Abstract
The particular mechanisms of migraine anticipation by different precipitating agents are still unknown. The contingent negative variation (CNV) was recorded in the premenstrual and ovulation phases of the cycle in both rest and stress conditions in 17 migraine and 15 healthy women. In migraineurs a significant increase of amplitude of the initial CNV component in the premenstrual phase compared with ovulation was observed. During both the ovulation and premenstrual phases both migraineurs and controls demonstrated a significant increase of the CNV amplitude on stress. The increase of the amplitude on stress in the premenstrual phase was more pronounced in migraineurs. This study shows that stress and menstrual cycle are associated with changes of the initial CNV amplitude, probably indicating a higher probability of migraine attacks.
Introduction
Migraine is a neurovascular disorder with a paroxysmal course of disease. Attacks of migraine may be precipitated by a number of conditions such as hormonal changes, stress, fasting, fatigue, variations in sleep rhythm and weather (1), but the particular mechanisms of migraine precipitation by these factors are still unknown. It may be suggested that different precipitants exert their effect in the central nervous system (CNS) because migraine attacks start in the CNS, and vascular mechanisms are more likely to follow a cascade of centrally induced changes (2).
The contingent negative variation (CNV) is a slow cortical event-related potential which can be recorded from the scalp between two stimuli while the subject is waiting for the second event and preparing for task performance. The CNV is related to the level of cortical excitability following activation in the striato–thalamo–cortical loop (3). The early CNV component (initial CNV or iCNV, 500–1000 ms after the first warning stimulus) may represent the attack anticipation because its amplitude and habituation change during the headache-free interval with gradual increase of abnormalities a few days before a migraine attack, so that the maximal negativity and the most pronounced loss of habituation may be observed just before an attack (4). The changes of iCNV parameters before an attack are well replicated in children and adults (5–7) and correspond to increasing abnormalities of other parameters of cortical information processing (abnormal habituation of visual evoked potentials and abnormal slope of intensity dependence of auditory evoked potentials) before an attack (8, 9). According to studies on periodic changes of the CNV, the assumption can be made that the iCNV may reflect the increased susceptibility of the brain to migraine-precipitating and -provoking agents and can be studied as a possible surrogate marker for investigating central mechanisms temporally associated with migraine attacks, e.g. mechanisms of migraine precipitation (5–7). This hypothesis will be proven in the present study.
Menstrual cycle and stress are the most frequent precipitants of migraine. About 60–70% of women with migraines suffer from menstruation-related migraine attacks (10). The normal female life cycle is associated with a number of hormonal milestones: menarche, pregnancy, contraceptive use, menopause and the use of replacement sex hormones. All these events and interventions alter the levels and cycling of sex hormones and may cause a change in the prevalence and intensity of headache (11). Behavioural, biochemical and physiological data indicate that ovarian hormones exert powerful effects on brain function (12). Oestrogens enhance and progesterone diminishes neuronal excitability, as demonstrated in human transcranial magnetic stimulation (TMS) studies (13). Various EEG and evoked-potential studies have demonstrated changes in power spectrum, lantencies and amplitudes of spontaneous and evoked cortical activity (14–16). CNV was also found to be sensitive to oestrogen/progesterone changes. Nagel-Leiby et al. (17) have shown that in migraineurs (with and without aura) the iCNV increases in the premenstrual phase of the cycle (when the probability of an attack is higher) and decreases in the ovulation phase. Therefore, the CNV seems to be an adequate method for investigating mestrual-related brain functioning.
According to different epidemiological studies, stress is a significant provoking agent in about 40–50% of patients (1, 18). It has been demonstrated that a few days before a migraine attack a greater number of daily hassles and pronounced mood changes may be observed (19–23). It is difficult to describe whether the increased stress susceptibility, anxiety and depressed mood before an attack in migraineurs is the cause or the effect of neurophysiological attack anticipation; however, it was proposed that the increasing stress exposure before a migraine episode may possibly explain increasing iCNV abnormalities (24). The influence of stress on CNV in migraine has yet to be investigated. This study used experimental achievement stress. Previous studies have shown that migraineurs possess a pronounced achievement motivation with marked fear of failure (25, 26). The CNV reaction time paradigm provides a possibility for manipulating the pressure to achieve because the CNV is directly related to performance and the paradigm is aimed at improvement of performance (3). Therefore, the CNV design enables the investigation of an additive effect of precipitants being typical to migraine such as the achievement stress and the menstrual cycle, by recording of the CNV in the rest and stress conditions during the premenstrual and ovulatory phases of the cycle.
Methods
Patients
Seventeen women suffering from migraine without aura (mean age 34.6 ± 6.0 years; mean attack frequency 3.16 ± 1.6 days with migraine/month; mean attack duration 16.5 ± 8.1 h; mean analgesics or triptan use 5.9 ± 4.4 tablets/month; mean frequency of tension-type headache 4.0 ± 4.4 days/month; mean duration of disease 18.1 ± 6.4 years) were recruited from the general population using advertisements. The diagnosis was made according to the criteria of the International Headache Society (27) (code 1.1) by the first author (neurologist) trained in diagnosis and treatment of headaches. Before being included in the study, all migraineurs were asked to keep a headache diary and a diary of the menstrual cycle for 3 months in order to collect clinical data and assess the relationship between migraine and menstruation. Fifteen age- and gender-matched healthy women (mean age 31.4 ± 6.7 years) without any family history of migraine or other types of headache were enrolled from the population using advertisements.
The exclusion criteria were similar for both groups and included the following: consumption of alcohol or other drugs (substance abuse and sporadic intake on the day of examination), history of psychopathological or neurological disorders, developmental and school retardation in the past, major medical problems, current medication, and non-corrected sensory deficits (obtained from medical history). Migraineurs had not used prophylactic migraine medication for at least 3 months prior to the investigation. All women had a regular menstrual cycle. None of the participants had used oral contraceptives for at least 3 months prior to admission in the study. The occurrence of the premenstrual syndrome was assessed anamnestically. None of the women investigated was characterized by a marked premenstrual syndrome. All migraineurs reported that the menstrual cycle and stress were the most frequent precipitating agents of their migraine attacks. However, none of the women suffered from only menstrual-related migraine (not all menstrual cycles and ovulation phases were accompanied by a migraine attack and not all attacks occurred in relation to the cycle). The study was approved by the Ethics Committee of the Faculty of Medicine, University of Kiel, Germany. The subjects were informed about the course of the experiment and gave written informed consent according to the Helsinki Declaration.
Experimental design
In each subject the recordings were performed in the premenstrual and ovulatory phases of the menstrual cycle. In nine of the women suffering from migraine and in six healthy women, data were collected first in the premenstrual phase. In other subjects the CNV was recorded first in the ovulatory phase. A rise in the basal body temperature (BBT) was used to confirm the occurrence of ovulation. For this purpose, patients were given basal thermometers and charts for recording BBT. The onset of the next menses was estimated by information supplied by patients relating to the length of the last three menstrual cycles and the observation of a 0.3–0.5 rise in BBT. All premenstrual recordings were carried out 1–2 days before the first day of menstruation. The occurrence of menstruation as well as the onset of a migraine attack during the premenstrual, menstrual and ovulatory phases after the recording was controlled by telephone contact. The mean length of the menstrual cycle in the migraine group was 27.4 ± 1.2 days and in the healthy group 27.8 ± 0.8 days. The migraine attack occurred on the next day after the recording in four women, in three women on the second day, in two women on the third day, in four women the onset of the migraine attack was registered on days 4–7 after recording but during the menstrual phase, and only in four women did the attack occur after menstruation. None of migraine women experienced a migraine attack during the ovulation phase of the cycle. Only in five women did the migraine attack occur within the period of 4 days before or after ovulation. The mean number of days passed between the recording and the next migraine attack during the premenstrual phase was 4.9 ± 4.6, and during ovulation −9.1 ± 7.2 days (two-tailed t-test: t(14) =−2.432, P = 0.029).
The influence of the actual state of the participants on the CNV during both the premenstrual and ovulatory phases of the cycle was considered by recording the quality (subjective evaluation on a digital scale from 0 to 9) and duration of sleep during the night before, mood (evaluation on a digital scale from 0 to 9), vigilance (on a digital scale from 0 to 9), medication, drug, alcohol and caffeine intake as well as smoking on the day of recording using a questionnaire.
During both the premenstrual and ovulatory phases of the cycle the CNV recordings were performed in the rest and stress conditions. During the rest condition the participants were asked to press a key as quickly as possible whenever they detected the imperative tone. In order to enhance stress and pressure, the mean reaction time (RT) obtained during the rest condition was then divided by 2. This substantially shortened RT was put into the feedback program as a target parameter. The subjects were instructed to try hard to press a key still more quickly than in the rest condition in order to achieve the target RT. After each trial, the subjects received feedback on a monitor about success of their attempts. When the subjects could not press the key at this reduced target time, they received the negative reinforcement and saw the signal ‘more quickly’ on the monitor. When the subjects fulfilled the task, they received no negative reinforcement. The subjects were told that success in the task was possible, and they must try to improve their performance by using feedback. The real probability of fulfilling the task was estimated as 10.5 ± 4.6%, therefore the patients were hard pressed for time and received predominantly negative reinforcement.
Recordings
All participants were seated in an armchair with eyes open in an electrically shielded sound-attenuated room. The auditory warning (S1) and imperative (S2) stimuli (75 dB(a)) were produced by a loudspeaker located behind the subject. The interval between S1 and S2 was 3 s. A CNV session consisted of 32 trials in which the subject had to react to the imperative stimulus (GO-response). Additionally, eight trials were randomly presented where no reaction was expected (NO-GO-response). The warning stimulus (S1) for the GO-response had a frequency of f = 1000 Hz and lasted 100 ms. The warning tone for the NO-GO-response had a frequency of f = 200 Hz. The imperative stimulus (S2) had a frequency of f = 2500 Hz, lasted a maximum of 1500 ms and was deactivated by pressing the button. Reaction time was defined as the period between onset of S2 and the pressing of the button. S1 and S2 pairs were offered at random intervals of 10–15 s. The duration of one recording was 6 s (the recording began 1 s before S1 and ended 2 s after S2). The period between recording onset and S1 was taken as the baseline for all measurements.
The EEG was recorded using non-polarizable Ag/AgCl electrodes over Cz according to the International 10–20 System with linked mastoids as reference. The electrode site on the scalp was prepared with alcohol and scraped with rough paper, resulting in an electrode impedance of <7 kΩ. The EEG signals were amplified using a Nihon Kohden amplifier with a time constant of 5 s (equivalent to a high-pass filter of 0.03 Hz) and a low-pass filter of 35 Hz digitized at a rate of 100 Hz for each channel. Vertical eye movement artefacts were excluded by a parallel recording of the electrooculogram (EOG) with electrodes (Ag/AgCl) positioned 1–1.5 cm above and below the right eye. The trial was rejected if EOG deflections >20 μV interfered with 5 s of the EEG recording. A protocol listed the number of rejected trials for each recording. Approximately 12% of the EEG trials were rejected due to eye movement artefacts. There were no significant differences between groups according to the number of rejected trials.
Data analysis
The GO-trials were averaged and the amplitudes of the total CNV, iCNV and late components and postimperative negative variation (PINV) were calculated. This study considers only the iCNV. This limitation of data presentation is based on the literature (4–6, 28) and on observations in this sample of subjects, whereby only the iCNV showed amplitude and habituation differences between groups. Indeed, as demonstrated in the Introduction, only the iCNV has a relevance to migraine. The iCNV was defined as the mean amplitude in a window of 200 ms duration around the maximal amplitude of the expectancy wave between 550 and 750 ms after S1 (28). Each recording was divided into eight blocks of four consecutive trials to determine the course of habituation and trends in the iCNV amplitudes. Habituation was indicated by a negative, whereas dishabituation was marked by a positive slope calculated by linear regression (y = ax + b, where a is the slope of habituation and b the intercept of linear regression) (29).
Statistical evaluation
Because the CNV data were normally distributed and characterized by homogen variances, an analysis of variance with two within-subjects factors ‘menstruation’ (ovulation vs. premenstrual phase of the cycle) and ‘stress’ (rest vs. achiement stress conditions) and a between-subjects factor ‘group’ (migraineurs vs. healthy controls) was applied for all CNV parameters. The analysis of variance (with factors ‘menstruation’ and ‘group’) and χ2 test were used to characterize differences between menstrual phases and groups according to mood, sleep, vigilance as well as smoking and caffein intake. In order to demonstrate the additive effect of precipitants, the CNV data obtained during the stress condition in the ovulation, rest and stress conditions in the premenstrual phases of the cycle were compared with CNV characteristics recorded during the rest condition of the ovulatory phase using two-tailed t-tests with the appropriate Bonferroni α adjustment. The differences were considered as significant by the α level of 0.05.
Results
Analysis of potential confounding variables
Table 1 shows results of evaluation of mood, vigilance and fatigue, duration and quality of sleep for women suffering from migraine and healthy women obtained on the day during the premenstrual phase and the day of the ovulatory phase of the cycle. None of the parameters listed differs significantly between the groups and phases of the cycle, so that any CNV changes could not be attributed to differences in mood, attention and/or sleep. The proportion of subjects who smoked cigarettes or drank caffeinated beverages on the day of recording was not significantly different in migraineurs and controls in either phase of the menstrual cycle (χ2, non-significant). The proportions of subjects who habitually smoked cigarettes [three (18%) of 17 migraineurs vs. three (20%) of 15 controls, χ2, non-significant] were also not significantly different. All investigations of all subjects were performed after 12.00 h.
Characteristics of mood, vigilance/fatigue, sleep quality and duration obtained by women suffering from migraine and healthy women, as well as number of subjects who smoked cigarettes or drank caffeinated beverages on the day of recording in both groups (NS, non-significant as proven by using analysis of variance with factors ‘menstruation’ and ‘group’ or χ2)
Mood, vigilance and quality of sleep were evaluated using visual-numeric scales from 0 (depressed mood, fatigue and poor attention, very poor sleep quality) to 9 (euphoric mood, excellent concentration, very good sleep quality).
The analysis of variance with the reaction time as a dependent variable (see Table 2) revealed the only significant main effect ‘stress’ (F(1,30) = 55.257; P < 0.001). No other effects or interactions were significant. The reaction time changed only during the stress condition compared with rest condition and independently of phases of the menstrual cycle. Migraineurs and controls demonstrated equal reaction times in all conditions investigated. The shortened reaction time during stress is a logical consequence of the instruction and the task. The reaction time data demonstrate that the subjects from both groups in both phases were motivated to achieve better reaction time results and that any changes in CNV in different conditions can not be attributed to behavioural differences.
Amplitudes of the early and late contingent negative variation (CNV) components (μV), the slope of CNV habituation (coefficient of linear regression) as well as the mean reaction time in patients and healthy subjects in premenstrual and ovulatory phases of the menstrual cycle and in rest and stress conditions
Contingent negative variation
Table 2 demonstrates amplitudes of the early and late CNV components, the slope of CNV habituation as well as the mean reaction time in patients and healthy subjects in premenstrual and ovulatory phases of the menstrual cycle and in rest and stress conditions. Figure 1 shows changes of the early CNV during different phases and conditions of the experiment. Women suffering from migraine were characterized by larger CNV amplitudes of both components and reduced CNV habituation compared with healthy controls, especially in the premenstrual phase of the cycle. Stress conditions caused a marked increase of all CNV amplitudes but no sufficient changes in CNV habituation in both groups with respect to both phases of the menstrual cycle.
Changes of the early contingent negative variation (iCNV) during different phases and conditions of the experiment (rest and stress conditions as well as premenstrual and ovulatory phases of the menstrual cycle) in migraineurs (□) and healthy controls (▪).
The analysis of variance for the amplitude of the early CNV component demonstrated significant main effects ‘menstruation’ (F(1,30) = 5.809; P = 0.02) and ‘stress’ (F(1,30) = 19.429; P ≤ 0.001) and significant interaction ‘menstruation × group’ (F(1,30) = 3.961; P = 0.05). The interactions ‘stress × group’ (F(1,30) = 0.018; P = 0.894), ‘menstruation × stress’ (F(1,30) = 0.001; P = 0.972) and ‘menstruation × stress × group’ (F(1,30) = 0.119; P = 0.732) were non-significant. This means that the early CNV changes during the menstrual cycle, and that these cycle-dependent variations of the amplitude differ between migrainous and healthy women (Table 2 and Fig. 1). Indeed, in healthy participants CNV characteristics did not differ substantially by comparison of premenstrual and ovulatory phases of the cycle. In migraineurs a significant increase of the early CNV in the premenstrual phase compared with the ovulatory phase could be observed (t(16) = 2.421; P = 0.028). Stress conditions led to a marked increase of the early CNV amplitude. However, this increase was equally pronounced in both groups and both phases of the menstrual cycle.
According to the amplitude of the early CNV, a marked additive influence of different precipitating agents on this CNV characteristic could be demonstrated (Fig. 1). When the CNV parameters in the rest and stress conditions did not differ between migraineurs and controls during the ovulatory phase of the cycle, the women suffering from migraine were characterized by much more pronounced negativity than healthy women during the premenstrual phase, especially in the stress condition (a tendency for the rest and a significant result for the stress conditions after Bonferroni adjustment for four comparisons). The more precipitating agents exerted an effect on the subject, the more negative was the amplitude of the early CNV component. This could be demonstrated for migraine but not for healthy women. The comparison of the ovulatory phase (as a condition with a minimal risk of a migraine attack) with all other conditions revealed an increase in significance of differences dependent on the risk of a migraine attack for migraineurs, but not for healthy women (Table 3). Moreover, in the migraine group the amplitude of the early CNV during the stress condition in the premenstrual phase differed significantly from the stress condition in the ovulatory phase (t(16) = 3.12; P = 0.007) and rest condition in the premenstrual phase (t(16) = 2.86; P = 0.011).
Comparison (P-values) of contingent negative variation (CNV) parameters obtained in the rest condition of the ovulatory phase (rest (Ov.)) with those recorded in the stress condition of ovulation (stress (Ov.)) and in both conditions of the premenstrual phase (rest and stress (Prem.)) (results of two-tailed t-tests with appropriate Bonferroni adjustment)
The analysis of variance for the amplitude of the late CNV component and slope of the CNV habituation demonstrated the only significant main effect ‘stress’ (F(1,30) = 23.561; P < 0.001) for the late CNV. No other effects or interactions were significant. The comparison of these CNV parameters between women suffering from migraine and healthy women revealed no significant results in any of the conditions or phases investigated. The analysis of additive effect by the comparison of the ovulatory phase of the menstrual cycle with other conditions of the study (Table 3) provided no significant effects for the slope of CNV habituation, and was significant but independent of the risk of a migraine attack for the late CNV. For this component, stress in the ovulatory phase caused a more pronounced increase of cortical negativity than in the premenstrual phase.
Discussion
In summary, a significant increase of amplitude of the early CNV component (iCNV) was observed in migraineurs in the premenstrual phase compared with ovulation. No changes in CNV during the menstrual cycle were found in healthy women. During both the ovulation and premenstrual phases both migraine patients and controls demonstrated a significant increase of the iCNV amplitude on stress compared with rest conditions. No changes in iCNV habituation were found. The increase of the amplitude on stress in the premenstrual phase was more pronounced in migraineurs, so that the patients differed significantly from healthy controls. Furthermore, an additive effect of the putative precipitating factors could be shown: the increase of the iCNV in migraineurs was more pronounced during exposure to both stress and premenstrual periods compared with exposure to each factor separately or the ovulatory phase of the menstrual cycle. The additive effect of precipitating agents could be described for migraine patients but not for healthy controls.
The study demonstrates that stress and menstrual cycle influence the iCNV amplitude and in this way are associated with an increased probability of migraine attacks. This statement is based on observations that the amplitude of the iCNV rises between migraine attacks with maximally pronounced cortical negativity just before an attack (4–7). The iCNV amplitude is thought to have predictive value—the more negative the amplitude, the larger the probability of the migraine attack (4). It has even been demonstrated that the prophylactic effect of some agents such as β-blockers and calcium antagonists is associated with a decrease of CNV amplitude (30, 31). The direct reduction of the CNV through the neurofeedback of slow cortical potentials may also lead to improvements in the clinical course of migraine (32). Therefore, as shown in a number of studies, the CNV is an appropriate method which may be used to describe the attack anticipation by the brain or, in other words, disposition of the brain to a migraine attack (24). Because the CNV corresponds to the level of cortical excitability and the iCNV represents the early stages of preparatory activity related to the orienting to relevant stimuli necessary for an appropriate performance (3), and because the typical migraine precipitants such as stress and menstrual cycle lead to increase of the iCNV amplitude as shown in this study, migraine anticipation could then be discribed as gradual changes in cortical excitability corresponding to an increase in orienting activity under the influence of migraine-precipitating agents. Which mechanisms are responsible for this neurophysiological migraine anticipation?
It can be proposed that stress is one of a number of factors influencing the changes in internal neurophysiological disposition to a migraine attack. It has been shown that 1–3 days before a migraine attack an increased number of daily hassles and stressful events as well as significant changes in mood and anxiety can be observed (19–23). It has even been demonstrated that experimental stress and anxiety may lead to an increase of CNV amplitude (33). Because the iCNV has been often proposed to correspond to the activity of the noradrenergic system (3, 34) and because emotional stress and sympatho-adrenergic function are closely related to each other, the link between the iCNV and stress on the basis of noradrenergic activity seems to be obvious, especially in migraine (35).
The investigations of physiological reactivity of different parameters in stress have shown increased vulnerability and sensitivity of migraineurs to laboratory stressors, although the results of these studies are inconsistent and even contradictory [for review (36, 37)]. This inconsistency of results could be related to both non-specific stresses used in most of the studies (mental arrythmetics, imagery of unpleasant situations, etc.) and investigation of mostly peripheral physiological parameters (heart rate, blood preasure, skin conductance and temperature, etc.), although migraine has been shown as a disease of the central nervous system (2). In this study we first used modelling of a specific stress situation based on achievement motivation which is abnormal in migraineurs (25, 26). Second, we recorded the reactivity of a central parameter—a slow cortical potential, the CNV—which is relevant to the attack disposition/anticipation (24). However, in our study no increased reactivity on stress in migraine patients compared with healthy controls was observed. Healthy women demonstrated also a pronounced enhancement of the iCNV amplitude in an achievement situation. In women suffering from migraine the CNV amplitude on stress was much greater in the premenstrual phase, representing an additive effect of stress and menstruation. It is important to emphasize that most migraineurs investigated experienced migraine attacks during the menstrual but not ovulatory phase of the cycle. It can be suggested that the combination of precipitating events is necessary to achieve the threshold of a migraine attack and to increase susceptibility of the migrainous brain to different provoking agents. However, the non-significant effect ‘stress × group’ and the significant effect ‘menstruation × group’ demonstrate that the menstrual cycle influences iCNV amplitude to a greater degree than stress. This provides an additional explanation for contradictions in studies on neurophysiological stress reactivity, because most of these studies did not consider the menstrual cycle.
Although the premenstrual abrupt fall of plasma oestradiol appears to precipitate migraine attacks, the exact relationship between these two phenomena remains unclear. The menstrual cycle is the result of a carefully orchestrated sequence of interactions between the hypothalamus, pituitary, ovary and endometrium, with the sex hormones acting as modulators and effectors at each level (10). Oestrogen and progesterone have potent effects on central serotoninergic and opioid neurons, modulating both neuronal activity and receptor density, and influence cortical excitability, metabolism and autonomic function (38). Surprisingly, the changes of neurophysiological parameters in the premenstrual phase of the cycle when compared with ovulation resemble migraine-specific abnormalities which could be observed between attacks in patients and correspond to the variation of these abnormalities before an attack. For example, in healthy women the increase in α- and θ-activity of the EEG was observed at premenses compared with ovulation (14), and the same changes were found in women suffering from migraine before an attack compared with the postattack recordings and healthy controls (5). In addition, in the premenstrual phase the increased amplitudes of visual evoked potentials (VEP) and prolonged VEP latencies have been described (14, 39, 40). This corresponds to increased amplitudes and latencies of VEP found in migraine, especially before a migraine attack (8).
The changes in catecholaminergic and glutamatergic activity as well as endogenous opiate function during the menstrual cycle could possibly explain the menstrual related variations in EEG, VEP and CNV (38, 41, 42). For example, Nagel-Leiby et al. (17) demonstrated an increase in the iCNV amplitude in migraineurs in the premenstrual phase of the cycle accompanied by changes in noradrenergic and domaminergic activity before menstruation. In another study, the same group has even shown that stress (psychological and physical threat) and the menstrual cycle have a potentiating effect on the CNV in healthy subjects (43). However, this study did not test the additive influence of stress and menstruation in migraineurs, the question discussed in the present article.
In summary, this study demonstrates the additive effect of stress and menstrual cycle on the amplitude of the iCNV. It could be suggested that migraine-precipitating agents influence the disposition to a migraine attack through changes in cortical excitability and information processing. The study emphasizes that our knowledge of mechanisms of migraine-precipitating agents are insufficient and that the investigation of these mechanisms could provide new information about the pathophysiology of migraine and open new possibilities of migraine prophylactic treatment.