Current Practice of Motor Evoked Potential Monitoring: Results of a Survey

Transcription

1 Journal of Clinical Neurophysiology 19(5): , Lippincott Williams & Wilkins, Inc., Philadelphia 2002 American Clinical Neurophysiology Society Current Practice of Motor Evoked Potential Monitoring: Results of a Survey Alan D. Legatt Departments of Neurology and Neuroscience, Montefiore Medical Center and the Albert Einstein College of Medicine, Bronx, New York Summary: Centers responding to a survey of MEP monitoring practices predominantly used transcranial electrical brain stimulation (TCES) with brief pulse trains and/or spinal cord stimulation (SCS) to elicit MEPs; transcranial magnetic stimulation and single-pulse TCES were not techniques of choice. Most centers using TCES had patient exclusion criteria (e.g., cochlear implants, cardiac pacemakers, prior craniotomy or skull fracture, history of seizures). Adverse effects included rare tongue injuries or seizures from TCES, and minor bleeding from needle electrodes in muscle. Spinal cord, peripheral nerve, and muscle recording sites were all employed. TCES with recording of muscle responses was the preferred MEP monitoring technique at the plurality of the centers. MEPs suitable for monitoring were obtained in about 91.6% of patients overall. Most of the failures were attributed to technical factors; preexisting neurologic dysfunction precluded MEP monitoring in approximately 1.7% of patients. Almost all centers monitored SEPs concurrently with MEPs. Overall, both measures remained stable during about 90.2% of cases. Adverse MEP changes occurred in about 8.3%; a little over half of these were accompanied by SEP changes. Adverse SEP changes without MEP changes occurred in about 1.5% of cases. SEPs and MEPs should be used together to optimally monitor the spinal cord. Keywords: MEPs SEPs Intraoperative monitoring Transcranial brain stimulation Spinal cord stimulation. Intraoperative recording of somatosensory evoked potentials (SEPs) generated in the brain following stimulation of peripheral nerves in the legs has been used for three decades to monitor the integrity of the spinal cord during surgery, and has been shown to improve surgical outcomes (Nuwer et al., 1995). However, SEPs only directly monitor the ascending dorsal column somatosensory pathways within the spinal cord (Emerson, 1988) and thus may fail to reflect spinal cord compromise that affects the descending motor pathways but spares the dorsal columns. This has led to the development of a variety of techniques that are intended to Address correspondence and reprint requests to Dr. Alan Legatt, Department of Neurology, Montefiore Medical Center, 111 East 210th Street, Bronx, New York ; Legatt@aecom.yu.edu monitor the motor pathways directly and have been labeled motor evoked potentials (MEPs). In contrast with SEP monitoring, which is a relatively mature discipline with fairly standardized techniques, MEP monitoring is a much newer field that is still in a state of flux; different MEP techniques are used in different institutions, and the techniques are still evolving. As MEP monitoring becomes more widely used, it is important to determine the optimal protocols for performing it. This study uses a survey of institutions where MEP monitoring is being performed to examine the current state-of-the art of MEP monitoring: It was designed to assess the extent to which the various MEP recording techniques are being used at different institutions and how they are being performed, to evaluate the effectiveness of the various techniques, and to see if a consensus technique is beginning to emerge. 454

2 CURRENT PRACTICE OF MEP MONITORING 455 METHODS TABLE 1. Stimulating electrodes for TCES-MEPs A survey form, approved by the Institutional Review Board (IRB) of the Montefiore Medical Center, was distributed to the entire membership of the American Society of Neurophysiological Monitoring (ASNM) and to the entire membership of the American Clinical Neurophysiology Society; in both cases the survey form accompanied other materials in regularly scheduled mailings of these societies. A copy of the survey form was also posted on the Internet, on the ASNM s Web site. Respondents were asked to provide details about the MEP techniques used at their centers and to estimate the number of cases monitored per year and the results obtained. Responses were received by mail, fax, and . RESULTS Completed survey forms were received from a total of 57 centers. Of these, 52 were located in the United States and 5 were located in other countries (2 in Canada, 1 in Mexico, and 2 in the Netherlands). Thirty-nine centers replied that they were currently performing MEP monitoring. Some of the 18 centers that were not performing MEP monitoring said that they hoped to do so in the future, and some also stated that the lack of IRB or other institutional approval had preventing them from performing MEP monitoring. One responder performed MEP monitoring at multiple hospitals and reported on a total of 6,000 cases per year. The other 38 centers that utilized MEPs monitored them during an aggregate total of 2,763 cases per year, with the number of MEP cases per center per year ranging from 4 to 535 (median 38). If the MEP techniques were analyzed solely by the number of patients in whom they were employed, the results of the survey would be dominated by the practices of the single large multihospital provider. Therefore, survey results will be predominantly presented as the number of centers utilizing each technique. Stimulation techniques Centers were asked what type of stimuli they used to elicit MEPs: transcranial magnetic brain stimulation, transcranial electrical brain stimulation, spinal cord stimulation utilizing electrodes outside the spinal canal, and spinal cord stimulation utilizing electrodes inside the spinal canal. For all four stimulation methods, centers were asked whether they used single pulses or pulse trains. For the two spinal cord stimulation methods, Stimulating electrode type centers were asked whether the stimulating electrodes were placed by the surgeons. Of the 39 centers performing MEP monitoring, 24 used the same stimulation protocol for all patients, and 15 used multiple stimulation protocols. Transcranial electrical stimulation No. of centers Corkscrew 9 Needle electrode 4 Surface cup or disk 3 Either corkscrew or surface disk 1 (no response to this question) 3 Transcranial electrical stimulation (TCES) of the brain was used to elicit MEPs at 20 centers. All of these used brief pulse trains rather than single stimuli, and all but one reported that TCES was their preferred stimulation technique for MEP monitoring. Nine of these centers utilized TCES exclusively; 11 of them used other stimuli to elicit MEPs during some operations. The stimulating electrode types used for TCES-MEP recordings are shown in Table 1. The mean number of pulses per train for TCES was approximately 4.8 (range 3 7). Most of the centers used an interpulse interval of 2.0 msec; two centers reported shorter interpulse intervals (0.3 or 0.4 msec) and four used longer interpulse intervals (2.5, 3.0, or 4.0 msec). Nine of these 20 centers used the same stimulus train parameters in all patients, and 11 of them individualized the train parameters based on the patients responses. The locations from which TCES-MEPs were recorded are listed in Table 2. The numbers total more than 20 because some centers recorded MEPs simultaneously from multiple sites following TCES. Both needle and surface recordings were used to record muscle responses following TCES; peripheral nerve recordings all utilized needle electrodes. In most centers, the muscle responses that were monitored were single epochs, recorded following a single brief TCES stimulus train. In contrast, TABLE 2. Recording sites for TCES-MEPs Recording site No. of centers Spinal cord electrodes outside the spinal canal 2 Spinal cord electrodes inside the spinal canal 4 Peripheral nerve 5 Muscle 19

3 456 A. LEGATT TABLE 3. Stimulating electrodes for SCS-MEPs TABLE 4. Recording sites for SCS-MEPs Stimulating electrode type No. of centers Recording site # of centers Outside the spinal canal, placed percutaneously 9 Outside the spinal canal, placed by the surgeons 17 Inside the spinal canal, placed percutaneously 1 Inside the spinal canal, placed by the surgeons 15 Spinal cord (caudal to region at risk) 11 Peripheral nerve 16 Muscle 22 several of the peripheral nerve and spinal cord recordings were averages of up to 50 epochs. Transcranial magnetic stimulation Only one center reported having used transcranial magnetic brain stimulation, utilizing a multi-pulse magnetic stimulator and either a butterfly or a single circle coil. The staff of this center reported that they had used transcranial magnetic stimulation only rarely, and that TCES is currently their MEP stimulation technique of choice. Direct cortical stimulation One center reported recording MEPs to direct cortical stimulation during intracranial surgery. They delivered pulse train stimuli through electrodes in a subdural grid that was placed on the brain by the surgeons. Spinal cord stimulation Electrical stimulation of the spinal cord (SCS) rostral to the region at risk, using electrodes either outside or inside the spinal canal, was used to elicit MEPs at 30 centers. Eleven of the centers that utilized SCS also recorded MEPs to TCES during some surgeries; 19 centers used exclusively SCS. As opposed to TCES, most of the SCS was typically accomplished using single stimuli rather than brief pulse trains; only four centers reported using pulse trains (of 2 to 5 pulses) for SCS. The electrode types used to stimulate the spinal cord are listed in Table 3. The numbers total more than 30 because nine centers used multiple (two or three) types of spinal cord stimulating electrode. Most centers used electrodes placed near the spinal cord by the surgeons. The sites from which SCS-MEPs were recorded are listed in Table 4. The numbers total more than 30 because some centers recorded MEPs simultaneously from multiple sites following SCS. Thirteen of the 16 centers that monitored peripheral nerve responses to SCS recorded them from needle electrodes. The three centers that used surface electrodes to record SCS-MEPs from peripheral nerves all monitored other responses as well, either from muscle or from the caudal spinal cord. As was the case with TCES, both needle and surface recordings were used to record muscle responses following SCS. In contrast to TCES-MEPs, where the muscle responses that were monitored were predominantly unaveraged, most of the centers used signal averaging to monitor muscle responses to SCS, with a mean of 28 epochs per average (range 2 to 100). Muscle recording and neuromuscular blockade Five of the 31 centers that monitored muscle responses to either TCES or SCS reported that the patients were not paralyzed during MEP monitoring; 25 centers used controlled neuromuscular blockade (NMB), and one center did not answer this question. The most common target value for the degree of NMB was the presence of 2-3 twitches following train-of-four stimulation. In eight centers, the paralytic drugs were administered as intermittent boluses; six of these eight centers reported that MEP testing typically produced patient movements. In 14 centers, continuous infusions of paralytic drugs were used; eight of these 14 centers reported that MEP testing typically produced patient movements. Three centers that used controlled NMB did not provide information about the method of paralytic drug administration. Centers were asked how they prevented MEP-related patient movements from interfering with the surgery. The commonest response was to warn the surgeon prior to MEP stimulation. Some responders watched the surgeons and performed MEP runs at opportune times, such as when the surgeon was not working within the surgical field. Only four centers reported adjusting the stimulus intensity as a strategy for reducing MEP-related patient movements. Safety and medicolegal considerations Most centers using TCES had patient exclusion criteria, including a history of seizures, a prior craniotomy or skull fracture, a metallic implant in the head, or an implanted stimulator (e.g. cochlear implant or cardiac pacemaker); one center also excluded patients with a

4 CURRENT PRACTICE OF MEP MONITORING 457 history of a stroke. Since TCES may activate the muscles of mastication, several centers reported using a bite block or tongue padding to prevent injury to the tongue. One center monitored the patient s EEG for afterdischarges following TCES. Several centers had criteria for limiting the stimulus intensity. One center reported that two patients had suffered tongue lacerations when they first began recording TCES-MEPs. They subsequently began using bite blocks, and have had no further morbidity from MEP monitoring. Tongue injuries were also reported by one other center, in some patients in whom the tongue padding was not in properly positioned during surgery. The latter center, which had the second largest number of cases per year (535), also reported some cases of minor bleeding from needle recording electrodes in muscle, and the occurrence of seizures in two cases. The patients with the seizures both had a prior history of seizures and were described as under-medicated, and only one of the seizures was temporally associated with transcranial brain stimulation. All of the other centers answered No to the question Have there been any adverse effects from MEP monitoring at your institution? One strategy for reducing the number of stimuli delivered to the central nervous system and the patient movements they produce is to limit the number of MEP recordings that are made. Seventeen centers recorded MEPs continuously, interspersed with recordings of SEPs or other monitoring modalities. Four centers recorded baseline MEPs and then did not record them again until explicitly requested by the surgeons. The other 18 centers recorded MEPs intermittently, with the interval between MEP runs depending on what was happening during the surgery. Thirteen of these 18 centers provided information about the typical interval between MEP runs during parts of the surgery that are not considered high-risk; the median interval was 9 minutes. Twenty-four of the 39 centers (including nine centers performing TCES) reported that they did not explicitly obtain patient consent for MEP monitoring. In seven centers consent for the monitoring was included in the surgical consent, and in eight centers a separate consent for MEP recordings was obtained by the monitoring team. Seven of the 39 centers were performing MEP monitoring under an IRB-approved protocol. MEP stimuli were generated by FDA-approved evoked potential recording systems at 26 centers. Six centers used a non-fda-approved high-intensity stimulator manufactured outside the United States under an Investigational Device Exemption (IDE); three centers reported using that same stimulator without an IDE. Three centers (two of them located outside the United TABLE 5. Preferred (or sole) MEP monitoring technique Monitoring technique No. of centers States) reported using non-fda-approved equipment, but did not specify the type of stimulator used. One center did not answer the question about the type of equipment used. Preferred MEP monitoring technique One survey question asked centers to identify their preferred MEP monitoring technique. The results are shown in Table 5; insufficient information denotes centers that used multiple techniques and for which the information provided was insufficient to determine a single preferred technique. The technique preferred by the most centers was TCES with recording of myogenic MEPs, the latter either from needle or from surface electrodes. SCS with recording of peripheral nerve signals (usually with needle electrodes), which has been labeled neurogenic motor evoked potentials (NMEPs), was the second preference in terms of the number of centers, but had the largest number of patients because it was the preferred technique of the aforementioned large multi-hospital monitoring provider. No center reported using spinal cord recordings as part of their preferred monitoring technique, though several reported that spinal cord recordings were performed together with MEP recordings at other sites. As part of their reply to this question, some centers wrote that they had previously used NMEPs as their preferred technique but had subsequently switched to TCES with EMG recording. Type of surgery monitored Total no. of cases per year TCES with spinal cord recording 0 0 TCES with peripheral nerve recording 0 0 TCES with muscle recording 15 1,035 SCS with spinal cord recording 0 0 SCS with peripheral nerve recording 9 6,347 SCS with muscle recording Insufficient information 9 1,190 One survey question asked about the types of surgery during which MEP monitoring was employed. The responses of the centers that provided this information, covering a total of 8,252 cases per year, are shown in Table 6. The choices provided in the questionnaire are shown in quotation marks, and the descriptions that the centers provided for the Other category are shown in

5 458 A. LEGATT TABLE 6. Applications of MEP monitoring Type of surgery the lower portion of the Table. Not surprisingly, the commonest application of MEP monitoring was during orthopedic spine surgery, such as for scoliosis, the same application for which SEP monitoring was initially developed. Some centers listed cervical spine operations, such as anterior cervical diskectomies, cervical corpectomies, and cervical spinal fusions under the Other category; it is possible that cervical spine operations may have been included in the numbers given for Orthopedic spine surgery by some of the other centers. In the vast majority of cases, MEPs were used to monitor the integrity of the spinal cord. However, two centers reported monitoring MEPs during peripheral nerve and plexus operations, two centers monitored MEPs during craniotomies for intracranial lesions (arteriovenous malformations, aneurysms, and brain tumors), one center used MEPs to monitor endovascular interventional radiology procedures, and once center reported using MEPs to monitor carotid surgery. Success rate of MEP monitoring No. of cases Orthopedic spine surgery (e.g. for scoliosis) * 6,000 Extramedullary spinal surgery (e.g. for 724 neurofibroma) Surgery for intramedullary spinal cord lesion 799 Surgery for posterior fossa or brainstem lesions 104 Other 625 Cervical spine surgery 305 Brain tumor, AVM, and aneurysm surgery 151 Peripheral nerve and plexus surgery 56 Carotid surgery 50 Endovascular radiology procedures 37 Spondylolisthesis surgery 15 Aortic aneurysm surgery 11 * the phrases in quotes appeared in the survey questionnaire The reported percentage of cases in which intraoperative MEP recordings were technically successful (i.e., elicited MEP waveforms suitable for intraoperative monitoring) ranged from 10% to 100% among the responding centers, but the distribution was highly skewed; only five centers reported a success rate lower than 80%. Combining the number of cases monitored at each center with the success rate for that center permits estimation of the overall success rate of MEP monitoring across all centers approximately 91.6% of the cases during which MEP monitoring was attempted. MEP monitoring failures were predominantly attributed to technical factors such as anesthesia, neuromuscular blockade, electrical artifacts, insufficient stimulus intensity, and misplaced or TABLE 7. Results of concurrent MEP and SEP monitoring-percent of cases at a center with each possible set of SEP and MEP findings, expressed as the average (range) for the 31 centers that provided this information Intraoperative monitoring findings dislodged electrodes. The absence of MEPs suitable for intraoperative monitoring was attributed to preexisting neurologic dysfunction in approximately 1.7% of the patients overall. Concurrent SEP and MEP monitoring Almost all of the centers monitoring both SEPs and MEPs; only two centers (collectively monitoring 100 cases/year) reported that they monitored MEPs without SEPs. Thirty-one centers provided complete information about the intraoperative findings in the two modalities and the degree to which they were concordant. The percentage of cases at each of the centers for the four possible classes of findings are shown in Table 7. These percentages were multiplied by the annual case load of each center to estimate the actual number of patients in each group at that center; these numbers were then combined across centers (see Table 8). Both measures were stable throughout the surgery in over 90% of the cases. Adverse changes (not necessarily persistent or signifying postoperative neurologic deficits) could occur in either measure alone or in both; MEP changes without SEP changes were more common than SEP changes without MEP changes. DISCUSSION % of cases at one center SEPs and MEPs both stable 90.5 (64 100) SEPs stable, adverse MEP changes 3.5 (0 30) MEPs stable, adverse SEP changes 2.0 (0 15) Adverse changes in both SEPs and MEPs 4.0 (0 36) Techniques for MEP monitoring have evolved significantly. The results of this survey show that TCES using brief pulse trains is an effective stimulation technique for TABLE 8. Results of concurrent MEP and SEP monitoring-estimated number of cases per year Intraoperative monitoring findings No. of cases %of total cases SEPs and MEPs both stable 7, SEPs stable, adverse MEP changes MEPs stable, adverse SEP changes Adverse changes in both SEPs and MEPs

6 CURRENT PRACTICE OF MEP MONITORING 459 MEP monitoring, and has essentially replaced the singlepulse technique that was used in the early trials of TCES-MEP monitoring (Levy et al., 1984). The latter technique may not reliably elicit MEPs because trains of action potentials in the spinal cord motor pathways are necessary to fire the alpha motor neurons (by temporal summation of the excitatory postsynaptic potentials in these neurons) and produce reliable peripheral nerve- or muscle-response MEPs under surgical anesthesia (Sloan, 2002). The D-waves and I-waves elicited by a single cortical stimulus may suffice in the unanesthetized state, but surgical anesthesia suppresses the production of cortical I-waves (Hicks et al., 1992, Woodforth et al., 1999). Trains of cortical stimuli produce trains of D- waves in the pyramidal tracts that are sufficient to depolarize the alpha motor neurons and permit MEP monitoring. Transcranial magnetic stimulation of the brain is even more susceptible to suppression by anesthesia (Sloan, 2002); the results of this survey suggest that it is not an MEP monitoring technique of choice at the present time. The results of this survey indicate that the NMEP technique is still widely used for intraoperative monitoring of the spinal cord, and it is still considered by many to be an MEP monitoring technique, i.e. a technique that directly monitors the descending motor pathways. However, it has been shown that action potentials recorded from peripheral nerves in the legs following rostral spinal cord stimulation are conducted through the spinal cord predominantly within the dorsal columns, in retrograde fashion (Toleikis et al., 2000), and that monitoring of NMEPs may fail to detect motor tract damage sufficient to cause paraplegia (Minahan, 2001). The other protocol that was reported as the preferred technique by some of the centers in this study is rostral SCS with muscle recording. These responses most likely represent a mixture of activity that was conducted in retrograde fashion through the dorsal columns and in anterograde fashion through the descending motor tracts of the spinal cord (Legatt and Emerson, 2002). Since the stimulators of evoked potential averagers are FDA approved for peripheral nerve stimulation (to record SEPs), using them to stimulate the brain or the spinal cord constitutes off-label use of the equipment. The other alternative is to use non-fda approved equipment. This situation has prevented several of the centers that responded to this survey from recording MEPs, denying their patients the benefits of MEP monitoring. There is a clear need for equipment that is FDA-approved for the stimulation protocols of MEP monitoring. However, it should be noted that the majority of the centers responding to this survey used FDA-approved evoked potential recording systems in an off-label manner to provide the stimuli for their MEP recordings. The U.S. Food and Drug Administration (1998) says that off-label use of FDA-approved medical devices is legally permissible and when the intent is the practice of medicine does not require the submission of an Investigational New Drug Application (IND), Investigational Device Exemption (IDE) or review by an Institutional Review Board (IRB). The most common current practice of MEP monitoring, used at the majority of the centers responding to this survey, is that the monitoring is not performed under an IRB protocol, and a separate consent for MEP monitoring is not obtained. The results of this survey indicate that state-of-the-art MEP monitoring techniques are effective (yielding MEPs suitable for monitoring in over 90% of cases) and, when appropriate precautions are taken, acceptably safe. The only adverse effects that were reported in this survey were two seizures in patients with epilepsy, minor bleeding from needle electrodes in muscle, and tongue injuries that could have been prevented by the use of properly positioned bite blocks or tongue padding during TCES. Insults that affect the anterior spinal cord usually affect the dorsal cord as well, which is why SEP monitoring improves neurologic outcomes (Nuwer et al., 1995). Slightly more than half of the adverse MEP changes that were encountered in this study were accompanied by adverse SEP changes (Table 8), though the survey design did not permit ascertainment of whether these changes were persistent or transient, or whether they were associated with new postoperative neurologic deficits. Some of the adverse MEP changes without SEP changes may have been false-positive changes, due to the greater anesthetic sensitivity of MEPs as compared to SEPs (Sloan, 2002) or to interpretation of the more labile MEPs by the stricter amplitude and latency criteria appropriate for SEPs. However, some of them probably reflected isolated anterior spinal cord compromise; it has been clearly demonstrated that SEP monitoring may fail to detect descending motor tract damage sufficient to cause paraplegia (Ben-David et al., 1987a; Zornow et al., 1990). Adverse SEP changes without MEP changes occurred in about 1.5% of cases. The dorsal columns may be damaged without compromise of the motor tracts (Ben- David et al., 1987b); thus, SEP monitoring remains necessary even if successful MEP monitoring is performed. SEPs and MEPs should be used together for optimal intraoperative monitoring of the spinal cord. Acknowledgments: The author would like to thank all who responded to this survey. He would also like to thank the

7 460 A. LEGATT American Society of Neurophysiological Monitoring and the American Clinical Neurophysiology Society for distributing the survey questionnaire. REFERENCES Ben-David B, Haller G, Taylor P. Anterior spinal fusion complicated by paraplegia: a case report of a false-negative somatosensoryevoked potential. Spine 1987a;12: Ben-David B, Taylor PD, Haller GS. Posterior spinal fusion complicated by posterior column injury. A case report of a false-negative wake-up test. Spine 1987b; 12: Emerson RG. Anatomic and physiologic bases of posterior tibial nerve somatosensory evoked potentials. Neurol Clin 1988;6: Hicks RG, Woodforth IJ, Crawford MR, Stephen JP, Burke DJ. Some effects of isoflurane on I waves of the motor evoked potential. Brit J Anaesth 1992;69: Legatt AD, Emerson RG. Motor evoked potential monitoring it s about time. J Clin Neurophysiol 2002;19: Levy WJ, York DH, McCaffrey M, Tanzer F. Motor evoked potentials from transcranial stimulation of the motor cortex in humans. Neurosurgery 1984;15: Minahan RE, Sepkuty JP, Lesser RP, Sponseller PD, Kostuik JP. Anterior spinal cord injury with preserved neurogenic motor evoked potentials. Clin Neurophysiol 2001;112: Nuwer MR, Dawson EG, Carlson LG, Kanim LEA, Sherman JE. Somatosensory evoked potential spinal cord monitoring reduces neurologic deficits after scoliosis surgery: results of a large multicenter survey. Electroencephalogr Clin Neurophysiol 1995;96: Sloan TB. Evoked potentials. Anesthesia and motor evoked potentials monitoring. In: Deletis V, Shills J, eds. Neurophysiology in Neurosurgery, San Diego, Academic Press, 2002: Toleikis JR, Skelly JP, Carlvin AO, Burkus JK. Spinally elicited peripheral nerve responses are sensory rather than motor. Clin Neurophysiol 2000;111: U.S. Food and Drug Administration. Off-Label and Investigational Use Of Marketed Drugs, Biologics, and Medical Devices. FDA Information Sheets 1998; Woodforth IJ, Hicks RG, Crawford MR, Stephen JPH, Burke D. Depression of I waves in corticospinal volleys by sevoflurane, thiopental, and propofol. Anesth Analg 1999;89: Zornow MH, Grafe MR, Tybor C, Swenson MR. Preservation of evoked potentials in a case of anterior spinal artery syndrome. Electroencephalogr Clin Neurophysiol 1990;77: