Intraoperative neuromonitoring of the spine refers to a group of tests that allow observation of the integrity of the spinal cord during surgery. The use of intraoperative neuromonitoring is associated with a decreased rate of paralysis after deformity surgery, and has been shown to have value as a predictor of neurologic outcome in cervical spine surgery and thoracic and lumbar laminectomy.1–3
The main modalities used for intraoperative neuromonitoring are somatosensory evoked potentials (SSEPs) and motor evoked potentials (MEPs). Because each test acts as a monitor for distinctly different areas of the spinal cord, they are typically used in concert during operations that place the spinal cord at risk.4,5 Intraoperative recordings of MEPs were first used in the mid-1980s as a means to assess the function of the motor tracks of the spinal cord, which are not as reliably evaluated by SSEPs. Over the past 2 decades, MEPs have become widely used during spine surgery.6 The wide use of intraoperative neuromonitoring has necessitated that anesthesiologists alter their choice of anesthetic technique to facilitate SSEP and MEP recording. Several studies have examined the influence of anesthetics on SSEPs and MEPs with contradictory results. MEPs seem in general to be more sensitive to anesthetics than SSEPs.6 Furthermore, lower extremity signals can be more difficult to obtain than upper extremity signals.7 Our experience suggests that patient factors may have an important role as well. Only a few studies have examined whether patient factors affect MEP signals. Chen et al.8 noted that age, presence of a preexisting deficit, or an intraspinal lesion in combination with a preexisting deficit is associated with difficulty in obtaining reliable MEPs. Kim et al.9 found that body mass index (BMI) and increased length of surgery are associated with the appearance of intraoperative MEP alarm criterion, but not associated with postoperative deficit.
In this study, we determined whether age, BMI, presence of diabetes and/or hypertension, and procedure type can predict failure to obtain baseline lower extremity MEP signals. We also evaluated the interaction between these factors and anesthetic technique.
Intraoperatively, only a very small percentage of patients lose signals; an even more miniscule number will have a postoperative deficit. Nuwer et al.1 studied >15,000 patients and found a 0.55% incidence of deficit after scoliosis correction. Experienced monitoring teams have half that rate. Therefore, to study neurophysiologic correlates of postdeficits is to analyze extremely rare events. Although a large study would be a worthwhile endeavor, a smaller study examining the effect of comorbid conditions on the loss of intraoperative signal is likely to be underpowered. Additionally, there is not necessarily a relationship between difficulty in obtaining baseline and loss of signal intraoperatively. Surgery, arterial blood pressure, preexisting myelopathy, and the hapless addition of random anesthetics are often to blame. Our study was not designed to examine the question of intraoperative loss of signal; the purpose of this study was to show that in the absence of surgical factors, patient comorbidities correlate with MEP monitoring difficulty, especially in the case of a less than conducive anesthetic technique.
After IRB approval, a retrospective chart review was conducted of all anesthetic records of patients who underwent spine surgery and MEP monitoring of the lower extremities from August 2001 to December 31, 2005. Patients with preexisting paralysis in the lower extremities were excluded. All patients received an anesthetic that consisted of either nitrous oxide with propofol or nitrous oxide with a volatile anesthetic (isoflurane, sevoflurane, or desflurane). Within this time period, MEPs were elicited in a uniform manner. Documentation of MEP baseline signals obtained after induction of anesthesia but before skin incision was coded on a scale of 0 to 2, representing the number of lower extremities in which a baseline signal was documented. All monitoring was performed by staff neurophysiologists, confirmed by the supervising neurophysiologist, and recorded in a database in Excel spreadsheet form. Staff neurophysiologists placed neuromonitoring leads, and obtained baseline signals as described below. The onsite supervising neurophysiologist (DJW) reviewed all tracings, which were then documented with comments in the database at the end of the procedure. Type of monitoring requested, success obtaining baseline, and whether alarm criterion occurred during the procedure were noted. The database was queried for patients who underwent MEP monitoring during 2001 to 2005, a period during which a single type of cortical stimulator was in use. In cases in which the record noted difficulty obtaining intraoperative neuromonitoring or the record was ambiguous, archived raw data were analyzed by 2 experts. Raw data were not examined for all patients because it was assiduously analyzed before entry into the database. Failure to obtain MEP signals was defined as the failure to obtain lower extremity MEP signals. Lower extremity MEP signals were analyzed because they are more difficult to obtain than upper extremity MEP signals.
Stimulating and recording electrodes were applied after the induction of general anesthesia. Spiral-design needle electrodes (Nicolet Biomedical, Madison, WI) were placed on the scalp at C3 and C4. C4 served as the cathode and C3 as the anode to preferentially activate the right motor cortex. The cathode and anode were reversed to activate the left motor cortex. Electrical stimulation was generated by a cortical stimulator from Digitimer Model D185 (Digitimer Ltd., Hertfordshire, UK). Trains of 4 to 9 pulses with 0.1-millisecond pulse width were delivered at an interpulse interval of 2 milliseconds. Electromyogram (EMG) was recorded from pairs of subdermal needle electrodes (2.0-cm interelectrode distance) placed bilaterally in the anterior tibialis muscle in the leg or the abductor hallucis in the foot and in the abductor pollicis muscle at the base of the thumb. EMG was amplified, digitized, and stored using NeuroNet, a commercial monitor from Computational Diagnostics, Inc. (Pittsburgh, PA). Filter settings were 3 to 3000 Hz, and amplification was 500. The stimulus intensity was gradually increased from an initial level of 25 V until stable responses were elicited. The presence of motor evoked response in any lower extremity muscle was considered as the presence of a motor evoked response. Lower extremity MEP signals were deemed present if a repeatable response was clearly discernible from the background noise. In general, responses as small as 25 μV peak to peak were detectable, responses that typically were at least 4× the level of background noise. For technical reasons, latencies were not usually considered meaningful. The stimulus intensity was never higher than 450 V for the patients in this study. Free-run EMG was monitored if requested by the surgeon for the purpose of detecting spontaneous nerve root activity during surgery. Both the digitized recording and an assessment of the signal by the monitoring professional were recorded.
Age, BMI, gender, presence of hypertension and/or diabetes, type of spinal procedure, and anesthetic regimen at the time of initial baseline MEP signal attainment were collected from patients’ charts using the CompuRecord anesthesia database (Philips Electronics, Best, The Netherlands). The diagnoses of hypertension and diabetes were made by each patient's care physician and noted during the history and physical examination and confirmed with the information gathered by the anesthesiologist and reported in the CompuRecord. Baseline signals were examined after tracheal intubation but before surgical incision. Therefore, no surgical procedure had been performed at the time of the baseline; however, for the purpose of understanding the demographics of our population, the patients were grouped according to surgical procedure and the presence of tumor or scoliosis. No patient had preoperative paralysis. More detailed information regarding neuropathy or presence and grade of myelopathy was not consistently available.
A univariate analysis was performed to examine the distributions of diabetes, hypertension, anesthesia technique, planned surgical procedure, age, gender, and BMI for patients for whom baseline lower extremity MEPs were successfully obtained. The χ2 test and the 2-sample t test were used to test the associations between the MEP status and the potential risk factors. Cochran-Armitage test was used to analyze trends in BMI and age by quartile. Bivariate analysis of the data was performed to analyze a potentially synergistic deleterious effect of diabetes and hypertension, and anesthetic technique using the Breslow-Day test for homogeneity of the odd ratios. Logistic regression analysis through stepwise selection was performed to form a model of the data.
Demographic information is presented in Table 1. The majority of our sample was patients presenting for cervical spine surgery. In the univariate analysis, age, BMI, presence of hypertension, anesthetic regimen, and/or diabetes was significantly associated with failure to obtain MEP signals. Gender and procedure type were not. The results of this analysis and the associated P values are summarized in Table 2. Scheduled surgical procedure did not influence the ability to obtain baseline MEPs. Additionally, the average age of patients in whom the team was able to obtain baseline lower extremity MEP signals was significantly younger by average and by quartile. BMI was also statistically significant both by average and by quartile. The overall trend was that in heavier patients, there was an increased risk of failure to obtain MEP baseline signals.
Individual and joint effects of diabetes and hypertension, compared with patients with neither condition, were presented under each anesthetic technique (Table 3). The bivariate analysis failed to detect a synergistic effect among any pair of diabetes, hypertension, and anesthetic technique variables on MEP signal attainment. Controlling for hypertension, the adjusted odds ratio of failure to obtain MEP baseline signal for diabetes was 6.97 (95% confidence interval: 2.98–14.38). The Breslow-Day test for homogeneity of the odds ratios with and without hypertension was not significant (P = 0.46). Controlling for diabetes, the adjusted odds ratio of failure to obtain MEP baseline for hypertensives was 6.04 (95% confidence interval: 2.77–13.18). The Breslow-Day test for homogeneity of the odds ratios with and without diabetes was also not significant (P = 0.47). The results indicated that hypertension and diabetes are independent risk factors. This signifies that when both conditions coexist, their individual contribution effect on MEP signal attainment effect does not depend on the other factor. Similarly, the detrimental effect of anesthetic technique on baseline MEP signal was independent of whether the patient was diabetic or hypertensive. Therefore, when the conditions coexist, their joint effect is additive rather than synergistic. The respective hypertension- and diabetes-adjusted odds ratio for failure to obtain MEP baseline for nitrous oxide and volatile anesthetic versus nitrous oxide with IV anesthesia supplementation was 2.95 (1.41–6.15) and 4.16 (1.81–9.55), respectively.
The results of a regression analysis showed that diabetes, hypertension, and anesthetic technique were the 3 most important variables to predict failure to obtain a MEP baseline signal. As seen in the bivariate analysis, no significant 2-way interaction was detected among these 3 variables. Age, BMI, and surgical procedure type were no longer associated with failure to obtain a MEP baseline signal after considering diabetes, hypertension, and anesthetic technique.
The results of this study clearly indicate that diabetes and hypertension are the main patient risk factors associated with failure to obtain MEP signals. Age and BMI were also significant but these factors were not independent risk factors because older, more obese patients may be more likely to have diabetes and hypertension. We were not able to calculate BMI for all patients because of missing height data; however, analysis of unadjusted patient weight found the same trends. The confidence intervals were notably wide; however, the upper and lower confidence limits are the exponential of the confidence limits of the logarithm of the odds ratio. Therefore, when the odds ratio is large, the interval tends to be asymmetrical and wide.
In this study, anesthetic technique affected the ability to obtain MEP signals. Our result is similar to previous studies that showed that nitrous oxide in the presence of volatile inhaled anesthetics resulted in an increase in failure to obtain MEP baseline.10,11 In our study, all patients received nitrous oxide; the addition of volatile anesthetic was associated with a deleterious effect on baseline MEP signals.
Several studies have examined the influence of IV and inhaled anesthetics on intraoperative neuromonitoring signal attainment with contradictory results.12 Pechstein et al.10 found that distinct motor responses were recorded in 88% of patients (15 of 17) undergoing total IV anesthesia (TIVA) but in only 8% of patients (1 of 13) undergoing anesthesia with nitrous oxide as well as volatile anesthetics. Pelosi et al.11 found that MEP signals could be recorded in 97% of patients undergoing anesthesia with propofol and nitrous oxide (29 patients total) but in only 61% of patients undergoing anesthesia with isoflurane and nitrous oxide (23 patients total). However, other studies in healthy scoliosis patients have not always found a significantly detrimental effect of inhaled agents.13 Obviously, there are circumstances in which MEP monitoring can be performed when inhaled anesthetics are used. A limitation of our study is that our database did not note anesthetic levels at the time of the baseline signals; hence, our study considers only the effect of the presence of a volatile anesthetic, not its concentration.
Finally, logistic regression analysis revealed that the 3 most significant independent variables are diabetes, hypertension, and anesthetic technique. The effect of diabetes on MEP signals is not surprising because nerve conduction is inversely related to hemoglobin A1c values, potentially because poor glycemic control is associated with development of neuropathy.14 The effects of hypertension may be related to exaggerated decreases in blood flow with the usual decrements in arterial blood pressure that are associated with general anesthesia. Hypertension has also been associated with a reduction in nerve fiber conduction velocity. The influence of anesthetics on arterial blood pressure is well established and may be significant in hypertensive patients, resulting in decremental effects on intraoperative neuromonitoring. However, there may be an additional effect of chronic hypertension on the nervous system that has not previously been appreciated. All anesthetics have effects on both blood pressure and neural conduction; the question is whether some patient characteristics are associated with a greater likelihood of intraoperative neuromonitoring failure.
In 1 recent study, cutaneous sensory thresholds were higher and sensory action potential amplitudes smaller in hypertensive patients than in their normotensive counterparts.15 Sensory and motor nerve conduction velocities did not differ between groups. This suggests that hypertension may reduce the number of active sensory nerve fibers without affecting the myelin sheath. Sensory action potential amplitudes were inversely related to cutaneous sensory thresholds, suggesting that subclinical axonal neuropathy of sensory afferents may help account for perceptual deficits that characterize hypertension.15 If hypertension affects motor nerve conduction similarly, then hypertensive patients may have increased susceptibility to anesthetic technique. It is unclear, however, whether hypertensive patients have a nontreatable issue of neural conduction or whether aggressive maintenance of arterial blood pressure would eliminate this difference. It has been demonstrated that there is a threshold relationship between regional cerebral blood flow and cortical evoked responses.16 Additionally, patients with secondary Raynaud phenomenon have slower lower extremity motor nerve conduction.17
Less intuitively, local factors may contribute to acute or chronic decreases in blood flow and regional ischemia even at “safe” systemic blood pressure. For example, during spine surgery, the effects of hypotension may be aggravated by spinal distraction such that an acceptable limit of systemic hypotension cannot be determined without monitoring.18,19 Because we had not synchronized the time stamp on the computerized anesthesia record with the neuromonitoring equipment, we are not able to comment on the possibility that effectively managed blood pressure might mitigate the negative effects of inhaled anesthesia on lower limb MEP signals. Future prospective studies should be designed to evaluate this important issue.
In any case, it is not true that patients undergoing anesthesia consisting of nitrous oxide and propofol reliably have higher arterial blood pressures than patients who receive inhaled anesthetics. Hypertensive patients may have some combination of subclinical neuropathy and decreased blood flow that may predispose them to ischemia of the motor tracts. The reason for the greater sensitivity of MEP signals to ischemia is that in contrast to SSEPs, highly metabolically active motor neuron cell bodies are concentrated in the lumbar and cervical cord. Additionally, because the motor and sensory tracts are removed topographically from one another, MEPs and SSEPs may show differential sensitivity to an ischemic event. It is well established that the single anterior spinal artery can be a tenuous supplier for the motor tracts especially in the thoracic region of the spinal cord. In contrast, the paired posterior spinal arteries provide the dorsal columns with substantial perfusion.
By establishing previously undescribed clinical predictors of intraoperative neuromonitoring failures, this study suggests anesthesiologists can anticipate the patients in whom MEP monitoring may be challenging and with this knowledge create a tailored anesthetic plan. We assert that intraoperative anesthetic technique can be optimized to produce better quality monitoring and to facilitate discrimination between false positives and true positives.
An anesthetic technique compatible with consistent intraoperative neuromonitoring needs to be established well in advance of the need to obtain monitoring signals. This is important because the transition from the relatively rapid dissipation of inhaled drug to the relatively slower onset of steady-state blood levels of drug infusion may take significantly longer than several minutes when not temporally associated with an induction dose.20 When monitoring difficulties are anticipated, the anesthesiologist should select a combination of drugs with a beneficial effect profile and start their maintenance infusion around the time of induction. Examples include the addition of ketamine to a propofol infusion and use of dexmedetomidine with the accompanying decrease in propofol and opioid requirements. If it is necessary to use inhaled anesthetics (e.g., propofol shortage, patient history of adverse reaction, or expense of TIVA), then there must be clear communication with the monitoring team. A single gas, either volatile anesthetic or nitrous oxide, should be used. If baseline signal is not available, then a risk-benefit analysis should be discussed with the monitoring team and the surgeon. If the monitoring is compulsory, then the drug may need to be discontinued. If signals do not appear, the surgeon must decide whether to proceed without monitoring. In any case, a technique that is compatible with the attainment of a neuromonitoring baseline must be continued throughout the case to avoid loss of signals at a critical portion of the procedure. The increased cost of TIVA may be justified for patients with 1 or more of the risk factors identified by this study. Young and healthy patients are generally less sensitive to inhaled anesthetics. Future studies should examine MEP failure rates after optimizing anesthetic technique based on patient characteristics. A prospective trial with an anesthetic protocol and time correlation with baseline MEP signals would enable a better comparison.
SGD helped in study design and manuscript preparation; SGK helped in conduct of study and manuscript preparation; H-ML helped in data analysis and manuscript preparation; and DJW helped in study design, data collection, and manuscript preparation.
1. Nuwer MR, Dawson EG, Carlson LG, Kanim LE, Sherman JE. Somatosensory evoked potential spinal cord monitoring reduces neurologic deficits after scoliosis surgery: results of a large multicenter survey. Electroencephalogr Clin Neurophysiol 1995;96:6–11
2. Kelleher MO, Tan G, Sarjeant R, Fehlings MG. Predictive value of intraoperative neurophysiological monitoring during cervical spine surgery: a prospective analysis of 1055 consecutive patients. J Neurosurg Spine 2008;8:215–21
3. Chandanwale AS, Ramteke AA, Barhate S. Intra-operative somatosensory-evoked potential monitoring. J Orthop Surg (Hong Kong) 2008;16:277–80
4. Devlin VJ, Schwartz DM. Intraoperative neurophysiologic monitoring during spinal surgery. J Am Acad Orthop Surg 2007;15:549–60
5. Sloan TB, Janik D, Jameson L. Multimodality monitoring of the central nervous system using motor-evoked potentials. Curr Opin Anaesthesiol 2008;21:560–4
6. Deletis V, Kiprovski K, Morota N. The influence of halothane, enflurane, and isoflurane on motor evoked potentials. Neurosurgery 1993;33:173–4
7. MacDonald D. Intraoperative motor evoked potential monitoring: overview and update. J Clin Monit Comput 2006;20:347–77
8. Chen X, Sterio D, Ming X, Para DD, Butusova M, Tong T, Beric A. Success rate of motor evoked potentials for intraoperative neurophysiologic monitoring: effects of age, lesion location, and preoperative neurologic deficits. J Clin Neurophysiol 2007;24:281–5
9. Kim DH, Zaremski J, Kwon B, Jenis L, Woodard E, Bode R, Banco RJ. Risk factors for false positive transcranial motor evoked potential monitoring alerts during surgical treatment of cervical myelopathy. Spine 2007;32:3041–6
10. Pechstein U, Nadstawek J, Zentner J, Schramm J. Isoflurane plus nitrous oxide versus propofol for recording of motor evoked potentials after high frequency repetitive electrical stimulation. Electroencephalogr Clin Neurophysiol 1998;108:175–81
11. Pelosi L, Stevenson M, Hobbs GJ, Jardine A, Webb JK. Intraoperative motor evoked potentials to transcranial electrical stimulation during two anaesthetic regimens. Clin Neurophysiol 2001;112:1076–87
12. Sloan TB, Heyer EJ. Anesthesia for intraoperative neurophysiologic monitoring of the spinal cord. J Clin Neurophysiol 2002;19:430–43
13. Lo YL, Dan YF, Tan YE, Nurjannah S, Tan SB, Tan CT, Raman S. Intra-operative monitoring in scoliosis surgery with multi-pulse cortical stimuli and desflurane anesthesia. Spinal Cord 2004;42:342–5
14. Huang CC, Chen TW, Weng MC, Lee CL, Tseng HC, Huang MH. Effect of glycemic control on electrophysiologic changes of diabetic neuropathy in type 2 diabetic patients. Kaohsiung J Med Sci 2005;21:15–21
15. Edwards L, Ring C, McIntyre D, Winer JB, Martin U. Cutaneous sensibility and peripheral nerve function in patients with unmedicated essential hypertension. Psychophysiology 2008; 45:141–7
16. Sloan T. A Textbook of Neurosurgical Anesthesia with Neurosurgery and Neuroscience Perspectives. New York: McGraw Hill, 1997
17. Kostopoulos D, Rizopoulos K, Vartholomeos N. Nerve conduction velocities in the lower extremity in patients with Raynaud's phenomenon and clinical applications. J Bodyw Mov Ther 2008;12:58–66
18. Brodkey JS, Richards DE, Blasingame JP, Nulsen FE. Reversible spinal cord trauma in cats. Additive effects of direct pressure and ischemia. J Neurosurg 1972;37:591–3
19. Dolan EJ, Transfeldt EE, Tator CH, Simmons EH, Hughes KF. The effect of spinal distraction on regional spinal cord blood flow in cats. J Neurosurg 1980;53:756–64
© 2010 International Anesthesia Research Society
20. Sadean MR, Glass PSA. Pharmacokinetic–pharmacodynamic modeling in anesthesia, intensive care and pain medicine. Curr Opin Anaesthesiol 2009;22:463–8