 years). In addition to gender and age, BMI was also recorded. The Finns are particularly suitable for genetic association studies being a homogenous Caucasian population, which results from genetic isolation.18

Genotyping

In the BBS, DNA was isolated from samples of whole blood, which had been stored frozen at −20°C, with a commercially available kit (Qiagen Inc., Valencia, CA). In the autopsy series, DNA isolation was performed either from frozen blood samples (TASTY) with the salt precipitation method or from frozen (−70°C) cardiac muscle samples (HSDS) with the standard phenol-chloroform method. Genotyping was carried out by using the 5′ nuclease assay and fluorogenic allele-specific TaqMan MGB probes in the ABI Prism 7900 HT sequence detection. The DNA and PCR master mix (together 5 μl) were pipeted into the 384-well plates using Tecan Freedom EVO 100 instrument and instrument software V4.8 (Tecan Schweitz AG, Switzerland). To monitor genotyping errors, random duplicates were run in parallel with unknown samples. Allele-specific fluorescence generated from each probe during the PCR amplification was measured with the allelic discrimination analysis module. The nucleotide sequences of primers and probes used in the PCR were deduced from the public databases and synthesized in conjunction with Applied Biosystems. The SNPs evaluated in this study were APOE ɛ2/ɛ3/ɛ4, −219G/T (rs405509) and +113G/C (rs440446). Because of technical problems in the PCR of the +113G/C SNP and a tight linkage disequilibrium (D′=0.86 r2=0.44; calculated by the Stata 8.0 program, STATA Corporation, TX, USA) between the –219G/T and +113G/C SNPs, only the −219G/T SNP was analyzed from all of the autopsy cases. The mean genotyping success was >95%.

Statistical analysis

Data were analyzed using the SPSS software (version 12.0, SPSS Inc., Chicago, IL, USA). The clinical data were compared between the IS cases and controls, using a binary logistic regression analysis with age as continuous covariate. To examine the effect of the APOE ɛ4 allele, study populations were divided into ɛ4 allele carriers (ɛ3/4 and ɛ4/4) and non-carriers (ɛ2/2, ɛ2/3 and ɛ3/3). Since the ɛ2/ɛ4 genotype was rare (n=18) and difficult to assign in a group (ɛ2 and ɛ4 allele carriers usually have opposite effects), it was excluded from the analyses. A binary logistic regression analysis with smoking, hypertension, alcohol consumption, obesity, diabetes and hyperlipidemia as dichotomous covariates and age as continuous covariate was used to evaluate the association of the genetic variants with IS. To study the association of the promoter polymorphism with IS excluding the confounding effects of the APOE ɛ2/ɛ3/ɛ4 genetic variation, we performed the analysis within the most common APOE ɛ3/ɛ3 genotype group. Furthermore, analysis of covariance (ANCOVA) with age and BMI as continuous covariates was used to compare the mean atherosclerosis scores between the studied genotypes. We analyzed four SNPs that are in linkage disequilibrium and located in the same gene. Therefore, applying a multiple comparison statistics, such as Bonferrroni correction, is inappropriate, because the individual SNPs are not independent.

Results

BSS series

The frequencies of the ɛ2, ɛ3 and ɛ4 alleles were 0.07, 0.80 and 0.14, respectively (ɛ2/ɛ4 was left out from the calculations). Within the APOE ɛ3/ɛ3 carriers, the frequencies of the −219G and −219T alleles were 0.55 and 0.45, and the frequencies of the +113G and +113C alleles were 0.56 and 0.44, respectively. All genotype frequency distributions were in Hardy–Weinberg equilibrium in cases and in controls. Stroke patients had higher prevalence of conventional cardiovascular risk factors, such as hypertension, hyperlipidemia, current smoking and alcohol consumption than controls (Table 1). The APOE ɛ4 carrier frequency did not differ significantly between the IS patients and controls (Table 2). Within the most common APOE ɛ3/ɛ3 genotype group, both the −219G and +113G allele carriers were more common among the IS cases than controls (OR=6.2; 95% CI 1.6–24.3, P=0.009 and OR=7.1; 95% CI 1.7–29.9, P=0.007, respectively) after adjustment for all recorded risk factors and age (Table 2). With a frequency of ∼50% for the at-risk allele at an α level of 0.05, our sample was evaluated to have 95% power to detect an RR of 1.6 for heterozygote and 3.2 for homozygotes (‘genetic power calculator’: http://statgen.iop.kcl.ac.uk/gpc/cc2.html).

Table 1 Clinical characteristics of Belgian stroke study population and odds ratios for ischemic stroke
Full size table
Table 2 Genotype frequencies (%) for Ischemic stroke cases and controls in the Belgian stroke study population
Full size table

Finnish autopsy series

The frequencies of the ɛ2, ɛ3 and ɛ4 alleles were 0.05, 0.79 and 0.17, respectively (ɛ2/ɛ4 was left out from the calculations). Within the APOE ɛ3/ɛ3 carriers, the frequencies of the −219G and −219T alleles were 0.62 and 0.38, respectively. All APOE allele frequencies were similar to the frequencies in the BSS series. The carriers of the APOE ɛ4 tended to have higher intracranial atherosclerosis score (5.4 vs 4.8, P=0.051) compared to non-carriers, but this association was statistically significant only among men (5.4 vs 4.6, P=0.044). Within the APOE ɛ3/ɛ3 genotype group, there was no significant difference in the circle of Willis atherosclerosis score between carriers and non-carriers of the APOE −219G allele. (Table 3)

Table 3 Mean atherosclerosis score in Helsinki Sudden Death Study (HSDS) and Tampere Autopsy Study (TASTY) series
Full size table

Discussion

Several studies have addressed the role of APOE ɛ4 in stroke, but the results have been inconsistent.7 In our selected Belgian stroke population, no association was seen between the APOE ɛ4 allele and IS. However, in the Finnish autopsy study, we found out that men carrying the APOE ɛ4 allele had significantly higher mean intracranial atherosclerosis score compared to the ɛ4 non-carriers. Previous autopsy studies have revealed a link between the APOE ɛ4 allele and larger coronary and aortic atherosclerotic lesion areas in men.17, 19 Thus, it seems that the APOE ɛ4 allele may have a gender-specific role in the development of atherosclerosis in different vascular beds. The ɛ4 allele is known to be associated with high LDL cholesterol level,6 which is an important factor in the early development of atherosclerosis. Therefore, it can be hypothesized that the ɛ4 allele mainly affects the initial stages of cerebral atherosclerosis. In addition, ɛ4 also seems to have an independent role in cerebral small vessel disease, as suggested by its effect on MRI white matter hyperintensities in a study by Hogh et al.20. These authors propose that ɛ4 may interact with other cardiovascular risk factors, such as hypertension, in affecting lipid metabolism and cellular repair mechanism.20 In fact, a previous study showed that APOE ɛ4 has a decreased antioxidant activity compared to other alleles.21

The role of the APOE promoter polymorphisms −219G/T and +113G/C in IS or cerebral atherosclerosis had not been studied before. Our results suggest that G-allele carriers of both polymorphisms are at an increased risk of IS. This association was significant within the most common APOE ɛ3/ɛ3 genotype group, indicating that it is most probably independent of the APOE ɛ2/ɛ3/ɛ4 genotype. The APOE gene promoter polymorphism −219G/T affects the transcriptional activity of the APOE gene, in particular the G-allele is associated with higher transcription than the T-allele.11 Consequently, G-allele carriers have significantly higher plasma concentrations of apoE.12 Recent studies suggested that the apoE level is related to stroke.9, 8 Our results provide the evidence that a genetic determinant of higher levels of apoE increases the risk of IS.

ApoE seems to have a multifaceted role on atherosclerosis and vascular events in humans. Our findings suggest that the association of the G-alleles of both promoter SNPs with the risk of IS would involve mechanisms other than those leading to an accelerated development of stable intracranial atherosclerosis. One hypothesis would be that increased apoE level predispose to an unstable atherosclerotic plaque. This is supported by studies on apoE −/− mice. Although these animals develop severe atherosclerosis, they rarely have spontaneous plaque rupture and thrombosis.22 Furthermore, a recent study showed that increased apoE deposits in early atherosclerotic lesions distinguish symptomatic from asymptomatic patients.23 A second mechanism may be related to the finding that apoE has proinflammatory properties.10, 24 Elevated apoE level lead to chronic inflammation that may contribute to atherosclerosis. Another possibility is that apoE influences the risk of IS by a mechanism independent of atherosclerosis. ApoE is known to play a role in the coagulation pathway, in particular, by affecting vitamin K1 metabolism. Vitamin K1 is a chylomicron-bound essential cofactor for the synthesis of several blood coagulation factors. Chylomicron uptake and clearance are affected significantly by APOE genotype,25 leading to marked fluctuations in plasma concentrations of vitamin K1.26 Low vitamin K1 levels have been measured, especially in patients carrying the ɛ4 allele26 and are associated with a poor outcome in hemorrhagic stroke patients.27 Thus, possible interactions between APOE variants and vitamin K1 metabolism might be related to the risk of IS independently of the development of atherosclerosis.

A possible explanation of the observed difference of APOE effect on cerebral atherosclerosis in men and women could be related to the gender-dependent effect of apoE isoforms on immune system activation28, 29 or to the influence of sex hormones on apoE protein production.30

In summary, this study revealed a multifaceted role of APOE gene on IS and subclinical intracranial atherosclerosis. The APOE ɛ4+ genotype was associated with the severity of subclinical intracranial atherosclerosis in men, but was not a predictor of IS. On the other hand, the promoter variants affecting apoE synthesis were significant predictors of IS, suggesting that quantitative rather than qualitative variation of apoE is related to IS.