White Matter Structure

Conventional diagnostic MRI is used to detect gross structural abnormalities. T2-weighted imaging, a standard clinical pulse sequence, is often used to highlight white matter (WM) hyperintensities (WMH). WMH in migraine have been well documented.Reference Hamedani, Rose and Peterlin^10–Reference Park, Lee, Kim, Yun and Kim¹³ A central question regarding WM changes in migraine is whether these are a biomarker of susceptibility or byproduct of the disease.

WMH have been widely studied in relation to cognitive impairment, stroke, and dementia.Reference Debette and Markus¹⁴ and are often presumed to have a vascular origin.Reference Wardlaw, Valdés Hernández and Muñoz-Maniega¹⁵ While WMH have been widely studied, different quantification methods (e.g., automated detection, visual inspection, and volume determination) exist. In the pediatric population, there is no consensus on measurement methods. Migraine also has associations with cognitive impairment,Reference de Araújo, Barbosa, Lemos, Domingues and Teixeira¹⁶ stroke,Reference Kurth, Chabriat and Bousser¹⁷ and dementia.Reference Chuang, Lin, Lin, Sung and Kao¹⁸ WMH are often interpreted as demyelination and axonal injury, which may be an underlying problem in migraine.Reference Fazekas, Kleinert and Offenbacher¹⁹ Several adult studies show WMH are more common in migraineurs than in control populations, and WMH have been shown to be progressive in migraineurs.Reference Palm-Meinders, Koppen and Terwindt^12, Reference Erdélyi-Bótor, Aradi and Kamson²⁰ Specifically, a 9-year follow-up study showed adults with migraine developed significantly more WMH than the control population.Reference Palm-Meinders, Koppen and Terwindt¹²

While the documented progressive nature of WMH in migraine suggests that WMH are a byproduct of migraine, WMH are also found in children with migraine, which may indicate otherwise.Reference Hämäläinen, Autti, Salonen and Santavuori^21, Reference Yılmaz, Çeleğen, Yılmaz, Gürçınar and Ünalp²² Studies assessing WMH in children often focus on the more severe WML, a potential bias in the literature. WML are a WMH that are caused by a degradation of myelin, whereas WMH may also have other origins such as normal structural variations, including lacunae or perivascular spaces. The prevalence rates of WML in childhood migraine patients are 6–67%Reference Candee, McCandless, Moore, Arrington, Minich and Bale^23–Reference Bayram, Topcu, Karaoglu, Yis, Cakmakci Guleryuz and Kurul²⁷ (Table 1), and WML in populations as young as 6 yearsReference Candee, McCandless, Moore, Arrington, Minich and Bale²³ may suggest that WML are not a byproduct of chronic life-long migraines. One study suggests WML development is related to migraine frequency compared to a cumulative lifetime risk.Reference Eidlitz-Markus, Zeharia, Haimi-Cohen and Konen²⁴ However, there are discrepancies in data regarding the prevalence and frequency of WML in relation to pediatric migraines. WML are also found in control populations of children, sometimes at prevalence rates of 4%, which may not significantly differ from pediatric migraine patients.Reference Mar, Kelly, Isbell, Aung, Lenox and Prensky²⁵ There are two major limitations regarding these childhood migraine WMH studies. First, there is little consensus of how to accurately measure and quantify WMH in pediatric populations. Second, the results are based on retrospective chart analysis as opposed to prospective studies. As neuroimaging is not routinely done in migraine, these results are likely from more severe cases of migraine.

Table 1. Pediatric migraine white matter studies

Advanced imaging methods can interrogate brain anatomy and physiology. Particularly relevant to WM is diffusion tensor imaging (DTI). The diffusion signal is based on free diffusion of water, which is hindered by WM fibers.Reference Tamnes, Roalf, Goddings and Lebel²⁸ DTI data enable maps depicting (a) the overall level of diffusion, typically shown as mean diffusivity (MD) and (b) the level of diffusion anisotropy or directionality associated with diffusion. Diffusion anisotropy is measured as: axial diffusivity (AD), the level of diffusion parallel to a WM tract; radial diffusivity (RD), diffusion across or perpendicular to the WM tract; and fractional anisotropy (FA), a composite where 1 is completely anisotropic diffusion (single orthogonal direction of diffusion) and 0 represents completely isotropic diffusion (Figure 1). Each of these metrics provides complementary information about WM structural integrity. For example, MD decreases with cytotoxic edema and increases with cell death;Reference Alexander, Hurley and Samsonov²⁹ RD can index axonal diameter and increases with demyelination;Reference Alexander, Hurley and Samsonov²⁹ AD values decrease with axonal injuryReference Alexander, Hurley and Samsonov^29, Reference Sun, Liang, Le, Armstrong, Cross and Song³⁰ and increase with brain maturation;Reference Alexander, Hurley and Samsonov^29, Reference Ashtari, Cervellione and Hasan^31–Reference Gao, Lin and Chen³³ and more aligned tracts show larger FA values, and disruptions in WM tracts cause decreases in FA.Reference Alexander, Hurley and Samsonov²⁹ FA also increases with age and is a marker of brain maturation; it increases from 10% to 25% between the ages of 5 and 25 years.Reference Lebel and Deoni^34, Reference Lebel, Walker, Leemans, Phillips and Beaulieu³⁵

Figure 1: Sample DTI maps.

To date, only one study has used DTI in childhood migraine (Table 1).Reference Messina, Rocca and Colombo³⁶ The study found decreases in MD, RD, and AD in the corpus callosum, cingulum, corticospinal tract, and superior longitudinal fasciculus (nociceptive pathways), indicating WM tract disruption in the absence of any WMH.Reference Messina, Rocca and Colombo³⁶ Similarly, the optic tract and optic radiations also showed decreased MD, RD, and AD but increased FA. As this study did not differentiate between migraine with aura (MA) and migraine without aura (MO), this may be an area for future studies to compare. From this single study, the authors were unable to conclude whether WM abnormalities stemmed from overuse because migraine experience or prior alteration of these tracts predisposes patients to migraine. While this study did have an age-matched control group, brain development and maturation is an important consideration in interpreting this data.

Some adult studies using DTI techniques show demyelination and axonal injury in the nociceptive pathways and migraine-related areas, though other studies show disruptions in different regions.Reference Chong and Schwedt^37–Reference Rocca, Pagani and Colombo³⁹ One study found demyelination in the left corticospinal tract, the right inferior longitudinal fasciculus, and the anterior thalamic radiations, but no group differences in FA values.Reference Chong and Schwedt³⁷ Another study found demyelination and axonal injury in the corpus callosum, and the right anterior and posterior limb of the internal capsule.Reference Yu, Yuan and Qin³⁸ A final study found demyelination in the visual pathway with reduced FA values compared to controls.Reference Rocca, Pagani and Colombo³⁹ All these studies show decreased WM tract density in pathways associated with migraine, contrasting observations in pediatric migraine (Table 1).

There are discrepancies in DTI results in the nociceptive pathways between pediatric and adult migraineurs. However, it is important to consider that myelination and therefore DTI metrics change over the lifetime; and in particular, pronounced changes of DTI metrics occur throughout development.Reference Bava, Thayer, Jacobus, Ward, Jernigan and Tapert^32, Reference Gao, Lin and Chen³³ As such, contrasts between these studies need to be interpreted cautiously. In pediatric migraine, results show the nociceptive pathway are denser with greater axonal injury compared to controls.Reference Messina, Rocca and Colombo³⁶ In adults, DTI data have shown the opposite, with results indicating demyelination.Reference Chong and Schwedt³⁷ This is potentially an interesting finding for future research as it could be a surrogate imaging marker in migraine and should be further validated. If nociceptive pathways are injured in pediatric migraine, decreased MD, RD, and AD may indicate inflammation. In adult migraine patients, increased MD, RD, and AD suggests axonal injury and demyelination. Inflammation in pediatric migraine may progress to axonal injury in adult migraine and may explain the transition differences in migraine, though normal developmental trajectories confound these investigations. Data for these conclusions are, however, based on limited pediatric evidence, and further studies using DTI in pediatric migraine would be important to support what is reviewed here.

Gray Matter

Conventional anatomical MRI, typically T1-weighted or T2-weighted imaging, have been used to investigate gray matter. This imaging provides detailed structural information, enabling morphometric measurement of structures directly, or more complex analysis can investigate cortical thickness and gray matter density.

In pediatric migraine, gray matter structure has been less studied than WM; to date, there are only three studies looking at gray matter structure in pediatric migraine (Table 2), all using different methods and examining different brain areas.

Table 2. Pediatric migraine gray matter studies

The brainstem, and in particular the periaqueductal gray matter (PAG), has long been theorized as a migraine generator.Reference Rocca, Ceccarelli and Falini^40, Reference Weiller, May and Limmroth⁴¹ A study comparing the diameter of the brainstem in the coronal and midsagittal planes in childrenReference Hämäläinen, Autti, Salonen and Santavuori²¹ has found the diameter of the pons to be significantly greater in the migraine group compared with age- and sex-matched controls, although still within the normal range. No differences were seen in the diameter of the medulla oblongata or the midbrain. The authors argued this result supports the theory that the brainstem acts as a migraine generator, and its overactivity and neurogenic inflammation contribute to the slightly larger size of the pons in pediatric migraineurs compared with controls;Reference Hämäläinen, Autti, Salonen and Santavuori²¹ however, this result is limited as it is based on a single measure of the diameter, not a volumetric assessment. Changes in brainstem gray matter have also been observed in adult migraine, specifically, increased gray matter density in the PAG as well as increased gray matter density of the dorsolateral pons in adult patients with MA compared with controls.Reference Rocca, Ceccarelli and Falini⁴⁰ More recently, diffusion kurtosis imaging (DKI) was used to investigate PAG in adult migraine. DKI is an extension of DTI that aims to assess non-Gaussian components of diffusion and is used to study gray matter. Higher mean kurtosis (MK) indicates diffusional heterogeneity, and MK increases tend to be associated with cell density or tissue complexity.Reference Zhuo, Xu and Proctor⁴² This study showed increased MD and increased MK values (both indicative of increased cell density) in the periaqueductal gray matter in patients, and MK was correlated with age and duration of disease without treatment.Reference Ito, Kudo and Sasaki⁴³ Consistencies between adults and pediatrics in migraine presentation demonstrate hallmarks of the disorder or aspects that are affected early and remain throughout the lifetime. In both pediatrics and adults with migraine, there is evidence of brainstem gray matter increases. This early evidence of brainstem increases lends evidence to the theory that the brainstem is highly involved in migraine pathophysiology and suggests that brainstem changes are characteristic of the disorder (Table 2).

Beyond the brainstem, a second study using a more global approach to investigate gray matter changes found widespread differences in pediatric migraine patients (ages 9–18).Reference Rocca, Messina, Colombo, Falini, Comi and Filippi²⁶ Compared to age-matched controls, migraine patients showed significantly less gray matter density in the frontal and temporal lobes. Interestingly, these reductions in gray matter were not correlated with disease duration or attack frequency. Adult migraine studies have also demonstrated decreased gray matter density and decreased cortical volume in the frontal cortex of migraineurs compared with controls.Reference Dai, Zhong and Xiao^44–Reference Schmitz, Admiraal-Behloul and Arkink⁴⁷ Components of the pain-processing network are located in the frontal cortex; thus, it has been suggested that frontal cortex atrophy is related to the reorganization of the pain network as a result of migraine,Reference Hougaard, Amin, Magon, Sprenger, Rostrup and Ashina⁴⁵ which is supported by the demonstration of atrophy in other parts of the pain network such as the subgenual cingulum.Reference Rocca, Messina, Colombo, Falini, Comi and Filippi²⁶ Alternatively, frontal atrophy may be a biomarker of migraine,Reference Rocca, Messina, Colombo, Falini, Comi and Filippi²⁶ which is supported by the fact that, in both children and adults with migraine, decreased gray matter volume in the frontal cortex is not associated with headache frequency or duration.Reference Kim, Suh and Seol^46–Reference Messina, Rocca and Colombo⁴⁸ The frontal cortex atrophy seen in migraine patients is an especially important aspect of pediatric migraine disorder. Adult migraine patients have an elevated risk of cognitive dysfunction with complaints regarding attention and memory,Reference de Araújo, Barbosa, Lemos, Domingues and Teixeira¹⁶ but similar findings have also been seen in children and adolescents (age 10–18 years).Reference Costa-Silva, de Prado, de Souza, Gomez and Teixeira⁴⁹ The connection between frontal cortex atrophy and executive function deficits is obvious and warrants further investigation. One important question to answer would be which comes first, the frontal atrophy or migraine. If migraine is the cause of frontal atrophy resulting in cognitive deficits, then preventing or limiting migraine in pediatrics becomes critically important. Investigating migraine in younger children, or in children before the onset of migraine, in a longitudinal study would be an interesting avenue to explore this question.

Gray matter in the temporal lobe was affected differently in pediatric patients with and without migraine aura. In MA patients, the left fusiform gyrus had greater gray matter volume compared with controls and MO patients. Also MO patients showed reduced volume of the fusiform gyrus compared with controls. Aura in migraine is a visual disturbance. Because the fusiform gyrus contributes to high-order visual processing, a difference between MA and MO patients is not surprising.Reference Rocca, Messina, Colombo, Falini, Comi and Filippi²⁶ The increased size of the fusiform gyrus in pediatric MA patients may be related to inflammation, though it is unclear why there would be decreased volume in pediatric MO. Interestingly, this exact finding has either not been studied or not been demonstrated in adults, and a recent meta-analysis showed no evidence of changes in that area.Reference Jia and Yu⁵⁰ It is possible based on the large volume of investigations of gray matter in adult migraine patients and the lack of evidence demonstrating that this is a purely pediatric phenomenon and may be partially responsible for the variation in migraine presentation between adults and children.

A recent study investigating gray matter in pediatric migraine examined the interactions of sex and age on whole-brain gray matter cortical thickness, comparing migraine patients and controls (Table 2).Reference Faria, Erpelding, Lebel, Johnson, Wolff and Fair⁵¹ Cortical thickness differences between female and male migraine patients appeared to be an important distinction to investigate. Compared with males, females have greater gray matter cortical thickness in the primary somatosensory cortex. Female adolescents with migraine had greater cortical thickness compared with male adolescents with migraine and all healthy controls.Reference Faria, Erpelding, Lebel, Johnson, Wolff and Fair⁵¹ Interestingly, in adults with migraine, there have been conflicting results with studies demonstrating increased somatosensory cortical thickness,Reference DaSilva, Granziera, Snyder and Hadjikhani⁵² no difference in thickness of the somatosensory cortex in migraine patients,Reference Datta, Detre, Aguirre and Cucchiara⁵³ and also thinning of the somatosensory cortex in migraine patients.Reference Hougaard, Amin and Arngrim⁵⁴ Somatosensory cortical thickness in female adults appears to be important, however, as somatosensory cortical thickness is correlated negatively with response to medications for migraines.Reference Hubbard, Becerra and Smith⁵⁵ Female children and adolescents in general also had greater cortical thickness in areas associated with nociception, including the precuneus, supplementary motor area, basal ganglia, and amygdala.Reference Faria, Erpelding, Lebel, Johnson, Wolff and Fair⁵¹ The authors found sex × disease interactions in structural findings that indicated females with migraine have significant gray matter thickening in the right supplementary motor area, and right precuneus compared with males with migraine and healthy controls.Reference Faria, Erpelding, Lebel, Johnson, Wolff and Fair⁵¹ Female adults have also been found to have more gray matter in the precuneus compared with male adults with migraine and controls.Reference Maleki, Linnman, Brawn, Burstein, Becerra and Borsook⁵⁶ Precuneus findings in females are especially relevant, as the precuneus has been correlated with pain sensitivity in adults.Reference Goffaux, Girard-Tremblay, Marchand, Daigle and Whittingstall⁵⁷ Baseline pain sensitivity is an important measure as it is closely related to the risk of developing chronic pain.Reference Granot^58–Reference Strulov, Zimmer, Granot, Tamir, Jakobi and Lowenstein⁶¹ As migraine is a form of chronic pain, precuneus data indicate that increased size may be a risk factor for experiencing migraine. Overall, these findings indicate that sex differences in migraine are important and demonstrate that sex differences in gray matter volumes of pain-related areas in the brain might predispose females to developing migraine or being less responsive to medication.

Perfusion

Perfusion imaging aims to depict and quantify tissue perfusion and can be quantified in terms of cerebral blood flow, amount of blood passing through a tissue capillary bed (quantified in units of mL/100 g tissue per minute), cerebral blood volume, average volume of blood within a tissue bed (mL/100 g tissue), or mean transit time, that is, time (per second) taken by the blood to pass through a tissue. Perfusion can be measured with MRI using (a) dynamic susceptibility contrast, in which a venous injection of contrast agent is administered and its passage through tissue is imaged to determine perfusion characteristics or (b) arterial spin labeling (ASL) in which blood is magnetically tagged and the passage of the magnetically tagged blood compared to imaging without magnetic tagging is subtracted to reveal perfusion maps. For research studies, ASL is more appealing as it does not require any injections. Clinically, dynamic susceptibility imaging is often used, or other imaging modalities, such as single photon emission computed tomography (SPECT), have also been applied.

Perfusion in pediatric migraine has most commonly been described through case studies of hemiplegic migraine, a rare form of MA involving motor weakness. By their nature, it is difficult to draw conclusions from case studies, but these can show trends when taken in aggregation. In the case of pediatric hemiplegic migraine, there is a trend of transient hypoperfusion contralateral to the side of aura (Table 3).Reference Altinok, Agarwal, Ascadi, Luat and Tapos^62–Reference Toldo, Cecchin and Sartori⁶⁶ While inconsistencies in this trend have been reported,Reference Kumar, Topper and Maytal⁶⁷ factors such as perfusion being measured multiple days after onset of symptoms confound these results. Consistent with these single case studies of altered contralateral perfusion, a small study (n = 4) using susceptibility-weighted imaging (SWI) showed asymmetry in the right and left cerebral vasculature in the early stages of hemiplegic migraine (within 6 h of symptom onset) with abnormalities in SWI contralateral to the hemiparesis, which resolved in follow-up SWI assessments.Reference Fedak, Zumberge and Heyer⁶⁸ While SWI is not designed to detect tissue perfusion, it is highly sensitive to iron deposition, microbleeds, and vasculature. A larger (n = 10) case–control study showed that cerebral blood flow decreased in brain regions associated with aura symptoms when MRI was performed <14 h after onset and increased if MRI was performed ≥17 hours after onset. Time course alterations were, however, not linked with the persistence of aura symptoms.Reference Boulouis, Shotar and Dangouloff-Ros⁶⁹ Using ASL and MR angiography in retrospective chart analysis, it was found that in pediatric patients with MA, there was homolateral hypoperfusion of the side of vasospasm.Reference Cadiot, Longuet and Bruneau⁷⁰ A more recent study also using ASL found differences between pediatric migraine patients with and without perfusion changes during migraine. This study has found that patients who had perfusion abnormalities were more likely to have aura, motor disabilities, confusion, and hospitalization. In this study, however, there were differences in MRI application (24 h, 6 days, and 7 days) between participants, which could have altered the findings.Reference Uetani, Kitajima and Sugahara⁷¹ The type of migraine may also influence observations in perfusion imaging; the last two studies investigated atypical MA, whereas all the previously mentioned case studies examined familial hemiplegic migraine (Table 3). Adult literature shows regional hypoperfusion during the aura phase of migraine and hyperperfusion during the headache phase,Reference Friberg, Olesen, Lassen, Olsen and Karle^72–Reference Pollock, Tan, Kraft, Whitlow, Burdette and Maldjian⁷⁶ consistent with the above findings in children.

Table 3. Pediatric migraine perfusion studies

Both adults and children with migraine appear to have differences in perfusion between migraines (interictally) as well during episodes (ictally). A recent study investigating pediatric and adult migraines has found increased blood flow to the somatosensory cortex in the interictal period of migraine as measured by ASL.Reference Youssef, Ludwick and Wilcox⁷⁷ These findings mirrored earlier findings in adults that demonstrated increased cerebral blood flow (measured using ASL) in the primary somatosensory cortex in MO patients compared with controls. Primary somatosensory cerebral blood flow was also correlated with migraine frequency.Reference Hodkinson, Veggeberg and Wilcox⁷⁸ These observations may explain increased sensory sensitivity in migraineurs and/or may indicate adaptive or maladaptive changes in the sensory cortex to explain migraine pain. The increase in blood flow to the somatosensory cortex may be constant as no differences in blood flow were seen between and during a migraine attack within the patient group.Reference Gil-Gouveia, Pinto, Figueiredo, Vilela and Martins⁷⁹ These findings indicate that differences in blood flow are likely chronic and not attack-dependent. The correlation between pediatric and adult data also suggests that increased blood flow in the somatosensory cortex is a feature of migraine as opposed to a consequence of chronic migraine.

As perfusion investigation techniques improve, particularly with the more widespread use of ASL, it is expected that more studies will investigate perfusion differences between migraineurs and controls. This will assist in investigating hypotheses underlying migraine, including perfusion response to cortical spreading depression. Currently, the available literature suggests that perfusion differences are still a valuable avenue for exploration in the realm of migraine and deserve further investigations in pediatric migraine particularly.

Metabolites

Magnetic resonance spectroscopy (MRS) provides a method to measure the concentration of metabolites in localized volumes of tissue. Different target nuclei enable the quantification of different metabolites, the most typical of which are ¹H followed by ³¹P MRS. Metabolite measurements give a different insight into brain structure, function, and metabolism. For example, metabolites can inform us about energy utilization, WM degradation, and the activity of neurotransmitters.

There are only two studies that investigated brain metabolites in childhood migraine (Table 4). The first study has found that the relative rate of mitochondrial oxidation, as determined by low levels of phosphocreatine concentration and high cytosolic pH, was higher in the occipital lobes of patients than in controls. They also found a lower phosphorylation potential in the brains of migraine patients, and that muscle mitochondrial respiration was abnormal. The authors concluded that there is a bioenergetics deficit in young migraine patients, which is possibly a defining feature of the disorder.Reference Lodi, Montagna and Soriani⁸⁰ This bioenergetics deficit is consistent with findings in adult studies where phosphocreatine levels in the occipital lobe were reduced in patients with migraine, which may suggest mitochondrial dysfunction.Reference Barbiroli, Montagna and Cortelli^81–Reference Welch, Levine, D’Andrea, Schultz and Helpern⁸⁷ The same study also showed a 25% reduction in magnesium ion concentration in juvenile migraine patients, and decreased magnesium levels have been associated with an increased odds of having a migraine attack.Reference Welch, Levine, D’Andrea, Schultz and Helpern^87, Reference Younis, Hougaard, Vestergaard, Larsson and Ashina⁸⁸ Decreased serum levels of magnesium in adult migraine populations have previously been suggested as an independent risk factor for migraine attacks,Reference Assarzadegan, Asgarzadeh, Hatamabadi, Shahrami, Tabatabaey and Asgarzadeh⁸⁹ and three adult studies have also found decreased brain magnesium in migraineurs in the occipital lobe,Reference Lodi, Iotti and Cortelli⁹⁰ the anterior-posterior region,Reference Boska, Welch, Barker, Nelson and Schultz⁹¹ and the frontal and temporal lobes.Reference Ramadan, Halvorson, Vande-Linde, Levine, Helpern and Welch⁹² As such, decreased magnesium ion levels have been suggested to contribute to reduced mitochondrial oxidation as well as reduced bioenergetics seen in patients with migraine, as magnesium is a cofactor in oxidative phosphorylation and stabilizes the mitochondrial membrane.Reference Younis, Hougaard, Vestergaard, Larsson and Ashina^88, Reference Welch and Ramadan⁹³ In two double-blinded, placebo-controlled adult trials, oral magnesium was effective as a prophylactic treatment for migraine.Reference Peikert, Wilimzig and Köhne-Volland^94, Reference Facchinetti, Sances, Borella, Genazzani and Nappi⁹⁵ A third study did not see the same effects, but this was suggested to be due to the absorbability of the magnesium supplement.Reference Pfaffenrath, Wessely and Meyer⁹⁶ The sole pediatric study saw a reduction in headache days and headache intensity using oral magnesium.Reference Wang, Lirng, Fuh and Chen⁹⁷ Evidence suggests that magnesium may be an effective supplement for pediatric migraine patients. As pediatricians and neurologists seek treatments appropriate for migraines in adolescents, magnesium offers an area of research that should be further pursued, as it has good tolerability and minimal side effects.

Table 4. Pediatric migraine metabolites studies

The second study to investigate metabolites in pediatric migraine was a case study, which applied proton MRS (Figure 2) in a case of hemiplegic migraine 15 days after the migraine (Table 4). The authors found a decreased N-acetylasparatate (NAA)/creatine ratio in conjunction with contralateral hemisphere swelling and a mild hyperintensity visible on diffusion-weighted imaging.Reference Toldo, Cecchin and Sartori⁶⁶ NAA is a neuronal marker, and reduction in NAA is often interpreted as decreased neuronal integrity (though this interpretation is incomplete; for a detailed review, see ref. Reference Rae⁹⁸). Creatine provides a measure of energy stores. A single case study is not convincing to argue that this is typical of pediatric migraineurs, although several studies of migraine in adults have shown decreased NAA in various regions of the brain;Reference Dichgans, Herzog, Freilinger, Wilke and Auer^99–Reference Zielman, Teeuwisse and Bakels¹⁰³ one study observed an increase in NAA in the pons in episodic migraineurs,Reference Lai, Fuh, Lirng, Lin and Wang¹⁰⁴ and many studies have not observed any differences in NAA levels in various regions of the brain.Reference Reyngoudt, Paemeleire, Descamps, De Deene and Achten^84, Reference Schulz, Blamire, Corkill, Davies, Styles and Rothwell^85, Reference Wang, Lirng, Fuh and Chen^97, Reference Becerra, Veggeberg and Prescot^105–Reference Watanabe, Kuwabara, Ohkubo, Tsuji and Yuasa¹¹⁴ Location, in addition to type of migraine, and migraine severity may also confound results and explain some of these discrepancies. A recent review argued that, instead of indicating neuronal loss, these reductions in NAA could indicate mitochondrial dysfunction, as mitochondria are proposed to synthesize NAA.Reference Younis, Hougaard, Vestergaard, Larsson and Ashina⁸⁸ If the pediatric case study is replicated and generalized, however, it would lend evidence to mitochondrial dysfunction as one of the mechanisms involved in migraine pathogenesis and could also perhaps provide further support for magnesium supplementation.

Figure 2: Sample proton MRS.

Functional Imaging

Blood oxygen level-dependent (BOLD) functional MRI (fMRI) is an extensively used method to examine correlations in brain activation, either with a task paradigm (task-fMRI) or within the brain at rest (resting-state fMRI). Task-fMRI is generally used to examine brain activation in response to a stimulus, while resting-state fMRI examines functional connections and coherent activation in the absence of a task. Despite the widespread use of BOLD-fMRI to investigate alterations in functional activation and functional connectivity, only one BOLD-fMRI study has been performed in pediatric migraine (Table 5).Reference Faria, Erpelding, Lebel, Johnson, Wolff and Fair⁵¹ The analysis included dichotomizing participants by sex. Results using fMRI showed differences in resting-state functional connectivity (rsFC) between females with migraine compared with male patients and healthy controls. Females with migraine exhibited greater rsFC in the pain network, specifically between the right precuneus and the left putamen, right caudate, left thalamus, and left amygdala compared with males with migraine and healthy controls. Female patients also showed greater rsFC between the left amygdala and the bilateral thalamus, right supplementary motor area, and bilateral anterior midcingulate cortex compared with male patients and healthy controls. Studies on adults with migraine also showed increased connectivity in pain-processing networks. For example, a study applying graph theory analysis to fMRI data showed abnormal connectivity nodes in female adult migraine patients in the precentral gyrus, orbital part of the inferior frontal gyrus, parahippocampal gyrus, anterior cingulate gyrus, thalamus, temporal pole of the middle temporal gyrus, and the inferior parietal gyrus.Reference Liu, Zhao and Li¹¹⁵ Other literature in adults found similar results with greater connectivity in migraineurs in the orbital frontal gyrus, medial frontal cortex, inferior frontal cortex, insula, supplementary motor area, precentral gyrus, postcentral gyrus, inferior parietal gyrus, and occipital cortex.Reference Liu, Zhao and Li^115, Reference Liu, Zhao and Lei¹¹⁶ Also, greater amygdala connectivity to the visceroceptive insula in migraine patients was seen compared with controls and other chronic pain groups.Reference Hadjikhani, Ward and Boshyan^117, Reference Chen, Chen, Liu, Dong, Ma and Yu¹¹⁸ However, not all adult studies have found similar results. One study has found decreased functional connectivity in pain-related regions of the brain in female adult migraine patients, with lower functional connectivity in the bilateral hippocampus, bilateral insula, right amygdala, right anterior cingulate cortex, bilateral putamen, bilateral caudate nucleus, and prefrontal cortex.Reference Gao, Xu and Jiang¹¹⁹ Overall, the underlying state of connectivity is not clear, particularly with heterogeneous methods and analyses. What can be concluded is that abnormal connectivity changes in the pain network underlie the migraine condition; how those changes manifest appears to generally trend towards greater connectivity in the pain network in migraine patients, but further studies are required to confirm these findings.

Table 5. Pediatric migraine functional imaging studies