Functional Imaging in Primary Headache Disorders


  • F. Michael Cutrer MD


Over the past two decades, the development of new functional neuroimaging techniques has improved our understanding of the brain events underlying several primary headache disorders. In migraine and cluster headache, the advent of these techniques has shifted the emphasis in pathophysiological research away from the vessel and back to the brain.


Dr. Cutrer is an Assistant Professor of Neurology at the Mayo Clinic in Rochester, Minnesota.


Attempts to investigate the pathophysiology of primary headache disorders have been hampered by their transient and unpredictable nature. The lack of any demonstrable gross pathological change or any other readily identifiable biomarker has slowed progress in headache research for decades. These factors combined with a paucity of effective treatments stigmatized headache sufferers, relegating migraine and other headache disorders in the minds of many to the category of psychosomatic complaint. At the start of the last quarter of the 20th century, there was a limited (at best) understanding of even the most common and devastating primary headaches such as migraine and cluster headache (CH). Most of the available data came from investigations of vessels, structures much more easily studied than the brain. Understandably, the predominant theory of headache was based on a vascular hypothesis and any head pain that throbbed was called a “vascular” headache. The rise of new functional imaging techniques, however, in the closing decades of the 20th century revolutionized our understanding of headache pathophysiology and shifted the emphasis away from the vessel and toward the brain.

Functional neuroimaging breakthroughs in migraine with aura

Some of the most intriguing information gleaned from neuroimaging has pertained to migraine. Throughout most of the 20th century, the dogma of headache pathogenesis was that migraine was caused by changes in the caliber of cerebral vessels. The aura was thought to be caused by vasoconstriction and a resulting relative shortfall in cerebral blood flow (CBF) to visual and/or sensory cortices. The headache was thought to be caused by a rebound vasodilatation which resulted in mechanical activation of nociceptive neurons within the walls of the engorged vessels. However, in the early 1980s, Jes Olesen et al employed intra-arterial 133Xe blood flow imaging to investigate changes that occurred in the brain during aura-like symptoms and headaches that were sometimes induced during carotid angiography. Olesen's findings flew in the face of the widely accepted vascular theory of migraine and were more consistent with neurogenic explanations for migraine aura proposed by Lashley and Milner in the 1940s and 1950s. These studies demonstrated that reductions in regional CBF (rCBF) observed during the aura were not of sufficient magnitude to cause ischemia and that there was an apparent anterior spread of the blood flow decrements which did not respect neurovascular boundaries. Perhaps most damning to the vascular theory of migraine was the fact that in some patients, the migraine headache appeared while blood flow was still decreased and before the appearance of hyperemia. Olesen's findings refocused attention on the possibleimportance of cortical spreading depression (CSD) in migraine. CSD, described by Leao in the 1940s, is an electrophysiological phenomenon of cortical hyper-excitation followed by suppression, which, once initiated by mechanical or chemical stimulation, migrated over the cortical surface of experimental animals at a slow rate of 3-5 mm per minute. Not surprisingly, Olesen's findings were controversial and some investigators proposed that a measurement artifact known as Compton's scatter was responsible for both an underestimate of the observed CBF decrements and the apparent spreading of the phenomenon across neurovascular boundaries. In the early 1990s, Woods et al published the case report of a subject who was being studied with positron emission tomography (PET) as control in an unrelated protocol who had the onset of an attack of migraine without aura. Although the visual symptoms reported by the subject were not typical for visual aura, spreading drops in blood flow of a magnitude similar to those observed by Olesen were detected in the occipital lobes on PET. In the mid 1990s, another functional imaging technique known as perfusion weighted imaging (PWI) was applied to the study of spontaneously occurring migraine visual auras. PWI, a gadolinium-based functional magnetic resonance imaging (fMRI) technique which is particularly sensitive to changes in the cortical microvasculature detected decrements in blood flow of about 35% in the occipital cortex contralateral to the visual hemifield affected by the aura. The area of blood flow change enlarged anteriorly with time. Therefore, within a decade, Olesen's original findings were confirmed by 2 independent groups, in several patients with spontaneous migraine attacks employing distinct functional imaging techniques, neither of which was vulnerable to Compton's scatter artifact. Interestingly, another fMRI technique, diffusion weighted imaging (DWI) that is based on changes in net translational movement of water across neuronal membranes and that has been documented in both ischemia and animal models of induced CSD, showed no change either during or just after spontaneous migraine visual aura in humans.

More detailed information about migraine aura has come to us from Blood Oxygen Level-Dependent (BOLD) imaging, a technique based on changes in flow-related-deoxyhemoglobin concentration and the corresponding in shift MRI signal. In a study reported by Hadjikhani et al, BOLD imaging was carried out during both spontaneous migraine visual auras (n = 2) and visual auras induced in a single individual by exercise. In all subjects, there was a loss of activation to visual stimulus that appeared in the occipital lobe contralateral to the symptomatic visual field at the onset of the aura and which resolved at the end of visual symptoms, while in the occipital lobe ipsilateral to the visual field defect there was a continuous normal response to stimulation. The identification of a subject whose auras were inducible by exercise allowed imaging: before the onset of symptoms, for the full duration of visual aura symptoms and into the headache phase after resolution of any visual disturbance. At the onset of aura symptoms, suppression of activation was observed first in an area of visual association cortex known as V3a. The suppression then spread across areas of contiguous occipital cortex at a rate of 3.5 mm per minute to involve both primary visual and association cortices, a rate similar to that seen during CSD. The area of BOLD perturbation within striate cortex closely matched the observed retinototopic visual disturbance. PWI performed immediately after completion of the BOLD imaging revealed perfusion defects similar to those observed in previously studied auras in same areas that had exhibited abnormal BOLD activation during the visual symptoms.

Functional imaging is likely to be a major driver in our ongoing quest for an understanding of primary headaches. The possibilities are numerous.

At around the same time, Cao et al investigated activation patterns within the occipital cortex induced by visual stimulation. In 5 subjects who developed visual changes (but not classical aura) and/or headache after stimulation, the headache or visual symptoms were preceded by a suppression of the initial activation pattern produced by the stimulus. This area of suppression spread across the occipital lobe at a slow rate (3-6 mm per minute). In a later case series published by the same group, migraine attacks (with and without aura) triggered by visual stimuli, 75% were found to have increases in BOLD signal within the red nucleus and substantia nigra prior to changes seen in occipital cortex, implicating these subcortical structures in migraine both with and without aura.

Functional neuroimaging breakthroughs in migraine without aura

Functional neuroimaging techniques including PET and fMRI have also been applied to the study of migraine without aura. Perhaps the most important functional imaging study of migraine without aura to date employed PET performed in a series of patients during spontaneous unilateral migraine headaches not preceded by aura and again after effective treatment of the headaches. An 11% increase in rCBF was measured in the medial brainstem contralateral to the headache as well as in cingulate, auditory, and visual association cortices. Successful treatment of the headache with sumatriptan resulted in a normalization of rCBF in cortical areas but did not reverse the brainstem CBF increases. These findings led Weiller et al to propose that a “generator” located within the superior brainstem might initiate the development of migraine headache. In a later PET study, brainstem activation was not observed after activation of trigeminal nociceptive fibers in the forehead elicited by injection of the chemical irritant, capsaicin. Similar increases in medial brainstem CBF were reported in a subsequent case of a patient who experienced a typical migraine headache without aura while in the PET scanner for an unrelated protocol.

Investigations using PWI have not shown changes in occipital hemodynamic parameters during migraine without aura. Similarly, no DWI changes were detectable in any area of cortex or white matter during attacks of migraine without aura.

Functional neuroimaging breakthroughs in CH and autonomic cephalalgias

Functional neuroimaging has also contributed to our understanding of the pathophysiology of CH and related autonomic cephalalgias. For years, CH along with migraine was considered a “vascular” headache. In the mid 1990s, May et al employed PET imaging to reshape current thought about CH and shift the focus back to the brain. They studied the patterns of brain activation in 9 CH patients during acute cluster attacks and compared their patterns to those seen in the headache-free state. Using this approach, they identified areas of activation within the posterior hypothalamus during acute CH which were ipsilateral to the side of pain, suggesting an important role for the hypothalamus in CH and implying that the periodicity in cluster attacks might arise from hypothalamic circadian rhythms. A subsequent study employing MRI voxel-based morphometric analysis has shown an increase in the averaged gray matter volume in these same hypothalamic areas in CH patients when compared to healthy controls.

Another particularly interesting series of data indicate that functional imaging may eventually prove to be a mechanism-based means of distinguishing primary headache disorders. Several of the trigeminal autonomic cephalalgias (TAC) have patterns of activation on functional imaging that are distinct from CH and each other. For example, unlike CH, PET studies in paroxysmal hemicrania show no differences in activation when acute attacks are compared to the pain-free state between untreated attacks. However, when the acute attack was compared to the indomethacin-treated pain-free state, differences were seen in the posterior hypothalamus and midbrain contralateral to the pain. A functional MRI study of SUNCT, another TAC, has shown bilateral hypothalamic activation although PET-based investigations have indicated ipsilateral activation similar to CH. A related headache disorder, hemicrania continua, when studied with PET, showed contralateral posterior hypothalamic, ipsilateral dorsal rostral pons, and ventrolateral midbrain, as well as bilateral pontomedullary junction activation. While functional imaging at this point is not a practical clinical tool for diagnosis, it points us in the direction of the underlying pathophysiology in many disorders which, until now, have been defined solely on characteristics of the clinical attack.

Possibilities for the future

Functional imaging is likely to be a major driver in our ongoing quest for an understanding of primary headaches. The possibilities are numerous. Developing magnetic resonance spectroscopic techniques promises to aid us in the investigation of the biochemical events that underlie migraine and the other primary headaches. Other potential future breakthroughs may lie in the combination of expanding genomic information with detailed functional neuroimaging data to evaluate changes in brain areas, mediating neurochemical pathways implicated by genetic polymorphisms linked to specific headache subtypes. Functional neuroimaging may ultimately be used to follow changes in brain, allowing us to monitor or even predict prophylactic treatment response. Increasing resolution of imaging beyond the level of gross visual inspection promises the next stage of breakthroughs. The future is bright and ripe with possibility.