Structural MRI in Migraine: A Review of Migraine Vascular and Structural Changes in Brain Parenchyma

Abstract

Migraine is a complex and common disorder that affects patients around the world. Despite recent advances in this field, the exact pathophysiology of migraine is still not completely understood. Structural MRI sequences have revealed a variety of changes to brain parenchyma associated with migraine, including white matter lesions, volume changes, and iron deposition. This Review highlights different structural imaging findings in various types of migraine and their relationship to migraine characteristics and subtypes in order to improve our understanding of migraine, its pathophysiologic mechanisms, and how to better diagnose and treat it.

Introduction

Migraine is a common neurological disorder that presents with recurrent headaches and contributes to significant disability around the world, affecting approximately 15% of the American population alone.^1,2 There are various types of migraine that have been recognized, such as chronic migraine (CM) and episodic migraine (EM), both of which can present with aura or without aura.^3-5 CM is defined as occurring at least 15 days a month with at least 8 of those days involving symptoms that meet diagnostic migraine criteria. ⁶ EM, on the other hand, is defined as fewer than 15 days of headaches per month and it may further progress to CM with time. ⁷ CM and EM may present with or without aura, which is characterized as one or more reversible symptoms that affect vision, sensation, speech, movement, and more; these symptoms may involve flashing lights, paresthesia (burning or prickling sensations), difficulty finding words, and dizziness or vertigo. ⁸ The International Headache Society (IHS) has described in detail the presentation of migraine and many other headache disorders, ⁹ but there is still much left to learn about the pathophysiology and underlying mechanisms of migraine—in particular, advancements in structural imaging may improve our understanding of migraine and how to treat it.

Structural imaging involves a variety of imaging techniques that focus on analyzing the anatomy of the brain and how pathological processes—such as migraine—may cause changes or damage to various brain structures. Utilizing different and particularly advanced MRI sequences can play a significant role in the diagnosis or even prognosis of migraine headache and help to tailor treatment based on different underlying changes. The common structural sequences required to exclude other underlying and secondary causes of headache and help with diagnosis of migraine are: T1-weighted images, T2/fluid-attenuated inversion recovery (FLAIR), susceptibility-weighted imaging (SWI) or other T2*-weighted sequences. Several quantitative analytic techniques have also been utilized in recent years including quantitative susceptibility mapping (QSM) and voxel-based morphometry (VBM) analyses. SWI and other T2*-weighted sequences are sensitive to iron in the brain and can be used to evaluate local deposition of iron in brain parenchyma or changes in the appearance of cortical veins, while T2-FLAIR sequence is used to investigate the effect of small vessel changes in migraine, including PVS and WMH lesions. VBM is an MRI technique that examines focal changes in brain structures and measures local differences in concentrations of brain parenchyma; it involves segmenting the brain into grey matter, white matter, and cerebrospinal fluid (CSF) while fitting MRI sections of the brain onto a template to eliminate anatomical differences between patients. Knowledge about the subtle pattern of changes on imaging, utilizing new advanced imaging techniques, and conducting research on ultra-high field MRI might help us to understand the underlying complex pathophysiology of migraine. While we have summarized relevant literature on structural imaging of migraine, we still have very little knowledge regarding imaging changes during migraine itself. Imaging patients during migraine presents with several practical and ethical challenges, one of which is the exacerbation of participant suffering from migraine due to visual and auditory stimuli during MRI. As a result, there is limited published data on intra-ictal imaging in migraine. In this review, we summarize the structural changes associated with different types of migraine, discuss the potential limitations of prior studies, highlight gaps in our current knowledge and suggest future direction for imaging in this field.

Utility of Different MRI Sequences in Migraine T1/T2 weighted images: Changes in Grey and White Matter Volume

Neurological disorders such as migraine can affect the volume of brain tissue in many ways, as the brain changes to adapt to pathological processes; these can manifest as increases or decreases in GMV, WMV, or overall intracranial volume.

Many studies found decreased volume of brain structures in different migraine subtypes. In patients with vestibular migraine, Messina et al (2017) found decreases in cerebellar and cortical GMV compared to healthy controls. ¹⁰ In EM, various studies reported decreased GMV in the cerebellum, left second somatosensory cortex, and visual processing areas as well as general volume decreases in the posterior hypothalamus.^11-13 Similarly, studies involving CM showed decreases in cerebellar WMV, cortical GMV, and brainstem and anterior hypothalamus volume.^12,14,15 In contrast, numerous brain regions displayed increased volume in the setting of different subtypes of migraine. Messina et al (2017) and Yang et al (2018) demonstrated that patients with (migraine with aura) MA had increased GMV of the cerebellum and visual processing areas compared to controls, as well as increased left interior frontal gyrus GMV compared to (migraine without aura) MWoA.^10,16 With regards to MWoA, various studies found that GMV as well as WMV were increased in the cerebellum, along with brainstem volume and cortical GMV.^16-18

Compared to controls, Messina et al (2017) and Wang et al (2019) found that patients with vestibular migraine showed increased volume in the left superior occipital gyrus, left middle temporal gyrus, left middle frontal gyrus, left thalamus, right subgenual cingulum, right medial superior frontal gyrus, and right angular gyrus.^10,19 Some patients with EM exhibited GMV increases in the periaqueductal grey and cerebral cortex, while patients with CM had reduced putamen GMV. In MA, there was an increase in left superior temporal gyrus volume, while the same studies in MWoA showed an increase in left calcarine cortex volume, both groups demonstrating an increase in cortical GMV.^10,16

Interestingly, some groups found no significant volume change in brain structures in the setting of migraine. One group observed no change in WMV between controls and vestibular migraine. ¹⁰ Other groups also found no significant difference in WMV as well as GMV in the setting of EM.^13,21 Compared to migraine-free controls, 2 studies found no significant difference in thalamus and nuclei volume in MA and no significant change in whole brain GMV in MWoA; comparing both MA and MWoA, there was also no significant whole brain GMV discrepancies.^22,23

Overall, prior studies seem to show decreased GMV in CM patients with a few studies showing decreased GMV in EM patients as well.^11,13-15,20 Chen et al (2019) also found decreased hypothalamus volumes in chronic and EM patients compared to controls, but no significant volume differences between the 2 groups. ¹² These studies also found that there was no correlation between disease duration and volume change, save for Messina et al (2018) which discovered decreased migraine duration in EM with decreased extrastriate cortex GMV and increased duration with increased frontotemporoparietal GMV.^{10,11,14,17,18,23} Messina et al (2018) did note that their study population was relatively small and heterogenous, as well as the fact that 25 of their patients were on preventative therapies that could have affected GMV measurements; this could explain their contrasting findings to that of other studies. ¹¹ Age was typically not significantly different between migraine subgroups^10,15,19,21 or considered a confounding/nuisance variable.^18,23 Valfre et al (2008) and Chen et al (2019) both note a significant age difference between chronic and EM patients, which could complicate the relationship between volume changes and age.^12,15 In terms of aura’s effect on volume changes, only Qin et al (2019) found an associated decrease in GMV with MWoA; other groups discovered that neither MA nor MWoA were linked to volume changes.^17,18,22,23 Reviewed literature shows that GMV and WMV can be increased or decreased in various different migraine subtypes; as a result, further investigation may help confirm some of these changes and clarify whether or not migraine affects GMV and WMV.

SWI/T2*: Iron Accumulation in the Brain

SWI and other T2*-weighted MRI can be used to assess iron accumulation in the brain which is believed to be present in migraineurs more than controls. Several studies showed that migraineurs have lower T2 signal or alternatively, increased R2*/R2’ signal, which represent higher levels of iron accumulation in various regions that can be diffuse and may or may not be secondary to CMB.

In particular, Dominguez et al (2019) and Chen et al (2021) showed that CM displayed lower T2 signal in the red nucleus, periaqueductal grey, right precuneus, insula, supramarginal gyrus, dorsolateral superior frontal gyrus, postcentral gyrus, and cuneus compared to EM and controls. Chen et al (2021) also showed a significant positive correlation between increased pain intensity and susceptibility values for brain regions with increased iron, but other clinical migraine variables showed no such significant correlation. Dominguez et al (2019) and Welch et al (2001) also found that EM patients had decreased T2 signal ²⁴ and increased R2*/R2′ ²⁶ in the periaqueductal grey compared to controls. Between MA and MWoA, Kruit et al (2008) found that the red nucleus of both MA and MWoA patients had lower T2 signal compared to controls. ²⁷ Contrary to this finding, Palm-Meinders et al (2017) showed an unexpected increase in T2 signal in the putamen of MWoA patients instead of a decrease; however, the baseline study that preceded this investigation found decreased T2 signal in deep nuclei of patients with MWoA. ¹ Granziera et al (2014) focused instead on the thalamus, finding significantly shorter mean T1 and T2* relaxation times in the thalamus of MA patients compared to MWoA patients and controls, indicating an increase in either cellular structures (such as neurons or glia) or iron. ²⁸ In MWoA, Kruit et al (2008) also showed increased iron deposition in the red nucleus compared to healthy controls. ²⁷

Conversely, Tepper et al (2011) discovered significantly increased T2 signal in the globus pallidus of EM patients compared to CM, which could mean higher iron deposition in CM cases. ²⁹ In addtion, Welch et al (2001) observed decreased R2* and R2’ signal in the red nucleus and substantia nigra of CM cases, likely demonstrating decreased levels of iron in these regions compared to that of EM and healthy control patients. ²⁶

Interestingly, Dominguez et al (2019) did not find a significant difference in T2 signal between EM and controls in the globus pallidus and red nucleus. Kruit et al (2008) also showed that mean T2-values did not differ between MA, MWoA, and controls. Amongst patients with either MA or MWoA, age younger than 50 years was significantly associated with lower T2-values in the red nucleus. Similarly, in migraine-free controls older than 50 years, they demonstrated lower T2-values in the putamen, head of the caudate nucleus, and globus pallidus. Overall, most of the literature analyzed seems to show that various migraine subtypes are associated with increased accumulation of iron in brain parenchyma particularly in CM cases, pointing to iron as a potential migraine biomarker of interest in future studies.^24-28

Microbleeds

CMB are defined as being small accumulations of blood in brain tissue that are detectable by SWI/T2* MRI (Figure 1). CMBs may appear over time with increasing age and are thought to be secondary to weakening of small cerebral vessels, increasing risk for stroke, and other neurological pathologies. ³⁰ While CMBs also contain iron, which would appear as increased iron signal on quantitative brain imaging such as QSM, they are focal regions of deposited blood product with various etiologies, such as vasculitis or amyloidosis.

Figure 1.

A. CMB visualized as round, dark lesions (arrow) on SWI sequence in the left temporal lobe in a migraine case with aura. B. Asymmetry in the appearance of the cortical vessels is more prominent on the left side (arrow) ipsilateral to the CMB.

Arkink et al (2015) examined the prevalence of CMBs in 3 different groups: MA, MWoA, and healthy controls. They found that the overall prevalence of CMB did not differ between these groups. However, patients with HTN or DM as well as MA or MWoA tended to have higher numbers of infratentorial CMBs compared to controls without HTN or DM, however neither DM nor HTN significantly increased the odds of infratentorial CMBs in controls without migraines. They also showed that patients with cerebral infarcts and either MA or MWoA had more CMBs overall than controls with cerebral infarcts, and that migraineurs tended to have CMBs and infarcts together more often than separately. ³¹ According to a study by Nakagawa et al (2018) on microbleeds in sentinel headache (SH) and migraine patients with aneurysms, SH patients with aneurysms showed more CMBs compared to migraine patients with aneurysms. ³²

Although the existing literature does not yet show a clear connection between CMBs and primary migraine, there are studies on migraine headache in the context of Cerebral Autosomal Dominant Arteriopathy with Subcortical Infarcts and Leukoencephalopathy (CADASIL). Nannucci et al (2018) showed lower prevalence of CMB with MA and MWoA, while Shindo et al (2020) found that 13.6% of their study population with CADASIL and CMB also had MA.^33,34 Another caveat to consider is presence of CM in cases with underlying cerebral amyloid angiopathy (CAA) or spontaneous superficial siderosis (SS). ³⁵ Some studies on hereditary type CAA showed that more than half of this population suffers from MA ³⁶ and as expected, they have lobar CMB or SS on their MRI.

Reviewing the current literature highlighted that migraine might have an accumulative effect on other causes of small vessel disease, increasing the prevalence of CMB. The conflicting results might be due to heterogeneity between different cohorts and presence of concurrent under-diagnosed other vasculopathies in some studies. Therefore, additional investigation into this relationship, particularly with ultra-high field MRI, might yield interesting insights into the microvascular mechanisms of migraine.

Changes in Cerebral Venous Structures

Migraine may also manifest in patients as changes to vascular structures in the brain, especially in the cerebral veins (Figure 1); examining these changes in cerebral venous circulation could in turn reveal more about the vascular aspects of migraine pathophysiology.

Breiding et al (2020) carried out a retrospective analysis of MA patients using SWI; by performing both a visual analysis and quantitative volume calculation of cerebral veins in MA, they found that 19 of the 20 unilateral headache patients had prominent focal veins in one hemisphere (the visually dominant hemisphere, or VDH) also had higher total cerebral vein volumes in that hemisphere (the calculated dominant hemisphere, or CDH). They were also able to demonstrate a significant association of headache location with VDH in these patients, as well as a negative correlation between CDH volume and time elapsed since headache onset. ³⁷

This correlation was reinforced by various case studies; one such study by Gocmen et al (2015) showed that in 2 pediatric patients with potential MA, episodes of right-sided weakness without headache were accompanied by more conspicuous left-sided cerebral veins, particularly in the parieto-temporal and posterior regions. Repeat SWI was performed after resolution of their symptoms, revealing normal SWI findings with no venous enlargement. ³⁸ Karaarslan et al (2011) performed yet another review of 2 MA patients, one adult and one pediatric; while both patients exhibited right-sided numbness, the former had prominent venous vasculature in the right parieto-occipital lobe, while the latter showed prominent veins in the left parietal and occipital lobes. As with the prior 2 studies, any abnormal SWI findings resolved after migraine symptoms subsided. ³⁹ Finally, Fedak et al (2013) reviewed a series of cases involving 4 female pediatric patients with sporadic hemiplegic migraine, all of whom experienced additional unilateral weakness or numbness. Interestingly, only one of the 4 patients exhibited prominent cerebral veins on the same side as their reported weakness. While distinct changes in venous vasculature appeared on SWI within 6 hours of symptom onset, they disappeared with symptom resolution in similar fashion to the previous studies. ⁴⁰

Overall, these findings point towards SWI’s potential as a diagnostic tool for different types of migraine, especially MA. These studies reveal a potential relationship between the opposing location of prominent cerebral venous structures and unilateral neurological symptoms such as weakness and numbness; further investigation into this relationship could even reveal novel mechanisms of migraine as they relate to cerebral vasculature. SWI is a relatively quick MRI sequence that does not require contrast and coupled with the lack of literature on SWI usage for analyzing cerebral veins in migraine, future exploration of this imaging modality could lead to the creation of novel diagnostic tools for migraine and a better understanding of its relation to cerebral blood flow.

T2/FLAIR sequences: White Matter Hyperintensities

White matter lesions appear as hyperintensities (bright areas of increased signal) on T2-weighted MR sequences (Figure 2). These lesions increase in volume and quantity with age and can be secondary to various underlying pathologies and neurological disorders such as migraine.

Figure 2.

A. Punctate deep white matter hyperintensities (WMH) (arrows) on sagittal T2-weighted MRI, more prominent in the frontal lobes. B. Migraine-free control without punctate WMH.

Studies have found that migraine may be associated with increased WMH in various locations within the brain; for example, Negm et al (2018) demonstrated increased overall volumes of WMH in MA compared to MWoA and controls. ⁴¹ Another group found that there was a significant association between the duration of MA and the number of migraine attacks with increased global WMH. ⁴² In contrast, when examining patients with MWoA, Xie et al (2018) observed a generally higher incidence of WMH compared to patients with MA, but this result did not reach statistical significance. ⁴³ Palm-Meinders et al (2012) showed that specifically female patients with MWoA possessed higher levels of deep WMH compared to controls, whereas Kruit et al (2010) found that female patients with migraine independent of subtype exhibited increased levels of deep WMH.^44,45

However, there are some conflicting findings on the relationship between WMH, migraine, and other factors, with some papers finding no significant association of WMH progression with different migraine clinical characteristics. When examining MA patients vs control patients, Cheng et al (2018) found no difference in total WMH burden. Two other studies found that for both MA and MWoA, there was also no difference in total ⁴⁷ and periventricular ⁴⁴ WMH quantity.

Negm et al (2018) and Honningsvåg et al (2018) report no significant correlations between sex or migraine duration with WMH, while Xie et al (2018) shows significantly higher disease durations in WMH patients and Palm-Meinders et al (2012) shows increased deep WMH in female migraineurs.^41,43,44,47 One factor that could explain these findings regarding sex and duration, however, is sample size; small sample sizes could make finding consistent trends difficult,^41,43 as does imbalance between the number of male and female patients studied.^46,47 As for population-based studies, loss of patient follow-up can be another complicating factor. ⁴⁴ Presence of aura with migraine was another area of contention, with 2 studies finding WMH to be significantly more common in MA^41,42 and 2 others reporting that there was either no significant difference in WMH between MA and MWoA.^41,43 Sample size is yet again a complicating factor for these studies, with Cheng et al (2018) reporting that all MA patients in their study also had MWoA.^41,46,47

Given these conflicting results, there is a need for larger cohort studies including cases with both chronic and EM with and without aura to assess if the severity and distribution of white matter changes are associated with any subtypes of migraine and whether they correlate with severity or frequency of clinical symptoms. With the association between WMH and age, future research could reveal more about the role of WMH in migraine, either as a causative injury to brain parenchyma or as an end-result of vascular changes in brain tissue due to migraine.

Perivascular Spaces (PVS)

PVS, otherwise known asVirchow-Robin spaces, are a series of fluid-filled cavities surrounding cerebral blood vessels (Figure 3). They are thought to be involved in the clearance of waste products and interstitial fluid in the brain. Recent studies have shown that changes in this PVS system play a role in various neurological conditions, in particular migraines (Figure 4). One such study uses an animal model to demonstrate that cortical spreading depression, a phenomenon thought to contribute to migraine aura, leads to closure of the PVS and impairment of glymphatic flow. ⁴⁸ However, much is still not known about the underlying mechanisms by which PVS quantity and volume contribute to the pathophysiology of migraine, which drives further investigation into how these microvascular changes contribute to migraine.

Figure 3.

A 3D map of centrum semiovale (CSO) and basal ganglia (BG) PVS generated from physician-reviewed automated segmentation of PVS. ⁵³

Figure 4

. A. Enlarged Centrum semiovale (CSO) PVS (arrows) on sagittal T1-weighted MRI in a case with CM B. Migraine-free control without enlarged CSO PVS.

A study by Machado Jr. et al (2001) found that between equal numbers of age-matched patients suffering from migraine of unspecified subtypeand and headache-free controls ranging from adolescence to middle age, 40% of migraine patients showed enlarged PVS compared to only 7.1% of controls. ⁴⁹ After comparing migraine of unspecified subtype in pediatric patients to adult patients with post-pubertal migraine onset, Schick et al (1999) showed that 61% of pediatric migraineurs had larger clusters of PVS compared to only 27% of pediatric controls and 13% adult migraineurs. ⁶ In contrast to these findings and other prior studies, a more recent investigation by Husøy et al (2016) discovered a decrease in PVS quantity for MWoA patients in the basal ganglia (BG) and hemispheric white matter (HWM); additionally, they showed only small differences in PVS number between headache groups and headache-free patients, as well as finding no significant association between PVS and headache category, frequency, or disease duration. ⁵⁰ Overall, the existing literature on PVS in migraine seems to indicate that migraine is generally associated with an increase in PVS volume; though the findings from Husøy et al (2016) contradict this trend, their study notes that the finding of fewer PVS with MWoA could be spurious, especially considering that it was the only significant finding between migraine groups; since MWoA and MA are thought to be closely related, one would also expect a similar decrease in PVS for MA within the same study. ⁵⁰

Based on current literature on small vessel changes in migraine there are significant discrepancies which require further investigation. Ultra-high field MRI and more quantitative studies on WMH and PVS in different types of migraine, considering duration of symptoms, sex and underlying cardiovascular disease will help improve understanding of the role of small vessel disease in the course of migraine headache.

Discussion

Despite years of investigations on migraine imaging, there is much more to learn about structural imaging findings of migraine. Certain studies have shown that these findings could potentially be used as biomarkers for migraine; Chen et al (2017) found that periaqueductal grey matter (PAG) volume expansion might be useful as a diagnostic marker of EM, as well as for treatment prognosis. Further exploration of these findings could enhance current diagnostic techniques and improve treatment of different migraine types in the future. Xie et al (2018) shows that findings such as increased WMH might even be able to predict migraine prognosis in the short term, which could further inform treatment decisions. ⁴³

The small sample size, retrospective design, and difficult circumstances surrounding participant recruitment such as timing of the imaging in relation to headache could explain many discrepancies seen across different studies. ⁴³ However, studies done with smaller participant pools could help set the stage for future studies with larger, more representative sample sizes. Large sample-size studies can overcome some of the abovementioned limitations but may have their own bias if not corrected for several underlying factors, resulting in more contradictory outcomes as summarized in this review.^{13,42,44,47,50} These contradictory results might be secondary to underlying differences between various types of migraine, chronicity of the symptoms or differences in the demographics of the studied population. In the future, additional population-based studies of larger scale including different types of migraine with and without aura, chronic or episodic, correcting for other underlying vascular pathologies as confounders, could help clarify the seemingly contradictory results of many of these investigations. One review by Chen et al (2022) describes how performing additional studies and analyzing more existing literature may also help us better understand how specific migraine subtypes lead to certain structural changes on MRI. ⁵¹ There is also a lack of published literature on the role of small vessel changes including CMB, WMH and PVS in migraine; The limited studies that do exist on these topics are over a decade old and may not be reflective of current findings due to rapid advancements and dramatic improvements in imaging technology in recent years. Utilization of more updated imaging technology, such as ultra-high-field 7T MRI, could give us insight into structural brain changes that was inaccessible in the recent past. Utilizing 7 T MRI provides several advantages over clinically available 1.5 T and 3T MRI, including higher contrast-to-noise and signal-to-noise ratios; this helps generate higher-resolution images and allows for better visualization of fine structural changes in the brain. ⁵² As of now, it is still unclear whether certain structural changes in the brain are precursors to migraine or consequences of migraine; for that reason, more longitudinal studies would be helpful to clarify when exactly these changes occur in the timeline of migraine development, information that could help improve diagnostic and treatment decisions in the future.

Conclusion

Structural imaging has shown us the innumerable ways in which brain anatomy can change in association with neurological disease and disorders, especially migraine. Some of these structural changes hint at pathological alterations of metabolic processes and pathways within the brain, and others indicate damage done to brain structures or relationships with migraine characteristics. Though not all studies necessarily agree with each other, and literature is scarce for certain topics, these challenges only further highlight the need for additional investigation into structural changes in the brain and their contribution to migraine. With cutting edge imaging technology and a wealth of information at our fingertips, continued research will pave the way for our understanding of migraine pathophysiology and how to effectively treat it.