Resting cranial and upper cervical muscle activity is increased in patients with

Archived at the Flinders Academic Commons:

http://dspace.flinders.edu.au/dspace/

‘This is the peer reviewed version of the following article:

Janani, A. S., Pope, K. J., Fenton, N., Grummett, T. S., Bakhshayesh, H., Lewis, T. W., … Willoughby, J. O. (2018).

Resting cranial and upper cervical muscle activity is increased in patients with migraine. Clinical

Neurophysiology, 129(9), 1913–1919. https://

doi.org/10.1016/j.clinph.2018.06.017 which has been published in final form at https://doi.org/10.1016/j.clinph.2018.06.017

© 2018 International Federation of Clinical Neurophysiology. Published by Elsevier B.V.

This manuscript version is made available under the CC- BY-NC-ND 4.0 license:

http://creativecommons.org/licenses/by-nc-nd/4.0/

Accepted Manuscript

Resting cranial and upper cervical muscle activity is increased in patients with migraine

Azin S. Janani, Kenneth J. Pope, Nicole Fenton, Tyler S. Grummett, Hanieh Bakhshayesh, Trent W. Lewis, Dean H. Watson, Emma M. Whitham, John O.

Willoughby

PII: S1388-2457(18)31139-8

DOI: https://doi.org/10.1016/j.clinph.2018.06.017

Reference: CLINPH 2008582

To appear in: Clinical Neurophysiology Accepted Date: 13 June 2018

Please cite this article as: Janani, A.S., Pope, K.J., Fenton, N., Grummett, T.S., Bakhshayesh, H., Lewis, T.W., Watson, D.H., Whitham, E.M., Willoughby, J.O., Resting cranial and upper cervical muscle activity is increased in patients with migraine, Clinical Neurophysiology (2018), doi: https://doi.org/10.1016/j.clinph.2018.06.017

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

1

Resting cranial and upper cervical muscle activity is increased in patients with

migraine

Azin S. Janani MSc^a,b, Kenneth J. Pope PhD^a,b , Nicole Fenton MD^c, Tyler S. Grummett BBSc^a,b,d, Hanieh Bakhshayesh BEE^a,b, Trent W. Lewis PhD^a,b, Dean H. Watson PhD^e, Emma M. Whitham FRACP, PhD^c, John O. Willoughby FRACP, PhD^c,d,*

a College of Science and Engineering, Flinders University, Adelaide, Australia

b Medical Device Research Institute, Flinders University, Adelaide, Australia

c Department of Neurology, Flinders Medical Centre, Adelaide, Australia

d Centre for Neuroscience, College of Medicine and Public Health, Flinders University, Adelaide, Australia

e Watson Headache Institute, Adelaide, Australia

*John O. Willoughby: Corresponding author at: Centre for Neuroscience, College of

Medicine and Public Health, Flinders University, Adelaide, GPO Box 2100, Adelaide 5001, South Australia. Tel: +618- 82044187. E-mail address: john.willoughby@flinders.edu.au

2 Abstract

Objective: To compare comprehensive measures of scalp-recorded muscle activity in migraineurs and controls.

Method: We used whole-of-head high-density scalp electrical recordings, independent component analysis (ICA) and spectral slope of the derived components, to define muscle (electromyogram-containing) components. After projecting muscle components back to scalp, we quantified scalp spectral power in the frequency range, 52-98 Hz, reflecting muscle activation. We compared healthy subjects (n=65) and migraineurs during a non-headache period (n=26). We also examined effects due to migraine severity, gender, scalp-region and task (eyes-closed and eyes-open). We could not examine the effect of pre-ictal vs inter-ictal vs post-ictal as this information was not available in the pre-existing dataset.

Results: There was more power due to muscle activity (mean ± SEM) in migraineurs than controls (respectively, -13.61 ± 0.44 dB vs -14.73 ± 0.24 dB, p=0.028). Linear regression showed no relationship between headache frequency and muscle activity in any combination of region and task. There was more power during eyes-open than eyes-closed (respectively, - 13.42 ± 0.34 dB vs -14.92 ± 0.34 dB, p=0.002).

Conclusions: There is an increase in cranial and upper cervical muscle activity in non-ictal migraineurs versus controls. This raises questions of the role of muscle in migraine, and the possible differentiation of non-ictal phases.

Significance: This provides preliminary evidence to date of possible cranial muscle involvement in migraine.

Keywords: quantification; cranial muscle activity; migraine; scalp electrical recordings;

Independent Component Analysis.

3 Highlights

• We applied a new method for quantifying electromyogram in scalp recordings.

• We measured scalp muscle activity in headache-free migraine patients and in controls.

• Scalp muscles exhibited more activity in migraineurs than in controls.

4

1. Introduction

The International Classification of Headache Disorders (ICHD) considers muscle to be relevant in tension-type headache, but does not address its relevance in migraine (Headache Classification Committee of the International Headache, 2013). There have been many studies since the 1970s examining the role of muscle in migraine as well as in tension-type headache, mostly qualitative (Bakal et al. , 1977, Tfelt‐Hansen et al. , 1981, Bakke et al. , 1982, Clifford et al. , 1982, Lous et al. , 1982, Ahles et al. , 1988, Celentano et al. , 1990, Lebbink et al. , 1991, Jensen et al. , 1993, Blau et al. , 1994, Burnett et al. , 2000, Hagen et al.

, 2002, Ebinger, 2006, Leistad et al. , 2006, Fernández‐de‐las‐Peñas et al. , 2008, Hung et al. , 2008, Oksanen et al. , 2008, Blaschek et al. , 2012, Watson et al. , 2012, Didier et al. , 2015, Landgraf et al. , 2015). Their conclusions disagree, but many do conclude there is a link between migraine and muscle activation. Here we limit ourselves to reviewing quantitative studies of migraine.

While many studies have focussed on tension-type headache, they sometimes have included migraine groups for comparison. The methods of quantitation of muscle activity in these studies differ, such that it is difficult to make robust comparisons and identify a clear conclusion. There are differences in recording electromyogram (EMG) (surface or needle recordings), differences in muscles sampled, differences in headache phase (during inter- headache or headache periods) and differences in activity state (at rest or during instructed contraction or head postures). The extracted measures of electromyographic activity include median or mean frequency and root mean square power.

The findings have not pointed to a consistent alteration in muscle activity. Bakal et al., (1977) reported migraine patients had higher frontalis EMG as well higher neck EMG activity than tension-type headache patients and headache-free controls. McArthur et al., (1980) reported

5

migraine patients had higher frontalis EMG activity than tension-type headache patients and headache-free controls. Anderson et al., (1981) reported that frontalis EMG does not

distinguish between tension-type and migraine headaches. Clifford (1982) reported during attacks, that migraineurs had activity in the anterior temporal muscles which exceeded the patient's own baseline recordings and that all muscles were activated more strongly than in the control period. Similarly, Bakke (1982) reported a rise in activity from control levels shortly before migraine patients experiencing maximal pain. Ahles et al., (1988) reported higher levels of muscle activity (frontalis, trapezius) in three headache groups, including migraineurs, than in the non-headache group, but which did not differ from each other.

Jensen et al., (1994) could not identify EMG measures corresponding to ‘migraine severity in the previous year’ (though increased EMG measures were seen in patients with ‘chronic headache’), and there was no relationship between muscle activity and migraine generally.

Given modern methods of signal analysis, it is now possible to extract EMG activity and other contaminants (such as electrooculogram (EOG), electrocardiogram (ECG), mains noise, white noise and other artefacts) from scalp electrical recordings, usually with the aim of extracting clean electroencephalogram (EEG). Using such methods, therefore, it is equally feasible to obtain ‘clean’ EMG and accurately quantitate power corresponding to EMG activity. We now report a topographic quantitative study on cranial and cervical muscle activity in a group of migraine sufferers and headache-free controls using recordings from an EEG cap with 128 electrodes covering the scalp. The cap included electrodes over muscles such as the frontalis, orbicularis and temporalis, and also close to nuchal (upper cervical) muscles. At frequencies above 10-20 Hz, EMG signals are the largest contributors to scalp electrical recordings, with the highest power typically in the range 60-90 Hz (Goncharova et al. , 2003, Whitham et al. , 2007, Whitham et al. , 2008, Zhang et al. , 2015). Recent research indicates that this is true even when the subject is relaxed and phasic muscle has been excised

6

from the recordings (Whitham et al. , 2007, Whitham et al. , 2008, Pope et al. , 2009, Yilmaz et al. , 2014) .

A standard approach in identifying components in electrical scalp recordings is to use Independent Components Analysis (ICA) to reveal components that correspond to separate, different sources of electrical activity, e.g. neurogenic, myogenic and others (Makeig et al. , 1996, Vigário et al. , 2000, Delorme et al. , 2007, Prakash et al. , 2016). The artefactual components (e.g. non-neurogenic sources if analysing EEG) are then discarded, and “clean”

scalp EEG recordings can be reconstructed. In this study, we utilise the myogenic sources from ICA, enabling us to quantitate muscle activation in migraineurs versus headache-free controls. We then tested for differences in the level of muscle activity due to gender (female and male), region (frontal, left-temporal, central, right-temporal, occipital), task (eyes closed and eyes open), and condition (migraine, headache-free control). Finally, we tested for a linear relationship between muscle activity and headache severity.

2. Subjects and Procedure

2.1. Subjects

The data used in this study was drawn from an existing dataset collected from participants with a range of neurological and psychological disorders and controls, whose purpose was to investigate changes in brain rhythms with disease. Both studies were approved by the Clinical Research Ethics Committee of Flinders Medical Center and Flinders University, application number: OFR # 382.13, and each participant signed a consent form. All participants were recruited from the clinics and staff of the Flinders Medical Center, or their relatives, between 2004 to 2007. All patients were evaluated by a neurologist and those with a single neuro- psychiatric diagnosis were included. A power analysis for the original study recommended recruiting 10 migraine participants with aura and 10 without. Additionally, some participants

7

recruited as controls were diagnosed as migraineurs after the initial medical examination by a neurologist. For this study, we recorded sufficient clinical information for accurate diagnosis;

we did not use migraine diaries, so the full suite of migraine expression was not known. In addition, while all patients were without headache on the day of recording, we do not know how long they remained headache-free. Diagnosis was validated by another neurologist using the 2013 ICHD-III-beta diagnostic criteria (Headache Classification Committee of the

International Headache, 2013), based on review of their records.

This resulted in a dataset consisting of two groups: 65 healthy participants with no history of headache and 26 migraine participants. Table 1 shows the demographic details of participants which shows gender distributions for migraineurs and controls that are close to the population expectations of 67% and 50% female. Given the inter-individual variability of muscle

activity, we chose to include the maximum number of participants. The migraine participants all described their pain intensity as three or four out of five (moderate or severe), mean intensity of 3.9, and the maximum number of attacks per year was 104. 50% had migraine with aura. About 70% of the migraine participants had a frequency of one or two attacks in a month, and the mean frequency was 0.9 per month. 11 participants reported migraines lasting for a few hours, and 15 reported durations of a few days.

2.2. Migraine Severity

We characterized severity using three measures: duration, frequency and intensity. The patient-estimated duration was quantised to hours or days. Patient-estimated frequency was input as a count per year. Patient-estimated intensity was quantised to the range 1-5, using descriptors dull, mild, moderate, severe or excruciating.

8 2.3. Data Collection

Scalp electrical activity was recorded from each participant while sitting, head un-supported, in a stable 23-24 C temperature, and completing a series of tasks including: eyes closed, eyes open, auditory and visual discrimination, visual rotation, finger tapping, maze, serial

subtraction, auditory verbal learning task, and reading (Whitham et al. , 2008). One

technician applied the electrodes and undertook recordings. No adaptation time was provided, nor judged to be necessary: the recording environment was shown to participants before the electrode cap was applied – they knew where they would be sitting. Instructions were both verbal and written (using a computer-based bio-behavioural instruction program).

Participants typically took 22 minutes to complete all the tasks. The sampling frequency was 2000 Hz.

2.4. Pre-processing

EEG electrodes were labelled according to the 10-5 international system (Oostenveld et al. , 2001). Data from each subject were resampled to 500 Hz and passed through a high-pass filter with a cut-off frequency of 0.5 Hz to attenuate electrode drift. Transient, high-

amplitude, phasic muscle activity was marked and excluded from analyses. All processing of the data was performed using Matlab toolbox.

2.5. Quantitating Cranial and Upper Cervical Muscle Activity

We used a method from Fitzgibbon and colleagues who showed that muscle components have an increasing spectral slope between 7-75 Hz, whereas the spectral slope of brain components decreases in this frequency band (Fitzgibbon et al. , 2016). We are exploiting this observation to distinguishing muscle components from brain components automatically, as follows:

 Scalp measurement (EEG) during all tasks in each subject was subjected to ICA (Infomax (Bell et al. , 1995)) to provide components and a mixing matrix.

9

 For the resting tasks (eyes closed and eyes open), the spectral gradient of each

component was calculated by fitting a straight line to the log-log spectrum between 7 Hz and 75 Hz (logarithmic power vs logarithmic frequency, units of bel/decade).

 Components with a gradient greater than a specific threshold (-0.31 bel/decade) (Fitzgibbon et al. , 2016) were identified as muscle-containing components and were kept while the remaining components were discarded.

 We then reconstructed surface EEG for the two resting tasks (eyes closed and eyes open) using the preserved muscle components and the mixing matrix.

Thus, the power of reconstructed EEG at each electrode location determines the contribution of cranial and upper cervical muscle activity to that specific location. We separated our EEG channels in to five groups based on their location in the 10-5 system as shown in Figure 1.

Frontal Fp1, Fpz, Fp2, AFp1, AFp2, AF7, AF3, AFz, AF4, AF8, F3, F1, Fz, F2, F4.

Left temporal FT9, FT7, FTT9h, TPP9h, T9, T7, TP9, TP7.

Central FC1, FCz, FC2, FCC1h, FCC2h, C1, Cz, C2, CCP1h, CCP2h, CP1, CPz, CP2.

Right temporal FT10, FT8, FTT10h, TPP10h, T10, T8, TP10, TP8.

Occipital O1, Oz, O2, O11h, O12h, PO9, I1, Iz, I2, PO10.

The electrodes in the ‘Frontal’ region are closest to and would record activity predominantly from frontalis, procerus, and orbicularis oculi, ‘Temporal (left and right)’ from temporalis and superior auricularis, ‘Occipital’ from occipitalis as well as trapezius, splenius capitis and sub-occipital muscles, and ‘Central’ would record activity volume conducted from remote muscles. We then calculated the average power spectrum in dB (zero reference level 0 dB = 1 (uV)²/Hz) for each region.

10 2.6. Power Spectra

Welch’s modified periodogram with one-second Hanning windows was implemented to compute the power spectra of components and also of reconstructed scalp signals for the resting tasks. To estimate the level of muscle activity, we chose the average power (dB) in the band 52-98 Hz. This band, which we refer to as the “muscle band”, covers the frequencies where muscle power is highest, while avoiding mains power interference.

2.7. Statistical Methods

Since the Lilliefors test confirmed the data were normally distributed, a regular four-way parametric analysis of variance (ANOVA) was used to compare the power of tonic muscle activity against factors of gender (female, male), task (eyes closed, eyes open), location (frontal, left-temporal, central, right-temporal, occipital) and condition (migraine, control).

Post hoc tests on significant factors used Tukey's honest significant difference criterion for the multiple comparisons Muscle activity of migraineurs with aura was not statistically significantly different to migraineurs without aura (migraine with aura: n=13, -13.62 ± 0.56 dB, migraine without aura: n=13, -13.20 ± 0.62 dB, p=0.71, data not shown), therefore we used a single migraine group.

A five percent level of significance was utilized throughout.

2.8. Linear Regression

Linear regression analysis was implemented to test for a relationship between the severity of the headache and the power of muscle activity. 89 of 91 participants reported intensity as none (controls) or severe, and duration was reported as either “hours” or “days”, i.e. these measures provided only two sufficiently populated abscissa points. Hence we excluded intensity and duration from the linear regression analysis, retaining only frequency. Muscle

11

activity was regressed against frequency for all combinations of task and region, and adjusted for multiple comparisons using Bonferroni correction.

3. Results

3.1. Derived Components

The distribution of the spectral slope of the components derived from applying ICA on the EEG data of each group (migraine and control) is shown in figure 2. It can be observed that for both tasks (eyes closed and eyes open), the migraine group has more muscle-containing components (components with gradient above -0.31 bel/decade) than the control group, and that the distributions of the two groups appear quite different. We elected to test this final observation by comparing the means of the distributions. As the data is not normally distributed, we used a Wilcoxon signed rank test, and found that the mean of the spectral slope of components in the migraine group is significantly different from the control group for both tasks (eyes closed: migraine = -0.33 ± 0.01 bel/decade, control = -0.61 ± 0.008 bel/decade, p<0.001, eyes open: migraine = -0.12 ± 0.01 bel/decade, control = -0.41 ± 0.008 bel/decade, p<0.001). This result is consistent with there being more muscle activity in the migraine group than control group.

3.2. Topographic Map of Muscle Power

The mean muscle power for each group, migraine and control, for each baseline task (eyes closed and eyes open) at muscle band (52-98 Hz) has been illustrated topographically in figure 3. Visual inspection reveals a largely peripheral distribution of muscle power, with the most intense power being seen posteriorly, adjacent to the upper cervical and occipital muscles.

We used a four-way ANOVA to test these observations statistically. The results of the statistical analysis are shown in tables 2 to 5.

12 3.3. ANOVA Results

Table 2 shows the results of statistical analysis. The ANOVA revealed a significant difference in muscle power based on condition, task, and location, but no significant differences were found for gender, nor in any interaction of the significant factors. We consider each significant factor in turn below.

3.3.1. Migraine versus Control

Table 3 shows the result of the post hoc test comparing muscle activity between the migraine and control groups. The muscle activity in the migraine group is statistically greater than muscle activity in the control group (F=4.85, p=0.028). Overall, migraine subjects have approximately 30% more cranial and upper cervical muscle activity than control subjects.

3.3.2. Eyes Closed versus Eyes Open

As shown in table 4, the amount of muscle activity during baseline eyes open is significantly more than in baseline eyes closed (F=9.78, p=0.002), by a factor of about 1.5 overall.

3.3.3. Region

The amount of muscle activity compared among five regions (frontal, left temporal, central, right temporal, and occipital) Table 5, illustrates there is a significant difference in muscle power based on location (F=13.34, p<0.001). The power in the occipital region is statistically greater than all other locations. Central power is also statistically less than frontal and left temporal, consistent with the absence of muscle in the central region.

3.4. Relationship between Severity of Headache and Cranial and Upper Cervical Muscle Activity

Using linear regression, we tested for a relationship between frequency of headache and muscle activity. Tests were undertaken twice, once without control participants, and once including. Control participants with a frequency of zero since they had no history of

13

headache. Based on the results of the ANOVA, we included data from males and females together, and grouped the left and right temporal locations, but otherwise tested separately for each combination of location and task. No statistically significant relationship was found in any test.

4. Discussion

The main finding is that there is more cranial and upper cervical muscle background activity in migraineurs than in non-headache controls (1.12 dB or 30% increase). The differences between this and previous studies are that our migraine participants were not selected for severity and, therefore, mostly had a frequency of headache less than 2 per month, and that they were recorded in their non-headache phase. These differences strengthen the finding because it is not confounded either by the complicating intra-individual reactions to acute headache pain, nor to the selection of an atypical sub-group of migraineurs. However, we did not document when the patients experienced their next migraine, so that some patients may have been preictal when studied. If so, these unrecognised pre-ictal cases do compromise this study. Given the infrequent migraines in many of our cases, and the lack of relation of muscle activity to frequency, we suggest this is unlikely. The finding adds to the evidence that

increased cranial and upper cervical muscle activity needs to be recognised as a feature in migraine and, therefore, less different from tension-type headache than normally recognised.

While the results are consistent with some previous studies, discussed in the Introduction, we have presented a measure of muscle activity that has a robust basis (ICA plus spectral slope plus muscle-frequency-band quantitation) combined with comprehensive topographic mapping. The use of a high-density EEG cap gives good spatial coverage, and avoids the issue of where to measure specific muscles, and enables us to evaluate the activity of all cranial and cervical muscles, not just one or two specific ones. Hence, we have a high degree

14

of confidence in our finding that migraine sufferers have more cranial and upper cervical muscle activity than controls.

A limitation of the ICA approach is that the separation of components into purely myogenic versus purely neurogenic sources is not perfect (Nam et al. , 2002, Delorme et al. , 2012), and the results could therefore be improved by an algorithm that separates the components more effectively. However, the same approach has been applied to all participants in this study, diminishing a systematic error in the comparison due to this limitation.

Our study does not address muscle activity during an acute headache but, given others’ work, it would be surprising to find a reduced level during headache than that recorded in the headache-free period. Our participants were studied during a non-ictal period, and further studies comparing post-ictal, inter-ictal and pre-ictal may reveal more about the nature of the association. Additionally, our study has no participants with chronic migraine,

hence extending our finding would require testing with further studies.

While our study reveals an association between the presence of migraine and resting muscle activity, we do not know the nature of this association. If the finding is of pathophysiological significance, it would provide some support to the now standard use of botulinum toxin in the treatment of severe migraine. Impaired sensory control by brainstem mechanisms (currently proposed as a primary feature in migraine (Goadsby et al. , 2017) and possibly present in tension-type headache (Ashina et al. , 2012)), are thought to magnify trigeminal perivascular and other sensations. We speculate, therefore, that there might also be impaired brainstem control mechanisms for cranial muscle activity in migraine, driving muscle metabolism, so impacting on perivascular sensory nociceptive nerves. Another possible reason for increased cranial muscle activity might be that individuals who have experienced

15

headaches learn (consciously or unconsciously) to use subtle adjustments of head posture or expression as means to control the development of their headache (Martins et al. , 2001).

A finding, previously reported (Whitham et al. , 2008, Boĭtsova et al. , 2009, Ben‐Simon et al. , 2013, Yilmaz et al. , 2014), is that there is more cranial muscle activity when the eyes are open than when eyes are closed. The findings show that the usual way of recording EEG, with the eyes closed, does diminish EMG contamination. Presumably, the act of opening the eyes is an alerting process that incorporates readiness for flight or fight and, therefore, muscles generally are somewhat activated.

The greater muscle activity in the occipital region is almost certainly due to EMG from the powerful trapezius, splenius capitis and sub-occipital muscles that insert along or under the nuchal ridge. These muscles maintain posture of the neck and head, and hence are active while sitting regardless of task. In contrast, the temporalis and frontalis muscle are small muscles, responsible for jaw position and facial expression and would be less active at rest.

Although the incidence of headache is twice as high in females and the percentage of females using botulinum toxin to treat their headache is much higher than males (85% vs 15%) (Aydinlar et al. , 2017), our findings suggest that males and females exhibit the same amount of resting muscle activity, whether they have migraine or not.

5. Conclusion

We used a new approach to quantitate muscle activity over the whole head in scalp electrical recordings that are ordinarily used for measuring EEG. This non-invasive topographic approach provides a better measure of activity of cranial and upper cervical muscle activity than other methods. We applied the approach to people with migraine, showing that muscle activity is elevated in migraineurs versus non-headache controls. We did not address

16

differences between migraine phases. This result diminishes one of the accepted conceptual differences between migraine and tension-type headache.

Acknowledgements

This work was supported by the National Health and Medical Research Council, the Flinders Medical Centre Foundation, the Clinician’s Special

Purpose Fund of the Flinders Medical Centre, and an equipment grant from the Wellcome Trust, London, U.K.

Funding

AJ is funded by an Australian Postgradiate Award.

Declarations of interest

None of the authors have potential conflicts of interest to be disclosed.

17 References

Ahles TA, Martin JB, Gaulier B, Cassens HL, Andres ML, Shariff M. Electromyographic and vasomotor activity in tension, migraine, and combined headache patients: The influence of postural variation.

Behav Res Ther. 1988;26:519-25.

Ashina S, Bendtsen L, Ashina M. Pathophysiology of migraine and tension-type headache. Tech Reg Anesth Pain Manag. 2012;16:14-8.

Aydinlar EI, Dikmen PY, Kosak S, Kocaman AS. OnabotulinumtoxinA effectiveness on chronic migraine, negative emotional states and sleep quality: a single-center prospective cohort study. J Headache Pain. 2017;18:23.

Bakal DA, Kaganov JA. Muscle contraction and migraine headache: psychophysiologic comparison.

Headache: J Head Face Pain. 1977;17:208-15.

Bakke M, Tfelt-Hansen P, Olesen J, Møller E. Action of some pericranial muscles during provoked attacks of common migraine. Pain. 1982;14:121-35.

Bell AJ, Sejnowski TJ. An information-maximization approach to blind separation and blind deconvolution. Neural Comput. 1995;7:1129-59.

Ben‐Simon E, Podlipsky I, Okon‐Singer H, Gruberger M, Cvetkovic D, Intrator N, et al. The dark side of the alpha rhythm: fMRI evidence for induced alpha modulation during complete darkness. Eur J Neurosci. 2013;37:795-803.

Blaschek A, Milde-Busch A, Straube A, Schankin C, Langhagen T, Jahn K, et al. Self-reported muscle pain in adolescents with migraine and tension-type headache. Cephalalgia. 2012;32:241-9.

Blau JN, MacGregor EA. Migraine and the neck. Headache. 1994;34:88-90.

Boĭtsova I, Dan'ko S. EEG changes in comparison of rest states with open and closed eyes in complete darkness. Fiziol Cheloveka. 2009;36:138-41.

Burnett C, Fartash L, Murray B, Lamey PJ. Masseter and temporalis muscle EMG levels and bite force in migraineurs. Headache: J Head Face Pain. 2000;40:813-7.

Celentano DD, Stewart WF, Linet MS. The relationship of headache symptoms with severity and duration of attacks. J Clin Epidemiol. 1990;43:983-94.

Clifford T, Lauritzen M, Bakke M, Olesen J, Møller E. Electromyography of pericranial muscles during treatment of spontaneous common migraine attacks. Pain. 1982;14:137-47.

18

Delorme A, Palmer J, Onton J, Oostenveld R, Makeig S. Independent EEG sources are dipolar. PloS one. 2012;7:e30135.

Delorme A, Sejnowski T, Makeig S. Enhanced detection of artifacts in EEG data using higher-order statistics and independent component analysis. Neuroimage. 2007;34:1443-9.

Didier HA, Di Fiore P, Marchetti C, Tullo V, Frediani F, Arlotti M, et al. Electromyography data in chronic migraine patients by using neurostimulation with the Cefaly(R) device. Neurol Sci. 2015;36 Suppl 1:115-9.

Ebinger F. Exteroceptive suppression of masseter muscle activity in juvenile migraineurs.

Cephalalgia. 2006;26:722-30.

Fernández‐de‐las‐Peñas C, Cuadrado M, Arendt‐Nielsen L, Pareja J. Side‐to‐side differences in pressure pain thresholds and pericranial muscle tenderness in strictly unilateral migraine. Eur J Neurol. 2008;15:162-8.

Fitzgibbon S, DeLosAngeles D, Lewis T, Powers D, Grummett T, Whitham E, et al. Automatic determination of EMG-contaminated components and validation of independent component analysis using EEG during pharmacologic paralysis. Clin Neurophysiol. 2016;127:1781-93.

Goadsby PJ, Holland PR, Martins-Oliveira M, Hoffmann J, Schankin C, Akerman S. Pathophysiology of migraine: a disorder of sensory processing. Physiol Rev. 2017;97:553-622.

Goncharova II, McFarland DJ, Vaughan TM, Wolpaw JR. EMG contamination of EEG: spectral and topographical characteristics. Clin Neurophysiol. 2003;114:1580-93.

Hagen K, Einarsen C, Zwart JA, Svebak S, Bovim G. The co‐occurrence of headache and musculoskeletal symptoms amongst 51 050 adults in Norway. Eur J Neurol. 2002;9:527-33.

Headache Classification Committee of the International Headache S. The International Classification of Headache Disorders, 3rd edition (beta version). Cephalalgia. 2013;33:629-808.

Hung C-I, Liu C-Y, Chen J-J, Wang S-J. Migraine predicts self-reported muscle tension in patients with major depressive disorder. Psychosomatics. 2008;49:502-10.

Jensen R, Rasmussen BK, Pedersen B, Olesen J. Muscle tenderness and pressure pain thresholds in headache. A population study. Pain. 1993;52:193-9.

Landgraf MN, Ertl-Wagner B, Koerte IK, Thienel J, Langhagen T, Straube A, et al. Alterations in the trapezius muscle in young patients with migraine--a pilot case series with MRI. Eur J Paediatr Neurol.

2015;19:372-6.

19

Lebbink J, Spierings E, Messinger HB. A questionnaire survey of muscular symptoms in chronic headache. An age-and sex-controlled study. Clin J Pain. 1991;7:95-101.

Leistad R, Sand T, Westgaard R, Nilsen K, Stovner L. Stress-induced pain and muscle activity in patients with migraine and tension-type headache. Cephalalgia. 2006;26:64-73.

Lous I, Olesen J. Evaluation of pericranial tenderness and oral function in patients with common migraine, muscle contraction headache and ‘combination headache’. Pain. 1982;12:385-93.

Makeig S, Bell AJ, Jung T-P, Sejnowski TJ. Independent Component Analysis of

Electroencephalographic Data. In: Touretzky DS, Mozer MC, Hasselmo ME, eds. Advances in Neural Information Processing Systems 8, 1996. pp 145-51.

Martins IP, Parreira E. Behavioral Response to Headache: A Comparison Between Migraine and Tension‐type Headache. Headache: J Head Face Pain. 2001;41:546-53.

Nam H, Yim TG, Han SK, Oh JB, Lee SK. Independent component analysis of ictal EEG in medial temporal lobe epilepsy. Epilepsia. 2002;43:160-4.

Oksanen A, Pöyhönen T, Ylinen JJ, Metsähonkala L, Anttila P, Laimi K, et al. Force production and EMG activity of neck muscles in adolescent headache. Disabil Rehabil. 2008;30:231-9.

Oostenveld R, Praamstra P. The five percent electrode system for high-resolution EEG and ERP measurements. Clin Neurophysiol. 2001;112:713-9.

Pope KJ, Fitzgibbon SP, Lewis TW, Whitham EM, Willoughby JO. Relation of gamma oscillations in scalp recordings to muscular activity. Brain Topogr. 2009;22:13-7.

Prakash A, Roy V. A Review on EEG Artifacts and its Different Removal Technique. IJSIP. 2016;9:291- 302.

Tfelt‐Hansen P, Lous I, Olesen J. Prevalence and significance of muscle tenderness during common migraine attacks. Headache: J Head Face Pain. 1981;21:49-54.

Vigário R, Sarela J, Jousmiki V, Hamalainen M, Oja E. Independent component approach to the analysis of EEG and MEG recordings. IEEE Trans Biomed Eng. 2000;47:589-93.

Watson DH, Drummond PD. Head Pain Referral During Examination of the Neck in Migraine and Tension‐Type Headache. Headache: J Head Face Pain. 2012;52:1226-35.

Whitham EM, Lewis T, Pope KJ, Fitzgibbon SP, Clark CR, Loveless S, et al. Thinking activates EMG in scalp electrical recordings. Clin Neurophysiol. 2008;119:1166-75.

20

Whitham EM, Pope KJ, Fitzgibbon SP, Lewis T, Clark CR, Loveless S, et al. Scalp electrical recording during paralysis: quantitative evidence that EEG frequencies above 20Hz are contaminated by EMG.

Clin Neurophysiol. 2007;118:1877-88.

Yilmaz G, Ungan P, Sebik O, Uginčius P, Türker KS. Interference of tonic muscle activity on the EEG: a single motor unit study. Front Hum Neurosci. 2014;8:504.

Zhang C, Tong L, Zeng Y, Jiang J, Bu H, Yan B, et al. Automatic artifact removal from electroencephalogram data based on a priori artifact information. Biomed Res Int.

2015;2015:720450.

Figure legends

Figure 1: EEG channels, illustrating the five groups, based on their location in the 10-5 system (modified from (Oostenveld et al. , 2001)).

Figure 2: Histogram of the spectral slope of components derived from applying ICA to the EEG data of the control and migraine groups during the eyes closed task (left) and eyes open task (right).

Figure 3: Mean muscle power in the muscle band for 26 migraine subjects and 65 control subjects.

21

22

23

Table 1: Subject Demographics

Participants Age (years)

(mean ± SEM) Females Males Number of EEG channels

Migraine 48.6 ± 13 19 7 124

Control 46.2 ± 17.2 33 32 124

Table 2: Results of ANOVA

Analysis of Variance

Source Sum of squares Mean square F p value

Condition 196.8 196.76 4.85 0.028 **

Task 396.8 396.83 9.79 0.002 **

Region 2163 540.74 13.34 < 0.001 **

Gender 29.7 29.65 0.73 0.392

Task*Region 35.4 8.84 0.22 0.928

Task*Condition 35.9 35.88 0.89 0.347

Region*Condition 103.7 25.92 0.64 0.634

Gender*Region 290.1 72.53 1.79 0.128

Gender*Task 12.8 12.75 0.31 0.575

Gender*Condition 1.1 1.13 0.03 0.867

Table 3: Conditions powers (mean ± SEM) and p values from post hoc test for conditions.

Condition Mean of muscle power (dB) p value Migraine -13.61 ± 0.38

0.028 **

Control -14.73 ± 0.24

Table 4: Tasks powers (mean ± SEM) and p values from post hoc test for tasks.

Task Mean of muscle power (dB) p value Eyes open -13.42 ± 0.30

0.002 **

Eyes closed -14.92 ± 0.30

Table 5: Regional powers (mean ± 0.54 dB from ANOVA) and p values from post hoc test for region.

Region Occipital Right temporal Central Left temporal Frontal

Power (dB) -11.21 -14.66 -16.64 -14.39 -13.42

Frontal 0.002 ** 0.880 0.003 ** 0.979

Left temporal 0.003 ** 0.996 0.023 **

Central < 0.001 ** 0.064 Right temporal < 0.001 **

24