Pathophysiology by migraine phases

Premonitory. The premonitory phase involves nonpainful symptoms that occur days to hours prior to the headache phase. There is a wide range of reported prevalence of premonitory symptoms, from 9% to 88% (21). It is likely underreported due to patients being unaware of the connection of these seemingly unassociated symptoms with their headaches (21). Common symptoms include fatigue, nausea, irritability, neck stiffness, pallor, excessive yawning, food cravings, increased thirst or frequency of micturition, sensory sensitivities, concentration difficulties, and other cognitive defects (39; 09; 21; 35; 22). Children can display changes in mood and/or behavior, fatigue, pallor, and dark circles (35).

The most likely source of these symptoms is the hypothalamus, which is involved in mood, appetite, and energy (35). Functional imaging has found hypothalamic involvement during the premonitory phase and greater connectivity from the hypothalamus to pain and autonomic areas between headaches (31; 35; 22). Network mapping data support that hypothalamic activity is associated not only with inducing a migraine but also with maintaining migraine chronicity through atrophy of brain regions including the insula and hypothalamus, not unlike other chronic pain states (07).

Hypothalamic neurotransmitters have also been linked in migraine research, including during the premonitory phase; neuropeptide Y and dopamine (secreted through the hypothalamus) are likely involved in mediating some of these premonitory symptoms (21). For example, dopamine agonists can induce yawning and dopamine antagonists have demonstrated the ability to abort migraine attacks while still in the premonitory phase (21; 35). More research is needed to uncover the roles of neuropeptide Y and dopamine, along with other hypothalamic hormones and neurotransmitters (orexins, somatostatin, cholecystokinin) in mediating premonitory symptoms.

The locus coeruleus and raphe nucleus have also shown selective activation during the premonitory phase. As these regions are responsible for sensory stimuli modulation (eg, light and sound), they could be responsible for sensory sensitivities like phonophobia, photophobia, and osmophobia (sensitivities to sound, light, and smell, respectively) during the premonitory phase (13). These areas also have input into cerebral blood flow, cortical excitability, and pain networks and may be partially responsible for their alterations during migraines (13).

Aura. Aura is defined by the International Classification of Headache Disorders, Third Edition (ICHD-3) as “recurrent attacks, lasting minutes, of unilateral fully-reversible visual, sensory or other central nervous system symptoms that usually develop gradually and are usually followed by headache and associated migraine symptoms”. These can occur from many brain areas and are divided into the following types: visual, sensory, speech and/or language, motor, brainstem, and retinal (ICHD). These auras can occur with or without corresponding headaches and the headache can occur with or without a corresponding aura (35). The prevalence of aura in migraine patients ranges from 20% to 40% in population studies, with 90% or more of those patients reporting visual symptoms (09). These auras generally last less than an hour, with only a fifth of patients reporting longer durations (35).

Auras have been studied for much longer than many migraine phases. An early study done by Dr. Hubert Airy in 1870 occurred when he drew out his own visual auras and how they changed over time (28). Gorden Holmes and Aristides Leão individually discovered that there was a wave of spreading depolarization at approximately 3 mm/min, what is currently known as cortical spreading depression (26; 27; 25; 37). Later researchers put these 2 pieces together and hypothesized that the migraine scotomas were the conscious experience of the spreading depolarization, which is still considered true today (28).

Cortical spreading depression (CSD) can be triggered in animal models by multiple modalities (electrical stimulation, mechanical stimulation, potassium, endothelin, or ouabain) (11; 10). This trigger causes a depolarization with a shift of ions (increase in extracellular potassium, decrease in extracellular sodium, changes in chloride, protons, magnesium, and zinc) and a large release of neurotransmitters (for example glutamate and dopamine) (11; 10). This ion shift causes neuronal dysfunction resulting in a regional increase in blood flow (hyperemia) followed by a relative lack of blood flow (oligemia) facilitated by glutamate. Glutamate builds up resulting in a loss of membrane potential and a subsequent current shift, which releases vasoactive substances like nitric oxide. These substances are responsible for the increase in blood flow (hyperemia) to meet the increase in energy requirements. Those with visual auras have cortical spreading depression-caused vascular changes starting in the occipital pole, then propagating outwardly generally stopping by the central sulcus (24; 01; 35). Vascular changes caused by cortical spreading depression have been visualized using fMRI (utilizing BOLD signal changes), PET, flow studies, and MRI with arterial spin labeling (11; 09; 10; 08; 35). Hyperemia is followed by a period of oligemia with a gradual shift back to baseline blood flow (39; 11; 09; 10; 35). The prolonged phase of oligemia generally does not reach a threshold for ischemia but it has been reported to increase the ischemic area of a stroke when superimposed.

The cortical spreading depression activation of the headache phase of migraine is still under investigation. It has been previously suggested that cortical spreading depression can activate neurons in the trigeminovascular system via 2 distinct pathways: one involving molecule-activated meningeal nociceptors (pain pathway), and the other being an inhibitory pathway originating in the primary sensory cortex (18). Therefore, it would be possible for cortical spreading depressions to either activate or inhibit the trigeminal pain pathway depending on the anatomical location (18). Although not the only mechanism for pain activation (not all migraines have aura and not all auras have headache), cortical spreading depression is one of the many triggers of the headache phase (13; 23).

Headache. Migraine headache pain is multifactorial and involves genetic predisposition, prior experiences, and amount of pain mediators, among many others (35). The predominant “vascular theory” of migraine involving dilation of vessels leading to a vascular-derived pain has been largely debunked (17). Vascular changes occur but are secondary to an abnormal CNS (17; 39). Current theories involve the trigeminocervical complex and cortical spreading depression. Migraine pain is closely associated with activation of trigeminal sensory pathways and cortical spreading depression has been shown to activate the trigeminovascular system along with many other triggers (13).

Trigeminal activation (via inflammatory mediators and vasoactive peptides including substance P, neurokinin A, adenylate cyclase-activating polypeptide, and calcitonin gene-related peptide) is a key piece in understanding migraine pain and processing (18; 22). It starts with peripheral afferents (nociceptive, trigeminal ganglion nonmyelinated C-fibers, dural thinly myelinated Ad-fibers) that connect centrally to specific pain structures that contribute to migraine pain (35). These pain structures include the eye, meninges, nearby cerebral and pial vasculature, and the dural venus sinuses (13; 35). Those peripheral afferents also innervate the trigeminal nucleus caudalis, or the caudal part of the spinal trigeminal nucleus, which combined with upper cervical afferents, create the trigeminocervical complex. This pathway explains the characteristic pain of migraines: periorbital pain that distributes to frontotemporal regions followed by occipital/nuchal regions (13).

Second-order nociceptive neurons from the trigeminal nucleus caudalis then connect to the thalamus. The thalamus (specifically the medial nucleus of the posterior thalamus and ventroposteromedial thalamus) has bidirectional connections to numerous cortical and autonomic systems such as the hypothalamus, somatosensory cortex, insula, limbic system, and amygdala (13; 35). This complex pain matrix correlates with our knowledge of migraine impairments including cognitive, affective, autonomic, and otherwise.

Understanding the trigeminal pain pathway has led to exciting new treatments and advancements in identifying possible migraine biomarkers (03). For example, calcitonin gene-related peptide and adenylate cyclase-activating polypeptide are the primary neurotransmitters that activate the trigeminal pathway and lead to migraine pain (02; 03; 35; 40; 14). The neurotransmitter induced pain may be facilitated through a common pathway of ATP-sensitive potassium (KATP) channels, which, when opened using a drug called levcromakalim gave all test patients a migraine headache (02). Release of these neurotransmitters also causes nearby vascular dilation and mast-cell degranulation that further activates nociceptors in the meninges and vasculature, potentiating the pain (13). Direct blockage of calcitonin gene-related peptide has shown promise in preventing and aborting migraines and has led to new therapies: anti-calcitonin gene-related peptide monoclonal antibodies and calcitonin gene-related peptide receptor antagonists (03; 13; 02; 40; 14). Calcitonin gene-related peptide is also one of the possible migraine biomarkers currently being studied, along with many others including glutamate, serotonin, S100β, neurokinin A, substance P, vasoactive intestinal peptide (VIP), β-endorphin, and nerve growth factor (03; 38).

The hypothalamus is not only implicated in the premonitory phase but also in the headache phase. The hypothalamus has many connections to pain systems, including the spinal trigeminal nuclei (32). There are specific cells within the hypothalamus that have been shown to inhibit pain signals coming from the trigeminocervical complex, an effect that was blocked by a selective dopamine antagonist, eticlopride (32). In one study Schulte and colleagues followed a patient with migraines for an entire month with daily functional imaging (36). They found increasing hypothalamic activity in response to trigeminal nociceptive input in the 24 hours leading up to an attack as well as increased functional connectivity between the hypothalamus and spinal trigeminal nuclei during the premonitory and pain phases (32; 36). Imaging studies suggest that the anterior thalamus is associated with the premonitory phase and chronic migraine whereas the posterior hypothalamus is involved with the headache phase (36).

Postdrome. A common but relatively unstudied phase of migraine is known as the “postdrome” or colloquially the “migraine hangover”. The postdrome follows the resolution of the headache phase and has symptoms like fatigue, nausea, neck stiffness, difficulty concentrating, irritability, and light sensitivity (35). These symptoms can last an entire day following the attack (average duration 25.2 hours). Up to 94% of patients report at least 1 of these symptoms, with the predominant symptom being fatigue (04; 35). Functional imaging during the postdrome phase found alterations in the same neural networks as the premonitory phase. It is suggested that the whole brain is involved in the postdrome but especially the frontal lobes and hypothalamus (04). Studies have found reduction in blood flow during migraine episodes including the postdrome phase, which may be contributing to the symptoms.

A couple of theories have been brought forward to explain this effect. The reduction in blood flow may be caused by activation of brainstem nuclei (eg, locus coeruleus) during and after pain stimuli processing, causing activation of alpha-2 adrenoceptors and, therefore, leading to vasoconstriction (04; 35; 22). Cortical spreading depolarization as described above for migraine aura could also be implicated in the postdrome. The microvascular dilator calcitonin gene-related peptide is involved in brain circulation and nociception (along with many other functions) and may be a key aspect in cortical spreading depression and therefore the postdrome phase (03; 13; 35).

Chronic migraine. Chronic migraine is a relatively new diagnosable condition associated with frequent headaches, a large burden of disability, and medication overuse (33). These patients typically have over 15 headache days per month, with the predominant headache being migrainous (33). The trigeminal pain system can shift to what’s known as “chronic migraine” through a variety of factors including impaired modulation, peripheral sensitization, central sensitization, decreased threshold, and alterations in connectivity (35). First, it is believed that the descending modulatory pain pathways are mainly regulated by the periaqueductal gray, the locus coeruleus, and the rostral ventromedial medulla. Dysregulation can amplify preexisting pain signals leading to an increased pain state. Increased pain states therefore have increased activation of the beginning of the pain matrix, the afferent trigeminal ganglion neurons, called peripheral sensitization (35; 38). Increased first-order afferent activation causes increased sensitivity of second-order trigeminal nucleus caudalis neurons and therefore increased third-order thalamic neuron sensitivity, deemed central sensitization (35; 38). For an acute migraine, the threshold of nociception inputs at the trigeminal nucleus caudalis decreases while the receptive field for nociception increases, causing acute activation of migraine. In the chronic form, there is altered resting state connectivity (for example, in the amygdala and insula) as well as persistent interictal changes (35). Altered connectivity includes changes to make networks “more weakly integrated, less efficient, less central, and abnormally segregated” (12). Network mapping research also found a possible feedback loop with hyperactivity of the visual cortex and associated gray matter atrophy/volume loss, causing a decreased threshold for cortical spreading depression and, therefore, an increased chance of repeated attacks and chronification of migraine (07).

Triggers. A common theme of many migraine triggers are metabolic derangements including changes in hormones (emotional stress, menstruation), sleep, weather, diet, physical activity, and alcohol intake (38; 22). These can create an unbalanced energy equation leading to oxidative stress. White matter hyperintensities due to increased oxidative stress can be found in migraine patients and are associated with changes in oxidative biomarker levels (22). It is postulated that the body uses cortical spreading depression (CSD) and/or activation of the trigeminovascular system to balance the energy equation and maintain homeostasis (19). The validity of self-described triggers is debated; in one study only 3 of 27 patients (with a history of migraine with aura) had a migraine attack after exposure to their perceived trigger(s) (02).

Right to left shunt. A new and active area of migraine research is the connection between migraine and right to left shunting. It is still being determined whether there is a causal relationship or simply an increased coexistence (20; 29). Conditions include intracardiac shunting like a patent foramen ovale (PFO) or atrial septal defect (ASD) as well as extra-cardiac shunting like a pulmonary arteriovenous malformation (PAVM) (34). A causal relationship is supported by the still evolving data finding improved migraine symptoms with shunt correction (20).

Predominating theories about pathogenesis include: hypoxemia with an associated increase in hemoglobin and cardiac output (which can initiate CSD), an increase in trigger chemicals (like serotonin) that avoid pulmonary capillary filtration, and microemboli/microthrombi that are shunted to the brain from the venous circulation causing tiny infarctions (and therefore low blood flow and/or CSD) (34; 30). Migraine patients with right to left shunts may be responsible for a large proportion of those with migraine that have clinical or subclinical brain infarction (34).