Abstract
Migraine is a risk factor for cerebral infarction in young women. The nature of the connection between these diseases remains however essentially unknown. Abnormalities of haemostasis leading to an increased thrombotic risk would provide a logical link. Platelets, antiphospholipid antibodies and more recently congenital thrombophilia have thus successively been implicated. The different studies concerning these topics have been reviewed. Because of the conflicting results obtained and because of the numerous methodological shortcomings of many of these studies, no definite conclusion can be reached. It is possible that these 3 factors play a role in the ischemic risk of migraine, but it is as likely or even more likely that other factors (inside or outside the hemostatic system) play a more important role. Further studies are thus deeply needed.
Introduction
Migraine has long been thought to be a potential cause for cerebral infarction (1) (so called ‘migrainous infarction’) and recently, it has been shown to be a risk factor for cerebral infarction in young women (2–6). There is thus an undisputable connection between migraine and cerebral ischaemia but the nature of this connection remains essentially unknown. Although mitochondrial dysfunctions (7) and cardiac abnormalities, such as mitral valve prolapse (8) and more recently patent foramen ovale (9), have been thought to play a role, most studies have been devoted to the detection of markers of a prothrombotic tendency in migrainous subjects. We review these studies, particularly as regards platelets, antiphospholipid antibodies and congenital thrombophilia, starting with an emphasis on the numerous methodological shortcomings which exist in the majority of these studies.
Methodological shortcomings
As in all studies devoted to migraine, there is a major problem in the definition of patients since there is no biological marker or definite diagnostic test for migraine. The situation has greatly improved with the publication of the IHS classification and diagnostic criteria (10), but since then some studies have gone on using their own criteria and not those of the IHS. Within migraine itself, there has often been no proper definition of migraine with aura (MA) and migraine without aura (MwA), and yet the increased risk of cerebral ischaemia is mainly observed in MA (3). Furthermore, there has been no effort to separate within patients with MA those who have only attacks of MA and those who have a variable mixture of attacks of MA and MwA. The IHS classification itself does not suggest such a distinction, which might be important for the ischaemic risk.
The timing of studies is a crucial factor in migraine. Some studies have been devoted to attacks, others to the interval between the attacks. This also raises the difficult questions of how long after an attack abnormalities may persist and of the potential interactions of treatments with haemostasis. The frequency of attacks may also play a role and has usually not been taken into account.
Finally, as regards migraineurs, there is always a major selection bias, hardly avoidable in biological studies, due to the fact that migrainous subjects coming to hospitals or to headache clinics are highly selective and not representative of the whole migrainous population. Furthermore, in many of the studies, the number of patients included was very small.
Not only is the selection of migrainous patients a difficult issue, but that of controls is an even greater problem. In many studies, there were no controls at all and in those who had a control group, there was often no effort to take into account the usual vascular risk factors, such as smoking, hypertension, etc.
Together with these major difficulties pertaining to the studied subjects, we will see there that are a number of technical difficulties related to the methods of measurement of the various haemostatic parameters.
Migraine and platelets
Because of their high serotonin content, platelets have been the subject of a long-standing interest in migraine. They have been considered to be a good model for neurones (11) and migraine was even once suggested to be a platelet disorder (12). A number of biochemical abnormalities of serotonin metabolism have been reported in migraine, but they are not directly connected to the haemostatic properties of platelets to which this review is restricted. Platelets play a crucial role in assuring a normal primary haemostasis after their activation by physiological stimuli. Following a lesion of the endothelium, they adhere to the site of injury and they undergo different reactions consisting in shape change, release of substances from alpha granules and dense bodies, synthesis of pro-aggregant prostaglandins such as thromboxane A2 and formation of platelet aggregates. Platelets can aggregate spontaneously, but they also do so in response to numerous agonists such as thrombin, collagen, adrenaline, ADP, platelet-activating factor (PAF) and serotonin. Platelet aggregation has been by far the most studied aspect of platelet function in migraine and particularly according to the technique with plasma rich in platelets (PRP). It was found to be increased in a majority of studies during as well as between the attacks. This applies to ADP (13–16), 5 HT (17–20), PAF (21) and epinephrine (20, 22, 23) induced platelet aggregation, although conflicting results exist with aggregation unchanged (15, 16, 24–27) or even diminished (22). Collagen-induced aggregation was found to be similar to control subjects (16, 25, 26). Thrombin-induced aggregation was found to be diminished during attacks and increased between the attacks (28). Spontaneous platelet aggregation was also found to be increased during and between attacks (13).
The reasons for this possible increased platelet aggregation in migraine are unclear. Studying migraine patients between attacks, Kozubski et al. found that the number and affinity of fibrinogen receptors on their platelets were significantly increased (29). Palowska et al. reported that GPIIb was significantly increased on the platelet membrane of migraineur patients during and between the attacks (30). Lastly, changes in the viscosity of the platelet membranes due to difference in lipid composition has been suggested in the participation of the functional abnormalities (31).
The interpretation of these conflicting results is hampered by the methodological shortcomings already mentioned, particularly since most of these studies were performed before the IHS classification. Furthermore, there are technical problems with the widely used optical aggregometer, which is based on PRP and has serious disadvantages. White and red blood cells are eliminated by centrifugation, while in vivo they exert a modulator role on platelet function; separation of platelets from other blood cells by centrifugation can alter their size, density and behaviour; centrifugation can also eliminate the heaviest platelets, which are also the youngest and the most active. Besides these technical difficulties, different methodologies were used in these studies: different concentrations of inducers, different timings of blood sampling in relation to the occurrence of the attacks or of the menstrual cycle and different blood sample procedures. To overcome these difficulties, other studies have been performed of platelet aggregation on native blood. Using an aggregometer with impedance technique, platelet response to collagen of migraine patients appeared to be normal (32) or even significantly reduced (33). Another study using a haemostatometer (in vitro technique with non-anticoagulated native blood) also found no difference between platelets of migraineurs outside the attacks and platelets of control subjects (34).
It is impossible to draw any firm conclusion regarding platelet aggregation in migraine. Results are too conflicting, studies too different and often methodologically insufficient. There might be a tendency towards increased platelet aggregation but it seems far too inconsistent to explain, in itself, an increased thrombotic risk. Furthermore, it is now well established how variable platelet aggregation is, varying for instance with stress or with smoking one cigarette (35).
Other tests reflecting platelet activation have been studied in migraine. Increased plasma levels of β-thromboglobulin and plasma factor 4 (PF-4) have been found (14, 36, 37). These proteins are normally contained in the alpha-granules and their presence in the plasma indicates that platelets have been activated and have undergone a release reaction. However, in vivo, PF-4 is normally trapped by the vascular endothelium and should not be increased. Moreover, there were no modifications of plasma levels of stable metabolites of thromboxane B2 and prostacyclin (37), as would have been expected in the presence of platelet activation.
If platelet aggregation plays a major role in migraine pathogenesis, it would be expected that drugs which inhibit platelet activation would be effective in both headache attacks and migraine prevention (22). According to the animal model of Moskowitz, neurogenic inflammation has been proposed as a possible mechanism for migraine with a role of platelet activation leading to among other things the release of thromboxane A2. Concerning the attacks, the TXA2 liberation could entirely explain the action of aspirin and non-steroidal anti-inflammatory drugs (38). Furthermore, dazoxiben, a highly specific thromboxane synthetase inhibitor, has been shown to be effective in migraine attack with a significantly greater reduction of TXA2 observed in patients who reported good symptomatic relief (39). On the other hand, the actions of ergotamine, dihydroergotamine (DHE) and triptans, very well-established active drugs in aborting migraine attacks, have been shown to bind with high affinity to 5-HT1 receptors (40). This suggests that their action is mediated by activation of 5-HT receptors present on sensory fibres innervating blood vessels in dura-mater and not at all with platelets. Concerning migraine prevention, there are also some data suggesting a beneficial preventive effect of aspirin alone (41, 42) or combined with dipyridamole (43) or DHE (44). In the same way, methysergide (45) also has some inhibitory effects on platelet aggregation. On the other hand, β-adrenergic blockers, which belong to a group of drugs of major importance in migraine prophylaxis, have different effects on platelets behaviour of migraine patients. Indeed, in equivalent doses, propanolol, a non selective β-blocker, increases aggregation and platelet secretion, whereas metoprololol, a selective β1-blocker, inhibits aggregation. Administration of both these drugs gives elevated thromboxane B2 levels and stable metabolites of TXA2, leading the authors to the conclusion that migraine was not a platelet disorder (46).
In summary, it is again impossible to draw any conclusion from drug trials about the platelet–migraine connection. Trials of aspirin are methodologically insufficient and, even if aspirin is shown to be effective, it has so many actions that it is impossible to know which one is implicated in such a multifactorial condition as migraine. As regards other very effective classes of drugs such as triptans in the acute treatment and betablockers in prophylactic treatment, there is no evidence at all that their anti-migrainous action has anything to do with platelets.
On the basis of what we have seen so far, the so often emphasized link between platelets and migraine seems rather tenuous: a suspicion of increased platelet activity and of a beneficial effect of antiplatelet drugs, mainly those which concerned TXA2. Yet, there are some intriguing observations regarding essential thrombocytemia that suggest an increased frequency of migrainous auras (or of transient ischaemic attacks presenting as migrainous auras) disappearing with antiplatelet drugs, such as ticlopidine, without a concomitant decrease in the number of platelets (47, 48). Essential thrombocytaemia might thus induce both cerebral ischaemia and migrainous aura (symptomatic migraine) and might be worth looking for in cases of transient focal neurological events in which it is impossible to differentiate migrainous auras and TIAs.
In conclusion, despite numerous studies, the reality of platelet hyperaggregability during migraine can not be considered as evidence based. Whereas the platelet theory provides an attractive link between migraine and cerebral ischaemia, the present data do not provide strong support for a primary role of platelets in migraine, even if the extreme complexity of platelet behaviour and of its assessment can not be ignored. Further studies are needed, exploring other aspects of platelet function and taking into account all the methodological difficulties that we have stressed.
Antiphospholipid antibodies and migraine
Antiphospholipid antibodies (aPL) are a group of circulating, usually polyclonal, serum immunoglobulins associated with thrombosis. They are of different isotypes, binding in vitro with negatively charged or neutral phospholipids. They include anticardiolipin antibodies (aCL), lupus anticoagulant (LA) and false positive serologic test for syphilis. They can develop either in patients with various diseases, including underlying autoimmune diseases such as systemic lupus erythematosus (SLE) defining the secondary aPL syndrome (49), or in patients without such diseases defining the primary aPL syndrome in association with arterial and/or venous thrombosis (50–52). Many neurological diseases have been described in patients with aPL (53, 54). They include cerebral infarction, transient ischaemic attacks, optic nerve infarction, vascular dementia, cerebral venous thrombosis, chorea, myelopathy and migraine. The relationships between aPL and migraine have been studied both ways: frequency of migraine in aPL syndromes and frequency of aPL in migraine.
The first association between migraine and aPL was reported in 1978 by Brandt and Lessel on the basis of two patients out of 11 with SLE and migraine (headache and/or visual disturbances evoking visual aura) who had LA (55). Since migraine seems to be more common in patients with SLE, some investigators have looked for a possible association with aPL. Asherson et al. reported 35 patients with cerebrovascular disease and aPL and indicated that vascular headaches were frequent in patients with aPL and SLE, but they gave no precise percentage (56). In a prospective study of 103 patients with SLE, Montalban et al. found no association between the presence of aCL and migraine in patients with SLE (57). On the other hand, in retrospective studies of selected populations of patients evaluated for aPL after a cerebral ischaemia or transient focal neurological event, the frequency of migraine in patients with positive aPL appeared to be twice that of the general population (54, 58–62). These much-debated results might once more be due to the difficulty in diagnosing migraine and in differentiating TIAs from migrainous auras, particularly atypical auras. Yet, it would be important to know whether migraine is indeed a prominent feature of primary or secondary aPL syndrome or not because it could help to identify patients with a high risk of ischaemic stroke.
The frequency of aPLs in patients with migraine has been evaluated in a number of studies summarized in Table 1 (63–68). In four of these studies, aPLs were not associated with migraine (63–66). It should, however, be noted that these studies usually included small numbers of patients, mixing up migraine with or without aura and that they did not did not always use a control group. Furthermore they excluded patients with any history of thrombosis, manifestation of autoimmune diseases or episode of thrombocytopenia, thus selecting a population that by definition would not have an aPL syndrome. By contrast, Robbins et al. looked for the presence of aPLs in 68 patients with migraine and 22 controls without clinical or biological exclusion criteria; they found in the migraine group 16 patients with IgG aCL (admittedly with a low titre in 13), but only one in the control group (67). In a recent study of aCL in consecutive patients with migraine without aura or with recent transient focal neurological events, eventually compatible with migrainous auras, the frequency of aCL did not significantly differ from that of controls. Nevertheless, the authors stress that despite the large number of patients with migraine included, the statistical power was not sufficient to dismiss an association between aCL and migraine (68).
Frequency of antiphospholid antibodies in patients with migraine
aPL, antiphospholipid antibodies; aCL, anticardiolipin antibodies; LA, lupus anticoagulant; MA, migraine with aura; MwA, migraine without aura.
In summary, the role of aPLs in migraine remains to be elucidated. There is no good evidence that the frequency of aPLs is increased in migraine. By contrast, in primary as well as in secondary aPL syndromes, there seems to be an increased frequency of transient focal neurological events, some of them having features of migrainous auras, others typical of TIAs.
Genetic prothrombotic abnormalities and migraine
Mutations in genes encoding proteins which are involved in thrombus formation play a major role in the predisposition to thrombosis. Most of these genetic variations play a major role in venous thrombosis and little or no role in arterial thrombosis: factor V Leiden (FVL) and factor II 20210 A mutations, antithrombin, protein C and protein S deficiencies. More recently, new inherited defects have been recognized and shown to play a role in arterial thrombosis: platelet GPIIIa and GPIbα polymorphisms (HPA-1 and HPA-2, respectively), factor VII (FVII), fibrinogen and homocysteine.
Few studies have so far investigated the prevalence of these prothrombotic genetic risk factors in patients with migraine. A very high frequency of MA (6 of 9; 67%) was found in Finnish patients carrying the mutation of FVL and suffering an ischaemic stroke. This contrasted with 60 of 227 (26%) in patients with ischaemic stroke but no mutation (69). By contrast, in a case-control study including 106 patients with migraine (49 with aura, 57 without aura), the frequency of several prothrombotic risk factors (FVL, factor II 20210 A mutations, FVII, HPA-1 et HPA-2 polymorphisms) (70) did not significantly differ in patients with migraine and in controls. A particularly high frequency of FVL was found in migraine with aura (6, 1%), but this association did not reach statistical significance (2, 8%; P = 0.323, OR 2.24, CI 0.34–14.58). A rather similar study was recently undertaken by D'Amico et al. who investigated the frequency of genetic abnormalities of the protein C system (i.e. protein C and protein S deficiencies, activated protein C resistance (aPC-R) and FVL) in 83 patients with migraine with aura, 31 patients with ischaemic stroke and 124 healthy controls, all under 45 years of age (71). They found that the frequency of aPC-R due to FVL and of protein S deficiency was significantly increased in migraine and ischaemic stroke (respectively 12% and 16%; 4.87% and 6.4%) compared with that of the controls (3.2%; 0%). These results were, however, not confirmed by other studies which found no increased aPC-R or FVL in various groups of migrainous subjects, such as children or young subjects aged 8–28 years (72) or migrainous subjects suffering a cerebral infarction: only one heterozygous carrier of FVL mutation was found in a series of 20 ‘migrainous infarcts’ (73).
In summary, once again, no definite conclusion can be drawn from these studies. Although usually more recent and of better quality than the platelet studies, they still face methodological studies involving particularly the definitions of the different groups of subjects: migraine, ischaemic stroke and migrainous infarctions. In contrast to the conclusion of some of these studies (70, 71) it is far too early to conclude that congenital thrombophilia (particularly FVL) plays a role in the migraine–ischaemic stroke connection. It should be remembered that these factors are essentially venous risk factors and that no increase in the risk of venous thromboembolic disease has been demonstrated in migraine. Haemostatic risk factors for arterial thrombosis would certainly be more interesting to study.
Conclusion
Since there is an increased risk of ichaemic stroke in young women with migraine and since ischaemic strokes are most often on a thrombo-embolic basis, abnormalities of haemostasis leading to an increased thrombotic risk would provide a logical link between migraine and cerebral infarction. Platelet activation, aPLs and more recently congenital thrombophilia have thus successively been implicated. However, because of the conflicting results obtained so far and because of the numerous methodological shortcomings of many of these studies, no definite conclusion can be reached. It is possible that these three factors play a role in the ischaemic risk of migraine, but it is as likely or even more likely that other factors (inside or outside the haemostatic system) play a more important role. Further studies are thus deeply needed. In the meantime, there is no justification for performing a systematic study of haemostatic parameters in subjects with typical migraine, even with aura, unless there is a personal or familial history of venous thromboembolism or of cerebral ischaemic events. Furthermore, it seems justified to look for essential thrombocytemia and aPL in patients with transient focal neurological deficits among which it is difficult to differentiate migrainous auras and TIAs. Because of this difficulty, any suspicion of ‘true’ TIA should be investigated with the usual vascular, cardiac and haematological evaluations.