Migraine is associated with an increased risk of deep white matter lesions, subclinical posterior circulation infarcts and brain iron accumulation

Abstract

Background

Previous studies suggested that migraine is a risk factor for brain lesions, but methodological issues hampered drawing definite conclusions. Therefore, we initiated the MRI “CAMERA” study.

Procedures

We summarize our previously published results. A total of 295 migraineurs and 140 controls were randomly selected from a previously diagnosed population-based sample (n=6039), who underwent an interview, physical examination, and a brain MRI-scan.

Findings

Migraineurs, notably those with aura, had higher prevalence of subclinical infarcts in the posterior circulation (OR=13.7; 95%CI 1.7–112). Female migraineurs were at independent increased risk of white matter lesions (WML; OR=2.1; 95%CI 1.0–4.1), and migraineurs had a higher prevalence of brainstem hyperintense lesions (4.4% vs. 0.7%, p=0.04). We observed a higher lifetime prevalence of (frequent) syncope and orthostatic insufficiency in migraineurs; future research needs to clarify whether autonomic nervous system dysfunction could explain (part of) the increased risk of WMLs in female migraineurs. Finally, in migraineurs aged <50, compared to controls, we found evidence of increased iron concentration in putamen (p=0.02), globus pallidus (p=0.03) and red nucleus (p=0.03). Higher risks in those with higher attack frequency or longer disease duration were found consistent with a causal relationship between migraine and lesions.

Conclusion

This summary of our population-based data illustrates that migraine is associated with a significantly increased risk of brain lesions. Longitudinal studies are needed to assess whether these lesions are progressive and have relevant (long-term) functional correlates.

Introduction

Migraine has been considered for decades as an episodic disorder without long-term consequences to the brain. However, over the past 30 years several studies have been carried out looking at and delivering arguments for a possible association between migraine and brain changes. For this association, three lines of evidence can be recognized.

First, several cases of ‘migrainous infarction’ or ‘migraine-induced stroke’ have been described, suggesting that migraine can act as an acute precipitant of ischemic stroke.^1–6 In such cases stroke is assumed to be directly and causally related to an acute migraine attack. Migrainous infarction has been estimated to occur with a low incidence (1.4-3.4/100.000)^1;7;8 and is probably an overdiagnosed condition,⁹ accounting for <40% of clinical ischemic stroke in women with migraine under age 50.¹⁰ Second, data from several hospital-based stroke case-control studies suggest that migraine is a risk factor for clinical ischemic stroke. A meta-analysis summarized the evidence and calculated a pooled relative risk of 2.2 for migraineurs.¹¹ It seems that higher risks apply for female patients with aura, with higher attack frequency, who smoke, who have hypertension, or who use oral contraceptives. Third, a number of clinic-based MRI studies found an increased prevalence of cerebral white matter hyperintense lesions (WMLs) in migraine patients; a meta-analysis calculated a pooled increased risk of 3.9 (95% CI 2.3–6.7).¹² Although most of these studies showed an apparent association between migraine and ischemic brain lesions, some results were discrepant: few did not find a statistically significant difference;^13–15 some found no difference in prevalence between MO and MA,^15–18 and two studies found a higher prevalence in MA;^13;19 only one study found an increased prevalence with increasing migraine attack frequency, however, the other studies did not assess the influence of attack frequency as an indicator of migraine severity.¹⁹ In addition selection bias and various other methodological problems prohibited drawing definite conclusions from those results.²⁰ It remained uncertain whether migraineurs are at independent risk for permanent brain changes, and if so, there are still questions about whether there are specific subgroups of patients who are most at risk, and what etiologic mechanisms are involved.

However, in non-migraine patients, it has been shown that both clinical and subclinical brain lesions can lead to negative sequellae, affecting both physical and cognitive function.^21–24 If this also translates to lesions that develop as a consequence of migraine (as a condition with a high prevalence), migraine would be an important contributor to neurological deficits in the general population. Therefore, it is of primary importance to investigate whether migraine is an independent risk factor for (subclinical) white matter lesions and infarcts.

We planned and performed a cross-sectional MRI study in an already existing population-based sample of adults with migraine and controls with no a headache history: the Cerebral Abnormalities in Migraine, an Epidemiological Risk Analysis (CAMERA) study.^25;26 We assessed the presence of several types of structural brain changes, compared these between migraineurs and controls, and correlated presence and extension of these changes also to various demographic, medical and migraine characteristics. In addition, we collected data from neurologic physical examination and cognitive tests, to correlate these data with lesion load. The population-based sample allowed us to study a range of migraine presentation, different from clinic studies that are usually based on more severe cases.

In the paragraphs below, we summarize the published, primary results from the CAMERA-study. We conclude with remarks on the relevance of the findings presented, and describe the necessity of future research initiatives.

Migraine as a risk factor for subclinical brain lesions: the CAMERA study

The CAMERA studyis a population-based case-control MRI study in an unbiased sample of 295 migraine cases (n=161 migraine with aura [MA]; n=134 migraine without aura [MO]) and 140 age-and sex-matched non-migraine controls.²⁶ The participants were randomly-selected from a previously diagnosed sample (n=6039) from the Dutch general population.²⁵ This epidemiologic approach minimized the role of potential selection bias, and a proper, multistep method to establish the migraine diagnosis minimized the possibility of diagnostic misclassification and recall bias. Sensitive imaging and lesion rating methods, including full brain covering thin (3mm) T2 and FLAIR slices, blinded expert reading, and semi-quantitative lesion volume quantification, minimized the possibility of lesion misclassification. The complete description of the cohort, with several baseline measurements available from the GEM-study, allowed to statistically control for relevant confounders. This is important as many brain lesions in population based samples result from cardio-vascular risk factors. These measures resulted in a study population that was well characterized using standardized measures. The cohort includes migraine sufferers with general migraine symptomatology and average migraine severity.

From the analyses of the MRI images in conjunction with socio-economic data, cardiovascular risk factors, neurologic exam, and migraine characteristics, it appeared that silent brain infarction in the posterior circulation territory occurred far more frequently in migraine cases than expected.²⁶ Findings were most pronounced in cases with migraine with aura: 8% had subclinical cerebellar infarcts (OR 13.7; 95% CI 1.7–112; compared to controls, controlling for cardiovascular risk factors). Female migraine cases were at increased risk of high-DWML-load (top 20^th percentile of the distribution of DWML-load), independent of migraine subtype and independent of the effects of cardiovascular risk factors (OR 2.0; 95% CI 1.0–4.2). This risk was higher in those with higher attack frequency (in those with ≥1 attack/month: OR=2.6; 95% CI 1.2–6.0), suggestive of a causal relationship. We also confirmed that concurrent smoking, hypertension and long oral contraceptive use increased the risk further.

These population-based findings suggest strongly that migraineurs with and without aura are at independently increased risk for sub-clinical brain lesions. We concluded that – given the high prevalence of migraine - these findings are of potential public-health importance and may have implications for management guidelines for migraine; further study into the possible etiologic mechanisms, but also into the potential functional consequences of brain lesions in migraine patients is required.

Posterior circulation territory infarcts

Based on our finding of an increased prevalence of posterior circulation infarct-like lesions in migraineurs, we further focused on these lesions.²⁷ Of all 60 identified subclinical brain infarcts, 39 (65%) were in the posterior circulation territory, and a majority of these (85%) were located in the cerebellum. Lesions were often multiple, and round or oval shaped, with a mean diameter of 7 mm. The majority (88%) of infratentorial infarct-like lesions had a vascular border zone location in the cerebellum. Prevalence of these border zone lesions differed between controls (0.7%), cases with migraine without aura (2.2%) and cases with migraine with aura (7.5%). Besides higher age, cardiovascular risk factors were not more prevalent in migraineurs with posterior circulation infarct-like lesions compared to those without lesions. The combination of vascular distribution, deep border zone location, shape, size and imaging characteristics on MRI makes it likely that the lesions have an ischaemic origin.

The most likely etiologic mechanism seems to be hypoperfusion and/or embolism, rather than atherosclerosis or small vessel disease. During and after migraine attacks, sluggish low cerebral flow below an ischemic threshold has been described.^28–32 A decrease in brain perfusion pressure (e.g. during migraine) theoretically affects the clearance and destination of embolic particles; narrowing of the arterial lumen and endothelial abnormalities stimulate formation of thrombi; occlusive thrombi further reduce blood flow and brain perfusion.³³ Because the deep cerebellar territories have a pattern of progressively tapering arteries with only few anastomoses present, they are likely to be particularly vulnerable to hypoperfusion related border zone infarct mechanisms.^34;35

No previous study specifically reported on cerebellar lesions in migraine. But interestingly, subclinical cerebellar dysfunction has been reported in migraineurs,^36;37 which raises the question whether more advanced functional tests would have identified cerebellar dysfunction in our cases.

Infratentorial hyperintense lesions

We also evaluated the prevalence, frequency and distribution of infratentorial hyperintense lesions (IHLs). Lesions were identified in 13/295 (4.4%) migraineurs and in 1/140 (0.7%) controls (P=.04).³⁸ Twelve cases had IHLs, mostly bilaterally, in the dorsal basis pontis, seeming to notably affect the transverse fibers on this location. The increased prevalence of IHL in migraineurs extends the knowledge about vulnerable brain regions and type of lesions in migraine brains.

Although the brainstem is involved in migraine pathophysiology, and activation of the dorsal rostral pons and periaqueductal gray matter has been described during attacks, these regions seemed not to be affected by hyperintense signal changes. IHL are likely due to small-vessel disease (arteriosclerosis) and/or repetitive perfusion deficits, similar to the proposed mechanism(s) for the occurrence of similar lesions in non-migraine subjects with pronounced cardiovascular risk profile³⁹ or cases with CADASIL.⁴⁰ The precise etiology of these lesions in migraine and the reason why migraineurs seem to be more susceptible for this kind of vasculo-ischemic brain changes remains still unknown.

Etiologic considerations

There is now strong evidence that migraine is indeed an independent risk factor for DWMLs, silent posterior circulation territory infarcts and IHL. The earlier medical opinion that only female migraine patients below age 45 with MA are at increased clinical stroke risk,^10;41–44 likely underestimates the real extent of brain injury in migraine patients in the general population.

Several features of migraine could contribute to the pathogenesis of both hyperintense lesions and infarcts in migraine, but a combination with other factors may be necessary to finally result in focal lesions. An increased coagulatory propensity, vasoconstriction,^45–48 local excessive neuronal activation, neurogenic inflammation, neuropeptide and cytokine release⁴⁹ or excitotoxity,⁵⁰ and cardiac abnormalities (like a patent foramen ovale, PFO) have been suggested to act as potential contributing or causal factors. Reversible MRI abnormalities during migraine aura, including regions of cerebral vasogenic edema⁵¹ and evidence of vasogenic blood-brain barrier (BBB) leakage in prolonged aura have been reported.^52;53 These abnormalities were linked to (temporary) impairment of the BBB integrity and enhanced permeability of meningeal microvessels.^53;54 It has been suggested that cortical spreading depression (CSD) causes disruption of the BBB through a matrix metalloproteinase-9 dependent cascade mechanism., that may result in local tissue damage.⁵⁵ Penetration of any toxic agent through a temporarilyreduced BBB, might be an explanation for the development of (focal) WML or (silent) infarcts. From studies in experimental and human cerebral ischaemia it is known that peri-infarct CSD-like depolarization potentiates infarct growth.⁵⁶

With the established relationship between migraine and brain lesions, we now need to plan studies assessing etiology and relevance of brain lesions in migraine.

Iron deposits in migraine

In two small MRI studies, higher brain iron levels were found in the periaqueductal grey matter (PAG) and red nucleus of patients attending an headache clinic for migraine or chronic daily headache.⁵⁷ As the PAG is activated during migraine attacks, the increased local iron load was suggested to reflect an impairment of brainstem structures involved in the central anti-nociceptive neuronal network in patients with severe migraine. We examined iron concentration in deep brain nuclei with MRI in participants from the CAMERA study.⁵⁸ T2 values derived from dual-echo MR images (1.5 T) were measured in seven deep brain nuclei in migraine cases (n=138) and controls (n=75); this allows a quantitative assessment of differences in iron concentration in vivo. We separately analyzed subjects under (n=112) and above age 50 (n=101), because measurements in older subjects are increasingly influenced by non-iron-related factors. In migraineurs aged <50, compared to controls, T2-values were lower in the putamen (P=.02), globus pallidus (P=.03) and red nucleus (P=.03). Controlling for age, those with longer migraine-history had higher concentrations in the putamen (P=.01), caudate (P=.04) and red nucleus (P=.001). We found no differences between MA and MO.

The findings suggest that repeated migraine attacks are associated with increased iron concentration in multiple deep brain nuclei that are known to be involved in central pain processing and migraine pathophysiology, and not only in the periaqueductal grey matter. It remains unclear whether the increased iron concentration is just a physiological response induced by repeated activation of nuclei involved in central pain processing, or whether the increased iron concentration could also damage these structures secondarily, e.g. due to formation of free radicals in oxidative stress. Theoretically, damage to these pain-processing nuclei might explain the occurrence of chronification of the disease in a minority of migraine patients. Recent observations from other MRI studies also point at disturbances in the pain-processing-network, and support the concept that changes in a chain of brain locations occur in migraineurs.^59;60 Further study into pain mechanisms in migraine should not be limited to the brainstem areas, but should cover a broader scope, and consider the whole involved pain network.

Syncope, migraine and brain lesions

Many earlier studies evaluated the function of the autonomic nervous system (ANS) in migraineurs. Studies were mostly performed interictally in clinic-based samples,^61–69 However findings are inconsistent with reports of either increased or decreased sympathetic or parasympathetic function. Because of these discrepancies, and because previous studies did not address clinical symptoms of ANS failure, including syncope, orthostatic insufficiency (OI) and postural tachycardia syndrome (POTS), we assessed the prevalence of these entities in migraine using a population-based design.⁷⁰

Migraineurs (n=323) and control subjects (n=153) from the CAMERA study answered a systematic questionnaire and underwent cardiovascular measurements during rest, while standing and after venipuncture, together addressing the prevalence of syncope, OI, orthostatic hypotension (OH) and POTS. The data showed that migraineurs had a higher lifetime prevalence of syncope (46% vs. 31%, P=.001), frequent syncope (≥5 attacks; 13% vs. 5%, P=.02) and OI (32% vs. 12%, P<.001). There was no association between ANS symptoms and the severity of migraine or migraine subtype. Cardiovascular measurements and the prevalence of POTS and OH did not differ significantly between migraineurs and controls. This first population-based study assessing clinical presentations of ANS dysfunction as well as BP and HR reflexes in migraine, showed an increased prevalence of syncope-related ANS symptoms in migraineurs. Our next step is to assess whether those with symptoms of ANS dysfunction are at increased risk for (subclinical) brain lesions and to assess whether it explains part of the increased risk of WMLs in female migraineurs.

Significance of brain lesions in migraine

The data from the population-based CAMERA study support and extend the results from earlier studies, and indicate that migraine is associated with a significantly increased risk of subclinical and clinical ischemic brain lesions.²⁶ The robustness of the methods and the validity of the findings have been acknowledged.^71;72

Our cross-sectional findings suggest migraine attacks may lead to brain lesions and iron depositions. Our conclusions regarding the temporality of the associations are supported by the finding of a higher risk of lesions in those with higher attack frequencies or longer migraine history. To show ongoing migraine attacks lead to progression of lesions, we need follow-up data showing that there is a higher rate of developing new lesions, and more lesion progression over time in migraine cases, compared to (e.g. age related) progression of lesions in controls. In addition, identification of a linear relationship between migraine severity (e.g. attack rate) and volume of lesions (progression), will increase the likelihood of a causal relationship.

Numerous studies in non-migraine elderly subjects demonstrated that silent brain infarcts and WML are associated with increased risk of stroke, dementia, and cognitive decline.^21–24 Findings in several studies also suggest that impairment of cortical brain function occurs more often in clinic-based migraine patients than in controls,⁷³ and some authors described significant cognitive changes during and between migraine episodes,^74–78 but others failed to confirm these findings.^79–82 Cerebellar dysfunction in migraineurs has also been suggested in a few studies,^36;83 and is a known entity in familial hemiplegic migraine (a rare type of inherited migraine).^84;85 However, with respect to migraine patients, it is still unknown whether brain lesions do have any negative (long-term) functional consequences. For these reasons together, it is now essential to measure cerebral or cerebellar function in migraineurs, and correlate the results with eventual presence and progression of brain lesions.

Confirmation that recurrence of migraine attacks is indeed associated with an increasing risk of brain lesions and/or brain dysfunction, will change migraine from an episodic disorder to a chronic-episodic or chronic progressive disorder.⁷¹ Such a shift in conceptualization of the disease also will change goals of treatments, and prevention of migraine may then potentially need to become an important target for secondary prevention in the general population.

Identification of specific factors that increase the risk of brain lesions in migraine patients, such as migraine type, migraine severity, sympathetic nervous system dysfunction, cardiovascular risk factors and PFO, may allow identification of specific subgroups to be treated. For instance, demonstration of an association between MRI lesions and the presence of a PFO would promote the (renewed) initiation of prospective randomized clinical trials on the effect of closure of PFO on migraine severity and associated brain lesions.

Future perspectives

Whether progression of migraine is related to progression of brain lesion can only be demonstrated in a longitudinal design. We therefore started in 2008 to rescan and re-evaluate the entire CAMERA population to measure the change in brain lesion load in migraineurs vs. controls after 8 years. Because of the high prevalence of PFO in MA, we included a cardiac right-to-left shunt-screening with transcranial Doppler, to assess the contribution of PFO in the risk of migraine-related brain lesions. Another aim is to study the functional consequences of the brain lesions. Another approach to assess whether migraine attacks change the brain is to perform MRI scans during migraine attacks. Several types of MRI data are of interest, including structural information, (regional) brain perfusion values, permeability of BBB, brain activation measures and patterns, and concentration values of several metabolites.