Discussion
The purpose of the present study was to extend our electrophysiological investigations of visual cortex reactivity in migraine by searching for differences between two distinct phenotypes of migraine with aura.
First, we confirm the previous finding that during continuous stimulation amplitude of the VEP N1–P1 component, but not of P1–N2, does not habituate over sequential blocks of averaged responses in migraine with aura patients between attacks while it does so in healthy volunteers [15]. An additional novel finding is that, relative to HV, VEP N1–P1 habituation is deficient both in migraine with pure visual aura (MA) and in patients with complex aura (MA+).
A second striking result is that the amplitude of visual responses differs between patients having pure visual aura and those with complex auras. MA+ patients consistently have greater N1–P1 VEP amplitudes than MA patients. Contrary to MA+, MA patients do not differ from healthy volunteers in VEP N1–P1 and P1–N2 block amplitudes, although they have reduced habituation over the 6 sequential blocks of 100 averaged VEP responses.
To the best of our knowledge, this is the first study of visual evoked responses in patients with different migraine aura phenotypes. It identifies within the migraine spectrum a subgroup of patients with complex neurological auras in whom excitability of the visual cortex appears genuinely increased, as evidenced by an increased VEP N1–P1 amplitude and decreased habituation. Previous VEP studies have yielded conflicting results in groups of migraine with aura (MwA) patients without phenotype distinction. In some reports the grand-average of VEP N1–P1 and/or P1–N2 amplitudes was found greater in MwA patients than in controls [30–35] and/or in migraine without aura (MO) patients [31, 36, 37]. The amplitude of steady-state VEP harmonics was also larger in MwA than in MO or HV [38]. In other studies, on the contrary, VEP amplitudes were found reduced in MwA [39], even when compared to MO [40]. Most often, VEP amplitudes in MwA were reported to be in the normal range [13–15, 41–44]. Our finding of low or normal visual cortex excitability in patients with pure visual auras, which is similar to migraine without aura patients [45] but contrasts with increased VEP amplitude and deficient habituation in patients with complex auras, may help to explain some of the abovementioned discrepant results. In fact, pooling patients with different migraine phenotypes (MO and MwA or different MA subgroups) in different proportions increases the variance of VEP results between studies. This probably fueled in part the controversy about the presence or not of interictal cortical hyperexcitability in migraine. In a previous review paper, we reasoned that in its strict physiological definition of a stimulus–response curve, the cortex would be hyperexcitable if it generates a response to a subliminal stimulus and/or if its response to a supraliminal stimulus is increased in amplitude. Because in most previous studies VEP amplitude in MO patients and, according to the present results, also in MA patients, increases during stimulus repetition, while remaining within a normal range (see Fig. 2), we proposed to abandon the general term “hyperexcitability” in favour of “hyperresponsivity” to characterise the response pattern of the migrainous brain to repeated stimulations [46]. As shown here, the functional abnormality is clearly different in MA+ patients in whom the initial VEP amplitude to a low number of stimuli is increased compared to both HV and MA, indicating that their visual cortex is genuinely hyperexcitable.
From a pathophysiological point of view, it is interesting to compare MA+ and chronic migraine that is also thought to be associated with true cortical hyperexcitability. The evidence in chronic migraine comes from studies of somatosensory evoked potentials [47] and magnetoencephalographic visual evoked responses [48]. The difference with MA+ is that in the latter VEP amplitude was increased in virtually all blocks of averagings and habituation was deficient over 6 blocks, while in chronic migraine only the 1st block of averaged visual or somatosensory responses was increased in amplitude, but not the subsequent blocks, leaving habituation normal. The electrophysiological pattern in migraine with complex neurological auras may therefore suggest that the visual cortex is locked in a state of persistent hyperexcitability.
The pathophysiological determinants of different aura phenotypes and related differences in interictal visual evoked potential profiles remain speculative. Cortical spreading depression (CSD) is thought to be the pathophysiological substrate of the migraine aura. CSD is an electrochemical wave that usually starts in the posterior regions of the brain and spreads anteriorly at approx. 3 mm/min, accompanied by biphasic cerebral blood flow changes [4]. In several brain imaging studies performed during attacks, though not in all, [49, 50] the vascular and metabolic changes accompanying the migraine aura spread more anteriorly in patients with complex neurological symptoms and hemiplegia than in those with only visual disturbances. The recovery from CSD depends largely on intact neurovascular coupling to match the increased energy demand and to restore ion gradients via the Na+/K+ ATPase pump [51]. The distance, to which CSD spreads during MwA attacks, and thus the clinical phenotype of the aura, depends on the balance between factors that predispose the brain to CSD and others that inhibit CSD and allow the parenchyma to recover.
The neurovascular tone is modulated by local factors such as oxygen availability or lactate concentrations, and by subcortical monoaminergic projections [52, 53]. During continuous visual stimulation neurovascular coupling is impaired in migraine patients between attacks, especially in migraine with aura [34, 54, 55]. There is also circumstantial evidence from biochemical and functional neuroimaging studies that monoaminergic, in particular serotonergic, transmission from the brainstem to the thalamus and cortex is altered in migraine [56]. Finally, convergent data from various laboratories have shown that the mitochondrial energy reserve and ATP levels are significantly reduced in the brain of migraineurs between attacks [27, 28]. Based on these biochemical and functional data, we have proposed that migraine is characterized interictally by a cycling dysregulation of the serotoninergic control of thalamo-cortical activity that causes varying degrees of cortical hyperresponsivity and thus increased energy demands, which, under the influence of triggering or aggravating factors, may disrupt homeostasis and lead to an attack [14, 57].
Several studies suggest that the abnormalities of energy metabolism could be more pronounced in migraine with complex neurological aura. The phosphocreatine/phosphate (PCr/Pi) ratio, a marker of the brain’s energy reserve, differed significantly between patients with different aura phenotypes and was lowest in those with more complex auras [58]. In a 1H-MR-spectroscopy study [20] MA+ patients had a significant increase of lactate in the visual cortex during sustained visual stimulation, while this was not the case in HV and MA patients. Variants in the mitochondrial DNA, such as those that distinguish responders from non-responders to preventive anti-migraine treatment with riboflavin [59], could play a role in the metabolic differences between aura phenotypes.
That genetic load can influence CSD patterns and severity is evidenced by the studies of the “knock-in” mice wearing CACNA1A [60] or ATP1A2 [61] mutations found respectively in familial hemiplegic migraine (FHM) type 1 and 2. In FHM1 mice having the S218L mutation that causes a more severe clinical phenotype in patients, CSD are more frequent and more spread out (up to the striatum) than in mice with the R192Q mutation. As mentioned, the common form of migraine with aura is not associated with the mutations found in FHM, but merely with common variants in a number of loci identified on genome-wide association studies (GWAS) that are seemingly not much different from those found in migraine without aura [62]. It remains to be determined whether the combination of such common genetic variants and their association with mitochondrial DNA variants may influence the clinical migraine phenotype, including that of the aura.
One can only speculate on the possible relation between the ictal phenomena, i.e. CSD and its spreading, and the VEP abnormalities found interictally. We know of only one study in photosensitive subjects with a photo-paroxysmal response to intermittent photic stimulation where increased VEP amplitude was correlated with spread of the paroxysmal EEG activity to more anterior brain areas [63]. In photo-paroxysmal responses and photically induced seizures, this could be the electrophysiological correlate of increased functional connectivity between occipital and parieto-temporo-frontal networks under the control of the thalamus [64–66]. A recent study showing in animals that CSD can activate the thalamic reticular nuclei that controls the flow of sensory information to the cortex, is therefore of major interest [67]. Translated to migraine pathophysiology, one may hypothesize that repeated thalamic activation by CSD could worsen the interictal impairment of thalamic/thalamocortical activity in migraine with complex auras [14, 68–71]. Studies correlating aura frequency and duration of the disorder with thalamic/thalamocortical activity in MwA are necessary to test this hypothesis.
Whatever the possible relation between ictal CSD and interictal VEP might be, the pathophysiological mechanisms underlying VEP habituation are not permanently influenced by the ictal phenomena, even in MA+ patients. In the latter, indeed, like in MwA patients overall and in migraine without aura [15], the VEP habituation deficit is obvious between attacks. Moreover, in MA+, but not in MA patients, it worsens progressively with time elapsed since the last migraine attack and decreases with increased attack frequency; in other words, VEP habituation increases with proximity to an attack. To explain this difference between MA+ and MA, we speculate that MA+ patients are carrying the most pronounced genetic load predisposing them to more prominent pathophysiological dysfunctions. For instance, we intend to explore the possibility that MA+ is the migraine with aura phenotype with the most pronounced deficit of short-range lateral inhibition within the visual cortex, an abnormality that we also found directly related to the distance from the last attack in a previous study of a mixed group of migraine with and without aura patients [15]. Taken together with our present results, this would indicate that the inhibitory performance and habituation with stimulus repetition decreases with the distance from the last migraine attack. A psychophysical study using visual metacontrast masking test, found a similar correlation between inhibitory processes and the number of days elapsed since the last attack [72]. The biochemical correlate of impaired inhibitory mechanisms could be lactate-induced downregulation of GABA activity in the visual cortex. As mentioned, in MA+ lactate levels increase in the occipital cortex during visual stimulation [20] there is emerging evidence that lactate, besides its role as energy substrate, has a concentration-dependent downregulating effect on GABAergic neurotransmission [22].
As other neurophysiological studies, ours has some methodological shortcomings. For instance, the investigators were blinded during off-line analyses of VEP data, as applied in previous studies by independent groups [15, 73], but not to diagnosis during the recording session, although this is probably only a minor risk for bias. As a matter of fact, in a clinical setting it is quasi impossible to totally blind a neurophysiological study. Even in VEP studies that found no abnormalities in migraineurs and were claimed to be blinded to diagnosis [12, 74, 75], blinding was not perfect for various reasons according to the reported methodology: 1) the neurophysiologist knew which set of responses belonged to each of the 6 blocks of averagings [74], which allows a selection bias in favour of low amplitude responses and thus normal habituation [76]; 2) the neurophysiologist was not blinded to check size [74], to which VEP amplitudes are quite sensitive [77]; 3) after each recording, the investigators guessed the correct diagnosis in more than half of subjects [12]; 4) although the investigators were blinded to diagnosis during the first examination, they were not during the 3 subsequent recording sessions [75].
Though we refer herein to habituation as the common feature of responses to any type of repeated sensory stimuli and to its classical definition of “a behavioral response decrement that results from repeated stimulation and that does not involve sensory adaptation/sensory fatigue or motor fatigue” [78, 79], we cannot totally exclude that changes in the level of attention and contrast pattern adaptation may have influenced our results. This is nonetheless unlikely for the following reasons. In previous studies VEP amplitudes after full field stimulation were not significantly influenced by attention and reaction time task conditions [80, 81]. Moreover, the effects of contrast adaptation on the P1 peak are small (peak time shift approx. 3 ms, amplitude unchanged) and require to take place stimulations lasting about 25 min [82, 83], contrasting with a 3 min 20 sec duration of a VEP recording session in our study.
We also are aware that our samples are relatively small and that clinical correlations are retrospective. Further studies are needed to repeat the analysis in a larger clinical sample with various migraine phenotypes and with a longitudinal, prospective follow-up of patients, allowing to record them during attacks as well as at different time points between attacks.