Abstract
Background
Whether use of oral contraceptives is a risk factor for arterial ischaemic stroke is controversial. In particular, few data are available on what criteria should be adopted to establish an individual profile of risk before the start of oral contraceptives.
Patients and methods
The effects of oral contraceptives and their interaction with the G1691A polymorphisms of the factor V gene, the G20210A polymorphisms of the prothrombin gene and the C677T polymorphisms of the MTHFR gene on the risk of cerebral ischaemia were determined in a series of 108 consecutive women aged <45 years with ischaemic stroke and 216 controls, in a hospital‐based case–control study design.
Results
Use of oral contraceptives was associated with an increased risk of cerebral ischaemia (odds ratio (OR) 3.95; 95% confidence interval (CI) 2.29 to 6.78). ORs for stroke were 2.25 (95% CI 1.15 to 4.40), 3.94 (95% CI 2.28 to 6.81) and 8.87 (95% CI 3.72 to 21.1) for non‐oral contraceptive users with the TT MTHFR genotype, oral contraceptive users without the TT MTHFR genotype and oral contraceptive users with the TT MTHFR genotype, respectively, when compared with non‐oral contraceptive users without the TT MTHFR genotype, with a multiplicative independent effect. Compared with non‐oral contraceptive users, ORs for stroke were 2.65 (95% CI 1.46 to 4.81) for oral contraceptive users with none of the studied polymorphisms and 22.8 (95% CI 4.46 to 116.00) for oral contraceptive users with at least one of the studied polymorphisms, with a synergistic effect.
Conclusions
Exposure to the effects of oral contraceptives may increase the risk of ischaemic stroke in women with an inherited prothrombotic background. Testing for these genetic variants may allow more accurate stratification of the population at risk before long‐term use of oral contraceptives is prescribed.
The cardiovascular safety of widely used oral contraceptives is still debated. The risks of venous thromboembolism have been well established,1,2 but there is controversy over whether oral contraceptives may also increase the risk of vascular arterial events. In particular, many studies have attempted to assess the risk of stroke related to use of oral contraceptives, but findings have been somewhat contradictory. Methodological differences among studies, the potential influence of confounding factors, the use of different oral contraceptive formulations and their changes over time mean the results of these analyses are not strictly comparable, making it difficult to draw definitive conclusions.3,4,5 Despite these differences, evidence suggests that a relationship between oral contraceptives and stroke exists. However, although this association is fairly robust across studies and meta‐analyses,6 the crude risk seems low, with only an additional 4.1 ischaemic strokes per 100 000 non‐smoking, normotensive women using low‐dose oestrogen oral contraceptives.4 Epidemiological studies have shown that the risk of cardiovascular events associated with oral contraceptives increases considerably in women with major vascular risk factors such as arterial hypertension or smoking. Accordingly, these conditions are now considered contraindications to the use of oral contraceptives, particularly after age 35 years.7,8 However, an issue that has received little attention is whether other conditions that are not currently taken into consideration by doctors when prescribing oral contraceptives might contribute to further increase the risk of arterial ischaemic diseases, because of an independent (additive or multiplicative) or an interacting (partial or complete) effect with these drugs.
Recently, in line with the results of previous reports showing the effect of a synergistic interaction between inherited thrombophilic defects and oral contraceptives on the risk of venous thrombosis,9 three independent studies suggested that the risk of arterial ischaemic stroke related to oral contraceptives might be higher in women with such a predisposing genetic background,10,11,12 prompting speculation that even a small degree of increased risk for stroke associated with oral contraceptives in the general population might be greatly amplified in such women and be clinically relevant. When confirmed, this might allow a more accurate stratification of the population at risk before oral contraceptives are prescribed.
Owing to the potential implications of screening for these genetic markers, we undertook the current case–control study of young women aged 18–45 years to investigate the contribution of oral contraceptive use and inherited thrombophilic defects to the occurrence of ischaemic stroke, by analysing their independent and interactive effects.
Subjects and methods
Data were obtained in the setting of a single‐centre, hospital‐based study, designed at the Department of Neurology, University of Brescia, Brescia, Italy, for the evaluation of gene–environment interactions in the development of ischaemic cerebrovascular disease. A detailed description of the methods and the standardised protocol for patients' documentation has been given elsewhere.13,14 For the purpose of this study, unselected, unrelated women from a series of consecutive patients admitted to our department with first‐ever acute ischaemic stroke occurring under the age of 45 years were enrolled. After completion of the diagnostic work‐up, cases were subtyped into four major aetiological categories, according to a classification based on the Trial of ORG 10172 in Acute Stroke Treatment criteria, accommodated and validated for the aetiology of stroke in the young15:
Atherosclerotic vasculopathy: cerebral infarction caused by large‐vessel atherosclerotic vasculopathy or small‐vessel disease
Non‐atherosclerotic vasculopathy
Cardiac/transcardiac embolism: also including cases with a cardioembolic source combined with a proven thrombophilic disorder
Other: cerebral infarction that did not meet the criteria for one of the categories outlined above.
Women from the staff of our hospital with no known history of vascular disease and matched to the cases by age in 3‐year bands served as controls.13 Both cases and controls were white and were from the same geographical area.
Assuming a gene–environment interaction model with a dominant mode of inheritance, significance level equal to 0.05 (two‐sided), disease prevalence equal to the prevalence of stroke in the Italian population aged 0–44 years (65/100 000),16 frequency of oral contraceptive use equal to 0.174,17 and prevalence of the studied prothrombotic variants equal to 0.262 in a representative population of northern Italy,18 we estimated that a sample of approximately 100 cases and 200 controls was necessary to meet the commonly accepted standard value of 80% for sufficient power, when the ORs of the genotype–environment interaction effect of genetic modelling are >5.19
Demographic data and history of hypertension, diabetes mellitus, cigarette smoking and hypercholesterolaemia were obtained from each subject. Hypertension was considered present if systolic blood pressure was >160 mm Hg and diastolic pressure >95 mm Hg in two separate measurements after the acute phase, or if the subject was under treatment with antihypertensive drugs before recruitment. The diagnosis of diabetes mellitus was established according to WHO criteria.20 Cigarette smokers were categorised as current smokers or non‐smokers (non‐smokers included former smokers who had quit smoking for at least 6 months before the study). Hypercholesterolaemia was considered present if cholesterol serum concentrations were >220 mg/dl or if the subject was under treatment with cholesterol‐lowering drugs. A detailed history of oral contraceptive use was recorded and the subjects categorised as users or non‐users (non‐users included former users who had stopped taking these drugs at least 1 month before the index event).
The study was designed and carried out in accordance with the ethical principles established by the local institutional guidelines on clinical investigation. Written, informed consent was provided by all study participants.
Genetic analyses
Venous blood sampling for biochemical determinations was performed in the early morning (before 07.00 h) after overnight fasting in all subjects. In patients, blood samples were obtained 7–10 days after the acute event. Genomic DNA was isolated in all subjects from −20°C frozen samples of EDTA anticoagulated whole blood using standard DNA extraction. The G1691A mutation in the factor V gene (factor V Leiden) and the G20210A mutation in the prothrombin gene were determined according to a standardised multiplex polymerase chain reaction method.21 The C677T MTHFR genotypes were determined according to the method of Frosst et al,22 using polymerase chain reaction amplification and restriction digestion with HinfI to distinguish mutant from wild‐type allele.
Statistical analysis
Bivariate mean differences (MD), odds ratio (OR) and 95% confidence intervals (CI) were estimated for conventional risk factors and polymorphisms. A binary variable was determined for each polymorphism (“yes” or “no” based on the status of “carrier” or “non‐carrier” of the FV 1691A allele, the PT 20210A allele and the TT677 MTHFR genotype). On the basis of this definition, the entire study group was stratified into two subcategories corresponding to subjects carrying at least one such genetic marker (prothrombotic, P) and subjects carrying none (non‐prothrombotic, non‐P), respectively.
The 4×2 table approach23 was used to estimate the effect of interaction between genetic thrombophilic background and the use of oral contraceptives on the risk of ischaemic stroke. In particular, the interactive effect of the TT677 MTHFR genotype with oral contraceptives, and that of the inherited prothrombotic background (defined as the presence of at least one genetic thrombophilic defect) with oral contraceptives were tested.
Five genotype–environment interaction models were fitted using logistic regression: (1) genotype–environment complete interaction; (2) genotype as moderator of the environment effect; (3) environment as moderator of the genotype effect; (4) genotype–environment partial interaction; and (5) genotype and environment independent effect.24 The referent categories were the following: (a) non‐TTMTHFR genotype and non‐oral contraceptive users for models 1, 4 and 5; (b) overall non‐oral contraceptive users for model 2; and (c) overall non‐TTMTHFR genotype for model 3. The modelling strategies included assessment of interaction models without and with adjustment for covariates (age, smoking habit, hypertension and hypercholesterolaemia). Diabetes mellitus was not entered into the final analysis as covariate owing to the low frequency of this condition in the present series. To compare such competing models, Akaike's information criterion (AIC = −2×model log‐likelihood+2×number of model parameters) and the Bayesian information criterion (BIC = −2×model log‐likelihood+log(n)×number of model parameters) were computed. The selected model was the one minimising either AIC or BIC.25 Because assessment of interaction by logistic regression models was performed assuming multiplicativity of the effects, Rothman's synergy (S) measure26 was also computed using the maximum likelihood estimates ORs of the selected model. The S index is the ratio of the observed effect with joint exposure to the effect predicted for joint exposure assuming additivity of the effects. No interaction corresponds to S = 1, whereas S>1 (S<1) can be interpreted as measure of relative increase (decrease) in the additive effect among those exposed to both factors.
Pooled ORs (95% CIs) from our data and those tabulated by Slooter et al11 (for the TT677 MTHFR genotype–oral contraceptive use interaction) and by Aznar et al10 (for the FV 1691A allele/PT 20210A allele–oral contraceptive use interaction) were also calculated using the same modelling approach. As frequencies of oral contraceptive users who were carriers of prothrombotic genotypes/alleles were not reported by Martinelli et al,12 data from this study were not included in the pooled analysis. Data analyses were conducted with SPSS V.11.1 software.
Results
Characteristics of the study group
As DNA could not be amplified in one subject, data from 108 cases were entered into the final analysis. Thus, 216 controls were recruited for a 1:2 case–control design. Table 1 presents the demographic characteristics and prevalence of selected risk factors of the participants. A cardioembolic aetiology was presumed in 36 (33.3%) patients, non‐atherosclerotic vasculopathy in 20 (18.5%), atherosclerotic vasculopathy in 11 (10.2%), and 41 patients (38.0%) met the criteria for other aetiologic categories. As expected, patients with stroke were more often smokers, those with hypertension and those with hypercholesterolaemia as compared with controls. The use of oral contraceptives was associated with a more than threefold increased risk of cerebral ischaemia (OR 3.95; 95% CI 2.29 to 6.78).
Table 1 Demographic and clinical characteristics of patients with stroke and controls.
| Characteristic | Patients with stroke (n = 108) | Controls (n = 216) | Crude OR | 95% CI |
|---|---|---|---|---|
| n (%) | n (%) | |||
| Current smokers | 44 (40.8) | 27 (12.5) | 4.89 | 2.79 to 8.54 |
| Hypertension | 13 (12.0) | 9 (4.2) | 3.14 | 1.30 to 7.62 |
| Diabetes mellitus | 0 (0) | 5 (2.3) | 0.18 | 0.01 to 3.23 |
| Hypercholesterolaemia | 27 (25.0) | 23 (10.7) | 2.99 | 1.61 to 5.54 |
| Use of oral contraceptives | 43 (39.8) | 31 (14.4) | 3.95 | 2.29 to 6.78 |
| Mean ± SD | Mean ± SD | Crude MR | ||
| Age (years) | 33.9 (7.4) | 34.6 (6.6) | −0.70 | – |
Distribution of the prothrombotic genotypes
The observed distribution of the genotypes closely resembled those reported previously in other series from northern Italy.18,27 An increase in stroke risk was associated with the TT677 MTHFR genotype when compared with the CC genotype (OR 3.36; 95% CI 1.60 to 7.05) and with the PT 20210A gene variant when compared with the PT 20210G gene variant (OR 6.52; 95% CI 1.73 to 24.6). By contrast, the frequency of the FV 1691A gene variant was not different in the two groups (table 2). The prevalence of subjects carrying at least one procoagulant genotype was higher in the group of patients than in the group of controls (30.6% v 14.4%; OR 2.63; 95% CI 1.50 to 4.59).
Table 2 Characteristics of the study group according to genotype distribution.
| Genotype distribution | Patients with stroke (n = 108) | Controls (n = 216) | Crude OR | 95% CI | ||||
|---|---|---|---|---|---|---|---|---|
| Methylenetetrahydrofolate reductase C677T | ||||||||
| CC | 25 (23.1) | 84 (38.9) | 1 | |||||
| CT | 61 (56.5) | 110 (50.9) | 1.86 | 1.00 to 3.21 | ||||
| TT | 22 (20.4) | 22 (10.2) | 3.36 | 1.60 to 7.05 | ||||
| FV G1691A | ||||||||
| GG | 103 (95.3) | 210 (97.2) | 1 | |||||
| GA | 5 (4.7) | 6 (2.8) | 1.70 | 0.51 to 5.70 | ||||
| AA | 0 (0.0) | 0 (0.0) | – | |||||
| PT G20210A | ||||||||
| GG | 98 (90.7) | 213 (98.6) | 1 | |||||
| GA | 9 (8.4) | 3 (1.4) | 6.52 | 1.73 to 24.6 | ||||
| AA | 1 (0.9) | 0 (0.0) | – | |||||
| FV 1691A or PT 20210A or TT MTHFR | ||||||||
| No | 75 (69.4) | 185 (85.6) | 1 | |||||
| Yes | 33 (30.6) | 31 (14.4) | 2.63 | 1.50 to 4.59 | ||||
Interaction analysis
Tables 3 and 4 summarise the results of the analyses undertaken to find the effect of potential interactions of both the TT677 MTHFR genotype and the overall inherited prothrombotic background with the use of oral contraceptives on the risk of ischaemic stroke. Based on the AIC or BIC indices (not shown), the following two models were selected: the “TT MTHFR genotype and oral contraceptive use independent effect” model (table 3) and the “inherited prothrombotic background as moderator of oral contraceptive use effect” model (table 4).
Table 3 4×2 table of TTMTHFR × oral contraceptive use × disease status, and maximum likelihood estimate odds ratios (95% confidence interval) of the model “TT MTHFR and oral contraceptives use independent effects on disease status”.
| Patients with stroke (n, %) | Controls (n, %) | Unadjusted | Adjusted* | |||||
|---|---|---|---|---|---|---|---|---|
| OR | 95% CI | p Value | OR | 95% CI | p Value | |||
| Non‐oral contraceptive users and non‐TT | 53 (24.3) | 165 (75.7) | 1 | — | 1 | |||
| Non‐oral contraceptive users and TT | 12 (37.5) | 20 (62.5) | 2.25 | 1.15 to 4.40 | 0.018 | 3.02 | 1.43 to 6.41 | 0.004 |
| Oral contraceptive users and non‐TT | 33 (53.2) | 29 (46.8) | 3.94 | 2.28 to 6.81 | <0.001 | 3.89 | 2.12 to 7.15 | <0.001 |
| Oral contraceptive users and TT | 10 (83.3) | 2 (16.7) | 8.87 | 3.72 to 21.1 | <0.001 | 11.8 | 4.46 to 31.0 | <0.001 |
Synergy index = (8.87−1)/((2.25−1)+(3.94−1)) = 1.88.
Adjusted synergy index = (11.8−1)/((3.02−1)+(3.89−1)) = 2.20.
p Values of the square root of Wald test with df = 1 (two‐tailed).
*Covariates: age, smoking, hypertension and hypercholesterolaemia.
Table 4 4×2 table of genetic background x oral contraceptive use × disease status, and maximum likelihood estimate odds ratios (95% confidence interval) of the model “genetic thrombophilic background as moderator of oral contraceptives use effect on disease status”.
| Patients with stroke (n, %) | Controls (n, %) | Unadjusted | Adjusted* | |||||
|---|---|---|---|---|---|---|---|---|
| OR | 95% CI | p Value | OR | 95% CI | p Value | |||
| Non‐oral contraceptive users and non‐P | 48 (23.5) | 156 (76.5) | 1 | – | 1 | |||
| Non‐oral contraceptive users and P | 17 (37.0) | 29 (63.0) | 1 | 1 | ||||
| Oral contraceptive users and non‐P | 27 (48.2) | 29 (51.8) | 2.65 | 1.46 to 4.81 | 0.001 | 2.57 | 1.33 to 4.98 | 0.005 |
| Oral contraceptive users and P | 16 (88.9) | 2 (11.1) | 22.8 | 4.46 to 116 | 0.007 | 23 | 4.19 to 125 | 0.008 |
P, prothrombotic.
Synergy index = (22.8−1)/(2.65−1) = 13.2.
Adjusted synergy index = (23.0−1)/(2.57−1) = 14.0.
p Values of the square root of Wald test with df = 1 (two‐tailed).
*Covariates: age, smoking, hypertension and hypercholesterolaemia.
Non‐oral contraceptive users carrying the TT MTHFR genotype had an increased risk of stroke in comparison with non‐oral contraceptive users without this genotype (OR 2.25; 95% CI 1.15 to 4.40). A similar effect was determined by oral contraceptive use alone (OR 3.94; 95% CI 2.28 to 6.81), whereas the combination of the TT MTHFR genotype and oral contraceptive use was associated with about a ninefold greater odds of ischaemic stroke (OR 8.87; 95% CI 3.72 to 21.1), as predicted by the multiplicativity of the genotype–environment effect (2.25×3.94 = 8.87). The combined effect of the TT MTHFR genotype and oral contraceptive use on stroke risk was about 90% greater than that predicted by assuming additivity of effects (S = 1.88). Adjustment for other covariates did not significantly change the results (table 3).
A synergistic effect was observed when the presence/absence of an inherited prothrombotic background (P/non‐P) was considered (table 4). Compared with the “corner‐point” reference category of non‐oral contraceptive users (both with or without an inherited prothrombotic background), current use of oral contraceptives resulted in about threefold increased risk of stroke in subjects with no predisposing thrombophilic background (OR 2.65; 95% CI 1.46 to 4.81), and in about 23‐fold increase in the disease risk (OR 22.8; 95% CI 4.46 to 116) when in combination with at least one thrombophilic genotype (table 4). Such a combined effect was about 13‐fold greater than that predicted by assuming additivity of effects (S = 13.2). Results remained substantially unchanged after adjustment for covariates.
Pooled data analysis
The model selection based on the AIC/BIC indices of the pooled data from our study and that of Slooter et al11 confirmed the multiplicative independent effect of the TT677 MTHFR genotype and oral contraceptive use in increasing the risk of stroke (OR 2.67×1.65 = 4.41; 95% CI 2.85 to 6.84; S = 1.47). The same effect resulted from the combination of the FV 1691A allele/PT 20210A allele and oral contraceptive use according to the pooled data from the present study and that of Aznar et al10 (OR 3.49×3.81 = 13.3; 95% CI 5.17 to 34.2; S = 2.32).
Discussion
Although oral contraceptive use is generally perceived as a predisposing condition for stroke in young women, the absolute risk seems to be small and even less relevant when the non‐negligible rate of maternal mortality during pregnancy is taken into account.28 As a general rule, premenopausal women can safely be prescribed oral contraceptives, provided they have no increased cardiovascular risk at baseline. However, apart from major cardiovascular risk factors such as hypertension and cigarette smoking, which are now established contraindications to the use of oral contraceptives,7,8 few data are available on the influence that other conditions might have in increasing the risk of stroke when combined with these drugs. The identification of these conditions has implications for public health, given the large number of women currently using oral contraceptives worldwide, and, at an individual level, would allow “at‐risk” subjects to be stratified and an individual profile of stroke risk to be established before any decision to start long‐term use of oral contraceptives. The results of our study emphasise the possibility of better defining such an individualised risk. Although an overall increased risk of disease was associated with the use of these drugs, women carrying inherited thrombophilic defects seem to be exposed to a greater risk of stroke when under treatment with oral contraceptives in comparison to those without such a genetic predisposing background.
These findings are in line with previous observations made by Aznar et al10 in a small subgroup of women with cryptogenic stroke, by Slooter et al11 in a separate analysis of the Risk of Arterial Thrombosis in Relation to Oral Contraceptives study, and by Martinelli et al12 in a recent case–control study, and further strengthen the assumption that a pre‐existing prothrombotic state should be considered an additional risk factor when oral contraceptives are prescribed. In particular, according to our results, an interactive effect with genotype seems to be operant when prothrombotic variants are analysed in combination. Methodological differences in the study design of these reports are noteworthy (table 5). Firstly, a population‐based setting, such as that of Slooter et al, has the theoretical advantage of reducing the risk of selection bias in comparison to a hospital‐based setting, such as that of Aznar et al, of Martinelli et al and ours. The observation that the summary risk estimates are generally lower in studies using population‐based controls than in studies using hospital‐based controls also supports this assumption.29 As Slooter pointed out, the fact that only a limited number of the eligible patients was included in their final analysis cannot exclude a survival bias. By contrast, as patients were both included and sampled during the acute phase of stroke, the possibility of biased observations due to a survival effect is unlikely in our study. Secondly, to obtain a homogeneous group in terms of clinical phenotype, Aznar et al focused on patients with stroke with no recognised cause of infarct, whereas Slooter et al applied specific predefined criteria for patient selection and Martinelli et al investigated a highly selected population of stroke survivors referred to their centre for thrombophilia screening. Unlike these authors, we decided to include in our analysis all women with cerebral infarct of arterial origin irrespective of stroke subtype, as we believe this may increase the external validity of our results. Finally, the sample size of the four studies is not strictly comparable. One rule in genetic epidemiology is that small initial studies (such as that of Aznar et al) are likely to overestimate the true effect size30 and their findings would need to be replicated in other, ideally larger, samples. In this regard, Slooter's, Martinelli's and our own analyses served as confirmatory.
Table 5 Comparison of study design and characteristics of the four studies investigating the thrombophilia–oral contraceptives–ischaemic stroke hypothesis.
| Aznar et al10 | Slooter et al11 | Martinelli et al12 | This study | |
|---|---|---|---|---|
| Study design | Case–control analysis | Case–control analysis | Case–control analysis | Case–control analysis |
| No of patients/controls | 34/68 | 193/767 | 105/293 | 108/216 |
| Patients | ||||
| Age, years, mean (SD) | 18–50 | 18–49, 38.6 (8.0) | Fertile age, 34.7 (9.1) | 18–45, 33.9 (7.4) |
| Diagnosis | Ischaemic stroke | First ever ischemic stroke | First ever ischaemic stroke | First ever ischaemic stroke |
| Inclusion criteria | Consecutive patients with cryptogenic stroke | Consecutive patients with acute stroke | Consecutive patients with history of stroke | Consecutive patients with acute stroke |
| Exclusion criteria | Atherosclerosis, heart disease, foramen ovale, occlusive vessel disease | TIA, cerebral venous sinus thrombosis, carotid artery dissection, history of cardiovascular or cerebrovascular disease, terminal illness, aphasia or cognitive impairment interfering with the questionnaire, not speaking Dutch | ||
| Controls | ||||
| Age, years, mean (SD) | <50 | 18–49, 39.7 (7.7) | Fertile age, 34.9 (8.6) | 34.6 (6.6) |
| Selection criteria | Not defined | From the general population by random digit dialling | From partners and friends who accompanied patients to the centre | From the hospital staff |
| Inclusion criteria | Age‐matched healthy subjects | Age‐matched subjects, no history of coronary heart disease, cerebrovascular event or peripheral vascular disease | Healthy subjects of fertile age | Age‐matched subjects (±3 years), no known history of vascular disease |
| Setting | Unit of thrombophilia | Stroke centre | Thrombosis centre | Stroke centre |
| OC users | Not defined | Within 1 month before the acute event | Within the 2 weeks before referral | Within 1 month before the acute event |
| Thrombophilic tests | FV G1691A | FV G1691A | FV G1691A | FV G1691A |
| PT G20210A | PT G20210A | PT G20210A | PT G20210A | |
| C677T MTHFR | C677T MTHFR | C677T MTHFR | ||
| Hyperhomocysteinaemia | ||||
| AT, PC, PS deficiency |
AT, anti‐thrombin; OC, oral contraceptive; PC, protein C; PS, protein S; TIA, transient ischaemic attack.
Overall, despite differences in the study design, there is substantial consistency among the results of these four reports, which supports the assumption that a combined effect of inherited thrombophilia and oral contraceptives on the risk of cerebral ischaemia is likely to exist.
In conclusion, if oral contraceptives are prescribed to women with an increased absolute risk at baseline, or for reasons other than contraception, the benefits might not justify the risks. Our findings suggest that the existence of an inherited prothrombotic state may increase such a baseline risk and may be associated with a higher rate of cerebral ischaemic events in women exposed to the effects of these drugs. The fact that a synergistic interaction with oral contraceptive use was found only when the thrombophilic genotypes were analysed in combination, as opposed to the independent multiplicative effect observed from the analysis of single polymorphisms in the present study, clearly indicates that studying an individual polymorphism in isolation would not uncover the overall predisposing influence of a hypercoagulable state. Although this study seems to be sufficiently powered in case of ORs of the gene–environment interaction effect >5, we do not dispute the fact that the large confidence intervals make the results of our interaction analysis statistically unstable. This puts further emphasis on the need for large national and international networks to overcome sampling limitations and come to conclusive evidence. A larger sample would also allow a more accurate characterisation of the study group with regard to other relevant variables such as the effect of the dose of oestrogen and the chemical composition of progestogen, or the effect of progestogen‐only compounds on stroke risk.
When confirmed, our observations may provide the opportunity to identify subgroups of individuals in whom the exposure to oral contraceptives might significantly increase the incidence of ischaemic outcomes and prompt consideration of alternative therapeutic or contraceptive approaches.
Acknowledgements
We thank Professor Deodato Assanelli, Dipartimento di Scienze Mediche e Chirurgiche, Medicina Interna, Università degli Studi di Brescia, for helpful discussions, Drs Biagio Troianiello and Giorgio Dalla Volta, Neurologia, Istituto Clinico Città di Brescia, and Dr Eugenio Magni, Neurologia, Casa di Cura Poliambulanza, Brescia, for their support. We also thank all individuals who participated in the study.
Abbreviations
AIC - Akaike's information criterion
BIC - Bayesian information criterion
Footnotes
Competing interests: None declared.
References
- 1.Burkman R T, Collins J A, Shulman L P.et al Current perspectives on oral contraceptives use. Am J Obstet Gynecol 2001185S4–12. [DOI] [PubMed] [Google Scholar]
- 2.Kemmeren J M, Algra A, Grobbee D E. Third generation oral contraceptives and risk of venous thrombosis: meta‐analysis. BMJ 2001323131–134. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Chan W S, Ray J, Wai E K.et al Risk of stroke in women exposed to low‐dose oral contraceptives. A critical evaluation of evidence. Arch Intern Med 2004164741–747. [DOI] [PubMed] [Google Scholar]
- 4.Gillum L A, Johnston S C. Ischemic stroke risk with oral contraceptives. JAMA 200028472–78. [DOI] [PubMed] [Google Scholar]
- 5.Siritho S, Thrift A G, McNeil J J.et al Risk of ischemic stroke among users of the oral contraceptives pill. The Melbourne Risk Factor Study (MERFS) Group. Stroke 2003341575–1580. [DOI] [PubMed] [Google Scholar]
- 6.Bushnell C D. Oestrogen and stroke in women: assessment of risk. Lancet Neurol 20054743–751. [DOI] [PubMed] [Google Scholar]
- 7.World Health Organization Improving access to quality care in family planning: medical eligibility criteria for contraceptive use. 2nd edn. Geneva: World Health Organization, 2001
- 8.ACOG Committee on Practice Bulletins‐Gynecology ACOG Practice Bulletin. The use of hormonal contraception in women with coexisting medical conditions. Number 18, July 2000. Int J Gynaecol Obstet 20017593–106. [DOI] [PubMed] [Google Scholar]
- 9.Legnani C, Palareti G, Guazzaloca G.et al Venous thromboembolism in young women: role of thrombophilic mutations and oral contraceptives use. Eur Heart J 200223984–990. [DOI] [PubMed] [Google Scholar]
- 10.Aznar J, Mira Y, Vayà A.et al Factor V Leiden and prothrombin G20210A mutations in young adults with cryptogenic ischemic stroke. Thromb Haemost 2004911031–1034. [DOI] [PubMed] [Google Scholar]
- 11.Slooter A J C, Rosendaal F R, Tanis B C.et al Prothrombotic conditions, oral contraceptives, and the risk of ischemic stroke. J Thromb Haemost 200531213–1217. [DOI] [PubMed] [Google Scholar]
- 12.Martinelli I, Battaglioli T, Burgo I.et al Oral contraceptive use, thrombophilia and their interaction in young women with ischemic stroke. Haematologica 200691844–847. [PubMed] [Google Scholar]
- 13.Pezzini A, Del Zotto E, Archetti S.et al Plasma homocysteine concentration, C677T MTHFR genotype and 844ins68bp CBS genotype in young adults with spontaneous cervical artery dissection and atherothrombotic stroke. Stroke 200233664–669. [DOI] [PubMed] [Google Scholar]
- 14.Pezzini A, Del Zotto E, Magoni M.et al Inherited thrombophilic disorders in young adults with ischemic stroke and patent foramen ovale. Stroke 20033428–33. [DOI] [PubMed] [Google Scholar]
- 15.Johnson C J, Kittner S J, McCarter R J.et al Interrater reliability of an etiologic classification of ischemic stroke. Stroke 19952646–51. [DOI] [PubMed] [Google Scholar]
- 16.Stroke Prevention and Educational Awareness Diffusion Ictus cerebrale: linee guida di prevenzione e trattamento. Pubbl Catel 2001441–52. [Google Scholar]
- 17.Parazzini F, Negri E, Ricci E.et al Correlates of oral contraceptice use in Italian women, 1991–1993. Contraception 199654101–106. [DOI] [PubMed] [Google Scholar]
- 18.Cattaneo M, Chantarangkul V, Taioli E.et al The G20210A mutation of the prothrombin gene in patients with previous first episodes of deep‐vein‐thrombosis: prevalence and association with factor V G1691A, methylenetetrahydrofolate reductase C677T and plasma prothrombin levels. Thromb Res 1999931–8. [DOI] [PubMed] [Google Scholar]
- 19.Gauderman W J. Sample size requirements for matched case‐control studies of gene‐environment interaction. Stat Med 20022135–50. [DOI] [PubMed] [Google Scholar]
- 20.WHO Study Group on Diabetes Mellitus World Health Organization technical report. Series 727. Geneva, Switzerland: World Health Organization, 1985
- 21.Ripoll L, Paulin D, Drouet L O. Multiplex PCR‐mediated site‐directed mutagenesis for one‐step determination of Factor V Leiden and G20210A transition of the Prothrombin gene. Thromb Haemost 199778960–961. [PubMed] [Google Scholar]
- 22.Frosst P, Blom H J, Milos R.et al A candidate genetic risk factor for vascular disease: a common mutation in methylenetetrahydrofolate reductase. Nat Genet 199510111–113. [DOI] [PubMed] [Google Scholar]
- 23.Botto L D, Khoury M J. Facing the challenge of gene‐environment interaction: the two‐by‐four table and beyond. Am J Epidemiol 20011531016–1020. [DOI] [PubMed] [Google Scholar]
- 24.Ottoman R. An epidemiological approach to gene‐environment interaction. Genet Epidemiol 19907177–185. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Burnham K P, Anderson D R.Model selection and inference. A practical information theoretic approach. 2nd edn. New York: Springer‐Verlag, 2002
- 26.Rothman K J. Synergy and antagonism in cause‐effect relationship. Am J Epidemiol 197499385–388. [DOI] [PubMed] [Google Scholar]
- 27.Martinelli I, Cattaneo M, Taioli E.et al Genetic risk factors for superficial vein thrombosis. Thromb Haemost 1999821215–1217. [PubMed] [Google Scholar]
- 28.Hoyert D L, Arias E, Smith B L.et al Deaths: final data for 1999. Nat Vital Stat Rep 2001491–113. [PubMed] [Google Scholar]
- 29.Baillargeon J P, McClish D K, Essah P A.et al Association between the current use of low‐dose oral contraceptives and cardiovascular arterial disease: a meta‐analysis. J Clin Endocrinol Metab 2005903863–3870. [DOI] [PubMed] [Google Scholar]
- 30.Wacholder S, Chanock S, Garcia‐Closas M.et al Assessing the probability that a positive report is false: an approach for molecular epidemiology studies. J Natl Canc Inst 200496434–442. [DOI] [PMC free article] [PubMed] [Google Scholar]