Chirp stimulation: H-response short and dynamic

Introduction

Migraine is disabling headache, usually associated with nausea and sensory hypersensitivity, affecting up to 15% of the population. In 1959, Golla and Winter described a test to discriminate between patients with episodic disabling headaches and headache-free controls (1) using a custom-built amplifier to quantify photic driving responses. When they were recording EEG during steady-state stimulation from 10 to 24 Hz, they found that the EEG of the headache patients was driven to frequencies above 18 Hz (‘H-response’), whereas in healthy controls was limited to about 14 Hz. More than 40 years later Chorlton and Kane (2) confirmed the high specificity and relatively high sensitivity of the ‘H-response’ test, applying the classification criteria for migraine of the International Headache Society. A review of the literature regarding electroencephalographic research in migraine showed the prominent photic driving response at high flash rates in migraine patients to be consistently reported (3). Only a recent study by Bjork et al. showed conflicting results and suggested that the driving response of migraine patients might be overestimated because of altered cortical information processing depending on the temporal proximity to migraine attack, as well as the influence of habituation among controls (4).

We studied the H-response applying a short, fast and dynamic stimulus, which might be used in the clinical setting, or as a tool for acute interventional or longitudinal studies in migraine.

Methods

Subjects

This study has been approved by the joint research ethics committee of the Institute of Neurology and the National Hospital for Neurology and Neurosurgery, Queen Square, London, UK. Written informed consent was obtained from all participants. All subjects were devoid of any other medical condition and had no personal or family history of epilepsy. In regard to the high frequency of tension-type headache in the general population, it was not possible to find enough control participants who never had any headaches in their lives.

There were six women and five men with migraine without aura with a mean age of 30.9 ± 8.5 years. They had between one and six attacks per month (mean 1.6 ± 2.0). The mean duration of disease was 12.4 years. None of them had received prophylactic anti-migraine medication for at least 3 months before the study, nor used ergotamines or triptans as acute treatment. In the group of healthy control participants, we had five women and six men with a mean age of 28.3 ± 4.5 years, which was statistically not different from the patient group (Table 1).

OPEN IN VIEWER

Table 1.

Characteristics of migraine patients and controls without migraine headaches.

Subject	Age	Sex	Age of onset	Headaches per month	Duration (hours)	Severity (VRS)
Migraine	1	23	F	12	2	10	6
2	21	F	18	0.25	12	5
3	22	M	15	0.2	24	7
4	31	F	18	0.5	3	6
5	37	F	20	0.1	3	7
6	27	F	16	5	12	4
7	35	M	16	0.2	6	5
8	48	M	28	1.5	4	7
9	26	F	16	0.25	24	8
10	29	M	29	0.1	8	7
11	41	M	16	6	12	4
Mean	30.91	6/5	18.55	1.46	10.73 a	6.00 a
Control	1	36	F	0.1	3	3
2	28	M	1	4	4
3	25	M	0	0	0
4	34	M	1	1	1
5	27	F	6	2	2
6	24	M	2	3	3
7	25	M	0	0	0
8	25	F	0.5	1	3
9	26	F	0.2	3	4
10	35	F	0.4	0.5	5
11	26	M	4	3	2
Mean	28.27	5/6	1.38	1.86 a	2.45 a

a

Significant group differences. VRS: verbal rating scale.

All recordings were made in the headache-free interval, at least 3 days before and after an attack. This was checked by personal, phone or email communication.

Data analysis

EEG recordings were processed using custom written MATLAB programs. For the analysis we used a total of 12 seconds per sweep, with 3 seconds baseline before the chirp-stimulation and 3 seconds afterwards.

The HR-estimator was calculated as following (see also Figure 2):

For every single sweep of 12 seconds, power spectral density (PSD) was calculated in the time/frequency window, using fast fourier transformation (FFT), with a window of 128, an overlap of 64 and a resolution of 128.

The PSD data of every single sweep was corrected for baseline (averaged 3 seconds before the stimulation).

Thereafter the power in a predefined time window of the sweep (time between the four flashes plus 200 ms) was averaged for every stimulation frequency from 10 to 40 Hz to receive a single power estimate.

The HR-estimator was defined as the average EEG power between 18 and 26 Hz for every single sweep.

OPEN IN VIEWER

Figure 2.

The figure shows the calculation of the HR-estimator in a clockwise mode. The time-frequency analysis of a single sweep of chirp stimulation was corrected for baseline. An occipital EEG recording for a single sweep of 12 seconds (baseline 3 seconds, stimulation 6 seconds, post-stimulation 3 seconds) of a typical migraine patient. In a next step the power in a predefined time window of the sweep was averaged for every stimulation frequency from 10 to 40 Hz to receive a single power estimate (2D).

Results

In all participants, photic driving could be elicited. Looking at the EEG power spectrum, in the range from 10 to 18 Hz, photic driving was similar in both groups, with an amplitude difference when compared to baseline of 3 dB. In the range from 19 to 26 Hz, the power spectrum was markedly different, with a more pronounced photic driving in this frequency range in the migraine group, and a trend towards more photic driving response above 26 Hz. There was a significant group difference for the HR-estimator between migraineurs and control participants (p < 0.05). The HR-estimator was reliable over 10 repetitions with a Cronbach’s alpha of 0.94 (n of cases = 22, n of items = 10).

Discussion

In this study we used a novel type of stimulus design to elicit an H-response, applying flash stimuli with increasing frequencies from 10 to 40 Hz. We were able to discriminate migraine patients from healthy volunteers, finding a higher photic-driving power in migraineurs in the lower beta-frequency range. The 6-second ‘chirp response’ proved to be reliable and, in contrast to the previous types of steady-state stimulation, may be administered efficiently in clinical or research setting. It adds another information processing aspect than the well described habituation deficit to the study of brain function in migraine (5). In line with previous studies, our data showed the highest discriminating power in the frequency range of 20 to 25 Hz.

Currently, the gold standard to diagnose migraine is the clinical information in the framework of the International Classification of Headache Disorders. However, the sensitivity (86.4%) and specificity (97.5%) in the study of Chorlton and Kane was good for the identification of migraine with and without aura, but not for other migraine subtypes (2). The discriminating value of the H-response between migraine and tension-type headaches has been questioned by de Tommaso et al. who studied 120 migraineurs, 64 tension-type headache patients and 51 controls (5). In their study there was no difference of the mean amplitude of the first harmonic elicited by flash stimulation in the 15 to 27 Hz range between different types of headache. They suggested a common neuronal dysfunction in the two studied headache subtypes, or poor discrimination using ICHD-II criteria. Allowing tension-type headaches in our control group, we still had a significant group difference in the photic-driving response in the low beta frequency range, suggesting different underlying information processing in the groups we studied. We do not have any reliability data for repeated testing on different days. However, we found good intra-individual reliability of the H-response within the same day of recording. Therefore ‘chirp stimulation’ may be a valuable tool for interventional research with longitudinal design.

In regard to confounding factors, all subjects were at least 3 days before or after an attack, what minimizes the influence of the interictal state on information processing characteristics (6). As photic-driving responses decrease with age (7), this should not influence the interpretation of our results, where migraine patients were slightly older than the control participants. The sex differences are expected to be small, although higher photic-driving is reported in women, depending on their hormonal state (8).

Using our stimulus design, it was possible to avoid interference with habituation processes that have been shown to be abnormal in visual processing in migraineurs (9). In analogy to the interictal lack of habituation the photic driving response of migraineurs may be explained by abnormal cortical information processing. As a result of the dynamic design of the chirp stimulus the brain is forced to react with fast adaptive changes. This makes it less likely to be affected by habituation processes, although based in revised concepts of habituation it cannot be excluded (10). Angelini et al. investigated phase synchronization in EEG recordings from migraine patients (11). They found enhanced alpha band phase synchronization in the presence of visual stimuli and concluded that migraine patients have an overactive regulatory mechanism to external stimuli. In a recent study we showed that amplitudes of single trial visual evoked potentials were explained almost entirely by increases in local amplitude, implying a higher energy demand (12). Other authors showed a variation of evoked potentials towards a migraine attack (13).

These findings may suggest that the migraine brain is ‘running at higher levels’ when stimulated interictally and therefore might consume more energy. The normalization of both photic driving and evoked potentials during the attack may point towards a kind of ‘valve’ mechanism of the migraine attack itself.

Clinical implications

Migraine is a clinically relevant disorder, affecting up to 15% of the general population.