Abstract
Objective and methods
Migraine patients are characterized by increased amplitude and reduced habituation of contingent negative variation (CNV). Furthermore, the CNV underlies periodic changes during the pain-free interval, being maximal before attack. The periodicity of CNV is related to periodic changes in habituation, probably due to variation of orienting activity during the pain-free interval. CNV and orienting response (OR) were studied in 20 females suffering from migraine without aura and in 12 matched healthy females. The neurophysiological recordings in the group of patients were performed 1–4 days before and 4 days after a migraine attack. The amplitudes and habituation of early and late components and total CNV were calculated. The OR was assessed using the habituation of the skin conductance response (SCR) and alpha blocking (AB). The non-parametric tests were employed for statistical analysis.
Results
There were no differences between the two groups for habituation of all CNV components and of SCR following an attack. However, the habituation of AB was significantly reduced in migraine. Before attack we observed a significantly reduced habituation of the early and total CNV and of the AB compared to controls and recordings performed after an attack. The habituation of the late component and of SCR remained unchanged.
Conclusions
The abnormal habituation could be explained by the periodic changes of physiological parameters during the pain-free interval. The impaired habituation of early CNV in migraine is associated with increased orienting activity seen only in the central component (AB) of OR.
Keywords
It has been shown in a number of studies that contingent negative variation (CNV) is a reliable method for studying the central mechanisms of migraine (1, 2). The increased amplitude of CNV and its normalization after prophylactic treatment are important parts of the puzzle of migraine pathogenesis (3, 4). The CNV also appears to have a diagnostic value, especially for the differentiation of headache types and prediction of the clinical efficacy of some prophylactic agents (4–8). There is no doubt that studies of CNV have proven their relevance to research of migraine mechanisms. On the one hand, CNV reflects the extent of excitability in the dendritic trees of cortical pyramidal neurones, so that the highly excitable neural tissue corresponds with a higher amplitude of the surface negative potential (9). Biochemical studies of the activity of excitatory amino acids, MEG investigations, the identification of low magnesium level by means of MR spectroscopy and usage of transcranial magnetic stimulation confirm the assumption that the migraineous brain is hyperexcitable, which is one possible explanation for increased slow cortical negativity in migraine (10–13). On the other hand, it has been demonstrated that CNV abnormalities could be related to reduced habituation (4). The lack of habituation has been found in both neuropsychological and evoked-potentials studies with short and long visual and auditive stimulation (14–20), so that some authors suggest the abnormal habituation or even potentiation to repetitive stimuli as the most important features of the migraineous brain (1, 21).
However, not all interesting CNV findings in migraine have been fully researched. Surprisingly, Kropp and Gerber (4, 22) have shown, using a long inter-stimulus interval, that the significant CNV abnormalities found in migraine are mainly related to the early CNV component. This component was proposed to characterize, among other neuropsychological features such as timing and programming of mental or motor performance, the central manifestation of orienting response (OR) (9, 23, 24). It is well known that the basic characteristic of OR is the habituation of its central and peripheral components (25). However, it has not been demonstrated yet that the abnormal information processing observed in migraine can be related to the enhanced orienting activity. Investigations of CNV in orienting design or time-locked recordings of the slow negative potential and the OR have not been performed.
Using neurophysiological methods Nagel-Leiby et al. (26) have shown for the first time that the CNV amplitude and catecholamine level change during the pain-free interval in patients suffering from menstrual migraine and that the more pronounced abnormalities could be seen during menses, when the probability of attack is relatively high. Furthermore, Kropp and Gerber (27) and Siniatchkin et al. (28) demonstrated the increase of the early CNV component between attacks in migraine without aura, maximal a few days before a headache. These findings, described as the ‘phenomenon of neurophysiological periodicity’, seem to solve some methodological problems in studying the pain-free interval. Firstly, the time since the last migraine attack had to be taken into account by diagnostic or investigation measures of neurophysiological parameters because of the hazard of a time-related error. Secondly, the ‘phenomenon of neurophysiological periodicity’ offers a fertile field for studying the relationship between different findings in migraine, making it possible to observe the parallel changes of parameters during the pain-free interval. However, the described phenomenon is insufficiently investigated and there are still some unsolved questions. In the present study we tried to answer the following questions:
Could the changes in habituation be responsible for the periodicity seen in the amplitude of CNV during the pain-free interval?
Could the periodic changes of CNV amplitude be related to variation in orienting activity during the pain-free interval?
Could the changes in habituation of CNV refer to abnormal orienting response?
The aim of the study was to investigate the periodic changes in habituation of CNV and of components of orienting response during the pain-free interval and to correlate the CNV abnormalities with the orienting activity in migraine patients.
Methods
Materials
Twenty females suffering from migraine without aura (mean age 33.2 ± 6.5 years) participated in this study. They were referred to the study by departmental neurologists and investigated in the specialized headache centre in Moscow. The diagnosis was made according to the classification criteria of the International Headache Society (code 1.1.) (29). Patients' history (18.2 ± 12.4 years) included recurrent attacks of mostly unilateral headache accompanied by nausea or/and vomiting with pain-free periods between attacks. Prior to the neurophysiological investigation the patients recorded about 5.89 ± 4.7 days with migraine per month with a mean duration of each single attack about 31.5 ± 16.1 h for at least 2 months. None of the patients had used oral contraceptives or prophylactic medication within the 4 weeks prior to the beginning of investigation. Attack-related conventional analgesic medication was not considered. None of the patients had used ergotamine or sumatriptan for the treatment of acute migraine attacks. No subject had hearing impairment or drug or alcohol abuse. In order to exclude the influence of the menstrual cycle on the neurophysiological recordings, the menstrual status was recorded. None of the CNV recordings were performed in the premenstrual phase of the ovarian cycle. Twelve healthy females matched for age (mean age 33.3 ± 6.9 years), recruited among students and hospital staff, served as controls.
General design
In the group of patients the contingent negative variation (CNV) and orienting response (OR) were recorded 1–4 days before the first day of an attack and 4 days after the last day with migraine. More detailed analysis of the data in relation to the migraine attack was impossible because of inhomogeneous distribution of patients during 4 days before an attack. Three patients were investigated 1 day before an attack, nine patients 2 days, two patients 3 days and six patients 4 days before an attack. In order to predict the next attack and determine the possible day of investigation, the patients were asked to record in a headache diary the occurrence and precipitating symptoms of attacks for at least 2 months prior to the acquisition of physiological data. During this time the patients were trained to predict their attacks. If the attack did not occur during the 1–4 days after investigation the recordings were discarded. The investigation before an attack was repeated one time by eight patients and two times by two patients. Altogether 64 recordings were performed. The patients wrote a headache diary during the whole investigation. The second day of physiological data acquisition was determined accordingly. The recordings by the healthy controls were performed on a day chosen by the participants.
In order to consider the influence of the actual state of participants on the CNV and OR, the quality and duration of sleep in the night before investigation, mood, medication, drugs, alcohol and caffeine intake on the day of recording were taken into account using standardized questionnaires. If quality of sleep was low and substance intake had occurred, the investigation was repeated later.
The neurophysiological recordings were performed in the following order: CNV was recorded by half of the subjects first. The recording of the orienting reaction followed a pause of 15 minutes. The remaining subjects were examined in the opposite order. In order to exclude the influence of the sequence of recording in relation to an attack, half of the patients were investigated first before an attack and then following one, and the other half of migraineous women were investigated first following and then preceding an attack. It has been recently shown (30) that the CNV has a good test/retest reliability and that the intra-individual variation of the early CNV component is small. These observations enable a comparison of the pre- and post-attack recordings without a sequence-related error.
Psychophysiological recording
All participants were comfortably seated in a chair placed in a soundproof chamber. They were asked not to close and move there eyes or to blink. CNV and EEG were recorded using Ag/AgCl electrodes over C3 and C4 according to the International 10–20 System with linked mastoids as reference. Recordings with an electrode resistance more than 5 kOhm were rejected and electrodes were reapplied. The EEG signals were amplified by a Conan system amplifier, analogue-filtered with a bandpass of 0.05–40 Hz for CNV and 0.5–40 Hz for EEG and digitized at a rate of 100 Hz for each channel. In order to exclude the vertical eye movement artifacts, the electro-oculogramm (EOG) was recorded parallel to CNV and EEG using electrodes positioned above and below the left eye. If eye and eyelid movements identified by EOG deflections as greater than 20 µV interfered with the CNV or EEG measurement for 5 sec, the trial or segment was rejected. A protocol listed the number of rejected trials for each session. The groups did not differ according to the number of rejected CNV trials, so that EOG artifacts did not exert any influence on the CNV habituation.
During CNV recording the auditory warning (S1) and imperative (S2) stimuli were produced through a loudspeaker placed behind the subjects with an intensity 80 dB(a) and frequency 1000 Hz. The time between S1 and S2 was 3 sec. A CNV session consisted of 40 artifact-free trials in which the subject had to react to the imperative stimulus as soon as possible by pushing a button. S1 and S2 pairs were offered in random intervals of 5–10 sec. The length of recording was 5 sec (the recording began 0.5 sec before S1 and ended 1.5 sec after S2). The period between recording onset and S1 was taken as the baseline for all measurements. Forty CNV trials were averaged and the amplitudes of total CNV, early and late components were calculated. The total CNV was assessed between 500 and 3000 msec after S1. The early CNV component was defined as the mean amplitude in a window of 200 msec duration placed around the latency which fit the maximal amplitude of the expectancy wave between 550 and 750 msec after S1. The late component consisted of the mean amplitude during the 200 msec before S2. All CNV trials were divided into four trial blocks, each consisting of 10 consecutive single recordings which were than averaged. Habituation was calculated as a trend of the four consecutive blocks. Only habituation of the early CNV component is presented in this article. Other CNV components did not show any changes from block to block of CNV recording in our sample of patients, as shown in previously published studies (4, 22). Reaction time was not recorded during this investigation because of technical problems. However, all patients were highly motivated and compliant during all sessions of CNV recordings and were told to react as soon as possible before each recording began.
The orienting pattern consisted of a sequence of 20 pure sinus tones of 1000-Hz frequency, 5-sec duration, 70-dB intensity, and randomized inter-stimulus intervals (mean 10 sec, range = 5–15 sec). Tones were generated by a function generator with intensity controlled by a programmable attenuator. Stimuli were produced through a loudspeaker placed behind the subjects. Each was told he would hear simple tones, assured they would not be unpleasant, and was asked to sit still and remain awake. The first tone among the sequence, after which no response of a component was observed three times in succession, was used as an index of habituation for each of the OR components. Therefore, the number of responses for each component characterized the ability of subjects to habituate.
The central components of OR were studied using the alpha blocking (AB) criterion by assessment of EEG data. The AB was estimated by means of EEG power spectrum analysis using a Fast Fourrier transformation in 5-sec intervals before and during stimulus presentation. These 5-sec intervals were divided into five periods of 1 sec (100 samples) each. The alpha (8.00–12.5 Hz) and beta (12.6–32.0 Hz) bands were considered sufficient for AB analysis. The ratio of power for each frequency in each second obtained during stimulus presentation was calculated relative to the mean power of that frequency over the 5-sec pre-stimulus period, and was expressed as a percentage. The stimulus/pre-stimulus power ratios were averaged as a mean ratio of percentage of power over the whole 5-sec interval, obtained during the stimulus presentation. If the alpha power during 5-sec stimulus presentation was reduced by at least 20%, this suppression was considered as an alpha blocking. The 20% criterion was chosen for a more stable and relatively sensitive habituation index (31).
The peripheral components of OR were investigated using the analysis of the skin conductance response (SCR), measured from bipolar leads on the medial phalanges of the second and third fingers of the left hand using Ag/AgCl electrodes and 0.5% potassium chloride in 2% agar-agar as electrolyte. The signals were amplified by a Conan system amplifier and analogue-filtered with a bandpass of 0.01–75 Hz. Amplification allowed identification of all SCRs greater than 0.05 µS. A skin conductance orienting response was defined as an increase in conductance greater than 0.05 µS within a latency window of 1–3 sec following any tone onset. The frequency of non-specific skin conductance responses was measured in the 3-min rest period before stimulus presentation.
Analysis
All statistical tests were performed with non-parametric procedures. In the case
of the intra-subject design (comparison of neurophysiological data obtained
before and after attack in migraine patients) the Wilcoxon paired test was used.
Independent group comparisons (patients vs. controls) were conducted by the
non-parametric Mann–Whitney U-test. The relationship
between CNV and components of orienting reaction were analysed with the Pearson
rank order correlation. Analysis of variance was not possible because of
non-normal distribution of data in all groups (tested by Kolmogorov-Smirnov
test) and inhomogenity of variances (tested by F-test during performance of the
one-way
Results
CNV amplitudes and habituation
Figs. 1 (a, b) and 2 (a, b), and Table 1, show the averaged CNV amplitudes and habituation before and after an attack. There were no differences between patient and control groups after attack for amplitudes of all CNV components and habituation of the early CNV. Before attack a significant increase of amplitude of early and total CNV and reduction of habituation of the early CNV component compared to after-attack recordings were found. The patients differed significantly from the control group according to amplitude and habituation of early component and total CNV. The most pronounced differences in habituation between the groups were observed during the fourth block of assessment. However, significant differences in early CNV before and after attack were also found in the second and third blocks.
=
migraineurs after an attack; 
=
migraineurs before an attack.
Group comparisons for the four trials blocks each with 10 subsequent CNV
recordings for the total CNV, the early and the late CNV components
performed before and after a migraine attack. Abbreviations: Z = Z
cores, P = probability,
The values of the amplitude regression lines (slope of habituation) for the early CNV component in patients before and after an attack and in healthy controls are listed in Table 2. Using a linear equation (y = ax + b) the increase in slope of habituation (the greatest positive slope) just before a migraine attack compared to slopes evaluated for data obtained after an attack and in healthy subjects (negative slopes) can be observed.
Slope of habituation (regression coefficient “a” and initial value “b” from the equation of linear regression y = ax + b) in groups of healthy controls and migraine patients after and before a migraine attack
Orienting response
After attack (Fig. 3 (a, b) and Table 3) the patients differ from controls only according to habituation of alpha blocking (AB) response. We observed a significantly more pronounced reduction of habituation of the central component of OR (AB) 1–4 days before attack. Difference between patients and controls according to AB response habituation just before a migraine attack became more pronounced and significant. No abnormalities in habituation of SCR before or after attack were found.
Groups comparisons for number of responses of the alpha blocking and the
skin conductance response obtained during OR recording in controls and
migraine patients before and after an attack. Abbreviations: AB = alpha
blocking, SCR = skin conductance response,
According to hemispheric asymmetry, we observed no significant differences in habituation of CNV and AB between left and right hemispheres. No correlation between site of pain during the attack following the recording and hemispheric asymmetry in habituation of the CNV and the OR components were found.
Correlation analysis between CNV parameters and habituation of the OR components revealed weak positive correlations between amplitude of the early CNV as well as slope of the CNV habituation and habituation of the alpha blocking (described as difference between the blocking effect on the first and the last tone) (r = 0.31, P < 0.05 and r = 0.35, P < 0.05, respectively). No other significant correlations were found.
Discussion
As previously shown, migraine is characterized by periodic changes in CNV amplitude representing the variation of the cortical excitability and central hyperactivity during the pain-free interval. The periodicity observed in the CNV amplitude can be explained by dynamic changes in habituation of total and, in great part, of the early component, which do not habituate or even tend to increase during repetitive stimulation just before an attack. The changes in ability to habituate during the pain-free interval were not only related to CNV, but also to the central component of OR. These results suggest the following assumptions and emphasize the questions requiring further research.
The habituation of neurophysiological parameters in migraine underlies periodic changes during the pain-free interval and is maximally reduced before an attack. Taking into account these findings, it seems that the physiological mechanisms leading to an attack function like a ‘sensitization in sensitization’. The habituation of event-related potentials in migraine appears to be a secondary phenomenon depending on the mechanisms responsible for periodicity. However, these mechanisms are poorly investigated and our suggestions about these mechanisms are speculative.
It can be proposed that, on the one hand, subliminally environmental factors during the interval could lead to changes in activity of different neurotransmitters followed by increased susceptibility to migraine attack and modifications in information processing. Indirect evidence for this is the observation that a number of patients experience warning symptoms before a headache episode, consisting especially of neuropsychological abnormalities which can be explained by abnormal dopaminergic and serotoninergic function (32). It was suggested that the lack of habituation in migraine is related to low serotoninergic activity causing the low level of cortical preactivation, enabling detectable cortical over-excitation (21). The Detroit research group demonstrated that serum dopamine rises and platelet serotonine drops in migraine patients during menses, when the probability of attack onset is relatively high (26, 33). Further investigations have to be performed in order to explain why neurotransmitter abnormalities build up during the pain-free interval under environmental influences and do not return to the optimal homeostatic level.
On the other hand, the periodical lack of habituation can be explained by a metabolic defect of the brain. It has been demonstrated that the habituation of visual evoked potentials is related to the metabolic changes of cortical neurones and is accompanied by elimination of lactate in neural tissue (34). Sappey-Marinier et al. (1992) found that cortical lactate level began to decrease only after the amplitude of evoked potentials had diminished by >50% (34). In such a way, habituation of evoked and event-related potentials may represent a protective mechanism prohibiting an excessive increase in cortical lactate level (16). Migraine patients are characterized by impaired oxidative energy metabolism and elevated lactate in the brain during pain-free intervals (35, 36). It can be suggested that periodically occurring loss of habituation causes increase in cortical lactate and initiates a cascade of metabolic processes which result in migraine attack. However, do the metabolic abnormalities, if related to the lack of habituation, change during the pain-free interval?
From another point of view, it was proposed that the metabolic shift causes instability of neuronal function due to neuronal excitability (37). The cortical excitability as a result, for example, of a low magnesium level or impaired function of calcium channels, could influence the oxidative phosphorilation (38). Periodically occurring changes in cortical excitability may also be responsible for both reduced habituation and metabolic deficit in the migraineous brain. Further investigations on the causal relationship between metabolic defect, cortical hyperexcitability and habituation would be of great value.
The periodic changes of amplitude and habituation of CNV are related especially to early and total CNV, but not to the late component. Furthermore, the variability of CNV habituation is accompanied by abnormal extinction of central components of orienting response. Does the early component of the slow negative potential represent the OR and can we interpret early CNV abnormalities as a changed orienting activity? Indeed, the early wave can be elicited by conditions that often evoke an OR (for review 9, 23, 24, 39). It has been demonstrated that this CNV component is extremely sensitive to the stimulus' information content, to the relevance of that information and the rate of presentation. The early wave is reported to mirror the temporal course of facilitation in forewarned detection and reaction time tasks, consistent with speculations that a primary function of the OR is to facilitate perception and performance (24). On the other hand, there is also evidence that the early CNV often does not fulfil all criteria of an OR, so that some authors emphasize that the argument for linking the early CNV component to the OR is poor and the relationship between these two processes is indirect (40, 41). Our data support this since we found only a weak correlation between habituation of the early CNV and habituation of the alpha blocking. This means that the early wave of CNV is more than the simple representation of OR, but possibly shares common mechanisms and central pathways with some components of OR. It has been previously shown that both alpha blocking and early CNV are closely related to the function of the reticulo-cortical activation system (42). The reduced habituation of these components emphasizes the important role of the reticular formation in migraine pathogenesis, describing it as a possible research target perhaps responsible for central manifestations of ‘phenomenon of neurophysiological periodicity’.
Habituation of the early CNV component represents possibly the stages of information processing at the cognitive level. Contingent negative variation is a cognitive event-related potential which could be recorded primarily from association areas of the brain (9). In particular the early CNV component is often discussed as associated with activity of the frontal cortex (23, 24). However, it has been demonstrated that the habituation deficit in migraine occurs already at the level of cortical activity generated in primary and secondary sensory cortices (16, 21). This suggests that the dysfunction is already present at the level of perception, and that the abnormalities found at the cognitive level could just be secondary ‘down stream’ changes. The relationship between the habituation deficit at the cognitive and the perceptual levels has not been adequately investigated. The majority of studies on habituation in migraine were performed based on analyses of cognitive event-related potentials as the P300 component of auditory and visual evoked potentials and the contingent negative variation (4, 5, 8, 17–20). Almost all event-related potentials studies (P300) investigated habituation of the ‘target’ stimuli recorded during the ‘oddball’ paradigm. Complex studies related to the comparison of the habituation deficits at the perceptual and cognitive levels would be of importance in understanding the central pathophysiological mechanisms of migraine.
In conclusion, the present study supports the assumption that the reduced habituation in response to repeated stimulation is an important feature of the migraineous brain, also being involved in periodic changes of physiological parameters during the pain-free interval. One possible explanation for the lack of habituation is the increased orienting activity. Despite some evidence linking migraine with the orienting response, there are still a large number of questions about the role of orienting activity in migraine. The phenomenon of neurophysiological periodicity seems to be an appropriate tool for studying these questions. Further work investigating the possible relationship between habituation, excitability, central neurotransmission, information processing, and brain metabolism in headache is needed.