Brain, Vol. 126, No. 2, 326-342,
February 2003
© 2003 Guarantors of Brain
doi: 10.1093/brain/awg043
Abnormal corticomuscular and intermuscular coupling in high-frequency rhythmic myoclonus
1 Sobell Department of Clinical Neurophysiology, Institute of Neurology and 2 Neurosciences Unit, Institute of Child Health and Great Ormond Street Hospital for Children, London, UK, 3 Neurologische Klinik und Poliklinik, Charité, Campus Virchow-Klinikum, Berlin, Germany, and 4 Institute of Child Neurology and Psychiatry, University of Pisa and IRCCS Fondazione Stella Mars, Pisa, Italy
*Present address: Institute of Child Neurology and Psychiatry, University of Pisa and IRCCS Fondazione Stella Mars, Pisa, ItalyCorrespondence to: Dr Pascal Grosse, Sobell Department of Neurophysiology, Clinical Motor Physiology Group, Institute of Neurology, 8–11 Queens Square, London WC1N 3BG, UK E-mail: p.grosse{at}ion.ucl.ac.uk
Received June 14, 2002. Revised August 5, 2002. Accepted September 8, 2002.
 |
Summary |
|---|
 Top Summary IntroductionPatients and healthy controlsMethodsResultsDiscussionReferences |
|---|
Frequency analysis may have some advantages over back-averaging in the neurophysiological assessment of patients with suspected cortical myoclonus in whom myoclonic EMG bursts repeat rhythmically at high frequency. However, the clinical utility of EEG–EMG coherence and related EMG–EMG coherence is not established. Equally, there is an incomplete understanding of the physiology of the systems contributing to the coherence evident between signals in cortical myoclonus. Here we address these issues in an investigation of EEG–EMG and EMG–EMG coupling in proximal and distal muscles of the upper extremities in nine patients with multifocal high frequency rhythmic myoclonus due to non-progressive conditions. We found exaggerated coherence between EEG and contralateral EMG and between pairs of ipsilateral EMG signals. The results of frequency analysis of EMG–EMG mirrored those for EEG–EMG, but the former technique was superior in distinguishing a pathologically exaggerated common drive in distal upper limb muscles. Both techniques were more sensitive than back-averaging. Frequency analysis also revealed important disparities between proximal and distal upper limb muscles. In the latter case, the functional coupling between cortex and muscle was dominated by efferent processes. In contrast, there was considerable inter-individual variation in the extent to which EEG–EMG and EMG–EMG coupling in proximal muscles reflected afferent and efferent loops. Thus, the processes sustaining myoclonic discharges may differ for proximal and distal muscles and between patients.
Keywords: myoclonus; frequency analysis; EEG–EMG coherence; EMG–EMG coherence; back-averaging
Abbreviations: 1DI= first dorsal interosseus muscle; APB = abductor pollicis brevis muscle; MEP = motor evoked potentials; TMS = transcranial magnetic stimulation
|
Introduction |
|---|
|
TopSummaryIntroductionPatients and healthy controlsMethodsResultsDiscussionReferences |
|---|
The analysis of the coherence between scalp EEG and surface EMG shows promise as a new tool in delineating the functional coupling between oscillatory activity in the motor cortex and that in muscle in both physiological and pathological conditions (Leocani and Comi, 1999
; Mima and Hallett, 1999a; Brown and Grosse, 2002). In particular, in cortical myoclonus EEG–EMG frequency analysis may have methodological advantages in detecting a cortical correlate over the classical neurophysiological repertoire of back-averaging (Shibasaki and Kuroiwa, 1975) and the detection of a giant cortical sensory evoked potential. Many myoclonic patients do not have reflex myoclonus and giant cortical evoked potentials, and the identification of a cortical correlate that precedes jerks in back-averages relies on the absence of myoclonic events just prior to the trigger EMG burst. Yet, many patients with cortical myoclonus have rhythmic EMG bursts at relatively high frequency (Thompson et al., 1994; Brown and Marsden, 1996), especially those with minipolymyoclonus (Wilkens et al., 1985; Hallett and Wilkens, 1986) as in cortical tremor (Toro et al., 1993; Terada et al., 1997), Angelman syndrome (Guerrini et al., 1996) or autosomal dominant cortical myoclonus and epilepsy (Guerrini et al., 2001). In contrast, the increased signal content with repetitive myoclonic jerks favours detection using frequency analysis. In addition, the latter technique introduces no arbitrary trigger level so that jitter is less, statistical evaluation of the results is possible and the technique is quick and automated, so that long sections of data may be analysed. Thus in a recent study, EEG–EMG coherence and a cortical correlate in the cumulant density estimate were demonstrated in eight patients with a variety of conditions associated with cortical myoclonus, whereas only three had a time-locked EEG correlate upon back-averaging (Brown et al., 1999).
Nevertheless, the determination of EEG–EMG coherence still requires a relatively artefact-free EEG recording and EEG recording itself can be difficult and time-consuming in patients with involuntary jerks, some of whom are children. There is growing evidence that corticomuscular coupling is reflected in the pattern of coherence between muscles (Farmer et al., 1993; Kilner et al., 1999). This leads us to hypothesise that EMG–EMG coherence may also be used to identify pathological cortical drives to muscle and, if so, this technique may have practical advantages over the assessment of EEG–EMG coherence. Certainly there is preliminary evidence of a close correspondence between the pattern of EEG–EMG coherence and that of EMG–EMG coherence in cortical myoclonus (Brown et al., 1999).
The interpretation of the results of frequency analysis in myoclonus is, however, bedevilled by the sensitivity of this technique. This has two consequences. First, even healthy subjects may be found to have EEG–EMG and EMG–EMG coherence, and corresponding features in cumulant density estimates (Halliday et al., 1998; Kilner et al., 1999; Mima and Hallett, 1999b). To date, studies have failed to address which aspects distinguish the pathological cortico-muscular coupling found in cortical myoclonus from the physiological state. Secondly, afferent activities will be detected as well as efferent discharges but, so far, there has been the tacit assumption that the results of frequency analysis in patients with myoclonus may be satisfactorily interpreted solely in terms of descending drives from the motor cortex to muscles.
Here we compare the results of back-averaging to those of frequency analysis in patients with high frequency rhythmic myoclonus and systematically explore the extent to which EMG–EMG coherence can provide a practical alternative to EEG–EMG coherence, the factors that distinguish pathological from normal corticomuscular and intermuscular coupling, and whether coupling is always the product of cortical efferent activity. We based our assessments on minimal interventions that would lend themselves to incorporation into routine clinical neurophysiological practice and studied patients with the clinical syndrome of minipolymyoclonus as these are the cases that are most difficult to diagnose using standard back-averaging techniques. Our results confirm the clinical utility of both EEG–EMG and EMG–EMG coherence estimates in the assessment of myoclonus, but indicate that interpretation must take into account the physiological complexity of cortical myoclonus, which does not solely involve efferent cortico-muscular pathways.
|
Patients and healthy controls |
|---|
|
TopSummaryIntroductionPatients and healthy controlsMethodsResultsDiscussionReferences |
|---|
We examined nine patients (mean age 43 years; range 14–80 years) with jerks due to a variety of non-progressive syndromes associated which cortical myoclonus (Table 1). All had high-frequency, low-amplitude myoclonus consistent with minipolymyoclonus (Wilkens et al., 1985
). Case 9 also had some additional infrequent and less regular larger amplitude jerks. Three patients of a pedigree (Cases 1–3) had multifocal myoclonus in relation to the recently described and genetically defined syndrome of autosomal-dominant cortical reflex myoclonus and epilepsy linked to chromosome 2 (Guerrini et al., 2001). Three patients (Cases 4–6) had Angelman syndrome with different genetic defects [one with 15q11–13 deletion (Case 4), one with uniparental disomy for chromosome 15 (Case 5) and one with UBE3A mutation (Case 6)]. They exhibited continuous multifocal, high frequency myoclonic jerks associated with dystonic limb posturing as previously described for this syndrome (Guerrini et al., 1996). One patient (Case 7) had myoclonus in relation with Lennox–Gastaut-syndrome. In one patient (Case 8), the diagnosis of familial cortical tremor was made. Cortical tremor is a type of minipolymyoclonus consisting of cortical reflex myoclonus, often associated with epilepsy and posture and/or action-induced jerks at high frequencies and low amplitudes showing the neurophysiological features of cortical myoclonus (Ikeda et al., 1990; Toro et al., 1993; Terada et al., 1997). The last patient (Case 9) had myoclonus related to coeliac disease. More than half of the patients (Cases 1, 2, 4, 6, and 7) had epilepsy with focal and/or generalized seizures besides cortical myoclonus. All patients except two (Cases 3 and 8) received a variety of antimyoclonic and/or antiepileptic medication with different modes of action at the time of the neurophysiological examination. Ten healthy subjects (mean age 40 years, range 27–74 years) were also studied.
|
All subjects gave their informed consent. The study was approved by the National Hospital for Neurology and Neurosurgery and the Institute of Neurology Joint Research Ethics Committee.
|
Methods |
|---|
|
TopSummaryIntroductionPatients and healthy controlsMethodsResultsDiscussionReferences |
|---|
EEG and EMG recording
Surface EMG and EEG were recorded with 9 mm diameter silver–silver chloride electrodes. We opted for a bipolar EEG derivation rather than a Laplacian derivative, as the latter requires considerably more channels, limiting its utility in the setting of a routine clinical neurophysiological service. Both montages avoid the use of a common reference although bipolar electrodes may degrade phase information (Mima and Hallett, 1999
b). Electrodes were positioned according to the 10–20 system at C3-F3 and C4-F4. EMG was recorded bilaterally from deltoid, finger extensor and intrinsic hand muscle (Cambridge Electronic Design Ltd, Cambridge, UK) [abductor pollicis brevis (APB) and first dorsal interosseous (1DI)]. EMG electrodes were placed 2 cm apart on the muscle belly (except for intrinsic hand muscles where one electrode was sited over the metacarpo-phalangeal joint). EMG and EEG were bandpass-filtered at 16–300 and 0.53–300 Hz, respectively. The high pass filter for EMG was chosen to limit movement artefact. Signals were amplified and digitized with 12-bit resolution by a CED 1401 analogue-to-digital converter. The sampling rate was 1000 Hz. Signals were displayed and stored on a PC using the CED Spike 2 software package. Patients were recorded either at rest (Cases 4–7), so that no voluntary drive to muscles was present, or while voluntarily maintaining a posture (shoulder abduction, wrist extension, thumb adduction; Cases 1–3, 8 and 9). Record lengths averaged 183 ± 21 s (standard error of the mean). Data lengths were kept fixed within individual subjects. Healthy subjects were asked to co-activate recorded muscles over four periods of 
60 s. A period of 60–180 s rest was given between co-activations. Total data lengths used were fixed at 200 s in healthy subjects.
Analysis
Frequency analysis
EEG–EMG and EMG–EMG coherence were analysed off-line using a program written by J. Ogden and D. Halliday (Division of Neuroscience and Biomathematical Systems, University of Glasgow, UK) and based on methods outlined by Halliday and colleagues (Halliday et al., 1995). The EEG, denoted by a, and rectified EMG, denoted by b, were assumed to be realisations of stationary zero mean time series. The statistical tool used for data analysis was the discrete Fourier transform and the parameters derived from it. These were estimated by dividing the records into a number of disjoint sections of equal duration (1024 data points) and estimating spectra by averaging across these discrete sections. In the frequency domain estimates of the autospectrum of the EEG [faa(
)] and EMG [fbb()] and their cross-spectrum [fab()] were constructed. The frequency resolution of all spectra was 0.98 Hz. The coherence [|Rab()|2] was also estimated, where:
Coherence is a measure of the linear association between two signals. It is a bounded measure taking values from 0 to 1, where 0 indicates that there is no linear association (that is process B is of no use in linearly predicting process A) and 1 indicates a perfect linear association. The variance of the coherence was normalized by transforming the square root of the coherence (a complex valued function termed coherency) at each frequency using the Fisher transform. This results in values of constant variance for each record given by 1/2L, where L is the number of segment lengths used to calculate the coherence. For comparison between groups, muscles and across signals the area under the curve of transformed coherence was calculated for each subject over the band at which coherence was significant. Data were then pooled for each group and the 95% confidence limits of the mean calculated.
Cumulant density estimates, similar to the cross-correlations between signals, were calculated from the inverse Fourier transform of the cross-spectra. Confidence limits for autospectra, coherence and cumulant density estimates were calculated as described previously (Halliday et al., 1995). Coherence was considered to be significant if it exceeded the 95% confidence level.
The phase, defined as the argument of the cross spectrum may be estimated by:
ab() = arg{fab()}
Phase was formally assessed only where coherence was significant and extended over at least five data points. The constant time lag between the two signals was calculated from the slope of the phase estimate after a line had been fitted by linear regression, but only if a linear relationship accounted for >80% of the variance. In some instances, coherence and phase spectra appeared to consist of more than one component. In these cases, the limits of individual components were defined by the turning points of the best-fit second or third order polynomial fitted to all contiguous plotted points at which coherence was significant. The polynomials accounted for >80% of the variance and had 
8 data points per model order. Lags and leads were calculated using the equation:
Confidence limits (95%) for the spectral phase estimates were calculated as described previously (Halliday et al., 1995), but those given for temporal delays are the 95% confidence limits of the line fitted by linear regression.
Back-averaging
Back-averaging was performed off-line in Spike 2. EMG was rectified and myoclonic EMG bursts identified using a level of 100 µV (sufficient to exclude volume conduction or mains artefact) to produce a series of digital events. EEG and EMG signals were then re-aligned to these events and averaged. The dominant frequency of corticomuscular coupling was determined from the interval between the peak positivities of the largest serial cortical correlates in the back-averaged contralateral EEG. The interval with which EEG lead or lagged EMG was determined from the latency of the peak positivity in the contralateral EEG with respect to the time of the digital events, after subtraction of the latency of the EMG response with respect to the same events. This was only performed where a single positive-negative cortical correlate was unequivocally larger (>10%) than others.
|
Results |
|---|
|
TopSummaryIntroductionPatients and healthy controlsMethodsResultsDiscussionReferences |
|---|
Raw EMG
The raw EMG consisted of rhythmic EMG bursts of short duration with minimal or no pre-innervation between myoclonic bursts. Thus, in each case the rectified EMG level between myoclonic bursts was less than 40 µV in >95% of interburst intervals. Four patients were recorded at rest and their results were no different to those in patients recorded whilst they made a voluntary postural contraction. Fig. 1A is a representative example of the signal recorded in a patient (Case 3) with autosomal-dominant cortical reflex myoclonus and epilepsy linked to chromosome 2. Brief myoclonic bursts are evident at a frequency of
13–15 Hz.
|
Frequency analysis
Fig. 1B–E illustrates the results of frequency analysis in the same patient. Coherence between APB and the contralateral motor cortex is significant at frequencies between 6 and 19 Hz (Fig. 1C) with a linear phase slope between the two signals (Fig. 1D) and a delay of 15.7 ms with EEG preceding EMG. The corresponding cumulant density estimate is shown in Fig. 1E.
Fig. 2, showing data taken from the same patient, gives an example of the typical difference between EEG–EMG and EMG–EMG coherence for both proximal and distal muscles. The data are drawn from the same recording. While the frequency at which coherence peaks remains constant, centred around 13 Hz, the frequency content is broader and the extent of coherence is higher for EMG–EMG coherence than EEG–EMG coherence.
|
EEG–EMG coherence
Figure 3A–C summarizes the area of significant transformed coherence between EEG and EMG in the spectra from individual subjects for deltoid, finger extensors and intrinsic hand muscles. Both right and left sided muscles are included. Note the log scale. Transformed coherences showed substantial overlap between patients and controls. Thus, on an individual basis, EEG–EMG coherence using a simple bipolar montage had limited sensitivity.
|
Nevertheless, there were clear differences at the group level. Figure 3D summarizes the mean transformed coherence area and its 95% confidence limits across all patients and all healthy subjects for the different muscles. The mean transformed coherence area in the patient group is higher by a factor of 3 to 9 compared with normal values (Fig. 3D). Note too that coherences are considerably higher for distal muscles. Figure 4 compares the distribution of transformed EEG–EMG coherence across frequencies for the different muscles. For this purpose the individual spectra, rather than the cumulative area, have been pooled. Healthy subjects only show a discrete peak in the pooled spectra for the forearm and intrinsic hand muscles centred around 15 Hz. This is slightly lower than in our previous study of physiological cortico-muscular coupling (Brown et al., 1998
) and may be partly due to the use of EEG (which is affected by the low-pass filter characteristics of the skull and scalp) rather than magnetoencephalograpy in the current study. The frequency range in the present study is comparable with that found in other electroencephalographic studies of corticomuscular coherence (Mima and Hallett, 1999b), where coherence in the alpha band is not uncommon (Mima and Hallet, 1999a). Patients also had a peak in the pooled spectrum for deltoid, while their mean transformed coherence area was mostly above the 95% confidence limits for the healthy subjects in all three muscles. The peak frequency was similar in the different spectra, although peaks were broader in the patient group, where there was also a subsidiary peak at just above 30 Hz in the spectra for distal muscles.
|
The sensitivity and specificity of EEG–EMG coherence were both high (Table 2) when using the 95% confidence limit of the control group (derived from the areas under the transformed coherence graphs for each subject) as the cut-off.
|
EMG–EMG coherence
Similar analyses were performed for EMG–EMG coherence. Figure 5A and B summarize the area of significant transformed coherence between finger extensor and deltoid, and between finger extensor and intrinsic hand muscle EMG in the spectra from individual subjects. Transformed coherences show overlap between patients and controls, although this overlap is modest compared with EEG–EMG coherence (Fig. 3).
|
The differences were even clearer at the group level. Figure 5C summarizes the mean transformed coherence and its 95% confidence limits across all patients and all healthy subjects for the different muscle pairs. The mean transformed coherence in the patient group is very much greater than in normals and greater than the mean transformed coherence between EEG and muscles in the patients (Fig. 3D). Note too that coherences are considerably higher for the distal muscle pair. Figure 6 compares the distribution of transformed EMG–EMG coherence across frequencies for the different muscle pairs. Healthy subjects only show a peak in the pooled spectra for the distal muscle pair. Patients have a peak in both proximal and distal muscle pairs, and transformed coherences for both are above the 95% confidence limits of the mean for the healthy subjects. The peak frequency in the spectrum for finger extensor–intrinsic hand muscles is similar in controls and patients, although the peak is broader in the patient group.
|
In every single patient the coherence between muscle pairs, be they proximal or distal, as measured by the area of transformed coherence exceeded the coherence between the EEG and respective muscles. In addition, EMG–EMG coherence between finger extensor and intrinsic hand muscles was able to establish abnormal coupling in every case, whereas abnormal coupling could only be demonstrated in <90% of cases of EEG–EMG coherence (Table 2).
The sensitivity and specificity of EMG–EMG coherence were both high (Table 2) and slightly greater than for EEG–EMG coherence. The cut-off used for these assessments was the 95% confidence limit of the control group (derived from the areas under the transformed coherence graphs for each subject).
Phase
Hand muscles
The temporal difference between EEG and EMG calculated from phase spectra demonstrated a uniform pattern in the intrinsic hand muscles. EEG systematically led EMG by 8.3–18.8 ms (Fig. 7A). The mean EEG lead was 14.8 ± 2.3 ms (95% confidence limits), being shorter than the mean motor evoked potentials (MEP) latency in active 1DI following transcranial magnetic stimulation (TMS) of the motor cortex of 20.4 (Eisen and Shytbel, 1990). A comparable pattern was seen for the temporal difference between forearm extensor and intrinsic hand muscle EMG. The former led by 3.3–11.2 ms (Fig. 7B). The mean lead of the forearm extensors was 6.7 ± 1.4 ms and was therefore compatible with the difference of 5.2 ms between MEP latencies for forearm extensors and 1DI following TMS of the motor cortex in normal controls (Eisen and Shytbel, 1990).
|
Proximal muscles
In contrast, phase differences between EEG and EMG for more proximal muscles and between deltoid and finger extensor EMG were more complicated. In some cases, EEG led EMG, but in others EMG led EEG (Fig. 7C–E). The latter suggests an afferent drive from muscle to cortex. Although the picture varied between subjects, phase estimates concurred within each subject in every case in whom estimates were available for deltoid and forearm extensors on the same side. Thus, phase estimates were of the same sign in Cases 2, 5 and 9 (see Fig. 7C and D). This suggests that variations in the temporal difference between EEG and proximal muscles across patients were due to physiological differences rather than chance. In some subjects, EEG–EMG coupling for proximal muscles was dominated by a corticomuscular efferent drive as with 1DI, whereas in others coupling was dominated by an afferent drive from the periphery.
When delays were compatible with an efferent system they were somewhat shorter than the latency of TMS induced MEPs to the respective active muscle, as was the case for the intrinsic hand muscles. For example, in the one patient (Case 2) in whom EEG led deltoid EMG, this was by 9.6 ms (Fig. 7C). In Cases 2, 3 and 8, in whom EEG lead forearm extensor EMG, this was by 10 ms (Fig. 7D).
On the other hand, when delays were compatible with an afferent system they were seldom consistent with the latency of evoked potentials from the region of the respective muscle recorded in healthy subjects. For example, in Cases 5, 7 and 9 in whom deltoid EMG led EEG, this lead varied between –24.1 to –76.9 ms (Fig. 7C). In Cases 1, 4, 5, 6 and 9, in whom finger extensor EMG led EEG, the mean delay between the two signals exceeded 25 ms (Fig. 7D). EMG leads were therefore greater than the expected delay for an afferent loop, even allowing 10 or so ms for electromechanical delay (McAuley et al., 1997). This variation in the temporal delays was not systematically related to the semiology of the myoclonus or the clinical syndrome associated with it.
A similar pattern was observed in the temporal delay between deltoid and finger extensor EMG. In Cases 6 and 7, and Cases 2 and 5 on the right, deltoid EMG led that in the finger extensor, as would be expected for a simple efferent system. However, in Cases 1 and 3, and Case 2 on the left, forearm extensor EMG led by 20.1 ± 12.1 ms, compatible with an afferent loop, in which afferent activity from the distal upper limb drove a reflex response in deltoid.
Back-averaging
Back-averaging was performed in all nine patients (Table 2). A back-averaged cortical correlate could be discerned in only 50–77% of muscles, but corticomuscular coherence was above the significance level in 100% of cases for finger extensors and intrinsic hand muscles and in 88% of cases for deltoid. For those muscles in which back-averaging was successful, the back-averaged EEG consisted of a rhythmic series of cortical correlates and was similar in nature to the cumulant density estimate (Fig. 8B–D). The frequency of back-averaged cortical correlates was 16 Hz, while the peak frequency upon frequency analysis was 14 Hz. However, there were no statistically significant difference between the peak frequency derived from back-averages and that derived from coherence spectra in those patients in whom both estimates were available.
|
A single positive–negative EEG correlate exceeded others in peak-to-peak amplitude in a given series by >10% in 31–56% of muscles examined. Accordingly, we were only able to measure unambiguous time differences between EEG correlates and EMG onset in these patients. Examples of uninformative/negative (Fig. 8A), ambiguous (Fig. 8B and C) and unambiguous (Fig. 8D) back-averages are illustrated in Fig. 8. Temporal differences measured from unambiguous back-averages are summarized in Fig. 9. EEG led EMG in the intrinsic hand muscles in all but one case. In contrast, EEG could lead or lag EMG in the forearm extensors. EEG’s lead or lag over EMG was always the same in direction in those muscles where temporal delays could be calculated from back-averages and frequency analysis, suggesting that variability was physiological rather than technique dependent.
|
|
Discussion |
|---|
|
TopSummaryIntroductionPatients and healthy controlsMethodsResultsDiscussionReferences |
|---|
We have shown that the exaggerated functional cortico-muscular coupling in patients with cortical myoclonus is not only reflected in an exaggerated coherence between EEG and EMG, but also in an abnormally strong coherence between the EMGs of muscles co-activated by myoclonic jerks. In addition, our results demonstrate that the phase relationship between EEG and EMG and between pairs of EMG signals is complex, reflecting both efferent and afferent drives between cortex and muscles.
Clinical utility of frequency analysis in myoclonus
Hitherto, the electrophysiological characterization of cortical myoclonus has largely depended on the results of back-averaging, in which a positive result requires the demonstration of a cortical correlate that precedes myoclonic EMG bursts. As stated in the Introduction, frequency analysis of myoclonic activity has several potential advantages over back-averaging when myoclonic EMG bursts are frequent as in minipolymyoclonus. This was borne out by the present study in which the symmetry of cortical correlates upon back-averaging meant that unambiguous estimates of the temporal difference between cortical correlate and myoclonic EMG could only be made for 31–56% of the muscles examined, and a cortical correlate could not be discerned for 40% of muscles. In contrast, frequency analysis demonstrated abnormal EEG–EMG coherence and was able to establish a temporal delay in the vast majority of cases. Furthermore, the sensitivity and specificity of EEG–EMG and EMG–EMG coherence were both high.
It must be stressed that back-averaging and frequency analysis emphasise different aspects of the data. The results of frequency analysis reflect the coupling between motor cortex and muscle, and that between muscles due to common drive averaged over time. Here we have characterized how this coupling deviates from normal in patients with multifocal minipolymyoclonus. All signals within a recording are analysed, so there is the methodological concern that the index of coupling reflects both myoclonic activity and any pre-innervation. In practice, this was not a problem in our data sets, where EMG levels were very low between myoclonic bursts. Had this not been the case, however, then only back-averaging would have demonstrated the cortical correlate exclusively linked to myoclonic EMG bursts. In addition, non-reflex myoclonic bursts can be relatively infrequent in some conditions. In these instances, current frequency analysis techniques would be inappropriate as local data stationarity is not approximated and back-averaging would offer the only possibility of documenting a cortical origin. Thus, frequency analysis of myoclonus has advantages when myoclonic jerks occur at high frequency, as in minipolymyoclonus, but back-averaging is the analytical technique of choice when myoclonic bursts occur at low frequency.
The present study also suggests that the assessment of EMG–EMG coherence may be more useful in the future than EEG–EMG coherence in the routine neurophysiological evaluation of patients with myoclonus. As EEG is not required, the technique is less time-consuming and applicable when movement artefact or cranial EMG activity are likely to prevent satisfactory EEG recordings. More importantly, the technique appears to be more sensitive in distinguishing abnormal and normal common drives, and for this purpose we would recommend the simultaneous assessment of the coherence between the forearm extensor and ipsilateral intrinsic hand muscles. The greater sensitivity of EMG–EMG coherence may relate to the increased coherence levels seen between these signals compared with EEG and EMG. Importantly, the increased coherence levels found between EMG signals did not seem to relate to volume conduction, as coherence occurred over relatively narrow bands and did not involve zero phase delays between EMG signals. It is possible that other forms of recording of cortical activity, such as MEG or Laplacian derivatives of EEG may provide more sensitive measurements of EEG–EMG coherence, but these techniques are not universally available or are time-consuming and require multiple EEG channels.
We should also draw attention to the question of specificity of an elevated EEG–EMG or EMG–EMG coherence with regard to other pathologies. Inflated EEG–EMG coherences have also been reported in Parkinson’s disease (Hellwig et al., 2000) and essential tremor (Hellwig et al., 2001), but here coherence is narrow band in nature and centred on a tremor frequency of 
10 Hz. Thus EEG–EMG coherence occurs at generally lower frequencies in these tremor disorders, but a comparative study of coherence in tremor and minipolymyoclonus is necessary to establish whether there is any significant overlap in the frequency of peak coherence in these entities.
Possible homology between EEG–EMG and EMG–EMG coherence
If EMG–EMG coherence is to be useful in the clinical evaluation of cortical myoclonus, then it should be a reasonably faithful marker of the functional coupling between cortex and muscle. This may not necessarily be the case as the coherence between EMG signals could reflect oscillatory subcortical and spinal inputs as well as oscillatory cortical drives to 
-motoneurons. In practice, however, the pattern of EEG–EMG and EMG–EMG coherence was similar, suggesting that the major oscillatory influence on spinal motoneurons, at least in this pathological state, involves the sensorimotor cortex. The one notable difference between EEG–EMG and EMG–EMG coherences was the wider frequency band of the latter. However, this band was still centred on similar peak frequencies and may simply be the product of the improved signal to noise ratio and greater coherence between pairs of EMG signals.
Preferential projection of the oscillatory corticomuscular system to the distal limb
EEG–EMG coherence was greater distally than proximally in the upper limb. This was evident in patients, but also in healthy controls, where there was no detectable coherence in deltoid using bipolar EEG electrodes. It is tempting to interpret these observations as evidence in favour of the preferential projection of pyramidal pathways to distal upper limb muscles (Colebatch and Gandevia, 1989; Rothwell et al., 1991; Ferbert et al., 1992; Palmer and Ashby, 1992; Marsden et al., 1999; Turton and Lemon, 1999).
However, we must first consider an alternative suggestion, that it was our use of a bipolar EEG lead drawn from over the hand area of the motor cortex that led to the greater coherences for intrinsic hand muscles. Isocoherence maps of the coherence between cortical and muscle activity in studies using magnetoencephalography or surface EEG (Salenius et al., 1997; Hellwig et al., 2001) and a further study examining the distribution of coherence by Mima and colleagues (Mima et al., 2000) would suggest that the source of cortical activity coupled to EMG activity is relatively focal. On the other hand, the Laplacian derivations used in many of these studies have been criticized as applying an excessively high spatial filter (Srinivasan et al., 1998), and many MEG studies start from the assumption of a point source responsible for activity (Salenius et al., 1997; Brown et al., 1998). Studies using electrocorticography or tomographic modelling of EEG sources suggest a much more distributed source for the cortical inputs responsible for EEG–EMG coherence even in healthy subjects (Feige et al., 2000; Marsden et al., 2000a; Ohara et al., 2000).
In considering the possibility that our use of a bipolar EEG lead over the hand area may have contributed to the apparent preferential projection of fast conducting pyramidal pathways to distal upper limb muscles, we are fortunate in having a further measure of common inputs to motoneurons that is independent of the EEG. Importantly, EMG–EMG coherence in the upper limb was also greater for a distal muscle pair compared with a proximal muscle pair.
Afferent loops in proximal muscles
A previous frequency analysis of data from patients with cortical myoclonus (Brown et al., 1999) suggested that EEG consistently led EMG. This study limited itself to consideration of distal upper and lower limb muscles. Our frequency analysis and back-averaging results in the distal upper limb were in accord with this. However, here we also demonstrate that the temporal relationships between EEG and the EMG of more proximal upper limb muscles and between pairs of EMG signals from more proximal muscles is more complex, regardless of whether relationships are calculated from time or frequency domain estimates. In many patients, temporal relationships were inverted so that EMG led EEG or a distal muscle lead a proximal muscle. In these instances, an afferent loop is implicated and the myoclonic bursts in such proximal muscles may be the product of a complex interaction of cortical, subcortical and spinal influences. It is worth noting that despite the differences in phase relationships between proximal and distal muscles, coherence did not involve activity over systematically different frequency bands.
It is also notable that not all patients demonstrated temporal relationships in proximal muscles suggestive of afferent loops. The consistency of findings for different proximal muscles within the same subject argues that this is likely to represent biological variation. A variability similar to ours has been reported in the phase differences between cortex and forearm muscles in the physiological action tremor of healthy subjects (Marsden et al., 2001) and in patients with tremor due to Parkinson’s disease. Here estimated phase delays between cortex and forearm muscles were widely distributed with cortex leading or lagging by as much as 76 ms (Hellwig et al., 2000; Salenius et al., 2002). The implication is that individual variation in the organization and dominance of afferent and efferent loops to upper limb muscles occurs outside the hand. Some of this individual variation may be pathological, although in our patients there seemed no consistent correlation between the variation in phase and either the semiology of the myoclonus or the presence of concomitant epilepsy or drugs. On the other hand, physiological inter-individual variation in the organization of motor pathways to proximal muscles is increasingly recognized and may underlie the variability in recovery following stroke (Hamdy and Rothwell, 1998; Turton and Lemon, 1999).
The frequent finding of prolonged delays between EMG and EEG when estimates indicated afferent conduction may, at least in theory, be due to conduction delays of somatosensory pathways as documented in some myoclonic syndromes, such as Angelman’s syndrome (Guerrini et al., 1996). However, in our patients the N1-latency was within normal limits (mean 18.6 ms ± 1.6, 95% confidence limits; value not available in Case 9). Neither do delays due to cortico-cortical spread of afferent triggered cortical activity (Brown et al., 1991) seem sufficient to account for the very excessive delays found in some of our patients. One possibility is the involvement of afferent pathways with indirect projections to cortex.
Delays to distal muscles
Although cortical activity lead EMG in distal upper limb muscles, the sensorimotor cortex’s lead over muscle was, in our patients, slightly shorter than expected from the TMS-induced MEP latency in the respective active muscle, recorded in healthy controls. Similar observations have been made in studies of myoclonic patients using back-averaging (Cantello et al., 1997) and in healthy subjects regardless of whether EEG was recorded with bipolar electrodes as here (personal observations), Laplacian or current source derivations (Mima and Hallett, 1999a, b). There may be several explanations for this, including the additional synaptic delay during cortical activation by (submaximal intensity) TMS, the way in which delay is calculated from a single point rather than the whole EMG waveform in TMS and back-averaging studies, and the low-pass filtering (with phase delay) of EEG by the skull (Pfurtscheller and Cooper, 1975). However, these factors are alone unlikely to explain the shorter cortical lead in the present patients as, in earlier studies using similar analytical techniques, we found that the phase differences between EEG and EMG in distal muscles were consistent with TMS-induced MEP latencies (Brown et al., 1999; Marsden et al., 2000b). Our earlier studies involved patients with large amplitude multifocal jerks rather than minipolymyoclonus. These differences in phase relationship could be reconciled if we were to assume mixed afferent and efferent loops to the distal muscles of the upper limbs, with activity in these loops occupying overlapping frequency bands, as for proximal muscles. In patients with large amplitude cortical myoclonic jerks, the efferent system might dominate so that phase differences mirrored closely those expected from TMS. In patients with minipolymyoclonus, there may be mixed afferent and efferent influences on distal muscles. The latter still dominate, so that cortex still leads, but the lead will be an underestimate as it reflects two processes occurring over the same frequency range.
Conclusion
EEG–EMG and EMG–EMG frequency analysis for distal muscles has potential diagnostic value in patients with high frequency rhythmic myoclonus in providing a simple and sensitive measure of the strength of functional coupling between cortex and muscle. Detailed consideration of EEG–EMG and EMG–EMG phase spectra also provides important information regarding the mechanisms underlying myoclonic bursts in different muscles. In the hand, phase suggests that efferent pathways dominate and jerking seems to be an expression of spontaneous cortical discharges that are intrinsically rhythmic. In more proximal muscles, phase relationships may be dominated by either efferent or afferent loops arguing that myoclonus may arise from spontaneous rhythmic cortical discharges or self-sustaining myoclonic activity through afferent-efferent loops.
|
Acknowledgements |
|---|
This work was supported by The Medical Research Council and GlaxoSmithkline.
|
References |
|---|
|
TopSummaryIntroductionPatients and healthy controlsMethodsResultsDiscussionReferences |
|---|
Brown P, Grosse P. EEG–EMG coherence in movement disorders. In: Reisin RC, Nuwer MR, Hallett M, Medina C, editors. Advances in clinical neurophysiology. Clinical neurophysiology suppl 54. Amsterdam: Elsevier; 2002. p. 382–6.
Brown P, Marsden CD. Rhythmic cortical and muscle discharge in cortical myoclonus. Brain 1996; 119: 1307–16.
Brown P, Day BL, Rothwell JC, Thompson PD, Marsden CD. Intrahemispheric and interhemispheric spread of cerebral cortical myoclonic activity and its relevance to epilepsy. Brain 1991; 114: 2333–51
Brown P, Salenius S, Rothwell JC, Hari R. Cortical correlate of the piper rhythm in humans. J Neurophysiol 1998; 80: 2911–7.
Brown P, Farmer SF, Halliday DM, Marsden J, Rosenberg JR. Coherent cortical and muscle discharge in cortical myoclonus. Brain 1999; 122: 461–72.
Cantello R, Gianelli M, Civardi C, Mutani R. Focal subcortical reflex myoclonus. A clinical and neurophysiological study. Arch Neurol 1997; 54: 187–96.[Abstract]
Colebatch JG, Gandevia SC. The distribution of muscular weakness in upper motor neuron lesions affecting the arm. Brain 1989; 112: 749–63.
Eisen AA, Shytbel W. Clinical experience with transcranial magnetic stimulation. Muscle Nerve 1990; 13: 995–1011.[CrossRef][ISI][Medline]
Farmer SF, Bremner FD, Halliday DM, Rosenberg JR, Stephans JA. The frequency content of common synaptic inputs to motoneurons studied during voluntary isometric contraction in man. J Physiol 1993; 470: 127–55.
Feige B, Aertsen A, Kristeva-Feige R. Dynamic synchronization between multiple cortical motor areas and muscle activity in phasic voluntary movements. J Neurophysiol 2000; 84: 2622–9.
Ferbert A, Caramia D, Priori A, Bertolasi L, Rothwell JC. Cortical projection to erector spinae muscles in man as assessed by focal transcranial magnetic stimulation. Electroencephaplogr Clin Neurophysiol 1992; 85: 382–7.
Guerrini R, De Lorey TM, Bonanni P, MonclaA, Dravet C, Suisse G, et al. Cortical myoclonus in Angelman syndrome. Ann Neurol 1996; 40: 39–48.[CrossRef][ISI][Medline]
Guerrini R, Bonanni P, Patrignani A, Brown P, Parmeggiani L, Grosse P, et al. Autosomal dominant cortical myoclonus and epilepsy (ADCME) with complex partial and generalized seizures: a newly recognized epilepsy syndrome with linkage to chromosome 2p11.1-q12.2. Brain 2001; 124: 2459–75.
Hallett M, Wilkins DE. Myoclonus in Alzheimer’s disease and minipolymyoclonus. Adv Neurol 1986; 43: 399–405.[Medline]
Halliday DM, Rosenberg JR, Amjad AM, Breeze P, Conway BA, Farmer SF. A framework for the analysis of mixed time series/point process data – theory and application to the study of physiological tremor, single motor unit discharges and electromyograms. [Review]. Prog Biophys Mol Biol 1995; 64: 237–78.[CrossRef][ISI][Medline]
Halliday DM, Conway BA, Farmer SF, Rosenberg JR. Using electroencephalography to study functional coupling between cortical activity and electromyograms during voluntary contractions in humans. Neurosci Lett 1998; 241: 5–8.[CrossRef][ISI][Medline]
Hamdy S, Rothwell JC. Gut feelings about recovery after stroke: the organization and reorganization of human swallowing motor cortex. [Review]. Trends Neurosci 1998; 21: 278–82.[CrossRef][ISI][Medline]
Hellwig B, Häußler S, Lauk M, Guschlbauer B, Köster B, Kristeva-Feige R, et al. Tremor-correlated cortical activity detected by electroencephalography. Clin Neurophysiol 2000;111: 806–9.
Hellwig B, Häußler S, Schelter B, Lauk M, Guschlbauer B, Timmer J, et al. Tremor-correlated cortical activity in essential tremor. Lancet 2001;357: 519–23.
Ikeda A, Kakigi R, Funai N, Neshige R, Kuroda Y, Shibasaki H. Cortical tremor: variant of cortical reflex myoclonus. Neurology 1990; 40: 1561–5.
Kilner JM, Baker SN, Salenius S, Jousmaki V, Hari R, Lemon RN. Task-dependent modulation of 15–30 Hz coherence between rectified EMGs from human hand and forearm muscles. J Physiol 1999; 516: 559–70.
Leocani L, Comi G. EEG coherence in pathological conditions. [Review]. J Clin Neurophysiol 1999; 16: 548–55.[CrossRef][ISI][Medline]
Marsden JF, Farmer SF, Halliday DM, Rosenberg JR, Brown P. The unilateral and bilateral control of motor unit pairs in the first dorsal interosseous and paraspinal muscles in man. J Physiol 1999; 521: 553–64.
Marsden JF, Werhahn KJ, Ashby P, Rothwell JC, Noachtar S, Brown P. Organization of cortical activities related to movement in humans. J Neurosci 2000a; 20: 2307–14.
Marsden JF, Ashby P, Rothwell JC, Brown P. Phase relationship between cortical and muscle oscillations in cortical myoclonus: electrocorticographic assessment in a single case. Clin Neurophysiol 2000b; 111: 2170–4.[CrossRef][ISI][Medline]
Marsden JF, Brown P, Salenius S. Involvement of the sensorimeter cortex in physiological force and action tremor. Neuroreport 2001; 12: 1937–41.[CrossRef][ISI][Medline]
McAuley JH, Rothwell JC, Marsden CD. Frequency peaks of tremor, muscle vibration and electromyographic activity at 10 Hz, 20 Hz and 40 Hz during human finger muscle contraction may reflect rhythmicities of central neural firing. Exp Brain Res 1997; 114: 525–41.[CrossRef][ISI][Medline]
Mima T, Hallett M. Corticomuscular coherence: a review. [Review]. J Clin Neurophysiol 1999a; 16: 501–11.[ISI][Medline]
Mima T, Hallett M. Electroencephalographic analysis of cortico-muscular coherence: reference effect, volume conduction and generator mechanism. Clin Neurophysiol 1999b; 110: 1892–9.[CrossRef][ISI][Medline]
Mima T, Matsuoka T, Hallett M. Functional coupling of human right and left cortical motor areas demonstrated with partial coherence analysis. Neurosci Lett 2000; 287: 93–6.[CrossRef][ISI][Medline]
Ohara S, Nagamine T, Ikeda A, Kunieda T, Matsumoto R, Taki W, et al. Electrocorticogram-electromyogram coherence during isometric contraction of hand muscle in human. Clin Neurophysiol. 2000; 111: 2014–24.[CrossRef][ISI][Medline]
Palmer E, Ashby P. Corticospinal projections to upper limb motoneurons in humans. J Physiol 1992; 448: 397–412.
Pfurtscheller G, Cooper R. Frequency dependence of the transmission of the EEG from cortex to scalp. Electroencephalogr Clin Neurophysiol 1975;38: 93–96.
Rothwell JC, Thompson PD, Day BL, Boyd S, Marsden CD. Stimulation of the human motor cortex through the scalp. [Review]. Exp Physiol 1991; 76: 159–200.[ISI][Medline]
Salenius S, Pontin K, Kajola M, Salmelin R, Hani R. Cortical control of human motoneuron firing during isometric contraction. J Neurophysiol 1997; 77: 3401–5.
Salenius S, Avikainen S, Kaakkola S, Hari R, Brown P. Defective cortical drive to muscle in Parkinson’s disease and its improvement with levodopa. Brain 2002; 125: 491–500.
Shibasaki H, Kuroiwa Y. Electroencephalographic correlates of myo clonus. Electroencephalogr Clin Neurophysiol 1975; 39: 455–63.[CrossRef][ISI][Medline]
Srinivasan R, Nunez PL, Silberstein RB. Spatial filtering and neocortical dynamics: estimates of EEG coherence. IEEE Trans Biomed Eng 1998; 45: 814–26.[CrossRef][ISI][Medline]
Terada K, Ikeda A, Mima T, Kimura K, Nagahama Y, Kamioka Y, et al. Familial cortical myoclonic tremor as a unique form of cortical reflex myoclonus. Mov Disord 1997; 12: 370–7.[CrossRef][ISI][Medline]
Thompson PD, Bhatia KP, Brown P, Davis MB, Pires M, Quinn NP, et al. Cortical myoclonus in Huntington’s disease. Mov Disord 1994; 9: 633–41.[CrossRef][ISI][Medline]
Toro C, Pascual-Leone A, Deuschl G, Tate E, Pranzatelli MR, Hallett M. Cortical tremor. A common manifestation of cortical myoclonus. Neurology 1993; 43: 2346–53.
Turton A, Lemon RN. The contribution of fast corticospinal input to the voluntary activation of proximal muscles in normal subjects and in stroke patients. Exp Brain Res 1999; 129: 559–72.[CrossRef][ISI][Medline]
Wilkens DE, Hallett M, Erba G. Primary generalized epileptic myoclonus: a frequent manifestation of minipolymyoclonus of central origin. J Neurol Neurosurg Psychiatry 1985; 48: 506–16.[Abstract]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|