There are at least 12 reports from independent sources that describe cessation or decrease in the frequency of migraines after transcatheter closure of PFO. Some of these patients had chronic debilitating headaches that were refractory to multiple prophylactic pain medications [9–19]. Of patients who had migraine with aura, an average 60 % reported complete cessation of their headaches and aura. Another 15 % reported that although they still had migraines, there was at least a 50 % reduction in the frequency of headaches as measured by the number of headache days per month. Thus, a significant benefit from PFO closure is reported in approximately 75 % of individuals who have migraine with aura.

Our understanding of migraine headaches has changed within the last 30 years. Until the 1990s, it was thought that migraine was a “vascular headache.” The concept was that the headache was induced by intense cerebral arterial spasm resulting in relative ischemia in the brain; this was thought to account for the visual aura or other transient neurological deficits. The vasospasm was then followed by vasodilation of the cerebral vessels. The throbbing component to migraine headaches was thought to be from pulsation of blood through the dilated arteries. There was, however, a lack of evidence from any human or animal studies to support this hypothesis.

The concept of migraine headache changed significantly with newer methodologies that could image brain perfusion and metabolism in humans. At least 18 studies with positron emission tomography and functional magnetic resonance imaging (MRI) demonstrate the opposite of the original theory of migraine [20–25]. Initially, there is vasodilation of the cerebral vessels, then constriction. This is associated with a wave of neuronal depression that spreads over the cerebral cortex, corresponding to the transient deficit of neurologic function attributed to the aura in migraine. This cortical spreading depolarization initiates in the optical cortex and proceeds at approximately 3–5 mm per minute up to the motor and sensory cortices. In addition, there is a complex interaction with other brain centers as well as stimulation of nociceptive pain fibers.

Cortical spreading depression involves a complex interplay and activation of neurons, glia and the vasculature. Glial membrane depolarization is implicated as the primary driver of the electrical changes in migraine, with astrocytes causing widespread intercellular communication through propagated increases in intracellular calcium. It is also hypothesized that astrocytes may also contribute to the primary wave of cortical spreading depression through direct stimulation. Other than the effects of the primary wave of depression, a second phase involving neurophysiological and vascular changes may persist for an hour after the initial wave of cortical spreading depression. Imaging with magnetic resonance angiography and transcranial Doppler support the hypothesis that migraine is associated with neurovascular uncoupling where the affected hemisphere may either have vasodilation and hyperperfusion, or vasoconstriction and hypoperfusion, and sometimes both in the same individual. Such a neurovascular uncoupling may explain the mechanism through which nociceptive messengers are produced. Electrophysiological studies have demonstrated that trigeminal nociceptive neuronal activation can occur with a significant delay after the initial event of cortical spreading depression; this suggests that pain signals may be generated in the “second phase” of the cortical spreading depression, rather than with the initial primary wave [26].

The newer concept of migraine has also been confirmed in elegant experiments using a mouse transgenic model of migraine [27, 28]. A genetically linked subset of migraine has been described where multiple family members have severe migraine associated with transient hemi-paresis. The neurologic deficit is only temporary, without evidence of stroke on MRI. The gene for this familial condition has been cloned and inserted into a mouse DNA. Phenotypically, these mice develop transient hemiplegia, and visual imaging of their brain reveals cortical spreading depression. Therefore, migraine is now thought to be an interaction between the cerebral vasculature and neurogenic centers in the brain that are stimulated to induce a migraine headache. The potential role of PFO in this mechanism is that the PFO may be the route for chemical triggers to reach the appropriate neural receptors in the brain. The profound cellular depolarization that spreads over the cerebral cortex produces the transient neurologic deficit perceived as aura [29]. In the majority of cases, this is a visual dysfunction with zigzag white lines, star bursts of light or scintillating scotoma. Other neurologic manifestations include paresthesias, motor weakness, diminished thought processing, or global amnesia.

The transient neurologic deficits that occur during migraine with aura may be difficult to distinguish from a transient ischemic attack (TIA). Both migraine with aura and TIA by definition have a transient neurologic deficit without demonstrating any abnormalities on MRI brain imaging. In addition, the transient neurologic deficits caused by migraine can occur without any headache, making it more difficult to attribute this to a migraine. This has important implications for clinical trials of cryptogenic stroke if TIA is used as an end point and it is assumed that all transient neurologic deficits are due to embolic phenomena. In the observational studies of migraineurs, it has been reported that PFO closure improves both migraine headache and the transient neurologic deficits associated with it [5, 17]. In the clinical trials for stroke, if PFO closure is effective for preventing migraine, the group on medical therapy presumably could have more episodes of migraine with transient neurologic deficits. In other words, it would appear as if PFO closure is preventing recurrence of TIAs when in fact it is preventing transient neurologic deficits due to migraine. While from the patients’ point of view, they may be pleased as long as the symptoms are diminished, from a mechanistic perspective, it would be helpful to be able to differentiate between an embolic TIA due to a small thrombus versus a chemically induced migraine with transient neurologic deficit.

The underlying concept connecting migraine and PFO is that migraine may be triggered by chemicals that reach the brain through a transient right-to-left shunt. These chemicals normally would be metabolized in the pulmonary capillary circulation by passage through the lungs. By bypassing the pulmonary circulation through a transient right-to-left shunt, these chemicals avoid metabolic alteration in the lungs and reach the arterial circulation in a concentration that stimulates receptors in susceptible individuals and results in the migraine phenomenon. Many observational studies report that migraine headache improves when the right-to-left shunt is closed suggesting that after closure there is some substance that is no longer getting to the brain. In addition to chemicals that could trigger migraine, such as serotonin, other potential culprits for inducing migraine might be platelet plugs or other particulate substances that could enter the cerebral circulation and act as the trigger of migraine headaches. It is suspected that once these chemicals enter the brain, they may stimulate a neural receptor that precipitates the migraine. This is consistent with other observations that some chemicals found in foods such as nuts, chocolate or alcohol can trigger migraines in susceptible people [30–34].

Another possible mechanism for PFO related migraine may be transient hypoxemia. There are some people who state that their migraine is induced by exercise or straining. Presumably, the transient increase in right atrial pressure opens the PFO and permits bursts of venous blood to enter the arterial circulation. Dr. Mark Reisman described a person on the catheterization table who had an intense migraine as soon as his PFO was crossed with a guidewire. This person continued to complain of migraine with vivid aura until the end of the PFO closure procedure. Once the PFO was closed, the migraine disappeared immediately. These observations suggest that transient changes in oxygen delivery to the brain or pH content of venous blood, may play a role in induction of migraine for susceptible individuals. This is also consistent with reports from some patients that supplemental oxygen prevents the onset of migraine or aborts it once the migraine has started.

Is there something unique about a pathway through the atrial septum with the presence of a PFO, or are migraines associated with any mixture of venous blood directly with the arterial circulation? A natural experiment to test this hypothesis would be in adults with congenital heart disease. While these people have a variety of abnormal connections in the circulation, they can generally be categorized as having a right-to-left shunt, a left-to-right shunt, or no shunt. To test the hypothesis, adults were chosen because they would have lived long enough to have the possibility of experiencing migraine headache, which usually begins in the teenage years or early 20s. UCLA has a specialized Adult Congenital Heart Disease Clinic that has followed over 3,000 patients. Of these, 800 people were available and contacted by phone. A history of migraine headaches was obtained using the MIDAS questionnaire [35]. The frequency of migraine in this population was compared with a control population that was matched by gender.

The frequency of migraine was 11 % in the control population, which is consistent with previously published epidemiologic studies [36]. However, in people with a diagnosis of right-to-left shunt (such as Tetralogy of Fallot, pulmonary or tricuspid atresia, etc.) the prevalence of migraine headache was 52 % (p < 0.001). In those people with a primary diagnosis of left-to-right shunt (such as ASD, VSD, or PDA), the prevalence of migraine headache was 44 % (p < 0.001). In addition, people who had a history of migraines and had surgical correction of their congenital heart disease in their adulthood reported a dramatic reduction in the frequency of their migraine headaches as documented by the MIDAS questionnaire. This reduction in migraine frequency after closure of the shunt was similar to that observed in migraineurs with cryptogenic stroke who had their PFO closed percutaneously [16].

Unexpectedly, in people with a congenital heart lesion that is not normally associated with a shunt (such as bicuspid aortic valve, Marfan’s syndrome, aortic dilatation, etc.), the prevalence of migraine headaches was also elevated at 38 % (p < 0.001) [35]. These congenital heart disease patients without an obvious shunt by standard echo/Doppler (which does not include an echo contrast/bubble study) were found to have a higher incidence of right-to-left shunt demonstrated by TCD (38 %; p < 0.01 compared to controls) that may partially explain the high frequency of migraine headache [37].

There are also rare examples (1 % of our cases) of pulmonary arteriovenous malformation (AVM) in patients who do not have a PFO and present with cryptogenic stroke or migraine headache. As with PFO closure, the migraine may be reduced or eliminated when the pulmonary AVM is closed. Thus it appears that it is the presence of persistent right-to-left shunting or intermittent right-to-left shunting (in the case of predominant left-to-right shunt, e.g. with ASDs) that is associated with migraine headaches as distinguished from something unique about a PFO.

Although the observational studies have reported dramatic improvement or resolution of migraines in a large fraction of patients who have migraine and aura, there are still 25 % of patients who do not respond to PFO closure. Furthermore, not all patients with migraine and aura have a PFO. Many neurologists and cardiologists have questioned whether there is a subset of patients with migraine who may be more responsive to PFO closure. Peter Goadsby, Mark Reisman, Brian Whisenant, and others have attempted to define a subset of migraineurs who underwent PFO closure with abolition or dramatic reduction in the frequency of headaches. Their theory is that the common feature in these patients is their auras, which is usually a visual field defect; however, other transient neurologic deficits may also occur intermittently with or without the headache.

Migraineurs often have white matter lesions (WML) in the brain that can be detected by MRI. These abnormalities are 2–5 mm hyperintense signals that are usually secondary to axonal degeneration, gliosis and demyelination as a result of microvascular ischemia. WML can be detected anywhere along the white matter tracts of the cerebrum, cerebellum and the brainstem [38]. These hyperintensities, which appear as white spots on FLAIR or T2 sequences on MRI images, can often appear similar to lesions found in multiple sclerosis, vasculitis, and lacunar strokes.

In one study from France, 1,643 individuals older than 65 years had WML on MRI and were followed for 5 years [39]. The risk of developing a subsequent stroke in these patients was directly correlated with the volume of the WML and was 5× higher for the highest quartile of WML volume. However, the presence of WML was not predictive of other cardiovascular outcomes, indicating that the pathophysiology of WML is different from atherosclerosis.

The exact etiology of WML in migraineurs is currently unclear. While the lesions may represent a metabolic disturbance related to the physiology of migraine, they may also represent an ischemic insult due to embolic material. WML have also been observed in patients who have a PFO with no history of migraine headache. This observation may suggest that these hyperintensities are caused by chemicals or particles such as platelet plugs that bypass metabolism in the lungs and enter the cerebral circulation through a right-to-left shunt. In contrast to this theory is the observation that migraineurs without a PFO also have a high incidence of WML on MRI. Bosca et al. imaged 44 migraineurs with and without aura. Of the 44 included patients, 29 (66 %) had WML but only 7 (24 %) of patients with WML had a right-to-left shunt. Right-to-left shunting occurred at a similar frequency by TCD among migraineurs with WML compared to those without WML (both groups 50 %) [40]. While the study was underpowered, it suggested that there either may be multiple etiologies for WML in migraine or that right-to-left shunting is unrelated to WML. Del Sette et al. conducted a similar study with 80 patients and arrived at the same conclusion [41]. They believe that WML in migraineurs may only be a sign of age-related gliosis or scar formation, and that WML should not be used as an indication for PFO closure in patients with migraine.

An Italian group headed by Carlo Vigna conducted an observational study of 82 patients, all of whom had a PFO, severe migraine, and WML on MRI [42]. Patients in the study were offered PFO closure. Of 53 patients who had their PFO closed, the frequency of migraine reduced from 32 ± 9 in the 6 months before closure to 7 ± 7 in the 6 months after closure (p < 0.001). The 29 patients who chose to not have their PFO closed had no significant reduction in migraines, from 36 ± 13 to 30 ± 21. This study demonstrated that migraineurs with WML had a significant response to PFO closure. However, it is unclear whether this is because PFO is causally related to migraine, or whether the WML are a sign of greater responsiveness to PFO closure.

At UCLA’s Interventional Cardiology program, we attempted to determine if there is any association between WMLs and PFO by comparing the prevalence per decade of WMLs in subjects with a known right-to-left shunt versus a general population. Four hundred and sixty four patients who had a documented right-to-left shunt and underwent brain MRI were split into migraineur and non-migraineur groups, and were compared to a control population of 670 healthy individuals with brain MRIs (unpublished data). Across each decade of non-migraineurs with a right-to-left shunt, WMLs were not more prevalent compared to age matched controls. However, migraineurs with RLS had an increased prevalence of WMLs, which were statistically significant in 2 age groups: 30–39 years (21.05 % vs. 4.4 % p = 0.01) and 50–59 years (49.15 % vs. 21.2 % p = 0.0008).

There was a significant difference between migraineurs and non-migraineurs with respect to WML prevalence in the study group. Among 263 migraineurs with a right-to-left shunt, 78 (30 %) had WMLs. In comparison, in 212 non-migraineurs with right-to-left shunt, 45 (21 %) had evidence of WMLs (p = 0.037). The increased prevalence of WMLs in the control group correlates with increasing age. The presence of WMLs was observed as early as the third decade, with the peak in the sixth and seventh decades of life.

Our results suggest there is no greater incidence of WMLs in people with PFO without migraine compared to the general population. WMLs increase per decade in both groups. On the other hand, WMLs are more common in migraine with or without a right-to-left shunt (Fig. 10.1). Thus, it is something unusual about the physiology of migraine and not the right-to-left shunt that induces the WMLs.

A concern regarding the presence of WML in migraine is whether the lesions result in any long-term cognitive impairment. Although a number of studies have concluded conflicting results [43–58], two large well-controlled reports did not observe any cognitive impairment in migraineurs [59, 60].

Are migraine and PFO connected by some genetic predisposition? Peter Wilmshurst was the first to demonstrate an association between migraine headaches and septal defects, in families who had either an ASD or a PFO [61]. His group used contrast echocardiography to detect right-to-left shunts in 71 relatives of 20 probands who had either a large PFO or an ASD. The occurrence of atrial shunts was consistent with an autosomal dominant inheritance. When the proband had migraine with aura and an atrial shunt, 15 of the 21 (71 %) first-degree relatives with a right-to-left shunt also had migraine with aura, compared with 3 of 14 (21 %) without a significant shunt (p < 0.02).

It is possible that the atrial defect is an independent bystander that permits some chemical to pass through which triggers the migraine in these susceptible families. But the demonstration of both migraines and atrial defects persisting in several generations implies that there is a genetic component to both the susceptibility of migraine headaches as well as the development of cardiac septal defects. Either the same gene could be active in both the heart and brain, or separate genes are located closely on a section of a chromosome. It is anticipated that future research may identify the gene or genes, which could have a significant impact on the treatment of both migraines and PFO.

An intriguing study that demonstrates the potential connection between genetic determinants of cardiac structure and predisposition to migraine was provided by Zicari and co-workers from the University of Siena, Italy [62]. They studied 23 patients with the syndrome of CADASIL (cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy). This familial disease presents with recurrent strokes, dementia, and migraine with aura. Transcranial Doppler studies in 21 of these subjects demonstrated the presence of right-to-left shunting in 15 (71 %), although there was no correlation between the clinical presentation or the amount of MRI changes and the presence of right-to-left shunting. CADASIL is caused by a mutation in the Notch3 gene that regulates cell differentiation in vascular smooth muscles in the brain, and also affects development of the cardiovascular system [63]. Lacunar strokes and subsequent leukoencephalopathy are produced by degeneration of the vascular smooth muscles, leading to arterial wall thickening and luminal narrowing in the small penetrating arteries of the brain. Notch3 is also expressed in cardiac tissue during embryogenesis and regulates the morphogenesis of cardiac valves and septa. Although CADASIL is an unusual genetic disorder, it demonstrates how the predisposition to migraine and cardiac morphogenesis producing a PFO may be genetically linked (Fig. 10.2).

People who have migraine headache, especially migraineurs with aura, have an increased risk of cryptogenic stroke. A more complete discussion of PFO and cryptogenic stroke is presented in Chap.​ 11.

Cryptogenic stroke, or stroke of unknown etiology, is a diagnosis of exclusion after standard causes of stroke have been ruled out by an extensive workup. Typically this definition applies to people <60 years of age, above which it is assumed that atherosclerosis is present and is the most likely etiology of the stroke. Since the origin of the thrombus is usually not found unless a deep vein thrombosis is present, it is hypothesized that one etiology of cryptogenic stroke is a venous thrombus (perhaps from peripheral or pelvic varicose veins) that bypasses the lungs via a PFO and enters the arterial circulation. Occasionally, this paradoxical embolism can be documented by echocardiography where a large thrombus is seen trapped in a PFO. The venous clots that produce cryptogenic stroke are usually <3 mm in diameter (based on the size of the cerebral vessels that they occlude). However, once the clot passes to the brain, it is not possible to prove how it got there.

According to a meta-analysis of 6 case-control studies, the relative risk of ischemic stroke for migraine without aura is 1.8 (range 1.06–3.15), whereas migraineurs with aura have a 2.3 greater risk [7, 8, 64–76]. In addition, the relative risk for ischemic stroke in women with migraine using oral contraceptives is increased to 8.7 (range 5.05–15.05). The results of these epidemiologic studies suggest that women with migraine should not take oral contraceptives. Transdermal patches have lower doses of estrogen and their absorption from the skin bypasses the liver, and is associated with a lower incidence of ischemic stroke; however, this has not been fully evaluated. In the Women’s Health Study, 3,577 women had migraine at baseline. Of these, 40 % had migraine with aura. The greatest association between ischemic stroke and migraine was observed in younger women, <50 years old, HR = 6.2 (95 % CI 2.3–16.2) [39]. In addition, these findings were independent of the usual risk factors for atherosclerosis: women with migraine and aura who had the lowest Framingham risk score had a greater risk of ischemic stroke (age-adjusted HR 3.9, 95 % CI 1.9–8.1).

One study in the Netherlands used MRI to assess 134 patients who had migraine without aura and 161 patients who had migraine with aura compared to 140 matched controls [77]. In the study, the total percentage of patients with an ischemic infarct was not increased in migraine patients versus controls (5 % versus 8.1 %). However, when the data were analyzed by vascular supply, there was an increased incidence of posterior circulation infarcts in migraineurs with aura (8.1 % versus 0.7 % in controls). Although the absolute number of strokes was small, the relative risk of posterior circulation strokes was significant (13 versus 1).

In Iceland, the longest prospective study was conducted with the Age Gene/Environment Susceptibility (AGES)-Reykjavík study in which 4,689 people (57 % women; mean age 51 years) were followed an average of 25 years, and then received a brain MRI. 12.2 % of the participants had migraine and 63 % of them were identified as having migraine with aura. Patients who had migraine with aura had an increased risk of subsequent infarct lesions on MRI (OR 1.4; 95 % CI 1.1–1.8). These results were predominantly due to an association of migraine with aura and cerebellar lesions among women (OR 1.9; 95 % CI 1.4–2.6) [78]. However, this large study did not assess the presence of right-to-left shunting in these patients, so we do not know the relative frequency of PFO in those migraineurs who developed stroke versus the migraineurs who did not have a stroke.

According to the initial theory of migraines and auras, it was believed that aura associated with migraine was secondary to intense arterial constriction that produced ischemia as the cause of the transient neurologic deficit. In addition, it was thought that if the arterial constriction was prolonged and severe enough, it would account for the cerebral infarcts observed in migraineurs. In the course of time our understanding of the etiology of transient neurologic deficits and the time course of vasoconstriction in migraine has completely changed; and the etiology of stroke in migraine is also subject to a new analysis. Since the observational studies have demonstrated that the majority of people who have migraine with aura also have a right-to-left shunt, our hypothesis is that the cause of stroke in migraineurs is predominantly due to a paradoxical embolism through a PFO. This is consistent with the high prevalence of migraine in those people who present with cryptogenic stroke and are found to have a PFO.

The prevalence of migraine headache is approximately 30–50 % in people who present with cryptogenic stroke [79]. Future epidemiologic studies of migraine and stroke need to include a sensitive screening method for determining whether a right-to-left shunt is present in those migraineurs who develop stroke. The question is, in a population of migraineurs who develop stroke, what is the frequency of transient right-to-left shunting? Wilmshurst et al. demonstrated that the prevalence of right-to-left shunts in patients who had stroke and migraine with aura was significantly greater (84 %) compared to patients who had migraine with aura but no history of stroke (38.1 %, p < 0.001), in population controls (12.2 %, p < 0.001), and in patients who had stroke but no migraine (55.6 %, p < 0.05). In addition, the prevalence of shunting in patients who had stroke and migraine without aura (75 %) was also significantly greater than that in population controls (12.2 %, p < 0.001) and in subjects who had migraine with aura but no history of stroke (38.1 %, p < 0.05) [80].

This theory is also consistent with the higher risk of stroke in migraineurs who are on birth control pills or hormone replacement therapy. Estrogen increases the risk of venous thrombosis. If a PFO is present, as suggested by the history of migraine, then it is possible that a venous thrombus, induced by the additional estrogen, may permit a paradoxical embolism to occur and result in a cryptogenic stroke. If this hypothesis is corroborated by data on the incidence of PFO in migraineurs with stroke, then it will not only affect our understanding of the mechanism of stroke in migraineurs, but will also suggest that prophylactic closure of PFO should be evaluated in a randomized trial to test whether it is an effective method of treatment for prevention of stroke in migraineurs. However, this type of study would be difficult to perform as the absolute risk of stroke in migraineurs is small. Therefore, a large number of patients would need to be treated with PFO closure and followed over many years to show an effect.

Women are twice as likely to develop migraine headaches at some point in their lives compared to men. About 50 % of women state that migraines are more frequent or are typically induced in the perimenstrual time frame. Studies suggest that the onset of migraine is associated with the decrease in estrogen blood level that occurs during the luteal phase [81–83]. In addition, the administration of 1.5-mg estradiol gel can postpone the development of migraine in these susceptible women. There is no absolute level of estrogen that was associated with the induction of migraine, suggesting that it is the withdrawal of estrogen rather than a specific blood level that is the trigger [84]. However, there are other reports that estrogen can induce migraine in susceptible women [85]. It is unknown whether a fall in estrogen alone induces migraine, or whether other chemicals such as prostaglandins, which also fluctuate throughout the menstrual cycle, also play a role in triggering the headache.

Migraine without aura commonly improves or is eliminated after menopause, whereas migraine with aura is not reduced by menopause [86, 87]. Surgical oophorectomy also improves or eliminates migraine [88]. However, in one study, 9 % of women reported worsening of migraine following menopause [89].

The effect of estrogen replacement on migraine during menopause is also an important clinical question. In the Women’s Health Study, there were 21,788 postmenopausal women, of whom 30 % had never used hormone replacement therapy (HRT), and 48 % were current users [90]. Of the HRT users, 8 % complained of migraine within the preceding year. By multivariate analysis, the use of HRT was associated with a 42 % increased risk of having migraine. This was not a trial of HRT randomized for the presence of migraine, and it is possible that many of these postmenopausal women were given HRT as a treatment for their headaches [91].

Several large epidemiologic studies describe an increased incidence of cardiovascular events in men and women who have migraine (Table 10.2). The Physicians Health Study followed 20,084 male physicians to assess the risk of developing cardiovascular disease and correlated that with baseline clinical characteristics [92]. Of these men, 7.2 % had a history of migraine headache. The mean age of these physicians was 56 years and they were followed for a mean of 15.7 years. In these men with migraine, there was a 24 % increased risk for cardiovascular disease events compared to those without migraine. This included a 42 % increase in MI, a 12 % increase in ischemic stroke, and a 7 % increase in cardiovascular death compared to men who did not have migraine at baseline. These increased risks were independent of the standard Framingham risks for developing cardiovascular disease, which suggests that migraine is not associated with an increase in the usual risk factors for atherosclerosis.