The understanding of migraine pathophysiology continues to advance rapidly, bringing fresh opportunities for the development of novel acute and preventive therapies. It is convenient to describe the phases of a migraine attack (premonitory, aura, headache, postdrome) relative to the headache phase because headache is the most easily recognizable, stereotyped, and quantifiable feature of an attack. But for a significant number of patients, the other phases of a migraine attack can be more prolonged and even more disabling than headache. Growing evidence indicates that the phases of migraine do not occur in a discrete and linear fashion but rather reflect overlapping chemical, physiological, and anatomical mechanisms.
A migraine attack is an extraordinarily complex brain event that takes place over hours to days. This review focuses on recent human studies that shed light on the evolution of a migraine attack. It begins with a constellation of premonitory symptoms that are associated with activation of the hypothalamus and may involve the neurotransmitter dopamine. Even in the premonitory phase, patients experience sensitivity to sensory stimuli, indicating that central sensitization is a primary phenomenon. The migraine attack progresses to a phase that in some patients includes aura, which involves changes in cortical function, blood flow, and neurovascular coupling. The aura phase overlaps with the headache phase, which is associated with further changes in blood flow and function of the brainstem, thalamus, hypothalamus, and cortex. Serotonin receptors, nitric oxide, calcitonin gene-related peptide, pituitary adenylate cyclase-activating polypeptide, and prostanoids are demonstrated specific chemical mediators of migraine based on therapeutic and triggered migraine studies. A number of migraine symptoms persist beyond resolution of headache into a postdromal phase, accompanied by persistent blood flow changes in several brain regions. Although these phases of migraine have substantial temporal, neurochemical, and anatomical overlap, each represents an important window onto the pathophysiology of migraine as well as a target for therapeutic intervention. A comprehensive approach to migraine requires an understanding of the entire range of mechanisms and resultant symptoms that occur throughout the evolution of an attack.
The Premonitory Phase
It is well known that for many migraine patients, the first symptoms of an attack are “premonitory” that occur up to hours before aura or headache.[1-7] The reliable occurrence of these symptoms in a significant majority of patients indicates that complex brain events are taking place well before the events associated with aura and headache. A better understanding of the mechanisms underlying premonitory symptoms is critical to a complete understanding of how a migraine attack begins. This understanding is particularly important because the “premonitory phase” of an attack may represent an important window of opportunity for novel acute therapies.
The most commonly reported symptoms preceding headache are fatigue, irritability, difficulty concentrating, mood change, yawning, stiff neck, phonophobia, and nausea.[1-7] Other symptoms that have been reported include change in appetite, food cravings, bloating, piloerection, and change in facial expression or body perception among others. Both retrospective and prospective studies indicate that more than 80% of adults[6, 7] and a slightly lower percentage of children experience some type of premonitory symptoms, and electronic diary studies indicate that some patients can reliably predict the occurrence of migraine headache up to 12 hours before its onset based on awareness of premonitory symptoms. Some symptoms may come and go before the headache phase, whereas others may build in intensity leading up to the headache, occur during the headache, and persist well beyond the resolution of headache. Indeed, several of the symptoms that have been described as part of the migraine “postdrome” are the same as those occurring in the premonitory phase.[8, 9] Some of the premonitory symptoms also raise questions regarding the nature of migraine triggers. While light, sound, and smell sensitivity are identified as premonitory symptoms, exposure to bright light, loud sounds, and strong smells are also commonly identified as migraine triggers. Migraine patients may have baseline differences in sensory sensitivity that could make them susceptible to these triggers. It is also likely, however, that many patients identify as triggers particular sensory stimuli to which they are already more sensitive because their acute migraine attack has already begun.
A prevailing hypotheses regarding premonitory symptoms has been that they involve the neurotransmitter dopamine.[10-13] Part of the evidence supporting this hypothesis includes the observation that exogenously administered dopamine receptor agonists produce some of the same symptoms that are experienced by migraine patients in the premonitory phase, namely yawning, nausea, drowsiness, and lightheadedness.[11, 12, 14, 15] Conversely, dopamine receptor antagonists may reverse some of these symptoms and have been suggested to have the capacity to prevent the occurrence of subsequent headache. A study of the dopamine receptor antagonist domperidone given during the premonitory phase of the attack found that it was able to abort a significant number of migraine attacks in a dose-dependent fashion and that it had greater efficacy the earlier it was given (up to 12 hours) before an “imminent” headache. The fact that domperidone doses not cross the blood–brain barrier in significant concentrations (it is used clinically to reduce peripheral effects of dopaminergic agonists in patients with Parkinson's disease) is somewhat hard to reconcile with the fact that several of the “dopaminergic” premonitory symptoms of migraine are believed to be mediated by the central nervous system. Nonetheless, it is clear that other dopamine antagonists, such as metoclopramide, that do cross the blood–brain barrier can be effective acute therapies for migraine in some patients,[17-19] supporting the concept that dopaminergic mechanisms in the brain play a significant role in a migraine attack and potentially early in the attack.
Migraine patients have been reported to be more sensitive than non-migraneurs to dopaminergic agonists as indicated by increased symptoms of yawning, nausea, and hypotension. This has led to the idea concept that migraine patients have “dopaminergic hypersensitivity” that may play some role in the disorder.[11, 14] Even the patients with these increased “dopaminergic” symptoms, however, do not experience headache in response to dopamine agonists. Headache is also not a common adverse effect of dopamine agonists in patients with Parkinson's disease being treated with these drugs. Taken together, these observations suggest that while dopamine release may occur in the premonitory phase of a migraine attack and migraine patients may be more sensitive to its effects, dopaminergic mechanisms are only one component of a complex neurochemical cascade.
Another widely held hypothesis regarding the premonitory phase of migraine is that it involves changes in the activity of the hypothalamus. Although clinical neurophysiological studies indicate that widespread changes in brain excitability occur preceding headache, a specific role for the hypothalamus has been hypothesized based on the symptoms involving changes in mood, appetite, and energy, all of which could be attributed to this brain region. Recent imaging studies have begun to provide additional support for a significant role for the hypothalamus in migraine. A positron emission tomography (PET) study by Denuelle and colleagues showed increased blood flow in the hypothalamus during a migraine attack. Recent studies specifically examining the premonitory phase of headache have exploited the fact that the migraine trigger nitroglycerin (NTG) may evoke not only migraine headache but premonitory symptoms as well. Sprenger and colleagues have recently examined changes in brain activity during premonitory symptoms evoked by NTG using H2O PET. Preliminary reports of their findings indicate that indeed, there are increases in hypothalamic blood flow that are correlated with migraine premonitory symptoms. The exciting implication of these findings is that there may be specific hypothalamic mechanisms that are novel targets for therapies that could be administered before a headache takes hold. In addition to the multiple neurotransmitters and neuromodulators that regulate hypothalamic function, specific hypothalamic peptides may represent important new therapeutic targets. A good example is orexins, which show promise in animal models as potential mediators of migraine and targets for treatment.
The consistent occurrence of a premonitory phase raises multiple important questions. Given that the premonitory symptoms may be subtle, hard to quantify, and in some cases amplifications of sensations or behaviors that occur throughout the course of a normal non-migrainous day, at what point are these symptoms pathological and indicative of an impending headache? Are there specific symptoms that are more reliable than others at identifying the onset of a migraine attack? What occurs during the transition from the premonitory phase to the headache phase? At what stage is therapeutic intervention appropriate? Further quantitative study of the premonitory phase with prospective clinical studies, imaging, electrophysiological, and pharmacological approaches will yield key information regarding these important questions.
The Aura Phase
Several recent studies have focused on the migraine aura and its relationship to the remainder of the attack. As with the premonitory phase, the migraine aura has traditionally been viewed as a distinct phase of the attack that precedes the headache and other symptoms associated with the headache phase. This view was reinforced by animal studies showing that cortical spreading depression (CSD) can cause activation of nociceptive pathways, leading to the hypothesis that cortical activation associated with CSD can be a direct cause of headache. Subsequent studies of the responses of freely moving rodents, however, indicated that a single CSD event does not elicit significant pain behavior. Furthermore, quantitative examination of the timing of migraine symptoms relative to aura indicates that headache and other migraine symptoms commonly occur simultaneously with aura. In a large study of the efficacy of transcranial magnetic stimulation as a treatment for migraine, migraine symptoms were prospectively recorded in relation to aura symptoms. Analysis of the data generated by this study indicates that the majority of patients reported headache, photophobia, and phonophobia within the same initial time window (15 minutes) that they began to experience aura symptoms. This result suggests that the pain and associated symptoms of migraine are caused by parallel mechanisms occurring at the same time as the aura rather than as a direct downstream consequence of the aura.
Previous single-photon emission computed tomography (CT), PET, CT, and magnetic resonance imaging (MRI) studies have shown dramatic changes in blood flow, metabolism, and contrast enhancement during migraine aura, including prolonged aura and the aura of hemiplegic migraine.[30-45] Several recent studies added to the constantly expanding number of these case reports.[46-53] To briefly summarize, these studies show that a variety of physiological responses can be observed during migraine aura, including hypoperfusion, hyperperfusion, hypometabolism, vasogenic edema, and breakdown of the blood–brain barrier. Clinical experience, as well as a few cases reports, shows that many patients have normal standard CT or MRI studies during migraine aura, indicating that some of these changes may be the exception rather the rule and may only occur in cases of prolonged aura or aura associated with hemiplegic migraine.
Although both increased and decreased perfusion may occur with aura,[47, 51] most imaging studies indicate that hypoperfusion occurs first. A study by Hansen et al demonstrated hypoperfusion in 2 hemiplegic migraine patients who were imaged within an hour of onset of aura symptoms, consistent with the original studies of Olesen and colleagues showing that the onset of migraine aura is associated with hypoperfusion, which is then followed by hyperperfusion (still during the aura phase). Iizuka et al performed a series of imaging studies of multiple attacks of prolonged aura in patients with familial hemiplegic migraine type 2 (with a novel mutation in the ATP1A2 gene). In these patients, migraine aura lasted 4-12 days. Neuroimaging studies revealed both hyperperfusion and hypoperfusion in the same patients; hyperperfusion occurred in the first 4 days, affecting the hemisphere contralateral to hemiplegia and in some cases associated with middle cerebral artery dilation. Interestingly, hypoperfusion was observed both in the affected hemisphere, as well as in the opposite hemisphere not referable to the hemiplegic symptoms. All changes were fully reversible. The occurrence of asymptomatic hypoperfusion contralateral to the “affected” hemisphere is particularly intriguing and emphasizes the fact that extensive changes in perfusion may occur in the setting of migraine that are clinically silent.
Although the significant variability in the perfusion responses observed in imaging studies may be due to their timing in the course of the attack, another explanation may be that there is a disruption in the normal coupling between brain activity and blood flow during migraine aura. Hypometabolism in the presence of normal blood flow[31, 46, 55] has been demonstrated. Transcranial doppler techniques have also shown impaired vascular reactivity during a migraine aura or headache. A similar neurovascular “uncoupling” has been reported in association with CSD in the setting of human brain injury. Surface electrode recordings in patients in the intensive care unit for subarachnoid hemorrhage, stroke, and traumatic brain injury have shown that some CSD events may be associated with an increase in blood flow, whereas others are associated with a reduction in blood flow that may lead to worsening of the primary injury. These human studies are supported by animal studies that show that CSD can be associated with a profound “neurovascular uncoupling,” in which there is a disruption in the usual relationship between brain activity and blood flow. It is uncertain whether these vascular changes play any primary role in generating aura or headache symptoms, or rather are simply a secondary consequence of other more primary processes.
The Headache Phase
Imaging studies in evoked migraine have continued to provide interesting results regarding vascular changes that occur during a migraine attack. Schoonman and colleagues used magnetic resonance angiogram approaches to show that migraine headache triggered by NTG was not correlated with any significant dilation of the cerebral or meningeal arteries, and Nagata et al similarly reported no dilation of the middle meningeal artery during a spontaneous migraine attack. By contrast, Asghar et al found that both the middle meningeal and middle cerebral arteries were slightly dilated on the same side as migraine headache evoked by infusion of calcitonin gene-related peptide (CGRP) and that administration of sumatriptan resulted in amelioration of the headache as well as contraction of the middle meningeal but not the middle cerebral arteries. These different findings may be the result of different techniques or a reflection of the different triggers that were used to evoke migraine. Even if vasodilation does consistently occur with migraine headache, however, there is still no direct evidence that this dilation plays any role as a cause of pain rather than simply representing a parallel consequence of the same pathophysiological mechanisms that are causing headache. Indeed, there are drugs that cause significant cerebral vasodilation without causing headache, and conversely, some drugs that do not cause significant vasodilation do evoke migraine. Regardless of the relationship between vascular changes and pain, however, study of these vascular changes represents a tool for increasing our understanding of migraine pathophysiology.
Exogenous Migraine Triggers
Demonstrated migraine triggers include the nitric oxide donor glyceryl trinitrate, CGRP, pituitary adenylate cyclase-activating polypeptide (PACAP), sildenafil, and prostaglandins I2 and E2.[61, 62] The ability of endogenously occurring brain signaling molecules or modulators of their signaling pathways to evoke migraine when delivered intravenously provides strong indirect evidence for their potential role in the disorder. With the exception of prostacyclin I and prostaglandin E2 (PGE), however, each of these triggers evokes an immediate mild headache but a migraine only after a delay of a few hours.[61, 62] It is therefore unlikely that the migraine headache is a direct effect of the exogenously administered nitric oxide, CGRP, or PACAP but rather is an indirect response to these compounds. One explanation is that exogenous administration of these migraine triggers may push a finely regulated system out of balance, setting in motion a pendulum of neurochemical changes that eventually swings back into a full-blown migraine attack. Following this line of reasoning, the exogenous migraine triggers could evoke a compensatory release of neurotransmitters or neuromodulators like dopamine, epinephrine, acetylcholine, or adenosine triphosphate to name a few, which in turn would eventually lead to the downstream endogenous release of the CGRP, nitric oxide, and PACAP. This concept is supported by the observation that NTG provokes premonitory symptoms prior to headache, which, as discussed earlier, may involve dopaminergic mechanisms. Here again, investigation of the brain changes that are occurring in the hours leading up to the headache may be highly informative.
In the case of PGE, the occurrence of migraine-like headache during the infusion indicates that this compound is directly triggering migraine, and its mechanism of action may therefore be downstream from those of CGRP, nitric oxide, or PACAP.
Regardless of whether these triggers evoke migraine directly or indirectly, each represents an individual potential therapeutic target. In the case of CGRP, there is now strong evidence that CGRP receptor antagonists are effective migraine therapies. New strategies for inhibiting the effects of CGRP are in development, as are nitric oxide synthase inhibitors, PACAP receptor antagonists, and novel prostanoid antagonists.
Migraine-Associated Sensory Sensitivity
As with the premonitory symptoms, there has been progress regarding the pathophysiology of other migraine symptoms, particularly the sensitivity to sensory stimuli that is commonly observed in migraine patients. The mechanisms of photophobia and the exacerbation of migraine pain by light has been the topic of important recent studies. Noseda and colleagues found that even in patients in whom rod and cone damage in the retina resulted in loss of image migraine pain was worsened by light even in blind patients in whom rod and cone damage had caused loss of image formation. These studies indicated that there are alternative pathways for transmitting a nociceptive response to light. They then used tract-tracing techniques in animals to demonstrate a direct pathway from retinal ganglion cells (primary transducers of non-imaging forming photoregulation) to a region of the posterior thalamus that also responds to stimulation of the dura. They propose this as a pathway for modulation of migraine pain by light. The identification of a novel retinothalamic mechanism highlights the existence of non-trigeminal pathways for modulation of migraine pain. Inputs via these pathways to regions of the brain involved in migraine may not only exacerbate headache but conversely have the potential to be exploited for therapeutic purposes.
Other studies indicate that photophobia is also associated with hyperexcitability of pathways involved in image formation. Denuelle and colleagues used H215O PET to study photophobia in migraineurs during spontaneous migraine attacks, after treatment with sumatriptan, and between attacks. They found that a low intensity of light activated the visual cortex during migraine attacks and after resolution of headache with sumatriptan treatment but not during the attack-free interval. They concluded from these observations that visual cortical excitability can occur independently of trigeminal nociception and suggested that this excitability is driven by brainstem nuclei.
Studies by Burstein and colleagues examined changes in brain activity associated with mechanical and thermal allodynia (the perception of ordinarily innocuous stimuli as uncomfortable). They used functional MRI blood oxygenation level-dependent signals to show that during a migraine attack in which patients had allodynia extending beyond the head, mechanical or thermal stimulation of the hand produced greater posterior thalamic responses than those evoked by the same stimulation between attacks. These studies implicate sensitization of the thalamus as an important mediator of the sensory sensitivity associated with migraine.
Photophobia, phonophobia, and allodynia are generally considered to be symptoms of central sensitization, which refers to increased activity of brain or spinal cord sensory processing pathways that lead to increased or altered sensory perception. In traditional pain models, this occurs as a secondary consequence of peripheral nociceptive input. The early occurrence of sensory sensitivity (particularly photophobia and phonophobia) during a migraine attack and the imaging evidence that these phenomena can be independent of trigeminal input are not consistent with this model. Rather, these studies provide further evidence that central sensitization is a primary event in a migraine attack that does not require pain from the periphery as an initiating trigger.
The dramatic efficacy of the triptans, selective agonists at 5-hydroxytryptamine (5HT) 1B, D, and F receptors, has clearly established a critical role for serotonin in the inhibition of an acute migraine attack. However, the 5HT receptor subtypes responsible for the therapeutic effects of the triptans and the anatomical location of their site of therapeutic action remain uncertain. Clinical studies of new migraine therapeutic agents with more selective pharmacology are shedding some new light onto these questions. For example, a clinical trial of lasmiditan, a non-triptan-selective 5HT1F receptor agonist, have yielded results that have significant implications regarding migraine pathophysiology.[66, 67] Both intravenous and oral preparations of lasmiditan have shown efficacy as an acute migraine therapy in pilot studies, and neither preparation caused the chest, neck, or jaw tightness, or heaviness that are commonly observed with triptans.[66, 67] Lasmiditan did, however, cause dizziness, vertigo, and fatigue in a dose-dependent fashion. These results confirm that selective activation of 5HT1F receptors has therapeutic effects in patients with migraine. 5HT1F receptors are not widely expressed in the vasculature, and activation of these receptors does not have vascular effects in vitro. The effects of lasmiditan therefore provide strong evidence for a nonvascular mechanism for migraine-specific acute treatment. The results of this study also suggest that the site of therapeutic action of lasmiditan is central rather than peripheral. Although 5HT1F receptors are expressed in the trigeminal ganglion and therefore, a peripheral site of action cannot be excluded, the CNS symptoms caused by lasmiditan provide strong evidence that lasmiditan is indeed reaching the brain and suggest that it is treating migraine within the brain. A central mechanism of action is supported by animal studies showing that 5HT1B, 1D, and 1F receptors are all expressed in the brain and have antinociceptive functions at multiple central sites.
Recent imaging studies have also provided new information regarding serotonin receptor involvement in a migraine attack. A PET study using [18F]2′-methoxyphenyl-(N-2′-pyridinyl)-p-fluoro-benzamidoethyipiperazine (MPPF), a ligand that specifically binds the 5HT1A receptors, found increased 5HT1A receptor availability in the pons, orbitofrontal cortex, precentral gyrus, and temporal pole during olfactory-triggered migraine attacks. These studies indicate reduced serotonergic tone in general during a migraine attack but also raise the possibility that 5HT1A receptors could be playing more of a role during a migraine than has been previously appreciated.
The Postdrome Phase
Despite the fact that it can be equally or more disabling than the phases that precede it, relatively few studies have systematically examined the “postdrome” phase of a migraine attack, ie, the symptoms occurring for hours to days after resolution of headache.[6, 8, 9] These studies indicate that postdromal symptoms occur in the majority of patients, with tiredness, weakness, cognitive difficulties, and mood change being the most common. Other symptoms include residual head pain, lightheadedness, and gastrointestinal symptoms. Some of these symptoms become apparent upon treatment of headache, commonly leading patients to believe that they are an adverse effect of the acute medication when in fact they are part of the attack. Also, as mentioned previously, there is some overlap between premonitory symptoms and postdromal symptoms, raising the possibility that the postdromal symptoms have been present throughout the attack but simply overshadowed by headache, nausea, or aura symptoms. Imaging studies provide some clues into the postdrome phase. Early studies by Olesen et al indicated that hyperperfusion may outlast the headache in patients with migraine with aura. Conversely, a recent PET study by Denuelle and colleagues found that there was bilateral posterior cortical hypoperfusion in migraine without aura, and this hypoperfusion persisted after successful treatment of headache with sumatriptan. This same group found that midbrain and hypothalamic activation persisted after headache relief, as did increased light-induced activation of the visual cortex.[22, 73] These studies are in line with previous PET studies showing that activation of the dorsolateral pons in NTG-triggered migraine persists after amelioration of headache with sumatriptan. These functional imaging studies clearly demonstrate that there are persistent changes in the activity of multiple brain regions for hours after cessation of headache.
Ongoing quantitative clinical observations, imaging studies, electrophysiological studies, and therapeutic clinical trials continue to provide important new information regarding how a migraine starts and progresses. Although the majority of research regarding migraine attacks has focused on the aura and headache phases, increased attention to the premonitory and postdromal phases may also yield critically important information. A comprehensive approach of migraine demands appreciation of all of the phases of an attack, and the development of future therapies may hinge not only on an understanding of what goes on in the brain during a headache but also what happens in the hours before it begins and after it ends.