Prothrombotic coagulation defects and cardiovascular risk factors in young women with acute myocardial infarction
Prof Dr F. R. Rosendaal, Department of Clinical Epidemiology and Department of Haematology, Leiden University Medical Centre, Building 1, C9, PO Box 9600, 2300 RC Leiden, the Netherlands. E-mail: F.R.Rosendaal@lumc.nl
Summary. We investigated the effect of prothrombotic coagulation defects in combination with smoking and other conventional risk factors on the risk of myocardial infarction in young women. In 217 women with a first myocardial infarction before the age of 50 years and 763 healthy control women from a population-based case–control study, factor V Leiden and prothrombin 20210A status were determined. Data on major cardiovascular risk factors and oral contraceptive use were combined with the presence or absence of these prothrombotic mutations, and compared between patients and controls. The overall odds ratio for myocardial infarction in the presence of a coagulation defect was 1·1 [95% confidence interval (CI) 0·6–1·9]. The combination of a prothrombotic mutation and current smoking increased the risk of myocardial infarction 12-fold (95% CI 5·7–27) compared with non-smokers without a coagulation defect. Among women who smoked cigarettes, factor V Leiden presence versus absence increased the risk of myocardial infarction by 2·0 (95% CI 0·9–4·6), and prothrombin 20210A presence versus absence had an odds ratio of 1·0 (95% CI 0·3–3·5). We conclude that factor V Leiden and prothrombin 20210A do not add substantially to the overall risk of myocardial infarction in young women. However, in women who smoke, the presence of factor V Leiden increased the risk of myocardial infarction twofold.
Factor V Leiden and the prothrombin 20210A variant are prothrombotic mutations that predispose for venous thrombosis (Bertina et al, 1994; Poort et al, 1996; Bloemenkamp et al, 1995); however, their contribution to the risk of arterial thrombosis is less clear. Positive associations were found in specific groups of patients, especially young patients with other cardiovascular risk factors or patients with normal coronary angiographies (Rosendaal et al, 1997a, b; Inbal et al, 1999; Mansourati et al, 2000; Van de Water et al, 2000; Burzotta et al, 2002), but these could not be confirmed in several other studies (Emmerich et al, 1995; Prohaska et al, 1995; Ridker et al, 1995a, 1999; Van Bockxmeer et al, 1995; Ardissino et al, 1996; Corral et al, 1997; Amowitz et al, 1999; Croft et al, 1999; Irani-Hakime et al, 2001; Smiles et al, 2002). In a study of 84 women with myocardial infarction aged below 44 years of age, factor V Leiden increased the risk of myocardial infarction 2·4-fold, an effect which seemed to be confined to current smokers (Rosendaal et al, 1997b). A similar effect of other major cardiovascular risk factors was seen for carriers of prothrombin 20210A (Rosendaal et al, 1997a).
Factor V Leiden enhances the effect of oral contraceptives, with a synergistic risk of venous thrombosis 35-fold greater than that of non-users without factor V Leiden (Vandenbroucke et al, 1994). Moreover, the increased risk of thrombosis is most pronounced in the first year of oral contraceptive use, especially in women with thrombophilic defects (Herings et al, 1999; Bloemenkamp et al, 2000; Vandenbroucke et al, 2000). For the prothrombin 20210A mutation and oral contraceptive use, a synergistic risk has been found for deep vein thrombosis and cerebral venous sinus thrombosis (Martinelli et al, 1998; 1999).
Oral contraceptives increase the risk of myocardial infarction with estimated relative risks between 2 and 5, which may be reduced by abstinence from smoking and a blood pressure check prior to prescription of oral contraceptives (WHO Collaborative Study of Cardiovascular Disease and Steroid Hormone Contraception, 1997). We recently found that modern low-dose oral contraceptives still increase the risk of myocardial infarction twofold and that the risk associated with oral contraceptive use was similar in women with and without a prothrombotic mutation (Tanis et al, 2001).
In this nationwide population-based case–control study we examined the effect of factor V Leiden and prothrombin 20210A on the first myocardial infarction in women aged less than 50 years. Furthermore, we aimed to assess whether these mutations would help to identify women at a high risk when smoking cigarettes, with a history of hypertension, diabetes or hypercholesterolaemia, and during oral contraceptive use.
Design. The Risk of Arterial Thrombosis In relation with Oral contraceptives (RATIO) study is a population-based case–control study on the association between oral contraceptive use and acute myocardial infarction, performed in the Netherlands (Tanis et al, 2001). We included patients from all eight university hospitals as well as eight surrounding hospitals and a population-based control group matched for age, index year and area of residence. In the first phase of the study a standardized mailed questionnaire was completed by 248 patients and 925 controls, and in the second phase either blood samples were drawn or buccal swabs collected for the determination of factor V Leiden and prothrombin 20210A status. The Ethics Committees of all participating hospitals approved the study protocol.
Patients and control women. Patients were women aged between 18 and 49 years with a first myocardial infarction who were all hospitalized in university or general hospitals in the Netherlands in the period between January 1990 and October 1995 (see Appendix for participating centres). Medical records and discharge letters were reviewed for confirmation of the diagnostic criteria for myocardial infarction, which was defined by the presence of symptoms, diagnostic elevated cardiac enzymes, and electrocardiographic changes indicative of myocardial infarction (Fried et al, 1991). We obtained DNA from 204 patients by venous blood draw and from 13 patients by mailed-in buccal swabs. Control subjects were women aged between 18 and 49 years without a prior history of myocardial infarction, stroke or severe peripheral arterial disease, and recruited from the same areas using random digit dialling (Hartge et al, 1984). In the control group, DNA was obtained from 763 women (638 women participated in venous blood draw and from 131 women we obtained a buccal swab). Overall response was 75%.
Demographic characteristics. Data regarding cigarette smoking, a history of hypertension, diabetes and hypercholesterolaemia, height and weight, educational level, a family history of cardiovascular disease and oral contraceptive use were collected with standardized mail questionnaires in 1997 and 1998. All questions elicited information from a time period preceding the index date, i.e. the date of myocardial infarction in patients and the mid-year of the same year in controls. Current smokers were those who reported smoking regularly in the year before the index date. Women were classified as having a history of hypertension, diabetes or hypercholesterolaemia when they reported a physician's diagnosis or were taking medication for these conditions before the index date. At the time of venous blood draw, we were able to confirm the data on three metabolic risk factors: hypertension, diabetes and hypercholesterolaemia. A women was classified as hypertensive when using antihypertensive medication, having a systolic blood pressure of 160 mmHg or higher, or a diastolic blood pressure of 95 mmHg or higher at the time of venepuncture. Blood pressure was measured in a supine position after 5 min rest. Diabetes was defined as the use of glucose lowering medication or a non-fasting glucose above 11·0 mmol/l. Hypercholesterolaemia was defined as the use of medication for this condition or a non-fasting serum total cholesterol above 6·5 mmol/l. Body mass index was calculated as body weight (kg) divided by height squared (m2). Women who had a body mass index ≥ 27·3 kg/m2 were classified as obese to facilitate comparisons with earlier reports (Fried et al, 1991; Rosendaal et al, 1997a). A family history of cardiovascular disease was defined as the presence of myocardial infarction, stroke or peripheral vascular disease under 60 years of age in a first-degree relative. Women were classified as current oral contraceptive users if they had used oral contraceptives in the month before the index date.
Blood collection and laboratory analysis. Venepunctures and buccal swab collections were performed between June 1998 and May 2000. Blood samples were drawn from the antecubital vein into two Sarstedt Monovette® tubes containing 0·106 mol/l trisodium citrate and centrifuged for 20 min at 4500 r.p.m. High-molecular-weight DNA was isolated from the white blood cells by a salting-out process (Miller et al, 1988) or from buccal swabs (Thomson et al, 1992). Cotton swabs in sodium dodecyl sulphate (SDS)-proteinease K solution were incubated for 2 h at 65°C after arrival and centrifuged for 1 min at 1000 r.p.m. Potassium acetate was added to a final concentration of 1·6 mol/l and after 15 min incubation on ice, protein were removed using chloroform/isoamylalcohol (24 : 1) treatment. The DNA was stored at 4°C until amplification. Genetic analysis of the factor V Leiden mutation (1691 G→A) was performed with the polymerase chain reaction as we described previously (Bertina et al, 1994). The prothrombin variant (20210 G→A) was determined by the presence of the HindIII restriction site in the polymerase chain reaction fragment also described previously (Poort et al, 1996). The technician who performed DNA analyses was blinded to whether a blood sample was from a patient or a control.
Statistical analysis. The prevalence of factor V Leiden and prothrombin 20210A was calculated in patients and controls separately. Univariate odds ratios with 95% confidence intervals (95% CI) were calculated for the relationship between coagulation defects or other cardiovascular risk factors and myocardial infarction as a measure of relative risk. The effect of a combination of risk factors, i.e. coagulation defects and oral contraceptive use or coagulation defects and major cardiovascular risk factors were analysed by calculating odds ratios in subjects with either one or both of these risk factors, compared with those with neither risk factor. Adjustment, when appropriate, was performed by multivariate regression models that controlled for age, index year and area of residence.
Risk factor variables
The mean age of the 217 patients with a first myocardial infarction was 42·8 years, and that of 763 control women 38·7 years (Table I). As expected, patients more frequently had a history of hypertension (25%vs 6%), diabetes (5%vs 1%) or hypercholesterolaemia (11%vs 3%). Validation of the metabolic risk factors at the time of blood sampling confirmed the diagnoses of hypertension (82%vs 12%), diabetes (11%vs 2%) or hypercholesterolaemia (62%vs 16%). However, as the result of blood pressure-lowering drugs, as well as lipid-lowering drugs, prescribed in patients after myocardial infarction, the percentages for hypertension and hyperlipidaemia far exceeded those documented prior to myocardial infarction (physician's diagnosis or drug therapy). The most striking difference between patients and controls was the percentage smoking at the index date: 83% in patients versus 42% in controls.
Table I. Baseline characteristics of women with first myocardial infarction and control women.
| Mean (SD)||42·8 (6·1)||38·7 (8·0)|
|Caucasian ethnicity*||206 (95)||719 (95)|
|Oral contraceptive use*||85 (39)||276 (36)|
| Primary school or less||113 (52)||229 (30)|
| Secondary school||80 (37)||332 (44)|
| Higher education or university||24 (11)||198 (26)|
| Hypertension||55 (25)||47 (6)|
| Hypercholesterolaemia*||23 (11)||22 (3)|
| Diabetes||11 (5)||10 (1)|
|Currently treated for|
| Hypertension†||177 (82)||92 (12)|
| Hypercholesterolaemia†||134 (62)||125 (16)|
| Diabetes†||23 (11)||12 (2)|
|Body Mass Index (mean, SD)*||25·5 (5·0)||23·4 (3·7)|
| Never||19 (9)||248 (32)|
| Former||18 (8)||198 (26)|
| Current||181 (83)||319 (42)|
|Family history of cardiovascular disease*||141 (67)||262 (37)|
The factor V Leiden mutation was found in 13/217 patients (6·0%), and in 42/763 healthy control women (5·5%). The prothombin 20210A mutation was present in 5/217 (2·3%) patients and 18/763 control women (2·3%). This yielded an overall odds ratio of 1·1 (95% CI 0·6–2·1) for first myocardial infarction in the presence of factor V Leiden and 1·0 (95% CI 0·4–2·7) for the prothrombin mutation (Table II). There were no homozygous carriers of either factor V Leiden or prothrombin 20210A; two control women were heterozygous for both mutations. Eighteen patients (8·3%) and 58 controls (7·6%) were carriers of either factor V Leiden or prothrombin 20210A. Adjustment for the stratification variables (age, index year and area of residence) did not affect the estimates. In women with a prothrombotic mutation aged less than 35 years, the relative risk of myocardial infarction was 1·6 (95% CI 0·4–5·8) and in women older than 35 years it was 0·9 (95% CI 0·5–1·7). The difference in these odds ratios was completely determined by factor V Leiden (and not by prothrombin 20210A) with a relative risk of 2·3 (95% CI 0·6–8·6) for women less than 35 years and 0·9 (95% CI 0·4–1·9) for women older than 35 years.
Table II. Prevalence of factor V Leiden and 20210A mutation in the prothrombin gene in 217 women with first myocardial infarction and 763 healthy control women.
|Factor V Leiden*|
| 1691 GG||204 (94·0)||721 (94·5)||1·0†|
| 1691 AG||13 (6·0)||42 (5·5)||1·1 (0·6–2·1)|
| 20210 GG||212 (97·7)||745 (97·6)||1·0†|
| 20210 AG||5 (2·3)||18 (2·4)||1·0 (0·4–2·7)|
| 1691 GG or 20210 GG||199 (91·7)||705 (92·4)||1·0†|
| 1691 AG or 20210 AG||18 (8·3)||58 (7·6)‡||1·1 (0·6–1·9)|
Combined effect of prothrombotic mutations and cardiovascular risk factors
The risk of myocardial infarction was increased more than 12-fold among smoking women with a coagulation defect compared with non-smokers without a coagulation defect, indicating a synergistic effect of both risk factors. The adjusted odds ratio for the combination of hypertension and a coagulation defect was 11·1 (95% CI 1·1–114), although this was based on very small numbers. The results presented in Table III indicate that the combined effect of a coagulation defect and either hypercholesterolaemia or obesity did not exceed the separate effect of the two factors. For diabetes, either the small sample size or an absence of carriers of a coagulation defect did not allow analyses for interaction. When we combined the presence of one or more of the metabolic risk factors (hypertension, hypercholesterolaemia, diabetes and obesity), the odds ratio did not differ between women without (OR 3·3; 95% CI 2·4–4·6) or with (OR 3·3; 95% CI 1·4–7·7) a coagulation defect. The risk of myocardial infarction was increased twofold for oral contraceptive users without a prothrombotic mutation (OR 2·1; 95% CI 1·5–3·0) and was not more increased in those with a prothrombotic mutation (OR 1·9; 95% CI 0·6–5·5).
Table III. Effect of smoking, hypertension, hypercholesterolaemia, diabetes, obesity and oral contraceptive use with and without a coagulation defect on the risk of first myocardial infarction.
|No smoking||No coagulation defect||35 (16·1)||403 (52·1)||1·0*|
|FVL/20210A||2 (0·9)||38 (5·0)||0·5 (0·1–2·3)|
|Smoking||No coagulation defect||164 (75·6)||300 (39·5)||6·5 (4·3–9·7)|
|FVL/20210A||16 (7·4)||18 (2·4)||12·5 (5·7–27)|
|Non-hypertensive||No coagulation defect||147 (67·7)||657 (86·6)||1·0*|
|FVL/20210A||15 (6·9)||55 (7·2)||1·2 (0·7–2·2)|
|Hypertensive||No coagulation defect||52 (24·0)||46 (6·1)||4·2 (2·7–6·6)|
|FVL/20210A||3 (1·4)||1 (0·1)||11·1 (1·1–114)|
|Non-hypercholesterolaemic||No coagulation defect||178 (82·0)||682 (90·0)||1*|
|FVL/20210A||16 (7·4)||54 (7·1)||1·2 (0·6–2·1)|
|Hypercholesterolaemic||No coagulation defect||21 (9·7)||19 (2·5)||3·7 (1·9–7·3)|
|FVL/20210A||2 (0·9)||3 (0·4)||1·7 (0·3–10·4)|
|Non-diabetic||No coagulation defect||188 (86·6)||693 (91·2)||1*|
|FVL/20210A||18 (8·3)||57 (7·5)||1·2 (0·6–2·0)|
|Diabetic||No coagulation defect||11 (5·1)||10 (1·3)||4·0 (1·6–9·8)|
|Non-obese||No coagulation defect||143 (65·9)||604 (81·5)||1*|
|FVL/20210A||10 (4·6)||47 (6·3)||0·9 (0·4–1·9)|
|Obese||No coagulation defect||56 (25·8)||82 (11·1)||2·6 (1·7–3·8)|
|FVL/20210A||8 (3·7)||8 (1·1)||3·7 (1·3–10·2)|
|No oral contraceptive use||No coagulation defect||116 (54·2)||443 (58·7)||1*|
|FVL/20210A||13 (6·1)||36 (4·8)||1·4 (0·7–2·7)|
|Oral contraceptive use||No coagulation defect||80 (37·4)||256 (33·9)||2·1 (1·5–3·0)|
|FVL/20210A||5 (2·3)||20 (2·6)||1·9 (0·6–5·5)|
Table IV shows that, in the subgroup of smokers, the presence of factor V Leiden increased the risk of myocardial infarction by 2·0 (95% CI 0·9–4·6), with an adjusted odds ratio of 3·4 (95% CI 1·3–9·2). In carriers of prothrombin 20210A who smoked, no additional increase in risk was found, with an odds ratio of 1·0 (95% CI 0·3–3·5) (adjusted OR 1·1; 95% CI 0·3–4·7). The combined effect of smoking, oral contraceptive use and a coagulation defect had a high, adjusted odds ratio of 34 (95% CI 5·5–179) compared with women without any of these risk factors.
Table IV. The effect of factor V Leiden and prothrombin 20210A mutation in smokers with and without myocardial infarction.
|Factor V Leiden|
| 1691 GG||168 (92·3)||307 (96·5)||1†||1†|
| 1691 AG||12 (6·7)||11 (3·5)||2·0 (0·9–4·6)||3·4 (1·3–9·2)|
| 20210 GG||176 (97·8)||311 (97·8)||1†||1†|
| 20210 AG||4 (2·2)||7 (2·2)||1·0 (0·3–3·5)||1·1 (0·3–4·7)|
| 1691 GG or 20210 GG||164 (91·1)||300 (94·3)||1†||1†|
| 1691 AG or 20210 AG||16 (8·9)||18 (5·7)||1·6 (0·8–3·3)||2·5 (1·1–5·6)|
Combination of cardiovascular risk factors
The contribution of conventional risk factors (smoking, hypertension, diabetes, hypercholesterolaemia and obesity) was far more than that of factor V Leiden, and in only four patients (2%) none of these risk factors were found, compared with 128 controls (18%). The risk of myocardial infarction increased very sharply with the number of risk factors observed: those with one risk factor had a 2·8-fold increased risk, those with two risk factors had a 14-fold increased risk, those with three risk factors a 26-fold increased risk, those with four risk factors a 89-fold increased risk and those with five risk factors had an odds ratio exceeding 140, all relative to those with no risk factor at all.
In this large population-based case–control study the prevalence of factor V Leiden and prothrombin 20210A did not differ between patients and control women. Factor V Leiden did not exert an effect on the overall risk of myocardial infarction (OR 1·1; 95% CI 0·6–2·1), nor did prothrombin 20210A (OR 1·0; 95% CI 0·4–2·7). The risk of myocardial infarction was further increased between twofold and threefold by factor V Leiden among smokers, but not by prothrombin 20210A. Among very young women, defined as aged below 35 years, factor V Leiden increased the risk of myocardial infarction twofold. Users of low-dose oral contraceptives had a twofold higher risk than non-users (95% CI 1·5–2·8), which was not affected by the presence of a prothrombotic mutation (Tanis et al, 2001). The combination of a coagulation defect, smoking of cigarettes and oral contraceptive use increased the risk of myocardial infarction 34-fold.
Combined effect of prothrombotic mutations and cardiovascular risk factors
The influence of coagulation defects on myocardial infarction has been investigated since 1995 with varying outcomes (Emmerich et al, 1995; Kontula et al, 1995; Ridker et al, 1995a, b, 1999; Van Bockxmeer et al, 1995; Ardissino et al, 1996; Arruda et al, 1997; Ferraresi et al, 1997; Rosendaal et al, 1997a, b; Doggen et al, 1998; Eikelboom et al, 1998; Franco et al, 1998; Croft et al, 1999; Gardemann et al, 1999; Inbal et al, 1999; Coulet et al, 2000). Different results on the contribution of the coagulation defects seem to depend on the type of population which was studied, with positive associations in specific subgroups of young women (Holm et al, 1994; Rosendaal et al, 1997a, b), smokers and patients with metabolic risk factors (Doggen et al, 1998; Inbal et al, 1999) or patients with normal coronary angiographies (Mansourati et al, 2000).
Previously, we reported that factor V Leiden was a determinant of myocardial infarction in young American women, increasing the risk 2·5-fold and fourfold when other major risk factors for myocardial infarction were taken into account (Rosendaal et al, 1997b). For the 20210A mutation, an age-adjusted odds ratio of 4·0 (95% CI 1·1–15·1) for myocardial infarction was found (Rosendaal et al, 1997a). Both in the two studies by Rosendaal and colleagues and the present study, the prevalence of smoking in young women with myocardial infarction was very high, 74% and 83% respectively. In the present study, we confirmed the hypothesis that smoking is a prerequisite for the effect of factor V Leiden in arterial thrombosis, but found no effect for prothrombin 20210A. Possible explanations for the different conclusions concerning metabolic risk factors in the two studies are admixture of different cardiovascular risk factors, i.e. obesity, in a heterogeneous American population or variations in geographical distribution of the prothrombotic mutations between young women from the USA and the Netherlands (Ridker et al, 1997; Rosendaal et al, 1998).
Strengths and limitations of the study
The present study is one of the largest case–control studies among young women that has examined the association between genetic coagulation defects and the risk of myocardial infarction. Both for patients and control women the response was high, at 80% and 73% respectively. The high percentage of oral contraceptive use in our study, almost 40% in both patient and control group, also allowed us to study the combined effect of oral contraceptive use and two prothrombotic mutations on myocardial infarction. In contrast, with the risk of venous thrombosis we found no interaction between oral contraceptive use and these prothrombotic mutations with regard to the risk of myocardial infarction.
The RATIO study is limited by the fact that we only studied women who survived myocardial infarction. It is unlikely that coagulation defects play an important role in survival of acute myocardial infarction, which is high in this age group (Dunn et al, 2000). However, the possibility that the prothrombotic mutations were associated with more severe myocardial infarction and acute death cannot be excluded. In this case, we would have underestimated the effect of prothrombotic mutations on the risk of myocardial infarction.
Myocardial infarction is a multicausal disease and several risk factors are needed before acute myocardial infarction occurs in young patients (Sidney et al, 1996; Siscovick et al, 1997). From a clinical perspective, the results of our study may have important implications. In the absence of an increased overall risk of myocardial infarction in carriers of the factor V Leiden mutation or prothrombin 20210A, the benefit of screening young individuals for these coagulation defects seems absent. In contrast, our study underlines the importance of screening for high blood pressure, diabetes and hypercholesterolaemia before prescribing the pill. Firstly, because of interaction between these conventional risk factors and oral contraceptive use as described in the first phase of the RATIO-study (Tanis et al, 2001), and secondly, the absolute risk associated with oral contraceptive use increases exponentially because of the steeply rising incidence of myocardial infarction with age (Farley et al, 1998). Smoking was by far the most striking risk factor for myocardial infarction in young women, present in over 80% of the patients, compared with 42% of control women. Therefore, the most important message for young women starting to use oral contraceptives is to have their blood pressure measured and to stop smoking. Given the high percentage of smokers among young women, this is the most important issue for the prevention of premature atherothrombotic events.
This study was supported by a grant from the Netherland Heart Foundation (grant 97.063). The authors thank the cardiologists of the participating centres (see Appendix). We thank Mrs M. de Boer and Mrs E. van Lunteren in assisting random digit dialling and venepunctures, and Mrs A. van Dam and Mrs J. van Dam for administrative support. We are indebted to Mrs C. Krommenhoek-van Es who performed the DNA analyses and Dr H. Vos for advice. Finally, we particularly thank all the women who participated in the RATIO-project.
Appendix: participating centres
Sint Antonius Hospital Nieuwegein, Prof Dr N. M. van Hemel
University Medical Center Amsterdam, Dr R. J. G. Peters
Leiden University Medical Center, Dr V. Manger Cats
Rijnstate Hospital, Arnhem, Dr H. A. Bosker
Medical Center Haaglanden, Westeinde Hospital, The Hague, Dr J. Kolf
University Medical Centrum Nijmegen- Sint Radboud, Prof Dr F. W. A. Verheugt
Leyenburg Hospital, The Hague, Dr B. J. M. Delemarre
University Medical Center Rotterdam-Dijkzigt, Dr F. A. M. Jonkman
University Medical Center Maastricht, Dr F. Vermeer
Rijnland Hospital, Leiderdorp, Dr C. van Rees
Free University Medical Center, Amsterdam, Dr O. Kamp
University Medical Center Utrecht, Prof Dr E. O. Robles de Medina*
University Medical Center Groningen, Dr M. van den Berg
Bronovo Hospital, The Hague, Dr P. R. M. van Dijkman
Sint Franciscus Gasthuis, Rotterdam, Dr A. Schelling
Diaconessenhuis, Leiden, Dr S. A. G. J. Witteveen