Abstract
Between attacks, migraine with (MO) or without aura (MA) patients show deficient habituation of pattern-reversal visual evoked potentials (PR-VEP) and a strong intensity dependence of auditory evoked cortical potentials (IDAP). Clinical observations of migraine prodromes and previously published electrophysiological studies suggest that cortical information processing may vary in close temporal relationship to the attack. We studied PR-VEP and IDAP just before (11 MO pts), during (23 MO, 3 MA), 1 day following (27 MO, 1 MA) and 2 days following (14 MO) a migraine attack. The results were compared with a large group of MO patients recorded at a distance of at least 3 days from an attack (n = 66 for IDAP; n = 39 for VEP). Patients recorded the day before the attack had on average an habituation of −13.6 ± 20.5% (mean ±
Keywords
Introduction
Evoked and event-related potentials have been extensively studied in migraine patients during the interictal period and suggest that cortical information processing is abnormal compared to healthy volunteers. For instance, a strong intensity dependence of auditory evoked cortical potentials (IDAP) is found in migraine with (MA) or without aura (MO) patients (1) and might be a consequence of reduced central serotonergic transmission responsible for decreased cortical pre-activation levels (2). Amplitudes of averaged visual evoked potentials (VEP) with flash or pattern-reversal stimulation were higher in MA and MO than in healthy controls in a number of studies (3–8), but not in all (9–12). Amplitude changes during repeated visual stimulation differ between migraine patients and healthy volunteers. Habituation (i.e amplitude decrease), which can be observed in most healthy subjects, is deficient in migraineurs or even replaced by potentiation (i.e. amplitude increase) during pattern-reversal stimulation of short (13) and long durations (14). Lack of habituation in the attack interval also characterizes event-related potentials such as contingent negative variation (CNV) (15, 16), ‘novelty’ auditory P3a in a passive oddball paradigm (17) or visual P3 in an active oddball paradigm (18).
CNV was found to undergo marked changes in temporal relation to the attack. Its amplitude tends to normalize during the attack (15, 19) and to increase 1 and 2 days before the attack (20, 21). Evers et al. (22) have reported recently that latency habituation, i.e. shortening, of the visual event-related component P3 also normalizes during the attack. It is thus likely that part of the variability of evoked potential findings in migraine is due to the fact that in most previous studies the delay between the recordings and the next attack was not taken into consideration.
From the clinical point of view it is known that in about 20% of patients the migraine attack can be preceded by up to 24 h by so-called premonitory symptoms, such as changes in mood, behaviour, vigilance, appetite, bowel activity or fluid balance (23), which suggest that it is initiated long before occurrence of the aura or headache symptoms.
For these reasons we decided to search for changes of VEP habituation and IDAP in the peri-attack period.
Patients and methods
Subjects
Seventy-seven patients (14 males, 63 females, mean age: 34 years) were recruited from a specialized headache clinic and diagnosed according to the criteria of the International Headache Society (24). Sixty-nine patients had exclusively migraine without aura (MO) (code 1.1; mean disease duration: 13 years, mean attack frequency: four/month), four migraine with typical aura (MA) (code 1.2.1; disease duration: 15 years, attack frequency: six/month) and four had both types of attacks (disease duration: 5 years, attack frequency: four/month).
Eleven patients were studied the day before, 24 during, 28 the day after and 14 2 days after the attack (Table 1). The occurrence of an attack after the recordings was verified by telephone interviews. No patient received prophylactic anti-migraine therapy and acute treatment during an attack was not allowed before the recordings. Oral informed consent was obtained and the study was approved by the Ethics Committee of the Faculty of Medicine, University of Liège.
Number of patients in each study-group
A group of MO patients recorded at an interval of at least 3 days before and after an attack served as interictal controls: 67 patients for IDAP (60 females and seven males, mean age: 37 years, mean disease duration: 16 years, mean attack frequency: four/mth) and 37 for VEP (34 females and three males, mean age: 39 years, disease duration: 15 years, attack frequency: four/mth). We also studied a group of 23 healthy volunteers (HV) (20 females, three males; mean age: 35 years) recruited among hospital and laboratory staff.
Experimental design and recordings
Patients were examined in a quiet room with dimmed light (30 Lux). For VEP recordings subjects were seated 1 m in front of a television monitor (mean luminance 260 cd/m2, colour temperature 9500 Kelvin). Stimuli were presented as a checkerboard pattern of black and white squares (contrast 80%) subtending 1 degree, 8 min of arch at a reversal frequency of 3.1 Hz. With one eye patched, subjects were instructed to fix a point in the middle of the screen. Needle electrodes were inserted into the scalp in the midline over the occipital region 2.5 cm above the inion (Oz: active electrode) and over the frontal region (Fz: reference). The ground electrode was placed on the forearm. During uninterrupted stimulation sequential blocks of 50 responses were averaged for a total duration of 2 min using a Cadwell 8400 apparatus (band pass 1–100 Hz, analysis time 300 ms).
Auditory evoked cortical potentials (AEPs) were evoked by 1000 Hz tones(50 ms total duration, 10 ms rise and fall times) delivered binaurally through earphones at a random repetition rate of 0.53–0.61 Hz at four intensities(40, 50, 60 and 70 dB) above sensation level in a randomized order. The EEG was recorded with a needle electrode at Cz and referenced to linked mastoids. The amplifier system was the same Cadwell 8400 apparatus, filters were set at 1 Hz low- and 20 Hz high-cut. For each stimulus intensity 100 artefact-free responses were averaged over a 400-ms epoch.
Data analysis
For VEPs, the five blocks of 50 responses were analysed in terms of peak latencies and peak-to-peak amplitudes of the maximal negative and positive deflections determined by visual inspection: the N1 peak was defined as the most negative point between 60 and 90 msec post-stimulus, P1 as the most positive point following N1 between 80 and 120 msec post-stimulus. Habituation was expressed as percentage amplitude change between the 1st and 5th block.
The N1 (between 60 and 150 ms post-stimulus) and P2 (between 120 and 200 ms post-stimulus) components of the AEP were identified for each averaged recording of 100 responses. Peak-to-peak amplitude of N1-P2 was measured for each stimulus intensity and the linear amplitude/stimulus intensity function (ASF) slope was calculated and expressed in µV/10 dB.
IDAP and VEP differences between groups were analysed with
Contrast analysis of
Results
In all recordings N1 and P1 VEP components as well as N1 and P2 components of AEPs were clearly identified. There were no significant latency nor mean amplitude differences in VEP or AEP between groups.
VEP habituation
In patients examined the day before an attack (n = 8), VEP amplitudes showed a clear-cut habituation between 5th and 1st block averaging −13.6 ± 20.5% (mean ±
Habituation (mean ±
Intensity dependence of AEPs
A comparable pattern of changes was observed for IDAP. The day before the attack (n = 11), AEP ASF slopes tended to be flat (0.38 ± 1.06 µV/10 dB) (Fig. 2), which was significantly different compared to the mean ASF slope of the group of 66 MO patients recorded between attacks (P = 0.02). During the attack (n = 24), ASF slopes were also significantly flatter (0.32 ± 1.44 µV/10 dB) than in the interictal period (P = 0.002). One day after the attack (n = 28), ASF slopes were steeper (0.87 ± 1.39 µV/10 dB) and not different from those computed on interictal recordings (P = 0.16). On day 2 after the attack (n = 14) ASF slopes had even more increased (1.02 ± 1.17 µV/10 dB) reaching levels observed between attacks (P = 0.67).
Amplitude stimulus function slope (in µV/10 dB) of auditory evoked cortical potentials (mean ±
Discussion
This study indicates that cortical processing of sensory information as measured by VEP habituation and intensity dependence of AEPs tends to normalize just before and during a migraine attack. Abnormal interictal patterns resume by the 2nd day following an attack. Because it is not easy to motivate large numbers of patients for repeated longitudinal studies, we chose to collect recordings from a number of patients who consulted our headache clinic and were at different time points in relation to the attack. Our results thus need to be interpreted taking into account that the mean data were obtained from different patients. Inter-individual variability, which is well known for visual and auditory cortical evoked potentials (25), is probably also the major reason for the large standard deviations at the various time points and for the lack of statistical significance between some of them.
Notwithstanding these reservations they are in line with previous studies of event-related potentials, such as CNV (19, 26), which decreases in amplitude, i.e. normalizes, just before and during attack. The same authors have described cyclic changes of CNV with the highest amplitude recorded a day before the attack and the lowest during the attack (21). The apparent discrepancy in timing between increased CNV amplitude occurring 1 day before the attack in Kropp & Gerber's study and reduction of IDAP as well as normalization of VEP habituation found in our study at the same time period could have several explanations. Our patients may have been closer to the actual migraine attack than those sequentially recorded by Kropp & Gerber. In addition, the underlying neurobiological mechanisms modulating event-related and evoked cortical potentials may be different, although similar abnormalities, i.e. deficient habituation, are observed in the attack interval for both types of potentials. Deficient habituation of evoked cortical responses was found between migraine attacks with various sensory modalities, suggesting a general dysfunction in cortical information processing. Concordantly, its normalization in the peri-attack period seems to hold true for various evoked/event-related potentials. It is initiated the day before the attack, i.e. at a time point when prodromes/premonitory symptoms may appear. This may indicate that functional central nervous system changes which are associated with an attack, occur some time before any clinical symptom is recognized. Prodromes, which may present at that time point, are indeed reported only by a minority of patients.
Normalization of VEP habituation around the attack might in part explain previous contradictory results on VEP amplitudes in migraineurs. Because in all VEP studies around 200 responses were averaged, and thus 200 stimuli repeated, habituation is a major factor influencing amplitude of the averaged response. Thus, as most studies were controlled for time interval between recording and last attack, but not for interval between recording and next attack, they may have included patients in the immediate pre-attack phase where they display normal habituation, and thus reduced amplitude. This confounding factor would be more likely to play a role in patients with rather high attack frequencies, who are precisely those patients consulting the specialized headache clinics, where such studies were done.
Habituation and intensity dependence of evoked cortical potentials are probably to a major extent controlled by subcortical serotonergic afferents to the cortex (2, 27). Because of their diffuse innervation pattern and their regular, ‘pacemaker’ activity serotonergic neurones in the raphe nuclei are well suited to tune cortical excitability (28). Akin to low plasma levels of serotonin (29), central serotonergic activity could be low in migraine patients between attacks. This would explain the high intensity dependence of AEPs (30) and the decreased cortical excitability found interictally with transcranial electromagnetic stimulation (14). The normalization of VEP habituation, and particularly of IDAP, we have observed just before and during the migraine attack, might thus reflect an increase in central serotonergic activity. Interestingly, during attacks of migraine without aura hyperperfusion was found in a brain stem region encompassing the dorsal raphe (31). Plasma levels of serotonin also tend to increase during the attack. Whether the fluctuations in central serotonergic activity could be due to changes in the functional state of genetically abnormal P/Q calcium channels (32–34) remains to be determined.