Somatosensory evoked potentials (SSEPs) to stimulation of the tibial nerve at the knee (TN-K) and ankle (TN-A), and the sural nerve at the ankle (SNA), were recorded from 3 or 4 spinal levels during surgery for scoliosis in 11 neurologically normal subjects. With stimulation of all 3 nerves, the propagation velocity along the spine was nonlinear: it was faster over cauda equina and midthoracic cord than over caudal spinal cord. Over the midthoracic cord, TN-K SSEP propagation was faster than that of TN-A and SN-A SSEPs, whereas over the caudal spinal cord these values were sirnilar on stimulation of all 3 nerves. These data suggest that fast conducting second order afferent fiber systems contribute to spinal cord SSEPs evoked by stimulating both mixed and cutaneous peripheral nerves. Key words: SSEPs tibial sural propagation velocity MUSCLE & NERVE 14~253-258 1991
INTRAOPERATIVE RECORDINGS OF SPINAL SOMATOSENSORY EVOKED POTENTIALS TO TIBIAL NERVE AND SURAL NERVE STIMULATION LUCIANA PELOSI, MD, GIUSEPPE CARUSO, MD, ROGER Q. CRACCO, MD, JOAN B. CRACCO, MD, and PIER0 BALBI, MD
T h e conduction characteristics along the spine of somatosensory evoked potentials (SSEPs) to stimulation of mixed nerves in lower limb have been extensive1 studied in man. 3-5.7-9.1 3- 16,20-22,25,27,28,3 I,SY However, the results have been inconsistent, mainly because of differences in stimulating and recording techniques. These include the use of either tibial or peroneal nerve stimulation at the ankle or the knee and the use of surface or extradural recording techniques at many different levels. Consequently, there is disagreement concerning the generators of these SSEPs. T o date, only one study described the conduction velocity along the spine of SSEPs to stimulation, in man, of a pure cutaneous nerve in the lower lirnb.l3 In this study, SSEPs to independent stimulation of a cutaneous nerve (sural nerve) at the an-
From the Department of Cllnical Neurophysiology. II School of Medicine, University of Naples, Naples, and Fondazione Clinica del Lavoro. Centro Medico di Campoli. Campoli M T.,Italy (Drs. Pelosi, Caruso, and Balbi) and Department of Neurology, State University of New York Health SCIence Center at Brooklyn, Brooklyn, New York (Drs. R.Q. Cracco and J.B. Cracco). Address reprint requests to Professor Giuseppe Caruso, Cattedra di Neurofisiopatologia. Universita' degli Studi di Napoli, I1 Facolta' di Medicina e Chirurgia, Via Pansini' Nuovo Policlinico, 80131 Napoli. Italy. Accepted February 6, 1990 CCC 0148- 639X/91/030253-06 $04.00 0 1991 John Wiley & Sons, Inc
lntraoperative Spinal SSEPs
kle and a mixed nerve (tibial nerve) at the ankle and knee, were recorded at several spinal levels during surgery for scoliosis. T h e aim was to compare the waveform and conduction characteristics of the responses from these nerves. METHODS
Eleven neurologically normal subjects (7 male, 4 female, mean age 14.5 years) were studied during surgery for scoliosis. Informed consent was obtained. In 9 subjects the tibial nerve was stimulated with needle electrodes at the knee over the popliteal fossa (TN-K). In all subjects the tibial nerve (TN-A) and in 5 the sural nerve (SN-A) were stimulated at the ankle with surface disc electrodes. The stimulus rate was 3 to 5 Hz and the duration 0.2 ms. With T N stirnulation, intensity was adjusted to produce twitches of innervated muscles; with SN-A stimulation intensity was regulated to evoke a sensory nerve action potential at the popliteal fossa, where needle electrodes were placed along the course of the nerve. Patients were then curarized. Needle electrodes (Dantec 13L64, Dantec Electronik, Skovlunde, Denmark) were inserted into the ligamentum flavum at 2 adjacent intervertebra1 spaces after the spinal apophysis was removed. Three to 4 bipolar recordings were simultaneously obtained. Recording sites were designated by the position of the caudal electrode of
MUSCLE 8, NERVE
March 1991
253
at 2 different sites, the term propagation velocity (PV) was used for these values because the neural generators of spinal SSEPs may not be identical at different recording locations. PV data are presented for only 6 of 11 subjects, because it was possible to calculate these velocities with stimulation of 1 or more of the 3 nerves over 1 or more of the following segments; cauda equina, caudal cord, or rostral cord, in only 6 subjects.
each bipolar pair. In each subject, the recording sites were the same during stimulation of different nerves. Bipolar recordings may cause response amplitude attenuation and waveform distortion, especially of the segmentary postsynaptic potentials at the level of the lumbar enlargement. However, bipolar recordings should reduce potential error in latency measurement due to the response contamination with volume recorded activity arising at a distance from the recording electrode. This activity would be recorded at approximately the same amplitude by both electrodes of a close bipolar pair and therefore should cancel. Filter bandpass was 10 to 2000 Hz (-3 dB) and analysis time 50 ms (sampling rate 5000 Hz). Two trials of 250 to 512 responses were averaged and superimposed. The response amplitude was measured peakto-peak from the first prominent negative wave to the subsequent positive potential. However, since each of the 2 electrodes of the bipolar derivation are active and may not be recording from identical neural substrates, this represents only a rough estimate of response amplitude variation over the spine. Latency was measured at the onset of the first negative wave or, when the response began with a positive potential, at the peak of the latter. Distances in centimeters were measured between caudally positioned electrodes at each recording site. The overall speed of conduction was then calculated from the most caudal (varying from LIT12 to T10T9) to the most rostral (varying from T10T9 to T4T3) recording site over the spinal cord in each subject and, in 6 subjects, over 1 or more of the following segments: (1) cauda equina between L5L4 and T12T11, (2) caudal cord between LlT12 and T8T7, and (3) midthoracic cord between T8T7 and T4T3 (Table 1). Since the term conduction velocity implies that activity originating in the same fibers is recorded
RESULTS
In all subjects, SN-A SSEPs were lower in amplitude at all spinal levels than either TN-K or TN-A SSEPs, which were similar in amplitude. The mean amplitude values over cauda equina, caudal spinal cord and midthoracic cord for T N - K were: 2.25 2 0.75, 5.16 t 2.2 and 1.54 0.5 pV, respectively, and for TN-A, 2.17 0.9, 6.65 4.1 and 1.4 +- 0.4 pV. By contrast, the equivalent values for SN-A were 0.8 k 0.14, 1.58 1.1, and 0.28 2 0.1 p v . The responses recorded over the cauda equina on stimulation of all 3 nerves consisted of triphasic potentials (positive-negative- positive) in which the negativity was dominant. With T N - K stimulation, the negative wave sometimes had two peaks or inflections (Fig. 1, Ae). Numerous small inflections were superimposed on all 3 of its phases in 2 of 4 subjects with TN-A (Fig. 1, Bi) and in 1 of 2 subjects with SN-A stimulation (Fig. 1, Cm). The responses recorded over the caudal spinal cord from all 3 nerves consisted of a short duration positive- negative diphasic potential, followed by longer negative (N2) and positive (P2) waves (Fig. 1, Ac, Bh, and Cl). On the broad negative wave 3 peaks were consistently superimposed with TN-K stimulation (Fig. 1, Ac), but not always with TN-A or SN-A stimulation (Fig. 1B and C). At midthoracic sites, T N - K and TN-A SSEPs consisted of small, initially positive polyphasic po-
*
*
*
*
Table l . Segmental propagation velocities (m/s) along the cauda equina and spinal cord of tibia1 and sural SSEPs from 6 subjects. ~~~~
Cauda equina
Subject no.
Spine segment
Caudal spinal cord
Stimulation of:
TN-K
TN-A
85.7
70.6
SN-A
1
2
L3L,-L,L,
3 5 7 11
L3LZ-TiZTqq L,L3-L,T,,
254
lntraoperative Spinal SSEPs
81.8 60.7
56.3 62.9
51.4
60.7
Spine segment
TqZTqq-T8T, LiTqZ-TgT8 L1T1Z-T8T7 Tj,T1q-Ti,T, Ti,Tj,-T8T, LlTlz-TgT8
Midthoracic spinal cord
Stimulation of:
Spine segment
TN-K
TN-A
SN-A
32.2 33.3
36.8 35.3 43.3 37.0 30.0 27.5
38.2
TaT,-T,T,
36.4
T,T,-T,T,
42.8 40.0
TeT,-T,T,
37.0 31.0
Stimulation of:
TN-K
TN-A
SN-A
108.3
86.7
86.7
82.7
80.8
66.7
66.6
100.0
MUSCLE & NERVE
March 1991
TN-K
\\T5T4
TN - A
SN -A
bl e
1
l
0
10
r
I
20
30
1
1
4OmS 50
10
20
B
A
30
40ms
50
1
o
C
I
I
I
10
20
30
1
4 o m s 50
FIGURE 1. SSEPs to stimulation of (A) TN-K (bottom three traces from subject 6, top two traces from subject l ) , (6)TN-A (in subject 11) and (C) SN-A (in subject 11). In this Figure and in Figure 2, position of caudal electrodes of the bipolar pair is reported for each trace, When the lower electrode was positioned over the first lumbar vertebrae and the higher over caudal spinal cord segment (Ad), the broad negative (N2) and the positive (P2) potential of the caudal cord response were recorded with reversed direction because of the greater contribution from the grid 2 electrode placed over caudal spinal cord."
tentials (Fig. 1, Aa, Ab, Bf, and Bg, and Fig. 2A and 2B). With SN-A stimulation the response was similar, but more inflections were superimposed on, and followed the response (Fig. 1, Cj and Ck, and Fig. 2C), suggesting greater temporal dispersion of impulse propagation. In subjects whose responses were recorded at 2 levels over the midthoracic cord, both peak and interpeak latencies of the individual polyphasic components were greater in the more rostral lead. Mean values of overall PV from caudal (varying from LIT12 to T10T9) to rostral (varying from T10T9 to T4T3) recording sites over the spinal cord in all subjects were, 43.2 ? 5.4 m/s for 8.2 m/s for TN-A, and 46.4 9.7 TN-K, 42.5 m/s, for SN-A. High standard deviations would be expected because of the wide variation in the levels of the recordings and the short length of spinal segments. Table 1 shows PVs obtained over cauda equina, caudal cord and midthoracic cord. Although the PVs calculated were few and the distances were short, conduction along the spine was always nonlinear with stimulation of all 3 nerves;
*
Intraoperative Spinal SSEPs
*
it was greater over the cauda equina and midthoracic cord than over the caudal cord. However, over the midthoracic cord, these values were similar with TN-A and SN-A stimulation (Table 1: subjects 1, 3, and 7) and greater with T N - K stimulation (Table 1: subjects 1 and 7). Over caudal cord, PVs were similar whichever nerve was stimulated (Table 1). Over the cauda equina, PV values could be as high as 85.7 m/s with T N - K stimulation (Table 1). Such high values were not observed with TN-A and SN-A stimulation.
DISCUSSION
The triphasic potentials recorded over the cauda equina are to be expected when recording a nerve action potential in volume. The greater number of peaks or inflections present when nerves were stimulated at the ankle suggests a greater temporal dispersion of impulses. The initial positivenegative diphasic potential observed over the caudal spinal cord originates presynaptically from the intramedullary continuations of dorsal roots. The
MUSCLE & NERVE
March 1991
255
TN-A
TN-K
SN-A
11
T6T5
1
t 0
I
lo
I
I
I
20
30
40
mr
A
1 50
I 0
I
10
30
20
B
4 0 mr
50
I
0
1
10
I
I
30
20
40
~~
50
C
FIGURE 2. SSEPs to TN-K (A), TN-A (B), and SN-A (C) stimulation in the same subject (subject 7). Note the higher magnification in
(C).
subsequent large broad negative and positive waves reflect segmental postsynaptic activity at the spinal entry zone of the peripheral afferent volley,22'lz2' although primary afferent de olarization also contributes to the positive wave.3! Responses recorded at midthoracic levels were polyphasic in all subjects, regardless of nerve or site stimulated. Both peak and interpeak latencies of their individual components were greater in more rostral leads, suggesting propagation in fiber groups with different conduction velocities, rather than repetitive firing of the same fibers. PV along the spine was nonlinear whether the mixed nerve was stimulated at the ankle or the knee, or whether the cutaneous (sural) nerve was stimulated. It was greater over the cauda equina and midthoracic cord than over the caudal spinal cord. There is a small possibility that the stationary segmental potentials arising at the lumbar enlargement were recorded in the close bipolar leads above and below their level of origin and, therefore, could have been superimposed on the axonal potentials of the ascending volley. This might have caused an error in estimating the latency of the fastest traveling potentials in recordings obtained above the lumbar enlargement (ie, the true onset latency may have been longer). T h e nearer the recording electrode to the lumbar enlargement, the greater the error. Assuming that this is the case, then the true conduction velocity along the caudal cord must be considered even slower than that estimated from our measurements. Results from this study confirm the previous findings obtained with surface-recorded SSEPs to stimulation of mixed nerves in and ex-
256
lntraoperative Spinal SSEPs
tradurally recorded SSEPs to sciatic nerve stimulation in animals.6,10,26 More recently, with multiple epidural recording, Halonen and colleague^'^ also found a delay in conduction at the level of root entry of the first negative wave to stimulation of the tibial nerve at the knee. However, to the authors' knowledge, this finding has not been described for SSEPs to stimulation of a pure cutaneous nerve in man. T h e slowing in speed of conduction over the caudal cord may be accounted for by fiber branching and synaptic delayIg; its acceleration over the rostral spinal cord suggests that large diameter second order afferent fibers contribute to this response. In fact, conduction velocity along first order ascending fibers, such as those of the dorsal columns, does not increase over rostral spinal cord segments. 12,19 Therefore, other pathways, such as the dorsolateral columns, which are known to contain large second order muscle and cutaneous fibers in animals,23724 may well contribute to spinal SSEPs evoked by both mixed and cutaneous nerve stimulation. However, Seyal and Gabor,'* using surfaceelbow reference recordings, found that the ascending volley from stimulation of the tibial nerve at the ankle had a uniform conduction velocity from sacral to cervical levels. This discrepancy between our results and theirs'* may be explained on the basis of differences in recording technique. The small propagated axonal potentials which determine response latency would be expected to be more clearly defined in depth than in surface recordings because of a better signal-to-noise ratio and would not be significantly attenuated in bipo-
MUSCLE & NERVE
March 1991
lar recordings of the type we used. Additionally, the contaminating effects on determining latency of nonpropagated near field potentials recorded in surface distant reference leads32 are minimized in close bipolar depth recordings. Seyal and Gabor,28 in surface recordings performed in man, and Feldman et a1.l’ in dural recordings performed in nonhuman primates, found that spinal responses recorded over the upper thoracic cord had a very short refractory period (
INTRAOPERATIVE RECORDINGS OF SPINAL SOMATOSENSORY EVOKED POTENTIALS TO TIBIAL NERVE AND SURAL NERVE STIMULATION LUCIANA PELOSI, MD, GIUSEPPE CARUSO, MD, ROGER Q. CRACCO, MD, JOAN B. CRACCO, MD, and PIER0 BALBI, MD
T h e conduction characteristics along the spine of somatosensory evoked potentials (SSEPs) to stimulation of mixed nerves in lower limb have been extensive1 studied in man. 3-5.7-9.1 3- 16,20-22,25,27,28,3 I,SY However, the results have been inconsistent, mainly because of differences in stimulating and recording techniques. These include the use of either tibial or peroneal nerve stimulation at the ankle or the knee and the use of surface or extradural recording techniques at many different levels. Consequently, there is disagreement concerning the generators of these SSEPs. T o date, only one study described the conduction velocity along the spine of SSEPs to stimulation, in man, of a pure cutaneous nerve in the lower lirnb.l3 In this study, SSEPs to independent stimulation of a cutaneous nerve (sural nerve) at the an-
From the Department of Cllnical Neurophysiology. II School of Medicine, University of Naples, Naples, and Fondazione Clinica del Lavoro. Centro Medico di Campoli. Campoli M T.,Italy (Drs. Pelosi, Caruso, and Balbi) and Department of Neurology, State University of New York Health SCIence Center at Brooklyn, Brooklyn, New York (Drs. R.Q. Cracco and J.B. Cracco). Address reprint requests to Professor Giuseppe Caruso, Cattedra di Neurofisiopatologia. Universita' degli Studi di Napoli, I1 Facolta' di Medicina e Chirurgia, Via Pansini' Nuovo Policlinico, 80131 Napoli. Italy. Accepted February 6, 1990 CCC 0148- 639X/91/030253-06 $04.00 0 1991 John Wiley & Sons, Inc
lntraoperative Spinal SSEPs
kle and a mixed nerve (tibial nerve) at the ankle and knee, were recorded at several spinal levels during surgery for scoliosis. T h e aim was to compare the waveform and conduction characteristics of the responses from these nerves. METHODS
Eleven neurologically normal subjects (7 male, 4 female, mean age 14.5 years) were studied during surgery for scoliosis. Informed consent was obtained. In 9 subjects the tibial nerve was stimulated with needle electrodes at the knee over the popliteal fossa (TN-K). In all subjects the tibial nerve (TN-A) and in 5 the sural nerve (SN-A) were stimulated at the ankle with surface disc electrodes. The stimulus rate was 3 to 5 Hz and the duration 0.2 ms. With T N stirnulation, intensity was adjusted to produce twitches of innervated muscles; with SN-A stimulation intensity was regulated to evoke a sensory nerve action potential at the popliteal fossa, where needle electrodes were placed along the course of the nerve. Patients were then curarized. Needle electrodes (Dantec 13L64, Dantec Electronik, Skovlunde, Denmark) were inserted into the ligamentum flavum at 2 adjacent intervertebra1 spaces after the spinal apophysis was removed. Three to 4 bipolar recordings were simultaneously obtained. Recording sites were designated by the position of the caudal electrode of
MUSCLE 8, NERVE
March 1991
253
at 2 different sites, the term propagation velocity (PV) was used for these values because the neural generators of spinal SSEPs may not be identical at different recording locations. PV data are presented for only 6 of 11 subjects, because it was possible to calculate these velocities with stimulation of 1 or more of the 3 nerves over 1 or more of the following segments; cauda equina, caudal cord, or rostral cord, in only 6 subjects.
each bipolar pair. In each subject, the recording sites were the same during stimulation of different nerves. Bipolar recordings may cause response amplitude attenuation and waveform distortion, especially of the segmentary postsynaptic potentials at the level of the lumbar enlargement. However, bipolar recordings should reduce potential error in latency measurement due to the response contamination with volume recorded activity arising at a distance from the recording electrode. This activity would be recorded at approximately the same amplitude by both electrodes of a close bipolar pair and therefore should cancel. Filter bandpass was 10 to 2000 Hz (-3 dB) and analysis time 50 ms (sampling rate 5000 Hz). Two trials of 250 to 512 responses were averaged and superimposed. The response amplitude was measured peakto-peak from the first prominent negative wave to the subsequent positive potential. However, since each of the 2 electrodes of the bipolar derivation are active and may not be recording from identical neural substrates, this represents only a rough estimate of response amplitude variation over the spine. Latency was measured at the onset of the first negative wave or, when the response began with a positive potential, at the peak of the latter. Distances in centimeters were measured between caudally positioned electrodes at each recording site. The overall speed of conduction was then calculated from the most caudal (varying from LIT12 to T10T9) to the most rostral (varying from T10T9 to T4T3) recording site over the spinal cord in each subject and, in 6 subjects, over 1 or more of the following segments: (1) cauda equina between L5L4 and T12T11, (2) caudal cord between LlT12 and T8T7, and (3) midthoracic cord between T8T7 and T4T3 (Table 1). Since the term conduction velocity implies that activity originating in the same fibers is recorded
RESULTS
In all subjects, SN-A SSEPs were lower in amplitude at all spinal levels than either TN-K or TN-A SSEPs, which were similar in amplitude. The mean amplitude values over cauda equina, caudal spinal cord and midthoracic cord for T N - K were: 2.25 2 0.75, 5.16 t 2.2 and 1.54 0.5 pV, respectively, and for TN-A, 2.17 0.9, 6.65 4.1 and 1.4 +- 0.4 pV. By contrast, the equivalent values for SN-A were 0.8 k 0.14, 1.58 1.1, and 0.28 2 0.1 p v . The responses recorded over the cauda equina on stimulation of all 3 nerves consisted of triphasic potentials (positive-negative- positive) in which the negativity was dominant. With T N - K stimulation, the negative wave sometimes had two peaks or inflections (Fig. 1, Ae). Numerous small inflections were superimposed on all 3 of its phases in 2 of 4 subjects with TN-A (Fig. 1, Bi) and in 1 of 2 subjects with SN-A stimulation (Fig. 1, Cm). The responses recorded over the caudal spinal cord from all 3 nerves consisted of a short duration positive- negative diphasic potential, followed by longer negative (N2) and positive (P2) waves (Fig. 1, Ac, Bh, and Cl). On the broad negative wave 3 peaks were consistently superimposed with TN-K stimulation (Fig. 1, Ac), but not always with TN-A or SN-A stimulation (Fig. 1B and C). At midthoracic sites, T N - K and TN-A SSEPs consisted of small, initially positive polyphasic po-
*
*
*
*
Table l . Segmental propagation velocities (m/s) along the cauda equina and spinal cord of tibia1 and sural SSEPs from 6 subjects. ~~~~
Cauda equina
Subject no.
Spine segment
Caudal spinal cord
Stimulation of:
TN-K
TN-A
85.7
70.6
SN-A
1
2
L3L,-L,L,
3 5 7 11
L3LZ-TiZTqq L,L3-L,T,,
254
lntraoperative Spinal SSEPs
81.8 60.7
56.3 62.9
51.4
60.7
Spine segment
TqZTqq-T8T, LiTqZ-TgT8 L1T1Z-T8T7 Tj,T1q-Ti,T, Ti,Tj,-T8T, LlTlz-TgT8
Midthoracic spinal cord
Stimulation of:
Spine segment
TN-K
TN-A
SN-A
32.2 33.3
36.8 35.3 43.3 37.0 30.0 27.5
38.2
TaT,-T,T,
36.4
T,T,-T,T,
42.8 40.0
TeT,-T,T,
37.0 31.0
Stimulation of:
TN-K
TN-A
SN-A
108.3
86.7
86.7
82.7
80.8
66.7
66.6
100.0
MUSCLE & NERVE
March 1991
TN-K
\\T5T4
TN - A
SN -A
bl e
1
l
0
10
r
I
20
30
1
1
4OmS 50
10
20
B
A
30
40ms
50
1
o
C
I
I
I
10
20
30
1
4 o m s 50
FIGURE 1. SSEPs to stimulation of (A) TN-K (bottom three traces from subject 6, top two traces from subject l ) , (6)TN-A (in subject 11) and (C) SN-A (in subject 11). In this Figure and in Figure 2, position of caudal electrodes of the bipolar pair is reported for each trace, When the lower electrode was positioned over the first lumbar vertebrae and the higher over caudal spinal cord segment (Ad), the broad negative (N2) and the positive (P2) potential of the caudal cord response were recorded with reversed direction because of the greater contribution from the grid 2 electrode placed over caudal spinal cord."
tentials (Fig. 1, Aa, Ab, Bf, and Bg, and Fig. 2A and 2B). With SN-A stimulation the response was similar, but more inflections were superimposed on, and followed the response (Fig. 1, Cj and Ck, and Fig. 2C), suggesting greater temporal dispersion of impulse propagation. In subjects whose responses were recorded at 2 levels over the midthoracic cord, both peak and interpeak latencies of the individual polyphasic components were greater in the more rostral lead. Mean values of overall PV from caudal (varying from LIT12 to T10T9) to rostral (varying from T10T9 to T4T3) recording sites over the spinal cord in all subjects were, 43.2 ? 5.4 m/s for 8.2 m/s for TN-A, and 46.4 9.7 TN-K, 42.5 m/s, for SN-A. High standard deviations would be expected because of the wide variation in the levels of the recordings and the short length of spinal segments. Table 1 shows PVs obtained over cauda equina, caudal cord and midthoracic cord. Although the PVs calculated were few and the distances were short, conduction along the spine was always nonlinear with stimulation of all 3 nerves;
*
Intraoperative Spinal SSEPs
*
it was greater over the cauda equina and midthoracic cord than over the caudal cord. However, over the midthoracic cord, these values were similar with TN-A and SN-A stimulation (Table 1: subjects 1, 3, and 7) and greater with T N - K stimulation (Table 1: subjects 1 and 7). Over caudal cord, PVs were similar whichever nerve was stimulated (Table 1). Over the cauda equina, PV values could be as high as 85.7 m/s with T N - K stimulation (Table 1). Such high values were not observed with TN-A and SN-A stimulation.
DISCUSSION
The triphasic potentials recorded over the cauda equina are to be expected when recording a nerve action potential in volume. The greater number of peaks or inflections present when nerves were stimulated at the ankle suggests a greater temporal dispersion of impulses. The initial positivenegative diphasic potential observed over the caudal spinal cord originates presynaptically from the intramedullary continuations of dorsal roots. The
MUSCLE & NERVE
March 1991
255
TN-A
TN-K
SN-A
11
T6T5
1
t 0
I
lo
I
I
I
20
30
40
mr
A
1 50
I 0
I
10
30
20
B
4 0 mr
50
I
0
1
10
I
I
30
20
40
~~
50
C
FIGURE 2. SSEPs to TN-K (A), TN-A (B), and SN-A (C) stimulation in the same subject (subject 7). Note the higher magnification in
(C).
subsequent large broad negative and positive waves reflect segmental postsynaptic activity at the spinal entry zone of the peripheral afferent volley,22'lz2' although primary afferent de olarization also contributes to the positive wave.3! Responses recorded at midthoracic levels were polyphasic in all subjects, regardless of nerve or site stimulated. Both peak and interpeak latencies of their individual components were greater in more rostral leads, suggesting propagation in fiber groups with different conduction velocities, rather than repetitive firing of the same fibers. PV along the spine was nonlinear whether the mixed nerve was stimulated at the ankle or the knee, or whether the cutaneous (sural) nerve was stimulated. It was greater over the cauda equina and midthoracic cord than over the caudal spinal cord. There is a small possibility that the stationary segmental potentials arising at the lumbar enlargement were recorded in the close bipolar leads above and below their level of origin and, therefore, could have been superimposed on the axonal potentials of the ascending volley. This might have caused an error in estimating the latency of the fastest traveling potentials in recordings obtained above the lumbar enlargement (ie, the true onset latency may have been longer). T h e nearer the recording electrode to the lumbar enlargement, the greater the error. Assuming that this is the case, then the true conduction velocity along the caudal cord must be considered even slower than that estimated from our measurements. Results from this study confirm the previous findings obtained with surface-recorded SSEPs to stimulation of mixed nerves in and ex-
256
lntraoperative Spinal SSEPs
tradurally recorded SSEPs to sciatic nerve stimulation in animals.6,10,26 More recently, with multiple epidural recording, Halonen and colleague^'^ also found a delay in conduction at the level of root entry of the first negative wave to stimulation of the tibial nerve at the knee. However, to the authors' knowledge, this finding has not been described for SSEPs to stimulation of a pure cutaneous nerve in man. T h e slowing in speed of conduction over the caudal cord may be accounted for by fiber branching and synaptic delayIg; its acceleration over the rostral spinal cord suggests that large diameter second order afferent fibers contribute to this response. In fact, conduction velocity along first order ascending fibers, such as those of the dorsal columns, does not increase over rostral spinal cord segments. 12,19 Therefore, other pathways, such as the dorsolateral columns, which are known to contain large second order muscle and cutaneous fibers in animals,23724 may well contribute to spinal SSEPs evoked by both mixed and cutaneous nerve stimulation. However, Seyal and Gabor,'* using surfaceelbow reference recordings, found that the ascending volley from stimulation of the tibial nerve at the ankle had a uniform conduction velocity from sacral to cervical levels. This discrepancy between our results and theirs'* may be explained on the basis of differences in recording technique. The small propagated axonal potentials which determine response latency would be expected to be more clearly defined in depth than in surface recordings because of a better signal-to-noise ratio and would not be significantly attenuated in bipo-
MUSCLE & NERVE
March 1991
lar recordings of the type we used. Additionally, the contaminating effects on determining latency of nonpropagated near field potentials recorded in surface distant reference leads32 are minimized in close bipolar depth recordings. Seyal and Gabor,28 in surface recordings performed in man, and Feldman et a1.l’ in dural recordings performed in nonhuman primates, found that spinal responses recorded over the upper thoracic cord had a very short refractory period (
