CSD could explain some features of migraine but following unresolved points needs to be addressed.
Emergence of visual and somatosensory auras simultaneously or sensory auras that begin in the visual field and continue with somatosensory and language dysfunctions cannot be justified by the propagation of canonical CSD in the human cerebral cortex
In great majority of patients presenting with both visual and somatosensory aura, visual disturbances appeared first and paresthesias appeared later. Recent records of 162 auras prospectively collected [
38] showed that single aura symptom was visual in 97% of the patients, while auras consisting of two symptoms were visual followed by somatosensory in 81% of the patients. The second symptom started concurrently with the first aura symptom in 35% or on the course of the first symptom in 37% of the patients. Only in 5% of the patients, did the second aura started after the first symptom ceased [
38]. Regarding the duration, most of the visual aura symptoms (≈60%) lasted between 11 and 30 min [
38]. Manzoni and colleagues reported that the average duration of aura was less than 30 min in 76% of the cases [
39]. Dysphasia symptom, always followed visual and/or sensory auras, generally lasted less than 10 min [
38]. For instance, hemianopsia and aphasic aura characterized by anomia and semantic paraphasia lasted 10 min in migraine patients [
59].
Furthermore, considering the distance that CSD has to travel with the time interval between aura symptoms the issue gets more complex. Even in a flattened macaque brain the distance from V1 to somatosensory cortex is 100–120 mm [
69]. The distance between occipital pole and central sulcus in surface reconstructed/flattened human cortex is approximately 250 mm [
70]. A wave of CSD from visual areas to sensorimotor cortex or language areas should march faster than 10–25 mm/min. Such a high speed was never detected in any species. The usual propagation rate is between 2 and 6 mm/min for any CSD in mammalian cerebral cortex, which would take approximately 42 to 125 min for CSD to travel such a distance. The concomitant manifestation of visual and somatosensory aura symptoms is inconsistent with march, although simultaneous manifestations of CSD at two non-contiguous sites remains possible.
Additionally, the somatosensory symptoms typically manifest as cheiro-oral syndrome (ipsilateral fingertips and per oral distribution). Cheiro-oral somatosensory aura symptoms start as tingling sensation or pins and needles from acral part of digits and propagate to wrist, and involve lips and tongue. In the cortical sensory homonculus, each digit is oriented in a caudo-rostral direction with 10–20 mm apart, where the acral part lies rostrally and there is 20–30 mm distance between face area and digits [
71,
72]. Prospective studies also showed that somatosensory aura follows and/or accompanies visual aura in great majority of attacks [
38]. Accordingly, the involvement of distal parts of the fingers and perioral area, tongue and palate while skipping the head, forehead, eyes, and nose in the cortical sensory homonculus seems impossible, as CSD waves are considered to travel rostrally from occipital cortex in a continuous manner. This is not compatible with hand and finger somatotopy in the cortical homonculus [
72].
Canonical CSD propagating through the cortex would also be expected to invade other areas between visual and somatosensory cortex and yield symptoms such as acalculia, agraphia, finger agnosia, and right–left disorientation, contraversive eye deviation, buzzing noise in regard to data from cortical stimulation studies in human [
73]. The majority of somatosensory aura last 11–20 min [
38]. In the light of the above data, somatosensory symptoms starting simultaneously with visual aura symptoms or while visual aura symptoms still continue, do not match the known properties of canonical CSD phenomenon, unless they start in more than one place at the same time, triggered by other events. For example, decreased cortical noradrenergic signaling from locus ceruleus (LC) was shown to reduce the threshold for CSD induction [
74]. Considering the pontine dysfunction in the premonitory phase and noradrerenergic projections to the entire brain, LC involvement could provide explanation for some clinical features.
If a SD like event travels in a restricted manner along a gyrus or sulcus, how does a CSD in the visual cortex manifest as a visual aura and travel to the somatosensory cortex causing paresthesia within 20 min? Or how do visual and somatosensory cortical dysfunction occur simultaneously?
CSD is limited to a few gyri in the primate brain however for multiple aura symptoms to occur sequentially, CSD should propagate over numerous sulci along the human cerebral cortex. Simultaneous occurrence of visual and somatosensory auras in the majority of patients needs to be explained with a process other than spreading depolarization waves continuously traversing along the cerebral cortex.
Canonical CSD in human cerebral cortex during migraine aura has not been proved electrophysiologically
DC shift, suppressed ECoG/ electrical silence and consequent loss of neurological function, lasting minutes, within the involved cerebral cortical area are key features of CSD. CSD is hardly induced in healthy, metabolically uncompromised primate brain [
75‐
77]. This is opposite to the experimental findings in lissencephalic brains where CSD can easily be induced by many ways and invades the whole hemisphere. Many patients with awake cortical mapping reported no dysfunction lasting minutes resembling CSD. CSD or its functional consequences were not reported in those awake neurosurgical procedures [
76]. However even electrophysiological mapping studies of the human cortex in epileptic /tumor surgery did not reveal any phenomenon resembling canonical CSD seen in rodent brain [
78]. McLachlan reported that he could not elicit CSD in the human cerebral cortex of 23 patients using chemical, mechanical or electrical stimulation including intracortical KCl injection while similar approaches easily induced CSD waves in the rat brain [
75]. In humans, SD waves on the other hand are easily recognized in damaged cerebral cortex due to hemorrhage, trauma or severe ischemia [
6]. Even if CSD is induced, its propagation is easily blocked or limited to a few gyri in primates and humans [
36,
77]. Indeed, mapping of a migraine aura onto the visual cortex suggests that the event underlying visual aura can travel along a single gyrus or sulcus [
36]. Likewise, electrophysiological hallmarks of canonical CSD have not been detected convincingly in migraine.
MEG DC shifts were reported in migraine brain while they were also detected in the control group with lower amplitudes [
79,
80]. However it was barely convincing that those changes were due to an underlying CSD in the occipital cortex. MEG DC shifts were not specified to the aura period but were also observed during headache and interictal periods [
79]. The aura symptoms and their duration, lateralization and retinotopic correlation were not defined in the subjects and spreading nature were not shown [
79,
80].
MEG DC shifts were widely distributed in both hemispheres particularly in occipito- parieto-temporal cortices in migraine patients. During the spontaneous attacks in migraine with aura patients, greater activation was detected in the occipito-parieto-temporal association cortices instead of primary visual cortex, though, the visual stimulation induced activation was clearly seen in the occipital cortex [
80]. MEG DC shifts were more pronounced in the occipito-parietal-temporal, higher order association cortices while occipital cortex has very low activation even during the early stages of aura/HA [
80]. Those changes therefore cannot indicate an underlying CSD.
Detecting MEG DC shifts alone cannot be taken as a proof of CSD, considering the fact that DC shifts in the MEG and /or EEG are also detected during cortical activation by auditory stimuli [
81], cognitive tasks, changes in vigilance states, changes in brain CO2 levels, shifts of attention, hyperventilation [
82], hemodynamic changes [
83,
84] sensory stimulation, finger movements [
85] or epileptic seizures [
86‐
88]. MEG DC shifts have a good localizing value for epileptic seizures and they are timely locked to neuronal activity in the higher frequencies. Infraslow oscillations in the DC range are believed to represent a slow modulation of gross cortical excitability [
89]. Additionally generation of infraslow oscillations likely involves thalamic networks and/or astrocytic calcium waves [
90,
91]. MEG DC shifts in the migraine brain could indicate cortical hyper-excitability and also be associated with these factors.
The failure to show electrical imprint of CSD in migraine patients could be related to the fact that CSD is limited to a relatively few number of gyri in human cortex under normal physiological conditions. The latter is somehow contradictory, when oligemia progressing almost in the whole hemisphere is accepted as a result of CSD in migraine brain.
The occurrence of aura and headache without any temporal relationship, or aura without any pain, is not compatible with conventional notion of CSD
If the CSD induced depolarization occurs only a single or a few gyri in human brain, the area of activated trigeminal perivascular nerve endings in the overlying dura mater would be inadequate to fire trigeminal neurons for the development of lateralized head pain.
In animal models nociceptive and vasoactive substances released during CSD diffuse to the dura mater and induce trigeminal nerve endings and lead to the activation of the trigeminal ganglion and trigeminal 2nd order pain neurons in the ipsilateral hemisphere. However, most of the migraine attacks occur without aura and aura may occur without headache in 10% of patients [
38,
44]. Furthermore, headache may even start before aura (9%) or emerge simultaneously with aura (14%) [
38]. A prospective study reported that in only 36% of the attacks, headache started with an interval after the end of aura [
38]. Prospective clinical studies also prove that aura and headache can occur any time and no temporal relationship exists between them [
36,
38]. CSD in the gyrencephalic cerebral cortex is limited to a few gyri and can only activate a very limited portion of trigeminal nociceptors in the dura mater and MMA, therefore headache development secondary to the aforementioned process is very unlikely.
CSD associated vascular reactivity changes do not match the rCBF/BOLD alterations detected in migraine patients
It is critical to note that CSD induced vascular reactivity has species differences. For example, initial rCBF decrease accompanies DC shift in mice, followed by hyperemia phase and post SD oligemia. On the other hand, focal hyperemia was recorded in primate CSD models while long-term hypoperfusion/or persistent hypoperfusion was not detected [
77]. Additionally, in vivo whole-cell recordings showed that the neuronal silencing after CSD was due to reduced spontaneous synaptic activity and AP firing. Significant reduction in the spontaneous excitatory and inhibitory postsynaptic potentials lasted more than 90 min [
13] yielding an imbalance towards inhibitory tone. Therefore, the hyperemia phase coincides with the neuronal silence in the cerebral cortex. In other words, if the hyperemia detected in the migraine brain is due to a canonical CSD, then it should be associated with suppressed neuronal activity and possibly negative symptoms. Moreover, if the initial depolarization phase is missed by EEG studies then the detected hyperemia phase should be associated with negative symptoms.
Lauritzen and colleagues reported clinical features and rCBF measurements in patients who developed migraine with aura attacks after the introduction of a catheter into the internal carotid artery. Spreading occipital oligemia observed in all patients with approximately at a rate of 2 mm/min did not cross the lateral or central sulcus but appeared in the frontal lobe as well. Remarkably hemispheric blood flow changes were not correlated with lateralization, localization and /or beginning of aura symptom,e.g. during the oligemia spreading through the temporal, parietal and frontal lobes, paresthesia was the only symptom, oligemia lasted quite longer than the focal neurological symptoms or the left hemispheric rCBF changes resulted in left sided focal symptoms [
45]. Therefore, spreading oligemia was concluded not to be associated with aura symptoms spatially and temporally and not to be the cause of focal symptoms in migraine aura [
45]. So, oligemia begins before the aura and outlasts the aura symptoms and has no resemblence with CSD except for the propagating speed of approximately 2 mm/min [
9,
45].
Later studies reported reduced cerebral blood flow, volume or perfusion in the symptomatic occipital cortex during visual aura [
46]. Blood flow changes were usually not detected in migraine patients without aura [
47], although Gelmers et al. reported focal oligemia in two migraine without aura patients [
92] and Woods et al. reported bilateral occipitotemporal oligemia in one migraine without aura patient [
48].
Apparent diffusion coefficient (ADC) decrease with DC shift reflects water movement during cellular depolarization. Transient regional hyperperfusion and post-CSD oligemia were also demonstrated by MRI in rats [
93]. ADC decline and diffusion-weighted imaging (DWI) changes reported during CSD in animal studies [
93,
94] were never detected in human studies even when rCBF was reduced to 53% during a visual aura or in FHM patients with prolonged neurological deficits [
46,
95]. Similarly, BBB remains intact during migraine attacks in contrast to the CSD experiments in rodents [
28,
96].
The majority of migraine patients exhibit positive visual and somatosensory symptoms [
38,
65,
97,
98]. Positive visual symptoms were reported in 89% of the migraine patients [
65]. Even though positive symptoms constitute the majority of aura symptoms, negative aura symptoms may also occur. Viana et al. reported [
38] that 67% of visual aura included at least one positive phenomena whereas 38% of visual aura included at least one negative phenomena and 14% of the patients had negative sensory symptoms. In another study [
99] blind spots were observed in 42% of the patients and tunnel vision was seen in 27%. Hartl et al. reported positive visual symptoms in 55.6% of the migraine patients while negative visual hallucinations were reported in 51.9% [
97]. However, CSD research in lissencephalic brain show that light induced neuronal activity or synaptic evoked activity is reduced in parallel to the blood flow decrease manifesting negative symptoms. Actually, during the positive symptoms experienced as scintillations, BOLD response in the visual cortex is reduced [
34,
49]. BOLD signal, an indirect measure of neuronal activation, reflects the excitatory and inhibitory post synaptic potential activity rather than spike activity. Increased synaptic activity leads to a positive BOLD response while decrease in local neural activity results in negative BOLD signal in the cerebral cortex [
100‐
102]. Neural activity decrease is usually related to the activity of inhibitory interneurons [
103] which is also compatible with the findings of whole cell recordings during CSD [
13].
BOLD response to visual stimulus was decreased bilaterally, pronounced in V1, in a patient who developed flickering visual symptoms on the right side and throbbing feeling in the right hand followed by pricking sensation in the right arm. This is another example of dissociated aura symptoms and BOLD response in migraine [
37]. Remarkably BOLD responses were increased in both hemispheres after sumatriptan, a drug that has no effect on CSD [
37].
Both oligemia and hyperemia were shown in FHM2 patients. Multifocal hypoperfusion was prominent within the first 19 h (in 71.4% of the attacks) when the aura symptoms and headache were both present, and hyperperfusion developed after 18 h when neurological deficit still persisted and the headache was resolved in 89% of the attacks. Multifocal hypoperfusion detected during acute phase of hemiplegic migraine with prolonged aura is not compatible with canonical CSD propagation in the human cerebral cortex [
88]. Cerebral metabolic rate of glucose (CMRGlu) increases with the DC shift in CSD experiments and returns to normal levels after about 15 min [
104]. In FHM2 attacks, increased cerebral glucose metabolism was detected by FDG-PET in the regions corresponding to the areas of hyperperfusion on day 4 while aura symptoms still persisted.
Hypoperfusion as detected by delayed rMTT and TTP and decreased rCBF were reported in 14/20 migraine with aura patients who admitted to ER with the diagnosis of stroke [
105]. Perfusion defects usually exceeding a single vascular territory, bilateral hypoperfusion (3 out of 14 patients) or whole hemispheric hypoperfusion resembling to severe stenosis of extracranial vessels (4 out of 14 patients) were detected in migraine patients. In a child case of confusional migraine attack with conjugate eye deviation, ASL CBF image revealed a reduction in rCBF in right occipito-parietal blood flow to 34 mL/min/100 g while rCBF was 210 mL/min/100 g on the contralateral occipito-parietal region [
106]. ASL image of a migraine attack with bilateral headache revealed hypoperfusion in bilateral thalamus and hypothalamus and hyperemia in bilateral frontal convexity [
107]. On the other hand MR imaging of 4 migraine patients during aura attacks did not reveal any abnormality in cerebral perfusion [
108]. In Woods’ case, during a migraine without aura attack, bilateral hypoperfusion started from occipital pole gradually invading more rostral areas to temporal cortex [
37,
48]. The PET images taken during a visual stimuli for 3 h showed an estimated blood flow decrease up to 40% which was not associated with any positive or negative neurological deficit or caused any aura symptoms on the way from occipital V1 to temporal pole [
48]. Remarkably, oligemia propagated easily through parietooccipital sulcus and other sulci without any problem.
Vascular reactivity changes as an indirect way to observe neuronal changes does not provide a rational explanation for the occurrence of CSD or as a cause of CSD. Hemispheric oligemia solely cannot be accountable for a migraine headache development. The perfusion changes during and following migraine attacks seem to be variable though the predominant feature in the aura phase is accepted as a hypoperfusion which may extend into the headache phase. Hyperperfusion is proposed to occur during headache (HA) phase, which is also not a consistent finding.
Original rCBF studies did not support the notion that oligemia was associated with aura, on the contrary, it was concluded that oligemia propagating from posterior to anterior accompanied headache [
9,
45,
48]. It seems that reduced blood flow alterations are not primarily responsible for aura symptoms. Neither hypoperfusion nor hyperperfusion are thoroughly correlated temporally or spatially with aura symptoms or headache. So the rCBF studies do not support a canonical CSD as an underlying cause and remain inconclusive for the development of migraine headache. The above arguments indicate a need for a more complex theory to rationalize those changes.