Homozygosity for 807 T Polymorphism in α2 Subunit of Platelet α2β1 Is Associated With Increased Risk of Cardiovascular Mortality in High-Risk Women
Background—Platelet adhesion to collagen is the initial step in both hemostasis and thrombosis; this adhesion is mediated by α2β1 on the surface of platelet membranes. An 807 C to T single nucleotide exchange polymorphism close to the gene coding for the α2 subunit of α2β1 is associated with the density of α2β1 on the platelet membrane.
Methods and Results—We studied the relation of the α2β1 807 C/T genotype to cardiovascular mortality in a prospective cohort study of 12 239 women who were invited for the breast cancer screening program of Utrecht, the Netherlands. The initial age was between 52 and 67 years. Women were followed on vital status between 1976 and 1995 (168 513 women-years). Data were analyzed by using a nested case-control design. The α2β1 807 C/T genotype was not associated with cardiovascular mortality in the total population: the rate ratio for cardiovascular mortality in 807 TT homozygotes compared with 807 CC wild types was 1.2 (95% CI 0.8 to 1.7). However, the α2β1 807 T polymorphism was associated with an increased risk of cardiovascular mortality in women who smoked or in women who had indications of compromised endothelium, such as diabetes and microalbuminuria. In those who were exposed to ≥2 of these factors, the risk ratio (95% CI) between α2β1 807 TT homozygotes and 807 CC wild types was 14.1 (5.0 to 39.9).
Conclusions—α2β1 807 TT homozygosity, coding for increased α2β1 density on the platelet membrane, is associated with an increased risk of cardiovascular mortality in those women with indications of compromised endothelium.
Hemostasis and thrombosis are initiated by the adhesion of platelets to collagen.1 Platelet adhesion to collagen is a multistep process depending on the availability and function of von Willebrand factor (vWF), collagen, and the platelet integrins glycoprotein Ib and α2β1. Briefly, vWF binds to collagen in the subendothelium. Glycoprotein Ib on the platelet surface can interact with vWF, which is attached to collagen.2 3 This reduces the velocity of circulating platelets and allows the binding of α2β1, an integrin also present on the platelet surface, to collagen, which is necessary for firm attachment.4 5 The availability of integrin α2β1 on the platelet surface may play an essential role in platelet adhesion to collagen in the vessel wall (types I, III, IV, V, and VI).6
Patients deficient of the α2 subunit of α2β1 suffer from prolonged bleeding times and chronic mucocutaneous bleeding and have increased platelet adhesion to collagen types I, III, IV, and VI in vitro, under flow conditions.7 8 Moreover, individuals with autoimmune platelet dysfunction have been found to express serum antibodies against the α2 subunit of α2β1 that block in vitro platelet adhesion to collagens and collagen-induced aggregation.9 10 11
Large studies on the relation of platelet reactivity and cardiovascular disease are scarce because they are laborious and expensive. With molecular genetic techniques, candidate genes for platelet activity were identified. The role of variability of specific platelet integrins in cardiovascular disease can now be studied in large populations. Most studies on genetic variability of platelet function and cardiovascular disease have been performed on the PlA1/A2 polymorphism in αIIbβ3. Some studies showed an increased risk for arterial thrombosis for the PlA2 allele,12 13 whereas another study did not indicate any association.14 The interpretation of a relationship between the PlA1/A2 polymorphism and arterial thrombosis is difficult because there is no difference in platelet function between the genotypes. Recently, an 807 C to T single nucleotide exchange polymorphism has been identified in the gene encoding the α2 subunit of α2β1, which is associated with increased α2β1 density on the platelet membrane and with increased platelet adhesion to collagen types I and IV.15 16 17 This has led to the hypothesis that subjects with the 807 T genotype of α2β1 express more α2β1 on the platelet surface, leading to an increased potential of platelet adhesion and a tendency to arterial thrombosis and, hence, an increased risk of cardiovascular disease. This hypothesis is supported by findings of 2 independent case-control studies, which reported an increased risk of nonfatal myocardial infarction in men with the α2β1 807 T allele compared with men with the 807 C allele.18 19 The strongest relationship was found in smokers.19 The latter finding suggests that the effect of the 807 T polymorphism may be particularly pronounced in subjects with arterial damage due to other risk factors.
We studied the relationship between genetic variability in α2β1 and cardiovascular mortality in a cohort study of 12 239 women, initially aged 50 to 67 years, who were followed up for a maximum of 18 years.20
Between December 1974 and October 1980, 20 555 women who were born between 1911 and 1925 and lived in the city of Utrecht, the Netherlands, were invited for an experimental program for breast cancer screening.20 The women were invited for repeat examination at 1- to 6-year intervals. We chose as a baseline population the 12 239 (60%) women who attended the second examination (1976 to 1978) because this examination included a questionnaire on smoking. All women gave oral consent for the use of their data and urine samples for future scientific research. The study was approved by the Institutional Review Board of the University Medical Center Utrecht, the Netherlands.
Baseline examination included a questionnaire involving medication, prescribed diets, presence of cardiovascular disease, and smoking. In addition, blood pressure, height (m), and body weight (kg) were measured. Women were classified as having diabetes mellitus if they reported the use of insulin or oral blood glucose–lowering drugs or were on a diabetes diet. Women were classified as smokers if they reported themselves to be current smokers. Body mass index (kg/m2) was calculated as weight (kg) divided by height squared (m2). Obesity was defined as body mass index >30 kg/m2. Microalbuminuria was defined as the highest quintile of albumin excretion. The presence of cardiovascular disease, smoking, diabetes, hypertension, and excessive urinary albumin excretion was considered by markers of vascular dysfunction.
Municipal registries informed the Department of Epidemiology (presently the Julius Center for Patient Oriented Research) about migration and mortality of the cohort members. Causes of death were obtained from the general practitioners. Subjects who were lost to follow-up were censored at the time point that they were lost. The 9062 surviving women had a median follow-up time of 17 years, with a maximum of 18 years. One thousand four hundred forty-seven women (12.3%) had moved outside the recruitment area and had a median follow-up of 10 years, with a maximum of 18 years. During follow-up (182 976 women-years), 1714 women died: 608 from cardiovascular diseases (codes 390 to 459 of the International Classification of Diseases, Ninth Revision [ICD-9]), 601 from neoplasms (ICD-9 140 to 239), 299 from other causes, and 206 from unknown causes. Mortality from myocardial infarction was defined as ICD-9 codes 410 to 414, cerebrovascular mortality as ICD-9 codes 430 to 439, and other cardiovascular mortality as all remaining ICD-9 codes between 390 and 460.
A nested case-control approach was used.21 This allowed the use of all information from the entire cohort while keeping the laboratory work within manageable bounds. The cases were all 608 women who died of cardiovascular disease during the follow-up period; the controls constituted a random sample of 618 of the cohort of 11 631 women who did not die of cardiovascular disease (sampling fraction 1:18.8). Urine samples of 59 cardiovascular cases and 49 controls were not collected at baseline or were lost during follow-up. DNA samples of 69 cardiovascular cases and 73 controls were not suitable for analysis. Consequently, the final study group constituted 480 cardiovascular mortality cases and 496 women in the reference group.
DNA was isolated from 50-mL urine samples.22 A fragment containing nucleotide 807, located close to the gene coding for the α2 subunit of α2β1, was amplified in 20 mmol/L Tris-HCl, pH 8.0, 2.5 mmol/L MgCl2, 50 mmol/L KCl, 0.1 mg/mL BSA, 0.4 pmol of 3′ primer (5′-TGTTTAACTTGAACACATATAAAACC-3′), 0.4 pmol of 5′ primer (5′-GATTTAACTTTCCCAGCTGCCTTC-3′), 0.42 mmol/L of each nucleotide (Pharmacia, Biotech), 0.075 U superTAQ polymerase (HT Biotechnology LTD), and 5 μL DNA with the use of a PTC200 multicycler (MJ Research). Temperature cycles were as follows: 4 minutes at 94°C and 33 cycles of 40 seconds at 94°C, 40 seconds at 55°C, and 2 minutes at 72°C. The reaction was terminated with a 10-minute incubation at 72°C. Genotype was determined from each DNA fraction by dot blotting and hybridization with antigen-specific oligonucleotides.22 The antigen-specific oligonucleotide was 5′-γ32P-AATTGCTCCGAA-TGTGTT-3′ for the 807 C allele and 5′-γ32P-AATTGCTC-CAAATGTGTT-3′ for 807 T allele. Dots were visualized on x-ray films (DuPont) after overnight radiation. Mutation analysis was performed by 2 independent readers blinded for case or control status of the samples.
Means and proportions of baseline cardiovascular risk factors were computed for women with the α2 subunit genotypes of 807 CC, 807 CT, and 807 TT. The significance of mean difference was tested by ANOVA, and significance in proportions was tested by χ2 statistics. The χ2 goodness of fit test was used to determine whether the observed genotype distribution was in Hardy-Weinberg equilibrium.23 24
A nested case-control approach was used to estimate incidence rates and rate ratios.22 Because controls were selected at random at a known sampling fraction, all data from this group provide unbiased estimates of the entire cohort of women without cardiovascular mortality. We first determined the genotype of the α2 subunit of α2β1. Subsequently, we estimated the follow-up years for each genotype in the entire cohort by weighting the follow-up years of the reference group with a factor of 18.8 (the inverse of the sampling fraction) and applying the follow-up years from the cardiovascular mortality cases. Incidence rates were calculated as the number of incidents per genotype divided by the estimated follow-up years for that genotype.22 The incidence rate ratio was calculated as the incidence rate for cardiovascular disease in heterozygotes divided by the incidence rate in wild types. Poisson regression was used to estimate incidence rates and risk ratios; 95% CIs were calculated by the method of Hubers.25
Similarly, crude relative risks, incidence rates, and rate ratios were estimated separately for women who died of myocardial infarction (ICD-9 410 to 414), cerebrovascular disease (ICD-9 430 to 438), and other cardiovascular disease (all remaining ICD-9 codes between 390 to 459). Potential confounding by age at entry in the cohort, hypertension, body mass index, and smoking was investigated by adding these variables individually and once simultaneously to the Poisson containing the 807 T genotype to see whether the crude rate ratio of the 807 T genotype changed. The presence of effect modification was investigated by subgroup analysis on age (above or below the median), current smoking at baseline, diabetes, history of cardiovascular disease, and microalbuminuria.
Further subgroup analysis was performed on the numbers of indicators of endothelial perturbation, which were defined as 0 if subjects were nonsmokers and nondiabetic and had no microalbuminuria, as 1 if they had 1 of these indicators, and as 2 if they had 2 or 3 of these indicators. The significance of modifying effects was tested in a multivariate model, which includes interaction terms.
The genotype distribution of the control group was in Hardy-Weinberg equilibrium (χ2=0.30; 1 df, P=0.58). The α2β1 807 C/T genotype was not associated with age, diabetes mellitus, smoking, body mass index, or systolic and diastolic blood pressure or with a history of cardiovascular disease either in cardiovascular cases or in the control group (Table 1⇓).
The follow-up time of our cohort was 148 209 women- years: 58 443 years for women with the 807 CC genotype, 67 532 for women with the 807 CT genotype, and 22 234 for women with the 807 TT genotype. The estimated incidence rates for overall cardiovascular mortality and for fatal myocardial infarction, cerebrovascular mortality, and other cardiovascular mortality were similar among the α2 807 C/T genotypes (Table 2⇓).
The possibility that the relationship between the 807 C/T polymorphism and cardiovascular mortality was a consequence of confounding by age, blood pressure, body mass index, and smoking was excluded because adjustment for these variables individually or simultaneously did not change the crude mortality rate ratio of the 807 T genotype.
Subgroup analysis according to history of cardiovascular disease showed no modifying effect on the relationship between the α2 genotype and cardiovascular mortality. However, in those women who reported themselves to be current smokers at baseline, an increased risk of cardiovascular mortality was found for α2 807 TT homozygotes compared with 807 CC wild types (Table 3⇓); the mortality rate for women who were both smokers and α2 807 TT homozygotes was 7.1 (95% CI 3.8 to 12.9) per 1000 years compared with 3.5 (95% CI 2.8 to 4.4) per 1000 years for women who were 807 wild types and nonsmokers, 3.2 (95% CI 2.3 to 4.6) per 1000 years for women who were 807 wild types and smokers, and 3.2 (95% CI 2.3 to 4.6) per 1000 years for women who were 807 TT homozygotes but nonsmokers. In smokers, the risk ratio for cardiovascular mortality between 807 TT homozygotes and 807 CC wild types was 2.2 (95% CI 1.1 to 4.4). This risk tended to be increased for fatal myocardial infarction as well as cerebrovascular mortality and all other causes of cardiovascular mortality. Similarly, in those subjects who had diabetes or microalbuminuria, the mortality rate appeared to be increased in α2 807 TT homozygotes compared with 807 CC wild types, but these risk ratios did not reach statistical significance.
Further subgroup analysis on the number of risk indicators of endothelial perturbation is presented in Table 4⇓. If anything, in subjects who were nonsmokers and had no diabetes and microalbuminuria, the mortality rate appeared to be lower for α2β1 807 CT heterozygotes and 807 TT homozygotes than for 807 CC wild types. However, in subjects with one of those indicators, a significantly increased risk was found for cardiovascular mortality between 807 TT homozygotes and 807 CC wild types, whereas in subjects with 2 or 3 of these indicators, a markedly increased risk for cardiovascular mortality was found in 807 TT homozygotes compared with 807 CC wild types. The total follow-up time in the subgroup of subjects with 2 or 3 factors of compromised endothelia was 2493 years (1673 years in 807 C wild types, 792 years in 807 CT heterozygotes, and 28 years in 807 T homozygotes). In the 807 CC homozygotes, 17 women died during the 1673 follow-up years (mortality rate 10.4 per 1000 years), 20 women died during the 792 follow-up years in the α2β1 807 CT heterozygotes (mortality rate 25.3 per 1000 years), and 4 women died during the 28 follow-up years in the α2β1 807 TT homozygotes (mortality rate 143 per 1000 years). The incidence rates for the different subgroups are presented in the Figure⇓.
Our findings in a large population-based study on platelet collagen receptor α2β1 genotype and cardiovascular mortality show that the α2β1 807 C/T genotype has no overall association with cardiovascular mortality in women. However, in women who were current smokers at baseline, a significantly increased risk of cardiovascular mortality was found compared with 807 C wild types (relative risk 2.2, 95% CI 1.1 to 4.4). The relationship between 807 TT homozygotes and cardiovascular mortality was most pronounced in women who had ≥2 indicators of compromised endothelium: the risk ratio between 807 TT homozygotes and 807 CC wild types was 14.1 (95% CI 5.0 to 39.9).
To appreciate our findings, some characteristics of the study need to be addressed. Our data are obtained from a large prospective cohort study and are therefore not subject to selection bias because cases and controls were members of the same cohort. In addition, we do not expect information bias because questionnaire data, anthropometric data, and urine samples were collected at baseline before the cardiovascular events occurred. The prospective nature of the present study enabled us to investigate cardiovascular mortality, whereas conventional case-control studies rely on retrospectively collected data and are therefore limited to morbidity of cardiovascular disease or to intermediate markers of disease.
The probability of measurement error or misclassification on genotype is negligible because the 2 independent investigators who performed DNA diagnostics were blinded for case or control status of the DNA sample, and all measurements were performed in duplicate. The genotype distribution of α2β1 in our reference population was in Hardy-Weinberg equilibrium.
A disadvantage of full cohort study analyses is the large number of DNA samples to be screened on the α2 807 C/T genotype. This problem was solved by adopting a nested case-control approach: DNA samples of 480 women who died of cardiovascular disease were analyzed together with DNA samples of a random cohort sample of 496 women who did not die of cardiovascular disease. Because of random sampling, our control group constituted an unbiased estimate of the population time experience in the entire cohort. Nested case-control analysis takes full advantage of a cohort study, but it has the cost and labor effectiveness of a case-control study.21
Our findings of an increased risk of cardiovascular disease for α2 807 T homozygotes in women with compromised endothelium due to smoking, diabetes, and microalbuminuria are in agreement with the a priori hypothesis that women with the 807 T genotype of α2β1 express more α2β1 on the platelet surface, leading to an increased potential of platelet adhesion and, hence, an increased risk of cardiovascular disease. Normally, the vessel wall is protected against platelet-collagen interaction by the intact endothelium. This may explain the lack of association between the 807 T allele of the α2 subunit of α2β1 in the population at large. In smokers and in subjects with diabetes or with microalbuminuria, the compromised endothelium may allow platelet-collagen interaction and subsequent platelet activation.
Our findings build on 2 studies that have reported that the 807 T allele is associated with nonfatal myocardial infarction in men.18 19 Similar to our findings, smoking appeared to be a risk-modifying factor in the relationship between the 807 C/T polymorphism and cardiovascular mortality.19 Our findings are further supported by studies on indirect platelet adhesion to collagen via vWF.26 27 These studies suggested an association between plasma levels of vWF and increased risk of myocardial infarction in patients with angina pectoris or with reinfarction in a population of survivors of myocardial infarction.
We found evidence that genetically determined expression of excess α2β1 on the platelet surface may be associated with increased risk of cardiovascular mortality in subjects with perturbed endothelia, which is associated with diabetes, microalbuminuria, and smoking. Further research on the relationship between α2β1 and cardiovascular disease in high-risk populations, such as patients with angina pectoris, coronary stenosis, diabetes, microalbuminuria, and smokers, is indicated.
The study was supported by the Netherlands Heart Foundation (grant NHS 95-165). We thank Thomas Kunicki, MD, PhD, Paul de Jong, MD, PhD, and Dick de Zeeuw, MD, PhD, for reviewing the manuscript and their critical comments.
- Received February 16, 2000.
- Revision received April 28, 2000.
- Accepted May 8, 2000.
- Copyright © 2000 by American Heart Association
Houdijk WP, Sakariassen KS, Nievelstein PF, et al. Role of factor VIII-von Willebrand factor and fibronectin in the interaction of platelets in flowing blood with monomeric and fibrillar human collagen types I and III. J Clin Invest. 1985;75:531–540.
Wu YP, van Breugel HH, Lankhof H, et al. Platelet adhesion to multimeric and dimeric von Willebrand factor and to collagen type III preincubated with von Willebrand factor. Arterioscler Thromb Vasc Biol. 1996;16:611–620.
Verkleij MW, Morton LF, Knight CG, et al. Simple collagen-like peptides support platelet adhesion under static but not under flow conditions: interaction via alpha2 beta1 and von Willebrand factor with specific sequences in native collagen is a requirement to resist shear forces. Blood. 1998;91:3808–3816.
Saelman EU, Nieuwenhuis HK, Hese KM, et al. Platelet adhesion to collagen types I through VIII under conditions of stasis and flow is mediated by GPIa/IIa (alpha 2 beta 1-integrin). Blood. 1994;83:1244–1250.
Nieuwenhuis HK, Sakariassen KS, Houdijk WP, et al. Deficiency of platelet membrane glycoprotein Ia associated with a decreased platelet adhesion to subendothelium: a defect in platelet spreading. Blood. 1986;68:692–695.
Deckmyn H, Zhang J, Van Houtte E, et al. Production and nucleotide sequence of an inhibitory human IgM autoantibody directed against platelet glycoprotein Ia/IIa. Blood. 1994;84:1968–1974.
Montgomery RR, Kunicki TJ, Taves C, et al. Diagnosis of Bernard-Soulier syndrome and Glanzmann’s thrombasthenia with a monoclonal assay on whole blood. J Clin Invest. 1983;71:385–389.
Kunicki TJ, Orchekowski R, Annis D, et al. Variability of integrin alpha 2 beta 1 activity on human platelets. Blood. 1993;82:2693–2703.
Kunicki TJ, Kritzik M, Annis DS, et al. Hereditary variation in platelet integrin alpha 2 beta 1 density is associated with two silent polymorphisms in the alpha 2 gene coding sequence. Blood. 1997;89:1939–1943.
Kritzik M, Savage B, Nugent DJ, et al. Nucleotide polymorphisms in the alpha2 gene define multiple alleles that are associated with differences in platelet alpha2 beta1 density. Blood. 1998;92:2382–2388.
Santoso S, Kunicki TJ, Kroll H, et al. Association of the platelet glycoprotein Ia C807T gene polymorphism with nonfatal myocardial infarction in younger patients. Blood. 1999;93:2449–2253.
Moshfegh K, Wuillemin WA, Redondo M, et al. Association of two silent polymorphisms of platelet glycoprotein Ia/Iia receptor with risk of myocardial infarction: a case-control study. Lancet. 1999;353:9150:351–354.
Mietinen OS. Estimability and estimation in case-referent studies. Am J Epidemiol. 1976;103:181–235.
Roest M, Banga JD, Tempelman MJ, et al. Factor V Arg506Gln mutation is not associated with cardiovascular mortality in older women. Am J Epidemiol. 1999;149:665–670.
Hardy GH. Mendelian proportions in mixed population. Science. 1908;28:41–50.
Weinberg W. On demonstration of heredity in man. In: Papers of Human Genetics. Englewood Cliffs, NJ: Prentice-Hall; 1963.
Huber PJ. The behavior of maximum likelihood estimates under non-standard conditions. In: Proceedings of the Fifth Berkeley Symposium on Mathematical Statistics and Probability. 1967;1:221–233.
Jansson JH, Nilsson TK, Johnson O. von Willebrand factor in plasma: a novel risk factor for recurrent myocardial infarction and death [see comments]. Br Heart J. 1991;66:351–355.
Thompson SG, Kienast J, Pyke SD, et al. Hemostatic factors and the risk of myocardial infarction or sudden death in patients with angina pectoris: European Concerted Action on Thrombosis and Disabilities Angina Pectoris Study Group [see comments]. N Engl J Med. 1995;332:635–641.