Abstract
The interest of the medical community has recently focused on the relationship between gender and cardiocerebrovascular risk. In this paper, we will first review gender differences in cardiocerebrovascular disorders, then we will discuss the existing evidence on the links between migraine and stroke and cardiovascular disease in women, and speculate on the possible physiopathological interpretations of this emerging epidemiological link. In the third part of this work, we will address the issue of the effect of sex hormones on vascular risk, and consider the evidence concerning the safety of oral contraceptives and hormone-replacement therapy. Finally, we will outline the main trends of future research and its possible clinical and therapeutic implications
Keywords
Vascular disorders have a huge social and economic impact in industrialized countries: myocardial infarction is the first cause of death in the western world, while stroke ranks third in mortality and is the leading cause of permanent disability [1]. In recent years, the interest of the medical community has focused on the relationship between gender and cardiocerebrovascular risk, for several reasons. First, the incidence of stroke is increasing in females, as a consequence of longer life expectancy and variations in women's lifestyles [2]. Furthermore, the incidence of heart attack is lower in women of premenopausal age compared with men, while the risk of both genders becomes similar after menopause. This fact has suggested a possible role of estrogens in vascular diseases, and has boosted both bench and clinical research. Second, several studies have suggested a possible relationship between migraine and ischemic stroke. However, this association has been consistently demonstrated in women only, pointing to a sex-specific mechanism in the pathogenesis of this type of cerebrovascular event. Moreover, migraine is much more frequent in females than in males, so that the relevance of migraine headache as a risk factor for stroke is of particular importance for women.
Gender differences in vascular diseases
Epidemiological data
The incidence of myocardial infarction shows a marked sex relationship: men have a higher incidence in all age groups, but this difference decreases progressively after menopause [3]. On the other hand, the incidence of stroke is also higher in men than in women, but no variation with increasing age has been observed, and the sex-specific difference is stable over the years (Figure 1) [4]. Although stroke incidence is 20–30% higher in males, the majority of deaths due to stroke occur in females [5,6]. A possible explanation is that strokes are more severe in women, possibly as a result of different pathogenic mechanisms. Furthermore, strokes in women are more disabling and associated with worse functional outcomes [5]. For this reason and due to the longer life expectancy of women, the number of female stroke survivors with severe disability is higher than that of male survivors (2.1 million vs 1.9 million) in the USA. As a result of the aging of the population, the incidence of stroke and its associated mortality and disability will considerably increase in future years.
Incidence of myocardial and brain infarction in men and women according to age category: 40-year follow-up.
Sex differences in stroke pathophysiology
The available data suggest that, although the basic pathogenic mechanisms of ischemic stroke are the same in both genders, their relative importance may differ in women compared with men. In particular, women have been shown to be less frequently affected by strokes due to large-vessel atheromatosis and more frequently by those with cardioembolic etiology [7]. As this latter type of stroke is associated with a higher severity, this different etiological distribution may partly explain the epidemiological data discussed in the previous section.
The reduced incidence of atherothrombotic stroke in females may be due to several factors: first, women have been reported to be less frequently affected by stenosis of the inner carotid artery, with a reported prevalence ranging from 1.8–2.3 compared with men of similar age [8]. Second, atheromatous plaques are more echolucent and irregular in men [8], two factors associated with an increased embolic risk. Furthermore, the anatomic location is different between the genders, as plaques tend to be located at the carotid bifurcation, often involving external carotid artery (ECA) origin, in women, while they are generally situated distal to the carotid bulb in men [9]. Both the different site and composition of the plaques may in turn be a consequence of a different anatomy of carotid vessels. In fact, women have been shown to have a larger internal carotid artery to common carotid artery ratio and a smaller ECA ratio [9], likely leading to different hemodynamic features and localization of vessel-wall stress.
These anatomical differences may also have important consequences from a therapeutic perspective. For example, the largest trial on endoarterectomy so far conducted, the North American Symptomatic Carotid Endarterectomy Trial (NASCET) has indicated important sex differences in the effectiveness and safety of the procedure. While surgical therapy was shown to be superior to medical therapy in all patients with stenosis above 70% and to be ineffective for stenosis below 50%, the outcomes in the range 50–69% were different in the two genders [10]: while men in this group benefited from surgery, women did not (Figure 2). Sex has also been found to be relevant in relation to the timing of surgery: in a revision of European carotid surgery trial (ECST) data, endoarterectomy had a positive effect in females with tight carotid stenosis only if performed within 2 weeks from the qualifying event [11]. Similar data were reported in patients with asymptomatic stenosis: in the Asymptomatic Carotid Artery Stenosis trial [12], no surgical benefit was observed in women compared with men. Furthermore, the rates of severe adverse events were higher in women in this trial, and a higher frequency of embolization [13], monitored with transcranial Doppler, was observed.
Incidence of vascular events in women with and without aura and in controls.
Finally, women have been shown to be at higher risk of subarachnoid hemorrhage and to have a higher incidence of intracranial aneurysms [14]; as a consequence mortality from subarachnoid hemorrhage is four-times higher in females. These data may also contribute to the higher stroke-related mortality and disability associated with the female sex.
Migraine & cardiocerebrovascular risk in women
Association between migraine & cerebrovascular diseases
Migraine is a chronic disorder of the CNS characterized by recurrent episodes of intense headache, often associated with nausea and vomiting. Its prevalence in western countries is elevated and is characterized by a striking gender difference: up to 18% of women are affected by migraine compared with only 8% of men [15]. In approximately one patient out of four the pain is preceded by focal neurological symptoms that may last up to 1 h, most frequently affecting the visual system [15]. This latter form of the disease is termed migraine with aura (MA), while the more common form, without focal symptoms, is referred to as migraine without aura (MwA).
Several studies have established a link between migraine and cerebrovascular risk. Three case–control, retrospective studies have shown an increased risk of ischemic stroke in female patients with MA and age less than 45 years (relative risk [RR]: 6.2, 95% confidence interval [CI]: 2.1–18) [16–18]. Two other studies have found such a relationship in MA patients of both sexes [19,20]; only one study has so far demonstrated an increased stroke risk associated with MwA [16]. Among these, some studies have demonstrated that the risk of stroke is further increased if migraine is associated with cigarette smoking (RR: 10.2, 95% CI: 3.5–29.9) or with the use of oral contraceptives (RR: 13.9, 95% CI: 5.5–35.1) [16,17]. The combination of MA, smoking and estroprogestinic hormones considerably increased cerebrovascular risk (RR: 34.4; 32.7–36.1). Three other studies failed to confirm an association between stroke risk and migraine [21–23].
The evidence from retrospective studies has been confirmed in large, prospective studies: the Physician's Health Study [24], which included more than 22,000 subjects, showed a doubled risk of stroke in migraineurs, while the National Health and Nutrition Evaluation Survey [25], with a population of approximately 12,000 subjects, highlighted a relative risk for migraine sufferers of 2.5. Finally, the analysis of two large databases in the USA (30,411 people) [26] and the UK (63,575 people) [27], yielded similar results (relative risk for stroke in migraineurs: 1.13–1.67); unfortunately, however, these databases did not include a distinction between MA and MwA, making subgroup analysis impossible. All the epidemiological evidence so far discussed has been included in a systemic meta-analysis, with the following results: the relative risk of ischemic stroke for all migraineurs was 2.16 (95% CI: 1.89–2.48); the risk was greater for MA (RR: 2.27; 95% CI: 1.61–3.19) as compared with MwA (RR: 1.83; 95% CI: 1.61–3.19), and once again an increased hazard in case of assumption of oral contraceptives was demonstrated (RR: 8.27; 95% CI: 5.05–15.05) [28].
However the actual clinical relevance of this increased RR is limited, as the incidence of cerebrovascular disorders in women aged 45 years or under is very low. In fact, the absolute risk of stroke in young female migraineurs can be estimated as approximately 18/100,000 person-years.
Until recently, the association between migraine and stroke was thought to be limited to young women, but two large randomized trials seem to indicate that this link may also be present also in older age groups. The Women Health Study (WHS) enrolled 27,840 health professionals aged over 45 years, with the aim of verifying the usefulness of aspirin in primary prevention [29]. In a secondary analysis, a correlation between stroke and MA (odds ratio [OR]: 2.05; 95% CI: 1.51–2.78), but not MwA, was found. The incidence of cerebro- and cardio-vascular events in the WHS are shown in Figure 2. Similar findings were achieved in the Atherosclerosis Risk in Communities Study (ARCS) [30], a randomized trial with a study population of 12,750 subjects of both sexes: patients with MA were found to have a 2.81 RR of ischemic stroke (95% CI: 1.6–4.92) and a 4.28 RR of transient ischemic attack (TIA) (95% CI: 3.02–6.08). The absolute increase of cerebrovascular risk is also limited for women in this age group, but is considerably greater than that of younger women: in the WHS study, 3.8 strokes per 10,000 persons-years could be attributed to MA [29].
Finally, the existence of a possible link between migraine and stroke is suggested by the finding that migraine seems to be a risk factor for subclinical brain lesions. In a recent meta-analysis considering seven case–control studies, the incidence of areas of increased signal at nuclear magnetic resonance (NMR) of the brain was fourfold in migraine patients (OR: 3.9; 95% CI: 2.3–6.7) [31]. Signal abnormalities in the deep white matter of the brain were more frequent in women (OR: 2.1; 95% CI: 1.0–4.1) and in patients with frequent attacks (OR: 2.6; 95% CI: 1.2–5.7). Similar findings have been reported in a cross-sectional study including 161 MA patients, 134 MwA patients and 140 controls [32], in which a higher incidence of cerebellar lesions was found in migraineurs compared with controls (5.4 vs 0.7%; OR: 7.1; 95% CI: 0.9–55); this difference was even more marked for MA (OR: 13.7; 95% CI: 1.7–112). The highest frequency of cerebellar lesions was found in MA patients with more than one attack per month (OR: 15.8; 95% CI: 1.8–140). Both the location and the shape and signal intensity of these areas suggest that they may have an ischemic origin [33]. These data may suggest that a high frequency of migraine auras may also be correlated with cerebrovascular risk, although direct evidence of this link in epidemiological studies is missing.
To summarize, despite a few negative results, the available scientific evidence points to the existence of an association between MA and risk of cerebrovascular events. This link has been more consistently established in women, and also seems to be present in the older age groups. Association of smoking with oral contraceptives strongly increases the risk.
Association between migraine & cardiovascular diseases
Until very recently, data on the association between cardiovascular disorders and migraine were scant and inconsistence. The first work to report such an association was a community-based study conducted in Australia with 3654 subjects aged over 49 years, in which higher odds of myocardial infarction were shown in migraineurs [34]. In the ARCS, which enrolled 12,887 participants of both sexes, migraine patients were found to more frequently report chest pain or to have a history of angina, whereas the rates of myocardial infarction were similar in migraineurs and nonmigraineurs [35]. Similar results were obtained in the California Kaiser Permanence Care Programs, in which higher odds of angina or chest pain, but not myocardial infarction, were documented [36]. The higher incidence of chest pain had previously been attributed to a lower pain threshold in migraineurs, but more recent results seem to suggest a different interpretation.
To date, two large, prospective trials have specifically addressed the issue of cardiovascular risk in migraineurs: the Physician's Health Study [37] and the WHS [38]. The two studies yielded conflicting results: in the former the risk of myocardial infarction or any coronary event did not differ between migraineurs and nonmigraineurs, while in the latter several significant correlations were found. In this large cohort study, which enrolled only women, MA patients had multivariate adjusted hazard ratios for major cardiovascular disease of 2.15 (95% CI: 1.58–2.92; p < 0.001), for myocardial infarction of 2.08 (95% CI: 1.3–3.31; p = 0.002), for coronary revascularization of 1.74 (95% CI: 1.23–2.46; p = 0.002), 1.71 for angina (95% CI: 1.16–2.53; p = 0.007) and 2.33 for cardiovascular death (CI: 1.21–4.53; p = 0.01). The correlation between MA and cardiovascular disease was maintained after adjusting for conventional risk factors, such as smoking or hypertension (
Risk of venous thrombosis with conditions specific to women.
hx: History; OC: Oral contraceptive; VT: Venous thrombosis.
Several methodological factors may be responsible for the discrepancies between the various studies. First, some studies did not consider MA and MwA as separate entities. Since most studies on cerebrovascular disorders have shown that only the former is linked with increased vascular risk, failure to distinguish the two forms of migraine may have biased the results. Second, most works have studied the association in mixed cohorts of both genders. Since, so far, a correlation with vascular disorders has emerged only in women, targeting studies on female patients only may help increase statistical power.
The mechanisms through which MA increases cardiovascular risk are unknown, However, MA has been found to be more frequent in women affected by other vasospastic phenomena, such as Raynoud's syndrome [39]. Thus, vascular hyper-activity and vasospasm may be a possible pathogenic explanation. In accordance with this hypothesis, MA has been reported to be more frequent in patients with Prinzmetal angina [39]. Furthermore, a very recent study, in which coronary angiography was performed in women reporting chest pain, demonstrated that coronary atheromatous lesions are actually less frequent in women with migraine [40]. This apparently paradoxical finding may be explained by methodological issues, as the authors themselves point out, but may also reflect a different pathogenic mechanism of coronary syndromes in women with migraine, such as vasospasm.
Possible biological mechanisms of the association between migraine & stroke
Several mechanisms might theoretically explain the above-mentioned epidemiological evidence. The first possibility may be that the migraine attack itself may cause cerebral ischemia. During migraine auras, a wave of massive neuronal depolarization, followed by a prolonged hyper-polarization, arises in the visual cortex and then spreads over the brain cortex. This phenomenon, known as cortical spreading depression (CSD) is responsible for the focal neurological symptoms and is associated with a transient increase in brain perfusion, which is then followed by prolonged oligoemia [41]. Although the level of brain perfusion doesn't fall below the ischemic threshold [42], reduced blood flow during migraine auras combined with vascular alteration and increased platelet aggregability might contribute to the pathogenesis of ischemic strokes. In fact, MA has been associated with an increased level of prothrombic and vasoactive factors [43], such as protrombin 1 and 2 and von Willebrand factor [44], and of circulating and platelet serotonin [45], as well as with increased platelet aggregability [46] and polymorphisms of the methyltetrahydrofolate reductase genes [47], responsible for high blood homocysteine levels. All these alterations may lead to elevated blood coagulability and vasospasm, thus increasing the possibility of clot formation in conditions of low blood flow.
Ischemic strokes directly due to migraine aura are termed migrainous infarction; the International Headache Society has defined it as one or more typical aura symptoms lasting more than 60 min and with the demonstration of an ischemic lesion in an appropriate brain area; no other possible causes of cerebral ischemic events must be present after accurate analysis [48]. The incidence of this disorder varies across the different studies, but it is probably responsible for 1–2% of ischemic strokes in young patients [49–52]. The magnitude of the migraine–stroke association must be therefore explained by other phenomena. Furthermore, it should be noted that no sex difference has been reported in the incidence of migraineus infarcts.
A second possibility may be a different cardiovascular risk profile in migraineurs. A recent prospective study with 5755 subjects has shown that migraine patients have worse cardio- and cerebro-vascular conventional risk factors, with a larger percentage of smokers and a higher incidence of relatives affected by cardiac disorders [53]. Moreover, migraine patients had higher cholesterol and blood pressure levels and were more likely to take oral contraceptives. As a consequence, migraine patients were at a twofold risk of having an unfavorable Framingham cardiovascular score compared with nonemicranic subjects. However, it should be stressed that in both the WHS and the ARCS the correlation between MA and stroke was still present after a multivariate analysis for conventional risk factors.
Another possible mechanism may be a higher probability of artery dissection and distal embolization in migraine patients. An association between cervical artery dissection and MA has been found in a retrospective study [54]. This finding may be due to genetic differences in the structural proteins of vessel walls or to higher activity of enzymes involved in protein degradation [55]. An alternative hypothesis may be that repeated vasospasm and endothelial dysfunction during migraine auras may directly damage the vessel wall and induce arterial dissection. Interestingly, fibromuscolar dysplasia type 2, a disorder of connective tissues associated with an increased risk of dissection, has been found to be more common in women than in men [56]. Women may thus be more prone to artery dissection and stroke during vasospasms. However, cervical artery dissection is a rare phenomenon and may account for only a fraction of strokes.
Another frequent condition that may link migraine and brain ischemia is the presence of a patent foramen ovale (PFO), a congenital abnormality of the interatrial septum. Four retrospective studies suggested that a large PFO may be more frequent in MA patients than in the general population [57–60]; moreover, several studies suggest that PFO closure may lead to a reduction in aura frequency and severity [61,62]. However, these studies may be biased by several methodological flaws, such as the presence of a recall effect, lack of control groups and the pharmacological effect of platelet antiaggregants, routinely administered to operated patients. Thus, the actual clinical usefulness of PFO closure for migraine will be established only after the results of randomized trials, currently ongoing in several European countries, are available. PFO has also been reported to be associated with cryptogenic stroke: in a recent meta-analysis the risk of ischemic stroke was increased for patients all patients with PFO (RR: 1.83; CI: 1.25–2.26), and was further increased if an atrial septal aneurysm was also present and if the patient was aged below 55 years (RR: 15.59; 95% CI: 2.83–85.7) [63]. Although no difference in the incidence of PFO between men and women has ever been documented, PFO may indeed be indirectly responsible for the sex-specificity of cerebrovascular risk associated with migraine, as women are more likely to have peripheral venous thrombosis as a result of pregnancy or OC use. However, the presence of a PFO cannot explain the increased risk of coronary disease in MA patients, which is more likely to be linked to other mechanisms, such as platelet or vascular over-reactivity.
Finally, one last possibility that needs to be considered is that some of the cerebrovascular events that occur in migraineurs may be due to the use of triptans. Triptans have a vasoconstrictive effect that has raised concerns regarding their use in MA patients, in whom increased vascular reactivity is indicated. However, no increased incidence of cardio- or cerebro-vascular events in connection with current or previous triptan use has been documented in several postmarketing studies [64–68], and in data derived from large, community-based databases in the UK and USA [24,25]. Thus, although caution in triptan use in MA patients may still be advised from a clinical standpoint, there is currently insufficient scientific evidence to establish if the link between migraine and stroke may be partly explained by medical treatment.
The influence of sex hormones on cerebrovascular risk
Hormone-replacement therapy & risk of stroke
The relationship between sex hormones and vascular risk is complex. The risk of venous thrombosis, including cerebral venous thrombosis, is increased in several conditions of altered sex steroid balance, such as pregnancy/puerperion or hormone-replacement therapy (HRT), as shown in Table 1 [69]. The association with arterial stroke is less clearly established. Data on in vitro and animal models seem to indicate that estrogens may have a positive effect on coagulation, blood lipids and neuronal regeneration, and thus exert a protective role against vascular disease [70,71]. This interpretation is supported by the lower incidence of cardiovascular disease in fertile women compared with man [3], and by its rapid increase after menopause. However, the large randomized trials testing the protective effect of HRT have yielded disappointing results: in the Heart and Estrogen & Progesterone Replacement Study (HERS), which enrolled 2763 women, no evidence of reduced cardio- or cerebro-vascular risk was found [72,73], while in the Women's Estrogen for Stroke Trial (WEST), conducted on 664 women, the risk of ischemic stroke for women receiving active treatment was actually increased (RR: 2.3; 95% CI: 1.1/6.0) [74]. A third trial, with a much larger population (16,808 women), the Women's Health Initiative trial, was stopped because of an increased risk of stroke and breast tumor in women treated with HRT [75–77]. A post-hoc analysis of the HERS and WEST trials has highlighted a possible protective role of antiaggregants against venous thrombosis, but not against arterial stroke. The scientific evidence available is therefore against the use of HRT for the prevention of cardio- or cerebro-vascular disorders. No interaction between HRT and migraine, with or without aura, with regards to cerebrovascular risk, has so far been documented.
Oral contraceptives & stroke
As previously reported, the association between OC and MA significantly increases cerebrovascular risk. As a consequence, OCs should probably be avoided by women with MA, who may instead be treated with progesterone-only contraceptives. Since cerebrovascular risk is further increased by smoking, all women who suffer from migraine should be strongly advised to stop smoking.
Studies of the risk of stroke connected to OC in all women have yielded conflicting results. A WHO study has demonstrated an increased hazard only for doses greater than 50 μg (OR: 5.3; 95% CI: 2.5–11) [78]; similar results have been obtained in the Royal College General Practitioners Study [79] and in a case–control study in California [80], both of which confirmed the safety of low-dosage OC. However, two other studies have given different results: both a case–control Danish study [81] and the prospective Risk of Arterial Thrombosis in relation to OCs (RATIO) study [82] have demonstrated an increased stroke incidence in association with the use of third-generation OCs (OR: 1.8; 95% CI: 1.1–1.9 and OR: 2.3; 95% CI: 1.3–1.9, respectively). Finally, a meta-analysis of 16 studies has also confirmed an association between cerebrovascular risk and OC use, even taking into account the potential confounding effect of smoking and arterial hypertension (RR: 1.93; 95% CI: 1.4–2.7) [83].
Conclusion & future perspective
Although the link between MA and cerebro- and cardio-vascular risk has now been established, several issues of clinical relevance are still open to future research. The first point concerns risk stratification and use of antiplatelet drugs. Since the absolute risk of stroke is low in MA women below 45 years, antiaggregant therapy for all patients in this group is probably not indicated. However, the identification of specific groups at increased risk would allow treating only those patients with a favorable risk:benefit balance. Some indicators that could be used for such stratification include the presence of a large PFO with evidence of right–left-shunt or the presence of NMR signal abnormality of the deep white matter or cerebellum. Establishing if the vascular risks of MA, PFO or NMR abnormality may be additive could be an important goal of future investigation. The identification of patients at higher risk could also be useful to screen candidates for PFO closure. Several studies are currently investigating the usefulness of this procedure both for stroke prevention and for treatment of severe MA, and the first results will soon be available.
A second important issue concerns the different response to antiplatelet drugs between the sexes. A very recent meta-analysis, which included data from the WHS, the Hypertension Optimal Treatment Study and the Primary Prevention Project, demonstrated that treatment with aspirin can reduce the risk of stroke by 19% (RR: 0.81; 95% CI: 0.69–0.96; p = 0.01) in women older than 45 years, while the same finding has not been confirmed in males [84]. The reasons for this statistical difference are not clear, but may reflect a specific effect on platelet aggregation in females.
Finally, as the trials investigating a possible protective role of female sex steroids against vascular disorders have been disappointing, one of the major goals of pharmacological research in this field will be centered on highly specific estrogen-receptor agonists. These drugs may potentially have the positive vascular effects of steroid hormones without their negative action on coagulation.
Overall, the study of the relationships between migraine and cerebro- or cardio-vascular diseases have taught the medical community that gender may be important both from a pathogenetic and therapeutic standpoint. Moreover, since epidemiological projections show us that women will have to endure the higher social and economic burden of cerebrovascular disorders, understanding the specific patterns of these diseases in the female sex will be a major duty of healthcare workers in the years to come.
Executive summary
Women endure the heavier social burden of stroke, as the majority of strokes occur in women, and as strokes in females are more severe and associated with greater disability. Owing to demographic phenomena, this trend is likely to continue in future years.
Stroke pathophysiology is different between the genders: women have more cardioembolic strokes and fewer atheromatous lesions. Endoarterectomy for stenosis 50–70% is not effective and potentially harmful in women.
The relationship between migraine with aura and ischemic stroke has been consistently confirmed only in women. The risk is notably increased by smoking and oral contraceptives. Recent data extend this association to women older than 45 years; women of all ages with migraine with aura should be strongly advised to stop smoking.
A correlation between migraine with aura and coronary syndromes, as well as myocardial infarction, has emerged in recent years. This link may be due to increased vasospasm or platelet hyperaggregation.
Patent foramen ovale (PFO) has been reported to be more common in migraine with aura and some studies indicate that PFO closure may reduce migraine severity. Until data from randomized trials are available, however, PFO closure should not be performed for treatment of migraine.
Data from randomized trials reported an increased cardio- and cerebro-vascular risk in women receiving hormone-replacement therapy. Therefore, hormone-replacement therapy should not be used for vascular protection.
The safety of third-generation contraceptives (low estrogen content) has not been firmly established, as they may increase vascular risk. Since this risk is strongly increased in women with migraine with aura, combination oral contraceptives should not be used by women with this disease.
Recent data indicate that aspirin might reduce the risk of stroke in women aged over 45 years. Aspirin treatment might be considered for primary prevention of stroke only in a subgroup of patients, such as those with migraine with aura with high frequency of attacks, long duration of disease and other associated risk factors.