Stephen J. Fried, M.D. (1), Diane M. Smith, M.D. (1), Alan D. Legatt, M.D., Ph.D. (1,2)
Departments of (1) Neurology,  (2) Neuroscience, Montefiore Medical Center and the Albert Einstein College of Medicine, Bronx, NY , U.S.A


Intraoperative monitoring of corticospinal tract motor pathways with motor evoked potentials (MEPs) reduces the likelihood of neurological deficits following scoliosis surgery. The most commonly used alarm criterion for MEP changes is a significant amplitude drop, though what constitutes “significant” is not universally agreed upon and may, in fact, be different in different muscles. We compared the run-to-run variability of MEPs across three muscle groups (thenar/hypothenar, tibialis anterior, and abductor hallucis), recorded under steady-state propofol infusion during 30 surgeries for idiopathic scoliosis in which MEPs remained present throughout and there were no post-operative neurological deficits. For each muscle group, there was no significant difference in the coefficient of variation (CV) between the left-sided and right-sided MEPs (Wilcoxon rank-sum). The thenar/hypothenar MEPs showed the greatest variability, with an average CV of 48%. This was significantly larger than the average CV within the abductor hallucis MEPs, 35% (p<0.001). This, in turn, was significantly larger than the average CV within the tibialis anterior MEPs, 27% (p<0.001). We conclude that, due to differences in run-to-run variability across muscle groups, MEP alarm criteria might best be tailored to the particular muscle group being monitored.


Intraoperative neurophysiologic monitoring can detect spinal cord compromise and help to avert neurologic morbidity during spinal deformity surgery. Somatosensory evoked potential (SSEP) monitoring reduces post-operative neurological deficits, but paraplegia can occur despite normal SSEP findings because SSEPs monitor dorsal column function whereas the corticospinal tract motor pathways are in the anterior spinal cord, in a different vascular territory, and spinal cord damage is often the result of vascular injury. Thus, direct monitoring of the corticospinal tract motor pathways via motor evoked potentials (MEPs) has become the standard of care for spinal cord monitoring in scoliosis surgery.

While alarm criteria for SSEPs are almost uniformly accepted, MEP alarm criteria are still not widely agreed upon. The most common alarm criteria used are based on response amplitudes. Some have suggested that only complete loss of MEPs should be considered an alarm criterion. Others have suggested an 80% amplitude drop, while still others have more recently suggested a 70% amplitude drop, as alarm criteria. Additional suggested alarm criteria include analysis of the MEP waveform morphology and measuring the stimulus threshold for eliciting an MEP, with alarm values based on changes in the stimulus threshold; however, both of these methods may be more difficult to implement than a simple amplitude-based alarm.

One of the major reasons for the debate on MEP alarm criteria is the significant run-torun variability of MEPs. However, few studies have actually quantified the degree of run-torun MEP variability, and the underlying cause of such variability is uncertain. It is clear that D-waves show slight variability, whereas myogenic MEPs show substantial variability. The latter may reflect varying recruitment of lower motor neurons by the descending corticospinal tract volley.

The goal of our study is to assess MEP variability in “real world” neuromonitoring conditions. In otherwise neurologically normal patients, with neither postoperative deficits nor significant MEP changes during surgery, under steady-state propofol infusion, how variable are MEP amplitudes? If one uses alarm criteria based on deviation from the baseline, such as a 2, 2.5, or 3 standard deviation change as is often used in medical testing, how large an amplitude drop should be considered an alarm criterion? Additionally, are all muscles equally variable, such that the same alarm criteria can be applied to any monitored muscle group?


We retrospectively reviewed the intraoperative monitoring studies for posterior spinal fusion surgeries performed for scoliosis that were conducted at the Montefiore Medical Center over a 32 month period. We included only idiopathic scoliosis cases without preoperative neurological deficit, in which there were no significant adverse changes during the procedure and there were no post-operative neurological deficits. Significant adverse change was defined as a 90% amplitude drop or loss of MEPs in any of the monitored muscle groups. Inclusion criteria also required at least 90 minutes and/or 9 MEP runs during which the stimulation parameters and the propofol infusion rates remained constant. Additionally, no inhalational anesthetics or neuromuscular blocking agents could be in effect during the period of analysis, and the anesthetic regimen in all cases consisted of propofol and an opiate. A total of 30 subjects were included in the analysis.

The MEP data were recorded using Protektor evoked potential recording systems (XLTek Corporation, Oakville, Ontario, Canada). Motor evoked potential electrodes were placed in the C1 and C2 positions according to the international 10-20 system. A multipulse, constant-current trans-cranial electrical stimulation (TCES) technique was used, with pulse number varying between 5 and 7, and the intensity ranging between 100 and 200 mA. Data were filtered with a bandpass of 5 Hz – 3000 Hz for the muscle MEP recording channels. Impedances were maintained less than 2.5 kO. The responses contralateral to both the TCES anode and cathode were recorded, though for the analysis only those responses contralateral to the TCES anode were used.

We analyzed the MEPs that were recorded in the left and right tibialis anterior muscle, abductor hallucis muscle, and in a recording between the abductor pollicis brevis (APB) and abductor digiti minimi (ADM) muscles. The amplitudes of the MEPs were recorded, and coefficients of variation (CV) calculated for each of the muscle groups, in each recording channel. The resulting values in each muscle group were then compared to each other in the aggregate of cases, using the two-tailed Wilcoxon rank-sum test (MATLAB, Mathworks. Natick, Massachusetts).


Of the 30 subjects included in the analysis, there were 22 females and 8 males. Their ages ranged from 11 to 21 years. The steady-state propofol infusion rates ranged from 75 to 160 µg/kg/min. As described above, only periods during which the TCES stimulus and propofol infusion remained constant were taken for analysis. A median of 16 runs per case were taken for analysis, with a median time of analysis of 136 minutes.

Over the aggregate of cases, for the abductor hallucis muscle the average CV was 35% (33% on the left and 38% on the right). For the tibialis anterior muscle, the average CV was 27% (27% on the left and 27% on the right). Lastly, for the APB/ADM muscles, the average CVs was 48% (45% on the left and 51% on the right). A summary of the CV values is shown in Table 1.

Each muscle was compared to its counterpart on the contralateral side, as well as to the other muscle groups. Upon left-right comparison within a given muscle group, there was no statistically significant difference within the tibialis anterior (p=0.64), abductor hallucis (p=0.18), or APB/ADM (p=0.24) muscle groups.

However, clear significance was noted when comparing two different muscle groups. The thenar/hypothenar MEPs showed the greatest variability, with an average CV of 48%. This was significantly larger than the average CV within the abductor hallucis MEPs, 35% (p<0.001). This, in turn, was significantly larger than the average CV within the tibialis anterior MEPs, 27% (p<0.001).

If MEPs could be treated as having a standard distribution, what would be the corresponding two standard-deviation drop from average that might be regarded as a statistically significant deviation, and thus a cause for alarm? For the most variable recording channel (APB/ADM), the alarm value would be a drop of 96%. For the intermediately variable abductor hallucis muscle, the alarm value would be a drop of 70%. For the least variable tibialis anterior muscle, the alarm value would be a drop of 54%. Figure 1 illustrates the differences in MEP variability using data from one patient. If the above alarm criteria were applied to the dataset in figure 1, no monitoring alerts would have been generated.


TCES may be highly effective in stimulating the axons of the corticospinal tract, but it is difficult to fire the lower motor neurons due to anesthetic effects on spinal cord synapses; pulse train stimulation is used to facilitate this synaptic transmission. Under anesthesia, only a small percentage of the lower motor neuron pool fires, and it is a different subset each time. That is the etiology of the run-to-run variability in the MEPs.

The most striking finding of this study is the differential variability in MEP amplitudes across different muscle groups. This would suggest that, if one wanted to define a “statistically significant” change from baseline as an alarm criterion for MEPs, then the percentage change might be muscle-dependent. In particular, it might be appropriate to define the alarm criterion for the less variable tibialis anterior MEPs at a smaller percentage change than for the more variable abductor hallucis muscle. The muscle group with the most variable MEPs, the APB/ADM, would seem to warrant the largest percentage change alarm criterion.

If the differences in MEP variability between the various recording channels arose purely by chance, then one might expect that there would also be chance differences in variability between homologous muscle groups on the two sides of the body. But we found no significant right-left differences in variability. This suggests that the differences in variability in different recording channels truly differences in physiology.

While it is not entirely surprising that different muscle groups might show different patterns of variability, what is much less clear is why they do so in this particular pattern. Differential recording between two adjacent muscles might contribute to the greater variability of the MEPs in the hand muscles, but this would not explain the statistically significant difference in variability between the two lower extremity muscles under study. The number of motor units in the muscles has the pattern APB/ADM > TibAnt > Abd Hall. The density of motor units does have the pattern APB/ADM > Abd Hall > TibAnt, but since a larger number of motor units would be sampled in the muscle with the greatest density (APB/ADM), one would expect the least variability in the APB/ADM using the principle that the larger the sample size within a given distribution, the less variability will occur from random selection of elements.

The pattern of variability does parallel the degree to which each muscle is under direct corticospinal tract control. It is well known that MEP’s are more easily elicited in more distal muscles than in more proximal ones, because the more distal muscles are under more direct corticospinal tract control; the hand and fingers have a much larger cortical representation than do the leg muscles. If there are fewer corticospinal tract fibers, the strength of each excitatory synaptic connection from the upper motor neuron to the lower motor neuron may be greater, and thus synaptic transmission will be more reliable and the recruitment of lower motor neurons  will be more consistent from trial to trial.

We are left with no definite understanding as to why this particular order of MEP amplitude variability was present. However, it is clear that MEP variability differs significantly between different muscle groups. Our study suggests that MEP alarm criteria, rather than being uniform across all muscles tested, might best be tailored to each muscle group that is being monitored.


Ben-David B, et al.. Anterior spinal fusion complicated by paraplegia: a case report of a false-negative somatosensory-evoked potential. Spine 1987; 12:536–539.

Burke D, et al.. Trial-to-trial variability of corticospinal volleys in human subjects. Electroencephalogr Clin Neurophysiol1995; 97: 231–7.

Calancie B, et al.. ‘‘Threshold-level’’ multipulse transcranial electrical stimulation of motor cortex for intraoperative monitoring of spinal motor tracts: description of method and comparison to somatosensory evoked potential monitoring. J Neurosurg 1998; 88: 457–70.

Esformes JI, et al.. Measurement of human muscle volume using ultrasonography. Eur J Appl Physiol 2002; 87(1): 90-2.

Ginsburg HH, et al.. Postoperative paraplegia with preserved intraoperative somatosensory evoked potentials: case report. J Neurosurg1985; 63: 296– 300.

Jacobson MD, et al.. Architectural design of the human intrinsic hand muscles. J Hand Surg Am. 1992; 17(5): 804-9. 

Johns RK, Fuglevand AJ. Number of motor units in human abductor hallucis. Muscle Nerve 2011; 43(6): 895-6.

Kobayashi S, et al. A new warning point of transcranial electrical stimulation motor evoked potentials for intraoperative spinal cord monitoring: a prospective multicenter study from the Spinal Cord Monitoring Working Group of the Japanese Society for Spine Surgery and Related
Research. J Neurosurg Spine. 2014; 20(1):102-7.

Kothbauer K, et al.. Motor evoked potential monitoring for intramedullary spinal cord tumor surgery: correlation of clinical and neurophysiological data in a series of 100 consecutive procedures. Neurosurg. Focus, 1998; 15: 4(5):e1. 

Krieger D, et al. Pure motor hemiparesis with stable somatosensory evoked potential monitoring during aneurysm surgery: case report. Neurosurgery 1992; 31: 145–150.

Kura H, et al.. Quantitative analysis of the intrinsic muscles of the foot. Anat Rec. 1997; 249(1): 143-51.

Langeloo DD, et al.. Transcranial electrical motor-evoked potential monitoring during surgery for spinal deformity: a study of 145 patients. Spine 2003; 28: 1043–50.

Legatt AD, et al.. Loss of Lower Limb Motor Evoked Potentials and Spinal Cord Injury During the Initial Exposure in Scoliosis Surgery. J Clin Neurophysiol 2014.

Legatt AD, et al.. Technical Aspects of Transcranial Magnetic and Electrical Stimulation. In: Schomer DL, Lopes da Silva F (Eds.) Niedermeyer’s Textbook of Electroencephalography: Basic Principles, Clinical Applications, and Related Fields, 6th Edition. Lippincott William & Wilkins,
Philadelphia, 2010: 1129-1138.

Macdonald DB, et al. Intraoperative motor evoked potential monitoring – a position statement by the American Society of Neurophysiological Monitoring. Clin Neurophysiol. 2013;124(12): 2291-316.

Machida M, et al. Dissociation of muscle action potentials and spinal somatosensory evoked potentials after ischemic damage of spinal cord. Spine 1988; 13: 1119–1124.

Magistris MR, et al.. A clinical study of motor evoked potentials using a triple stimulation technique. Brain 1999;122 ( Pt 2):265-79.

McNeil CJ, et al.. Motor unit number estimates in the tibialis anterior muscle of young, old, and very old men. Muscle Nerve 2005; 31(4): 461-7.

Nuwer MR, et al.. Somatosensory evoked potential spinal cord monitoring reduces neurologic deficits after scoliosis surgery: results of a large multicenter survey. Electroencephalogr Clin Neurophysiol 1999; 96: 6—11. Nuwer MR, et al. Evidence-based guideline update: intraoperative spinal monitoring with somatosensory and transcranial electrical motor evoked
potentials. J Clin Neurophysiol 2012;29:101-108.

Penfield W, Jasper H. Epilepsy and the functional anatomy of the human brain. Boston: Brown L; 1954. Quinones-Hinojosa A, et al. Changes in transcranial motor evoked potentials during intramedullary spinal cord tumour resection correlate with postoperative motor function. Neurosurgery 2005; 56: 982—93.

Shefner JM, et al. Multipoint incremental motor unit number estimation as an outcome measure in ALS. Neurology 2011; 77(3): 235-4.

Woodforth IJ, et al.. Variability of motor-evoked potentials recorded during nitrous oxide anesthesia from the tibialis anterior muscle after transcranial electrical stimulation. Anesth Analg. 1996 ;82(4):744-9.

Zornow MH, et al. Preservation of evoked potentials in a case of anterior spinal artery syndrome. Electroencephalogr Clin Neurophysiol 1990; 77: 137–139.