Transcranial motor-evoked potentials (TcMEPs) monitor spinal cord motor tract integrity. Using a swine model, the authors studied the effects of vasodilatory hypotension, hemorrhage, and various resuscitation efforts on TcMEP responses.
Twelve pigs were anesthetized with constant infusions of propofol, ketamine, and fentanyl. Animals were incrementally hemorrhaged, until bilateral tibialis anterior TcMEP amplitude decreased to less than 40% of baseline or until 50% of the blood volume was removed. Mean arterial pressure (MAP), cardiac output (CO), and oxygen delivery (DO2) were examined. Resuscitation with phenylephrine, epinephrine, and colloid were evaluated. In seven animals, vasodilatory hypotension was examined. Paired comparisons and multivariate analysis were performed.
Hemorrhage significantly reduced (as a percentage of baseline, mean ± SD) TcMEPs (left, 33 ± 29%; right, 26 ± 21%), MAP (60 ± 17%), CO (49 ± 12%), and DO2 (43 ± 13%), P value less than 0.001 for all. Vasodilation reduced MAP comparably, but TcMEPs, CO, and DO2, were not significantly lowered. After hemorrhage, restoration of MAP with phenylephrine did not improve TcMEPs, CO, or DO2, but similar restoration of MAP with epinephrine restored (to percentage of baseline) TcMEPs (59 ± 40%), and significantly increased CO (81 ± 17%) and DO2 (72 ± 19%) compared with both hemorrhage and phenylephrine, P value less than 0.05 for all. Resuscitation with colloid did not improve TcMEPs. Multivariate analysis revealed that changes in TcMEPs were more closely associated with changes in CO and DO2 as compared with MAP.
Hypotension from hemorrhage, but not vasodilation, is associated with a decrease in TcMEP amplitude. After hemorrhage, restoration of TcMEPs with epinephrine but not phenylephrine indicates that CO and DO2 affect TcMEPs more than MAP. Monitoring CO may be beneficial in major spine surgery when using TcMEP monitoring.
Transcranial motor-evoked potentials monitor spinal cord function, but are also reduced by hypotension, complicating interpretation
The effects of various hemodynamic manipulations on this monitor were assessed in a swine model
Treatment of hemorrhage-induced hypotension and reductions in potential amplitude was less effective with phenylephrine than with epinephrine, which also increased cardiac output and oxygen delivery
This animal model suggests that monitoring cardiac output facilitates treatment of hemorrhage-induced reductions in motor-evoked potentials in spine surgery
TRANSCRANIAL motor-evoked potentials (TcMEPs) are used to monitor the integrity of the motor tracts in the spinal cord during spine surgery.1–3 Acute reductions of the TcMEP amplitudes compared with baseline (e.g., by >50% or 80% from baseline) is a common criterion used to indicate a possible intraoperative neural injury.4,5 However, reductions in amplitude and increased variability among TcMEP stimulation trials are observed during periods of intraoperative hemorrhage.6,7 These amplitude reductions may exceed established warning criteria, making it difficult to distinguish them from a surgically induced neurological injury. Many authors believe that it is a decrease in systemic blood pressure that produces these amplitude reductions seen during hemorrhage. Thus, augmenting blood pressure toward baseline levels is considered to be the most essential intervention for restoring TcMEPs during hemorrhagic hypotension.8–11 However, it is not clear that hypotension is the principal or only hemodynamic aberration responsible for TcMEP amplitude changes. Other hemodynamic factors are also affected by hemorrhage, such as cardiac output (CO), hematocrit, intravascular volume, and oxygen delivery. Correcting these variables may be more beneficial than increasing blood pressure at restoring TcMEPs during periods of hemodynamic instability.
Given the need to treat hemorrhage and to restore hemodynamic homeostasis quickly, it is not ethically feasible to perform a clinical human study in which selective manipulation of different hemodynamic variables are evaluated for their effects on TcMEPs. To our knowledge, no animal model exists that consistently produces TcMEP changes due to hemodynamic variation. Therefore, using our previously described porcine model for obtaining TcMEPs,12,13 we initially sought to establish a clinically relevant model of hemodynamically induced TcMEP changes. Next, we selectively altered several hemodynamic variables, using vasodilation, hemorrhage, or resuscitation with vasoactive agents or fluids. Through analysis of the data from these hemodynamic manipulations, we sought to determine which factors were most closely linked to changes in the TcMEP responses, based on our hypothesis that CO more strongly influences TcMEP amplitude than blood pressure.
Materials and Methods
Anesthesia and Hemodynamic Monitoring
The University of California, San Francisco Institutional Animal Care and Use Committee reviewed and approved the experimental design and this protocol. Adolescent female Landrace pigs (n = 12) weighing 48–52 kg were studied. Each animal was initially anesthetized with intramuscular 4.4 mg/kg tiletamin/zolazepam (Telazol®; Wyeth, Madison, NJ), 2.2 mg/kg xylazine, and 0.04 mg/kg atropine, followed by 1–3% of inhaled isoflurane. Endotracheal intubation was performed, and mechanical ventilation was initiated. After confirmation of an appropriate anesthetic depth, the animal was placed supine and we performed a surgical cut-down of the right neck to expose the jugular vein and carotid artery. We placed a 5-French arterial catheter directed proximally into the carotid artery. Central venous access was obtained using the same incision to gain exposure of the internal jugular vein. An 8-French sheath introducer was placed and sutured into the internal jugular vein. The neck wound was closed with the ports of the 8-French introducer externalized. A 4-lumen pulmonary artery catheter was then floated into the pulmonary artery until a clear pulmonary arterial pressure waveform was achieved and the catheter could yield a pulmonary capillary wedge pressure. Systemic mean arterial pressure (MAP), pulmonary artery pressure, central venous pressure, and heart rate were monitored continuously. CO based on thermodilution and pulmonary capillary wedge pressure were measured intermittently. At designated data points, we obtained samples for measuring arterial blood gas, hemoglobin, and hematocrit. These data were used to calculate oxygen delivery (DO2), using the equation:
The pig was then placed prone on soft bolsters to support the chest and hindquarters, minimizing abdominal compression to facilitate ventilation and to avoid compromising venous return.
Isoflurane was discontinued and anesthesia was maintained using a total intravenous anesthetic consisting of infusions of propofol (100–125 µg·kg−1·min−1), ketamine (10 µg·kg−1·min−1), and fentanyl (2 µg·kg−1·h−1). The dose of propofol was adjusted to achieve “adequate” anesthetic depth as determined by a lack of jaw and extremity tone as well as no response to painful stimuli. Once the hemorrhage protocol commenced, all anesthetic infusions were maintained at a constant rate.
Electrophysiological Recording and Stimulation Protocols
The technique we used for obtaining TcMEPs in pigs is described in previous articles.12–14 We placed needle electrodes (Medtronic-Xomed, Rochester, MN) into the tibialis anterior (TA) muscle of each hindleg, and cranial screw-electrodes for electrical stimulation. TcMEP responses were deemed acceptable if stimulation yielded reproducible peak-to-peak amplitudes of at least 500 µV in both TAs. Before hemorrhage, we performed multiple TcMEP trials to establish the degree of trial-to-trial variability. All TcMEP responses were saved in electronic files and analyzed (e.g., measuring peak-to-peak amplitude) on a separate day from the procedure by two independent investigators blinded to the other physiologic data obtained for that animal.
Blood was systematically removed from the arterial cannula as described below, until we observed decreased TcMEP amplitude in either TA to less than 40% of the baseline value or until we removed 50% of the animal’s total blood volume. Total blood volume was estimated as 69 ml/kg.15 Ten percent increments of the calculated total blood volume were removed over 21 min and stored in blood-storage bags. No intravenous fluid was given during hemorrhage. TcMEPs were elicited every 3 min during the period of blood removal; CO and pulmonary capillary wedge pressure were recorded when half of this volume was removed (at 10.5 min). After 21 min and completion of the 10% blood volume removal, we enacted a 5-min equilibration period. At the end of this period, we recorded all hemodynamic values, obtained blood samples to measure arterial blood gas and hematocrit, and recorded TcMEP responses. This process was repeated in additional 10% increments until one of the previously described endpoints was achieved.
If TcMEP responses in the TA did not decrease to less than 40% of baseline after 50% of the blood volume was removed, the study was terminated. Once the TcMEP responses decreased less than 40% of baseline, further blood removal was stopped and an infusion of either intravenous epinephrine (0–2 µg·kg−1·min−1) or phenylephrine (0–200 µg/min) was started. The infusion dose was increased incrementally every 5 min until: (1) the TcMEPs recovered to greater than 60% of baseline amplitude; (2) the MAP reached 120% of baseline; or (3) the predetermined maximal dose was achieved. Hemodynamic and TcMEP data were recorded at each infusion step change. The infusion was then stopped and the hemodynamic and TcMEP values were allowed to reach a new baseline. After physiologic equilibrium, the other agent (either phenylephrine or epinephrine) was infused. The order of infusion was randomized. After completing the second vasopressor, the animal was infused with intravenous colloid solution (Hextend®; Hospira, Inc., Lake Forest, IL) and hemodynamic, laboratory, and TcMEP values were recorded. Hextend was given in 10% of total blood volume increments each over 20 min, up to a maximum of 1,500 ml. All animals were then euthanized with an overdose of sodium pentobarbital followed by bilateral thoracotomy per institutional protocol.
In seven animals, we infused sodium nitroprusside (0–20 µg·kg−1·min−1) before hemorrhage. The dose was increased incrementally until: (1) the TcMEP amplitude decreased to less than 40% of baseline; (2) the MAP decreased to less than 40% of baseline; or (3) we achieved the predetermined maximum dose. After completion of this protocol, the drug was stopped and a 30-min recovery period was used to establish new baseline hemodynamic and TcMEP values before hemorrhage.
Data and Statistical Analysis
Before initiation of this study, there were no relevant data to use for a power analysis estimating the appropriate number of pigs to study. Consequently, we performed a power analysis based purely on an estimate that our model could produce a 60% decrease in TcMEP amplitude. On the basis of this assumption, we performed a power analysis, with a two-sided α of 0.05 and a power of 80% to detect a 60% difference between TcMEP amplitude at baseline and at maximal hemorrhage. That analysis indicated that data from 10 pigs would be sufficient. We studied 12 pigs in case there were unexpected experimental difficulties. After the first five animals, we confirmed that hemorrhage did in fact produce the expected drop in TcMEP amplitude, and that therefore the estimated sample size would be sufficient.
The effect of hemorrhage on TcMEP amplitudes was analyzed using the amplitude as a “percent of baseline” for each animal. Linear regression versus the percentage of blood volume removed was performed, accounting for repeated measures. Linearity was confirmed by testing the statistical significance of an X2 term (polynomial). This analysis was performed separately for the left and right sides, then together to test for any difference between the two sides (“side, right or left” entered as an additional variable in the linear model).
TcMEP amplitudes and all physiological variables (MAP, CO, DO2, hematocrit) were compared at baseline and maximal hemorrhage using paired t tests. Separately, physiological variables were compared before and after sodium nitroprusside for the seven animals where this experiment was performed. TcMEP amplitudes and all physiological variables were compared among baseline, hemorrhage, and the three resuscitation strategies (epinephrine, phenylephrine, and volume) using repeated-measures ANOVA. Tukey–Kramer honestly significant difference was used for multiple comparisons.
Multivariate analysis was used to examine the effect of different physiological factors (MAP, CO, DO2, hematocrit, volume resuscitation) on TcMEP amplitude. Data from all the experiments were combined, but included only the data at the beginning and end of each treatment. The right and left TA data were analyzed separately. We accounted for repeated measures, using animal no. as a random effect in the model. Models were developed starting with a single physiological variable, then adding other variables. Colinearity among the variables in the model is reported as correlation coefficients and P values.
Data were analyzed using JMP 10.0 (SAS Institute, Cary, NC). All statistical tests were two-tailed. Data are presented as mean ± SD or median (interquartile range) unless otherwise specified. Statistical significance was accepted at P value less than 0.05 for all tests.
All 12 animals survived until completion of the hemorrhage protocol. TcMEPs were recorded from bilateral TA in all animals, although in one pig we could only measure adequate baseline TA amplitudes from the left side, yielding a total of 23 TA myotomes from 12 pigs. Examples of TcMEP tracings for a single-animal experiment are shown in figure 1. Baseline hemodynamic and TcMEP values for each animal are shown in appendix 1.
Data from hemorrhage for each animal are shown in appendix 2 and figure 2A; combined data from all pigs are shown in figure 2B and tables 1 and 2. The final amount of blood removed from each animal ranged from 20 to 50% of the total blood volume. TcMEP amplitude decreased significantly and linearly (P < 0.001) during hemorrhage in all TA myotomes (fig. 2A). At maximal hemorrhage, TcMEP amplitude fell to an average of 32.9 ± 29.2% of baseline on the left and 26.2 ± 21.3% of baseline on the right (both P < 0.001 vs. baseline). The rate and degree of amplitude decline was symmetric; there were no statistically significant differences between left and right TA TcMEP amplitudes at any level of hemorrhage (fig. 2B; P = 0.93). In pigs no. 1 and no. 8, TcMEP failed to decrease to 40% of baseline in either TA, so the study was terminated for these two pigs at a predetermined maximal hemorrhage of 50% of total blood volume, with no vasopressors or inotropes given.
MAP and CO decreased substantially in all animals at maximum hemorrhage, to 60.4 ± 17.1% and 48.8 ± 12.2% of baseline, respectively (both P < 0.001). The decrease in hematocrit during hemorrhage was small but statistically significant, falling to only 86.4 ± 10.1% of baseline at maximum hemorrhage (P = 0.0013). Calculated DO2 decreased the most during hemorrhage, to 42.5 ± 13.1% of baseline (P < 0.001).
Data from vasoactive and volume infusions are shown in tables 1 and 2, and figures 3, 4 and 5. Sodium nitroprusside infusion was performed in seven pigs (fig. 3). MAP decreased to 58.1 ± 10.9% of baseline (P < 0.001), similar to the decrease seen during maximal hemorrhage. CO and DO2 were not significantly reduced. TcMEP amplitude was minimally affected by sodium nitroprusside (P = 0.082 and 0.11, left and right, respectively).
After hemorrhage, both phenylephrine and epinephrine infusions were given to the 10 pigs whose TcMEPs had decreased to less than 40% of baseline (tables 1 and 2; figs. 4 and 5). Phenylephrine restored MAP to 90.2 ± 25.9% of the original baseline value (P = 0.12 vs. baseline; P < 0.001 vs. the nadir MAP from hemorrhage). Phenylephrine led to a small change in CO (return to 61.9 ± 14.7% of original baseline, net increase of 29.3 ± 33.8% [P = 0.0096]) and DO2 (return to 55.0 ± 16.9% of baseline, a net increase of 34.6 ± 37.9% [P = 0.0077]). Epinephrine increased MAP to a similar degree as phenylephrine (P = 0.80 vs. phenylephrine). However, epinephrine augmented CO and DO2 to 80.8 ± 17.2% and 72.1 ± 19.5% of baseline (P = 0.011 and 0.0029, respectively vs. baseline; both P < 0.01 vs. phenylephrine). Phenylephrine infusion led to no significant improvement in TcMEP responses, but epinephrine restored TcMEP amplitudes to a statistically significantly higher level, 58.3 ± 39.3% of baseline (fig. 5; P < 0.05 vs. phenylephrine).
Hextend was administered to 10 pigs (tables 1 and 2 and fig. 4). One pig (no. 6) sustained a cardiac injury during hextend infusion, with changes noted on its electrocardiogram and a falling CO; volume data for this animal were excluded. Administration of colloid solution restored MAP to 86.9 ± 24.0% of the original baseline, an increase of 67.5 ± 18.6%. Hematocrit was reduced by the hemodilution to 13.4 ± 2.1. CO increased from 3.2 ± 1.2 to 9.2 ± 2.2 l/min. This produced a 68.3 ± 44.7% increase in calculated DO2, despite the reduced hematocrit. All hemodynamic changes after colloid administration listed above were statistically significant at P value less than 0.001. TcMEP amplitude increased significantly on the left (P = 0.035) but not on the right (P = 0.064).
Multivariate analysis, combining data from hemorrhage, nitroprusside, phenylephrine, and epinephrine revealed that decreases in TcMEPs were more closely associated with changes to CO and DO2 than to the MAP. Models beginning with a single variable (MAP, CO, or DO2 alone) showed statistical significance (P < 0.001). CO and DO2 correlated too highly with each other (r = 0.95) to enter both into a statistical model. Models based on either CO or DO2 continued to show statistical significance for CO or DO2 (P < 0.001) even after adding MAP. In the model based on DO2, MAP was not statistically significant for either the left or right TA (P = 0.53 and 0.068, respectively). For the model based on CO, MAP was also statistically significant for neither the left nor right TA (P = 0.30 and 0.074, respectively). CO and MAP (r = 0.69; P < 0.001), and DO2 and MAP (r = 0.57; P < 0.001), both remained highly correlated despite pharmacological manipulations separating these variables.
A model including volume replacement data found that DO2 had the lower P value (P < 0.001), although both CO (P = 0.0035) and MAP (P = 0.005) were also statistically significant.
MacDonald et al.16 were the first to describe the phenomenon of “fading” TcMEPs during scoliosis surgery. Prolonged exposure to anesthetic agents may lead to a gradual decline in TcMEP amplitude.17 In contrast, the reductions of TcMEPs due to hemorrhage are acute and may appear indistinguishable from actual mechanical trauma.5,7,8 Developing an animal model is essential to determine which hemodynamic factors best correlate with TcMEP changes. This model can also provide the means to determine where along the motor pathway the TcMEPs are disrupted: the motor cortex, the descending motor tracts in the spinal cord, the synapses in the ventral horn, or even at the level of the peripheral muscles. Most important, the model will allow a better understanding of which treatments can best resolve hemorrhage-induced TcMEP amplitude reduction.
Acute hypovolemic hemorrhage produced consistent and dramatic decreases in TcMEP amplitude (>60%) in 17 of the 23 myotomes monitored. Our choice of a hemorrhage model was based on clinical observation that TcMEPs are often depressed during the most complex deformity-correction procedures (e.g., pedicle subtraction osteotomies), in which persistent and massive hemorrhage is common. In human spine surgery, resuscitation is continuous; patients rarely experience decreases to circulating blood volume of this degree. We used extreme hypovolemic hemorrhage in our model in order to produce large TcMEP amplitude decreases in healthy animals. We have demonstrated that systemically induced changes in MEP amplitudes are fairly uniform and symmetrical; the differences in the decrease of the left TA compared with right TA during hemorrhage were negligible. Clinical identification of this pattern may help distinguish hemodynamically induced TcMEP alterations from TcMEP changes caused by actual iatrogenic injury.
The most intriguing finding in this study was that MAP was less important than both CO and DO2 in correlating with changes to TcMEPs. We base this conclusion on two clear findings. Hypotension induced by sodium nitroprusside did not notably reduce TcMEP amplitude. Sodium nitroprusside reduced CO and DO2 much less than the MAP. Moreover, when TcMEP amplitudes had faded after hemorrhage, epinephrine was much more useful in rapidly restoring TA amplitude than phenylephrine. This is likely because although both epinephrine and phenylephrine markedly improved MAP, epinephrine increased CO and DO2 much more than phenylephrine did. These results suggest that selecting pharmacologic and resuscitative interventions that increase the CO or DO2 may be superior to merely increasing the MAP with respect to preserving TcMEPs during hemorrhage. This finding seems to contradict claims that increasing MAP is the most essential step to recover reduced TcMEP responses. In the few published reports noted during a search of the literature, the resuscitative steps used to increase MAP likely also augmented CO.8,10,11 In a case report, Othman et al.11 described how lower extremity somatosensory evoked potentials and neurogenic MEPs were lost during a period of worsening hypotension. These electrophysiological signals recovered as MAP was restored. However, the agents used to increase MAP were the inotropes epinephrine and dopamine. CO was not measured during this procedure, but one expects CO was augmented by the interventions used in this patient. Schwartz et al.8 described the importance of normalizing intravascular volume status for maintaining the quality of intraoperative electrophysiology during deformity-correction surgery. Treatment of hypovolemia with appropriate fluids and blood products strongly affect the CO as well as MAP.
Analysis of the resuscitation data in this model revealed that DO2 is the hemodynamic variable that correlates best with TcMEP responses. Our administration of colloid augmented CO greatly, but at the expense of reducing hematocrit, resulting in minimal improvement to the low DO2. We observed only small augmentation of the TcMEP amplitudes, suggesting high CO is less effective if oxygen-carrying capacity of the blood is so diminished as to limit the rise of DO2. Thus, fluids may not be as effective as interventions that are more likely to increase DO2, such as transfusing blood or administering inotropic agents. Augmenting DO2 may better restore oxygen supplies to the brain, spinal cord, or musculature than by solely increasing the MAP. Myburgh et al.18 showed that infusion of epinephrine increased cerebral blood flow in anesthetized pigs, whereas phenylephrine or norepinephrine did not. The prone position has been shown to lead to lower cardiac stroke volume and CO,19 thus increasing the likelihood that hemorrhage will lead to even lower CO and DO2 levels intraoperatively. Our data also suggest that MAP may still have relevant effects, even when accounting for DO2.
Several limitations of the current study merit discussion. We relied on a specific degree of amplitude reduction in order to detect a meaningful change in our TcMEP monitoring. Acute reductions of the TcMEP amplitudes compared with baseline (e.g., by >50% or 80% from baseline) appear to be the most common criteria used to indicate a possible intraoperative neural injury.4,5 However, some practitioners require a complete loss of TcMEP responses (e.g., “all-or-none”) to warrant an alarm situation.4 Others rely on needing a specified increase of stimulation voltage in order to maintain the same amplitude response.20
Another major limitation of this study is that we were unable to determine which parts of the neural pathway were interrupted such that TCMEPs were altered. The systemic hemodynamic changes may impact the perfusion and function of the cerebral cortex, spinal cord, spinal nerve roots/peripheral nerves, and/or muscles. Another explanation for why decreased CO affects TcMEP response is because changes in CO and blood volume affect the metabolism and volume of distribution of propofol. Propofol is often used as a hypnotic agent for spine surgery as it causes less suppression of evoked potentials. However, acute increases in the serum level of propofol during hemorrhage may also contribute to the reduced TcMEP amplitude.21,22
Although an animal model allows extreme manipulation of physiology not acceptable in humans, verification that our conclusions apply clinically will require human data. We studied healthy adolescent animals that were able to compensate for extreme hypovolemia. Complex spinal deformity cases are often performed in an older patient population with reduced cardiac reserve. Thus, smaller degrees of hemorrhage may produce comparable changes to hemodynamic parameters. Multiple treatments in the same animal may have been excessive, which may limit our ability to distinguish all the different physiological variables in their impact on evoked potentials. Also, the levels to which we reduced hematocrit may be beyond levels where physiological compensation may occur. Decreasing hemoglobin may reduce oxygen delivery less when initial hemoglobin levels are higher.23 Additionally, oxygen dissolved in the blood may provide substantial oxygen to tissues when severe anemia is present.24
The study succeeded in establishing a successful animal model of TcMEP reduction during significant rapid hemorrhage without concurrent volume replacement. We have provided some initial evidence that certain physiological factors; namely, CO and DO2, may prove to be of equal or of greater importance than MAP in affecting TcMEPs. Given these results, the addition of CO monitoring in patients undergoing complex spine surgery who are likely to hemorrhage intraoperatively could be considered. Moreover, interventions to augment CO or DO2 may prove to be more efficacious at restoring TcMEP responses than those that solely increase MAP. Further study is warranted to better understand the mechanisms and sites of disruption that produce hemodynamic fade and improve the clinical intraoperative management of patients undergoing complex spine surgery. Hence, maintaining optimal hemodynamic stability will improve the reliability of TcMEPs to detect iatrogenic spinal cord injury.