Platelet glycoprotein Ibα Kozak polymorphism is associated with an increased risk of ischemic stroke

Ischemic stroke is a frequent cause of death worldwide and remains a common cause of persistent disability.1,2 A large number of risk factors have been identified and prevention strategies, such as smoking cessation, reducing elevated blood pressure, lowering cholesterol, and maintaining good diabetic control, have been successful in reducing the disease burden.3 However, only two thirds of all strokes can be attributed to these modifiable risk factors.4 A family history of stroke is also known to be a predisposing factor,4 and other genetic variations such as platelet and coagulation factor polymorphisms may contribute to a further proportion of strokes.

Platelets are pivotal to the process of arterial thrombosis that leads to ischemic stroke.5 Occlusive thrombus is almost exclusively initiated by plaque rupture and adhesion of platelets to subendothelial von Willebrand factor (vWf) by its specific platelet receptor, the α chain of the glycoprotein (GP) Ib-IX-V complex.6,7 This interaction generates powerful signals leading to platelet activation and further platelet recruitment that ultimately leads to thrombus formation.8,9 Alternatively, thrombus can also develop in partially obstructed atherosclerotic arteries through high-shear-induced binding of fluid phase vWf to platelet GP Ibα, which again activates platelets independent of other agonists.8,9 Platelet activation has been consistently shown to be present in patients with acute ischemic stroke, both at the time of stroke and subsequently in the chronic phase.10 It is clear that ongoing platelet activation is still occurring in these patients despite best medical management with the current antiplatelet agents or oral anticoagulants. One possible reason is that these drugs do not prevent shear-induced platelet activation mediated by GP Ibα and vWf.7,8 

The vWf and GP Ibα interaction is therefore a target for investigation in patients with ischemic stroke. Recently, 3 relatively frequent polymorphisms in GP Ibα have been described in the general population,11 each of which could increase the potential for shear-induced platelet activation by enhancing the efficiency of the binding of vWf to GP Ibα. The Kozak dimorphism detected by the presence of either thymine (T) or cytosine (C) at position −5 from the initiator ATG start codon, influences messenger RNA translation and the amount of GP Ibα on the platelet surface.12 The C allele proportionally increases the amount of GP Ibα expressed on the platelet surface.12 It is plausible that increasing the density of receptor would predispose to enhanced attachment of vWf, causing platelet activation. Two other polymorphisms affect the structure of GP Ibα. One causes a change in the length of the polypeptide by a variation in the number of 13 amino acid tandem repeats (VNTR) in the mucinlike macroglycopeptide region.13-15 The length varies by multiples of either 1 (D allele), 2 (C allele), 3 (B allele), or 4 (A allele).15,16This polymorphism may have important implications for the function of GP Ibα, as each added repeat would position the ligand-binding region further away from the platelet membrane surface, making it more accessible to ligand binding and more susceptible to shear forces. The other GP Ibα polymorphism within the vWf- and thrombin-binding leucine-rich repeat region is based on the presence of threonine or methionine at position 145.16-18 This polymorphism is the basis for the human platelet antigen 2 (HPA-2) (Ko) platelet allo-antigen system.16-18 It has been shown that methionine 145 (HPA-2b) is in linkage disequilibrium with the VNTR alleles A and B.16,18 

The Kozak polymorphism has not been examined in a large number of patients with ischemic stroke and conflicting results have been obtained in 4 reports regarding the importance of the VNTR and HPA-2 polymorphisms in ischemic stroke.19-22 To help clarify this issue, the present study examined the prevalence of the VNTR, HPA-2, and Kozak polymorphisms in a large number of patients with confirmed ischemic stroke classified by etiologic subtype and a similar number of people randomly selected from the electoral roll.

Consecutive patients presenting to a university teaching hospital in Western Australia between March 1996 and June 1998 with a first-ever ischemic stroke were approached for consent to participate in our study that was approved by the Ethics Committee of Royal Perth Hospital. Stroke was defined as a clinical syndrome characterized by rapidly developing clinical symptoms or signs of focal and, at times, global loss of brain function with symptoms lasting more than 24 hours or leading to earlier death and with no apparent cause other than that of vascular origin.23 Ischemic stroke was defined as a stroke with either a normal computed tomography (CT) brain scan or evidence of a recent infarct in the clinically relevant area of the brain on a CT or magnetic resonance imaging (MRI) brain scan performed within 3 weeks of the event or at autopsy.24 Patients with cerebral hemorrhage or cerebral venous thrombosis were not included. Baseline demographic data (age and sex), history of conventional vascular risk factors (hypertension, diabetes, hyperlipidemia, and current smoker), and history of previous vascular events (myocardial infarction, angina, claudication, and amputation) were obtained. All patients underwent a CT brain scan. Echocardiography and extracranial duplex ultrasound were performed at the discretion of the clinician. An overnight fasting blood sample was obtained for biochemical and genetic analyses within 7 days of the acute stroke event.

On the basis of clinical evaluation and results of imaging studies, the study neurologist (G. J. H.) (who remained blinded to the results of GP Ibα genotyping) classified all strokes into 4 major subtypes according to the following predefined criteria.24(1) Large-artery disease included ischemic stroke with (a) evidence of extracranial or intracranial occlusive large artery disease (eg, Doppler or angiographic), (b) no major cardioembolic source (atrial fibrillation, recent myocardial infarction [in the past 6 weeks], endocarditis, or prosthetic heart valve), and (c) clinical opinion that the most likely cause of brain infarction was atherothrombosis involving the aortic arch, carotid arteries, or major branches (main stem middle cerebral artery), or vertebral, basilar, and posterior cerebral arteries. (2) Small-artery disease included ischemic stroke with (a) consciousness and higher cerebral function maintained; (b) one of the classical lacunar syndromes (ie, pure motor hemiparesis, pure hemisensory loss, pure hemisensory-motor loss, or ataxic hemiparesis) or nonlacunar small-artery clinical syndromes (eg, basilar branch artery syndromes); and (c) CT or MRI brain scan, performed within 3 weeks of symptom onset, that was either normal or showed a small deep infarct in the basal ganglia, internal capsule, or brainstem. (3) Cardioembolic disease included ischemic stroke with (a) a major cardioembolic source, (b) no definite evidence of occlusive large artery disease, and (c) clinical opinion that the most likely cause of brain infarction was embolism from the heart. (4) Other included ischemic stroke that did not meet the criteria for one of the categories outlined above (eg, periprocedural, hypoperfusion, dissection, or procoagulant state), or when there was more than one likely explanation (eg, concurrent large-artery occlusive disease and major cardioembolic source).

Control subjects were randomly selected from the Western Australian electoral roll, stratified by 5-year age group, sex, and postal code. A letter of invitation to participate, together with a stamped and self-addressed envelope, was sent to potential controls. Nonresponders were contacted by telephone. Controls who agreed to participate in the study were given the option of attending the hospital outpatient clinic or being visited at home by the study nurse. Baseline demographic data (age and sex), history of conventional vascular risk factors, and history of previous vascular events were obtained for each control. A blood sample was obtained for genetic analysis.

Genomic DNA was prepared from peripheral blood leucocytes by use of a Triton X-100 (Merck, Melbourne, Australia) salt precipitation method.25 Polymerase chain reaction (PCR) was performed by using a Perkin Elmer Cetus DNA thermal cycler (Norwalk, CT). Approximately 100 ng genomic DNA was amplified using 10 μM of each primer, 1-2 U Taq polymerase (Biotech, Perth, Australia), and 20 μM each deoxynucleotide triphosphate in buffer containing 67 mM Tris-HCl (pH 8.8), 17 mM ammonium sulfate, 1.5 mM magnesium chloride, 0.45% Triton X-100, and 0.2 mg/mL gelatin. The DNA fragments were generated from a 35-cycle PCR consisting of 40 seconds at 95°C (denaturing), 40 seconds at 60°C (annealing), and 1 minute at 72°C (extension).

For detection of the Kozak polymorphism, the sequence of the upstream primer was 5′-GAGAGAAGGACGGAGTCGAG-3′ and that of the downstream primer was 5′-GGTTGTGTCTTTCGGCAGG-3′ as previously described.12Samples were restriction digested using 2 U Ppu MI (New England Biolabs, Beverly, MA) at 37°C for several hours. Digestion of the amplified product from T/T produced 3 bands (125 base pair [bp], 157 bp, and 175 bp), from C/C 2 bands (125 and 332 bp), and from heterozygotes C/T 4 bands (125 bp, 157 bp, 175 bp, and 332 bp).

The VNTR polymorphism was detected by using the upstream primer 5′-TCCACTGCTTCTCTAGACAG-3′ and the downstream primer 5′-GGCTGATCAAGTTCAGGGAT-3′.

The HPA-2 polymorphism was detected by allele-specific hybridization, using the common upstream primer 5′-GATGGGACGCTGCCAGTGCTG-3′ with either the downstream primer for Thr, 5′-CTTCTCCAGCTTGGGTGTGGGAG-3′, or the downstream primer for Met, 5′-CTTCTCCAGCTTGGGTGTGGGAA-3′.

All DNA fragments were subjected to electrophoresis on 2% agarose gels and visualized under ultraviolet light after staining with ethidium bromide.

The association of the Kozak, VNTR, and HPA-2 polymorphisms with ischemic stroke was assessed by using a logistic regression model with patient or control as the dependent variable, adjusting for conventional cardiovascular risk factors (smoking, hypertension, diabetes, hyperlipidemia, and previous vascular event). The results were expressed as the odds ratios (ORs) together with their 95% confidence intervals (CIs). Baseline differences between cases and controls were examined by means of the unpaired Student ttest for continuous variables and the chi-square test for categorical data. Differences were considered statistically significant forP values ≤ .05.

Clinical characteristics and vascular disease factor data for our patient and control groups are analyzed in Table1. Age and sex were similar for the 219 cases and 205 controls. As expected, cases were more likely to have had a history of hypertension (P < .001), diabetes (P < .001), smoking (P < .001), and previous vascular event (P < .001).

The genotypic distribution of the patients and controls are shown in Table 2. The prevalence of the heterozygous Kozak T/C genotype was 32.2% in patients, compared with 22.8% among the controls (P < .03; OR, 1.62; 95% CI, 1.03-2.54). There was no difference between cases and controls in the prevalence of the VNTR and the HPA-2 polymorphisms, although there was a trend in the prevalence of the HPA-2a/b in stroke patients (15.0%) compared with that of controls at 9.9% (OR, 1.58; 95% CI, 0.85-2.8;P = .14). Logistic regression was used to adjust for age, sex, and conventional cardiovascular risk factors. The Kozak T/C genotype was found to be an independent risk factor for stroke (adjusted OR, 1.61; 95% CI, 1.00-2.59; P = .05). A similar trend existed for the HPA-2a/b genotype (adjusted OR, 1.8; 95% CI, 0.94-3.4), but it was not significant (P = .07). No significant associations were seen with the VNTR polymorphism.

The distribution of the glycoprotein Ibα genotypes was examined according to the subtypes of stroke as seen in Table3. No association was seen between any of the polymorphisms and the subtype of stroke (large artery, small artery, cardioembolic, and other type of stroke).

This report identifies that the Kozak GP Ibα polymorphism is an independent risk factor for ischemic stroke. The frequency (22.8%) of the T/C genotype in the predominantly Caucasian Australian population was similar to a report of the French Caucasian population (23.1%) and within the frequency of 15% to 30% of 4 other different ethnic populations.12 We found that the T/C genotype was elevated in the stroke group (32.2%) with an unadjusted OR of 1.6 (95% CI, 1.03-2.54; P < .03). This increased relative risk was still apparent even when conventional cardiovascular risk factors such as hypertension, diabetes, hyperlipidemia, smoking, and previous vascular event were adjusted with an OR of 1.6 (95% CI, 1.0-2.59; P = .05). The CC homozygotes were uncommon, and the study was not powered to examine the role of the CC genotype alone. There appeared to be no overrepresentation of the Kozak polymorphism with the particular subtype of stroke.

The reason why the Kozak polymorphism may be important in the pathogenesis of ischemic stroke could be because the presence of the C allele increases the level of GP Ibα on the platelet surface.12 With the use of flow cytometry in washed platelets, the amount of GP Ibα on the surface of the platelet has been shown to be proportional to the amount of the C allele. Although this finding was not supported by another report,26 when the C allele was transfected into Chinese hamster ovary cells, it resulted in a proportional increase in surface GP Ibα expression.12 Compared with the common T/T genotype (100%), the C/C homozygotes expressed the most (157% ± 26%) and the T/C heterozygotes an intermediate amount (128% ± 16%) of GP Ibα.12 Because the GP Ibα molecule is important for platelet adhesion to the vessel wall and binds vWf under conditions of high shear stress,7 it is likely that increasing the relative density of GP Ibα on the surface of the platelet would make platelets more adhesive and possibly also to be more readily activated by shear stress, causing thrombosis and vessel occlusion. It may be one mechanism to explain why patients with ischemic stroke have evidence of ongoing platelet activation that is seen even in the chronic phase10,21,27 and is associated with poststroke mortality.21 

The strengths of our study are that we assembled an inception cohort of more than 200 patients with ischemic stroke and a similar number of community-based controls selected at random from the electoral roll. The diagnosis and etiologic subtype of stroke was made by a single experienced neurologist (G. J. H.) on predefined and accepted objective criteria.23,24 He remained blinded to the GP Ibα results. All causes of stroke were included, particularly those with cardioembolic stroke that can be overlooked and can confound the results of other case-control stroke studies. Another small study of 104 patients could not find a difference in the Kozak polymorphism and cerebrovascular disease.26 However, the recruitment was different because the control group consisted of patients admitted to the hospital with other undefined illnesses and a substantial number of patients had transient ischemic attacks (30%). Patients with transient ischemic attack were excluded from our cohort because they rarely present to the hospital, and there is a greater degree of imprecision in the diagnosis, diluting the focus on confirmed cases of ischemic stroke. Also in our study, a number of other design factors were considered that are likely to reduce bias and to diminish any differences between cases and controls. First, the clinical characteristics and demographics of our cohort are representative of typical cases admitted to a large metropolitan teaching hospital (Table1). They are mostly older than 60 years and frequently have other coexisting and significant cardiovascular risk factors. Second, socioeconomic status is an important cardiovascular risk factor, so we stratified the selection of controls by postal code, which is an established surrogate marker for socioeconomic status. Last, controls were included in the study irrespective of whether they had a past history of vascular disease. Even after adjustment for this bias, using a logistic regression model, the Kozak polymorphism was still a significant independent risk factor for ischemic stroke. In addition there was no significant difference between the frequency of the T/C polymorphism between those with (23.5%) or without (28.7%) vascular disease. It makes any positive finding in our series more significant and generally applicable to all patients with ischemic stroke.

Our study has several potential limitations. First, although cases were classified prospectively and recruited consecutively and controls were randomly selected from the community, potential confounding can never be entirely eliminated in an observational study. Second, the small number of cases among the etiologic subtypes of ischemic stroke may have limited the power of our study to detect potentially important differences in the prevalence of the GP Ibα polymorphisms between these groups.

Other gene polymorphisms of GP Ibα have been described. One is the VNTR that is in linkage disequilibrium with the HPA-2 polymorphism at position 145.16,18 The A and B VNTR variants are associated with methionine 145 (HPA-2b).16,18 We found a trend toward the association between the HPA-2a/b genotype and ischemic stroke that was more apparent when conventional risk factors were considered (adjusted OR, 1.8; 95% CI, 0.94-3.4; P = .07). Previous studies concerning the HPA-2 polymorphism are conflicting, but all show a trend of increased risk with the HPA2a/b polymorphism. The significance of the VNTR polymorphism data is also different in several reported studies. The results are summarized in Table4. Like the present study, one study of 609 stroke patients found no association between the genotype distribution of VNTR in patients and in controls.21 This is further supported by the finding that, although the levels of plasma markers of platelet activation (PF4 and β-thromboglobulin) were generally elevated in the stroke group, no differences could be demonstrated according to VNTR genotype.21 In a small study, there was a link between the B allele and the BC genotype (OR, 2.83) for cerebrovascular disease.19 Like another report21 the BC genotype was uncommon in our stroke population (2.1% control versus 2.4% patient). These differences in the apparent significance of the GP Ibα polymorphisms in stroke are therefore probably due to variations in the background genotype frequencies of these mutations among normal populations of different ethnic backgrounds, to different patient and control recruitment, to variation in the inclusion criteria, or to differences in attributable risks to other cardiovascular factors in the various cohorts.

We found little variation in the prevalence of the GP Ibα polymorphisms and the classification of ischemic stroke according to large-, small-artery, and cardioembolic stroke. One recent study suggests the HPA-2b allele may be more important in those patients with transient ischemic attack (OR, 4.3) followed by lacunar infarction (OR, 2.2) then by atherothrombotic stroke (OR, 1.5).22 However, in that study the distribution of type of stroke was different and the cohort was younger when compared with our group of patients. These differences in case demographics may account for the lack of association in our study. It suggests that larger studies are required to clarify the effect of the GP Ibα polymorphisms in subtypes of ischemic stroke or transient ischemic attack.

Our findings show that the Kozak GP Ibα polymorphism is an independent risk factor for ischemic stroke regardless of etiology. However, further studies are needed to define the role of the other platelet polymorphisms and the interaction of known cardiovascular risk factors with the GP Ibα polymorphisms.