The human cervical and lumbo-sacral evoked electrospinogram. Data from intra-operative spinal cord surface recordings

Electroencephalography and: clinical Neurophysiology, 80 ( 1991 ) 477-489 © 1991 Elsevier Scientific Publishers Ireland, Ltd. 0168-5597/91/$03.50

477

EVOPOT 90215

The human cervical and iumbo-sacrai evoked electrospinogram. Data from intra-operative spinal cord surface recordings D. Jeanmonod

a,b,,,

M. Sindou a and F. Mauguibre b

Departments of a Neurosurgery and b Clinical Neurophysiology, Neurological Hospital, 69003 Lyon (France) (Accepted for publication: 7 May 1991)

Summary We have undertaken the analysis of the human 'evoked electrospinogram' during intra-dural surgical explorations in 20 patients. Averaged spinal cord surface evoked potentials to peripheral nerve electrical stimulation were obtained from various restricted loci on the pial surface of the cervical and lumbo-sacral spinal cord. The brachial plexus P9 potential and its lumbo-sacral counterpart P17 were recorded as ubiquitous initial far-field positivities. The pre-synaptic compound action potentials N l l and N21 dwelt on the ascending slope of N13 and N24 respectively. They were composed of 1-5 sharp peaks and collected from the dorsal and dorso-lateral positions mainly, on the cervical and lumbo-sacral cord respectively. They are thought to be generated in the proximal portion of the dorsal root, the dorsal funiculus and the afferent collaterals to the dorsal horn. Compound action potentials could also be gathered from the surface of the dorsal roots, the cervical N10 and lumbo-sacral N19 potentials. The large cervical N13 and lumbo-sacral N24 waves originate from a dorso-ventral post-synaptic dipole, generated in deep laminae of the dorsal horn during the activation of large diameter afferent fibers. These waves were maximal on the main entry cord segments of the stimulated nerves and fell off on the 1-4 more rostral and caudal segments. The N2 wave is the dorsal component of another post-synaptic dorso-ventral dipole generated in deep laminae of the dorsal horn but activated by medium diameter afferent fibers. The latest event was the N3 wave, also possibly part of a dorso-ventral post-synaptic dipole, and generated by cells in the dorsalmost and deep dorsal horn laminae during the activation of small diameter afferent fibers. The P wave was a prolonged positive deflection which carried the N2 and N3 waves. It is the manifestation of pre-synaptic inhibition on primary afferent fibers. A supra-segmental ascending spinal cord volley was also described, composed of a long succession of sharp and low voltage peaks. Key words: Dipolar generators; Dorsal root entry zone (DREZ); Evoked electrospinogram; Far-field potentials; Microsurgical DREZ-otomy; Near-field potentials; Spinal cord field potentials

Spinal cord field potentials have been the subject of numerous experimental studies since their initial description by Gasser and Graham in 1933. They have been evoked by stimulation of cutaneous nerves (Bernhard and Wid6n 1953; Bernhard et al. 1953; Lindblom and Ottosson 1953a, b; Fernandez de Molina and Gray 1957; Beall et al. 1977), muscle afferents (Eccles et al. 1954; Foreman et al. 1979), cutaneous and muscle afferents (Bernhard 1953; Coombs et al. 1956), or dorsal root fibers (Austin and McCouch 1955). A recent and thorough review has been produced by Yates et al. (1982). The collection of a significant amount of data on the generation of the human spinal cord field potentials

* Supported financially by the 'Fondation pour la Recherche M6dicale', Paris, France.

Correspondence to: Dr. Daniel Jeanmonod, Labor f'tir Funktionelle Neurochirurgie, Neurochirurgische Klinik, Universitiitsspital Ziirich, R~imistrasse 100, 8091 Zurich (Switzerland).

has been performed by means of non-invasive skin or oesophageal computer-averaged recording, particularly using non-cephalic references (Cracco and Cracco 1976; Desmedt and Cheron 1981, 1983; Desmedt 1984; Desmedt and Nguyen 1984; Maugui~,re 1987; Maugui~re et al. 1987). Other groups have studied more directly these potentials by inserting recording macroelectrodes in epidural (Caccia et al. 1976; Shimoji et al. 1977, 1978; Ertekin 1978; Dimitrijevi6 et al. 1980; Maruyama et al. 1982; Jones et al. 1982; Beri6 et al. 1986; Cioni and Meglio 1986), intra-thecal (Ertekin 1976, 1978; Friedman et al. 1984; Jones and Thomas 1985; Nashold et al. 1985; Makachinas et al. 1988), and intra-medullary locations (Campbell and Lipton 1983; Campbell and Miles 1984). Ertekin (1976) has coined the term 'evoked electrospinogram' to define these human spinal cord field potentials, a qualification which we shall retain here and abbreviate 'EESG.' This study was conceived as a detailed analysis of the different components of the human EESG, including understanding of their various generators. To fulfil this purpose, recording conditions close to experimen-

478

tal ones were warranted. A small macroelectrode, positioned under visual control in direct pial contact, was thus used to record the activity in various specific sites on the spinal cord surface. High resolution and large amplitude records of the EESG potentials could thus be collected in the immediate vicinity of their generators. These potentials correlated well with experimental data and with the usual human skin-recorded EESG events. This study was performed during microsurgical DREZ-otomies (MDT), which consists of a microsurgical lesion in the dorsal root entry zone (DREZ). MDT was introduced on the basis of anatomical studies (Sindou 1972; Sindou et al. 1974) of the human DREZ, and its clinical results against chronic pain (Sindou and Daher 1988) and spasticity (Sindou et al. 1986; Sindou and Jeanmonod 1989) have been recently reviewed.

D. J E A N M O N O D E T AL.

Nll

1/1

A

RECORDED ON DORSAL

10 msec

FUNICULUS

B

10 msec Methods

RECORDED The patient population was composed of 20 patients, 10 males and 10 females, with ages between 21 and 74 years. The informed consent of the patients was obtained before operation. They suffered from either chronic neurogenic pain (13) or spasticity (7) and were treated by MDT at the cervical (9) or lumbo-sacral (11) cord levels. In 8 of ther 20 patients, the lesion producing the symptomatology was unilateral, and it was possible to carry out control recording from contralateral cord hemisegments, whose structure, inputs and outputs were thus known to be unaltered by any disease process. Seven patients presented, on the affected side, lesions of the descending pathways to the recorded segments. In 10 others, segments were deprived of their peripheral inputs. In 6 patients, there was clear evidence of more rostral, suprasegrnental dorsal column involvement. In 4 patients, there were various degrees of direct destruction o f some dorsal horn potential-generating structures. These pathologies accounted for the appearance of clearly abnormal waves recorded mainly in subjects with extensive deafferentations and cord lesions, leading at the extreme to a silent record. T h e s e records were excluded from our analysis. The comparison of all suitable records with the control ones demonstrated no significant differences, and thus a great stability of the potential wave forms and peak latencies, which allowed easy recognition of the recorded events. However, we noted large variations of wave amplitude in homologous segments between patients, and this even between homologous control hemisegments. Anesthesia was induced by a unique i.v. flush of a short-lasting barbiturate (thiopental, 3 mg/kg), then maintained by isoflurane and nitrous oxide at as low a concentration as possible, with the addition of a nar-

C

ON SKIN IN13 I I I I

10 msec

Fig. 1. Three records from the same patient. A and B are collected intra-operatively from the dorsal funiculus position in C7, A centered on the fasciculus cuneatus, B being only 2 - 3 m m away f r o m A towards the midline. The detailed recording of N i l obtained in A is lost in B. C shows a pre-operative skin recording on the spinous process of C6 in the awake patient. Note the large amplitude differences between the two pial and the one skin records.

cotic analgesic when needed (fentanyl, 0.05-0.1 mg per dose). These anesthetic conditions, added to the cooling of nervous structures due to their surgical exposure to theater temperatures, caused a reproducible latency shift of the recorded potentials of 1-2 msec, as compared with the values obtained during pre-operative skin somatosensory evoked potential (SSEP) recording (Fig. 1). Short-lasting curare derivatives were administered only at or just after induction, to allow for the later determination of unaltered motor thresholds, and for the intra-operative analysis of root motor function through stimulation. Cord segment determination was reached thanks to radiological and surgical landmarks and to the motor effects of ventral and dorsal root stimulation, systematically performed in all patients.

EESG recordingprocedure Bipolar stimulation with a proximal cathode was applied to the median nerve at the wrist and to the tibial nerve at the ankle or rarely in the popliteal fossa, using subcutaneous stainless steel needle electrodes.

INTRA-OPERATIVE STUDY OF HUMAN EVOKED ELECTROSPINOGRAM •

Monophasic square waves with a duration of 0.2 msec were delivered at a frequency of 6 Hz by a constantcurrent isolated stimulator. The intensities were just above motor threshold, and sometimes at multiple values (2, 2.5, 3 and 4) of this threshold. Higher stimulation frequencies were used on some occasions (8, 11, 16, 20 and 30 Hz). The active recording electrode was a silver ball measuring 7 5 0 / z m - 1 mm long by 5 0 0 - 7 0 0 / z m wide. It

479

was soldered at the tip of a teflon-insulated silver wire. The reference electrode, a subcutaneous stainless steel needle 0.4 mm in diameter, was always implanted in a non-cephalic position (Maugui~re et al. 1987), i.e., the knee for lumbo-sacral cord studies and the shoulder for cervical studies, both contralateral to stimulation. The active recording electrode was placed in contact with the pial surface of the spinal cord and was maintained in position by a wet (saline) small cotton pad.

c

~s

I

l__ Fig. 2. Synthetic diagram displaying one typical EESG sample for each recording position. A cervical cord segment is represented, containing one axon from a cervical dorsal root ganglion cell. Its collateral ends in a lightly shaded domain of the dorsal horn corresponding to its layers IV-VI. Another broken axon travels rostrally from the lumbo-sacral levels in the fasciculus gracilis. A: dorsal funiculus position, showing the succession of the P9, N i l , N13, N2 and P waves. N13 is at its maximal amplitude. B: dorsal root position, recording a dorsal root compound action potential, buried in the depth of the P9 wave. C: dorso-lateral position, displaying maximally the succession of sharp peaks composing the pre-synaptic N11 potential (between the two diverging arrows). D: lateral funiculus position showing two sharp peaks, possibly N11, followed by a small N13 wave. E: ipsilateral ventral funiculus position, with maintenance of P9 and probably N i l but reversal of N13, N2 and P into P13, P2 and N waves. F: contralateral ventral funiculus position. P9 is followed by two sharp peaks, possibly N11, and a small P13 wave. G: distant rostral position, with the supra-segmental ascending spinal cord volley, composed of a long succession of sharp and low voltage peaks. The first arrow points to the beginning of the volley, the second to the first of the two larger amplitude peaks. H: dorsal funiculus position, but more rostral than the main entry segment. The collateral from the axon to the dorsal horn of this level has not been represented. Note a smaller N13 than in A and an evident N11. The horizontal bars all represent 10 msec, and the vertical ones 10/~V except in G where it is 1 IzV.

480

D. JEANMONOD ET AL.

The strict limitations on intra-operative recording times imposed that each recording session was adapted to the given patient situation. Some recording positions took longer to install than others and were thus less often explored. An unavoidable heterogeneity ensued, reflected in the different numbers for the different recording positions. The following positions on the cervico-thoracic and lumbo-sacral spinal cord surface were explored (Fig. 2): (1) a dorsal root position (Fig. 2B), into the adjacent vertebral foramen at the cervical level only; (2) a dorsal funiculus position (Fig. 2A and H); (3) a dorso-lateral funiculus position (Fig. 2C); (4) a lateral position (Fig. 2D); (5) ventral funiculus positions, either ipsilateral (Fig. 2E) or contralateral (Fig. 2F) to stimulation. These ventral positions, the only ones not to be under complete visual control, were reached by sliding the electrode 2 - 3 mm out of sight toward the midline on the ventral aspect of the cord. Close contact of the silver ball with the cord was obtained by changing the position of the cotton pad and controlling the signal quality using repeated recording trials. (6) Dis-

A 'iF

Fig. 3. A: record of the N10 dorsal root near-field potential from the C7 dorsal root surface. B" record of the N19 dorsal root near-field potential from the L5 dorsal root surface. Note a small N24 wave. All horizontal bars are 10 msec, all vertical ones 10 /~V. Ce is for cervical and LS for lumbo-sacral cord levels.

tant suprasegmental positions (Fig. 2G), not recording

TABLE I Number of patients Total number of recordings Cervical recordings Lumbo-sacral recordings Total number of dorsal funiculus recordings Cervical dorsal funiculus recordings Lumbo-sacral dorsal funiculus recordings Dorsal root recordings Cervical dorso-lateral funiculus recordings Lumbo-sacral dorso-lateral funiculus recordings Recordings fromjunction of dorso-lateral and ventro-lateral funiculi lpsilateral ventral funiculus recordings Contralateral ventral funiculus recordings Distant suprasegmental recordings Number of identified P9 Number of identified N10 Number of identified N11 Number of identified N13 Number of identified P13 Number of identified cervical N2 Number of identified cervical P2 Number of identified cervical N3 Number of identified cervical P3 Number of identified cervical P Number of identified cervical N Number of identified PI 7 Number of identified N19 Number of identified N21 Number of identified N24 Number of identified P24 Number of identified lumbo-sacral N2 Number of identified lumbo-sacral P2 Number of identified lumbo-sacral N3 Number of identified lumbo-sacral P Number of identified lumbo-sacral N

20 256 140 116 155 77 78 22 9 11 2 12 20 23 95 3 57 85 14 24 5 3 4 15 5 40 10 17 83 10 32 3 6 44 4

segmental activities like all the preceding positions, but only suprasegmental potentials traveling in the spinal cord ascending tracts after stimulation of the tibial nerve. The recording electrode was located at the cervical level on the fasciculus gracilis and dorso-lateral funiculus ipsilateral to stimulation and on the ventral funiculus contralateral to stimulation. The numbers of recordings performed for each of the above-mentioned electrode positions are given in Table I. Six segments were explored in the cervicothoracic area (C4-T1), and 8 segments in the lumbosacral area (L1-$3). Between 20 and 200 sweeps were sampled with a bin width of 137/xsec and averaged on analysis times of 60 msec and 90 msec, respectively for cervical and lumbosacral studies. The filter bandpass was set between 2 Hz and 2 kHz (6 dB/octave). Peaks were labeled from their polarity and peak latency (P9, N10, N l l , and N13 for the cervical EESG, and P17, N19, N21 and N24 for the lumbo-sacral one), according to the nomenclature proposed by Desmedt and Cheron (1981 and 1983), except for the later N2, N3 and P waves, which were labeled according to the experimental designations (Yates et al. 1982). The wave form, low voltage and time overlapping of these latter events prevented indeed any satisfactory peak latency determination. Pre-operative (17 patients) and post-operative (5 patients) skin SSEPs were recorded, for the upper and lower limbs, respectively from Erb's point or popliteal fossa, over the C6 or L1 spinous processes or corresponding portions of the operative scar, and from contralateral parietal cortex or vertex positions (Maugui~re et al. 1987).

INTRA-OPERATIVE STUDY OF HUMAN EVOKED ELECTROSPINOGRAM

481

A review of typical traces for each cervical recording position after median nerve stimulation is displayed in Fig. 2.

at the lumbo-sacral level (Fig. 3B), which was usually followed by an N24 wave. This was because of the impossibility, at that level, of sliding the electrode in the appropriate vertebral foramen, far enough to cancel the activity originating in the spinal cord (Fig. 3B).

(A) Dorsal root recording

(B) Dorsal funiculus recording

Twenty-two recordings were performed in this position. The P9 and P17 potentials were identified in 13 records, immediately followed by a sharp, negative and low voltage peak, N10 at the cervical (Fig. 3A) and N19

A highly reproducible succession of sharp and slower events could be shown at this recording site (Figs. 2 and 4), where 155 records were obtained. The first was a quick positive potential, P9 at the cervical and P17 at

Results

I

Ce

N2

L,

N~

I

I

E N,H~

Fig. 4. Traces from the cervical (A-F) and lumbo-sacral (G-J) dorsal funiculus positions in 7 patients. A - F : the cervical records were all taken from the C7 cord segment. They display the presence of one (A and B), two (E) or more (C, D and F) sharp peaks composing the N11 potential. The diverging arrows point to the earliest and latest such phenomena. P9 and N13 are evident in all traces, N2 in 5 of them. G-J: G, H and I are recorded from the 1..5 and J from the S1 cord segments. P17 is often small or inconspicuous. N21 is discreet (G and H) or in the form of a hump (I). In J, 3 sharp N21 events can be distinguished (between the diverging arrows), in spite of some high frequency artefacts. H and I show evident P waves, J a large N2 wave. B and E are recorded from the same patient, C and F from another, G and H from a third. In all 10 traces, the horizontal bar is 10 msec, the vertical one 10 tzV. Ce is for cervical and LS for lumbo-sacral.

482

D. J E A N M O N O D E T AL.

the lumbo-sacral level (Fig. 4). P17, of smaller amplitude than P9 and consequently more often masked by the background noise, was identified in 34% of lumbosacral records, whereas P9 was found in 78% of cervical records, P9 peak latency variations ranged between 0.2 and 1 msec in all our recording cord positions. The following most evident events were the large and slow negative N13 and N24 waves, recorded respectively at cervical and lumbo-sacral levels. On the ascending slope, and sometimes at the top, of these negativities, a succession of small sharp peaks were visible. The sum of these peaks will be considered here to correspond to the cervical N11 and the lumbo-sacral N21 potentials in skin surface records (Desmedt and Cheron 1981, 1983). The N l l potential (Figs. IA and 4A-F) was composed of 1-4, most often 2, peaks (Fig. 4E). The N21 potential (Fig. 4G-J) was characterized by 1-3 sharp peaks. Fig. 1 shows, at the cervical level, how a medial displacement of 2-3 mm on the fasciculus cuneatus

toward the midline can smooth the N l l peaks and even preclude their identification. Fig. 1C shows that, in this patient, the skin records displayed only the presence of N l l as a notch on the ascending phase of N13. Table II presents the values of all accurately determined baseline-to-peak amplitudes and peak latencies of N13 and N24 as recorded longitudinally at different segmental levels in the dorsal funiculus position. We have included in this quantification only traces that had stable baselines, pure wave forms and low background noise levels. They first show constant peak latencies for both potentials, the higher values gathered from the rostral lumbar segments being probably non-significant and due (1) to the errors inherent in peak latency determination on small and slow N24 waves, and (2) to the small number of records at these levels. These data also show similar amplitude curves at cervical and lumbo-sacral levels, these amplitudes being maximal respectively at C7 and C8 for the me-

T A B L E II Cord segmental

Peak latencies of N 1 3 / N 2 4

Amplitudes of N 1 3 / N 2 4

Peak latencies of P 1 3 / P 2 4

A mpl i t ude s of P 1 3 / P 2 4

level

Mean

r

Mean

Mean

n

Mean

C4

14.7

13.8 15.5

2

31.4

C5 C6

13.7 15

C7

15.2

n

4.3

47

2

46.9

-2

15.7

1

3.1

15.7 14.8 16.3 L1

32.6

n

1 1 15.4

18

15.6

r

3

10.4 28.9

16.4 15.1

T1

r

5.2 1 1

15.4

n 2.5

4

14

C8

r

14.8 18

100.2 26.6 67.2 6.4 31.5 2.4 -

15.6

-

27.9

16.3

-

2

34.8

2

3

2

3.8

L2

30.2

L3

25.3

L4

24.1

L5

26.7

S1

26

$2

26

$3

25.3

1

24.1 26.5 23.9 24.3 26.1 29.2 25.4 26.5 24.7 26.6 24.3 26.6

1

6

5 5

14.5

2

21.6

13

40.4

10

41.4

4

24.8

3

5.9

39 20.4

6

26.5

1

6.7

1

3 22.9 13.9 75.5 21.9 75.7 5.7 42.8 2.1

22.3 13

-

11.5 3

14.4

26.3 11

5

4 10.3

23.7

24.7

-

3

16.8 1

6.8

1

INTRA-OPERATIVE STUDY OF H U M A N EVOKED ELECTROSPINOGRAM

dian nerve and at L5 and 81 for the tibial nerve, and falling off progressively rostrally and caudally away from these dominant segments. A more detailed statistical analysis of these data was made impossible by the large amplitude variations between cases. An increase in stimulation frequency up to 30 Hz evidenced, on 5 occasions, an amplitude loss (17.737.6%) of the N13 and N24 potentials, whereas N l l and N21 remained unaffected. After the N13 and N24 potentials, we noted the presence of a very prolonged and low voltage positive deflection, the P wave, often masked by background noise, illustrated in Fig. 4A, H and I, and more often seen in lumbo-sacral than in cervical records. Increasing the intensity of stimulation to more than twice the motor threshold revealed a second slow negative wave, N2, which followed N13 or N24 and sometimes, especially in cervical records, shouldered it (Fig. 5). This test was performed with success on 9 occasions in 7 patients. The amplitude of

A

Ce

483

A

,B

P3

"

7

J

1

Fig. 6. A: C7 dorsal funiculus record displaying the 3 negative waves N13, N2 and N3. B: C7 ventral funiculus record showing the reversal of N13, N2 and N3 into P13, P2 and P3. C: same as A, but at the L5 level. All horizontal bars are 10 msec, all vertical ones 10/zV. Ce is for cervical and LS for lumbo-sacral cord levels.

this N2 wave was exceptionally more than half that of the N13 or N24 potentials. The N2 wave was seen more often in lumbo-sacral than in cervical records. In a few, rarer (n = 9) samples, and not always in relation to higher stimulation intensities, we obtained a third slow negative, low voltage wave, N3 (Fig. 6).

(C) Dorso-lateral funiculus recording

Fig. 5. A and B: C7 dorsal funiculus record in one patient, A without and B with the N2 wave. C and D: L4 dorsal funiculus record in one patient, D with an intensity 2.6 times higher than C. They show the intensity-dependent appearance of the N2 wave. All horizontal bars are 10 msec, all vertical ones 10 ~V. Ce is for cervical and LS for lumbo~sacral cord levels.

Most of these records (20) were taken from the dorso-lateral funiculus (Fig. 2C), always ipsilateral to nerve stimulation. They revealed a pattern grossly similar to the dorsal funiculus traces (Fig. 7A and B), but with some substantial differences. Firstly, the amplitudes of the N13, N24, N2 and P waves were smaller. Secondly, the number of the successive N l l : o r N21 peaks (with a minimum of 3 and a maximum of 5 peaks) was higher than in the dorsal funiculus position, due to the regular presence of additional peaks at the top of N13 or N24, and the appearance of a small one in the depth of P9 or P17, at latencies of 10-11 msec and 19-21 msec respectively at the cervical and lumbo-sacral levels. A review of all dorsal and dorso-lateral records showed that N l l was composed of 1 sharp peak in 10%, 2 peaks in 49%, 3 peaks in 33%, 4 peaks in 4% and 5 peaks in 4%. N21 was composed of 1 peak in 57%, 2 peaks in 22% and 3 peaks in 21%. N l l and N13 were found in respectively 66% and 99% of the cervical dorsal and dorso-lateral records, while N21

484

D. JEANMONOD ET AL.

and N24 were obtained in respectively 19% and 93% of the lumbo-sacral records.

(D) Lateral recording These records (n = 2), at the junction of dorso-lateral and ventro-lateral funiculi (Fig. 2D), showed an initial sharp peak superimposed on P9, followed by a small N13 (Fig. 7C). On the ascending slope of N13, a well defined peak could be seen, which had grossly half the amplitude of the corresponding dorsal funiculus event.

I

(E) Ventralfuniculus recording Twelve such recordings were performed ipsilateral to the stimulated nerve and yielded traces (Fig. 2E) in which the P9 and P17 potentials were similar to those obtained on the dorsal aspect of the cord but with polarity reversal of the N13, N24, N2 and P waves into their P13, P24, P2 and N counterparts (Jeanmonod et al. 1989b). In 5 of these 12 records (42%), one or two small peaks were noted, which had similar latencies to the dorsal N i l and N21 potentials. Twenty ventral records were obtained, contralateral to the stimulation. C o m p a r e d to ipsilateral responses, they often differed by the presence of a larger P9 (or P17), of a less evident P13 (or P24), and of more prominent sharp peaks at latencies corresponding to these of N i l (Fig. 8) or N21 potentials. Such sharp events were noted in 60% of these contralateral ventral

Ce

Fig. 8. A: contralateral ventral funiculus record on the C7 cord segment, showing two sharp peaks between the P9 and P13 waves, possibly the Nll peaks. B: C4 contralateral ventral funiculus record in another patient showing two similar sharp peaks. All horizontal bars are 10 msec, all vertical ones 10 ~V.

funiculus records. Table II presents the mean values of the P13 and P24 peak latencies and baseline-to-peak amplitudes. The P13 and P24 waves were present in 75% of all our ventral records.

(F) Distant suprasegmental recording Stimulating the tibial nerve and recording from the surface of the spinal cord at the cervical level revealed a long (13-18 msec) succession of very sharp and low voltage peaks, gathered from the dorsal funiculus (Fig. 2G) and dorso-lateral positions. One record on the contralateral ventral funiculus remained silent. A total of 23 such distant suprasegmental records were made.

Discussion

c

Co

Fig. 7. A and B: dorso-lateral position on respectively the C7 and S1 cord segments. Note the 3 (B) or 5 (A) sharp peaks composing the Nil and N21 potentials, on the ascending slopes of the N13 and N24 waves. C: C7 lateral funiculus record showing two peaks, probably corresponding to Nil, followed by a small N13. All horizontal bars are 10 msec, all vertical ones I0 #.V. Ce is for cervical and LS for lumbo-sacral cord levels.

We have collected, in theater conditions, large amplitude and high resolution, reproducible E E S G records, the characteristics of which are specifically related to the position of the electrode on the spinal cord pial surface. This could be achieved in spite of a very artefact laden electrical environment such as can be found in an unshielded, conventional operating theater and in an open surgical ward. Moreover, the p e r f o r m e d microsurgical operation, during which E E S G data were collected, is an open and long-duration procedure most often applied to heavily debilitated patients, which imposed strict limitations on intra-operative recording times. We thus had to adapt each recording session to collect a maximal amount of significant data in a minimal amount of time. The observed large amplitude variations of the recorded potentials find two main probable explanations: first, the variations of contact between electrode and pia,

INTRA-OPERATIVE STUDY OF HUMAN EVOKED ELECTROSPINOGRAM depending on the pressure applied to the electrode to keep it in place and on the amount of cerebro-spinal fluid present; second, on the exact position of the electrode in relation to the orientation of the dorsoventral post-synaptic dipoles. A third, but probably secondary, factor could well be the influence of pathology on the recorded events. In spite of these amplitude variations, the stability of the obtained potential wave forms and peak latencies allowed easy recognition of the different normal recorded events as well as correlations with experimental data.

The P9, NIO, P17 and N19 potentials The P9 and P17 potentials represent a very ubiquitous and constant initial phenomenon of the EESG. There is now ample evidence (Desmedt and Cheron 1981, 1983; Desmedt 1984; Desmedt and Nguyen 1984) that these initial positive events are far-field or stationary potentials (Kimura et al. 1984) generated in the afferent fiber volley when it travels respectively in the proximal portions of the brachial (P9) and lumbo-sacral (P17) plexuses. This origin is well in keeping with the fact that they were obtained with similar latencies and wave forms in all of our recording positions. The fact that the P9 wave is identified much more often than the P17 wave could be due to different spatial relationships, in the cervical and lumbo-sacral areas respectively, between the longitudinal axes of (1) the spinal cord, and (2) the dorsal roots and proximal parts of the plexuses. The dorsal root volley in skin human records is mentioned in the study of Desmedt and Cheron (1981), where the N10 cervical potential was described. These dorsal root potentials have the aspect of low voltage near-field compound action potentials. N10 and N19 peak at 1 and 2 msec after the plexus P9 and P17 respectively. The first components of the N l l and N21 events were found in some cases to peak also at latencies of 10 and 19 msec respectively. This favors the hypothesis that they are the manifestation of a unique generator in the proximal part of the dorsal root. The N l l and N21 potentials The presence of an early fast and sharp event, often called the triphasic spike, has been described repeatedly in the experimental literature (Bernhard 1953; Beall et al. 1977; Yates et al. 1982). The genesis of this presynaptic event as a compound action potential in the primary afferent fibers of the dorsal root has been thoroughly documented, on the basis of its duration, shape,and constancy in polarity and shape (Austin and McCouch 1955), its spatial distribution (Campbell 1945), its insensitivity to high-frequency repetitive stimulation (Bernhard 1953), and its resistance to asphyxia (Austin and McCouch 1955). The last authors have moreover identified two such successive events, the

485

intra-medullary spike generated by the afferent volley in the dorsal columns, and the Nla potential, due to activity in the afferent terminals heading for the dorsal horn. The N l l and N21 near-field negative peaks described in the human literature (Ertekin 1976; Desmedt and Cheron 1981, 1983; Maruyama et al. 1982; Mauguibre et al. 1983; Nashold et al. 1985; Beri6 et al. 1986; Cioni and Meglio 1986) and in this publication correspond with certainty to these animal compound action potentials: (1) they are fast and sharp; (2) for a given transverse spinal cord level, they were best seen in an area including the dorsal column sector and the dorsalmost portion of the lateral funiculus; this area is centered on the DREZ and corresponds exactly with its animal counterpart (Campbell 1945); (3) they were not affected by increasing stimulus frequency up to 30 Hz; (5) after MDT section, interrupting the fine fibers and supposed to spare the larger ones which send their axon collaterals into the dorsal columns, the N l l / N 2 1 potentials are preserved (Jeanmonod et al. 1989a). Fig. 1 illustrates the fact that, when recording from the same cord segment in the same patient, slight changes in the electrode position cause substantial changes in the aspect of the N l l wave. This variability sets limits to the interpretations that can be drawn from one record at one moment in a given position, the critical factor being the exact position of the electrode. The comparison of two cord surface records (Fig. 1A and B) with a skin SSEP (Fig. 1C) illustrates moreover the relatively poor reliability of this latter technique for a detailed analysis of the N l l peaks of the EESG. We cannot find any evident explanation for the higher frequency with which N l l is found in comparison with N21. For such near-field potentials, variations in the orientation of a distant generator, as discussed for P9 and P17, can indeed not be invoked. Moreover, contrary to what happens at the cervical level, the whole medio-lateral extent of the lumbo-sacral dorsal funiculus should provide proper and easy capture of N21. A possible explanation could be given by pathology, as the majority of patients recorded at the lumbosacral level had either a peripheral or a central lesion of the primary afferent fibers. Destruction of ascending dorsal funiculus fibers rostral to the recorded segments could thus cause retrograde degeneration along these fibers and explain the disappearance of one or more components of N21. Skin SSEP recording (Maugui~re et al. 1987) has, however, shown the preservation of the far-field potential P l l after cervico-medullary lesions. Reference to experimental data (Austin and McCouch 1955) suggests that the multiple N l l subpeaks could represent action potentials generated in different segments of peripheral afferent fibers, i.e., first the proximal portion of the dorsal root, second the entry of

486

D. JEANMONOD ET AL.

the afferent volley to the dorsal column, and third its approach to the dorsal horn. Campbell (1945) and Austin and McCouch (1955) recorded experimentally pre-synaptic 'spikes' on the dorsal, lateral and ventral surfaces of the cord, but with minimal amplitudes in the central and ventral recording sites ipsilateral to stimulation. The very progressive decrease of the 'spike' amplitudes when recording away from the DREZ gives some indication that all recorded peaks may be generated in the primary afferent fibers. Our data (Fig. 2) are very comparable to these experimental ones and compatible with this possibility. The interpretation that some of the recorded spikes, on the lateral surface of the spinal cord, might be generated in spino-cerebellar pathways is to be mentioned but is, on neuroanatomical grounds (Grant 1982), probably not to be favored. One might raise moreover the question of the origin of the ventral contralateral peaks in the spino-thalamic tract. However, Campbell and Lipton (1983), when recording inside the contralateral ventral funiculus during cordotomies, did not discover any fast and sharp activity. We similarly report here a recording on the surface of a cervical ventral funiculus during contralateral tibial nerve stimulation, which remained silent. It might thus be argued that the spino-thalamic tract, although large in size, does not generate action potentials synchronized enough to be recorded as a sharp EESG component. This could well be in keeping with the known functional heterogeneity of this tract. It must, however, be mentioned that we have not collected ventral funiculus records with higher stimulus intensities. It is thus still possible that we were not in a condition to record a spino-thalamic event related to the summated activation of the large and small afferent fibers.

The N2 potential This event has been studied by Beall et al. (1977), who showed that it originates from a post-synaptic transverse dorso-ventral dipolar generator, distinct from the N1 generator and located in laminae IV-VI of the dorsal horn of the monkey. These authors suggested that N2 reflects the activation of small group II and large group III peripheral afferent fibers. A sec'ond negative wave in the human EESG, which should be considered as the analogue of the experimental N2 wave, has been described by Ertekin (1976), Shimoji et al. (1977) and Nashold et al. (1985). We have shown repeatedly that an increase of the stimulus intensity causes the appearance of the N2 wave (Fig. 5). This is well in keeping with its origin in medium diameter fibers, as discussed above for primates. We have moreover shown (Figs. 2E and 6) that the dorsal N2 wave has a ventral P2 counterpart, indicating, as in the monkey, a dorso-ventral dipolar generator for this pair.

The N13 and N24 potentials These large and relatively slow events are the hallmark of the EESG. N13 was present in 99% of all our cervical dorsal and dorso-lateral records, N24 in 93% of lumbo-sacral ones. We bring here converging evidence that they are the human analogue of the experimental N1 wave. The human N13/N24 is obtained at a low threshold, which indicates its activation by large primary afferent fibers. Beall et al. (1977) have shown experimentally that N1 is due to post-synaptic neuronal activity in laminae IV and V of the primate dorsal horn, and that its emergence is correlated with the activation of group I and II peripheral afferent fibers. The longitudinal distribution of the N13 (this study) and N24 waves (this study, and Shimoji et al. 1977) is identical to that of the N1 potential (Bernhard 1953; Bernhard et al. 1953; Lindblom and Ottosson 1953a). We have confirmed that the N13/N24 waves are affected in their amplitudes by increasing the frequency

The N3 potential Beall et al. (1977) have experimentally described the N3 wave, shown to be generated post-synaptically in two locations in the dorsal horn, i.e., the dorsalmost dorsal horn grey matter and the laminae IV-VI, both areas distinct from those generating the N1 and N2 waves. N3 was correlated with the activation of group III peripheral afferent fibers. We found in only 5% of all dorsal and dorso-lateral records a negative wave, peaking later than the N2 wave (Fig. 6), the amplitude of which was not clearly related to high stimulation intensities, as observed for N2. The latency of this negativity and its reversal into a positive potential on the ventral surface of the cord, that we observed on 3 occasions, suggest that this event, originating from a dorso-ventrally orientated dipole, is an analogue of the experimental N3 wave. Much more data are needed to confirm and, if correct, refine the definition of this human N3 wave.

of stimulation (Ibafiez et al. 1989), as reported for N1 (Bernhard 1953; Bernhard and Wid6n 1953). And finally, we show here (Fig. 2A and E) and in Jeanmonod et al. (1989b) that N13 and N24 are the dorsal components of two dipoles, N13/P13 and N24/P24 respectively. Our dorso-lateral, lateral and contralateral ventral records, showing smaller amplitude N13/N24 than the dorsal and ipsilateral ventral positions, bring converging evidence for a strict dorso-ventral orientation of these dipoles, as was demonstrated for the experimental N1 (Austin and McCouch 1955; Bea¿• et al. 1977). These data strongly confirm the conclusions of other human studies on corresponding skin-recorded potentials (Desmedt and Cheron 1981, 1983; Maugui~re et al. 1983; Seyal and Gabor 1985; Restuccia and Maugui~re 1991).

INTRA-OPERATIVE STUDY OF HUMAN EVOKED ELECTROSPINOGRAM

Although increasing the stimulation intensity was clearly correlated with the appearance of the N2 wave on 9 occasions, there were other records in which the N2 and N3 waves appeared at what seemed to be a low intensity, as based on the determination of the motor threshold. In some pathological situations, the threshold of these late negative waves may be modified, a late negative event possibly corresponding to N3 having for example been shown to appear at low stimulation intensities after experimental spinal cord lesions (Lindblom and Ottosson 1953b). On the other hand, overestimation of the motor threshold due to marked motor deficits in the territory of the stimulated nerve probably also accounts for the similar thresholds observed for N13/N24, N2 and even N3 waves in some patients. In these conditions, the possible mixture of the described sensory potentials with antidromic motor action potentials cannot be completely ruled out, although we have no evidence for it. The recording position most susceptible to be influenced by such phenomena is the ipsilateral ventral funiculus one, but the excellent matching of the wave forms and latencies in this position with those, of the dorsal funiculus records, in the context of the demonstrated dorsoventral orientation of .the dipoles, makes it unlikely that a motor parasitic slow wave is present. The sharp event recorded ventrally, as discussed above, has good, although by no means definitive, evidence of being of primary afferent origin. The P wave A correlation has been demonstrated in animals between the 'negative dorsal root potential' and the P wave (Yates et al. 1982). Both of them are manifestations of the processes of pre-synaptic inhibition affecting the primary afferent fibers inside the dorsal horn (Eccles 1964; Wall 1964). The exact location of the interneurons that subserve this function remains elusive. A phase reversal of the P wave has, however, been demonstrated during dorso-ventral intra-medullary explorations (Eccles 1964). The P wave was described in human EESG studies (Ertekin 1976; Shimoji et al. 1977; Maruyama et al. 1982; Nashold et al. 1985; Beri6 et al. 1986; Cioni and Meglio 1986). The data presented here and those from Shimoji et al. (1977) firmly support the dorso-ventral orientation of a P / N dipole. This orientation is confirmed by the absence of any sizeable P or N wave in the other positions we have recorded from (Fig. 2). The ascending spinal cord volley Recording at the cervical level during tibial nerve stimulation in 3 patients yielded a succession of irregular, sharp and small events, arising from desynchronized populations of axons. Jones et al. (1982) dis-

487

cussed the possibility for the initial part of this event to reflect the afferent volley in the dorsal spino-cerebellar tract, the following peaks originating in the dorsal funiculus. Campbell and Miles (1984) recorded the fastest event of their afferent cord volley inside the subpial part of the dorso-lateral funiculus, thus presumably originating from the dorsal spino-cerebellar tract. Our two records of the ascending cord volley in the dorso-lateral position are very similar to the more numerous dorsal ones, thus precluding any further comments on this point from our own data. We want to acknowledge the very kind support of Dr. C. Fischer and Dr. M. Magnin, the photographic help of Mr. S. Bello and the secretarial help of Mrs. V. Cucci. The electrodes used here were developed and provided by Dr. C. Fischer, D6partement de Neurophysiologie Clinique, H6pital Neurologique, Lyon, France. We are very grateful to the theater staff for their kindness and patience.

References Austin, G.M. and McCouch, G.P. (1955) Presynaptic component of intermediary cord potential. J. Neurophysiol., 18: 441-451. Beall, J.E., Applebaum, A.E., Foreman, R.D. and Willis, W.D. (1977) Spinal cord potentials evoked by cutaneous afferents in the monkey. J. Neurophysiol., 40: 199-211. Beri6, A., Dimitrijevi6, M.R., Prevec, T.S. and Sherwood, A.M. (1986) Epidurally recorded cervical somatosensory evoked potential in humans. Electroenceph. clin. Neurophysiol., 65: 94-101. Bernhard, C.G. (1953) The spinal cord potentials in leads from the cord dorsum in relation to peripheral source of afferent stimulation. Acta Physiol. Scand., 29 (Suppl. 106): 1-29. Bernhard, C.G. and Wid6n, L. (1953) On the origin of the negative and positive spinal cord potentials evoked by stimulation of low threshold cutaneous fibres. Acta Physiol. Scand., 29'(Suppl. 106): 42-54. Bernhard, C.G., Lindhlom, U.F. and Onosson, J.O. (1953) The longitudinal distribution of the negative cord dorsum potential following stimulation of low threshold cutaneous fibres. Acta Physiol. Scand., 29 (Suppl. 106): 170-179. Caccia, M.R., Ubiali, E. and Andreussi, L. (1976) Spinal evoked responses recorded from the epidural space in normal and diseased humans. J. Neurol. Neurosurg. Psychiat., 39: 962-972. Campbell, B. (1945) The distribution of potential fields within the spinal cord, Anat. Rec., 91: 77-88. Campbell, J.A. and Lipton, S. (1983) Somatosensory evoked potentials recorded from within the anterolateral quadrant of the human spinal cord. In: J.J. Bonica et al. (Eds.), Advances in Pain Research and Therapy, Vol. 5. Raven Press, New York, pp. 193-196. Campbell, J.A. and Miles, J. (1984) Evoked potentials as an aid to lesion making in the dorsal root entry zone. Neurosurgery, 15: 951-952. Cioni, B. and Meglio, M. (1986) Epidural recordings of electrical events produced in the spinal cord by segmental, ascending and descending volleys. Appl. Neurophysiol., 49: 315-326. Coombs, J.S., Curtis, D.R. and Landgren, S. (1956) Spinal cord potentials generated by impulses in muscle and cutaneous afferent fibres. J. Neurophysiol., 19: 452-467. Cracco, R.Q. and Cracco, J.B. (1976) Somatosensory evoked poten.tials in man" far-fie.ld potentials. Electroenceph. clin. Neurophysiol., 41: 460-466.

488 Desmedt, J.E. (1984) Non-invasive analysis of the spinal cord generators activated by somatosensory input in man: near field and far field potentials. Exp. Brain Res., Suppl. 9: 45-62. Desmedt, J.E. and Cheron, G. (1981) Prevertebral (oesophageal) recording of subcortical somatosensory evoked potentials in man: the spinal P13 component and the dual nature of the spinal generators. Electroenceph. olin. Neurophysiol., 52: 257-275. Desmedt, J.E. and Cheron, G. (1983) Spinal and far-field components of the human somatosensory evoked potentials to posterior tibial nerve stimulation analysed with oesophageal derivations and non-cephalic reference recording. Electroenceph din. Neurophysiol., 56: 635-651. Desmedt, J.E. and Nguyen, T.H. (1984) Bit-mapped colour imaging of the potential fields of propagated and segmental subcortical components of somatosensory evoked potentials in man. Electroenceph, clin. Neurophysiol., 58: 481-497. Dimitrijevi~, M.R., Lehmkuhl, L.D., Sedgwick, E.M., Sherwood, A.M. and McKay, W.B. (1980) Characteristics of spinal cord-evoked responses ii~man. Appl. Neurophysiol., 43: 118-127. Eccles, J.C. (1964) Presynaptic inhibition in the spinal cord. In: J.C. Eccles and J. Schadd (Eds.), Progress in Brain Research, Vol. 12. Elsevier, Amsterdam, pp. 65-91. Eccles, J.C., Fatt, P., Landgren, S. and Winsbury, G.J. (1954) Spinal cord potentials generated by volleys in the large muscle afferents. J. Physiol. (Lond.), 125: 590-606. Ertekin, C. (1976) Studies on the human evoked eleetrospinogram. Acta Neurol. Scand., 53: 3-20. Ertekin, C. (1978) Comparison of the human evoked electrospinogram recorded from the intrathecal, epidural and cutaneous levels. Electroenceph. clin. Neurophysioi., 44: 683-690. Fernandez de Molina, A. and Gray, J.A.B. (1957) Activity in the dorsal spinal grey matter after stimulation of cutaneous nerves. J. Physiol. (Lond.), 137: 126-140. Foreman, R.D., Kenshalo, D.R., Schmidt, R.F. and Willis, W.D. (1979) Field potentials and excitation of primate spinothalamic neurones in response to volleys in muscle afferents. J. Physiol. (Lond.), 286: 197-213. Friedman, A.H., Nashold, B.S. and Ovelmen-Levitt, J. (1984) Dorsal root entry zone lesions for the treatment of post-herpetic neuralgia. J. Neurosurg., 60: 1258-1262. Gasser, H.S. and Graham, H.T. (1933) Potentials produced in the spinal cord by stimulation of dorsal roots. Am. J. Physiol., 103: 303-320. Grant, G. (1982) Spinocerebellar connections in the cat with particular emphasis on their cellular origin. Exp. Brain Res., Suppl. 6: 466-473. Ibafiez, V., Deiber, M.P. and Mauguibre, F. (1989) Interference of vibrations with input transmission in dorsal horn and cuneate nucleus in man: a study of somatosensory evoked potentials (SEPs) to electrical stimulation of median nerve and fingers. Exp. Brain Res., 75: 599-610. Jeanmonod, D., Sindou, M. and Maugui~re, F. (1989a) Intra-operative spinal cord evoked potentials during cervical and lumbosacral microsurgical DREZ-tomy (MDT) for chronic pain and spasticity (preliminary data). Acta Neurochir. (Wien), Suppl. 46: 58-61. Jeanmonod, D., Sindou, M. and Maugui~re, F. (1989b) Three transverse dipolar generators in the human cervical and lumbo-sacral dorsal horn. Evidence from direct intra-operative recordings on the spinal cord surface. Electroenceph. clin. Neurophysiol., 74: 236-240. Jones, S.J. and Thomas, D.G.T. (1985) Assessment of long sensory tract conduction in patients undergoing dorsal root entry zone coagulation for pain relief. In: J. Sehramm and S.J. Jones (Eds.), Spinal Cord Monitoring. Springer, Berlin, pp. 266-273. Jones, S.J., Edgar, M.A. and Ransford, A.O. (1982) Sensory nerve conduction in the human spinal cord: epidural recordings made

D. JEANMONOD ET AL. during scoliosis surgery. J. Neurol. Neurosurg. Psychiat., 45: 446451. Kimura, J., Mitsudome, A., Yamada, T. and Dickins, Q.S. (1984) Stationary peaks from a moving source in far-field recording. Electroenceph. olin. Neurophysioi., 58: 351-361. Lindblom, U.F. and Ottosson, J.O. (1953a) Localization of the structure generating the negative cord dorsum potential evoked by stimulation of low threshold cutaneous fibres. Acta Physiol. Scand., 29 (Suppl. 106): 180-190. Lindblom, U.F. and Ottosson, J.O. (1953b) Effects of spinal sections on the spinal cord potentials elicited by stimulation of low threshold cutaneous fibres. Acta Physiol. Scand., 29 (Suppl. 106): 191208. Makachinas, T., Ovelmen-Levitt, J. and Nashold, B.S. (1988) Intraoperative somatosensory evoked potentials. A localizing technique in the DREZ operation. Appl. Neurophysiol., 51: 146-153. Maruyama, Y., Shimoji, K., Shimizu, H., Kuribayashi, H. and Fujioka, H. (1982) Human spinal cord potentials evoked by different sources of stimulation and conduction velocities along the cord. J. Neurophysiol., 48: 1098-1107. Maugui~re, F. (1987) Short-latency somatosensory evoked potentials to upper limb stimulation in lesions of brain-stem, thalamus and cortex. In: R.J. Ellingson, N.M.F. Murray and A.M. Halliday (Eds.), The London Symposia (EEG Suppl. 39). Elsevier, Amsterdam, pp. 302-309. Maugui~re, F., Challet, E. and Brechler, T. (1983) Les potentiels dvoquds somesthdsiques cervicaux chez le sujet normal: analyse des aspects obtenus scion le si~ge de l'dlectrode de rdfdrence. Rev. EEG Neurophysiol., 13: 259-272. Maugui~re, F., Ibafiez, V., Deiber, M.P. and Garcia Larrea, L. (1987) Noncephalic reference recording and spatial mapping of shortlatency SEPs to upper limb stimulation: normal responses and abnormal patterns in patients with nondemyelinating lesions of the CNS. In: C. Barber and T. Blum (Eds.), Evoked Potentials III. Butterworth, London, pp. 40-55. Nashold, B.S., Oveimen-Levitt, J., Sharpe, R. and Higgins, A.C. (1985) Intraoperative evoked potentials recorded in man directly from dorsal roots and spinal cord. J. Neurosurg., 62: 680-693. Restuccia, D. and Maugui~re, F. (1991) The contribution of median nerve SEPs in the functional assessment of the cervical spinal cord in syringomyelia. A study of 24 patients. Brain, 114: 361-379. Seyal, M. and Gabor, A.J. (1985) The human posterior tibial somatosensory evoked potential: synapse dependent and synapse independent spinal components. Electroenceph. clin. Neurophysiol., 62: 323-331. Shimoji, K., Matsuki, M. and Shimizu, H. (1977) Wave-form characteristics and spatial distribution of evoked spinal electrogram in man. J. Neurosurg., 46: 304-313. Shimoji, K., Shimizu, H. and Maruyama, Y. (1978) Origin of somatosensory evoked responses recorded from the cervical skin surface. J. Neurosurg., 48: 980-984. Sindou, M. (1972) Etude de la Jonction Radiculo-MdduUaire Post6rieure. La Radiceilotomie Postdrieure S61ective duns la Chirurgie de la Douleur. Medical Thesis, Lyon. Sindou, M. and Daher, A. (1988) Spinal cord ablation procedures for pain. In: R. Dubner, G.F. Gebhart and M.R. Bond (Eds.), Proceedings of the Fifth World Congress on Pain. Pain Research and Clinical Management, Vol. 3. Elsevier, Amsterdam, pp. 477495. Sindou, M. and Jeanmonod, D. (1989) Microsurgical DREZ-otomy for the treatment of spasticity and pain in the lower limbs. Neurosurgery, 24: 655-670. Sindou, M., Quoex, C. and Baleydier, C. (1974) Fiber organization at the posterior spinal cord-rootlet junction in man. J. Comp. Neurol., 153: 15-26. Sindou, M., Mifsud, J.J., Boisson, D. and GouteUe, A. (1986) Selective posterior rhizotomy in the dorsal root entry zone for treat-

INTRA-OPERATIVE STUDY OF HUMAN EVOKED ELECTROSPINOGRAM ment of hyperspasticity and pain in the hemiplegic upper limb. Neurosurgery, 18: 587-595. Wall, P.D. (1964) Presynaptic control of impulses at the first central synapse in the cutaneous pathway. In: Progress in Brain Research, Vol. 12. Elsevier, Amsterdam, pp. 92-115.

489

Yates, B.J., Thompson, F.J. and Parker Mickle, J. (1982) Origin and properties of spinal cord field potentials. Neurosurgery, 11: 439450.