JOURNAL OF NEUROTRAUMAVolume 6, Number 4, 1989Mary Ann Liebert, Inc., Publishers

Spinal Cord Evoked Injury Potentials in Patientswith Acute Spinal Cord Injury

JOHN A. HALTER,1 ISABEL HAFTEK,2MARZENA SARZYNSKA,2and MILAN R. DIMITRIJEVIC «

ABSTRACT

Six patients were examined in the acute stage of spinal cord injury, between 11 h and 12days posttrauma. Quadripolar epidural electrodes were positioned either percutaneouslyusing a Tuohy needle or directly into the epidural space during surgical intervention.These electrodes were combined with a common reference to obtain monopolar recordingsof spinal cord evoked potentials resulting from either median nerve stimulation at thewrist or tibial nerve stimulation at the popliteal fossa. Spinal cord evoked injurypotentials (SCEIPs), stationary potentials with positive polarity on the distal aspect ofthe lesion and negative polarity on the proximal aspect, were recorded in all cases. Theaverage amplitude (n = 3) of the SCEIP resulting from tibial nerve stimulation as

measured across the lesion was 13.5 uY with an average duration of 12.7 msec. Formedian nerve stimulation, the average amplitude (n = 3) of the SCEIP was 16.3 uY withan average duration of 6.7 msec. There was a change in polarity in all cases over a

distance of less than 6 mm, the distance between the electrode contacts on the epiduralelectrode. In one case, recordings were performed initially at 11 h and repeated at 21days posttrauma. In the latter recording, the SCEIP was still present but was five timessmaller in amplitude. Coincidentally, the patient also showed clinical signs of improve-ment in sensory and motor spinal cord function. This study demonstrates the feasibility ofrecording the SCEIP in patients with acute spinal cord injury, describes the features ofthese SCEIPs, discusses their origins, and explores the utility of recording the SCEIP as

an aid in determining the severity of the injury as well as a means of monitoring changesin spinal cord function.

'Division of Restorative Neurology and Human Neurobiology, Baylor College ofMedicine, Houston, Texas.2The Metropolitan Center for Rehabilitation and Medical Institute for Postgraduate Education, Konstancin,

Poland.

231

HALTER ET AL.

INTRODUCTION

Human spinal cord potentials can been recorded from the skin surface over the vertebralbodies (Dimitrijevic et al., 1980) or from the epidural space within the spinal canal,

thereby providing a more direct measure of function of the white matter spinal tracts, seen as

conducting wave potentials, as well as the function of interneurons of the spinal grey matter,seen as stationary (nonpropagating) potentials (Beric et al., 1986; Cioni and Meglio, 1986; Halteret al., 1983; Shimoji et al., 1971). When the spinal cord is traumatized, conduction in the spinaltracts is impaired and a stationary monophasic versus propagating polyphasic waveform can beseen when action potentials approach the lesion site. This is referred to as the spinal cordevoked injury potential (SCEIP) (Schramm et al., 1983b), also called the killed-end (Deecke andTator, 1973; Woodbury, 1965) or final potential (Schramm et al., 1979a). Several animal studieshave demonstrated the SCEIP at the site of an acute spinal cord lesion (Cracco and Evans,1978; Deecke and Tator, 1973; Haghighi et al., 1987; McDonald and Sears, 1970; Sarnowski etal., 1975; Schramm et al., 1979a,b, 1983a,b, 1984). Studies have also shown a correlationbetween the degree of injury and the amplitude of the SCEIP as well as a reduction of theSCEIP amplitude during recovery (Cracco and Evans, 1978; Haghighi et al., 1987; Schramm etal., 1979b).

The SCEIP has been recorded in cases of acute spinal cord injury (SCI) in humans (Katayamaet al, 1988; Tsuyama et al., 1978) with demonstrated clinical utility for recording the SCEIPwith epidural electrodes at and near the site of lesion as an aid in preoperative localization ofthe lesion (Katayama et al., 1988). Epidural electrodes have traditionally been used primarily formonitoring conduction potentials from spinal tracts intraoperatively to avoid damage to the cordduring scoliosis surgery (Jones et al., 1982, 1985) and further damage to the spinal cord in cases

of tumor removal (Baba et al., 1985; Ducker et al., 1985; Macón and Poletti, 1982). Tsubokawa(1987) argues for the safety and efficacy of using epidural electrodes to examine the functionalcondition of the spinal cord in cases of acute SCI.

In this paper, we describe our technique for recording the SCEIP with epidural electrodes,present the results of a series of recordings from patients in the acute stage of SCI, and discussthe characteristics and changes over time of the SCEIP in these patients.

PATIENTS AND METHODS

All patients were admitted to the Metropolitan Center for Rehabilitation in Konstancin,Poland under the care of neurosurgeon Jan Haftek, M.D., Ph.D. Table 1 lists the patients'clinical characteristics. The period between the time of injury and the initial examination rangedfrom 11 h to 12 days. All patients gave informed consent for the procedure.

Table 1. Patients' Clinical Characteristics

Patient SCI lesion_

Time interval, Initial position ofNeurologicexamination

no. Sex Age level Cause Motor Sensory onset-recording epidural electrode

1 M 19 C5 Diving 0 +-

3d 8h C52 M 22 C5 Diving 0 +

-

12d C2v3 M 37 C7 Diving 0 +

-

llh C74 F 37 T9 Fall 0 0 15h T9v5 M 33 C4 Diving 0 +

-

lOd C26 M 19 C5 Diving 0 +

-

14h C5

0, no function; + -, partial function; d, days; h, hours; v, ventral.

232

SPINAL CORD EVOKED INJURY POTENTIALS IN ACUTE SCI

Patients who showed evidence of an incomplete lesion underwent surgical procedures deter-mined by their neurologic and orthopedic condition. Orthopedic stabilization of the spine was

performed when indicated, involving decompression of the spinal cord by facetectomy, foramino-tomy, or vertebrectomy. This was followed by stabilization with bone grafts and, if necessary,internal fixation (wire, spring, hooks) (Weiss, 1977).

The electrodes used in this study were Neuromed 1980JF spinal cord stimulation electrodes,which have a diameter of 1 mm, an overall length of 55 cm, and four platinum-iridium contacts4 mm in length spaced at center-to-center intervals of 1 cm. The electrodes were introduced intothe epidural space either through a percutaneous technique or directly during the surgicalprocedure. When introduced percutaneously, a 16-gauge modified Tuohy spinal needle wasinserted between the vertebral bodies, the electrode was passed through the needle, and theposition in the spinal canal was monitored by fluoroscopy. Care was taken when the electrodetip was approaching the location of the lesion; if any abnormal resistance to the electrodeinsertion was noticed, the electrode was not advanced any further. In this group of six patients,no such difficulty was encountered. The electrodes would normally follow a path along thedorsal aspect of the epidural space. On occasion, the electrode would take a path to the lateralor ventral aspect of the epidural space. After the position of the electrode was verified, theelectrode was fixed to the skin at the point of entry with a small suture. The electrode was

usually removed within 7 days. No cases of infection related to the electrode were observed.When the electrodes were introduced during a surgical procedure, they were guided gently intothe dorsal, ventral, or lateral aspect of the epidural space, as desired.

Evoked potentials were recorded with a four-channel portable system (Biologic Corporationmodel Traveler Two). A monopolar recording technique was used, which connected eachepidural electrode contact to a given channel of the evoked potential system. The evokedpotential system routinely verified that the electrode impedances were low enough to provideaccurate recordings. The impedances present between electrode contacts were also measuredoccasionally, using a Grass EZM5 electrode impedance meter. A Beckman surface referenceelectrode was placed over the spinous process near the level of the epidural electrode wheneverpossible, on the shoulder of the subject in the case of cervical placement and as near the spineas possible for other cases. During surgery a subcutaneous reference was used, typically a clamp.A high-pass filter of 1 Hz and low-pass filter of 5 kHz were used. The acquisition parameterswere typically set to a duration of 40 msec using 512 points per channel, resulting in aneffective acquisition bandwidth of 6.4 kHz per channel.An electrical stimulus was applied to the peripheral nerve at a rate of 3 Hz with a pulse width

of 0.5 msec. The tibial nerve was stimulated at the popliteal fossa with one contact of a bipolarprobe used as the cathode and a 20 cm2 surface carbon-rubber electrode placed on the knee asthe anode. Median or ulnar nerve stimulation was performed by positioning a bipolar probe atthe wrist with the cathode proximal. The stimulus intensity was varied, but most commonly was

adjusted to achieve a large motor twitch of the related muscle group. For the recordingsperformed during surgery, the surface stimulation electrodes were positioned and tested foreffectiveness before induction of anesthesia. The time required to collect the data intraoperative-ly averaged 45 sec per acquisition trial (64-128 sweeps), with a total time of 20-30 mins.After the data were collected, they were transferred to a Hewlett Packard 1000F minicomputer

located in Houston for analysis and display. Conduction velocities at the epidural electrodelocation were determined by measuring the onset time of activity present at the epiduralelectrode contacts relative to the distance between the contacts. The recordings in the followingsection are presented either in the monopolar manner relative to a common reference, in whichcase only the electrode contact number (E0, El, E2, E3) is used, or in a bipolar manner, inwhich case both electrode contacts are listed.

233

HALTER ET AL.

RESULTS

An example of the spinal cord evoked potentials that can be recorded within the epiduralspace from the intact human spinal cord is shown in Figure 1. An example of ascendingcondition waves as recorded from the epidural space at the level of the T8 vertebral body isshown in Figure 1A; note the polyphasic character of the waveforms seen at each electrodecontact. An example of the stationary waves originating from interneurons in the spinal greymatter of the lumbosacral enlargement can be seen in Figure IB; note the similar shape of eachwaveform with a constant latency seen at each electrode contact and a diminishing amplitude asthe distance from the focus of interneuronal activity becomes greater.

Figure 2 presents typical spinal cord evoked injury potentials. This recording was obtainedfrom a patient (case 6) with a C5 SCI, 16 h after trauma. The patient's clinical neurologiccondition was motor paralysis with hypoesthesia present below T6, hypoalgesia below Til,vibration and temperature sensation present. The arrows present in the schematic diagram of thespinal cord in Figure 2 represent this neurologic condition. As seen in Figure 2A, when tibialnerve stimulation was applied, a negative stationary potential was present below the lesion and a

positive one above, in contrast to the multiphasic conduction potentials seen in Figure 1A. Theamplitude of the negative peak found just below the lesion was 7.5 p.V and the positive peak

0.00 16.00 32.00 148.00MILLISECONDS

0.00 16.00 32.00MILLISECONDS

48.00

FIG. 1. A diagram of the recording and stimulation paradigm. The position of the electrode relative to thespinal cord is seen in the transverse and sagittal sections; the electrode contacts are labeled E0, El, E2, andE3. The potentials were recorded using a monopolar paradigm and are presented using an upward defectionfor negative polarities. A. Ascending conduction volleys in the spinal tracts recorded in the thoracic regionresulting from tibial nerve stimulation. B. Activity in the lumbosacral grey matter seen as stationary(nonpropagating) potentials.

234

SPINAL CORD EVOKED INJURY POTENTIALS IN ACUTE SCI

C5

ED

E3

0.00 16.00 32.00 M8.00MILLISECONDS

0.00 16.00 32.00MILLISECONDS

48.00

FIG. 2. Evoked potentials recorded at the spinal cord lesion resulting from (A) tibial nerve stimulation and(B) median nerve stimulation. The patient's (case 6) status was motor complete and sensory incomplete.This is indicated in the schematic illustration of the spinal cord by the dark arrows above the lesionrepresenting intact ascending and descending spinal tract function, the light arrow below the lesionindicating some ascending function passing through the lesion, and no descending arrow below the lesionrepresenting no apparent descending function passing through the lesion. As in Figure 1, a monopolarrecording paradigm was used, and upward defections represent negative polarities. Note the large positivepotential recorded at electrode contact El compared with the negative potential seen at contact E2 incontrast to the evoked potentials present in Figure 1A. This is similar to the evoked injury potentials seenat the focus of the spinal cord injury in experimental animal studies.

just above the lesion was 15 uV. The bipolar SCEIP produced by taking the difference betweenthe potential found below the lesion and that above the lesion had an amplitude of 22.5 |xV anda duration of 14 msec. Across three patients, this potential had an average amplitude of 13.5uV, with an average duration of 12.7 msec. For one case (no. 4) of ventral recording, theamplitude of the bipolar SCEIP was 10 uV and the duration was 8 msec. A similar polarityreversal is present above and below the lesion resulting from median nerve stimulation (Figure2B). The amplitude of the negative peak was 12 uV and the positive peak was 18 uV. The

235

HALTER ET AL.

C5 EOE I

E2F 3

A0 CM

o'. 00 16.00 32.00MILLISECONDS

48.00

1CM-0CM

2CM-1CM3CM-2CM4CM-3CM5CM-4CM6CM-5CM

0.00 16.00 32.00 4ÍMILLISECONDS

00

FIG. 3. Recording of activity resulting from tibial nerve stimulation as the electrode was movedprogressively down from its initial position. Monopolar (A) and (B) equivalent bipolar recording across theelectrode contacts were calculated from the monopolar recording. Note the large potential generated bysubtraction of the recorded traces labeled 0 cm and 1 cm. This illustrates the very localized focus of theinjury. Notice also the recording of conduction waves as the electrode is moved caudal from the injuryzone, revealing the ascending volleys in the spinal tracts with conduction velocities of 62.5 m/sec.

duration of the potential was 6 msec. The average amplitude of the bipolar SCEIP as measuredin three patients was 16.3 pV and the average duration was 6.7 msec. For one case (case 2) ofventral recording, the amplitude of the bipolar SCEIP was 11 pV and the duration was 8 ms.

A measure of the longitudinal distribution of the spinal cord evoked potentials was accom-

plished in case 6 by removing the epidural electrode gradually and repeating the recordings (Fig.3). The SCEIP decreased in amplitude with increasing distance from the lesion. The spatialgradient (change in amplitude with distance) of the SCEIP was 2.03 uV/mm at the lesion site,0.32 pV/mm 1 cm away from the focus and 0.19 pV/mm at a distance of 2 cm from the lesionfocus. Multiphasic conduction waves could be seen below the lesion level, with a conductionvelocity from the leading components of 62.5 m/sec.

236

SPINAL CORD EVOKED INJURY POTENTIALS IN ACUTE SCI

In one patient (case 3) it was possible to perform a follow-up examination 3 weeks after a C7spinal cord injury. The patient had begun to experience pain and spasticity and an epiduralelectrode was again placed percutaneously to evaluate the potential therapeutic effects of spinalcord stimulation, thus presenting an opportunity to repeat the epidural recordings. The firstelectrode had been placed 11 h after injury and was removed 5 days later. The initial neurologiccondition of the patient was motor complete, with no sensation below the T2 level. Theneurologic examination at the time of the second recording (21 days postinjury) showedvibration sensation present, impaired touch below L2, altered pain below T7, and impairedtemperature sensation. At 1 year after injury the patient was ambulatory with a cane.

Figure 4 shows the recorded evoked potentials resulting from tibial nerve stimulation. Notethe similar distribution of the SCEIP with a smaller amplitude than in the previous recording(compare A and B in Fig. 4). The relative amplitude of the bipolar SCEIP, as seen betweenelectrode contacts El and E2, shows a reduction by a factor of approximately 5 (3 uV vs. 14uV). Examination of the relative peak-to-peak amplitude of the conduction waves superimposedon the bipolar SCEIP shows a reduction by a factor of approximately 2 (1.5 uV vs. 3 uV).Comparison of the positive component of the SCEIP seen at electrode contact El with thenegative component seen at electrode contact E2 shows a greater change in amplitude for thepositive component (1.25 uV vs. 6.5 uV) than the negative component (2.5 uV vs. 6.5 uV).

The impedances present at the time of the second recording were 3.1 K£2 as measuredbetween electrode contacts EO and El, 2.5 KQ between El and E2, and 2.5 KQ between E2 andE3. This is comparable to the impedances documented in case 5 at the time of electrodeplacement: 3.5 KQ between EO and El, 2.5 KQ between El and E2, and 2.6 KQ between E2and E3; and 5 days after placement: 3.3 KQ between EO and El, 2.4 KQ between El and E2,and 2.5 KQ between E2 and E3.

DISCUSSION

TechniqueWe did not observe any complications in this study relating to the use of epidural electrodes

in recording the SCEIP from the site of the lesion. We were concerned initially about possibleadditional injury to the spinal cord by passing the electrode up to and past the lesion site but wehad no difficulty in positioning the electrodes in this group of patients. Katayama et al. (1988)also reported no difficulty in positioning epidural electrodes past the lesion site in acute SCIcases. In a few other patients whom we examined but did not include in this study, we did notcontinue the electrode placement procedure when abnormal mechanical resistance was experi-enced or cerebrospinal fluid was detected in the epidural space. Other authors (Schramm, 1985;Tsubokawa, 1987) argue that the use of epidural electrodes is safe and superior to surfacerecording methods in signal quality, reliability, shorter data acquisition times, and less difficultywith the effects of anesthesia.

The recording method was adequate in for obtaining good-quality records both in theoperating room and at the bedside. Intraoperative recordings were subject to contamination byelectrocauterization, which required coordination with the surgical staff. Our use of a monopolarrecording method afforded us the ability to view the SCEIP in the way recommended by Baba etal. (1988b) and allowed us to generate the bipolar representation of the SCEIP from the originalcomputer records. The use of epidural electrodes containing multiple, regularly spaced contactsallowed for rapid examination of a longer region of the spinal cord than is possible with single-contact electrodes. This allowed us to "bracket" the region containing the lesion and examinethe evoked potentials present on the proximal and distal aspects of the lesion simultaneously. Inaddition, the fixed spacing also allowed for easy computation of the velocity of conductingwaves local to the recording site and could provide a means to examine changes in conduction

237

HALTER ET AL.

A

C7H

ED o1.00 16.00 32.00MILLISECONDS

148. 00

EZ

E3B

oo

CO

O

0.00 16.00 32.00 48.00MILLISECONDS

FIG. 4. Evoked injury potentials recorded at the lesion at (A) 11 h and (B) 21 days after the injury. Thepatient's (case 3) status was motor complete and sensory incomplete. The electrode was placed in the dorsalaspect of the spinal canal, and a monopolar recording paradigm was used. The tibial nerves were stimulatedbilaterally, with an intensity of 4 mA for left and 6.2 mA for right in the first case (A) and 12.4 mA for leftand 11 mA for right in the second case (B). The SCEIP is still present in the second recording and isreduced by a factor of 5 in amplitude. The patient also showed clinical signs of improvement in spinal cordfunction over this time interval.

velocity near the lesion site, as studied by Blight (1989). In future studies, it would beadvantageous to have electrode contacts that are shorter than 4 mm. Since the SCEIP was seen

to change polarity in less than 6 mm, the size of the present electrodes is most likely distortingthe true nature of the SCEIP very near the lesion site. Tator et al. (1988) also raise concernabout the spatial resolving ability of the epidural electrodes used to examine the SCEIP.

238

SPINAL CORD EVOKED INJURY POTENTIALS IN ACUTE SCI

Electrode contacts on the order of 1 mm in length, preferably as short as possible, spaced atintervals of 5 mm or less, would solve this problem.

Origin and Character of the SCEIP

Early work on the nature of conduction in nerve performed in the 19th century tookadvantage of the electric field generated by an action potential volley terminating on the crushedend of a peripheral nerve (Hermann, 1879). The biophysics of the potentials seen at and nearthe end of a damaged nerve were explored in detail by Lorente de No (1947). He performed aseries of experiments on the bullfrog sciatic nerve that demonstrated the types of potentials thatcould be recorded at various longitudinal and lateral distances from the damaged end of thenerve. As the end was approached, a demarcation potential could be seen resulting from acontinuous direct current sourced by the damaged nerve in the resting state. When a propagatingaction potential approached the end, the action potential shape changed from a triphasic to amonophasic positive polarity waveform, which he refers to as the "killed-end" potential. He alsodiscusses performing recordings on the nerve preparation after the demarcation potential haddecreased to negligible levels in order to have a stable preparation; this was a period of typically5-8 h after the nerve was crushed. A similar course of decrease in the demarcation potential was

recently reported by Goodman et al. (1985), who examined this potential in rat spinal cordsubjected to weight drop injury. Lorente de No (1947) describes the nature of the killed-endpotential in terms of the volume conductor field potentials generated by the action potentialcurrents that cease as the damaged end of the nerve is approached. This can be thought of interms of a current "sink" that occurs at the point of depolarization as the action potentialpropagates relative to the current "sources" presented by other sections of the nerve or theextracellular space surrounding the nerve. As the action potential reaches a point of termination,the potential seen beyond the end of the nerve fiber appears positive in polarity. It is importantto emphasize that the SCEIPs reported in this paper are believed to be generated by the killed-end potential mechanism as described by Lorente de No and not the mechanism producing thedemarcation potential.

The presence of an SCEIP in the spinal cord of cats subjected to an injection of diphtheriatoxin that resulted in a demyelinating lesion was demonstrated by McDonald and Sears (1970).They draw a distinction between the mechanism of the SCEIP recorded in acutely injured andchronically injured nerve fibers. In the latter case the demyelinated regions of the fibers act aspassive current sources for the current sinks present at the final active node of Ranvier. We donot see a distinction in the mechanism, aside from the possible presence of an intact axolemmalmembrane rather than a ruptured or sealed ending. In the case of an intact axolemmalmembrane that is dysmyelinated, hope may exist for restoration of conduction in these fiberswith application of 4-aminopyridine, as has been demonstrated by Blight (1989) in chronicspinal cord injured cats. Although not examined in his study, an SCEIP would be expected to bepresent and diminish with restoration of conduction in these dysmyelinated fibers.

Other experimental animal studies present the SCEIP secondary to SCI (Baba et al., 1988a;Cracco and Evans, 1978; Deecke and Tator, 1973; Fehlings et al., 1989; Haghighi et al., 1987;Sarnowski et al., 1975; Schramm et al., 1983 a,b, 1984). In these studies, the amplitude of theSCEIP as seen with monopolar recordings ranged from +4 to +140 pV. The recording andstimulation paradigms varied across the studies. Schramm et al. (1983a) report an amplitude forthe SCEIP that is 150% of the preinjury conduction wave amplitude. The duration of the SCEIPas seen in these studies ranged from 1 to 40 msec. Longer-duration SCEIPs resulted fromstimulation at sites that were more distant from the lesion site. This is most likely due to thearrival of the action potential volleys that were more temporally dispersed due to the distribu-tion of conduction velocities in the stimulated fibers.

239

HALTER ET AL.

SCEIP in Humans

The presence of the SCEIP in cases of acute SCI in humans has been reported by Katayamaet al. (1988) and Tsuyama et al. (1978). In the case reported by Tsuyama et al. (1978), a

monopolar recording from the epidural space near the level of a T8-9 SCI is presented resultingfrom epidural spinal cord stimulation of the dorsal columns at the LI level. The resulting SCEIPhas an amplitude of +50 to +75 uV and a duration of 2 msec. The time from onset of theinjury to the recording is not provided; however, the patient is described as being in spinalshock. The report of Katayama et al. (1988) presents an SCEIP recorded from the epidural spaceof a patient with a C6 SCI. A monopolar recording technique was used with epidural spinal cordstimulation at the lower thoracic level. Again, the time from onset of the injury to the recordingis not provided, although the patient is described as having been recently referred to the clinic.The overall amplitude of the SCEIP as seen in the recording at the lesion level is +6 |iV with aduration of 9 msec. In the recording from the C7 level, just below the level of the lesion, a

longer negative wave with an amplitude of 3.5 |o.V is present along with superimposedconduction waves. It is interesting to note in their study that these conduction waves also showevidence of inversion in the C6 recording, rostral to the lesion, indicating conduction block ofthese components. The amplitude of the SCEIPs we recorded, elicited by tibial nerve stimula-tion, is on the order of the +15 uV rostral to and -7.5 jxV caudal to the lesion as seen in Figure2. The +15 uV amplitude of the SCEIP is 550% of the 2.6 uV peak-to-peak amplitude for theconduction waves normally recorded from the intact spinal cord (Jones et al., 1982). The SCEIPsare of a longer duration (average of 12.7 msec) than those reported by Katayama et al. (1988)and Tsuyama et al. (1978) obtained through the use of epidural spinal cord stimulation. Thissupports the principle of the duration of the SCEIP being related to the temporal distribution ofthe action potential volleys arriving at the lesion site.

One aspect of the SCEIP recordings we obtained that appears to be different from most otherreports is the presence of slow negative potentials caudal to the lesion. These waves are

synchronous with the positive potentials recorded more rostral to the lesion and also appear todecrease in amplitude with increasing distance from the lesion site. This is in keeping with thepresence of a dipole located in a rostral-caudal orientation. The bipolar representation of theSCEIP as recorded across the lesion site accentuates this apparent dipole, as seen in Figure 3.Experimental animal studies have not discussed the presence of a slow negative potentialsynchronous with the positive potential seen on the other aspect of the lesion, although one

report by Schramm (1983b) illustrates such a potential without comment. Upon close examina-tion of the recording presented by Katayama et al. (1988), a negative potential is seen below thelesion. The amplitude of this negative potential is 50% of the positive potential seen above thelesion, which correlates with our observations. The negative potential could be due to localvolume conductor effects or activation of structures local to the lesion site that would not havebeen activated before the injury. Should it be generated by interneurons in the dorsal horns, adorsal-ventral orientation of the resulting dipole should be present (Halter et al., 1983;Jeanmonod et al., 1989; Willis, 1980). However, ventral recordings of the SCEIP made in cases2 and 4 in this study also show the same polarity as the dorsal recordings, which implies thatthe potentials are symmetrical with respect to the longitudinal axis. Another interesting feature isthe slight difference in latency of the peak of the positive potential seen just rostral to the lesionsite (Fig. 2A). Closer examination near the lesion site with smaller electrode contacts spaced atcloser intervals could help to resolve the detail of the SCEIP.In the case of epidural recording at the cervical level resulting from stimulation of a peripheral

nerve in the arm, the evoked potentials are a mixture of conduction waves, postsynapticstationary waves from interneurons, and the SCEIP from the lesioned section of the spinal cord(Fig. 2B). This is most likely due to a combination of afferent volleys arriving in variousbranches of the brachial plexus, some of which enter the spinal cord below or in the lesionedsegment and the others that enter the spinal cord rostral to the lesion. The resulting potentials

240

SPINAL CORD EVOKED INJURY POTENTIALS IN ACUTE SCI

reflect the SCEIP in the lower traces and a combination of the SCEIP and postsynaptic negativestationary waves in the upper traces. In the intact spinal cord, these potentials consist of aninitial triphasic wave that reflects the incoming afferent volley, followed by a negative peakreflecting interneuronal activity and then a positive wave from primary afferent depolarization,similar to what is seen in Figure IB (Beric et al., 1986; Willis, 1980).

Change in SCEIP Over Time

Perhaps the most interesting aspect of this study was the opportunity to record the SCEIP at11 h postinjury and then 3 weeks later. The SCEIP was still present in the latter recording andreduced in amplitude despite a larger stimulus intensity applied to the tibial nerves. Thereduction in the conduction waves by a factor of 2 between recordings seems to be in conflictwith the clinical condition of the patient, who was demonstrating recovery of previouslyincomplete sensory function below the lesion level. The patient was also showing evidence ofrecovery of motor function and was ambulatory with assistance of a cane 1 year postinjury.

The possible mechanisms for this change in the SCEIP are degeneration, or 'die-back' ofinjured nerve fibers, recovery of conduction in the previously nonconductive nerve fibers, or

remyelination of damaged nerve fibers (Blight and Young, 1989). In the case of recovery or

remyelination, diminution of the SCEIP would result from restoration of conduction through thelesion site, diminishing the amplitude of the SCEIP while preserving the amplitude of theconduction waves. In the case of degeneration, the fibers would no longer be present to generatethe SCEIP or the conduction waves. Some degeneration of axons is indicated since there was areduction in the peak-to-peak amplitude of the conduction waves arriving at the lesion site. Therelative differences in the amplitudes of the positive components of the scalp seen at electrodecontacts E0 and El and the negative components seen at electrode contacts E2 and E3 betweenthe two recordings could be due to redistribution of the effective current sources and sinks asthe injured fibers either degenerate or recover. Further examination of the biophysical mecha-nisms behind the change in character of the positive versus negative components is warranted.

Regarding the possible influence of technical aspects of the recordings on the results, theelectrode impedances as measured between electrode contacts appeared to be stable withinanother patient (case 5) over a 5 day period as well as comparable to those found in this patient(case 3). In both recordings presented in Figure 4, the time between electrode placement andperformance of the recording was several hours at the most. Therefore, the possible influence ofthe impedance seen at the electrodes relative to the amplitude of the SCEIP as measured at theelectrode contacts appears to be minimal.

Regarding the relative position of the electrode to the source of the SCEIP within the spinalcord, the distance from the electrode contact surface to the surface of the dura may have someinfluence on the recorded amplitude of the SCEIP. The posterior epidural space in themidcervical region is reported to be on the order of 2-6 mm thick, as determined frommeasurements made on sectioned frozen vertebral columns (Shapiro, 1975). Coburn and Sin(1985) present the epidural space as on the order of 1.5 mm thick, as determined frommeasurements on sectioned cadaver vertebral columns in which the subarachnoid space wasfilled with paraffin (Coburn, 1980). Given that the diameter of the electrode is 1 mm, theelectrode could have been at a different distance from the dura in both recordings. However, therelative difference in amplitude of the SCEIP by a factor of 5 and the conduction waves by afactor of 2 between both recordings implies a more complex mechanism than only a greaterdistance of the electrode from the dura in the latter recording. Controlled studies of theinfluence of the electrode impedances and electrode position on the recorded amplitude of theSCEIP are warranted to evolve a technique that can be used as an objective means ofassessment.

Changes in the SCEIP over time have been studied in experimental cases of SCI (Cracco andEvans, 1978; Haghighi et al., 1987; Schramm et al., 1979b). In their work on a SCI model in the

241

HALTER ET AL.

cat, Cracco and Evans (1978) note two cases of neurologic recovery with reduction of theSCEIP. They argue that appearance of the SCEIP does not indicate that the functional block atthe lesion is irreversible, and they continue with the observation that these methods might proveuseful as monitors of changes in the physiological integrity of the spinal cord in humans.Haghighi et al. (1987) show recovery of conduction waves coincident with disappearance of theSCEIP as well as return of neurologic signs of spinal cord function in their studies on cat.Schramm et al. (1979b) studied reversibility in evoked spinal responses in cats subjected tocompression of the spinal cord. They present recordings showing the appearance of the SCEIPon compression and disappearance on decompression of the spinal cord in 5 of the 21 animalssubjected to compression for different durations and in different locations (dorsal vs. ventral).

FUTURE STUDIES

Further study on the SCEIP in human and animals in relation to selective stimulation ofspinal tracts could provide a means to identify functional versus nonfunctional fibers. It hasbeen shown that the fastest components of the conduction waves as recorded in the thoracic orcervical region from caudal stimulation originate in the lateral tracts of the spinal cord and areassumed to reflect activity in the la afférents, whereas the slower components are conducted inthe dorsal columns and are assumed to reflect activity in the cutaneous afférents (Halonen et al.,1989; Tsuyama, et al., 1978). These can be selectively elicited by stimulation of a mixed (motorand sensory) versus a sensory peripheral nerve, such as the tibial nerve at the popliteal fossaversus the ankle. The SCEIP may also alter in character with conduction activity in the differentspinal tracts. Tator and colleagues (1988) warn of the hazards in monitoring spinal cord functionsolely by observation of conduction in the dorsal column, because functional integrity of theventral portion of the spinal cord is not documented. Levy and York (1983) used the evokedpotentials recorded epidurally from the motor tracts in humans as elicited by epidural stimula-tion of the motor tracts of the spinal cord to study the condition of ventral spinal cord function.They continued with this method using transcranial stimulation to generate the descendingaction potential volleys, and reported one case of SCI in which an "injury wave pattern"appeared (Levy et al., 1984). Subsequent animal studies by Levy et al. (1986) producedrecordings of an SCEIP to motor cortex stimulation. Recently, Fehlings et al. (1989) alsodemonstrated a case of the SCEIP produced by motor cortex stimulation in a rat model of SCI.The combination of SCEIP recording and pharmacologie manipulation may also prove to be a

powerful means of exploring the functional condition of the spinal cord. As Blight (1989) hasshown, 4-aminopyridine can improve the conduction ability of dysmyelinated residual nervefibers passing through the spinal cord lesion. Faden (1987) discusses the possible benefits of theadministration of naloxone (Faden et al., 1985) as well as other pharmacologie agents in theacute stage of SCI. The possible changes in spinal cord function resulting from these agents asreflected in the SCEIP have yet to be examined. This method of assessing spinal cord functionmay also prove useful in future studies on the functional efficacy of neural implants andtransplants (Das, 1987). This is an area of rapid growth and development as well as one inwhich much hope rests as a means for restoring spinal cord function in patients who havesustained SCI.

Integration of this method for epidural recording of spinal cord evoked potentials withmultiple modes of stimulation is proving to be a powerful approach to the difficult problem ofassessment of spinal cord function in the acute stages of SCI. The SCEIP as recorded in thisapproach has demonstrated utility in better defining the location of the lesion site (Katayama etal., 1988). We hope that this phenomenon may be exploited to allow us to define better thefunctional condition of the spinal cord, both to enable more accurate prognoses and to guidemethods for restorative intervention to improve those prognoses.

242

SPINAL CORD EVOKED INJURY POTENTIALS IN ACUTE SCI

ACKNOWLEDGMENTS

We are indebted to the physicians and staff of The Metropolitan Center for Rehabilitation andMedical Institute for Postgraduate Education, Konstancin, Poland, for their valuable assistancein our study.

Support for this work was provided by The Vivian L. Smith Foundation for RestorativeNeurology, Houston, Texas.

REFERENCES

BABA, H., SHIMA, I., TOMITA, K., et al. (1985). Clinical usefulness of spina cord evoked potentials. In:Spinal Cord Monitoring. J. Schramm and S.J. Jones (eds.). Springer-Verlag: Berlin, pp. 245-249.

BABA, H., TOMITA, K., UMEDA, S., et al. (1988a). Experimental ascending evoked potentials in spinalcord injury. In: Neurophysiology and Standards of Spinal Cord Monitoring. T.B. Ducker and R.H. Brown(eds.). Springer-Verlag: Berlin, pp. 46-51.

BABA, H., TOMITA, K., UMEDA, S., et al. (1988b). Clinical study of spinal cord evoked potential. In:Neurophysiology and Standards of Spinal Cord Monitoring. T.B. Ducker and R.H. Brown (eds.). Springer-Verlag: Berlin pp. 216-221.

BERIC, A., DIMITRIJEVIC, M.R., PREVEC, T.S., and SHERWOOD, A.M. (1986). Epidurally recordedcervical somatosensory evoked potentials in humans. Electroencephalogr. Clin. Neurophysiol. 65, 94-101.

BLIGHT, A. (1989). Effect of 4-aminopyridine on axonal conduction-block in chronic spinal cord injury.Brain Res. 22, 47-52.

BLIGHT, A.R., and YOUNG, W. (1989). Central axons in injured cat spinal cord recovery electrophysiolog-ical function following remyelination by Schwann cells. J. Neurol. Sei. 91, 15-34.

CIONI, B., and MEGLIO, M. (1986). Epidural recordings of electrical events produced in the spinal cord bysegmental, ascending and descending volleys. Appl. Neurophysiol. 49, 315-326.

COBURN, B. (1980). Electrical stimulation of the spinal cord: Two dimensional finite element analysis withparticular reference to epidural electrodes. Med. Biol. Eng. Comput. 18, 573-584.

COBURN, B., and SIN, W.K. (1985). A theoretical study of epidural electrical stimulation of the spinalcord—part I: Finite element analysis of stimulus fields. IEEE Trans. BME 32, 971-977.

CRACCO, R.Q., and EVANS, B. (1978). Spinal evoked potential in the cat: Effects of asphixia, strychnine,cord section and compression. Electroencephalogr. Clin. Neurophysiol. 44, 187-201.

DAS, G.D. (1987). Neural Transplantation in normal and traumatized spinal cord. Ann. N.Y. Acad. Sei.495, 53-70.

DEECKE, L., and TATOR, C. (1973). Neurophysiological assessment of afferent and efferent conduction inthe injured spinal cord of monkeys. Neurosurgery 39, 65-74.

DIMITRIJEVIC, M.R., LEHMKUHL, L.D., SEDGEWICK, E.M., and SHERWOOD, A.M. (1980). Charac-teristics of spinal cord evoked responses in man. Appl. Neurophys. 43, 118-127.

DUCKER, T.B., THATCHER R.W., CANTOR, D.S., and MCALESTER, R. (1985). Spinal cord monitoringin neurosurgery in the United States. In: Spinal Cord Monitoring. J. Schramm, and S.J. Jones (eds.).Springer-Verlag: Berlin, pp. 245-249.

FADEN, A.I. (1987). Pharmacotherapy in spinal cord injury: A critical review of recent developments. Clin.Neuropharmacol. 10, 193-204.

FADEN, A.I., and JACOBS, T.P. (1985). Experimental spinal cord injury: treatment with naloxone andthyrotropin-releasing hormone. In: Trauma of the Central Nervous System. R.G. Dacey, Jr., et al. (eds.).Raven Press: New York pp. 259-265.

FEHLINGS, M.G., TATOR, C.H., and LINDEN, R.D. (1989). The relationship among the severity ofspinal cord injury, motor and somatosensory evoked potentials and spinal cord blood flow. Electroence-phalogr. Clin. Neurophysiol. 74, 241-259.

243

HALTER ET AL.

GOODMAN, R.M., WACKS, K., KELLER, S., and BLACK, P. (1985). Spontaneous spinal cord "injurypotential" in the rat. Neurosurgery 17, 757-759.

HAGHIGHI, S., CHEHRAZI, B., HIGGINS, R., REMINGTON, W., and WAGNER, C. (1987). Effect oflidocaine treatment on acute spinal cord injury. Neurosurgery 20, 536-541.

HALONEN, J-P., JONES, S.J., EDGAR, M.A., and RANSFORD, A.O. (1989). Conduction properties ofepidurally recorded spinal cord potentials following lower limb stimulation in man. Electroencephalogr.Clin. Neurophysiol. 74, 161-174.

HALTER, J.A., DOLENC, V., DIMITRIJEVIC, M.R., and SHARKEY, P.C. (1983). Neurophysiologicalassessment of electrode placement in the spinal cord. Appl. Neurophys. 46, 124-128.

HERMANN, L. (1879). Allgemeine nervenphysiologie. In: Handbuch der Physiologie. F.C.W. Vogel (ed.).Ister Theil: Leipzig, pp. 1-196.

JEANMONOD, D., SINDOU, M., and MAUGUIERE, F. (1989). Three transverse dipolar generators in thehuman cervical and lumbo-sacral dorsal horn: Evidence from direct intraoperative recording on the spinalcord surface. Electroencephalogr. Clin. Neurophysiol. 74, 236-240.

JONES, S.J., EDGAR, M.A., and RANSFORD, A.O. (1982). Sensory nerve conduction in the human spinalcord: Epidural recordings made during scoliosis surgery. J. Neurol. Neurosurg. Psychiatry 45, 446-451.

JONES, S.J., CARTER, L., EDGAR, M.A., MORLEY, T., RANSFORD, A.O., and WEBB, P.J. (1985).Experience of epidural spinal cord monitoring in 410 cases. In: Spinal Cord Monitoring. J. Schramm andS.J. Jones (eds.). Springer-Verlag: Berlin, pp. 215-220.

KATAYAMA, Y., TSUBOKAWA, T., YAMAMOTO, T., HIRAYAMA, T., and MAEJIMA, S. (1988).Preoperative determination of the level of spinal cord lesions from the killed end potential. Surg. Neurol.29, 91-94.

LEVY, W.J., and YORK, D.H. (1983). Evoked potentials for the motor tracts in humans. Neurosurgery 12,422-429.

LEVY, W.J., YORK, D.H., MCCAFFREY, M., and TANZER, F. (1984). Motor evoked potentials fromtranscranial stimulation of the motor cortex in humans. Neurosurgery 15, 287-302.

LEVY, W., MCCAFFREY, M., and YORK, D. (1986). Motor evoked potentials in cats with acute spinalcord injury. Neurosurgery 19, 9-19.

LORENTE DE NO, R. (1947). A study of nerve physiology. Analysis of the distribution of the actioncurrents of nerve in volume conductors. Stud. Rockfeiler Inst. Med. Res. 132, 384-477.

MCDONALD, W.I., and SEARS, T.A. (1970). The effects of experimental demyelination on conduction inthe central nervous system. Brain 93, 583-598.

MACÓN, J.B., and POLETTI, C.E. (1982). Conducted somatosensory evoked potentials during spinalsurgery. J. Neurosurg. 57, 349-353.

SARNOWSKI, R.J., CRACCO, R.Q., VOGEL, H.B., and MOUNT, F. (1975). Spinal evoked response inthe cat. J. Neurosurg. 43, 329-336.

SCHRAMM, J. (1985). Spinal cord monitoring: Current status and new developments. Cent. Nerv. Syst.Trauma 2, 207-227.

SCHRAMM, J., HASHIZUME, K, FUKUSHIMA, T., and TAKAHASHI, H. (1979a). Experimental spinalcord injury produced by slow, graded compression. (Alterations of cortical and spinal evoked potentials.)Neurosurgery 50, 48-57.

SCHRAMM, J., TAKAHASHI, H, and KRAUSE, R. (1979b). Reversibility of changes in evoked spinalresponse following graded spinal cord compression. Acta Neurochir. Suppl. 28, 599-604.

SCHRAMM, J., KRAUSE, R., SHIGENO, T., and BROCK, M. (1983a). Experimental investigation on thespinal cord evoked injury potential. J. Neurosurg. 59, 485-492.

SCHRAMM, J., SHIGENO, T., and BROCK, M. (1983b). Clinical signs and evoked response alterationsassociated with chronic experimental cord compression. J. Neurosurgery, 58, 734-741.

244

SPINAL CORD EVOKED INJURY POTENTIALS IN ACUTE SCI

SCHRAMM, J., KRAUSE, R., SHIGENO, T., and BROCK, M. (1984). Relevance of spinal cord evokedinjury potential for spinal cord monitoring. In: Fundamentals and Clinical Application of Clinical SpinalCord Monitoring. S. Homma and T. Tamaki (eds.). Saikon Publ. Co: Tokyo, pp. 113-124.

SHAPIRO, R., (1975). Myelography. Year Book Medical Publishers: Chicago, pp. 90-92.

SHIMOJI, K, HIGASHI, H., and KANO, T. (1971). Epidural recording of spinal electrogram in man.

Electroenceph alogr. Clin. Neurophysiol. 30, 236-239.TATOR, C.H., LINDEN, R.D., FEHLINGS, M.G., BENEDICT, CM. and BELL, I. (1988). Overview of

fundamental and clincial aspects of monitoring the spinal cord during spinal cord surgery. In: Neurophysi-ology and Standards of Spinal Cord Monitoring. T.B. Ducker and R.H. Brown (eds.). Springer-Verlag:Berlin, pp. 368-383.

TSUBOKAWA, T. (1987). The clinical value of multimodality spinal cord evoked potentials in theprognosis of spinal cord injury. In: Thoracic and Lumbar Spine and Spinal Cord Injuries. Advances inNeurotraumatology, Vol. 2. R.P. Vigouroux and P.H. Harris (eds.). Springer-Verlag: New York, pp. 65-92.

TSUYAMA, N., TSUZUKI, N., KUROKAWA, T., and IMAI, T. (1978). Clinical application of spinal cordaction potential measurement. Int. Orthop. 2, 39-46.

WEISS, M. (1977). Clinical Results: Early Therapeutic, Social and Vocational Problems in the Rehabiliationof Persons with Spinal Cord Injuries. Plenum Press: New York.

WILLIS, W.D. (1980). Spinal cord potentials. In: The Spinal Cord and Its Reaction to Traumatic Injury.W.F. Windle (ed.). Marcel Dekker: New York, pp. 159-187.

WOODBURY, J.W. (1965). Potentials in a volume conductor. In: Neurophysiology, Ed. 2. T.C. Ruch, H.D.Patton, and J.W. Woodbury (eds.). W.B. Saunders: Philadelphia, pp. 83-91.

Address reprint requests to:Milan R. Dimitrijevic, M.D., D.Sc.

Division of Restorative Neurology andHuman Neurobiology

Baylor College of MedicineOne Baylor Plaza

Houston, TX 77030

245