The pathophysiology of giant SEPs in cortical myoclonus: a scalp topography and dipolar source modelling study

Electroencephalography and clinical Neurophysiology 104 (1997) 122–131

The pathophysiology of giant SEPs in cortical myoclonus: a scalp topography and dipolar source modelling study Massimiliano Valeriani a, Domenico Restuccia a ,*, Vincenzo Di Lazzaro a, Domenica Le Pera a, Pietro Tonali a , b a

Department of Neurology, Universita` Cattolica del Sacro Cuore, Roma, Italy b The CSS Hospital IRCCS, San Giovanni Rotondo, Italy Accepted for publication: 6 December 1996

Abstract Somatosensory evoked potential (SEP) recordings in patients suffering from cortical myoclonus (CM) are characterised by evidence of abnormally enhanced scalp components. Our aim was to verify whether enhanced activity in giant SEPs arises from the same generators as in healthy subjects. We used the brain electrical source analysis (BESA) to compare scalp SEP generators of healthy subjects to those calculated in 3 patients with CM of varying causes. Firstly, we built a 4-dipole model explaining scalp distribution of early SEPs in normal subjects and then applied it to traces recorded from CM patients. Our model, issued from the right median nerve grand average and applied also to recordings from single individuals, included a dipole at the base of the skull and three other perirolandic dipoles. The first of the latter dipoles was tangentially oriented and was active at the same latencies as the N20/P20 potentials and, with opposite polarity, the P24/ N24 responses; the second dipole explained the central P22 distribution and the third had a peak of activity corresponding to the N30 component. When we applied our 4-dipole model to CM recordings, the first perirolandic dipole had a third peak of activity in all patients at the same latency as a parietal negativity and a frontal positivity, both following giant P24/N24 components; on the other hand, in one patient the second perirolandic dipole showed a later activation corresponding to a high central negativity, following a giant P22 response. We suggest that only the initial giant SEPs correspond to physiological potentials evoked in healthy subjects. The occurrence of late giant SEPs could be explained by hyperpolarization, following the postsynaptic excitatory potentials responsible for the early giant components.  1997 Elsevier Science Ireland Ltd. Keywords: Giant SEPs; Cortical myoclonus; Brain electrical source analysis (BESA)

1. Introduction Cortical myoclonus (CM) is a fairly common involuntary movement due to various causes and characterised by irregular, sudden, jerky contractions of a muscle or group of muscles without loss of consciousness. It is commonly classified as reflex or spontaneous CM depending on its sensitivity to different stimuli. Since the first report by Dawson (1946), it is well known that evoked responses of enormous amplitude can be recorded in patients suffering from CM, although the pathogenesis of this abnormality has still to be clarified. Nevertheless, on the basis of the similar distribu-

* Corresponding author. Istituto di Neurologia, Policlinico A. Gemelli, L.go A. Gemelli 8, 00168 Roma, Italy; fax: +39 6 35501909.

0168-5597/97/$17.00  1997 Elsevier Science Ireland Ltd. All rights reserved PII S0921-884X(97)9661 6-9

tion of normal amplitude and giant somatosensory evoked potentials (SEPs) previous studies have suggested that the latter may result from a pathological enhancement of certain normal early cortical SEPs, such as ‘P25’, ‘N33/35’, ‘N30’ and ‘P30’ (Shibasaki et al., 1985; Kakigi and Shibasaki, 1987; Ebner and Deuschl, 1988; Shibasaki et al., 1990; Ikeda et al., 1995). Since many components overlie each other in the first 40 ms of SEPs and giant responses may ‘cover’ normal amplitude potentials that are close in latency and distribution, a simple visual inspection of the scalp distribution of different amplitude potentials cannot be unambiguous, even when using brain mapping of SEPs. The spatial and temporal resolutions of cortical responses can be improved in both healthy subjects and CM patients by using the brain electrical source analysis (BESA). Though BESA does not provide a precise anatomical loca-

EEP 96616

M. Valeriani et al. / Electroencephalography and clinical Neurophysiology 104 (1997) 122–131

tion of the dipolar sources, it proved to be accurate enough to localise correctly the generators of various modalities of EPs reflecting responses from well identified primary sensory areas (Scherg et al., 1989; Simpson et al., 1990; Franssen et al., 1992; Valeriani et al., 1996). In particular, BESA is well adapted to separate clearly the activities of neighbouring cerebral structures and, therefore, the sources of cortical SEPs. The aim of our study was to determine whether giant SEPs in CM patients could be generated by the same sources which are usually active in the first 40 ms of normal scalp SEPs. In order to achieve our purpose, we built a BESA model which explained well the scalp SEP distribution in normal subjects and applied it to the giant SEPs recorded from our CM patients.

2. Materials and methods 2.1. Healthy subjects and patients We recorded scalp SEPs to 12 right and 9 left median nerve stimulation from 14 healthy subjects and scalp SEPs to right median nerve stimulation in two CM patients and to bilateral median nerve stimulation in a third patient. 2.2. Patient no. 1 This 21 year old woman had presented involuntary movements of hands and eyelids since the age of 12 years and, later, began to suffer generalised seizures with loss of consciousness during sleep. The seizures were controlled by therapy with clonazepam and phenobarbital. In the 2 years before our examination, she had developed dysarthria and ataxia and myoclonic jerks had spread to all 4 limbs. Family history was consistent with a recessive autosomal inheritance. EEG recording showed epileptic spikes all over the scalp. Cerebral MRI, muscle biopsy and EMG were negative. A probable diagnosis of Ramsay-Hunt syndrome was made. 2.3. Patient no. 2 This 44 year old woman had for 10 years suffered hallucinatory episodes and loss of consciousness with clonic movements of all 4 limbs. More recently, she began to suffer myoclonic jerks of limbs and face, which were worse when she closed her eyes, without loss of consciousness. She also had slight intellectual deterioration. Therapy with valproic acid reduced seizure frequency. EEG recording showed spikes induced by closure of eyes throughout the scalp. MRI showed diffused cerebral atrophy; muscle biopsy and EMG were negative. The diagnosis was progressive myoclonus epilepsy of unknown origin.

123

2.4. Patient no. 3 This 40 year old woman had developed myoclonic jerks in all four limbs, especially during intended movements, dysarthria and tremor. The symptoms had progressively worsened and the patient was unable to walk at the time of our examination. EEG recording showed infrequent spikes all over the scalp. Cerebral CT scan, muscle biopsy and EMG were normal. Diagnosis was unknown. 2.5. SEP recording For SEP recording, subjects lay on a couch in a warm and semidarkened room. Stimulations of median nerve (0.2 ms duration, 3 Hz) were delivered by skin electrodes at the wrist; the stimulus intensity was adjusted slightly above the motor threshold. Disk recording electrodes (impedance below 5 kQ) were placed at 20 locations of the 10-20 system (excluding Fpz). The reference electrode was at the lobe of the ipsilateral ear, as suggested by previous reports (Desmedt et al., 1987; Tomberg et al., 1991), and the ground at the stimulated arm. The analysis time was 64 ms, including also 5 ms of preanalysis, with a bin width of 250 ms. The amplifier bandpass was 1–3000 Hz (12 dB roll off). An automatic artifact-rejection system excluded from the average all runs containing transients exceeding ±65 mV at any recording channel. In order to ensure baseline stabilisation, SEPs were digitally filtered off-line by means of a digital filter with a bandpass of 19–1900 Hz. Two averages of 1000 trials each were obtained and printed out by the computer on a desk-jet printer. Frozen maps showing the distribution of the responses over the scalp were obtained by linear interpolation from the 4 nearest electrodes. 2.6. Data analysis SEPs were identified on the basis of latency, polarity and scalp distribution. Amplitudes and peak latencies were measured on the average of the two runs obtained for each median nerve. Amplitudes were measured from the baseline. Limits of normal values of both latencies and amplitudes were defined as mean plus 3 SD. Peak latencies were compared by paired t test. A grand average of the 12 normal right median nerve SEPs was also obtained. All traces were made coincident at the latency of the P14 potential prior to grand averaging, the preanalysis delay being unaffected by this manoeuvre. We performed this normalisation of latencies to avoid artificial ‘smoothing’ due to different latencies of the responses between subjects. 2.7. Brain electric source analysis Detailed description of BESA is reported elsewhere (Scherg, 1990). Basically, BESA is a program that calculates potential distributions over the scalp from preset voltage dipoles within the brain. It also shows the degree of

124

M. Valeriani et al. / Electroencephalography and clinical Neurophysiology 104 (1997) 122–131

agreement between the recorded and calculated maps and, particularly, the percentage of data that cannot be explained by the model card is expressed as residual variance (RV). The model suggested by BESA is a hypothesis that does not exclude other solutions; nevertheless, BESA results can be validated when applicable to individual data. The different solutions which can be found by BESA depend mostly on the strategy by which the model is obtained. Since our attempt to build a model which explains the distribution of scalp SEPs to upper limb stimulation in healthy subjects is not the first, we thought it reasonable to apply initially to our data the solutions proposed by previous studies (Franssen et al., 1992; Buchner et al., 1995).

3. Results 3.1. SEP data in healthy subjects As shown in Fig. 1, the lemniscal P14 response was recorded by all scalp electrodes. N20 and P20 responses were well identifiable in the centro-parietal region contralateral to the stimulation and in the frontal traces respectively. The P20 amplitude reached its maximal value at Fz. The P22 potential, significantly different in latency from the frontal P20 (P , 0.01), was recorded by the central electrode contralateral to the stimulation site in 16/21 median nerve SEPs. A parietal positivity and a frontal negativity without any significant difference in latency (P = 0.1) were identified in 13/21 median nerve SEPs. These last potentials were labelled as N24 and P24 in a previous paper (Garcia-Larrea et al., 1992). The P24 and P22 latencies were significantly different (P , 0.01); the mean N24 amplitude was maximal at Fz. Lastly, the N30 response was identifiable in all subjects in the fronto-central region and its maximal mean amplitude was recorded by the Fz electrode. This potential was constantly present in frontal traces and at Cz, while it was recorded only in 17/21 median nerve SEPs by the central electrode contralateral to the stimulation. Table 1 shows means and SD of all scalp responses; amplitudes were measured in the traces where they were maximal in the grand average waveforms. 3.2. SEP data in CM patients In all patients the N20 potential was normal in both latency and amplitude. Patients nos. 1 and 2 showed a simi-

Fig. 1. Grand average of right median nerve SEPs across all our normal subjects. The figure shows traces recorded by 20 scalp electrodes in the 1020 system locations. The P14 scalp far-field is largely diffused over the scalp and is indicated in the Fz trace. The N20 and P20 responses are shown in the P3 and F3 traces, respectively, while the P22 potential reaches its maximal amplitude in the left central region. Frontal N24 and parietal P24 potentials are indicated in F3 and P3 recordings, respectively, and the N24 response is represented by a shoulder on the rising branch of the N30 potential. Lastly, the N30 potential is labelled in the Fz and Cz traces, although it is identifiable also in the other frontal and C3 traces.

lar SEP pattern (Fig. 2), characterised by a giant parietal positivity (Pp) following the N20 potential and almost coincident in latency with a giant frontal negative response (Nf). Two lower responses, which were labelled according to their polarity as Np and Pf, were recorded later in parietal and frontal traces, respectively. The central electrode contralateral to the stimulation recorded first a positive (Pc) and later a negative (Nc) response, showing almost the same latency as Pp and Np potentials, although smaller in amplitude. Brain maps, calculated at the latencies of the Pp/Nf and Np/Pf responses, showed a dipolar distribution. In patient no. 3 (Fig. 3), parietal and frontal traces showed the same features as in the other patients, while the Pc potential was different in latency from the Pp response; conversely, the Nc latency was almost coincident with the Np/Pf latency. Brain maps showed a dipolar distribution at the latency of the Pp/Nf response, whereas a high positivity and negativity were highly focused in the central region contralateral to the stimulation at the same latency as the

Table 1 SEP data in normal subjects

Latency (ms) Amplitude (mV)

P14

N20

P20

P22

N24

P24

N30

15.2 ± 0.99

20.6 ± 1.31 N20 (Pc) 1.7 ± 0.82

21.6 ± 1.4 P20 (Fz) 1.2 ± 0.61

23.7 ± 2.21 P22 (Cc) 1.2 ± 0.6

26.6 ± 2.4 N24 (Fz) 1.1 ± 0.58

26.3 ± 2.31 P24 (Pc) 0.9 ± 0.58

32 ± 3.7 N30 (Fz) 1.3 ± 0.7

Data are mean ± SD. Cc, central lead contralateral to the stimulation; Pc, parietal lead contralateral to the stimulation.

M. Valeriani et al. / Electroencephalography and clinical Neurophysiology 104 (1997) 122–131

Pc and Nc potentials, respectively. SEP data in CM patients are shown in Table 2. 3.3. Brain electrical source analysis in healthy subjects and patients We analysed the grand average of normal right median nerve SEPs over the time interval, where all the abovedescribed responses could be identified. First, we tested a 3-dipole model, which, it has been suggested, provides a satisfactory explanation of the cortical distribution of early scalp SEPs to upper limb stimulation (Franssen et al., 1992; Buchner et al., 1995). This model included a dipole at the base of the skull, which was active at the latency of the P14 potential, and two other perirolandic dipoles, the first tangentially and the second radially oriented. The time courses of these dipoles were in accordance with that found by the above-mentioned authors. The tangential dipole showed a first peak of activity at the same latency as the N20/P20 responses and a later activation of inverted polarity at the latency of the P24/N24 potentials. The radial perirolandic dipole showed a first peak of activity at the latency of the central P22 and later inverted its polarity at the same latency as the N30 potential. After fitting each dipole in the time interval where it was active, this model gave a RV of 7% and therefore we considered other solutions to explain our data. Since in the three-dipole model RV was generated mostly in the N30 latency range, we added another dipole and allowed it to change freely in both orientation and location. After fitting, the RV decreased to 4.4% (Fig. 4). The dipole at the base of the skull (no. 1) and the tangential perirolandic dipole (no. 2) retained almost the same location, orientation, and time course as in the three-dipole model. Conversely, the dipole (no. 3) that was radially oriented in the previous model assumed a more tangential orientation and served only the first peak of activity corresponding to the P22 latency. The last dipole (no. 4) reached a radial orientation and a medial location in the perirolandic region and showed a late peak of activity at the latency of the fronto-central N30. In order to test our model, we applied it to right median nerve SEPs recorded in each individual. Firstly, we allowed only the orientation of the dipoles to change, their location remaining constant, and ensured that time course of each dipole, as found in the grand average model, did not significantly change in each individual model. We obtained RV values ranging from 3.4% to 15.9% and, particularly, in 9/12 median nerve SEPs, which showed a good signal/noise ratio, RV did not exceed 10%. Co-ordinates of dipoles in models obtained from the grand average and from each individual are shown in Table 3. In order to determine whether other locations of dipoles could be excluded in individual models, we allowed also dipoles location to change freely, but failed to obtain any model with physiologically plausible dipole location giving better RV values than those issued from analysis of interindividual grand average.

125

We applied the 4-dipole model obtained in healthy subjects to right median nerve SEPs recorded in CM patients. In patients nos. 1 and 2, we allowed only the orientation of the dipoles to change and obtained RVs of 4.7% and 4.5%, respectively. Time course and strengths of dipoles were similar in the two patients (Fig. 5a). Dipole 2 showed a first minor peak of activity corresponding to the N20 latency and a second major peak of inverted polarity at the latency of the giant Pp/Nf responses; it then again inverted its polarity at the same latency as the Np/Pf potentials. The strength of the second peak of dipole 2 in these patients (2.7 meff and 1.8 meff in patients nos. 1 and 2, respectively) was out of our normal limits (mean + 3 SD = 0.7 meff). Dipole 3 showed a small peak of activity (0.4 meff and 0.5 meff in patients nos. 1 and 2, respectively) slightly preceding the Pc latency. Dipole 4 showed no activity. In a second session, we allowed also the locations of dipoles to change freely to avoid the exclusion of other locations of the giant components. Nevertheless, we did not obtain any model with physiologically plausible dipole location giving better RV values than those described above. Moreover, we observed that, also in the other models we tested, a same source was activated with opposite polarities at the same latencies of both the Pp/Nf and Np/Pf responses. In patient no. 3, we initially allowed only the orientation of the dipoles to change freely, but did not obtain a good RV value. We then allowed also the location of the dipoles to change and obtained a RV of 4.8% (Fig. 5b). It is interesting that the dipoles showed only minor variations in both orientation and location in comparison with the grand average model. Also in this case, as in the model obtained for patients nos. 1 and 2, dipole 2 showed a first small peak of activity at the latency of the parietal N20 and a second one of opposite polarity corresponding to the Pp/Nf latency; it then again inverted its polarity at the same latency as the Np/Pf potentials. The strength of the second peak of dipole 2 (2.8 meff) exceeded our normal limits. Conversely, the time course of dipole 3 differed substantially from that in the model of the first two patients. Indeed, dipole 3 showed in this case a major peak of activity simultaneously with the giant Pc potential and then inverted its polarity at the latency of the Nc response. The strength of the first peak of dipole 3 (1.9 meff) was out of our normal limits (mean + 3 SD = 0.7 meff). Dipole 4 showed no activity. When we compared orientations of all dipoles, expressed by means of the v and J angles, between the healthy subject and CM patient groups, we did not find any significant difference (MannWhitney U test, P . 0.05). Co-ordinates of the dipoles in models of patients are shown in Table 3.

4. Discussion In this study we show that a 4-dipole model explaining the scalp distribution of median nerve SEPs in healthy subjects may also be applied satisfactorily to recordings from

126

M. Valeriani et al. / Electroencephalography and clinical Neurophysiology 104 (1997) 122–131

Table 2 SEP data in CM patients Patients

1 2 3

Side

R L R R

N20

Pp

Np

Pc

Nc

Nf

Pf

Lat.

Amp.

Lat.

Amp.

Lat.

Amp.

Lat.

Amp.

Lat.

Amp.

Lat.

Amp.

Lat.

Amp.

18.8 19.5 18.5 19.3

2.3 2.1 4 2

25 25.3 23 25

15.4 16 16.2 7.3

40.3 40 40.8 31.8

8.5 7.5 5.6 4.6

24.3 24.6 22.3 23.7

7 8.2 12.2 9.2

40.3 39.8 40.8 31.3

3.2 2.9 3.5 7.5

25.5 25.3 23 25

11.9 12 13.3 5

41.8 40.5 38.8 30.5

7.8 6 4.5 3.9

Lat., latency; Amp., amplitude.

CM patients, although the time course of the dipoles differs from that observed in healthy subjects. In agreement with previous reports on dipolar modelling of early scalp SEPs to upper limb stimulation (Franssen et al., 1992; Scherg and Buchner, 1993; Buchner et al., 1995), our model, issued from healthy subjects, included a dipole in the basal region of the skull, which probably corresponds to the brainstem source of the P14 far-field potential. This dipole was localised almost 7 cm in the back of the centre of the sphere representing the head. This finding could be due, firstly, to an intrinsic fault of dipolar modelling, in which mean spatial resolution is generally 12 mm (Cuffin et al., 1991), but increases for deep sources. Moreover, the posterior location with orientation to forward of the dipole representing the P14 generator may be also explained if we take into account that the P14 response shows a larger amplitude in the frontal than in the parieto-occipital region (Restuccia et al., 1995). We did not observe, as other authors did, a phase-reversal of the dipole no. 1 at the latency of the brainstem N18. It is possible that the absence of any dipolar activity corresponding to the N18 in our model could be due to the minimal contribution that the N18 source provides to scalp traces recorded with an earlobe reference montage (Desmedt et al., 1987). For physical reasons a far-field source, such as that of the P14, should show a biphasic activity (Kimura et al., 1984); however, our finding of a single-phase source could be explained by the poor volume conductor description in the skull base, so that the two opposite phases of the same far-field source cannot have the same distribution. The first peak of activity of our dipole 2, which shows a perirolandic location and is tangentially

oriented, has the same latency as the N20/P20 potentials and thus may well correspond to the firing of the neuronal population generating these responses. Later, dipole 2 inverts its activity at the latency of the parietal P24 and frontal N24 potentials. It has previously been suggested that a tangential generator in the 3b area explains the dipolar distribution of the latter potentials (Allison et al., 1980; Garcia-Larrea et al., 1992) and both the location and orientation of our dipole 2 are consistent with this hypothesis. Dipole 3 of our model offers a satisfactory explanation of the scalp distribution of the central P22 potential. Although other studies on giant SEPs hypothesise a common generator for both P22 and N30 responses (Furlong et al., 1993; Ikeda et al., 1995), we failed to achieve a satisfactory RV using a single dipole with two opposite peaks of activity. Conversely, RV decreased in the grand average model when another radial perirolandic dipole was added to the model. The activity of this last dipolar source is similar in latency to the N30 potential and, therefore, dipole 4 probably represents the N30 cortical source. In individual models, we tried also to exclude dipole no. 4 and observed that dipole no. 3 assumed a second peak of activity at the latency of the N30 and RV increased significantly (paired t test, P , 0.001). A low RV is not obviously a sufficient criterion for a proper model result, nevertheless a RV decrement obtained by adding a further dipole, which does not interfere with the others and assumes a physiologic location, has to be considered as a valid result in the philosophy of BESA. Two previous studies employing BESA proposed a 3-dipole model, where a single radial dipole explained both the P22 and N30 components. In the former (Buchner et al., 1995), the authors

Fig. 2. Right median nerve SEPs in patient no. 2. The left part of the figure shows traces recorded by the F3, Fz, C3, Cz and P3 electrodes. Negativity is upward. On the right, the figure shows frozen maps calculated at different latencies corresponding to the Pp/Nf and Np/Pf responses. The N20 component is well identifiable in the P3 trace. Later, a high positivity (Pp) at the parietal location is almost coincident in latency with the giant negative potential labelled in the F3 trace (Nf). The corresponding map shows the dipolar distribution of these responses. Lastly, a parietal negative component (Np) and a frontal positive potential (Pf) are indicated in the P3 and F3 traces, respectively; their distribution, shown in the corresponding map, is similar to that of the Pp/Nf components, though inverted in polarity. Fig. 3. Right median nerve SEPs in patient no. 3. The same presentation as in Fig. 2 on the left; on the right, frozen maps calculated at the same latencies as the Pc, Pp/Nf, and Nc-Np/Pf responses are shown. The N20 component is well identifiable in the P3 trace. Later, the P3 trace shows a positive (Pp) and a negative (Np) component with almost the same latency as a frontal negative (Nf) and a frontal positive (Pf) response, respectively. Both Nf and Pf components are labelled in the F3 trace. The C3 trace shows high positivity (Pc), preceding the Pp/Nf responses and followed, in the same trace, by a giant negative potential (Nc) with almost the same latency as the Np/Pf components. The map calculated at a latency corresponding to the Pc response shows a highly focused central positivity, while the map recorded at the same latency as the Pp/Nf potentials shows the dipolar distribution of these components. Lastly, the third map calculated at a latency corresponding to both the Nc and Np/Pf responses shows a scalp potential distribution apparently similar to that of the first map, though inverted in polarity; the dipolar distribution of the Np/Pf potentials is masked at this latency by the giant central negativity.

M. Valeriani et al. / Electroencephalography and clinical Neurophysiology 104 (1997) 122–131

127

128

M. Valeriani et al. / Electroencephalography and clinical Neurophysiology 104 (1997) 122–131

Fig. 4. Four-dipole spatiotemporal solution for the grand average of right median nerve SEPs. The residual variance is 4.7%. On the left, the source potentials of dipoles are shown. On the right, 3 head views illustrate the location and orientation of the dipoles. The top row shows source potentials and location of the dipole at base of the skull (dipole 1). The source potential and location of the tangential perirolandic dipole (dipole 2) are shown in the second row. Note that the activity of this dipole has two peaks of inverted polarity during the analysis time. The third and fourth rows show source potentials and locations of the other perirolandic dipoles (dipoles 3 and 4, respectively).

found good RV values, but referred to a time interval excluding the N30 potential; on the other hand, in the latter (Franssen et al., 1992), where the N30 latency range was included in the dipolar analysis, a number of electrodes lower than ours was used to record scalp SEPs. When we applied our model issued from grand average waveforms to traces recorded in single individuals we obtained higher RVs, thus confirming previous observations (Bromm and Chen, 1995; Valeriani et al., 1996). The interindividual variability is attributable to several factors: head shape, differences in cortical gyri organisation and noise level. However, the fact that we obtained acceptable values in the 9/12 right median nerve SEPs by applying the model and allowing only the dipole orientation to change freely supports the validity of conclusions concerning the locations and activation sequence issued from grand average analysis. In CM patients, our model could well explain the scalp distribution of both normal amplitude and giant potentials, but the time course of the dipoles differed from that observed in healthy subjects. In particular, dipole 2 presented a late third peak of activity at the latency of the Np/Pf responses in all 3 patients, and in patient no. 3 dipole 3 showed a second peak of activity at the latency of the Nc potential. We propose that the Pp/Nf potentials in all 3 patients and the Pc component in patient no. 3 correspond, albeit with a much higher amplitude, to normal P24/N24 and P22 responses, respectively. This conclusion is suggested by the similar scalp distribution and by the fact that, as for the corresponding normal amplitude potentials, Pp/Nf and Pc components are explained by the activities of

dipoles 2 and 3, respectively. In patients nos. 1 and 2, a Pc potential with almost the same latency as and a lower amplitude than the Pp response probably represents the last component picked up by the central electrode contralateral to the stimulation. Dipolar analysis seems also to confirm this hypothesis, since the maximal strength of dipole 3, which probably corresponds to the firing of the neuronal population generating the P22 response, is much smaller than that of dipole 2 and precedes the peak latency of the Pc potential. In all three patients, parietal and frontal traces show two late giant components labelled, according to their polarity and location, as Np and Pf. These responses have scalp distribution that are very similar, though inverted in polarity, to the preceding Pp/Nf potentials and are explained by dipole 2 activation. In patient no. 3, also the central trace contralateral to the stimulation shows a late negative component, that is distributed as the Pc potential and corresponds to the second peak of activity of dipole 3. In patients nos. 1 and 2, the Nc response probably represents the Np potential picked up by the central electrode. No components corresponding to the Np/Pf and Nc potentials can be identified in SEPs recorded from healthy subjects. Therefore, the visual inspection of waveforms fits well to dipolar analysis, both suggesting that Np/Pf and Nc responses are not generated by physiologic SEP sources. Although the pathophysiology of giant SEPs is still far from being clarified, it is largely accepted that they are an expression of hyperexcitability of the cerebral cortex (Rothwell et al., 1984; Shibasaki et al., 1985). In epileptic patients, cerebral cortex hyperexcitability entails the EEG recording of interictal spikes in periods free from seizures. It has been demonstrated that in the genesis of interictal spikes a hyperpolarisation follows the excitatory postsynaptic potentials responsible for the cortical paroxysm (Ayala et al., 1973). We suggest that a similar mechanism may be involved in the building of both interictal spikes and giant SEPs. Seen in this light, the Np/Pc components in all 3 patients and the Nc potential in patient no. 3 can be interpreted as being due to the hyperpolarisation phases following the excitatory postsynaptic potentials responsible for the Pp/Nf and Pc responses, respectively. We may wonder why this late hyperpolarisation does not develop in normal SEPs and why normal amplitude potentials corresponding to Np/ Pf and Nc responses are not identifiable. Although further studies employing intracellular recordings are needed to answer this question definitely, we can presume that the small, or even absent, hyperpolarising afterpotential demonstrated in normal cortical neurones (Spencer, 1977) may explain the different SEP findings in normo- and hyperexcitable cortices. A recent magnetoencephalographic study suggested that ‘the same neural population seems to generate both the giant and the normal SEFs’ (Karhu et al., 1994). Indeed, the authors demonstrated that in UnverrichtLundborg (ULD) disease there is a hyperreactivity of the thalamo-cortical loop, but not of the cortical neurones; moreover, they did not exclude that the sensorimotor cortex

129

M. Valeriani et al. / Electroencephalography and clinical Neurophysiology 104 (1997) 122–131 Table 3 Coordinates of dipoles in normals and patients RVs (%)

GAR HS 1 2 3 4 5 6 7 8 9 10 11 12 Pts. 1 2 3

Dipole 1

Dipole 2 v

J

x

y

z

v

−3

−42

−69

−39

−14

47

−78

50

−69 −69 −69 −69 −69 −69 −69 −69 −69 −69 −69 −69

−3 −3 −3 −3 −3 −3 −3 −3 −3 −3 −3 −3

−49 −43 −30 −27 −37 −64 −68 −49 −31 −37 −86 −27

−72 −7 −44 −3.1 −67 51 54 −70 −8 −7 52 −27

−39 −39 −39 −39 −39 −39 −39 −39 −39 −39 −39 −39

−14 −14 −14 −14 −14 −14 −14 −14 −14 −14 −14 −14

47 47 47 47 47 47 47 47 47 47 47 47

−92 −87 −89 −92 −78 −97 −97 −29 −82 −77 −68 −99

64.2 42 21 66.4 46 59 65.6 −10 40 78 45 42

−69 −69 −59

−3 −3 −12

−10 −42 −6

−37 −69 −85

−39 −39 −29

−14 −14 −10

47 47 42

−86 −78 97

49 50 −83

x

y

z

v

J

x

y

4.4

10

−69

7.1 15.9 6 3.4 7.2 3.6 8.9 13.5 4.9 11.3 9.8 4.4

10 10 10 10 10 10 10 10 10 10 10 10

4.7 4.5 4.8

10 10 10

z

Dipole 3

GAR HS 1 2 3 4 5 6 7 8 9 10 11 12 Pts. 1 2 3

J

Dipole 4 v

J

x

y

z

−39

2

38

74

85

−19

−12

36

−27

48

−39 −39 −39 −39 −39 −39 −39 −39 −39 −39 −39 −39

2 2 2 2 2 2 2 2 2 2 2 2

38 38 38 38 38 38 38 38 38 38 38 38

63 138 87 −36 74.8 66 126 −55 43 44 −70 45

82 34 51 −58 76.5 81 −85 −70 75 89 −83 52

−19 −19 −19 −19 −19 −19 −19 −19 −19 −19 −19 −19

−12 −12 −12 −12 −12 −12 −12 −12 −12 −12 −12 −12

36 36 36 36 36 36 36 36 36 36 36 36

−22 −29 −16 −14 −23 −3 43.1 −18 −52 39 −91 −47

−25 8 84 −20 29.9 84 87.7 20 53.6 −63 35 54

−39 −39 −38

2 2 1

38 38 51

60 −74 −80

80 −55 −38

−19 −19 −17

−12 −12 −19

36 36 40

−27 −27 −152

48 48 89

GAR, grand average of right median nerve SEPs; HS, healthy subjects; Pts., patients.

could be hyperexcitable in other types of myoclonic epilepsy. We did not study patients suffering from ULD and, therefore, we may admit that the pathophysiology of giant SEPs in our cases is different from that in ULD patients. In our CM patients, N30 potential was not recognisable and dipole 4, which we indicated as the N30 generator, showed no activity. The lack of N30 response could be explained in different ways: (1) a ‘real’ N30 absence is possible due to dysfunction of the corresponding source; (2) a normal amplitude N30 may be masked by the giant Pf and Nc responses, occurring almost at the same latency as the N30 component in healthy subjects. Present data do not allow to choose between these hypotheses. In conclusion, our study shows 4 cortical dipoles

which explain the topographic distribution of early scalp SEPs; in particular, 3 perirolandic dipoles would appear to be activated in the latency range of the cortical components following the P14 far-field. BESA findings should not be overinterpreted. BESA represents an attempt to separate cortical sources overlap in time and/or space; nevertheless, multiple solutions are possible depending on the preset constraints on electrical brain events adopted from physiological and anatomical knowledge. BESA has a limited spatial resolution and the assignment of BESA findings to brain morphology is speculative if MRI scans are not coupled with individual BESA results and more accurate head models are not employed. On the other hand, the precise localisation of the scalp SEP sources in the brain is

130

M. Valeriani et al. / Electroencephalography and clinical Neurophysiology 104 (1997) 122–131

ism, similar to that involved in the genesis of interictal spikes.

References

Fig. 5. Four-dipole spatiotemporal solution for right median nerve SEPs recorded from patients nos. 2 (a) and 3 (b). In both (a) and (b) the presentation is the same as in Fig. 4. RVs are of 4.5% and 4.8% for patients nos. 2 and 3, respectively. (a) Note the different calibration of dipole 1, that is early activated. Dipole 2 shows a first small peak of activity followed by large peak of inverted polarity. Lastly, this dipole is activated again with opposite polarity. Dipole 3 shows a peak of activity, earlier in latency and much smaller in strength than the second peak of dipole 2. Dipole 4 has no activity. (b) Note the different calibration of dipole 1, that is early activated. The time course of dipole 2 is very similar to that in (a). Conversely, dipole 3 shows a large peak of activity preceding the second peak of dipole 2 and, later, is reactivated with inverted polarity. Dipole 4 has no activity.

outside the aims of this study and, conversely, we used BESA as mathematical mean to compare SEP findings in healthy and in CM subjects. Seen in this light, the main conclusion from this study is as follows: not all giant SEPs, but only the initial components, correspond to the physiological scalp potentials evoked in healthy subjects. The time course of the dipoles, where late peaks of activity of dipoles 2 and/or 3, absent in normal 4 dipole model, fit well to the scalp distribution of Np/Pf and Nc potentials, confirms our hypothesis. We think that the later giant responses can be explained by a hyperpolarisation mechan-

Allison, T., Goff, W.R., Williamson, P.D. and Van Gilder, J.C. On the neural origin of early components of the human somatosensory evoked potential. In: J.E. Desmedt (Ed.), Progress in Clinical Neurophysiology, Vol 7, Clinical Uses of Cerebral, Brainstem and Spinal Somatosensory Evoked Potentials. Karger, Basel, 1980, pp. 51–68. Ayala, G.F., Dichter, M., Gumnit, R.J., Matsumoto, H. and Spencer, W.A. Genesis of epileptic interictal spikes. New knowledge of cortical feedback systems suggests a neurophysiological explanation of brief paroxysms. Brain Res., 1973, 52: 1–17. Bromm, B. and Chen, A.C.N. Brain electrical source analysis of laser evoked potentials in response to painful trigeminal nerve stimulation. Electroenceph. clin. Neurophysiol., 1995, 95: 14–26. Buchner, H., Adams, L., Mu¨ller, A., Ludwig, I., Knepper, A., Thron, A., Niemann, K. and Scherg, M. Somatotopy of human hand somatosensory cortex revealed by dipole source analysis of early somatosensory evoked potentials and 3D-NMR tomography. Electroenceph. clin. Neurophysiol., 1995, 96: 121–134. Cuffin, B.N., Cohen, D., Yunokuchi, K., Maniewski, R., Purcell, C., Cosgrove, G.R., Ives, J., Kennedy, J. and Schomer, D. Tests of EEG localization accuracy using implanted sources in the human brain. Ann. Neurol., 1991, 29: 132–138. Dawson, G.D. Investigations on a patient subject to myoclonic seizures after sensory stimulation. J. Neurol. Neurosurg. Psychiatry, 1946, 49: 796–807. Deiber, M.P., Giard, M.H. and Mauguie`re, F. Separate generators with distinct orientations for N20 and P22 somatosensory evoked potentials to finger stimulation. Electroenceph. clin. Neurophysiol., 1986, 65: 321–334. Desmedt, J.E., Nguyen, T.H. and Bourguet, M. Bit-mapped color imaging of human evoked potentials with reference to the N20, P22, P27 and N30 somatosensory responses. Electroenceph. clin. Neurophysiol., 1987, 68: 1–19. Ebner, A. and Deuschl, G. Frontal and parietal components of enhanced somatosensory evoked potentials: a comparison between pathological and pharmacologically induced conditions. Electroenceph. clin. Neurophysiol., 1988, 71: 170–179. Franssen, H., Stegeman, D.F., Moleman, J. and Schoobar, R.P. Dipole modelling of median nerve SEPs in normal subjects and patients with small subcortical infarcts. Electroenceph. clin. Neurophysiol., 1992, 84: 40–47. Furlong, P.L., Wimalaratna, S. and Harding, G.F.A. Augmented P22-N31 SEP component in a patient with a unilateral space occupying lesion. Electroenceph. clin. Neurophysiol., 1993, 88: 72–76. Garcia-Larrea, L., Bastuji, H. and Mauguie`re, F. Unmasking of cortical SEP components by changes in stimulus rate: a topographic study. Electroenceph. clin. Neurophysiol., 1992, 84: 71–83. Ikeda, A., Shibasaki, H., Nagamine, T., Xu, X., Terada, K., Mima, T., Kaji, R., Kawai, I., Tatsuoka, Y. and Kimura, J. Peri-rolandic and frontoparietal components of scalp-recorded giant SEPs in cortical myoclonus. Electroenceph. clin. Neurophysiol., 1995, 96: 300–309. Kakigi, R. and Shibasaki, H. Generator mechanism of giant somatosensory evoked potentials in cortical reflex myoclonus. Brain, 1987, 110: 1359– 1373. Karhu, J., Hari, R., Paetau, R., Kajola, M. and Mervaala, E. Cortical reactivity in progressive myoclonus epilepsy. Electroenceph. clin. Neurophysiol., 1994, 90: 93–102. Kimura, J., Mitsudome, A., Yamada, T. and Dickins, Q.S. Stationary peaks from a moving source in far-field recording. Electroenceph. clin. Neurophysiol., 1984, 58: 351–361.

M. Valeriani et al. / Electroencephalography and clinical Neurophysiology 104 (1997) 122–131 Restuccia, D., Di Lazzaro, V., Valeriani, M., Conti, G., Tonali, P. and Mauguie`re, F. Origin and distribution of P13 and P14 far-field potentials after median nerve stimulation. Scalp, nasopharyngeal and neck recording in healthy subjects and in patients with cervical and cervico-medullary lesions. Electroenceph. clin. Neurophysiol., 1995, 96: 371–384. Rothwell, J.C., Obeso, J.A. and Marsden, C.D. On the significance of giant somatosensory evoked potentials in cortical myoclonus. J. Neurol. Neurosurg. Psychiatry, 1984, 47: 33–42. Scherg, M. Fundamentals of dipole source potential analysis. In: F. Grandoni, M. Hoke and G.L. Romani (Eds.), Auditory Evoked Magnetic Fields and Electric Potentials, Advances in Audiology, Vol. 6. Karger, Basel, 1990, pp. 40–69. Scherg, M. and Buchner, H. Somatosensory evoked potentials and magnetic fields: separation of multiple source activities. Physiol. Meas., 1993, 14: A35–A39. Scherg, M., Vajar, J. and Picton, T.W. A source analysis of the human auditory potentials. J. Cognit. Neurosci., 1989, 1: 336–354. Shibasaki, H., Yamashita, Y., Neshige, R., Tobimatsu, S. and Fukui, R. Pathogenesis of giant somatosensory evoked potentials in progressive myoclonic epilepsy. Brain, 1985, 108: 225–240.

131

Shibasaki, H., Nakamura, M., Nishida, S., Kakigi, R. and Ikeda, A. Wave form decomposition of ‘giant SEP’ and its computer model for scalp topography. Electroenceph. clin. Neurophysiol., 1990, 77: 286–294. Simpson, G.V., Scherg, M., Ritter, W. and Vaughan, H.G. Localization and temporal activity functions of brain sources generating the human visual ERP. In: C.H.N. Brunia, A.W.K. Gaillard and A. Kok (Eds.), Psychophysical Brain Research. Tilburg University Press, Tilburg, 1990, pp. 99–105. Sonoo, M., Shimpo, T., Takeda, K., Genba, K., Nakano, I. and Mannen, T. SEPs in two patients with localized lesions of the postcentral gyrus. Electroenceph. clin. Neurophysiol., 1991, 80: 536–546. Spencer, W.A. Physiology of mammalian supraspinal neurons. In: Handbook of Physiology, Section I: The Nervous System, Vol. 1. The American Physiological Society, 1977, pp. 969–1021. Tomberg, C., Desmedt, J.E. and Ozaki, I. Right or left ear reference changes the voltage of frontal and parietal somatosensory evoked potentials. Electroenceph. clin. Neurophysiol., 1991, 80: 504–512. Valeriani, M., Rambaud, L. and Mauguie`re, F. Scalp topography and dipolar source modelling of potentials evoked by CO2 laser stimulation of the hand. Electroenceph. clin. Neurophysiol., 1996, 100: 343–353.