Abstract
A dysbalance of the cerebrovascular response during functional activation of the brain has been postulated as a factor in the pathophysiology of migraine. To determine the dynamic pattern of the cerebrovascular response in migraineurs compared with a control group, changes of the cerebral perfusion during cerebral activation were studied with high temporal resolution by functional transcranial Doppler sonography (fTCD). The cerebral blood flow velocity (CBFV) in the right posterior cerebral artery (PCA) and the left middle cerebral artery (MCA) was measured simultaneously during visual stimulation in 19 interictal migraineurs and in 19 age- and sex-matched control subjects. Data were analysed with a previously validated technique based on automated stimulus-related averaging of the CBFV. The MCA migraineurs exhibited a steady increase of CBFV during the stimulation, while normal subjects showed a habituation of the CBFV response. The lack of habituation in migraineurs was significantly (P ≤ 0.05) more pronounced across patients with a high attack frequency (≥ 4 per month) compared with migraineurs with a low attack frequency (< 4 per month). In the PCA, compared with normal subjects, migraineurs showed significantly (P ≤ 0.05) stronger CBFV changes at the beginning and after the end of stimulation, with a slower decline to baseline. Data are in accordance with electrophysiological findings in migraineurs. It is assumed that a lack of habituation of the cerebrovascular response in migraineurs might contribute to a disturbance of the metabolic homeostasis of the brain that might induce migraine attacks.
Introduction
Increased metabolic demands during cerebral activation lead to arteriolar vasodilatation, prompting an increase of blood flow velocity (CBFV) in the basal cerebral arteries. Functional transcranial Doppler sonography (fTCD) has been shown to render this information about cerebral activation (1–8).
The recent concept of the pathophysiology of migraine implies a dysfunction of the cerebrovascular coupling (9). The cerebrovascular response in interictal migraine during cerebral activation in cognitive, motor and visual paradigms has been previously studied by means of fTCD. Results, however, are heterogeneous, showing a stronger or unaltered cerebrovascular response in migraineurs compared with normal subjects (10, 11). This might be due to a lack of temporal resolution applied in previous studies. As the cerebrovascular coupling is dynamic (12, 13) the dysfunctional features of the cerebrovascular response in migraine might be phasic in nature and thus overlooked when the time-pattern of cerebral perfusion changes is not investigated during the whole time of stimulation. Previous studies have shown the ability of fTCD to study the time-course of cerebrovascular responses with a high temporal resolution (4, 5). In the present study we employed fTCD to study the dynamic pattern of the functional activation of the cerebrovascular system during visual stimulation in migraineurs in the interictal phase and in normal subjects. In order to investigate the cerebrovascular response due to activation of the visual cortex, the CBFV was measured in the posterior cerebral artery (PCA). As migraineurs show a higher cortical excitability in the interictal phase (14), changes of CBFV were simultaneously monitored in the middle cerebral artery (MCA) to gain information about more global changes of the cerebral perfusion.
Methods
Subjects
The study was performed with 19 migraine patients (22–63 years of age) from the outpatient pain clinic of the Technical University Munich and with an age- and sex-matched group of 19 healthy subjects. All subjects participated in the study after giving informed consent. Data from the study population are shown in Table 1. All migraineurs had at least a 12-month history of migraine according to the diagnostic criteria of the International Headache Society (15). The frequency of migraine attacks was monitored by headache diaries kept by the patients over a period of 3 months before the Doppler measurement. The mean attack frequency was 3.9 per month. Patients with (n = 4) and without (n = 15) migraine prophylaxis were included in the study. The migraineurs were studied during the headache-free interval, with at least 3 days since the last migraine attack.
Descriptive data of observed migraineurs
Experimental design
The examination was performed in a darkened, quiet room and subjects were comfortably sitting in an armchair. They were requested to keep their eyes opened during the experiment. Light stimulation was performed with a light-proof stimulating goggle used in clinical routines for visual evoked potentials containing multiple-array light-emitting diodes applying flash stimuli with a frequency of 14 Hz. The stimulation period lasted for 57 s, followed by a resting phase of 57 s without any stimulation. In every subject three on–off cycles were performed after an initial adaptation phase of 5 min without any stimulation. A commercially available dual transcranial Doppler ultrasonography device (Multi-Dop X, DWL, Sipplingen, Germany) was used for the recording of the CBFV. Two transducer probes (2-MHz) were attached to a headband and placed at the temporal bone windows bilaterally. During the stimulation the CBFV was measured in the P2 segment of the right PCA and in the left MCA simultaneously. The PCA was insonated at a depth of 66 mm, the MCA was insonated at 54 mm. For technical details, particularly the correct identification of the arteries, see Conrad & Klingelhofer (3). The spectral envelope curves of the Doppler signal were recorded with a rate of 28 sample points per second. Additionally, arterial blood pressure was recorded non-invasively according to the Penaz method (Finapress, Ohmeda, USA) (16).
Analysis
The analysis was performed off-line with the analysis-software AVERAGE (7). Heart rate was calculated by a peak analysis of the envelope curve of the Doppler frequency spectrum in the raw data. The data were segmented into epochs that related to the stimulation. The epochs were set from 20 s before the onset of light stimulation to 35 s after the end of stimulation. The mean velocity in the 20-s pre-stimulation-interval (Vpre.mean) was taken as the baseline value. The relative CBFV changes (dV) during cerebral activation were calculated by the formula: dV = (V(t) − Vpre.mean)∗100/Vpre.mean where V(t) is the CBFV over the course of time (5). To reduce modulations of the CBFV due to the pulsatility of the TCD velocity curves, they were integrated over the corresponding heart cycles and replaced by a step function (7). Finally data were averaged time-locked to the stimulation for single subjects and for all subjects of the respective group (n = 19) (grand average). Additionally the patients were split into groups according to their frequency of attacks and their side preference of the headache. For the high frequency group, patients with a frequency of ≥ 4 attacks per month were accepted. The low-frequency-group consisted of patients with an attack frequency of < 4 attacks per month. Data for the groups were then analysed separately.
The velocity values were averaged for each time-point for MCA and PCA and evaluated for normality distribution by the Kolmogorov-Smirnov test. Differences between groups were analysed by the two-tailed Students t-test for unpaired samples. CBFV differences within the PCA and MCA were analysed by the t-test for paired samples (17).
Describing the averaged velocity curve by M data points the t-test was repeated for all sampled time points ti of the epoch (i = 1,2, …. M) (7). For this procedure the assumption was made that the velocity values in different epochs are statistically independent for the latency ti. Differences of dV between PCA and MCA in each group and differences in dV in the respective artery between the two groups were classified as significant at P ≤ 0.05. To account for the Bonferoni problem only time periods of significant differences longer than 1 s were accepted. All values are presented as mean ± standard error of the mean (
Results
During visual stimulation there was a relative increase of perfusion compared with baseline in the PCA and the MCA in migraineurs and normals across subjects. Perfusion increases were significantly (P < 0.01) stronger in the PCA than in the MCA in both groups.
Cerebrovascular response of the MCA in migraineurs and normals
In both groups an initial increase of CBFV could be observed during the first 15 s of the light stimulation. Across normal subjects the CBFV then habituated and remained on baseline values during the rest of the stimulation. In contrast, CBFV of the MCA across migraineurs did not habituate, but exhibited a steady increase of CBFV from the first peak increase to 10 s after the light was off. This perfusion pattern was not paralleled by changes in the arterial blood pressure, which in both groups showed an initial, transient decrease of 2% across subjects 10 s after the beginning of stimulation and remained on baseline values during the rest of the stimulation (Fig. 1b).
Grand average of the cerebrovascular response in the right PCA (a) and the left MCA (b) in migraineurs and normals. The dotted line (.......) represents the relative increase of CBFV in migraineurs (n=19), the solid line (––) represents the relative increase of CBFV in normal subjects (n=19). Thick vertical bars show onset and offset of the light stimulation. Hatched areas reflect significant (P < 0.05) differences between the curves of migraineurs and normal subjects. The data represent the grand average of 19 subjects.
The differences between the two groups reached statistical significance (P < 0.05) from 38 to 67 s post-stimulation onset. The maximum of the perfusion increase across normal subjects was 2.2% (± 0.09). In migraineurs the initial increase of CBFV was significantly stronger with 3.6% (± 0.11). A second peak of CBFV (3.2%, ± 0.17)) was reached 2 s after the light was off.
Cerebrovascular response of the PCA in migraineurs and normals
A distinct modulation pattern of the CBFV of the PCA could be observed in both groups. This pattern consisted of an initial peak of CBFV followed by a plateau-phase until the end of the stimulation and a final undershoot of the decline after the light was switched off (Fig. 1a).
The comparison of CBFV changes of the PCA between migraineurs and healthy subjects showed significant (P ≤ 0.05) differences in the cerebrovascular response at the initial peak and after the end of the stimulation, while during the plateau-phase (15–58 s post-stimulation onset) no significant differences could be detected.
While CBFV across normal subjects showed an initial maximum increase of 7.2% (± 0.16), the initial change of CBFV in the migraineurs was slightly but significantly (P < 0.05) stronger, with a maximum increase of 8.5% (± 0.18). The mean increase of CBFV during ongoing stimulation in the plateau-phase was 5.4% (± 0.14) in migraineurs and 5.0% (± 0.16) in normal subjects. After the end of stimulation across normal subjects the CBFV remained on the plateau level for a further 4 s before declining to baseline. In the migraineurs a second increase of CBFV could be observed with a maximum of 6.8% (± 0.18) followed by a slower decline of CBFV compared with normal subjects after the light was switched off. While the CBFV across normal subjects returned to baseline within 10 s (± 0.32) after the end of stimulation, the CBFV across migraineurs returned to baseline within 14.3 s (± 0.27) after the light was switched off.
Migraineurs with high and low attack frequency
Group-analysis of migraineurs with high and low attack frequency showed significant differences between the two groups. The lack of habituation in the MCA was significantly more pronounced in the high-frequency group compared with the low-frequency group (Fig. 2b). Significant (P ≤ 0.05) differences in the cerebrovascular response could mainly be observed during the second half of the stimulation (38–50 s post-onset) and after the end of the stimulation (62–70 s post-onset).
Grand average of the cerebrovascular response in migraineurs with high (≥ 4 per month) and low (< 4 per month) frequency of migraine attacks. (a) shows the relative increase of the CBFV in the right PCA, (b) shows the CBFV modulation of the MCA. Dotted lines (.......) represent perfusion changes in the low-frequency group, solid lines (––) represent perfusion changes in the high-frequency group. Thick vertical bars show onset and offset of the light stimulation. Hatched areas reflect significant (P ≤ 0.05) differences between the curves of the two groups.
Similarly, in the PCA (Fig. 1a) there was a stronger increase of the CBFV mainly during the second half of the stimulation period in the high-frequency group compared with the low-frequency group. This difference was statistically significant (P < 0.05) for the period of 40–51 s post-stimulation onset. Other characteristics of the CBFV response, including the initial overshoot, the post-stimulation undershoot and the decline to baseline, did not differ significantly between the two groups.
Side preference of the headache
In the group analysis there was no significant difference in the relative changes of the CBFV in PCA and MCA, whether patients had a right- or a left-sided preference of their headache.
Discussion
The aim of the study was to compare the dynamic pattern of the cerebrovascular response to visual stimulation in migraineurs and healthy subjects. Our data reveal that migraineurs compared with normal subjects exhibit a different cerebrovascular reaction across subjects. In the MCA, migraineurs exhibited a steady increase of CBFV until the end of stimulation, while in normal subjects a habituation of the CBFV response could be observed after an initial, transient increase. This lack of habituation was more marked across migraineurs with a high frequency of attacks compared with those with a low attack frequency.
For the PCA, perfusion differences could be observed in the beginning and at the end of the stimulation. This was reflected by an initial stronger overshoot, a stronger post-stimulus peak, as well as a delayed decline of CBFV. During the ‘plateau-phase’, however, there was no significant difference between the two groups.
The observed dynamic pattern of perfusion changes might explain why previous studies testing the cerebrovascular response of the PCA in migraineurs showed divergent results depending on the temporal resolution that had been applied. Since the differences of PCA-perfusion increases between migraineurs and normals are transient in nature, they might have been overlooked when calculating mean changes of CBFV (11). However, if peak values are taken into account migraineurs show a significant stronger cerebrovascular response (10). In contrast to other studies (18) we could not find a statistically significant influence of the headache-side preference on the cerebrovascular response. This might be due to the relatively small study population. Additionally, comparing CBFV in the left and the right PCA might be more sensitive than comparing perfusion changes in the PCA and the MCA.
Compared with previous findings the observed changes of CBFV in the present study were small. This might be due to the simple stimulation paradigm (flash stimuli) compared with more complex tasks used in previous studies (3, 19). It is remarkable that despite small overall changes of CBFV there were statistically significant differences between the cerebrovascular response in migraineurs and normal subjects. Since a more complex visual task leads to a higher increase of CBFV (3), perfusion changes in migraineurs might show more pronounced abnormalities with a higher complexity of the perceptual surrounding as experienced in daily living. This could be important for diagnostic purposes. In the present data there was a high inter-individual variability of the CBFV changes, which caused a broad overlap of the individual responses between the two groups. Thus there is a low discriminative power for screening single subjects. Further studies will be performed to investigate if the discriminative value of fTCD might be enhanced when using more complex paradigms.
Changes in the CBFV are not only determined by regional changes in cerebral activation, but also by autonomous variables. Thus it can not be excluded that breathing had an influence on global changes in the CBFV. However, since heart rate and arterial blood pressure did not parallel the CBFV patterns, it is not likely that changes in the vegetative state were a major confounding factor.
Functional evoked changes of the CBFV during the activation of the brain measured by fTCD are comparable with results gained by other functional imaging methods such as functional magnetic resonance imaging (fMRI) (20). It can therefore be assumed that perfusion changes seen in the present study are at least in part due to changes in the cerebral activation. In this case another explanation for the different cerebrovascular response in migraineurs and healthy subjects could be an altered processing of the sensory input, with changes of the CBFV-response as an epiphenomenon of an underlying altered neural substrate in migraineurs.
Migraineurs show a lack of habituation to repetitive stimuli in event-related potentials (21). This phenomenon tends to increase during the interictal interval and normalizes after the migraine attack. Additionally migraineurs show a higher cortical excitability in the interictal phase (22). These electrical abnormalities have been interpreted as a dysfunction of cerebral information processing resulting from a high level of cortical arousal in migraineurs (23).
Data obtained from electrophysiological measurements of the brain and cerebral perfusion data have to be compared carefully. Possible explanations, however, for the stronger CBFV changes in the MCA could be a higher arousal or an altered visual processing in migraineurs, since the MCA supplies wide areas associated with non-visual and some areas involved in complex visual processing. Moreover, the lack of habituation of the CBFV seen in the MCA in migraineurs is in accordance with the hypothesis of a perceptual dishabituation postulated from the ERP findings. Since this dishabituation appeared to be more pronounced in migraineurs with a high attack-frequency it might be a predisposing pathophysiological factor which fosters a high frequency of migraine attacks.
The altered cerebrovascular response observed in migraineurs, either caused by an altered perceptual processing or by a dysbalanced reaction of the vessel diameter to vasoactive stimuli, might contribute to a metabolic strain on the brain, which is thought to play a role in triggering migraine attacks (24). A disturbance of the metabolic homeostasis cumulating over days may induce a migraine attack via the activation of the trigeminovascular system (9). The spreading depression and corresponding changes of the cerebral perfusion found in the headache phase might restore the homeostasis of the system. It will be interesting to see if the altered cerebrovascular response in migraineurs observed in the present study tends to normalize after a migraine attack. Additionally, monitoring of the cerebrovascular response in migraineurs under prophylactic treatment might provide insight into the mode of action of the respective treatment modality. Preliminary data show a normalization of the cerebrovascular response under a prophylactic treatment with acupuncture.
Footnotes
Acknowledgements
M. Bäcker was sponsored by the Deutsche Ärztegesellschaft für Akupunktur (DÄGFA). The authors would like to thank Mr Artur Schikowski for his support in data-acquisition.