|Home | About | Journals | Submit | Contact Us | Français|
With great interest we read the article by Koppen et al.1 In contrast to the authors’ opinion, we believe that there is substantial evidence for a strong relationship between a right-toleft shunt (RLS) and migraine with aura (MA).
Our belief was initially based on the empirical experience with hereditary haemorrhagic teleangiectasia (HHT) patients, who frequently exhibit a permanent RLS through a pulmonary arteriovenous malformation (PAVM). Clinicians noticed that patients with HHT frequently complained of migraine headaches, and a PAVM was found remarkably often in this group. More importantly, there is redundant scientific evidence for a strong relation between RLS and MA.
Firstly, the association appears to be bidirectional with a high prevalence of RLS in MA patients, and otherwise a high prevalence of MA in patients with a symptomatic RLS.2
Secondly, the association exists independently of shunt anatomy. We and others have not only reported a high prevalence of MA in the presence of an RLS through a symptomatic patent foramen ovale (PFO),3 but also in the presence of an atrial septal defect, occasionally with an RLS,4 or a permanent RLS through a PAVM.2
Thirdly, MA is related to the size of the RLS.5 Migraine is more prevalent in patients with an RLS at rest compared with those with a provocable RLS.6 Likewise, we have shown that a large intrapulmonary RLS increases the risk for MA more than sevenfold.7 To explain the relationship between RLS and migraine, it has been proposed that an RLS permits micro-thrombi to bypass the lung filter, enter the cerebral circulation and trigger a migraine attack. This ‘thrombi-hypothesis’ might be the link between migraine and ischaemic stroke. This is supported by the finding that MA patients have a higher risk for both clinical and subclinical brain infarctions, 8,9 and as reported by the authors, a good response on aspirin or clopidogrel. Furthermore, paradoxical embolism is responsible for the high incidence of stroke in patients with PAVM.
Koppen et al. correctly pointed out that the initial retrospective studies reporting a positive effect of PFO closure on MA occurrence have methodological restrictions. However, also in several prospective observational PFO closure studies we and others have reported a decrease in MA occurrence. 3,5 Moreover, a decrease in the occurrence of MA has been found after PAVM embolisation.2 Interestingly, these patients were not treated with aspirin or clopidogrel. Therefore, this could not have influenced the occurrence of migraine. As Koppen et al. describe, the randomised MIST trial did not show a significant reduction in migraine six months after PFO closure.10 However, besides the limitations that Koppen et al. indicate, the lack of benefit in this study might be influenced by a different population compared with other studies. The patients in the observational studies suffered a symptomatic RLS (history of paradoxical embolic event). In the MIST trial these patients were excluded. Other limitations are the inclusion of a high frequency of patients without a PFO during invasive evaluation, a higher than usual number of complications, and an unclear number of residual shunts after closure. Therefore, in our opinion the MIST study failed to refute the concordant findings of the observational studies.
In conclusion, in contrast to Koppen et al. we are convinced that there is a strong association between the presence and size of an RLS and MA, independent of its origin. We do agree with the authors that, at present, PFO closure is not an option for migraine prophylaxis. More randomised studies with appropriate inclusion criteria11 and clinical endpoints should be performed to determine the effect of RLS closure on migraine.
We read the letter by Luermans with interest. In our editorial on patent foramen ovale (PFO) and migraine,1 we mainly disputed the causal relation between PFO and migraine, as there is no convincing evidence (level 1 or 2) for any positive effect following the closure of PFO on migraine. Furthermore, also the pathophysiological grounds for a causal relation remain highly speculative. This makes us conclude that PFO closure is not an option for migraine prophylaxis.
We did report that several studies found an association between PFO and migraine with aura; however, we pointed out that these studies were all clinical-based with the known methodological limitations.1 Recently the first populationbased study using transthoracic echocardiography (TTE) was published, which could not demonstrate an association between PFO and migraine with aura (MA).2 In our recently finished population- based study investigating 286 persons with transcranial Doppler with air-contrast (TCD-c), a technique which is highly sensitive for detecting right-to-left shunts, we could only demonstrate a modestly increased prevalence of right-to-left shunt (RLS) in MA, being only half-fold increased compared with controls.3 This association was much weaker than the two- or threefold increased prevalence reported in most clinical-based studies. Furthermore, we could not demonstrate an increased prevalence for spontaneous RLS in MA compared with controls.3 This strongly suggests that the clinical-based studies indeed partly suffered from methodological short comings as referral, selection and investigator bias.
We do agree with Luermans et al. that different causes of RLS are associated with MA, but whether PFO, atrial septal defect (ASD), or pulmonary arteriovenous malformation (PAVM) were topics of investigation, the methodological shortcomings are just the same. Furthermore also non-RLS cardiovascular diseases such as Marfan syndrome or myocardial infarction have been associated with migraine with aura,4 which favours the argument that migraine is just more common in different cardiovascular diseases independent of the presence of an RLS.
It is interesting to read that after PAVM embolisation patients complained of less migraine, and no aspirin or clopidrogel use could have attributed to this observed positive effect. However, all other major methodological limitations still do account for this study, for example the uncontrolled design.
In summary we conclude that we agree with the fact there is an association between MA and RLS, but this association is probably not as strong as reported by the previous clinical-based studies. Secondly, but more important, there is still no convincing evidence for a causal relation between RLS and MA. Only theoretically speaking, the opposite is even possible, a PFO being the result of frequent vomiting in MA and thereby increasing right-side heart pressure which opens the formerly closed PFO. More likely the association is genetic as suggested by family studies.5
In our opinion the decision to close an RLS, whether it is caused by a PFO, ASD, or PAVM, as migraine prophylaxis, should be based on solid evidence derived from randomised controlled trials, which until this moment is lacking for all these different causes of RLS.