Arterial and Venous Thrombosis Is Not Associated With the 4G/5G Polymorphism in the Promoter of the Plasminogen Activator Inhibitor Gene in a Large Cohort of US Men
Background The 4G allele of the 4G/5G polymorphism in the promoter of the plasminogen activator inhibitor (PAI-1) gene is associated with increased PAI-1 activity. In a small group of young Swedish men, this allele has been reported to predict risk of myocardial infarction. Whether this polymorphism increases risk of arterial and venous thrombosis among middle-aged men is unknown.
Methods and Results Among 14 916 men 40 to 84 years old participating in the Physicians' Health Study who provided baseline blood samples for DNA analysis, 374 suffered first myocardial infarction and 121 had venous thromboembolism during 8.6 years of follow-up. Distributions of the 4G/5G polymorphism in the PAI-1 gene promoter were assessed in these men as well as in a sample of study participants matched on age and smoking who did not develop vascular occlusion during the prospective follow-up period. The distributions of the 4G/4G, 4G/5G, and 5G/5G genotypes among men who developed myocardial infarction (0.27, 0.51, 0.22; P=.7) or venous thromboembolism (0.30, 0.49, 0.21; P=.5) were virtually identical to those of men who remained free of vascular disease (0.27, 0.50, 0.23). Thus, the relative risk of future thrombosis among those with the 4G/4G genotype compared with those without the 4G/4G genotype was 1.02 (95% CI, 0.8 to 1.3). There was no effect modification by age, smoking status, family history of premature thrombosis, history of hypertension, hypercholesterolemia, or aspirin use.
Conclusions These data indicate that the 4G/5G polymorphism in the promoter of the PAI-1 gene is not a major pathogenetic risk factor for arterial or venous thrombosis among middle-aged men.
Basic research and prospective cohort studies support the hypothesis that baseline fibrinolytic capacity is an important marker of risk for clinical thrombosis.1 2 3 4 5 6 In particular, impaired fibrinolytic function, as assessed by elevated plasma concentration of plasminogen activator inhibitor (PAI-1), is a marker for first and recurrent myocardial infarction among young patients1 2 and for ischemic events among individuals with prevalent atherosclerosis7 and may play a role in the pathogenesis of venous thromboembolism.8
At present, however, genetic and environmental determinants of PAI-1 expression are incompletely understood. Recently, a specific insertion/deletion polymorphism in the promoter region of the PAI-1 gene has been described in which one allele sequence has four guanosines (4G) and the other has five (5G).9 In vitro assays of promoter activity suggest that the 4G allele has higher activity than the 5G allele, and clinical studies indicate that individuals homozygous for the 4G allele have higher PAI-1 concentrations in plasma.9 10 11 12 Most recently, in a small case-control study of myocardial infarction among Swedish men <45 years old, the 4G allele was found to be associated with a twofold increase in risk of coronary thrombosis.10 This observation is intriguing and of potential clinical importance, because an easily detectable genetic determinant of PAI-1 function might be useful in the selection and dosing of patients being treated with anticoagulant and fibrinolytic agents, in particular tissue plasminogen activator.
At this time, however, no evaluation of the 4G/5G polymorphism has been performed in a large and well-characterized population of healthy individuals followed prospectively for arterial thrombosis, a critical test for any genetic hypothesis dealing with a complex, multifactorial, and polygenic disorder such as myocardial infarction. In addition, no data regarding the 4G/5G allele are available for venous thrombosis. Thus, to directly address both these issues, we evaluated the 4G/5G polymorphism in a cohort of nearly 15 000 US men initially free of cardiovascular disease who were followed for an average period of 8.6 years for the occurrence of first myocardial infarction, deep venous thrombosis, and pulmonary embolism.
In the US Physicians' Health Study,13 14 916 men initially free of reported vascular disease provided a baseline whole-blood sample at study initiation. Details of the Physicians' Health Study, a randomized, double-blind, placebo-controlled trial of aspirin and β-carotene in the prevention of cardiovascular disease and cancer, have been described elsewhere,13 as have the methods used to collect, store, and recover DNA from the baseline blood specimens.14 15
For this analysis, case patients were those study participants who subsequently developed a first myocardial infarction, deep venous thrombosis, or pulmonary embolism during a mean follow-up period of 8.6 years. Confirmation of each reported end point was determined from detailed reviews of hospital records, death certificates, and autopsy reports by an end-points committee of physicians using prespecified standardization criteria.14 For each case, a control subject was selected at random from those study participants who remained free of reported cardiovascular disease during the follow-up period, provided a baseline blood sample for DNA analysis, and met the matching criteria of age (within 1 year of the case patient), time since study initiation (in 6-month intervals), and smoking habit (current, former, or never smoker). Study participants provided baseline information concerning other cardiovascular risk factors, including systolic and diastolic blood pressures, parental history of MI before age 60 years, history of hypercholesterolemia, and presence of diabetes.
For each case patient and control subject, blood collected at enrollment was thawed, underwent DNA extraction, and was assayed for the 4G/5G genotype. The region around the 4G/5G polymorphism was amplified from genomic DNA, and the 188- and 189-bp products were denatured, separated by electrophoresis, and visualized by silver staining with a Pharmacia PhastGel system. The 5′ and 3′ primers were TAACCCCTGGTCCCG-TTC and CAGAGGACTCTTGGTCTTTCC, respectively. A PTC-200 with a heated lid (+15°C) from MJ Research was used to cycle 5-μL reactions in 200-μL thin-walled tubes without mineral oil. Each tube contained buffer (50 mmol/L KCl, 2 mmol/L MgCl2, 0.1% Triton X-100, 10 mmol/L Tris, pH 9.0), the four dNTPs (200 μmol/L each), both primers (1 mmol/L each), Taq polymerase (0.05 U), and sample DNA (5 to 50 ng). The program included denaturation at 94°C for 1 minute; 20 cycles of 92°C for 15 seconds, 65°C (−0.5°C per cycle) for 20 seconds, and 72°C for 30 seconds; 25 cycles of 92.0°C for 15 seconds, 55.0°C for 20 seconds, and 72°C for 30 seconds (±2 seconds per cycle); and a final 72°C for 5 minutes. After an equal volume of 98% formamide containing 0.05% bromphenol blue and 0.5% xylene cyanol in 20 mmol/L EDTA was added, samples were denatured for 3 minutes at 95°C, immersed in a 4°C ice bath, and loaded into 1-μL applicator combs. Electrophoresis using native buffer strips was at 15°C, with voltage and watts limited to 400 V and 2.5 W. Homogeneous 12.5% acrylamide gels were prerun for 100 V-h (10-mA limit), followed by a sample application step of 2 V-h (1-mA limit) with the applicator down at 1.2 and up at 1.3 V-h, respectively, and a separation step of 150 V-h (10-mA limit). Automated silver staining involved fixing in 20% trichloroacetic acid for 5 minutes at 20°C; sensitizing with 25% fresh glutaraldehyde for 6 minutes at 40°C; washing twice with MilliQ water for 2 minutes at 40°C; staining with 0.5% silver nitrate for 10 minutes at 30°C; washing once with MilliQ water for 2.5 minutes and twice more for 0.5 minute at 20°C; developing with fresh 12.5% sodium carbonate containing 40 μL of 37% formaldehyde/100 mL, first for 0.5 minute and then for 5 minutes at 30°C; stopping with 5% acetic acid for 0.5 minute at 30°C; and finally, preparation for air drying with 5% acetic acid/5% glycerol for 3 minutes at 30°C.
A picture of two gels (16 samples) is shown in Fig 1⇓. Three patterns are apparent; representative PCR products were sequenced to identify the 4G and 5G alleles.
Means and proportions for baseline cardiovascular risk factors were computed for case patients and control subjects, and differences were tested by Student's t test or the χ2 statistic. The proportions of case patients and control subjects with the 4G/4G, 4G/5G, and 5G/5G genotypes were compared by χ2 analysis, as were the 4G and 5G allele frequencies. Relative risks of thrombosis were calculated with logistic regression models. All probability values are two-tailed.
Baseline clinical characteristics of case and control subjects are shown in Table 1⇓. Because age at time of study entry and smoking status were matching criteria, these variables were identical among case and control subjects. As previously reported, men who developed myocardial infarction during the study period had a higher prevalence of diabetes, hypertension, and hypercholesterolemia than did those who remained free of reported vascular disease during follow-up.14 The mean age at time of first infarction was 62.9 years, and the mean age at time of venous thrombosis was 62.3 years. Study participants were almost entirely white. Blacks, Asian Americans, and Hispanic Americans made up 1.0%, 1.3%, and 2.1% of the cohort, respectively, and were equally distributed between the case and control groups.
Among control subjects who remained free of cardiovascular disease during follow-up, 133 (26.9%) were found to carry the 4G/4G genotype, 247 (49.9%) the 4G/5G genotype, and 115 (23.2%) the 5G/5G genotype, such that the overall allele frequencies among control subjects were 0.52 and 0.48 for the 4G and 5G alleles, respectively. These data are consistent with the distribution predicted by the Hardy-Weinberg equilibrium.
As shown in Fig 2⇓, the distributions of the 4G/4G, 4G/5G, and 5G/5G genotypes among case patients were virtually identical to those of control subjects both for any thrombosis (0.27, 0.51, 0.22; P=.9) and for those who developed either myocardial infarction (0.27, 0.51, 0.22; P=.7) or venous thrombosis (0.30, 0.49, 0.21; P=.5). No differences in allele frequencies were found between case and control groups (4G, 0.53 and 5G, 0.47 versus 4G, 0.52 and 5G, 0.48, respectively, P=.9). As shown in Table 2⇓, there was no association between thrombosis occurrence and the 4G/4G genotype (compared with the non-4G/4G genotypes) for the entire study group or for those ≤60 years old, the median age; with nonsmokers; with those without hypercholesterolemia or hypertension; or with those with no parental history. Identical null findings were obtained in analyses assuming autosomal-dominant and allelic-additive models of inheritance. In analyses limited exclusively to white men, the relative risk of thrombosis associated with the 4G allele was 1.03 (95% CI, 0.8 to 1.4), virtually identical to that of the study group as a whole (relative risk, 1.02; 95% CI, 0.8 to 1.3).
Because prior data have described an age-specific decline in the frequency of the 4G/4G genotype,11 we further analyzed the genotypic proportions by age. The 4G/4G genotype was present among 26.5% of study subjects <50 years old, compared with 27.4%, 25.4%, and 29.5% of study subjects 50 to 59, 60 to 69, and ≥70 years old, respectively (all values of P=NS). No difference in frequency of the 4G allele was observed in these four age categories (0.52, 0.51, 0.53, and 0.53, respectively). Further, no significant difference in the distribution of the 4G/5G polymorphism was found in comparisons of those with fatal (n=130) or nonfatal (n=365) thromboses.
To address whether or not aspirin use had any effect on the association between the 4G allele and thrombosis, we performed stratified analyses based on randomized aspirin assignment for those events that occurred before January 25, 1988, the date of unblinding of the aspirin component of the Physicians' Health Study. Overall, the relative risk of thrombosis associated with the 4G/4G genotype among individuals randomized to aspirin was 0.91 (P=.6), and the relative risk among those randomized to aspirin placebo was also 0.91 (P=.7).
In a large cohort of apparently healthy men, there were no material differences in allele frequency or genotype distribution for the 4G/5G polymorphism in the PAI-1 gene promoter comparing those who subsequently developed either arterial or venous thrombosis with those who remained free of reported vascular disease during almost 9 years of follow-up. Further, we found no evidence that the 4G allele prevalence declines with age or is preferentially associated with fatal compared with nonfatal events. Indeed, the narrow CIs surrounding the observed relative risks of 1.02 (for the whole cohort) and 1.03 (for analyses limited to white men) excludes with assurance effects of even small to moderate size.
The present data contrast with those of earlier publications concerning the 4G/5G polymorphism in the promoter of the PAI-1 gene and vascular risk. In particular, in a study of 94 Swedish men with first myocardial infarction at <45 years old, 43% had the 4G/4G genotype, compared with 26% of control subjects, a difference associated with a statistically significant 2.2-fold increase in risk (95% CI, 1.2 to 4.0).10 In addition, the 4G/4G genotype was associated with higher levels of PAI-1 activity, an important finding, because PAI-1 levels have previously been shown to be predictive of first and recurrent infarction among these individuals.1 2
As in any analytical epidemiological study, chance, bias, and confounding must be viewed as possible alternative explanations.16 Chance seems unlikely to explain our null result, because the present data include almost five times as many case patients as in prior studies and the 95% CIs for any potential risk elevations associated with the 4G allele are narrow. Selection bias is also not a plausible explanation for our findings because, in contrast to prior retrospective studies in which case and control subjects are selected by the investigators, our data derive from a prospective cohort of initially healthy men. Further, the overall allele distribution in our study (4G, 0.52; 5G, 0.48) is virtually identical to that previously published (4G, 0.53; 5G, 0.47), making it unlikely that our population is unique or poorly representative. Finally, confounding is unlikely in a prospective study in which the exposure variable is genetic.
With regard to potential effect modification, we found no evidence of association between the 4G allele and thrombotic risk when the analyses were limited to men <60 years old, to nonsmokers, to those free of hypertension and hypercholesterolemia, to those with no parental history of premature coronary heart disease, to whites, or to those assigned at random to active aspirin or aspirin placebo. We also failed to find any age-related decline in 4G allele frequency, as has been reported for a group of healthy Swedish blood donors.11 Our null observation in this regard suggests that the 4G allele does not have a major selective effect on all-cause mortality. Because our study included few patients <45 years old, these data do not address whether the 4G/5G polymorphism results in high risk for premature coronary disease.
Although the 4G/5G polymorphism may not be a major pathogenetic factor underlying arterial or venous thrombosis among middle-aged men, these data do not imply that PAI-1 is unimportant as a predictor of thrombotic risk. Prior work in this cohort has demonstrated that fibrinolytic capacity as assessed by plasma level of endogenous tissue-type plasminogen activator (TPA:ag) is positively associated with risk of myocardial infarction and stroke.3 4 Since TPA:ag and PAI-1:ag are directly correlated, those findings are in general consistent with those of other prospective studies demonstrating an association between altered fibrinolytic potential as assessed by PAI-1 concentration and risk of vascular occlusion.1 2 Thus, further studies will be necessary to determine the genetic and environmental factors responsible for alterations in baseline fibrinolytic potential.
Dr Ridker is the recipient of a Clinician Scientist Award from the American Heart Association.
Reprint requests to Dr Paul Ridker, Brigham and Women's Hospital, 900 Commonwealth Ave E, Boston, MA 02115. E-mail firstname.lastname@example.org.
- Received April 10, 1996.
- Revision received May 13, 1996.
- Accepted May 21, 1996.
- Copyright © 1997 by American Heart Association
Cortellaro M, Cofrancesco E, Boscheti C. Increased fibrin turnover and high PAI-1 activity as predictors of ischemic events in atherosclerotic patients: a case control study. Arterioscler Thromb. 1993;13:1412-1417.
Dawson SJ, Wiman B, Hamsten A, Green F, Humphries S, Henney AM. The two allele sequences of a common polymorphism in the promoter of the plasminogen activator inhibitor (PAI-1) gene respond differently to interleukin-1 in HepG2 cells. J Biol Chem. 1993;268:10739-10745.
Eriksson P, Kallin B, van't Hooft FM, Bavenholm P, Hamsten A. Allele-specific increase in basal transcription of the plasminogen-activator inhibitor 1 gene is associated with myocardial infarction. Proc Natl Acad Sci U S A. 1995;92:1851-1855.
Ye S, Green FR, Scarabin PY, Nicaud V, Bara L, Dawson SJ, Humphies SE, Evans A, Luc G, Cambou JP, Arveller D, Henney AM, Cambien F. The 4G/5G genetic polymorphism in the promoter of the plasminogen activator inhibitor-1 (PAI-1) gene is associated with differences in plasma PAI-1 activity but not with risk of myocardial infarction in the ECTIM study. Thromb Haemost. 1995;74:837-841.
Hennekens CH, Buring JE. Epidemiology in Medicine. Boston, Mass: Little, Brown, & Co; 1987.