Migraine and Genetic and Acquired Vasculopathies

Abstract

It is remarkable that migraine is a prominent part of the phenotype of several genetic vasculopathies, including cerebral autosomal dominant arteriopathy with subcortical infarcts and leucoencephalopathy (CADASIL), retinal vasculopathy with cerebral leukodystrophy (RVCL) and hereditary infantile hemiparessis, retinal arteriolar tortuosity and leukoencephalopahty (HIHRATL). The mechanisms by which these genetic vasculopathies give rise to migraine are still unclear. Common genetic susceptibility, increased susceptibility to cortical spreading depression (CSD) and vascular endothelial dysfunction are among the possible explanations. The relation between migraine and acquired vasculopathies such as ischaemic stroke and coronary heart disease has long been established, further supporting a role of the (cerebral) blood vessels in migraine. This review focuses on genetic and acquired vasculopathies associated with migraine. We speculate how genetic and acquired vascular mechanisms might be involved in migraine.

INTRODUCTION

A vascular component in the aetiology of migraine has been debated for many years (1). Recent experimental data suggest that vasodilatation of intra- and extracerebral vessels is no more than an epiphenomenon of migraine or even may not occur at all (1–4). On the other hand, several studies provide evidence for a strong vascular component in migraine (5–7), which is supported by the clinical observation that migraine, in particular migraine with aura (MA), and several genetic and acquired vasculopathies can co-occur. Acquired vasculopathies including ischaemic stroke (8), ischaemic heart disease (8), and arterial dissection (9) have been associated with migraine. In addition, it is known that vascular anomalies such as arteriovenous malformations or small angiomas can cause MA (10–12). Also, migraine has been associated with reversible cerebral vasoconstriction syndromes (RCVS) (13). Genetic vasculopathies where migraine can co-occur in patients include cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathies (CADASIL) (14), retinal vasculopathy with cerebral leukodystrophy (RVCL) (15, 16) and hereditary infantile hemiparesis, retinal arterial tortuosity, and leukoencephalopathy (HIHRATL) (17). The aim of this review is to discuss acquired and genetic vasculopathies that are associated with migraine in the light of the vascular component in the migraine pathophysiology.

CEREBRAL BLOOD VESSELS AND MIGRAINE

Traditionally, two theories regarding the aetiology of migraine exist: the neuronal and the vascular theory (18). In the classical vascular hypothesis the migraine aura is related to ischaemia caused by intracerebral vasoconstriction, whereas the headache is attributed to (rebound) vasodilatation of cerebral and meningeal blood vessels (19). This purely vascular hypothesis has nowadays been discarded (1). Instead, migraine is considered a primary brain disorder with neuronal events (e.g. cortical spreading depression (CSD)) causing the aura (20). Dysfunction of certain brainstem nuclei, associated with nociceptive processing, may also play a role (21). In this theory, vasodilatation of cerebral and meningeal vessels, if present, is considered a secondary phenomenon that may occur after activation of the trigeminovascular system (TGVS) (22). Activation of trigeminovascular efferents leads to the release of vasoactive neuropeptides (e.g. CGRP, Substance P and NO), which are believed to cause neurogenic inflammation, central pain transmission and headache.

Although the role of neurogenic inflammation in patients has never been shown, neurogenic inflammation of the dura and around the meningeal vessels has clearly been demonstrated in experimental animal models of migraine (23). The same experiments also have shown vasodilatation of these vessels (23). A role for vasodilatation in migraine in humans is supported by the fact that specific antimigraine drugs, the triptans and ergotamine, deactivate the TGVS, can inhibit the release of vasoactive neuropeptides from perivascular nerve terminals, and have a vasoconstrictive effect (24). Recent studies have casted considerable doubt on whether our idea about the role of vasodilatation in migraine is correct. For instance, CGRP antagonists do not have a vasoconstrictive effect, but are very effective in the treatment of migraine (25).

In addition, despite common believe, there is very little factual evidence that vasodilatation really exists during migraine headache attacks. This is probably because studies were hampered by the fact that sensitive non-invasive techniques to assess intracranial blood flow were not available at the time. A recent study by Schoonman and co-workers, however, using a sensitive 3 Tesla MRA-technique, failed to show in vivo cerebral and meningeal vasodilatation in humans during migraine headache (2). Nevertheless, although this study did not provide evidence for vasodilatation of large meningeal and cerebral vessels during migraine headache, it does not rule out a role for small cerebral vessels.

The latter is relevant because small cerebral vessels were shown to cause blood flow changes that occur during CSD. CSD is characterized by a brief (seconds) wave of intense neuronal and glial depolarization that slowly (2–5 mm/min) propagates across the cerebral cortex. The wave is followed by a potent, relatively long-lasting (≥20 min) neuronal suppression (20, 26). There is increasing evidence that CSD is the underlying mechanism of the clinical migraine aura (27). An experimental study in rats showed that CSD can activate the TGVS, and thereby trigger headache mechanisms (23). In humans, however, there is no direct proof for the relevance of CSD in triggering headache mechanisms. Functional neuroimaging studies that showed similarities between blood flow changes in patients during visual auras and CSD in experimental animal models suggest that CSD may also occur in humans (28); the blood flow changes consisted of an initial hyperaemia during depolarization of neuronal and glial membranes, followed by a reduction of cerebral blood flow during hyperpolarization, and suppression of neuronal and glial membranes (20, 28–31). This reduction in cerebral blood flow does not seem to reach ischaemic threshold and therefore is referred to as oligaemia (28–31). Changes in blood flow are considered to follow the virtually increased and then reduced metabolic demand of neurons and glial cells during CSD.

Recently, it was demonstrated that a reverse order of events, i.e. a vascular event that is able to trigger CSD, may also be possible (5–7). Dreier et al. showed that the vasoconstrictive peptide endothelin-1 (ET-1) induced changes characteristic of CSD in the rat cortex, which were mediated by NMDA receptors (5). As ET-1 is not capable of inducing CSD in rat brain slices without intact perfusion, it was suggested that a vascular-mediated event acted as a trigger for CSD. Dreier and co-workers also showed that ET-1 induces neuronal damage that was due to ET1A receptor activation (6). They proposed that a vascular neuronal process is able to cause the migraine aura: a clinical insignificant ischaemic micro-area (caused by vasoconstriction or a small embolus) giving rise to CSD and neuronal dysfunction in a larger volume of tissue. They put this theory forward as an explanation for the association between migraine and patent foramen ovale (PFO) (32), which increases the risk of small cerebral emboli. Changes in vessel diameter and blood flow changes secondary to neuronal activity were also the subject of a study by Brennan and co-workers (7). Using optical intrinsic signal imaging combined with electrophysiological techniques, they demonstrated that vasomotor changes in the cerebral cortex travel: (i) at a significantly greater velocity than the neuronal changes; (ii) with a different pattern (circuitous along arterioles as opposed to the concentric parenchymal CSD pattern); and (iii) dissociated from neuronal changes (it extended beyond the margins of the spread of parenchymal CSD). Thus, although it is generally accepted that during a migraine attack alterations in neuronal activity (e.g. CSD, dysfunction in ion transport and brainstem dysfunction) precede vascular changes, their data suggest that vascular alterations may trigger neuronal dysfunction, and not the other way around.

The clinical observation that migraine is associated with several acquired vasculopathies (i.e. ischaemic stroke, ischaemic heart disease, arterial dissection, arteriovenous malformation (AVM) and RCVS) as well as with monogenetic cerebral small vessel diseases (RVCL, CADASIL and HIHRATL) further indicates that vascular changes may increase susceptibility to migraine. Here, we will discuss these vasculopathies in more detail.

GENETIC VASCULOPATHIES

Cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy

Cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL) is an autosomal dominant late-onset arteriopathy that is clinically characterized by recurrent transient ischaemic attacks (TIAs) and strokes leading to cognitive decline, psychiatric symptoms and dementia (33). In about one-third of patients migraine with aura occurs, often as the presenting symptom several years before the other symptoms (34).

CADASIL has distinct neuroradiological features that are usually manifest prior to the first stroke. The most important features are white matter hyperintensities (WMH) and lacunar infarcts. WMH are symmetrically distributed and located in the deep and periventricular white matter. Typical for CADASIL is the bilateral involvement of the anterior temporal lobes and external capsule. Other neuroradiological features are subcortical lacunar lesions, lacunar infarcts and microbleeds (35, 36).

Histopathologically, CADASIL is characterized by abnormalities of the wall of small arteries and arterioles. Typical for CADASIL is the deposition of granular osmiophilic material (GOM) in the basement membrane and surrounding extracellular matrix of vascular smooth muscle cells (VSMCs) as well as the degeneration and eventual disappearance of VSMCs. These specific pathological changes are not limited to the cerebral vasculature, but are ubiquitously present in the arterial system, allowing for histopathological confirmation of the diagnosis by skin biopsy, in addition to genetic screening (37). Endothelial structural changes have also been reported in CADASIL (38).

CADASIL is caused by mutations in the NOTCH3 gene (located on chromosome 19p13.2-p13.1), which encodes a cell surface receptor that, in human adult tissue, is solely expressed on vascular smooth muscle cells (39, 40). The Notch3 protein is a heterodimeric receptor with an extracellular, ligand binding domain, a transmembrane domaine and a cystoplastmic domain. It is part of a signal transduction pathway, critical for aspects of vascular development, homeostasis and VSMC differentation (41).

Most mutations in CADASIL are missense mutations and involve the loss or gain of a cysteine residue in the Notch3 protein (specifically the extracellular epidermal-growth-factor-like repeat). The exact pathogenic mechanism by which NOTCH3 mutations lead to CADASIL has not yet been deciphered. More insight comes from the study of a Notch3 transgenic mouse model harbouring the archetypical R90C mutation, enabling the in vivo study of mutated Notch3 and the time-course of its effects (42). The mouse model showed that degeneration of VSMCs precedes the accumulation of GOM and Notch3 as well as a disruption of anchorage of VSMCs to adjacent cells and extracellular matrix. The current idea is that mutations in the NOTCH3 gene have a gain-of-function effect on the protein and that disrupted anchorage triggers VSMC degeneration (43). The variable disease course of CADASIL, ranging from relatively mild to very severe (34, 44), might be an indication for the involvement of other genetic (45) and environmental modifying factors.

The prevalence of migraine with aura, but not without aura, is significantly increased in CADASIL. This suggests that CADASIL is more associated with aura than with migraine headache. It is unknown why migraine with aura prevalence is increased in CADASIL. Possible mechanism will be discussed later in this review.

Retinal vasculopathy with cerebral leukodystrophy

Retinal vasculopathy with cerebral leukodystrophy (RVCL) is a neurovascular syndrome that primarily involves the retina and the central nervous system (15). The most prominent symptom is a vascular retinopathy, which often is difficult to distinguish from diabetic or hypertensive retinopathy. This may lead to a delay in the diagnosis when other symptoms of the disease are (still) absent, or when the family history is not carefully evaluated. Neurological manifestations may include cognitive disturbances, depression, migraine (mainly without aura) and focal neurological symptoms. In later disease stages, cerebral MRI scans often show characteristic contrast-enhancing intracerebral mass lesions. Several systemic symptoms can be present as well, including renal and liver dysfunction, Raynaud's phenomenon and gastro-intestinal bleeding.

What is now called RVCL was originally described in three families under different disease names and abbreviations (i.e. cerebroretinal vasculopathy (CRV) (46), hereditary vascular retinopathy (HVR) (16) and hereditary endotheliopathy, retinopathy, nephropathy and stroke (HERNS) (47)). Several years after mapping of the disease locus in the three families to chromosome 3p21.1-p21.3 (48) the disease causing gene, encoding the 3′-5′exonuclease TREX1, was identified (15). With the discovery of the gene, it is now well established that CRV, HVR and HERNS are overlapping phenotypes of the same disease entity.

Histopathological examination of cerebral tissue (46, 47, 49) showed white matter necrosis, fibrinoid necrosis, thrombosis of microvessels with perivascular inflammatory infiltrates (plasma cells and lymphocytes) and reactive gliosis of astroglia. In one of the families, electronmicroscopic examination showed a multilaminated capillary basement membrane (47). In another family fine calcium deposits within necrotic foci were described (46).

The exact function of the TREX1 gene has not been elucidated yet. Currently, three different roles have been attributed to the protein. First, it encodes a 3′-5′exonuclease, which might suggest a role in DNA editing and repair (50). However, because Trex1 knockout mice do not have an increased spontaneous mutation rate or higher cancer incidence, it is less likely that this exonuclease exerts such a function. Because these mice develop an inflammatory myocarditis it seems that Trex1 is involved in auto-immune mediated processses (51). Secondly, the Trex1 protein is part of the SET complex, which is involved in a specific form of apoptosis induced by cytoxic T-cells and killer lymphocytes (i.e. Granzyme A-mediated cell death) (52, 53). Normally, the SET complex resides in the cytoplasm attached to the membrane of the endoplasmatic reticulum, but in response to Granzyme A-induced oxidative stress, it can move to the nucleus and attack nuclear DNA, leading to apoptosis. Thirdly, recently, a role for Trex1 in cell cycle homeostasis was proposed (54). Trex1-deficient cells have an impaired G1/S-transition and a 60–65 bp ssDNA species accumulates in the cytoplasm (54). At present, five different TREX1 mutations in nine different RVCL families have been reported, all causing a truncation of the C-terminus of the Trex1 protein, resulting in altered cellular localization, while leaving the exonuclease activity intact (15).

Next to RVCL, mutations in TREX1 have been identified in several auto-immune related diseases (55–58), including Aicardi-Goutieres syndrome (AGS) (55, 56) and familial chilblain lupus (FCL) (56, 57). Also, some patients from a large cohort of SLE patients carried a TREX1 mutation (58). Both heterozygous frameshift mutations causing C-terminal truncations, similar to those found in RVCL, and missense mutations were identified in SLE patients (58). Although most of the missense mutants have not functionally been tested and it therefore is unclear whether they indeed cause SLE, the R114H mutation that has a deficient exonuclease function and was previously found in AGS, can also cause SLE. The possible relation with migraine comes from the fact that migraine, especially MA, can co-occur with SLE (59).

The current idea is that either dysfunctional Trex1 (that is with impaired exonuclease acitivity) or Trex1 functioning in the wrong cellular context because of mislocalization (in the case of truncating RVCL mutations) can lead to accumulation of DNA intermediates in the cell, which may trigger an abnormal immune response and cause disease.

Because of the other prominent clinical features, especially the retinopathy and intracerebral mass lesions, not much attention was given to the occurrence of migraine in the RVCL families. Only one of the nine TREX1 families has been systematically investigated for migraine. In this large Dutch RVCL family (n = 54) genetic evidence was obtained that TREX1 might act as a genetic modifier for migraine (16, 60). Currently, a study is being undertaken to evaluate the presence of migraine in the other RVCL families.

Hereditary infantile hemiparesis, retinal arteriolar tortuosity and leucoencephalopathy

A third genetic vasculopathy that affects both cerebral and retinal small vessels is caused by a mutation in the COL4A1 gene that is located on chromosome 13qter (61). MA has been described in a family with hereditary infantile hemiparesis, retinal arteriolar tortuosity and leucoencephalopathy, carrying COL4A1 mutation G652E (17, 62, 63).

Besides causing HIHRATL, mutations in COL4A1 are associated with a diverse clinical spectrum in both humans and mice, including familial porencephaly and intracerebral haemorrhage (61, 62, 64). Familial porencephaly is considered to be caused by peri- and antenatal cerebral haemorrhage (65, 66). In addition, ocular anterior chamber abnormalities, including congenital cataract (67), are part of the clinical spectrum. Systemic small vessels and larger cerebral vessels (intracranial aneurysms) may also be affected, as is seen in HANAC syndrome (hereditary angiopathy, nephropathy, aneurysms and muscle cramps) (68). Differences in type of COL4A1 mutations may explain this variability in clinical spectrum.

The COL4A1 gene encodes the α1 chain of type IV collagen, a ubiquitous expressed basement membrane protein, including the vascular basement membrane. Complex basement membrane defects are associated with COL4A1 mutations and reported in human skin (68), skin capillaries (65, 68) and kidney (including basement membrane of tubules, interstitial capillaries and of Bowman's capsule) (68). Histopathological examination of brain tissue of COL4A1 mutation carriers has not been performed yet. However, in Col4a1 mutant mice the cerebral vascular basement membrane is also affected, in addition to kidney and eye where similar abnormalities were found (68, 69). Abnormalities include local disruptions, irregular thickening and enlargement of endothelial cells.

After diagnosing a child with familial porencencephaly, the affected parent may appear to have leukoencephalopathy, lacunar infarcts, microbleeds and macrobleeds later in life in the absence of porencephaly, suggesting modifying genetic and environmental influences (65). In line with this, in Col4a1 mutant mice (Col4a1 ^+/Δex40), a strong gene-environment interaction was seen with a large effect of stressors such as birth trauma and hypertension (62).

In the family with hereditary infantile hemiparesis, retinal arteriolar tortuosity and leucoencephalopathy and COL4A1 mutation G652E, three out of six mutation carriers had MA (17, 62, 63). The association with MA has not been reported in phenotypes caused by other mutations, therefore the co-occurrence with MA may either be coincidental or mutation specific. Of note, in a family with hereditary porencephaly, carrying a COL4A1 mutation (a mutation affecting the methionine start codon that results either in an absent or an abnormal truncated protein), MO was described in one patient, while in two other patients the migraine subtype was not specified (70). However, as in six other reported families with hereditary porencephaly migraine was not described this finding may well be coincidental.

Possible mechanisms for the association of migraine and genetic small vessel diseases

RVCL, CADASIL and the syndrome associated with COL4A1 mutations share several features (Table 1). The most important of these for the current review are that all three affect the integrity of cerebral and systemic small vessels and that migraine is part of the phenotype, at least for RVCL and CADASIL. The mechanisms by which these vasculopathies can increase the risk of migraine is unknown, but they may be similar. Several explanations can be put forward that may partly overlap.

Table 1

Clinical, neuroradiological and pathological features of CADASIL, RVCL and HIHRATL

	CADASIL	RVCL	HIHRATL
Inheritance	Autosomal dominant	Autosomal dominant	Autosomal dominant
Gene	NOTCH3	TREX1	COL4A1
Function protein	Cell-surface receptor	3′-5′ exonuclease	α1 chain of type IV collagen (bm protein)
Effect of mutated protein	Disrupted anchorage and degeneration of vascular smooth muscle cells	Altered subcellular localization of Trex1 causes accumulation of DNA intermediates and abnormal immune response	Disruption of vascular integrity
Onset	Between 30–60 years	Between 30–60 years	Childhood, but may be diagnosed at later age
Migraine	MA	MA and MO (mainly MO)	MA
Neurological	Recurrent TIAs and strokes, cognitive decline, psychiatric symptoms, epileptic seizures, dementia	Cognitive and psychiatric disturbances, infarcts, focal neurological symptoms	Seizures, infantile hemiparesis
Neuroradiologal	Leucoencephalopathy, subcortical lacunar lesions, lacunar infarcts, microbleeds	Contrast-enhancing cerebral mass lesions, calcifications, leukodystrophy	Leucoencephalopathy, dilated perivascular spaces, intracerebral haemorrhage, cerebral and brainstem atrophy, microbleeds
Pathological	GOM in lamina basalis of VSMCs; degeneration of VSMCs	Multilayered capillary basement membranes of small cerebral vessels, skin capillaries and kidney capillaries; coagulative necrosis; obliterative vasculopathy	Complex bm abnormalities in skin (capillaries) and kidney. In mice: bm abnormalities in eye, kidney and cerebral vasculature
Extracerebral manifestations	Skin, Heart (myocardial infarction) Retina	Vascular retinopathy, renal impairment, liver impairment, Raynaud's phenomenon, GI-bleeding	Arteriolar retinal tortuosity

bm, basement membrane; CADASIL, cerebral autosomal dominant arteriopathy with cerebral leukodystrophy; GI, gastro-intestinal; GOM, granular osmiophilic material; HIHRATL, hereditary infantile hemiparessis, retinal arteriolar tortuosity and leukoencephalopathy; MA, migraine with aura; MO, migraine without aura; TIA, transient ischaemic attack; RVCL, retinal vasculopathy with cerebral leukodystrophy; VSMCs, vascular smooth muscle cells.

Spurious association: 

It is unlikely that the comorbidity of migraine with CADASIL and RVCL may be explained solely by the high population prevalence of migraine (71). It cannot be excluded that the association of migraine and COL4A1 mutations may be coincidental, considering the high population prevalence of migraine and the low number of reported mutation carriers with migraine. For CADASIL, investigation of large patient series firmly established that the prevalence of MA is increased (34). For RVCL, few families have been reported so far, thus no large studies investigating the prevalence of migraine have been performed. Evidence for an association with migraine comes from a large Dutch family where migraine clearly is part of the clinical spectrum (16).

Shared genetic factors: 

The co-occurrence with migraine may be causally related to (certain) NOTCH3, TREX1 and COL4A1 mutations. In this case, the genes may be considered to increase the susceptibility for migraine (i.e. serve as genetic modifiers). For the TREX1 gene a genetic, family-based, association study demonstrated that the RVCL locus slightly enhances the susceptibility for migraine (60). However, other risk factors must be involved as well, because migraine frequency was also increased in branches from the family in which the mutation was not transmitted. For the NOTCH3 gene, the evidence for a genetic association with migraine is conflicting. One study found an association between the NOTCH3 polymorphism G684A and migraine, but this has to be confirmed (72). On the other hand, no association was found between the NOTCH3 polymorphism T6746C and migraine (73). Unfortunately, studies performed thus far have important limitations, including small sample size. The COL4A1 gene has not been studied for genetic association with migraine.

Vascular: endothelial dysfunction: 

There is evidence that migraine attacks are associated with endothelial dysfunction.

Migraine prevalence is increased in persons with polymorphisms linked to endothelial function (74–80). Circulating endothelial progenitor cell (EPC) numbers and function have recently been shown to be reduced in migraine patients, which suggests that migraine patients may have dysfunctional endothelial regeneration (81). This may provide a biological link between migraine and the increased cardiovascular risk found in migraine patients. In addition, the brachial arteries of migraine patients were shown to have a decreased capacity for endothelial-dependent vasodilatation (82–84). In CADASIL and RVCL, pathological changes in endothelial cells may interfere with endothelial function and alter vascular reactivity. This may be a third possible explanation for the increased risk of migraine. Endothelial function has been studied in CADASIL. Mice expressing the mutated protein display early dysfunction in vasoreactivity with decreased flow-induced dilatation and increased pressure-induced myogenic tone (85). In CADASIL patients, impaired endothelial function was also found in two studies (86, 87). It would be interesting to study endothelial function in RVCL and HIHRATL as well.

Neuronal: 

CSD is thought to be the underlying mechanism of the migraine aura. A hypothesis for the increased prevalence of MA in CADASIL could be that vascular changes make the cortex more susceptible to CSD, thereby increasing the risk for MA. Indeed in CADASIL patients a reduced cerebral baseline flow (87) and decreased cerebrovascular reactivity (88–90) and an impaired endothelial function (86, 87) were demonstrated. Furthermore, it was recently shown that vasomotor changes can precede neuronal dysfunction (7) and that a vascular event, such as a clinical insignificant ischaemia, is able to trigger CSD (5, 6). These observations strengthen this hypothesis for CADASIL. Increased susceptibility for CSD seems a less plausible mechanism for RVCL, as this syndrome is mainly associated with MO. However, it could also form an explanation for the occurrence of MA in three out of six family members with HIHRATL.

ACQUIRED VASCULOPATHIES

From the above mentioned genetic vasculopathies it becomes clear that genetic defects that give rise to loss of vessel wall integrity and perhaps endothelial dysfunction may predispose to migraine. Similarly, vasculopathic changes leading to endothelial dysfunction may explain the association between migraine and cardiovascular disease. Evidence for the association between migraine and ischaemic stroke and ischaemic heart disease, as well as possible mechanisms, will be discussed in the next section. In addition, other acquired vasculopathies that can result in or are associated with migraine, such as arterial dissection, arteriovenous malformations and reversible cerebral vasoconstriction syndromes, are discussed.

Ischaemic stroke

Several studies have shown that migraine, specifically MA, is a risk factor for ischaemic stroke (91–96). A meta-analysis of 11 retrospective case-control and three prospective cohort studies showed pooled relative risks of 1.83 (95% confidence interval (CI) 1.06–3.15) for MO, and 2.27 (95% CI 1.61–3.19) for MA (91). This risk was higher in young female migraineurs below 45 years old [2.76 (95% CI 2.17–3.52)] and highest when using oral contraceptives [8.72 (95% CI 5.05–15.05)]. Since 2004, after the meta-analysis, two large prospective studies (92, 93) and two case-control studies (94, 95) have been performed. The prospective Women's Health Study also showed an increased risk for ischaemic stroke for women with MA [hazard ratio (HR) 1.71 (95% CI 1.11–2.66)], which was more increased in the youngest women in the cohort, aged between 45 and 55 (92). In a retrospective case control study in much younger women (aged 15–49) the increased risk of ischaemic stroke could also be detected [OR: 1.5 (95% CI:1.1–2.0)], increasing with higher attack frequency (OR for migraine patients with more than 12 attacks per year 2.2 (95% CI 1.5–3.3)) and even up to seven- to 10-fold with the combination of smoking and oral anticonceptive use (94). A case-control study in older adult migraineurs above 55 suggested that the higher risk of ischaemic stroke found in young and middle aged women with MA persists (in this study men and women were not separately analysed) (95). In the prospective Physician's Health Study the risk of ischaemic stroke in male migraineurs was not increased, but one of the main drawbacks of this survey is that no distinction between MA and MO could be made (93). When interpreting all these risk estimates, one must keep in mind that the absolute increase in risk for MA patients is still low. In all studies the increased risk for ischaemic stroke appeared to be independent from traditional cardiovascular risk factors (except OAC use, smoking or both) (92, 93, 95) or especially present among individuals without cardiovascular risk factors (94).

Although the above-mentioned studies are well designed and have included impressive numbers of patients, and long-term follow-up was performed in the two prospective studies, they were not specifically designed to study the migraine-stroke relationship. The main drawbacks of these studies are: (i) migraine and aura were self-reported and not diagnosed according to the IHS-criteria, a limited number of written questions were used and diagnosis was not verified by an interview; (ii) transient ischemic attacks (TIAs) could have confounded the results as they are a risk factor for ischaemic stroke and can be difficult to distinguish from the migraine aura, especially when no direct interview has been performed; and (iii) no information was given about the use of triptans or ergotamines, which may increase the risk of vascular events due to their vasoconstrictive effects. On the other hand, a retrospective study of a pharmacology register found that ergotamine overuse but not triptan overuse was associated with an increased risk of ischaemic complications (97).

Another line of evidence for an association between migraine and ischaemic stroke comes from imaging studies. In the population-based CAMERA study evidence was presented that MA patients have a 14-fold increased risk of silent infarct-like lesions in the posterior circulation territory of the brain (i.e. the cerebellar region), a risk which increases with increasing attack frequency (98). The specific MRI characteristics of these lesions suggest an infarct origin (99). As with the clinical association between ischaemic stroke and migraine, the risk was not modified by traditional cardiovascular risk factors (except for age), which again suggests the involvement of other than atherosclerotic mechanisms. A possible mechanism is that CSD, the underlying mechanism of the migraine aura, initiates a combination of hypoperfusion and thrombo-embolism formation leading to infarction (99).

Also, from the CAMERA study it became clear that in women with migraine the risk for deep white matter lesions is increased compared with controls, with no difference between MO and MA patients. This risk increases with higher attack frequency.

A strength of the CAMERA study is that migraine diagnoses were made according to the IHS criteria (100), after a telephone interview and in consultation with a neurologist specialist in headache. As the CAMERA study had a cross-sectional design a causal explanation for the increase in MRI abnormalities in migraineurs is speculative. The CAMERA follow-up study (CAMERA II), in which a second MRI scan will be made after a follow-up period of 8–9 years, should be able to clarify this relationship. First results from this study are expected at the end of 2009.

Ischaemic heart disease

Whether vascular changes in migraine are restricted to the cerebral vasculature or also systemically present is not yet clear. Systemic vascular dysfunction is suggested by the association of migraine with an unfavourable cardiovascular risk factor profile (101), prothrombotic and vasoactive factors (102), and the relation with systemic endothelial dysfunction (81). The association of migraine with ischaemic heart disease also points to systemic vascular dysfunction. This association has been investigated in a large number of studies, among which are the Women's Health Study and Physician's Health Study (93, 96). In the first study (mean follow-up 10 years) women aged 45 years or older with active MA, had an increased risk for major cardiovascular disease (CVD) after controlling for traditional cardiovascular risk factors [HR 2.15 (95% CI 1.58–2.92)]. CVD was defined as first instance non-fatal ischaemic stroke, non-fatal myocardial infarction, or death due to ischaemic CVD. In addition, in women with MA, after controlling for traditional cardiovascular risk factors, an increased risk was found for myocardial infarction, angina, coronary revascularization and for ischaemic CVD death. These risks were not increased for women with MO compared with controls (96). A study in an older cohort confirmed that in women, MA, but not MO, may be associated with an increased risk of coronary heart disease (CHD) (103). In the Physician's Health Study, which did not discriminate MO from MA, an increased risk for major CVD was found [adjusted HR 1.24 (95% CI 1.06–1.46)] for male migraineurs, which was driven by the increased risk for myocardial infarction [adjusted HR 1.42 (95% CI1.15–1.77)] (93).

Mechanisms for the association of migraine with stroke and ischaemic heart disease

Several mechanisms have been proposed that may underlie the increased risk of ischaemic stroke and ischaemic heart disease in migraine patients (8). One candidate is CSD, the underlying mechanism of the migraine aura, which is associated with decrease in cerebral blood flow (28–31). CSD (indirectly) alters blood–brain barrier permeability, which might lead to exacerbation of local cellular injury caused by ischaemia. Together with factors predisposing to co-agulopathy and release of local vasoactive neuropeptides, this could result in further changes in cerebral haemodynamics, arterial thrombosis and infarction (99). Other mechanisms that might predispose migraine patients to ischaemic cardiovascular events include endothelial dysfunction, migraine-specific medication, PFO or shared genetic factors. Alternatively, the increased risk may be mediated by an unfavourable cardiovascular risk profile. In the population-based GEM study, migraineurs were more likely to smoke and to report a parental history of early myocardial infarction. Migraineurs with aura were more likely to have an unfavourable cholesterol profile, have elevated blood pressure, and report a history of early onset coronary heart disease (CHD) or stroke. The odds of having an elevated Framingham risk score for CHD were approximately doubled for the migraineurs with aura. Thus, migraineurs, particularly with aura, have a higher cardiovascular risk profile than individuals without migraine (101). In the Women's Health Study this higher cardiovascular risk profile was found for women with MA and myocardial infarction, implying that the association with CHD is, at least in part, mediated by atherosclerosis (104). For angina this was not entirely the case, with both more risk of angina in the low risk and high risk category. The authors propose two mechanisms for angina, one mediated by atherosclerosis and another one that alters the vasculature independent of atherosclerosis.

Interestingly, in the Women's Health Study the association between ischaemic stroke and MA in women appeared to be significant only in women with a low (≤1%) Framingham risk score for coronary heart disease (CHD), which is particularly mediated by low age and low total cholesterol concentrations (104). As the Framingham risk score for CHD can be used as a proxy for vascular risk status, this would imply that the biological mechanism for the MA-ischaemic stroke association is not mediated by atherosclerosis, but involves alterations in the cerebral microvasculature. This hypothesis is supported by the data of the CAMERA study, where no association was seen between posterior circulation (PC) territory infarct-like lesions and types of supratentorial brain changes, such as deep white matter lesions or periventricular white matter lesions (99). Furthermore, there were no large differences in cardiovascular risk factors in those with and without PC territory infarct-like lesions. These two factors suggest that the lesions are not atherosclerotic in origin but reflect small vessel disease. An impairment in the adaptive cerebral haemodynamic mechanisms in the posterior circulation in migraine patients with aura might be part of the underlying mechanisms between migraine and brain infarcts (99).

In the Women's Health Study the relation between biomarkers of CVD and migraine was also analysed (i.e. lipids and a panel of inflammatory biomarkers: C-reactive protein (CRP), fibrinogen, intercellular adhesion molecule-1, homocysteine and creatinine). Overall, the data suggest that these biomarkers are an unlikely explanation for the observed association between MA and CVD in this cohort (105). Migraine, and specifically MA, has been associated with the T allele of the C677T polymorphism in the methylenetetrahydrofolate reductase gene (78, 106), which is also associated with moderately increased levels of homocysteine, a risk factor for CVD (107). Remarkably, in the Women's Health Study no association was found for migraine overall or for MA with increased levels of homocysteine (105), whereas in the GEM study homocysteine levels were increased only in the subgroup of male MA T/T homozygote carriers of the polymorphism (78).

Arterial dissection

Several studies found an increased migraine prevalence in patients with spontaneous cervical artery dissection (CAD) compared with controls with non-CAD ischaemic stroke, controls without vascular history or both (9, 108–111). CAD is an arteriopathy related to extracellular matrix abnormalities. This suggests that similar vessel wall defects form a risk factor for development of migraine. The association has also been put forward to explain (part of) the relation between migraine and ischaemic stroke. However, in two recent studies (9, 109), the association was merely significant for MO, but not for MA, while the risk of ischaemic stroke is only increased in patients with MA. Whether the CAD migraine association is true and indeed specific for MO remains to be confirmed in larger cohorts, as studies performed thus far had small sample sizes, were retrospective and had several limitations, including selection bias, lack of blind assessment and correction for confounders. Interestingly, in a family with familial aortic dissection carrying the R460H mutation in the transforming growth factor β receptor 2 (TGFβR2) gene, migrainous headache was reported in no less than 10 of 14 mutation carriers (112).

Arteriovenous malformations and angiomas

Occipital arteriovenous malformations have been reported to cause migraine with aura (11, 12). In addition, migraine can develop after bleeding from cavernous angiomas in the brain stem (10, 113). These observations once more suggest that vascular events are able to initiate neuronal deregulation, leading to migraine.

Reversible cerebral vasoconstriction syndromes

Reversible cerebral vasoconstriction syndromes (RCVS) is a descriptive term for a group of conditions, all characterized by multifocal areas of constriction involving the cerebral arteries that resolve within days to weeks (13). RCVS patients present with acute onset headache (‘thunderclap headache’) with or without neurological signs and symptoms due to ischaemia distal to severe vasoconstriction. Diagnostic evaluation to exclude other causes, such as subarachnoid haemorrhage, primary angiitis of the central nervous system and arterial dissection, is critical. Some 25% of RCVS patients have a history of migraine and the syndrome has been termed migrainous vasospasms or ‘crash migraine’ (13). Indeed parallels between migraine and RCVS exist, such as the throbbing headache and associated features (nausea, vomiting, photophobia), the topography of migraine-associated and RCVS-associated stroke (posterior circulation) and the serotonergic mechanisms that have been implicated in both disorders (114). However, important differences between migraine and RVCS exist as well. Migraine is not accompanied by cerebral angiography abnormalities. Also, migraine patients can distinguish the presenting acute headache in RCVS clearly from previous migraine attacks, and aura symptoms or premonitory symptoms are absent in RCVS. Diagnosis and management of RVCS in migraine patients is similar to other conditions. Simple observation may be justified. Because of the moderate risk for ischaemic stroke, treatment options such as calcium channel blockers (Verapamil and Nimodipine) or high dose glucocorticoids have been suggested; however, these options are based on observational data only (13).

CONCLUSION

A vascular component in the aetiology of migraine is supported by the association between migraine and genetic and acquired vasculopathies. In genetic small vessel diseases such as CADASIL and RVCL, vascular changes, including endothelial dysfunction, may directly or via neuronal pathways increase the risk for migraine. Alternatively, DNA variants in the susceptibility genes may directly form a risk factor for both migraine and the syndrome itself. The relation between migraine and increased risk of ischaemic stroke and cardiovascular disease may be explained by CSD, associated with decrease in cerebral blood flow and prothrombotic mechanisms. Other mechanisms include endothelial dysfunction, migraine-specific medication, PFO or shared genetic factors. For genetic as well as acquired vasculopathies research into endothelial function and shared genetic factors will be important in order to unravel the pathophysiological mechanisms that can explain their comorbidity with migraine.