mGl S
Original papers
9 Springer-Verlag 1992
Child's Nerv Syst (1992) 8:185 189
Electrical stimulation and multichannel EMG recording for identification of functional neural tissue during cauda equina surgery Alan D. Legatt 1,2,5, Charles E. Schroeder 1,2,5, Bhagwant Gill 3'5, and James T. Goodrich 4,s 1 Department of Neurology, z Department of Neuroscience, 3 Department of Urology, and 4 Department of Neurosurgery, Albert Einstein College of Medicine and 5 Montefiore Medical Center, Bronx, New York, USA Received August 5, 1991
Abstract. Electrical stimulation of structures within the surgical field was used to identify functional neural elements during 25 cauda equina operations. E M G responses from anterior thigh, posterior thigh, and anal sphincter muscles were recorded simultaneously using a multichannel signal averager. During nine operations, stimulation of a presumed filum terminale or other tissue produced clear E M G responses, p r o m p t i n g modification of surgical procedures. In one patient, this resulted in preservation of a flattened spinal cord which resembled a band of scar tissue. Some E M G responses were restricted to a single muscle group; these neural structures would p r o b a b l y not have been identified if only a single-channel E M G recording was used. Visual examination alone was not adequate for identifying functional neural elements, or for determining whether atretic-appearing nerve roots were functional. Electrical stimulation with multichannel E M G recording facilitates the preservation of functional neural elements and the optimization of surgical results in cauda equina surgery. Key words: C a u d a equina surgery - Electric stimulation Electromyography - Evoked potentials, somatosensory Intraoperative neurophysiologic monitoring - Tethered spinal cord
Electrical stimulation m a y be used to localize and define neural structures during a variety of neurosurgicaI procedures [9]. F o r example, the facial nerve m a y b e identified by electrical stimulation during surgery for acoustic neuroma; this has been shown to result in improved surgical outcomes [4]. Peripheral nerves and nerve roots are also at risk during surgical procedures on the c a u d a e q u i n a . M a t s o n [12] described the importance of identifying neural elements
Correspondence to."A.D. Legatt, Department of Neurology, Montefiore Medical Center, 111 East 210th Street, Bronx, NY 10467, USA
with electrical stimulation during meningomyelocele repair; grossly visible muscle twitching was the positive result [10]. Stimulation-induced changes in rectal, urethral, and bladder pressures have been used to identify the nerve supply to these structures [5, 6, 14, 17, 18], and anal sphincter E M G has also been used to identify lower spinal nerve roots during surgery for spinal d y s r a p h i s m [7]. We use electrical stimulation with simultaneous E M G recording at multiple sites to identify nerves innervating both limb musculature and anal sphincter during cauda equina surgery, and report herein on our experience in 25 operations.
Patients and methods
Patient population These data are from 25 consecutive operations in the cauda equina region during which electrical stimulation with EMG recording was used. Of the 24 patients (one patient was monitored twice), 14 were female and 10 were male. Their ages ranged from I week to 38 years (mean 13 years). Most of the operations were performed for release of tethered spinal cord in patients with progressive neurologic deficits and radiologically documented tethers; many of these patients also had sacral lipomas. Five operations were for other tumors of the cauda equina or the lower spinal cord (a teratoma, a myxopapillary ependymoma, an astrocytoma, and two neurofibromas). In 15 cases, there was a history of prior lumbar surgery, mostly meningomyelocele repairs within the first few days of life or debulking of sacral lipomas. Four patients had re-tethered after previous surgery for tethered spinal cord, performed 1-11 years earlier.
Electrophysiology For EMG recordings, standard EEG gold cup electrodes were attached to the skin over the quadriceps femoris and hamstring muscles of each leg, in the anterior and posterior midlines and approximately midway between knee and hip. Anal sphincter electrical activity was recorded by sterile platinum-iridium EEG needle electrodes placed in subcutaneous tissue lateral to the anus. All six electrodes were included in the recording montage. EMG signals
186 were amplified (gain 3000 6000, bandpass 5 Hz 3000Hz) and displayed by a Nicolet Pathfinder I signal averager (Nicolet Instrument Corporation, Madison, Wisconsin, USA). Constant-current electrical square pulses (amplitude 1- 3 mA, duration 200 gs, rate 2/s) were passed through a capacitor (to remove any direct current component) and connected to a standard bipolar cautery forceps, which was placed by the surgeons on selected structures within the surgical field. Paralyzing agents were discontinued 15-30 min prior to nerve identification, and the lack of neuromuscular blockade was ascertained prior to stimulation of structures within the surgical field. If the stimulating forceps were not making proper contact with the tissue, the stimulator indicated an open circuit condition; the surgeons were notified and repositioned the forceps. A short circuit between the two blades of the forceps was signaled by the absence of the electrical stimulus artifact from the raw data; the forceps blades were kept slightly separated to prevent this. Somatosensory evoked potentials (SEPs) to unilateral stimulation of each posterior tibial nerve at the ankle were monitored using standard intraoperative SEP recording techniques [9]. Signal averaging was also useful for recognizing small EMG responses. The technical aspects of electrical stimulation and EMG recording will be described in greater detail elsewhere (A. D. Legatt et al., in preparation).
2oo.v:/ 4 ms
A
L
anal s p h i n c t e r R anal s p h i n c t e r
L
anal s p h i n c t e r L a n t e r i o r thigh
I~
L
p o s t e r i o r thigh L a n t e r i o r thigh
~
_
R p o s t e r i o r thigh R anterior thigh
B
C
Results
Nerve identification Clear, reproducible E M G responses in anal sphincter or leg muscles were recorded during 21 operations. In the other four, none of the structures selected for testing by the surgeons gave an E M G response. Stimulation of nerve roots sometimes elicited movements of the legs or buttocks which were visible to, or felt by, the surgons through the surgical drapes. M a n y E M G responses were unaccompanied by palpable or grossly visible muscle movements, however. Responses were often present in multiple channels, either refecting activation of several muscle groups or the presence of gross movements which generated motion artifacts at a number of recording electrodes. In some cases, however, responses were present in only a single channel (Fig. 1 B, C). In one patient, stimulation at one site produced a response in right leg muscles (Fig. 1 C) which was considerably longer in latency than the response in the same channel to stimulation of other neural structures (Fig. 1 B). Electrical stimulation confirmed the visual identification of apparent neural structure during m a n y operations (Fig. 2 A). During surgery for tethered spinal cord where a filum terminale was present, the filum was tested by electrical stimulation prior to sectioning. In most cases, no E M G responses were obtained and the filum was divided. However, E M G responses were elicited by stimulation of what grossly appeared to be the filum in three patients. In two of these, stimulation of an apparent filum gave a large E M G response in the anal sphincter (Fig. 1 E), and the structure was preserved. In the third, stimulation of the rostral filum elicited a large response over left leg muscles, even when the stimulus intensity was reduced to 0.5 m A (Fig. 2 B). The dissection was continued caudally, to a point where stimulation of the filum
D
Fig. 1A-E. Averaged EMG responses (average of 5 epochs) to electrical stimulation of various structures within the surgical field in an 11-year-old girl with a tethered spinal cord. A No response to stimulation at this site. B A large early response in the right posterior thigh and/or right anterior thigh. C A small, longer-latency response in the right leg muscles. D A response in both left anterior thigh and left posterior thigh muscles. E The EMG response to stimulation of a structure which appeared to be the filum terminale, but which was preserved because of the electrophysiologic findings
did not produce a significant E M G response (Fig. 2 C), and the filum was divided at that point. Low-intensity stimulation of the rostral stump still produced a large E M G response (Fig. 2 D). Several other structures beside the filum terminale were found to contain functional neural elements which were not grossly visible, but which were preserved because of the electrical stimulation results. Overall, testing led to alterations of surgical procedures which preserved functional neural tissue in 9 out of the 25 cases. Nerve roots hidden within scar tissue from previous surgery were localized and spared in three patients. In two other cases, stimulation of the remnant o f a sacral lipoma (after most of the tumor had been removed) showed that it
187 A
B
L anal sphincter R anal sphinctel _
~
L anal sphincter L anterior thigh
__
~,
/
~
L f
~__
~._~_ . . . .
C
,
.
v
~
,
~ ,.,
~
L p~176 hL anterior thigh R posterior thigh R anterior thigh Fig. 2A-D. Averaged EMG responses (average
D
_
:
_
.
.
.
.
.
.
,oo< !
contained functional nerve roots, and the resection was stopped. During removal of a large sacral neurofibroma in a patient with central neurofibromatosis, a small neurofibroma was noted on another nerve root. This root was determined to be innervating the anal sphincter, so it was not resected. After the majority of a sacral lipoma had been resected in one patient, a band of fibrous-looking tissue was encountered deep (anterior) to it. Stimulation of this apparent scar tissue elicited E M G responses from the anal sphincter, so it was preserved. On further dissection, it was determined that this tissue was actually a thinned-out ribbon of spinal cord extending anterior to the lipoma. In one patient, stimulation of what appeared to be atretic and nonfunctional nerve roots gave clear E M G responses, and the roots were preserved. During three other operations, structures which appeared to be atretic nerve roots, and which were mechanically tethering the spinal cord, elicited no response when stimulated. They were therefore sectioned in order to maximally untether the spinal cord.
SEP recordings SEPs were monitored during 24 of the 25 operations. Clear cervical and/or cortical SEPs were identifiable in 14 cases. SEP changes related to alterations o f the anesthetic regimen were present in six of these. There were no SEP changes attributable to surgical maneuvers. During ten operations, neither cortical nor cervical SEPs were identifiable at any point, though clear peripheral nerve SEPs were present. Cortical and cervical components had been absent in preoperative SEP recordings in nine of these, and abnormal in the other. In nine of the ten cases, the patient had a history of meningomyelocele;
of 5 epochs) to electrical stimulation of various structures within the surgical field in another 1l-year-old girl with a tethered spinal cord. A higher amplifier gain was used, and some of the waveforms were truncated. A Stimulation of a structure which was clearly a nerve root. B Stimulation of the rostral filum terminale. C Stimulation of the ilium more caudally. D Repeat stimulation of the filum at the rostra1 site (stimulation site of B) after it had been divided at the caudal site (stimulation site of I~)
the other patient had cauda equina scarring from arachnoiditis.
Outcomes Several patients had transient impairment o f bowel or bladder function, or worsening o f preexisting abnormalities, during the first few days to weeks after surgery, most likely due to intraoperative manipulation of the spinal cord. In order to assess outcomes, neurologic findings were evaluated at 4 - 6 weeks after surgery. Most patients were at their neurologic baseline at this time; several others showed improvements in function in the immediate postoperative period. Persistent increased neurologic deficits were noted postoperatively in three cases. Two patients had postoperative sphincter dysfunction, with gradual improvement over several months. In the third patient, a functioning nerve root was deliberately sacrificed to permit total resection of a cauda equina neurofibroma. Postoperatively, right leg numbness was worse, but m o t o r function was improved, compared to the preoperative baseline. Two patients displayed delayed worsening of neurologic function many months after their operations. One had surgery for a lipomeningomyelocele and a tethered spinal cord; her bladder function subsequently deteriorated concurrent with the development o f massive obesity. The other patient developed vesicoureteral reflux and hydronephrosis 1 year after a meningomyelocele repair. A tethered spinal cord was diagnosed and surgery was performed; both of this patient's operations are included in this series. During three operations, structures which appeared to be atretic nerve roots, but whose stimulation elicited no E M G responses, were sectioned in order to maximally
188
cases.
ic-appearing nerve roots which were found to be functionally innervating muscle in another patient; they could not be differentiated by visual examination.
Discussion
Methodologic considerations
SEP findings
As in facial nerve identification [2], monitoring of EMG responses is clearly superior to observation for gross muscle twitching during cauda equina surgery. Visible or palpable limb movements are accompanied by deflections in the EMG, but smaller responses may only be recognizable electrically. While electrical stimulation with EMG recording can identify motor roots and mixed nerves, its ability to identify sensory nerve roots has been questioned [10]. The long latency of the response in Fig. 1 C suggests that the stimulated structure was a sensory nerve root, and that the EMG response represents muscle contraction mediated by a reflex arc. The latency difference between this EMG response and that elicited by motor nerve stimulation in the same region (Fig. 1 B) reflects propagation of the nerve impulses within more proximal portions of both motor and sensory roots as well as synaptic delays in the spinal cord. Thus, reflexive activation may permit identification of sensory as well as motor nerves by electrical stimulation. Sensory nerves might also be identified by stimulating them and recording SEPs over the spinal cord rostral to the surgical field [15], though the utility of this technique would be limited by the dorsal column somatosensory conduction blocks frequently found at the level of the neural plaque of a meningomyelocele [16]. Filum terminale and fibrous bands have also been differentiated from neural tissue by stimulating at one point and attempting to record a propagating action potential at another point along the same structure [10]. Maintaining contact with two pairs of electrodes simultaneously is more difficult, and dissection of a sufficiently long segment of the structure to be tested is not always possible. Moreover, this technique cannot be used to identify nerve roots embedded within tumor or scar tissue. Anal sphincter EMG recordings have been used for nerve identification during cauda equina surgery [7]. Our findings that some EMG responses are confined to a single muscle group points out the need to monitor several EMG channels concurrently. For example, a nerve root supplying leg muscles (Fig. 1 B - D ) would not be identified as a functional neural element, and might therefore be destroyed, if only anal sphincter EMG [7], or rectal, urethral, or bladder pressures [5, 6, 14, 17, 18] were monitored. Our EMG recording electrode placements were chosen for maximum coverage of the lumbar and sacral roots. Anterior thigh responses are elicited by stimulation of the L2, L3, or L4 roots, posterior thigh responses by stimulation of the L5, $1, or $2 roots, and external anal sphincter responses by stimulation of the $2, $3, or $4 roots [8]. The identification of functional neural tissue is based on the presence or absence of a clear response in the EMG waveform; the precise waveshape of the re-
untether the spinal cord. There were no new or increased neurologic deficits postoperatively in any of these three
The absence of SEPs generated rostral to the surgical site precluded intraoperative SEP monitoring in 40% of the overall patient group and in 9 out of the 12 patients with a history of meningomyelocele. In infants absence of SEPs may be due to maturational factors [3], but in older patients with meningomyeloceles it most likely reflects functional discontinuities of the dorsal column somatosensory pathways at the level of the neural plaque [16]. Although SEP monitoring has been shown to be useful during spinal cord surgery [9, 11, 13], surgical procedures were not altered by intraoperative SEP findings in this series of 25 cauda equina operations.
Intraoperative neurophysiology and surgical outcomes Increases in neurologic deficits which occur several months postoperatively do not reflect destruction of neural elements at the time of surgery. In one case, their correlation with massive weight gain suggests that they were due to associated overgrowth of residual lipomatous tissue in the region of the cauda equina [1]. The late increase in deficits in another patient was a manifestation of a delayed tethering of the spinal cord. Postoperative increased neurologic deficits lasting more than 4 - 6 weeks occurred in three patients. In one, they were caused by deliberate sacrifice of a nerve root which electrical stimulation demonstrated was functional. In the other two cases, improvement over the next several months suggested that some of these more persistent deficits were still due to reversible neurologic dysfunction rather than destruction of neural tissue. During 9 of the 25 operations, surgical procedures were altered in order to preserve areas or structures whose stimulation elicited EMG responses. In some cases, the structure being tested was then identified as a nerve root or as the spinal cord. In others, electrical stimulation demonstrated that functional neural elements were invisibly embedded within an area of tumor or scar tissue. Visual examination alone was not adequate for identification and localization of functional neural elements. Since the identified neural elements in these nine cases were functionally innervating limb or sphincter muscles, the patients presumably would have suffered persistent neurologic deficits if these tissues had not been preserved because of the intraoperative neurophysiologic testing. Conversely, in three cases the demonstration that apparent nerve roots were nonfunctional allowed the surgeons to divide them and optimally release tethered spinal cords without causing increased neurologic deficits. These nonfunctioning nerve roots looked like other atret-
189 sponse is n o t crucial. C o n t r a c t i o n o f a muscle distant f r o m the recording electrode m a y also cause a deflection in the electrophysiologic data due to m o t i o n artifact rather than direct E M G pickup, but that is equally useful as an indicator that the tissue being stimulated is a functioning nerve. W h e n the stimulating forceps are indeed t o u c h i n g a nerve, a false-negative E M G response m a y be caused by systemic paralysis, by technical factors that prevent the stimulus current f r o m passing into the tissue, or by technical factors that prevent recording o f an E M G Signal which is in fact present. The absence o f an E M G response can only be used to infer that the tissue being tested is n o t a nerve if (1) the lack o f systemic n e u r o m u s c u l a r blockade is verified, (2) the stimulator is n o t indicating an o p e n circuit condition, (3) and adequate stimulus artifact is present in the raw data, and (4) the recording electrodes are m a k i n g g o o d contact, as s h o w n by their impedances.
Conclusions D u r i n g surgery in the region o f the c a u d a equina, visual e x a m i n a t i o n alone is n o t a d e q u a t e for localizing and identifying functional neural elements, or for determining if atretic-appearing nerve roots are functional. Electrical stimulation with multichannel E M G facilitates the preservation o f functional neural elements and the optim i z a t i o n o f surgical results during c a u d a equina surgery.
Acknowledgements. The authors were supported in part by grants NS-25041 (A.D.L.) and MH-06723 (C.E.S.) from the United States Public Health Service. We would like to thank Joanne Gaudio, Shirley Seto, Jolynn Wagner, and Rosemary James for their technical assistance. We would also like to acknowledge the contributions of Drs. Kamram Tabaddor, Hugh Wisoff, Patrick LaSala, and Stephen Weitz, who provided some of the case data reported herein.
References 1. Bassett RC (1950) The neurologic deficit associated with lipomas of the cauda equina. Ann Surg 131:109-116 2. Delgado TE, Buchheit WA, Rosenholtz HR, Chrissian S (1979) Intraoperative monitoring of facial muscle evoked responses
obtained by intracranial stimulation of the facial nerve: a more accurate technique for facial nerve dissection. Neurosurgery 4:418-421 3. Gilmore R (1988) Use of somatosensory evoked potentials in infants and children. Neurol Clin 6:839-859 4. Harner SG, Daube JR, Ebersold MJ, Beatty CW (1987) Improved preservation of facial nerve function with use of electrical monitoring during removal of acoustic neuromas. Mayo Clin Proc 62:92-102 5. Hellbusch LC, Nihsen BJ (1989) Rectal sphincter pressure monitoring device. Neurosurgery 24:775-776 6. Ikeda K, Kubota T, Kashihara K, Yamamoto S (1986) Anorectal pressure monitoring during surgery on sacral lipomeningocele: case report. J Neurosurg 64:155-156 7. James HE, Mulcahy JJ, Walsh JW, Kaplan GW (1979) Use of anal sphincter electromyography during operations on the conus medullaris and sacral nerve roots. Neurosurgery 4:521 523 8. Kimura J (1983) Electrodiagnosis in diseases of nerve and muscle: principles and practice. Davis, Philadelphia 9. Legatt AD (1990) Intraoperative neurophysiologic monitoring. In: Frost EAM (ed) Clinical anesthesia in neurosurgery, 2nd edn. Butterworths, Stoneham, Mass, pp 63-127 10. Lindsay KW, Teasdale GM (1980) Electrophysiological identification of nerve roots during operations for spinal dysraphism. Surg Neurol 14:49-51 11. Macon JB, Poletti CE, Sweet WH, Ojemann RG, Zervas NT (1982) Conducted somatosensory evoked potentials during spinal surgery. II. Clinical applications. J Neurosurg 57:354-359 12. Matson DD (1979) Neurosurgery of infancy and childhood, 2nd edn. Thomas, Springfield, Ill, pp 5-60 13. McCallum JE, Bennett MH (1975) Electrophysiologic monitoring of spinal cord function during intraspinal surgery. Surg Forum 26:469-471 14. Pang D, Casey K (1983) Use of an anal sphincter pressure monitor during operations on the sacral spinal cord and nerve roots. Neurosurgery 13:562-568 15. Phillips LH II, Park TS (1990) Electrophysiological monitoring during lipomeningomyelocele resection. Muscle Nerve 13: 127132 16. Reigel DH, Dallmann DE, ScarffTB, Woodford J (1976) Intraoperative evoked potential studies of newborn infants with myelomeningocele. Dev Med Child Neurol 18 [Supp137]: 42-49 17. Toczek SK, McCullough DC, Boggs JS (1978) Sacral rootlet rhizotomy at the conus medullaris for hypertonic neurogenic bladder. J Neurosurg 48:193-196 18. Torrens MJ, Griffith HB (1976) Management of the uninhibited bladder by selective sacral neurectomy. J Neurosurg 44: 176185
Original papers
9 Springer-Verlag 1992
Child's Nerv Syst (1992) 8:185 189
Electrical stimulation and multichannel EMG recording for identification of functional neural tissue during cauda equina surgery Alan D. Legatt 1,2,5, Charles E. Schroeder 1,2,5, Bhagwant Gill 3'5, and James T. Goodrich 4,s 1 Department of Neurology, z Department of Neuroscience, 3 Department of Urology, and 4 Department of Neurosurgery, Albert Einstein College of Medicine and 5 Montefiore Medical Center, Bronx, New York, USA Received August 5, 1991
Abstract. Electrical stimulation of structures within the surgical field was used to identify functional neural elements during 25 cauda equina operations. E M G responses from anterior thigh, posterior thigh, and anal sphincter muscles were recorded simultaneously using a multichannel signal averager. During nine operations, stimulation of a presumed filum terminale or other tissue produced clear E M G responses, p r o m p t i n g modification of surgical procedures. In one patient, this resulted in preservation of a flattened spinal cord which resembled a band of scar tissue. Some E M G responses were restricted to a single muscle group; these neural structures would p r o b a b l y not have been identified if only a single-channel E M G recording was used. Visual examination alone was not adequate for identifying functional neural elements, or for determining whether atretic-appearing nerve roots were functional. Electrical stimulation with multichannel E M G recording facilitates the preservation of functional neural elements and the optimization of surgical results in cauda equina surgery. Key words: C a u d a equina surgery - Electric stimulation Electromyography - Evoked potentials, somatosensory Intraoperative neurophysiologic monitoring - Tethered spinal cord
Electrical stimulation m a y be used to localize and define neural structures during a variety of neurosurgicaI procedures [9]. F o r example, the facial nerve m a y b e identified by electrical stimulation during surgery for acoustic neuroma; this has been shown to result in improved surgical outcomes [4]. Peripheral nerves and nerve roots are also at risk during surgical procedures on the c a u d a e q u i n a . M a t s o n [12] described the importance of identifying neural elements
Correspondence to."A.D. Legatt, Department of Neurology, Montefiore Medical Center, 111 East 210th Street, Bronx, NY 10467, USA
with electrical stimulation during meningomyelocele repair; grossly visible muscle twitching was the positive result [10]. Stimulation-induced changes in rectal, urethral, and bladder pressures have been used to identify the nerve supply to these structures [5, 6, 14, 17, 18], and anal sphincter E M G has also been used to identify lower spinal nerve roots during surgery for spinal d y s r a p h i s m [7]. We use electrical stimulation with simultaneous E M G recording at multiple sites to identify nerves innervating both limb musculature and anal sphincter during cauda equina surgery, and report herein on our experience in 25 operations.
Patients and methods
Patient population These data are from 25 consecutive operations in the cauda equina region during which electrical stimulation with EMG recording was used. Of the 24 patients (one patient was monitored twice), 14 were female and 10 were male. Their ages ranged from I week to 38 years (mean 13 years). Most of the operations were performed for release of tethered spinal cord in patients with progressive neurologic deficits and radiologically documented tethers; many of these patients also had sacral lipomas. Five operations were for other tumors of the cauda equina or the lower spinal cord (a teratoma, a myxopapillary ependymoma, an astrocytoma, and two neurofibromas). In 15 cases, there was a history of prior lumbar surgery, mostly meningomyelocele repairs within the first few days of life or debulking of sacral lipomas. Four patients had re-tethered after previous surgery for tethered spinal cord, performed 1-11 years earlier.
Electrophysiology For EMG recordings, standard EEG gold cup electrodes were attached to the skin over the quadriceps femoris and hamstring muscles of each leg, in the anterior and posterior midlines and approximately midway between knee and hip. Anal sphincter electrical activity was recorded by sterile platinum-iridium EEG needle electrodes placed in subcutaneous tissue lateral to the anus. All six electrodes were included in the recording montage. EMG signals
186 were amplified (gain 3000 6000, bandpass 5 Hz 3000Hz) and displayed by a Nicolet Pathfinder I signal averager (Nicolet Instrument Corporation, Madison, Wisconsin, USA). Constant-current electrical square pulses (amplitude 1- 3 mA, duration 200 gs, rate 2/s) were passed through a capacitor (to remove any direct current component) and connected to a standard bipolar cautery forceps, which was placed by the surgeons on selected structures within the surgical field. Paralyzing agents were discontinued 15-30 min prior to nerve identification, and the lack of neuromuscular blockade was ascertained prior to stimulation of structures within the surgical field. If the stimulating forceps were not making proper contact with the tissue, the stimulator indicated an open circuit condition; the surgeons were notified and repositioned the forceps. A short circuit between the two blades of the forceps was signaled by the absence of the electrical stimulus artifact from the raw data; the forceps blades were kept slightly separated to prevent this. Somatosensory evoked potentials (SEPs) to unilateral stimulation of each posterior tibial nerve at the ankle were monitored using standard intraoperative SEP recording techniques [9]. Signal averaging was also useful for recognizing small EMG responses. The technical aspects of electrical stimulation and EMG recording will be described in greater detail elsewhere (A. D. Legatt et al., in preparation).
2oo.v:/ 4 ms
A
L
anal s p h i n c t e r R anal s p h i n c t e r
L
anal s p h i n c t e r L a n t e r i o r thigh
I~
L
p o s t e r i o r thigh L a n t e r i o r thigh
~
_
R p o s t e r i o r thigh R anterior thigh
B
C
Results
Nerve identification Clear, reproducible E M G responses in anal sphincter or leg muscles were recorded during 21 operations. In the other four, none of the structures selected for testing by the surgeons gave an E M G response. Stimulation of nerve roots sometimes elicited movements of the legs or buttocks which were visible to, or felt by, the surgons through the surgical drapes. M a n y E M G responses were unaccompanied by palpable or grossly visible muscle movements, however. Responses were often present in multiple channels, either refecting activation of several muscle groups or the presence of gross movements which generated motion artifacts at a number of recording electrodes. In some cases, however, responses were present in only a single channel (Fig. 1 B, C). In one patient, stimulation at one site produced a response in right leg muscles (Fig. 1 C) which was considerably longer in latency than the response in the same channel to stimulation of other neural structures (Fig. 1 B). Electrical stimulation confirmed the visual identification of apparent neural structure during m a n y operations (Fig. 2 A). During surgery for tethered spinal cord where a filum terminale was present, the filum was tested by electrical stimulation prior to sectioning. In most cases, no E M G responses were obtained and the filum was divided. However, E M G responses were elicited by stimulation of what grossly appeared to be the filum in three patients. In two of these, stimulation of an apparent filum gave a large E M G response in the anal sphincter (Fig. 1 E), and the structure was preserved. In the third, stimulation of the rostral filum elicited a large response over left leg muscles, even when the stimulus intensity was reduced to 0.5 m A (Fig. 2 B). The dissection was continued caudally, to a point where stimulation of the filum
D
Fig. 1A-E. Averaged EMG responses (average of 5 epochs) to electrical stimulation of various structures within the surgical field in an 11-year-old girl with a tethered spinal cord. A No response to stimulation at this site. B A large early response in the right posterior thigh and/or right anterior thigh. C A small, longer-latency response in the right leg muscles. D A response in both left anterior thigh and left posterior thigh muscles. E The EMG response to stimulation of a structure which appeared to be the filum terminale, but which was preserved because of the electrophysiologic findings
did not produce a significant E M G response (Fig. 2 C), and the filum was divided at that point. Low-intensity stimulation of the rostral stump still produced a large E M G response (Fig. 2 D). Several other structures beside the filum terminale were found to contain functional neural elements which were not grossly visible, but which were preserved because of the electrical stimulation results. Overall, testing led to alterations of surgical procedures which preserved functional neural tissue in 9 out of the 25 cases. Nerve roots hidden within scar tissue from previous surgery were localized and spared in three patients. In two other cases, stimulation of the remnant o f a sacral lipoma (after most of the tumor had been removed) showed that it
187 A
B
L anal sphincter R anal sphinctel _
~
L anal sphincter L anterior thigh
__
~,
/
~
L f
~__
~._~_ . . . .
C
,
.
v
~
,
~ ,.,
~
L p~176 hL anterior thigh R posterior thigh R anterior thigh Fig. 2A-D. Averaged EMG responses (average
D
_
:
_
.
.
.
.
.
.
,oo< !
contained functional nerve roots, and the resection was stopped. During removal of a large sacral neurofibroma in a patient with central neurofibromatosis, a small neurofibroma was noted on another nerve root. This root was determined to be innervating the anal sphincter, so it was not resected. After the majority of a sacral lipoma had been resected in one patient, a band of fibrous-looking tissue was encountered deep (anterior) to it. Stimulation of this apparent scar tissue elicited E M G responses from the anal sphincter, so it was preserved. On further dissection, it was determined that this tissue was actually a thinned-out ribbon of spinal cord extending anterior to the lipoma. In one patient, stimulation of what appeared to be atretic and nonfunctional nerve roots gave clear E M G responses, and the roots were preserved. During three other operations, structures which appeared to be atretic nerve roots, and which were mechanically tethering the spinal cord, elicited no response when stimulated. They were therefore sectioned in order to maximally untether the spinal cord.
SEP recordings SEPs were monitored during 24 of the 25 operations. Clear cervical and/or cortical SEPs were identifiable in 14 cases. SEP changes related to alterations o f the anesthetic regimen were present in six of these. There were no SEP changes attributable to surgical maneuvers. During ten operations, neither cortical nor cervical SEPs were identifiable at any point, though clear peripheral nerve SEPs were present. Cortical and cervical components had been absent in preoperative SEP recordings in nine of these, and abnormal in the other. In nine of the ten cases, the patient had a history of meningomyelocele;
of 5 epochs) to electrical stimulation of various structures within the surgical field in another 1l-year-old girl with a tethered spinal cord. A higher amplifier gain was used, and some of the waveforms were truncated. A Stimulation of a structure which was clearly a nerve root. B Stimulation of the rostral filum terminale. C Stimulation of the ilium more caudally. D Repeat stimulation of the filum at the rostra1 site (stimulation site of B) after it had been divided at the caudal site (stimulation site of I~)
the other patient had cauda equina scarring from arachnoiditis.
Outcomes Several patients had transient impairment o f bowel or bladder function, or worsening o f preexisting abnormalities, during the first few days to weeks after surgery, most likely due to intraoperative manipulation of the spinal cord. In order to assess outcomes, neurologic findings were evaluated at 4 - 6 weeks after surgery. Most patients were at their neurologic baseline at this time; several others showed improvements in function in the immediate postoperative period. Persistent increased neurologic deficits were noted postoperatively in three cases. Two patients had postoperative sphincter dysfunction, with gradual improvement over several months. In the third patient, a functioning nerve root was deliberately sacrificed to permit total resection of a cauda equina neurofibroma. Postoperatively, right leg numbness was worse, but m o t o r function was improved, compared to the preoperative baseline. Two patients displayed delayed worsening of neurologic function many months after their operations. One had surgery for a lipomeningomyelocele and a tethered spinal cord; her bladder function subsequently deteriorated concurrent with the development o f massive obesity. The other patient developed vesicoureteral reflux and hydronephrosis 1 year after a meningomyelocele repair. A tethered spinal cord was diagnosed and surgery was performed; both of this patient's operations are included in this series. During three operations, structures which appeared to be atretic nerve roots, but whose stimulation elicited no E M G responses, were sectioned in order to maximally
188
cases.
ic-appearing nerve roots which were found to be functionally innervating muscle in another patient; they could not be differentiated by visual examination.
Discussion
Methodologic considerations
SEP findings
As in facial nerve identification [2], monitoring of EMG responses is clearly superior to observation for gross muscle twitching during cauda equina surgery. Visible or palpable limb movements are accompanied by deflections in the EMG, but smaller responses may only be recognizable electrically. While electrical stimulation with EMG recording can identify motor roots and mixed nerves, its ability to identify sensory nerve roots has been questioned [10]. The long latency of the response in Fig. 1 C suggests that the stimulated structure was a sensory nerve root, and that the EMG response represents muscle contraction mediated by a reflex arc. The latency difference between this EMG response and that elicited by motor nerve stimulation in the same region (Fig. 1 B) reflects propagation of the nerve impulses within more proximal portions of both motor and sensory roots as well as synaptic delays in the spinal cord. Thus, reflexive activation may permit identification of sensory as well as motor nerves by electrical stimulation. Sensory nerves might also be identified by stimulating them and recording SEPs over the spinal cord rostral to the surgical field [15], though the utility of this technique would be limited by the dorsal column somatosensory conduction blocks frequently found at the level of the neural plaque of a meningomyelocele [16]. Filum terminale and fibrous bands have also been differentiated from neural tissue by stimulating at one point and attempting to record a propagating action potential at another point along the same structure [10]. Maintaining contact with two pairs of electrodes simultaneously is more difficult, and dissection of a sufficiently long segment of the structure to be tested is not always possible. Moreover, this technique cannot be used to identify nerve roots embedded within tumor or scar tissue. Anal sphincter EMG recordings have been used for nerve identification during cauda equina surgery [7]. Our findings that some EMG responses are confined to a single muscle group points out the need to monitor several EMG channels concurrently. For example, a nerve root supplying leg muscles (Fig. 1 B - D ) would not be identified as a functional neural element, and might therefore be destroyed, if only anal sphincter EMG [7], or rectal, urethral, or bladder pressures [5, 6, 14, 17, 18] were monitored. Our EMG recording electrode placements were chosen for maximum coverage of the lumbar and sacral roots. Anterior thigh responses are elicited by stimulation of the L2, L3, or L4 roots, posterior thigh responses by stimulation of the L5, $1, or $2 roots, and external anal sphincter responses by stimulation of the $2, $3, or $4 roots [8]. The identification of functional neural tissue is based on the presence or absence of a clear response in the EMG waveform; the precise waveshape of the re-
untether the spinal cord. There were no new or increased neurologic deficits postoperatively in any of these three
The absence of SEPs generated rostral to the surgical site precluded intraoperative SEP monitoring in 40% of the overall patient group and in 9 out of the 12 patients with a history of meningomyelocele. In infants absence of SEPs may be due to maturational factors [3], but in older patients with meningomyeloceles it most likely reflects functional discontinuities of the dorsal column somatosensory pathways at the level of the neural plaque [16]. Although SEP monitoring has been shown to be useful during spinal cord surgery [9, 11, 13], surgical procedures were not altered by intraoperative SEP findings in this series of 25 cauda equina operations.
Intraoperative neurophysiology and surgical outcomes Increases in neurologic deficits which occur several months postoperatively do not reflect destruction of neural elements at the time of surgery. In one case, their correlation with massive weight gain suggests that they were due to associated overgrowth of residual lipomatous tissue in the region of the cauda equina [1]. The late increase in deficits in another patient was a manifestation of a delayed tethering of the spinal cord. Postoperative increased neurologic deficits lasting more than 4 - 6 weeks occurred in three patients. In one, they were caused by deliberate sacrifice of a nerve root which electrical stimulation demonstrated was functional. In the other two cases, improvement over the next several months suggested that some of these more persistent deficits were still due to reversible neurologic dysfunction rather than destruction of neural tissue. During 9 of the 25 operations, surgical procedures were altered in order to preserve areas or structures whose stimulation elicited EMG responses. In some cases, the structure being tested was then identified as a nerve root or as the spinal cord. In others, electrical stimulation demonstrated that functional neural elements were invisibly embedded within an area of tumor or scar tissue. Visual examination alone was not adequate for identification and localization of functional neural elements. Since the identified neural elements in these nine cases were functionally innervating limb or sphincter muscles, the patients presumably would have suffered persistent neurologic deficits if these tissues had not been preserved because of the intraoperative neurophysiologic testing. Conversely, in three cases the demonstration that apparent nerve roots were nonfunctional allowed the surgeons to divide them and optimally release tethered spinal cords without causing increased neurologic deficits. These nonfunctioning nerve roots looked like other atret-
189 sponse is n o t crucial. C o n t r a c t i o n o f a muscle distant f r o m the recording electrode m a y also cause a deflection in the electrophysiologic data due to m o t i o n artifact rather than direct E M G pickup, but that is equally useful as an indicator that the tissue being stimulated is a functioning nerve. W h e n the stimulating forceps are indeed t o u c h i n g a nerve, a false-negative E M G response m a y be caused by systemic paralysis, by technical factors that prevent the stimulus current f r o m passing into the tissue, or by technical factors that prevent recording o f an E M G Signal which is in fact present. The absence o f an E M G response can only be used to infer that the tissue being tested is n o t a nerve if (1) the lack o f systemic n e u r o m u s c u l a r blockade is verified, (2) the stimulator is n o t indicating an o p e n circuit condition, (3) and adequate stimulus artifact is present in the raw data, and (4) the recording electrodes are m a k i n g g o o d contact, as s h o w n by their impedances.
Conclusions D u r i n g surgery in the region o f the c a u d a equina, visual e x a m i n a t i o n alone is n o t a d e q u a t e for localizing and identifying functional neural elements, or for determining if atretic-appearing nerve roots are functional. Electrical stimulation with multichannel E M G facilitates the preservation o f functional neural elements and the optim i z a t i o n o f surgical results during c a u d a equina surgery.
Acknowledgements. The authors were supported in part by grants NS-25041 (A.D.L.) and MH-06723 (C.E.S.) from the United States Public Health Service. We would like to thank Joanne Gaudio, Shirley Seto, Jolynn Wagner, and Rosemary James for their technical assistance. We would also like to acknowledge the contributions of Drs. Kamram Tabaddor, Hugh Wisoff, Patrick LaSala, and Stephen Weitz, who provided some of the case data reported herein.
References 1. Bassett RC (1950) The neurologic deficit associated with lipomas of the cauda equina. Ann Surg 131:109-116 2. Delgado TE, Buchheit WA, Rosenholtz HR, Chrissian S (1979) Intraoperative monitoring of facial muscle evoked responses
obtained by intracranial stimulation of the facial nerve: a more accurate technique for facial nerve dissection. Neurosurgery 4:418-421 3. Gilmore R (1988) Use of somatosensory evoked potentials in infants and children. Neurol Clin 6:839-859 4. Harner SG, Daube JR, Ebersold MJ, Beatty CW (1987) Improved preservation of facial nerve function with use of electrical monitoring during removal of acoustic neuromas. Mayo Clin Proc 62:92-102 5. Hellbusch LC, Nihsen BJ (1989) Rectal sphincter pressure monitoring device. Neurosurgery 24:775-776 6. Ikeda K, Kubota T, Kashihara K, Yamamoto S (1986) Anorectal pressure monitoring during surgery on sacral lipomeningocele: case report. J Neurosurg 64:155-156 7. James HE, Mulcahy JJ, Walsh JW, Kaplan GW (1979) Use of anal sphincter electromyography during operations on the conus medullaris and sacral nerve roots. Neurosurgery 4:521 523 8. Kimura J (1983) Electrodiagnosis in diseases of nerve and muscle: principles and practice. Davis, Philadelphia 9. Legatt AD (1990) Intraoperative neurophysiologic monitoring. In: Frost EAM (ed) Clinical anesthesia in neurosurgery, 2nd edn. Butterworths, Stoneham, Mass, pp 63-127 10. Lindsay KW, Teasdale GM (1980) Electrophysiological identification of nerve roots during operations for spinal dysraphism. Surg Neurol 14:49-51 11. Macon JB, Poletti CE, Sweet WH, Ojemann RG, Zervas NT (1982) Conducted somatosensory evoked potentials during spinal surgery. II. Clinical applications. J Neurosurg 57:354-359 12. Matson DD (1979) Neurosurgery of infancy and childhood, 2nd edn. Thomas, Springfield, Ill, pp 5-60 13. McCallum JE, Bennett MH (1975) Electrophysiologic monitoring of spinal cord function during intraspinal surgery. Surg Forum 26:469-471 14. Pang D, Casey K (1983) Use of an anal sphincter pressure monitor during operations on the sacral spinal cord and nerve roots. Neurosurgery 13:562-568 15. Phillips LH II, Park TS (1990) Electrophysiological monitoring during lipomeningomyelocele resection. Muscle Nerve 13: 127132 16. Reigel DH, Dallmann DE, ScarffTB, Woodford J (1976) Intraoperative evoked potential studies of newborn infants with myelomeningocele. Dev Med Child Neurol 18 [Supp137]: 42-49 17. Toczek SK, McCullough DC, Boggs JS (1978) Sacral rootlet rhizotomy at the conus medullaris for hypertonic neurogenic bladder. J Neurosurg 48:193-196 18. Torrens MJ, Griffith HB (1976) Management of the uninhibited bladder by selective sacral neurectomy. J Neurosurg 44: 176185
