The Role of Transcranial Motor Evoked Potentials in Predicting Neurologic and Histopathologic Outcome after Experimental Spinal Cord Ischemia

Animal care and experimental procedures were performed in compliance with The National Guidelines for Care of Laboratory Animals in the Netherlands. The study protocol was approved by the Animal Research Committee of the Academic Hospital at the University of Amsterdam, the Netherlands. Fifteen pigs, weighing 47 ± 5 kg, were included in the study.

Premedication consisted of ketamine 15 mg/kg intramuscularly. Anesthesia was induced with inhalation by mask of 2.0% isoflurane in a mixture of 50% O²in air. Two intravenous catheters (18 gauge) were placed in an ear vein, and normal saline was infused at a rate of 15 ml · kg⁻¹· h⁻¹. After induction of anesthesia, sufentanil 15 μg/kg and clonidine 2 μg/kg were given intravenously, isoflurane was discontinued, and anesthesia was maintained with a continuous infusion of ketamine (15 mg · kg⁻¹· h⁻¹), sufentanil (5 μg · kg⁻¹· h⁻¹), and clonidine (1 μg · kg⁻¹· h⁻¹). The tracheas were intubated, and animals were ventilated using intermittent positive pressure ventilation. End-tidal carbon dioxide concentration was measured by a mainstream capnograph (Hewlett-Packard, Boebingen, Germany), and Paco²was maintained between 36 and 40 mmHg. Adequacy of ventilation was confirmed by blood gas analysis at 37°C. The level of neuromuscular blockade was monitored electromyographically using a Datex Relaxograph (Datex-Ohmeda, Helsinki, Finland) placed at the animal's wrist equivalent after stimulation of the median nerve. A closed loop infusion system with pancuronium was used to maintain 40% relaxation as referenced to the control situation. This level of muscle relaxation was chosen to provide an optimal trade off between spontaneous movements of the animal and the minimal requirements for myogenic evoked potential recording and assessment of adequate depth of anesthesia. Arterial blood pressure and central venous pressure were measured from catheters placed in the right popliteal artery and the left cephalic vein, respectively. Oxygen saturation was continuously assessed by pulse oximetry. Nasopharyngeal temperature was monitored and kept at 38°C by means of a heating lamp. Urinary output was measured throughout the experiment. Before incision, the skin of the animal was pinched with a forceps, and absence of spontaneous movement or an increase of arterial blood pressure was an indication for adequate depth of anesthesia. Before the start of ischemic manipulations, arterial p  H, Pao², Paco², hemoglobin concentration, and hematocrit were measured.

Animals were placed in the right decubitus position. Under sterile conditions a laparotomy was made through a midline incision, and the viscera were transposed to the right. The abdominal aorta, the arising lumbar arteries, and the sacral artery were exposed. The thoracic duct, which in pig is much larger compared to the thoracic duct in humans, is at risk during the exposure of segmental arteries. Rupture may result in development of severe dyspnea during the postoperative phase, as observed in a pilot study. To avoid damage, we separated the entire lumbar part of the duct from the aorta. Incidental leaks were meticulously sutured, and all animals in the study remained free from pulmonary complications. Permanent occlusion of segmental arteries (group B) was carried out with a clip application instrument (ligaclip MCA; Ethicon, Somerville, NJ). After induction of SCI (see Experimental Design), the abdomen was closed in layers. The spinal cord was harvested at 72 h, during which procedure the interior of the aorta was inspected to determine whether all lumbar and intercostal segmental arteries (SAs) had been identified.

Figure 1shows the experimental setup. Fifteen minutes before ischemic manipulations, baseline values for tcMEP were obtained. Animals were randomized to undergo either transient tcMEP reduction (group A, n = 5) or prolonged tcMEP reduction (group B, n = 10). The target period of tcMEP loss was less than 10 min in group A and 60 min in group B. To induce SCI, SAs were clamped in a caudal-to-cranial direction. A time interval of 5 min was applied between the occlusion of two successive arteries. In group A, a number of SAs were occluded sequentially with removable arterial clamps, which was sufficient to abolish tcMEP completely. One or several clamps were then sequentially released, starting at the most cranial arterial clamp, until tcMEP returned to values greater than 75% of baseline. The remaining clamps were replaced by vascular clips that permanently occlude an artery. Permanent clipping of these arteries was performed to approach the clinical situation of aortic aneurysm repair, during which only SAs that are critical for spinal cord perfusion (as identified by tcMEP recording) are reattached to the graft. 17tcMEP monitoring was discontinued at 60 min after the initial loss of the tcMEP signal. In group B, all arteries were sequentially occluded with permanent vascular clips. tcMEP monitoring was discontinued at 60 min after the primary loss of tcMEP signals. After termination of tcMEP recording, anesthesia was discontinued in all animals. At 72 h after the initial tcMEP loss, animals underwent the final neurologic scoring, and spinal cords were harvested after the animals were killed.

Transcranial motor evoked potential stimuli were applied with a transcranial electrical stimulator (Digitimer D 185 cortical stimulator; Welwyn Garden City, UK) through four needle electrodes attached to the scalp. A train-of- four square wave pulses (peak voltage 600–1,200 V), with a duration of 50 μs and an interstimulus interval of 2 ms, was distributed over the motorcortex via  an anode located at the occiput and three interconnected cathodes placed behind the ears and in the soft palate. Compound muscle action potentials were recorded from the skin over the right upper limb triceps muscle and bilaterally over the lower limb quadriceps muscles using adhesive gel Ag/AgCl electrodes. The signals were amplified 5,000–20,000 times (adjusted to obtain maximum vertical resolution) and filtered between 30 and 1,500 Hz using a 3T PS-800 biologic amplifier (Twente Technology Transfer, Twente, the Netherlands). Stimulus intensity was adjusted to acquire maximal responses, and recording was performed at a setting of 10% above the stimulus level that obtained maximal amplitude. Amplitude of the compound muscle action potentials was defined as the peak-to-peak distance in microvolts. A reduction of tcMEP amplitude on the muscle groups monitored to less than 25% of the baseline value was considered an indication of ischemic spinal cord dysfunction. This criterion is based on the assumption that an amplitude decrease below three times the SD should provide optimal detection of ischemia while limiting the rate of false-positive monitoring results and our previous observation of a 26% within-patient variability of the tcMEP. 16Loss of tcMEP signal was defined as reduction of the amplitude to below 40 μV (this value represented approximately 2 times the peak-to-peak noise level in waveforms acquired without stimulation). This complete reduction of tcMEP amplitude is presented in Results as tcMEP absence. Baseline motor evoked potentials were obtained during laparotomy by averaging 10 consecutive responses before the clamping of SAs. Responses were displayed and stored on a MacIntosh Quadra computer (Apple Computer, Cupertino, CA) with 12-bit A/D conversion and acquisition software written in the LabVIEW programming environment (National Instruments, Austin, TX). 25 

Before extubation animals received buprenorphine 0.01 mg/kg intramuscularly and a nonsteroidal antiinflammatory drug (flunixin 2 mg/kg intramuscularly) for analgesia. During the night following surgery, the animals received buprenorphine 0.01 mg/kg intramuscularly at 8 and 16 h after awaking, and flunixin 2 mg/kg intramuscularly at 12 h after awaking. All animals received total parenteral nutrition (Intralipid 20%; Fresenius, Homburg, Germany) 100 ml, during the first hours after awaking.

Twenty-four, 48, and 72 h after the ischemic insult, the neurologic status of the animals was assessed by an observer blinded to the surgical intervention. After stimulation of the animals with the placement of food rewards several feet away from their resting location, neurologic function was evaluated according to a modified Tarlov score (five-point scale of motor function) 26: 0 = paraplegic with no lower extremity function; 1 = poor lower extremity function, weak antigravity movement only; 2 = good movement, not able to stand; 3 = able to stand, not to walk; 4 = normal motor function.

After final scoring of neurologic function at 72 h, the animals were premedicated (ketamine 15 mg/kg intramuscularly) and anesthetized with ketamine (15 mg · kg⁻¹· h⁻¹), sufentanil (5 μg · kg⁻¹· h⁻¹), and clonidine (1 μg · kg⁻¹· h⁻¹), and 2.0% isoflurane in a mixture of 50% O²in air. After administration of heparin, animals were killed with pentobarbital (100 mg intravenously) and KCL (10 ml intravenously), and the vertebral column containing the vertebral bodies L2-S1 was perfusion fixated with formalin 3.6%. The lumbosacral segment of the spinal cord was removed en bloc  and immersed in formalin for at least 10 days. The whole lumbosacral portion of the spinal cord was sampled systematically. 27Twelve equidistant transverse slices (1 mm thick) were dissected and embedded in paraffin. From each paraffin block, randomly selected 4-μm thick sections were cut and stained with hematoxylin and eosin. In the resulting series of sections, the spinal cord segment that gives rise to the lumbosacral plexus (L4–L6 and S1–S4) in pig 28was located between the 5th and the 11th section.

At a low magnification, all of the sections were digitized and the areas of total gray matter and infarcted gray matter were measured interactively using image analysis software (Qwin; Leica, Cambridge, UK) by two observers who were blinded to the experimental results. The areas (millimeters squared) were then integrated with the known distance between each transverse level to provide an estimate of the infarction volume of the spinal cord. In each animal, the extent of infarction was expressed as the percentage of necrotic tissue of the total gray matter volume. To further specify the localization of infarctions, gray matter area was separated into dorsal, intermediate, and ventral zones by dividing the dorsoventral axis of gray matter into three equal parts.

To quantify selective necrosis, eosinophilic neurons were counted in every section of the spinal cord using light microscopy (Leica). Individual counts were added to give an aggregate of eosinophilic neurons for all 12 sections. The effective magnification was 100×.

The total number of apparently viable ventral horn (α) motoneurons was determined in each section. Morphologic viability was defined according to the following criteria: fine granular cytoplasm with basophilic stippling (presence of Nissl substance), prominent nucleoli, and a soma diameter of 30–60 μm. 29Results were expressed as aggregates of 12 counts for each animal, one count being the total number of motoneurons for one section.

Power analysis was used to calculate the minimal group size that allowed for detection of significant differences in neurologic outcome between the two groups. We wished to have sufficient power (1-β= 0.8, α= 0.05) to be able to detect a reduction of the incidence of paraplegia from 80% in the permanently clamped group to 0% in the intervention group (transient tcMEP reduction). This required a group size of 5 in the intervention group (group A) and 10 in the permanently clamped group (group B). We assumed that the expected difference in neurologic events would agree with a relevant difference in duration of tcMEP loss between the groups. This would allow for a reliable determination of the predictive value of short and prolonged tcMEP loss for postoperative neurologic function. Because the degree of postoperative neurologic injury was the dependent variable in this study, the expected outcome in the intervention group (100% normal motor function) was less informative. Therefore, the number of animals in the intervention group was limited, resulting in unequal group sizes.

Arterial blood pressure, blood gases, weight, and temperatures are expressed as means ± SD. Neurologic scores, tcMEP amplitude, infarction volumes, and neuron counts are presented as medians and interquartile ranges. The physiologic variables were analyzed with a one-way analysis of variance, and when significant differences were identified, Student t  tests for intergroup comparisons with appropriate correction for multiple comparisons were carried out. Neurologic scores, infarction volumes, and neuron counts were analyzed with the use of a rank-sum statistical test (Mann-Whitney U test). A P  value < 0.05 was considered significant. Sensitivity (true positive/true positive + false negative), specificity (true negative/true negative + false positive), positive predictive values (true positive/true positive + false positive), and negative predictive values (true negative/true negative + false negative) of different reduction times of tcMEP for postoperative spinal cord injury were calculated.

There were no differences in animal weight between group A (40 ± 4 kg) and group B (41 ± 6 kg). Table 1summarizes the hemodynamic and blood gas data before the onset of arterial clamping and rectal temperature after placement of the clamps. No differences in p  H, Pao², Paco², hemoglobin, hematocrit, and rectal temperature were observed between the groups. Mean arterial blood pressure during and after tcMEP reduction was the same for animals in group A (82 ± 3 mmHg) and group B (83 ± 5 mmHg). Six lumbar SAs and the medial sacral artery were identified. Complete loss of the hind-limb tcMEP signal was established after clamping of 7 ± 1 (SD) arteries during sequential clamping. In group A, motor transmission recovered after declamping of 2–4 arteries following the initial tcMEP loss. The viscera were repositioned after termination of tcMEP recording to avoid the possible influence of increased circulating volume and subsequent reduction of the intensity of the induced ischemia.

Reproducible tcMEP were recorded in all animals. The median amplitude before ischemic manipulations was 1,934 μV (interquartile range: 1,498–2,727 μV). In one animal (group A), tcMEP could not be reduced to below 25% of baseline with the maximum numbers of SAs clamped. Staged aortic clamping was used to determine if all arteries were found. Postmortem observation showed one segmental artery that was not identified during the first surgical procedure. Because repeated aortic clamping was used, this animal was not included in the analysis. The time interval between the start of segmental artery clamping and reduction of tcMEP to below 25% of baseline was the same in group A (median, 16 min; interquartile range, 13–23 min) and group B (median, 19 min; interquartile range, 12–21 min). The subsequent interval, between the initial reduction of tcMEP to 25% of baseline and complete loss of tcMEP waveform, was not different in group A (median, 4 min; interquartile range, 3–5 min) and group B (median, 2 min; interquartile range, 1–5 min).

Figure 2shows the tcMEP for every successive segmental artery clamping stage for both groups. In group A, the declamping of segmental arteries after an initial reduction of tcMEP amplitude resulted in prompt recovery (within 10 min) of tcMEP signals to baseline values in three animals. No spinal cord infarction was present in these animals, and motor function was normal at 72 h after tcMEP reduction. In one animal of group A, tcMEP reduction was more prolonged (animal No. 3, tcMEP < 25%: 61 min, tcMEP absence: 43 min). This animal showed minimal infarction of the lumbosacral spinal cord and had a normal motor function at 72 h after tcMEP reduction. In seven (70%) animals in group B, the tcMEP waveform was rapidly abolished and remained absent. These animals had massive infarction of the spinal cord segments giving rise to the lumbosacral plexus, and were all paraplegic 72 h after the initial tcMEP loss. In three animals (30%) of group B, tcMEP returned to values above 25% of baseline after the primary decrease within 8, 10, and 42 min, despite the permanent occlusion of all lumbar SAs. Their total infarction volumes were 9.6%, 31.3%, and 95.1%, respectively, and their Tarlov scores were 4, 4, and 3, respectively. No additional SAs were found during postmortem observations at 72 h after tcMEP reduction.

The period of tcMEP reduction, neurologic score, and infarction volumes for individual animals are shown in table 2. Both in groups A and B, reduction of tcMEP to below 25% of baseline was a continuous period. Continuous absence of tcMEP waveform occurred in all animals except one in both groups (animal No. 3 in group A and animal No. 6 in group B). In these animals, three short periods of waveform absence were observed, the summation of which is given in table 2. Neuronal counts for three different periods of tcMEP reduction are shown in figure 3. The total number of viable motoneurons and eosinophilic neurons was not different among these groups. In animals with massive infarction of the spinal cord, individual neurons could often not be identified in the penumbral region of the lesion. In figure 4, the relation is shown between the time of tcMEP reduction and neurologic function at 72 h after the initial loss of tcMEP amplitude.

To determine the diagnostic characteristics of intraoperative tcMEP reduction for postoperative spinal cord damage, animals were regrouped according to their tcMEP reduction times. Four different periods of tcMEP reduction were defined, and the respective diagnostic characteristics for paraplegia and spinal cord infarction are shown in table 3. The positive and negative predictive values of fast recovery of tcMEP signals (within 10 min) for normal postoperative motor function were 1.00 and 0.64, respectively. The positive and negative predictive values of fast recovery of tcMEP for absence of spinal cord infarction were 1.00 and 1.00, respectively.

In the present study, intraoperative absence of tcMEP signals for more than 1 h consistently resulted in paraplegia and extensive spinal cord infarction 72 h later, while SCI followed by prompt restoration of tcMEP signals corresponded with normal histopathology and normal motor function. The data indicate that tcMEPs predict neurologic outcome after short and permanent intraoperative loss of the responses.

The present results are in agreement with observations following aortic occlusion in dogs, where loss of intraoperative spinal tcMEP correlated with worse neurologic outcome 24 h after reperfusion. 30The authors concluded that neurogenic motor evoked potentials were too sensitive to use as an indicator of spinal cord damage, because the responses disappeared within 1 min after the onset of SCI. However, to reduce adverse outcome, it is imperative that an intraoperative spinal cord monitoring technique detects ischemia before the occurrence of irreversible neuronal damage. The present data indicate that prolonged loss of myogenic tcMEPs predicts postoperative neurologic dysfunction, and that detection of acute SCI with tcMEPs before irreversible damage has occurred is feasible. In a porcine model of SCI, 10-min reduction of spinal cord evoked potentials recorded from the lumbar spine, followed by reperfusion, was associated with paraplegia in 2 of 4 animals at 3 h after reperfusion. 31In contrast, recovery of myogenic tcMEP within 10 min was associated with normal motor function in the present study. Our data support the hypothesis that myogenic tcMEPs provide a more sensitive and clinically relevant assessment of both ischemia and irreversible damage of the spinal cord in comparison with spinal cord evoked potentials. In a previous study we used tcMEP to assess adequacy of SCBF when SAs were selectively perfused during aortic cross-clamping. 32In all animals (n = 5) where SAs were not perfused, tcMEPs were permanently lost and paraplegia was observed 72 h later. Although a good correlation between intraoperative loss of tcMEP amplitude and postoperative neurologic deficit was present, that study was not designed to prospectively evaluate the validity of tcMEP criteria for the prediction of neurologic outcome.

Because the consequences of iatrogenic paraplegia are so devastating, spinal cord monitoring techniques should have a high sensitivity for the detection of intraoperative spinal cord dysfunction, in order to allow prompt intervention. In addition, false-negative monitoring results should be absent. For the prediction of adverse outcome, tcMEP can be regarded as a repetitive application of a diagnostic test, so that the prognostic power of a sequence of positive monitoring results is important. Loss of neurogenic motor evoked potentials recorded in dogs that were subjected to permanent segmental artery ligation predicted postoperative neurologic injury in 67% of the cases. 33Unfortunately, the difference in evoked potentials recording site and the extensive collateral circulation in dogs as compared to pig and humans 34hampers the comparison with the present data.

In the current study, a high positive predictive value and sensitivity of tcMEP loss for postoperative paraplegia were observed after loss of the responses for more than 60 min, while 30 min of tcMEP loss was already highly predictive of spinal cord infarction involving more than 20% of the total gray matter volume. The total number of paraplegic animals was less than expected, and one animal was excluded from the analysis, resulting in a smaller-than-proposed group size in the intervention group. However, it is unlikely that this influenced the predictive power of the present data because the overall incidence of paraplegia remained constant.

In some animals in group B (animal Nos. 1, 5, and 6;fig. 2), tcMEP did not remain absent during the entire study period, but gradually recovered. These animals had less extensive histopathologic damage and better neurologic score at 72 h after the initial tcMEP reduction. One possible explanation is the recruitment of collateral blood vessels that partially restored SCBF to a level sufficient for recovery of tcMEP signals. The postoperative motor score and histopathologic damage in these animals was similar to that observed after fast restoration of tcMEP signals in the animals of group A, which underwent reperfusion of the spinal cord through declamping of the most cranial lumbar SAs. In general, a good correlation was observed between histopathologic damage and neurologic function. However, in two animals (animal No. 3 in group A, No. 1 in group B) a similar intermediary duration of tcMEP loss resulted in normal motor function in both animals, while histopathologic damage was less in the animal of group A. A possible explanation for this discrepancy is the higher number of permanently clamped segmental arteries that was present in the animal in group B (seven compared to five in the animal of group A). The greater number of occluded SAs in the animal of group B might have resulted in a more compromised SCBF and subsequent more extensive neuronal loss during the postoperative period.

We reduced SCBF by sequential clamping of segmental arteries. Compared to aortic cross-clamping, this model minimizes the risk of false-positive responses due to ischemia of the peripheral nerve and muscles and avoids the major cardiovascular changes associated with aortic occlusion and the subsequent reperfusion period. The spinal cord blood supply of the pig resembles that of humans because it comprises a plurisegmental artery supply of a continuous anterior spinal artery. 35,36The vulnerable region of the spinal cord with regard to SCBF is located in the lower thoracic region, facilitating flow in the lumbar region, because of a large decrease in diameter of the anterior spinal artery in the cranial direction and the presence of extensive collateral circulation in the lumbar spinal cord. 35,36In a previous study we confirmed with laser Doppler flowmetry that SCBF of the anterior horn is severely reduced after segmental artery clamping. 23However, because placement of the flow probes on the ventral aspect of the spinal cord involved extensive removal of vertebral bone, this would preclude survival of the animals, and we therefore chose not to measure SCBF in the present study.

The rationale for the 25% tcMEP amplitude decrease criterion that is used clinically to detect SCI is based on theoretical considerations and has been confirmed empirically. 37In general, the criterion for SCI during somatosensory evoked potentials monitoring is a decrease of amplitude to less than 50% of baseline or a 10% latency increase. 38The more restrictive criterion for spinal cord dysfunction during tcMEP monitoring was based on the larger amplitude variability of tcMEP signals compared to somatosensory evoked potentials. 16The cortical and spinal motoneuronal components of tcMEP are vulnerable to agents that reduce neuronal excitability, and the myogenic component of tcMEP is severely depressed by agents that possess muscle-relaxing properties. The combination of ketamine-sufentanil anesthesia in the current study is used clinically 16,39and probably depresses myogenic tcMEP less than total intravenous anesthesia, such as low-dose propofol-opioid. 40 

The present experimental data support the validity of the empirical criterion (25% tcMEP amplitude reduction) for initiating interventions to maintain spinal cord perfusion pressure during surgical procedures on the thoraco-abdominal aneurysm or the spine.

The present observations in this porcine model demonstrate that prolonged reduction of intraoperative tcMEP amplitude predicts spinal cord injury following SCI. In contrast, immediate restoration of SCBF after tcMEP amplitude loss results in full tcMEP recovery and prevents spinal cord neuronal damage even in the presence of permanent occlusion of several segmental arteries. Our finding supports the concept that prompt restoration of spinal cord perfusion after disappearance of the tcMEP signal may prevent neurologic deficit.

The authors thank Marjolein Porsius, Research Assistant, and Heiko Botman, M.D. (Department of Anesthesiology), for monitoring of tcMEP; Marloes Klein, Biotechnician, and Henk de Wit, Biotechnician (Department of Experimental Surgery) for surgical assistance; Henk van Veen, Research Assistant, and Jan van der Marle, Ph.D. (Department of Microscopic Research), for image analysis assistance (all at the Academic Medical Center, University of Amsterdam, The Netherlands).