[page 66↓]

5.  Coherence analysis in the myoclonus of corticobasal degeneration

Myoclonus is a common finding in corticobasal degeneration (CBD), with more than half of patients exhibiting jerks during the course of the disease (Kosmopoliti et al., 1998). It has features in common with cortical myoclonus as it is predominantly distal, action-in­duced and stimulus sensitive with a rostro-caudal spread of the myoclonic bursts along the affected limb (Thompson et al., 1994). However, classical neurophysiological techniques have generally failed to confirm a cortical origin for the myoclonus in CBD. Only one pa­tient has been reported in whom back-averaged surface EEG disclosed a time-locked corti­cal correlate (Tanosaki et al., 1999) and a further patient has been reported in whom a corti­cal correlate was identified upon magnetoencephalography Mima et al., 1998a). Similarly, in a large study comprising 14 patients a giant sensory evoked potential was only found in one case (Thompson et al., 1994). The rarity of electrophysiological evidence of a cortical origin could be due to methodological difficulties, pathological involvement of the sensori­motor cortex (Brunt et al., 1995; Lu et al., 1998) or a subcortical origin for the myoclonus.

The myoclonus of CBD tends to be of high frequency and low amplitude. Under these circumstances we expected frequency analysis to have some advantages over time-domain methods, such as back-averaging, in the detection of a cortical correlate as shown in other conditions associated with cortical myoclonus (Guerrini et al., 2001, Brown et al., 1999).

5.1. Patients and methods

5.1.1. Patients

Five patients (table 5.1) who fulfilled the criteria of clinically probable CBD (Riley and Lang, 2000) were examined. All patients exhibited unilateral action-induced and stimulus sensitive myoclonus in the upper extremity except case 5 who had bilateral myoclonus. 4 [page 67↓]healthy subjects (mean age: 62 years, range: 55-65 years) were also studied. Patients and healthy subjects gave their informed consent to the study, which was approved by the local ethics committee.

Table 5.1.






duration (yrs)

Peak frequency of

myoclonic bursts [Hz]

Clinical features







unilateral asymmetric akinetic-rigid syndrome; limb apraxia, dystonia, dysarthria

L-Dopa, Baclofen, Sodium Valproate






unilateral asymmetric akinetic-rigid syndrome; dystonia, pain

Morphine, Gabapentin






unilateral asymmetric akinetic-rigid syndrome; dystonia, pain

Sodium Valproate






unilateral asymmetric akinetic-rigid syndrome; painful limb dystonia, numbness







unilateral asymmetric akinetic-rigid syndrome; dystonia, dysphasia


5.1.2. Recordings

Surface EMG and EEG were recorded with 9 mm diameter silver-silver chloride elec­trodes. Bipolar EEG was recorded from FC3/4-C3/4. EMG was recorded bilaterally from finger extensor and first dorsal interosseous (1DI) muscles in cases 1 and 2 and from the affected finger extensors and 1DI in cases 3-7. EMG and EEG were amplified, band-pass filtered and sampled at 600 Hz (cases 1-2) or 2 kHz (cases 3-7). Signals were displayed and stored on a PC using CED Spike 2, version 4 software. Patients were recorded during po­stural contraction which would trigger the myoclonus. Healthy subjects were asked to co-activate recorded muscles over 4 periods of about 60 seconds with 60 seconds rest between contractions.

[page 68↓]

5.1.3.  Frequency analysis and back-averaging

Analysed record lengths were kept constant at 200 seconds in all subjects. The EEG and rectified EMG were assumed to be realisations of stationary zero mean time series. EEG-EMG and EMG-EMG coherence were analysed using methods outlined in chapter II. Signals were interpolated or down-sampled to a sampling rate of 1 kHz and block size was 1024 data points. The phase was only assessed where coherence was significant and extended over at least 5 consecutive data points.

First-order partial coherence functions were also estimated to assess whether ‘partialisa­tion’ with a third process (the ‘predictor’) accounted for the relationship between two other processes (Halliday et al., 1995; Rosenberg et al., 1998). As used here, partial coherence can be viewed as representing the fraction of coherence between EMG signals that is not shared by EEG (Halliday et al., 1995; Rosenberg et al., 1998; Spauschus et al., 1999). Thus, if sharing of the signal between the different EMG signals and EEG were complete, then partialization of the coherent activity between the EMGs with contralateral EEG as the predictor would lead to zero coherence. It follows that if EEG had no influence on the coupling between EMGs then partialization with EEG would have no effect on EMG-EMG coherence.

Back-averaging was performed off-line in Spike 2. EMG was rectified and myoclonic EMG bursts identified using a level of 100 μ V to produce a series of digital events. EEG and EMG signals were then re-aligned to these events and averaged.

5.1.4. Statistics

The power in each bin of autospectra was expressed as the relative percentage of the to­tal power of each autospectrum. The variance of the coherence was normalised 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 [page 69↓] each record given by 1/2L where L is the number of segment lengths used to calculate the coherence.

To compare autospectra and coherences between patients and normals repeated mea­sures ANOVA (analysis of variance) was performed. When a significant difference was present, post-hoc pair-wise comparison with Scheffé correction was carried out.

5.2. Results

Fig 1A is an example of the raw EMG, showing high frequency myoclonic bursts in both finger extensor and 1DI at a rate of ~12 Hz, together with the accompanying EEG. Note the absence of a normal interference pattern in Fig 5.1A, with no pre-innervation between myoclonic bursts. Fig 5.1B shows the corresponding autospectra of EEG and EMG and Fig 5.1C is the coherence between finger extensor and 1DI, all in the same pa­tient (case 1). Intermuscular coherence is excessively exaggerated up to ~60 Hz. Partial co­herence between between finger extensor and 1DI with the EEG as predictor shows only a slight reduction of coherence in the frequency range between 6 and 11 Hz.

[page 70↓]

Fig. 5.1: (A): Raw EEG and EMG of case 1 exhi­biting irrregular short myoclonic bursts at an average frequency of ~12 Hz during a postural con­traction. (B) Normalised autospectra of EEG over FC3-C3 and EMG from finger extensor and 1DI. (C) Intermuscular cohe­rence (thick line) between finger extensor and 1DI discloses exaggerated co­herence up to 58 Hz. Note that the partial cohe­rence between the two muscles (thin line) with the EEG as predictor is only slight­ly lower from 6 to 12 Hz.

[page 71↓]

Similar to case 1, all the remaining patients had abnormally inflated EMG-EMG cohe­rence and the results of all 5 patients are pooled in Fig 5.2. Fig 5.2 A is the mean percen­tage total EEG power at each frequency in the five patients recorded over the sensorimotor area ipsilateral and contralateral to the affected limb, compared to that found in age-matched healthy subjects. The normal power increase in the beta range (15-30Hz) with a peak centered around 20 Hz is absent in the patients and there is a shift of EEG activity to the theta range, with a peak at ~8 Hz, which is, however, not statistically significant. Fi­gures 5.2 B and 5.2C are the mean percentage total EMG power at each frequency for the forearm extensors and 1DI, respectively. The patients show a large peak centred around 15 Hz. Figure 5.2D is the mean transformed coherence between the ipsilateral forearm exten­sor and 1DI EMG signals in the five patients, compared to that in high frequency cortical myoclonus (Grosse et al., 2003) and healthy subjects. The five patients with CBD have a grossly inflated EMG-EMG coherence over a broad band that even exceeds that in the pa­tients with established high-frequency cortical myoclonus. Transformed coherence reached up to 1.5 (range 0.4-1.5) and was above the 95%-confidence limit up to 60 Hz. Coherence for CBD patients was significantly different from normals (p<0.001) in both the 8-30 Hz (α and β bands) and 31-60 Hz (γ band), while patients with cortical myoclonus only differed form normals over the 8-30 Hz range (p<0.05) (Fig. 5.2E). In CBD, phase spectra sugges­ted that EMG activity in the forearm extensors preceded that in 1DI by 5.5 ± 0.7 ms (range 3.1 –7.7 ms), consistent with synchronisation through an efferent drive and excluding vo­lume conduction as an explanation for the high levels of EMG-EMG coherence (Fig. 5.2F). Similar lags were seen in cortical myoclonus and healthy subjects.

[page 72↓]

Fig. 5.2: Pooled results in the 5 pa­tients compared to age-matched heal­thy subjects. Normalised power for EEG (A), finger extensor (B) and 1DI (C). Pooled transformed coherence for finger extensor and 1DI is inflated in the range up to 60 Hz with a peak centered around 15 Hz (D). Note that EMG-EMG coherence for patients with established high frequency corti­cal myoclonus is less exaggerated and occupies a narrower frequency band. (E) In CBD patients coherence is sig­nificantly different from both normals and patients with cortical myoclonus across 8-30 Hz and 31-60 Hz, while for patients with cortical myoclonus only the 8-30 Hz band is statistically different from normals. Error bars in­dicate standard error of the mean. (*=p<0.05). (F) Time delays between the two muscles for patients with CBD, patients with cortical myo­clonus and healthy controls showing an appropriate delay between 1DI and finger extensors, thereby indicating that high levels of coherence were not due to volume conduction.

[page 73↓]

Despite the grossly inflated intermuscular coherence, significant cortico-muscular cohe­rence was only found for both finger extensor and 1DI in case 1 and occurred over a nar­row frequency range centred around 10 Hz (Fig 5.3 A) which corresponds to the drop in coherence when partialisation with EEG was performed (shown in Fig.5.1 C). On the affec­ted side the phase spectrum was suggestive of an afferent drive, as 1DI EMG lead EEG by 51.7 ± 6.4 ms (Fig 5.3 B) and finger extensor EMG lead EEG by 58.2 ± 10.8 ms. In con­trast, over the unaffected side EEG lead finger extensor EMG by 5.4 ± 2.6 ms, consistent with a predominantly cortico-spinal drive while phase for 1DI was not significant.

Fig. 5.3: (A) Transformed coherence between affected side (FC3/C3-right 1DI) and unaffected side (FC4/C4-left 1DI) in case 1 showing exaggerated co­herence on the affected side up to 18 Hz with a distinct peak at 10 Hz which is neither present on the unaffected side nor in averaged coherence in normals. Com­pare this narrow EEG-EMG coherence with the broad band of EMG-EMG cohe­rence in the same patient shown in Fig 1C. (B) Phase spectrum on the affected side discloses that right 1DI EMG leads EEG by 28.4 ± 7.7ms.

Back-averaged EEG and cumulant density estimates were negative or disclosed ambigu­ous results with no unequivocal EEG cortical correlate preceding the onset of averaged EMG.

[page 74↓]

5.3.  Discussion

It can be shown that in five patients with the clinical diagnosis of probable CBD a dis­tinctive pattern of inflated EMG-EMG coherence over a broad band without evidence of a comparably exaggerated EEG-EMG coherence was present. The very high levels of inter­muscular coherence in the affected limb are indicative of an abnormally strong common drive to these muscles. There are several reasons for believing that this common drive might differ in some respects from other forms of cortical myoclonus which have been stu­died so far. Using similar techniques, exaggerated EMG-EMG coherence can be seen in rhythmic high-frequency cortical myoclonus, but in this condition this feature is accompa­nied by abnormal EEG-EMG coupling to finger extensors and 1DI over a common fre­quency range (Grosse et al., 2003), in which EEG phase leads EMG (Farmer et al., 1993). We found no significant EEG-EMG coherence in four of our patients, and, in the one pa­tient in whom this activity was present, EMG lead EEG on the affected side, indicating re-afference rather than corticospinal drive. Although motor cortex pathology is common in CBD, affecting most likely both cortico-cortical and efferent cortical projections (Arm­strong et al., 2000; Armstrong et al., 2001) it is, however, unlikely that this could entirely obscure cortico-muscular coherence while leaving the structures generating myoclonus pre­served (Grosse et al., 2003). Cortical myoclonus is the result of the synchronised discharge of pyramidal neurones in the motor cortex and it is depolarisation of these neurones that likely accounts for the scalp negative cortical correlate that underlies cortico-muscular co­herence. High frequency cortical myoclonus is likely to represent an exaggeration of the physiological tendency of cortical pyramidal neurones to synchronise in the lower beta band (Grosse et al., 2003) and it is conspicuous that the normal peak at this frequency was absent from the EEG picked up over the sensorimotor cortex in our patients with CBD. By exclusion, then, it is possible to hypothesise that the elevated EMG-EMG coherence in the absence of any significant EEG-EMG coherence found in our patients was either subcorti­cal in origin or that it reflects a disturbed interaction and gradual disintegration of the net­work between the sensorimotor cortex and subcortical structures as has been previously [page 75↓]suggested (Carella et al., 1997). The involvement of a subcortical generator of the myoclo­nus would be consistent with the exceptionally short latency of reflex myoclonus in CBD compared to typical cortical reflex myoclonus (Thompson et al., 1994; Lu et al., 1998).

© Die inhaltliche Zusammenstellung und Aufmachung dieser Publikation sowie die elektronische Verarbeitung sind urheberrechtlich geschützt. Jede Verwertung, die nicht ausdrücklich vom Urheberrechtsgesetz zugelassen ist, bedarf der vorherigen Zustimmung. Das gilt insbesondere für die Vervielfältigung, die Bearbeitung und Einspeicherung und Verarbeitung in elektronische Systeme.
DiML DTD Version 3.0Zertifizierter Dokumentenserver
der Humboldt-Universität zu Berlin
HTML generated: