Outcomes of Combined Somatosensory Evoked Potential, Motor Evoked Potential, and Electroencephalography Monitoring during Carotid Endarterectomy Sean D. Alcantara,1 Joseph C. Wuamett,1 John C. Lantis II,1 Sedat Ulkatan,2 Philip Bamberger,3 Donna Mendes,1 Alan Benvenisty,1 and George Todd,1 New York, New York
Background: While much has been written about multiple methods of neuromonitoring during carotid endarterectomy (CEA), there has been relatively little discussion of the use of triple monitoring via somatosensory evoked potentials (SEPs) and motor evoked potentials (MEPs) in conjunction with electroencephalography (EEG). Our objective was to evaluate the rate of detection and prevention of neurologic events by multinerve SEP, MEP, and EEG in patients undergoing CEA while under general anesthesia. Methods: A prospective study of 181 consecutive patients undergoing CEA between June 2005 and September 2010 was reviewed. Intraoperative changes, including a 50% reduction in the amplitude of SEP waveforms, loss of MEP, and/or a 50% change in EEG frequency were noted as indications for shunting. This was correlated with the actual use of intraoperative shunting and postoperative neurologic sequelae at both 24 hours and 30 days. Median and tibial nerve SEPs and MEPs were also correlated. Results: Eleven patients (6%) experienced intraoperative monitoring changes (SEP: 11/11; MEP: 6/11). Five of 11 patients with MEP/SEP changes underwent shunting, while the other 6 had normalization with the elevation of their blood pressure. Of the 11 patients that had neurophysiologic changes, 54% (6/11) were patients with symptomatic disease. No patients had significant EEG changes. The total shunt rate was 2.7% (5/181). No postoperative neurologic sequelae were noted. Conclusion: The ratio of shunting at 2.7% is equal to the lowest rates reported in the awake patient literature. Interestingly, the predicted synergy of multimodality monitoring cannot be directly attributed to an increased specificity resulting from the addition of SEP and MEP to EEG, because no patients had EEG changes. In addition, in today’s cost-conscious world of health care, our results do not justify implementing this particular technique of neuromonitoring across the boarddbut it is apparent that the combination of these 3 modalities is both safe and effective with potential applications in symptomatic patients.
INTRODUCTION We continue to believe that carotid endarterectomy (CEA) has been shown to be the most effective
Presented at the 25th Eastern Vascular Meeting, National Harbor, MD, September 22e24, 2011. 1 Division of Vascular /Endovascular Surgery, St Luke’s-Roosevelt Hospital, New York, NY. 2
Division of Neurophysiology, St Luke’s-Roosevelt Hospital, New York, NY. 3
NY.
Department of Anesthesia, St Luke’s-Roosevelt Hospital, New York,
treatment in preventing cerebrovascular events in well-selected patients1e5; however, the methods of monitoring adequate cerebral perfusion during the procedure remain idiosyncratic to the center in Correspondence to: John C. Lantis II, MD, Division of Vascular/ Endovascular Surgery, 1090 Amsterdam Avenue, Suite 7A, New York, NY 10019, USA; E-mail: [email protected] Ann Vasc Surg 2014; 28: 665–672 http://dx.doi.org/10.1016/j.avsg.2013.09.005 Ó 2014 Elsevier Inc. All rights reserved. Manuscript received: October 1, 2012; manuscript accepted: September 18, 2013; published online: February 1, 2014.
665
666 Alcantara et al.
which the CEA is performed. Most centers that practice 1 particular method report excellent outcomes. The practice of cerebral monitoring and the use of carotid shunt placement remains variable. Some recent studies have suggested that monitoring and selective shunt use during CEA decreases the risk of perioperative stroke and can be superior to routine shunt placement,6,7 while other centers report excellent results with regional anesthesia and awake monitoring.8,9 Although no absolute data are available, it would appear that electroencephalography (EEG) has been established as one of the more commonly used tools for intraoperative cerebral monitoring because of its ability to detect cerebral ischemia. Multiple groups have shown that CEA performed with routine EEG monitoring and selective shunt placement is both safe and effective.10e16 The published literature contains significant detail about multiple methods of neuromonitoring during CEA; however, there is relatively little discussion regarding the use of somatosensory evoked potentials (SEPs)dand, in particular, multinerve SEPsdin conjunction with EEG.17,18 Our objective was to evaluate the rate of detection and prevention of neurologic events by multinerve SEP, motor evoked potentials (MEPs), and EEG in patients undergoing CEA while under general anesthesia.
Annals of Vascular Surgery
Anesthesia All patients were anesthetized with total intravenous anesthesia to facilitate the requirements of SEP monitoring. Volatile anesthetic agents were avoided because they are known to interfere with SEP monitoring by decreasing amplitude and increasing latency. We also avoided nitrous oxide. Patients were ventilated with FiO2 0.4e1.0. Anesthesia was induced with intravenous midazolam 2 mg, intravenous fentanyl 250 mcg (3e5 mcg/kg), propofol 2.0e2.5 mg/kg, or etomidate 0.2e0.3 mg/kg and succinylcholine 100 mg (1.0e1.5 mg/kg). Maintenance of anesthesia was accomplished with propofol 150e200 mcg/kg/min. Systolic blood pressure was maintained near preoperative levels with the administration of remifentanyl 0.1e0.2 mcg/kg/min and phenylephrine 10e75 mcg/min as needed. All patients were heparinized with 80e100 units/kg and the systolic blood pressure increased to approximately 120% of preoperative levels (180e200 mm Hg) just before cross-clamping. Pressure was allowed to decrease to preoperative levels plus or minus 10e20% just before unclamping. The anesthetic agent was converted to the volatile agent, desflurane, during wound closure after monitoring had ceased. All patients were extubated in the operating room without incident. Intraoperative Monitoring
METHODS A retrospective review of a prospectively maintained Institutional Review Board (IRB)-approved database was carried out. Five surgeons from the Division of Vascular Surgery at our hospital contributed to this database. Altogether, the records of 181 consecutive patients that underwent CEA between July 2005 and September 2010 were reviewed. Data related to clinical presentation, patient comorbidities, patient demographics, operative details, hospital stay, and outpatient follow-up were tabulated from the hospital, anesthesia, office, radiology, vascular laboratory, and intraoperative EEG, SEP, and MEP records. While duplex ultrasound scanning was used in 100% of patients preoperatively, 73% of patients underwent additional imaging, including computed tomography angiography (CTA), magnetic resonance angiography (MRA), and carotid angiography. The exact imaging choices were often based on the referral source but were at times chosen by the attending surgeon. Information regarding the collateral circulation did not influence the decision for intraoperative shunt placement.
Electrode placement on the skull was based on the international 10/20 EEG system. This was done after anesthesia had been successfully induced. SEP Method A modified Axon Sentinel 4 EP Analyzer (Axon Systems, Inc., Hauppauge, NY) was used for stimulation and recording of evoked potential responses. Corkscrew electrodes (Model CS; Nicolet Co., Madison, WI) were subcutaneously placed at FZ, C3, C4, CZ, and at Cz1 (International 10e20 EEG system) for cortical and subcortical median and tibial nerve cortical SEP (cSEP) recordings. Stimulating electrodes were placed at each nerve bilaterally. Stimulation parameters consisted of constant-current, square-wave pulses (median nerve stimulation: 40 mA [max]; 0.2-msec pulse duration, 2.3- or 3.5-Hz stimulation rate; tibial nerve stimulation: 40 mA [max], 0.5-msec pulse duration, 2.3- or 3.5-Hz stimulation rate). Signals were amplified 40,000 times and recorded within a 100-msec epoch length and filtered (bandpass: 30e300 Hz). A mean of 200 sweeps was completed per response. A reduction in median cSEP amplitudes of >50% from baseline
Vol. 28, No. 3, April 2014
response was considered to be a significant warning sign of cerebral ischemia. Tibial cSEP was used for control. MEP Method Corkscrew electrodes were placed subcutaneously at C1/C2 for left and at C2/C1 for right hemispheric stimulation. For eliciting MEPs, multipulse (train of 5 stimuli; single-pulse duration: 0.5 msec; interstimulus interval: 4 msec) anodal stimulation was used (200 mA [max] for transcranial electrical stimulation). The MEPs were recorded using needle electrodes inserted bilaterally in the abductor pollicis brevis and unilaterally on the extensor digitorum communis contralateral to CEA. For the lower extremity, we monitored the contralateral abductor hallucis brevis. The selection of appropriate muscles to record MEP is an important issue in the monitoring of MEPs. The small hand muscles (e.g., the abductor pollicis brevis or first dorsal interosseus muscle) are the optimal muscles to monitor for the upper extremities. In the lower extremities, the abductor hallucis brevis is the optimal muscle because of its quantitative representation in the motor cortex and pyramidal tract. The signals were recorded within a 100msec epoch, filtered (bandpass: 1.5e853 Hz), and amplified 10,000 times. Threshold level MEP responses were used, and the loss of the MEP response was considered to be a warning sign for cerebral ischemia. EEG Method A preoperative EEG was performed at baseline. In addition, a continuous digital EEG system (Axon Systems, Inc) that uses 4 channels and 8 separate electrodes was used. The EEG recording was started after anesthesia induction and continued throughout the procedure. A compressed spectral array (CSA) algorithm was used, with a 50% change in EEG frequencydeither a 50% decrease in fast EEG activity or a 50% increase in slow EEG activitydconsidered as significant. Raw digital EEG was also continuously assessed. Finally, motor status was evaluated immediately after the operation. Surgical Technique A standard anterior sternocleidomastoid incision was carried out and a longitudinal arteriotomy was created; no eversion endarterectomies were performed. Seventy percent of the CEAs had a saphenous vein patch placed, and 30% had a Hemasheild
EEG, SEP, and MEP during CEA 667
patch (Maquet, Wayne NJ); this was based on surgeon preference. In addition, 30% had distal tacking stitches placed. Close communication was maintained between the anesthesiologist, the neurophysiologist, and the operating team. A 50% reduction in amplitude of the SEP or a 50% change in EEG frequency was considered significant. Persistence or progression of these defined changes was addressed by increasing the mean arterial pressure. If this did not resolve the neurophysiologic change within 2 minutes, a Brenner shunt (Bard, Tempe, AZ) was placed. Outcomes The primary outcome measured was neurophysiologic change requiring shunt placement. Anatomic and clinical factors associated with shunt placement were evaluated. The risk factors investigated included severity of both ipsilateral and contralateral disease, the presence of ipsilateral preoperative symptoms, hypertension, coronary artery disease, diabetes, age, and sex. Elevation of intraoperative BP to 20% above baseline BP measure was standardized before clamping of the internal carotid artery. Statistics The ability of patient characteristics to predict outcomes was assessed using univariate analysis by logistic regression. P < 0.05 was considered statistically significant. The Medcalc Package (version 11.5.0.0; MedCalc Software, Mariakerke, Belgium) was used to perform statistical analysis.
RESULTS Our study group consisted of 181 primary CEAs performed in 176 patients (64% men) who were a mean age of 71.6 years (range: 47e92 years). Five patients underwent bilateral staged CEA for bilateral carotid stenosis. Demographics and comorbidities of the patients are listed in Table I. CEA was performed for asymptomatic carotid artery stenosis in 145 patients (82.4%). In total, 5 of 181 (2.76%) patients required shunt placement for significant neurophysiologic changes. There were a number of patients that had significant changes in SEP (11/181) or MEP (6/181); however, 6 of 11 improved with increased mean arterial pressure and did not require a shunt. The most frequent change observed with cross-clamping was a decrease of SEP of >50%. Six patients had changes in both SEP and MEP. Five of these patients were
668 Alcantara et al.
Annals of Vascular Surgery
Table I. Patient demographics and comorbidities Variable
Total patients Mean age, yr (range) Male Hypertension Diabetes mellitus History of cigarette smoking History of CHD/CHF Symptomatic according to NASCET criteria Stroke Transient ischemic attack Contralateral disease
n (%) unless otherwise indicated
176 71.6 112 156 44 130 88 31
(100) (47e92) (63.6) (88.6) (25.0) (73.9) (50.0) (17.6)
17 (9.7) 14 (8.0) 10 (5.5)
CAD, Coronary heart disease; CHF, congestive heart failure; NASCET, North American Symptomatic Carotid Endarterectomy Trial.
shunted. Notably, no patients in this series had significant EEG changes during the operation (Table II). The perioperative stroke and neurologic event rate was 0.0% (0/181). There was no operative mortality. The overall morbidity rate was 2.8%, including 4 (2.2%) patients with cardiac events (cardiac arrhythmia and myocardial infarct) and 1 (0.55%) patient returned to the operating room because of bleeding or hematoma. Postoperatively, 1 patient was found to have a distal clamp injury on a 1-month ultrasound scan; it was asymptomatic but was treated with a postoperative stent. Clinical factors, including male sex, age 80 years, hypertension, diabetes, coronary artery disease, and history of tobacco use were not significant predictors for shunt placement or neurophysiologic changes using logistic regression analysis. All patients had intraoperative precarotid clamping manipulation of blood pressure to 20% baseline, so this could not be evaluated as a variable. The noticeable clinical factor was that while only 17% (31/181) of all the CEAs performed were for symptomatic disease, 54% (6/11) of the total patients that had neurophysiologic changes were patients with symptomatic disease. Four of these 6 patients required a shunt, because the other 2 were successfully treated with additional elevation of their blood pressure and marked by a return to normalcy of their neurophysiologic parameters. The presence of symptomatic disease was a predictor of both change in neurophysiologic monitoring (odds ratio [OR]: 7.93; 95% confidence interval [CI]: 1.27e49.63; P ¼ 0.0269) and the need for a shunt (OR: 6.96; 95% CI: 1.97e24.55; P ¼ 0.0026). In this study, contralateral internal carotid artery (ICA) disease was not a significant predictor of
neurophysiologic change (OR: 1.6; 95% CI: 0.19e 13.77; P ¼ 0.685). In total, 5.5% of our population had contralateral disease. Only 1 of 11 (9%) patients with neurophysiologic changes had documented untreated significant contralateral disease before the operation. This single patient did require a shunt. It should be noted that the vast majority of the patients operated upon were asymptomatic (82.4%), and this population at our center all have >80% carotid stenosis. In addition, because of the low shunt rates in this seriesdand relatively low shunt rates in all selective shunt studiesdit was determined that a study designed to show superiority of 1 monitoring method over the other methods in selecting the appropriate patients to shunt (80% power; 5% significance level) would require between 3,000e5,000 patients.
DISCUSSION Various multicenter trials have shown CEA to be beneficial in symptomatic and asymptomatic carotid artery stenosis in select patients1e4; however, the method of intraoperative cerebral monitoring remains variable. When performed at centers of excellence, studies have suggested that selective shunting with routine monitoring, including EEG, carotid stump pressure measurement, and awake patients with local anesthesia, are all associated with a low perioperative stroke rate and morbidity.6,7 As we are aware, most carotid-related strokes, and certainly most perioperative strokes, are related to thromboembolic phenomena during carotid surgery.19e21 Many surgeons believe that CEA performed without a carotid shunt eliminates the potential risk of plaque dislodgement or distal vessel injury caused during shunt placement and allows for better visualization of the distal endarterectomy end point. Nonetheless, it must be noted that there are centers that practice routine carotid shunting with good results.22 Intraoperative EEG monitoring has been an established method to detect cerebral ischemia during carotid surgery, and various studies have shown that CEA performed with routine EEG monitoring and selective shunt placement is safe and associated with a relatively low rate of perioperative stroke.10e 15 This had been the practice in our group until 2004. At that point in time, through collaboration with our neurophysiology department, we initiated a protocol of SEP, MEP, and EEG monitoring during CEA. The addition of SEP monitoring theoretically allows for the increased detection of cortical and subcortical ischemia. While other groups have written about this protocol, we hoped to compare our
Vol. 28, No. 3, April 2014
EEG, SEP, and MEP during CEA 669
Table II. Characteristics of patients with neurophysiologic changes Age (yr)
Sex
Symptoms
Contralateral stenosis
MEP
SEP
MEP and SEP
Shunt
75 78 67 80
M M F F
No No No Yes
Yes No No Yes
Yes Yes Yes Yes
Yes No No Yes
Yes No No Yes
66 66 71 79 77 57 77
M F F M F M F
Yes (amaurosis fugax/TIA) Yes (L arm weakness) No Yes (speech problems, L hand weakness) No No Yes (L-sided weakness) Yes (TIA) Yes (L eye blindness) No Yes (L eye blindness)
LCEA (8 yrs ago) No RCEA (8 months ago) No No No No
No Yes No Yes Yes Yes Yes
Yes Yes Yes Yes Yes Yes Yes
No Yes No No Yes Yes Yes
No No No No Yes Yes Yes
F, Female; L, left; LCEA, left carotid endarterectomy; M, male; MEP, motor evoked potential; RCEA, right carotid endarterectomy; SEP, somatosensory evoked potential; TIA, transient ischemic attack.
need for shunt and to identify possible predictors of neurophysiologic change.17,18 Not surprisingly, we found that symptomatic disease was the only predictor of a need for shunting. Perioperative morbidity and mortality in our series was comparable with other series of CEA1e4; however, our overall shunt rate of 2.76% was very low. Other studies have found shunt rates as high as 15% with routine intraoperative EEG monitoring to direct selective shunt placement.14,22,23 Most surgeons would argue that the criterion standard for the true need for shunting would be the awake patient. Rates for carotid shunting in the awake patient have been reported to be in the 4e 6% range in patients with 80% contralateral disease. Therefore, in our experience, most patients with contralateral stenosis tolerated carotid crossclamping without neurophysiologic changes. Of the 5 patients who had bilateral staged CEAs in our series, 1 had MEP/SEP changes during their first operation. Others have found that patients with moderate ipsilateral ICA stenosis (range: 50e79%)
were more likely to be associated with EEG changes during clamping compared with those with severe ipsilateral ICA stenosis (range: 80e99%).23 While this study was not designed to look at this phenomenon, it is consistent with our previous observations. Our hypothesis, as well as the hypotheses of others, is that this reflects the greater relative loss of ipsilateral flow as a result of clamp placement in the moderate stenosis group. That being said, the only predictor of neurophysiologic change in this study was the presence of symptomatic disease. While there has been an association between symptomatic disease and the need for shunting,23 others have noted lower stump pressures in patients with symptomatic disease, but this has not been correlated with the actual need for shunting.26 As previously noted, symptomatic patients may have lower grades of stenosis than those that are asymptomatic, because symptomatic disease is predominantly secondary to distal embolic disease rather than flow-occluding lesions. Therefore, patients that have 80% carotid stenosis may be better preconditioned to tolerate the diminished flow of carotid clamping. One group has previously used SEP to decide to shunt in >1,000 operations. In that group, moderate to severe changes were noted within the first 3 minutes after carotid clamping, and in a subset of that series a shunt was used in 16.1% of 205 consecutive cases.27e29 This same group has used EEG and SEP in combination and noted that in general there was a close association between EEG and SEP. In their series, EEG changes were noted to occur 15e 30 seconds earlier than SEP changes in cerebral hypoperfusion. However, a review of the data available in the literature found that SEP has a sensitivity and specificity of 0.52 and 0.98 for detecting a
670 Alcantara et al.
neurologic event, which is comparable to the sensitivity and specificity of digital EEG (0.58 and 0.98, respectively). The sensitivity and specificity of analog EEG is much lower (0.27 and 0.87, respectively).18 In our institution’s experience, the interpretation of SEP/MEP is less subjective than EEG, which also could contribute to the discrepancy in our findings. We hypothesized that the added specificity of SEP and MEP would allow us to decrease our shunt rate even further. However, contrary to previous studies using multiple modes of neuromonitoring, there was not a significant correlation of EEG and SEP changes in the 181 operations performed, because none of our patients had EEG changes. This might be the result of the CSA EEG algorithm, because the processed EEG data could potentially mask significant analog EEG changes.30 This may also reflect the fact that these patients experienced primarily subcortical ischemia, because this is one of the theoretical advantages of SEP monitoring over EEG. When EEG, SEP, and MEP are combined, one would expect the sensitivity to detect intraoperative ischemia to increase; however, there is no large cohort study focused on combination SEP, MEP, and EEG monitoring. Recently, a multicenter study reported a subset of patients with postoperative neurologic deficits who had intraoperative MEP changes without significant SEP changes.31 The fact that we saw no neurologic sequelae in our patient cohort limits our ability to derive statistically significant conclusions regarding perioperative stroke prevention, and may be more a reflection on patient selection. Our cohort consisted mainly of asymptomatic patients with >80% stenosis. As stated previously, patients with higher levels of stenosis are likely to be preconditioned to tolerate diminished cerebral blood flow. Interestingly, given the low neurologic deficit rate recorded in large multicenter trials for CEA, we have estimated that it would take 3,000e5,000 patients to determine clinical superiority of one monitoring technique over another. We note that MEP and SEP appear to be effective monitoring modalities for CEA as well as other intracerebral vascular surgeries.32 Combining these modalities with EEG allows for a more complete assessment of cerebral ischemia. In the event of a true intraoperative stroke, all 3 of these modalities would eventually show significant changes. However, when one modality shows significant irreversible changes, a shunt must be placed. In the literature, one subject that is underemphasized is that SEP and MEP are indicative of specific cerebral functions, whereas EEG changes can only be attributed
Annals of Vascular Surgery
to a generalized region. In addition, CSA-based EEG monitoring can be prone to show changes even with steady anesthesia,33 although we did not witness this phenomenon in our cohort. It must be noted that SEP/MEP monitoring the way we perform it is quite labor intensive. We have a neurophysiologist physician monitoring the SEP, MEP, and EEG during the entire case. In this time of cost concerns, one must consider the additional cost burden. The cost effectiveness of SEP monitoring has been addressed in cervical spine surgery; in that series, it added 6% to the cost of the procedure but was looked at as a cost savings because it was cited to prevent neurologic complications.34 Given our low complication rate and the minimal cost impact of shunting, reduction in the shunt rate alone cannot be evaluated as a variable associated with a cost savings. While it is true that we had good outcomes, these cannot be attributed alone to the addition of SEP and MEP monitoring, which appears to add 8% to the cost of the procedure. Despite this, we feel that the benefit of having a less subjective modality of measuring brain ischemia, and the ability to identify the approximate territory involved is worth the added time and cost.
CONCLUSION CEA performed in conjunction with routine EEG, MEP, and SEP monitoring with selective shunt placement is safe and is associated with excellent patient outcomes at our institution. MEP, SEP, and EEG evaluate cerebral response to changes in perfusion; therefore, they are all theoretically good tools for intraoperative detection of patients that require shunting. By using triple modality monitoring, our group hoped to observe synergy resulting in increased specificity and greater precision in selecting patients for carotid shunting. Although a very low shunt rate was shown in our series, it cannot be directly attributed to an increased specificity resulting from the addition of SEP and MEP to EEG, because no patients had a postoperative deficit. In addition, as more data are collected, we will begin to better understand the relationship between each modality and how each can be used most effectively. With increased understanding, a targeted system of monitoring using 1 of the systems discussed here will hopefully be implemented based on the individual patient’s disease and risk profile. For example, the incidence of significant neurophysiologic changes and the need for shunt placement is significantly influenced by a patient’s symptomatic status. Presumably, missed neurophysiologic changes in
Vol. 28, No. 3, April 2014
this subset would have the greatest potential consequence for neurologic complications. It could be argued that triple monitoring in this scenario offers the physician the best chance at detecting neurophysiologic changes and the subsequent prevention of complications. The added benefit of using all 3 modalities remains unclear, and in today’s costconscious world of health care, our results do not justify implementing this particular technique of neuromonitoring across the board. Nonetheless, in centers with the appropriate expertise and infrastructure in place, our data suggest that this technique can be safely implemented to optimize patient outcome.
EEG, SEP, and MEP during CEA 671
15.
16.
17.
18.
19.
REFERENCES 1. North American Symptomatic Carotid Endarterectomy Trial Collaborators. Beneficial effect of carotid endarterectomy in symptomatic patients with high-grade carotid stenosis. N Engl J Med 1991;325:445e53. 2. Barnett HJM, Taylor DM, Eliasziw M, et al. Benefit of carotid endarterectomy in patients with symptomatic moderate to severe stenosis. N Engl J Med 1998;1415e25. 3. Executive Committee for the Asymptomatic Carotid Atherosclerosis Study. Endarterectomy for asymptomatic carotid artery stenosis. JAMA 1995;273:1421e8. 4. Halliday A, Mansfield A, Marro J, et al. Prevention of disabling and fatal strokes by successful carotid endarterectomy in patients without recent neurological symptoms: randomised controlled trial. Lancet 2004;363:1491e502. 5. Hobson RW, Mackey WC, Ascher E, et al. Management of atherosclerotic carotid artery disease: clinical practice guidelines of the Society of Vascular Surgery. J Vasc Surg 2008;48:480e6. 6. Woodworth GF, McGirt MJ, Than KD, et al. Selective versus routine intraoperative shunting during carotid endarterectomy: a multivariate outcome analysis. Neurosurgery 2007;1170e6. 7. Blackshear WM, DiCarlo V, Seifert KB, et al. Advantages of continuous electroencephalographic monitoring during carotid artery surgery. J Cardiovasc Surg 1986;27:146e53. 8. Shah DM, Darling RC 3rd, Chang BB, et al. Carotid endarterectomy in awake patients: its safety, acceptability, and outcome. J Vasc Surg 1994;19:1015e9. 9. Bekker AY, Basile J, Gold M, et al. Dexmedetomidine for awake carotid endarterectomy: Efficacy, Hemodynamic profile, and side effects. J Neurosurg Anesthesiol 2004;16: 126e35. 10. Faught E. Loss of high frequency EEG activities as a sensitive indicator of cerebral ischemia. Electroencephalogr Clin Neurophysiol 1984;58:119. 11. McGrail KM. Intraoperative use of electroencephalography as an assessment of cerebral blood flow. Neurosurg Clin North Am 1996;7:685e92. 12. Harada RN, Comerota AJ, Good GM, et al. Stump pressure, electroencephalographic changes, and the contralateral carotid artery: another look at selective shunting. Am J Surg 1995;170:148e53. 13. Salvian AJ, Taylor DC, Hsing YN, et al. Selective shunting with EEG monitoring is safer than routine shunting for carotid endarterectomy. Cardiovasc Surg 1997;5:481e5. 14. Schneider JR, Droste JS, Schindler N, et al. Carotid endarterectomy with routine electroencephalography and selective
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
shunting: Influence of contralateral internal carotid artery occlusion and utility in prevention of perioperative strokes. J Vasc Surg 2002;35:1114e22. Cho I, Smullens SN, Streletz LJ, et al. The value of intraoperative EEG monitoring during carotid endarterectomy. Ann Neurol 1986;20:508e12. Plestis KA, Loubser P, Mizrachi EM, et al. Continuous electroencephalography monitoring and selective shunting reduces neurological morbidity rates in carotid endarterectomy. J Vasc Surg 1997;25:620e8. Moritz S, Kasparak P, Arlt M, et al. Accuracy of cerebral monitoring in detecting cerebral ischemia during carotid endarterectomy. Anesthesiology 2007;107:563e9. Florence G, Guerit JM, Gueguen B. Electroencephalography (EEG) and somatosensory evoked potentials (SEP) to prevent cerebral ischaemia in the operating room. Neuophysiol Clin 2004;34:17e32. Jacobowitz GR, Rockman CB, Lamparello PJ, et al. Causes of perioperative stroke after carotid endarterectomy: special considerations in symptomatic patients. Ann Vasc Surg 2001;15:19e24. Smith JL, Evans DH, Fan L, et al. Interpretation of embolic phenomenon during carotid endarterectomy. Stroke 1995; 26:2281e4. de Borst GJ, Moll FL, van de Pavoordt HD, et al. Stroke from carotid endarterectomy: when and how to reduce perioperative stroke? Eur J Vasc Surg 2001;21:484e9. Rerkasem K, Rothwell PM. Routine or selective carotid artery shunting for carotid endarterectomy (and different methods of monitoring in selective shunting). Cochrane Database Syst Rev 2009;4:CD000190. Tan TW, Garcia-Toca M, Marcaccio EJ, et al. Predictors of shunt during carotid endarterectomy with routine electroencephalography monitoring. J Vasc Surg 2009;49: 1374e8. Karmeli R, Lubezky N, Halak M, et al. Carotid endarterectomy in awake patients with contralateral carotid artery occlusion. Cardiovasc Surg 2001;9:334e8. Bekker A, Gold M, Ahmed R, et al. Dexmedetomidine does not increase the incidence of intracarotid shunting in patients undergoing awake carotid endarterectomy. Anesth Analg 2006;103:955e8. Chiriano J, Abou-Zamzam AM, Nquyen K, et al. Preoperative carotid duplex findings predict carotid stump pressures during endarterectomy in symptomatic but not asymptomatic patients. Ann Vasc Surg 2010;24:1038e44. Guerit JM, Witdoeckt C, de Tourtchaninoff M, et al. Somatosensory evoked potential monitoring in carotid surgery. I. Relationships between qualitative SEP alterations and intraoperative events. Electroencephalogr Clin Neurophysiol 1997;104:459e69. Witdoeckt C, Ghariani S, Guerit JM. Somatosensory evoked potential monitoring in carotid surgery. II. Comparison between qualitative and quantitative scoring systems. Electroenchephalogr Clin Neurophysiol 1997;104:328e32. Kearse LA, Brown EN, McPeck K. Somatosensory evoked potential sensitivity relative to electroencephalography for cerebral ischemia during carotid endarterectomy. Stroke 1992;23:498e505. Kearse LA Jr, Martin D, McPeck K, et al. Computer-derived density spectral array in detection of mild analog electroencephalographic ischemic pattern changes during carotid endarterectomy. J Neurosurg 1993;78:884e90. Malcharek MJ, Ulkatan S, Marin o V, et al. Intraoperative monitoring of carotid endarterectomy by transcranial motor
672 Alcantara et al.
evoked potential: a multicenter study of 600 patients. Clin Neurophysiol 2013;124:1025e30. 32. Lopez J, Chang S, Steinberg G. The use of electrophysiological monitoring in the intraoperative management of intracranial aneurysms. J Neurol Neurosurg Psychiatry 1999; 66:189e96.
Annals of Vascular Surgery
33. Heyer E, Adams D, Moses C, et al. Erroneous conclusion from processed electroencephalogram with changing anesthetic depth. Anesthesiology 2000;92:603e7. 34. Ayoub C, Zreik T, Saway R, et al. Significance and costeffectiveness of somatosensory evoked potential monitoring in cervical spine surgery. Neurol India 2010;58:424e8.
Background: While much has been written about multiple methods of neuromonitoring during carotid endarterectomy (CEA), there has been relatively little discussion of the use of triple monitoring via somatosensory evoked potentials (SEPs) and motor evoked potentials (MEPs) in conjunction with electroencephalography (EEG). Our objective was to evaluate the rate of detection and prevention of neurologic events by multinerve SEP, MEP, and EEG in patients undergoing CEA while under general anesthesia. Methods: A prospective study of 181 consecutive patients undergoing CEA between June 2005 and September 2010 was reviewed. Intraoperative changes, including a 50% reduction in the amplitude of SEP waveforms, loss of MEP, and/or a 50% change in EEG frequency were noted as indications for shunting. This was correlated with the actual use of intraoperative shunting and postoperative neurologic sequelae at both 24 hours and 30 days. Median and tibial nerve SEPs and MEPs were also correlated. Results: Eleven patients (6%) experienced intraoperative monitoring changes (SEP: 11/11; MEP: 6/11). Five of 11 patients with MEP/SEP changes underwent shunting, while the other 6 had normalization with the elevation of their blood pressure. Of the 11 patients that had neurophysiologic changes, 54% (6/11) were patients with symptomatic disease. No patients had significant EEG changes. The total shunt rate was 2.7% (5/181). No postoperative neurologic sequelae were noted. Conclusion: The ratio of shunting at 2.7% is equal to the lowest rates reported in the awake patient literature. Interestingly, the predicted synergy of multimodality monitoring cannot be directly attributed to an increased specificity resulting from the addition of SEP and MEP to EEG, because no patients had EEG changes. In addition, in today’s cost-conscious world of health care, our results do not justify implementing this particular technique of neuromonitoring across the boarddbut it is apparent that the combination of these 3 modalities is both safe and effective with potential applications in symptomatic patients.
INTRODUCTION We continue to believe that carotid endarterectomy (CEA) has been shown to be the most effective
Presented at the 25th Eastern Vascular Meeting, National Harbor, MD, September 22e24, 2011. 1 Division of Vascular /Endovascular Surgery, St Luke’s-Roosevelt Hospital, New York, NY. 2
Division of Neurophysiology, St Luke’s-Roosevelt Hospital, New York, NY. 3
NY.
Department of Anesthesia, St Luke’s-Roosevelt Hospital, New York,
treatment in preventing cerebrovascular events in well-selected patients1e5; however, the methods of monitoring adequate cerebral perfusion during the procedure remain idiosyncratic to the center in Correspondence to: John C. Lantis II, MD, Division of Vascular/ Endovascular Surgery, 1090 Amsterdam Avenue, Suite 7A, New York, NY 10019, USA; E-mail: [email protected] Ann Vasc Surg 2014; 28: 665–672 http://dx.doi.org/10.1016/j.avsg.2013.09.005 Ó 2014 Elsevier Inc. All rights reserved. Manuscript received: October 1, 2012; manuscript accepted: September 18, 2013; published online: February 1, 2014.
665
666 Alcantara et al.
which the CEA is performed. Most centers that practice 1 particular method report excellent outcomes. The practice of cerebral monitoring and the use of carotid shunt placement remains variable. Some recent studies have suggested that monitoring and selective shunt use during CEA decreases the risk of perioperative stroke and can be superior to routine shunt placement,6,7 while other centers report excellent results with regional anesthesia and awake monitoring.8,9 Although no absolute data are available, it would appear that electroencephalography (EEG) has been established as one of the more commonly used tools for intraoperative cerebral monitoring because of its ability to detect cerebral ischemia. Multiple groups have shown that CEA performed with routine EEG monitoring and selective shunt placement is both safe and effective.10e16 The published literature contains significant detail about multiple methods of neuromonitoring during CEA; however, there is relatively little discussion regarding the use of somatosensory evoked potentials (SEPs)dand, in particular, multinerve SEPsdin conjunction with EEG.17,18 Our objective was to evaluate the rate of detection and prevention of neurologic events by multinerve SEP, motor evoked potentials (MEPs), and EEG in patients undergoing CEA while under general anesthesia.
Annals of Vascular Surgery
Anesthesia All patients were anesthetized with total intravenous anesthesia to facilitate the requirements of SEP monitoring. Volatile anesthetic agents were avoided because they are known to interfere with SEP monitoring by decreasing amplitude and increasing latency. We also avoided nitrous oxide. Patients were ventilated with FiO2 0.4e1.0. Anesthesia was induced with intravenous midazolam 2 mg, intravenous fentanyl 250 mcg (3e5 mcg/kg), propofol 2.0e2.5 mg/kg, or etomidate 0.2e0.3 mg/kg and succinylcholine 100 mg (1.0e1.5 mg/kg). Maintenance of anesthesia was accomplished with propofol 150e200 mcg/kg/min. Systolic blood pressure was maintained near preoperative levels with the administration of remifentanyl 0.1e0.2 mcg/kg/min and phenylephrine 10e75 mcg/min as needed. All patients were heparinized with 80e100 units/kg and the systolic blood pressure increased to approximately 120% of preoperative levels (180e200 mm Hg) just before cross-clamping. Pressure was allowed to decrease to preoperative levels plus or minus 10e20% just before unclamping. The anesthetic agent was converted to the volatile agent, desflurane, during wound closure after monitoring had ceased. All patients were extubated in the operating room without incident. Intraoperative Monitoring
METHODS A retrospective review of a prospectively maintained Institutional Review Board (IRB)-approved database was carried out. Five surgeons from the Division of Vascular Surgery at our hospital contributed to this database. Altogether, the records of 181 consecutive patients that underwent CEA between July 2005 and September 2010 were reviewed. Data related to clinical presentation, patient comorbidities, patient demographics, operative details, hospital stay, and outpatient follow-up were tabulated from the hospital, anesthesia, office, radiology, vascular laboratory, and intraoperative EEG, SEP, and MEP records. While duplex ultrasound scanning was used in 100% of patients preoperatively, 73% of patients underwent additional imaging, including computed tomography angiography (CTA), magnetic resonance angiography (MRA), and carotid angiography. The exact imaging choices were often based on the referral source but were at times chosen by the attending surgeon. Information regarding the collateral circulation did not influence the decision for intraoperative shunt placement.
Electrode placement on the skull was based on the international 10/20 EEG system. This was done after anesthesia had been successfully induced. SEP Method A modified Axon Sentinel 4 EP Analyzer (Axon Systems, Inc., Hauppauge, NY) was used for stimulation and recording of evoked potential responses. Corkscrew electrodes (Model CS; Nicolet Co., Madison, WI) were subcutaneously placed at FZ, C3, C4, CZ, and at Cz1 (International 10e20 EEG system) for cortical and subcortical median and tibial nerve cortical SEP (cSEP) recordings. Stimulating electrodes were placed at each nerve bilaterally. Stimulation parameters consisted of constant-current, square-wave pulses (median nerve stimulation: 40 mA [max]; 0.2-msec pulse duration, 2.3- or 3.5-Hz stimulation rate; tibial nerve stimulation: 40 mA [max], 0.5-msec pulse duration, 2.3- or 3.5-Hz stimulation rate). Signals were amplified 40,000 times and recorded within a 100-msec epoch length and filtered (bandpass: 30e300 Hz). A mean of 200 sweeps was completed per response. A reduction in median cSEP amplitudes of >50% from baseline
Vol. 28, No. 3, April 2014
response was considered to be a significant warning sign of cerebral ischemia. Tibial cSEP was used for control. MEP Method Corkscrew electrodes were placed subcutaneously at C1/C2 for left and at C2/C1 for right hemispheric stimulation. For eliciting MEPs, multipulse (train of 5 stimuli; single-pulse duration: 0.5 msec; interstimulus interval: 4 msec) anodal stimulation was used (200 mA [max] for transcranial electrical stimulation). The MEPs were recorded using needle electrodes inserted bilaterally in the abductor pollicis brevis and unilaterally on the extensor digitorum communis contralateral to CEA. For the lower extremity, we monitored the contralateral abductor hallucis brevis. The selection of appropriate muscles to record MEP is an important issue in the monitoring of MEPs. The small hand muscles (e.g., the abductor pollicis brevis or first dorsal interosseus muscle) are the optimal muscles to monitor for the upper extremities. In the lower extremities, the abductor hallucis brevis is the optimal muscle because of its quantitative representation in the motor cortex and pyramidal tract. The signals were recorded within a 100msec epoch, filtered (bandpass: 1.5e853 Hz), and amplified 10,000 times. Threshold level MEP responses were used, and the loss of the MEP response was considered to be a warning sign for cerebral ischemia. EEG Method A preoperative EEG was performed at baseline. In addition, a continuous digital EEG system (Axon Systems, Inc) that uses 4 channels and 8 separate electrodes was used. The EEG recording was started after anesthesia induction and continued throughout the procedure. A compressed spectral array (CSA) algorithm was used, with a 50% change in EEG frequencydeither a 50% decrease in fast EEG activity or a 50% increase in slow EEG activitydconsidered as significant. Raw digital EEG was also continuously assessed. Finally, motor status was evaluated immediately after the operation. Surgical Technique A standard anterior sternocleidomastoid incision was carried out and a longitudinal arteriotomy was created; no eversion endarterectomies were performed. Seventy percent of the CEAs had a saphenous vein patch placed, and 30% had a Hemasheild
EEG, SEP, and MEP during CEA 667
patch (Maquet, Wayne NJ); this was based on surgeon preference. In addition, 30% had distal tacking stitches placed. Close communication was maintained between the anesthesiologist, the neurophysiologist, and the operating team. A 50% reduction in amplitude of the SEP or a 50% change in EEG frequency was considered significant. Persistence or progression of these defined changes was addressed by increasing the mean arterial pressure. If this did not resolve the neurophysiologic change within 2 minutes, a Brenner shunt (Bard, Tempe, AZ) was placed. Outcomes The primary outcome measured was neurophysiologic change requiring shunt placement. Anatomic and clinical factors associated with shunt placement were evaluated. The risk factors investigated included severity of both ipsilateral and contralateral disease, the presence of ipsilateral preoperative symptoms, hypertension, coronary artery disease, diabetes, age, and sex. Elevation of intraoperative BP to 20% above baseline BP measure was standardized before clamping of the internal carotid artery. Statistics The ability of patient characteristics to predict outcomes was assessed using univariate analysis by logistic regression. P < 0.05 was considered statistically significant. The Medcalc Package (version 11.5.0.0; MedCalc Software, Mariakerke, Belgium) was used to perform statistical analysis.
RESULTS Our study group consisted of 181 primary CEAs performed in 176 patients (64% men) who were a mean age of 71.6 years (range: 47e92 years). Five patients underwent bilateral staged CEA for bilateral carotid stenosis. Demographics and comorbidities of the patients are listed in Table I. CEA was performed for asymptomatic carotid artery stenosis in 145 patients (82.4%). In total, 5 of 181 (2.76%) patients required shunt placement for significant neurophysiologic changes. There were a number of patients that had significant changes in SEP (11/181) or MEP (6/181); however, 6 of 11 improved with increased mean arterial pressure and did not require a shunt. The most frequent change observed with cross-clamping was a decrease of SEP of >50%. Six patients had changes in both SEP and MEP. Five of these patients were
668 Alcantara et al.
Annals of Vascular Surgery
Table I. Patient demographics and comorbidities Variable
Total patients Mean age, yr (range) Male Hypertension Diabetes mellitus History of cigarette smoking History of CHD/CHF Symptomatic according to NASCET criteria Stroke Transient ischemic attack Contralateral disease
n (%) unless otherwise indicated
176 71.6 112 156 44 130 88 31
(100) (47e92) (63.6) (88.6) (25.0) (73.9) (50.0) (17.6)
17 (9.7) 14 (8.0) 10 (5.5)
CAD, Coronary heart disease; CHF, congestive heart failure; NASCET, North American Symptomatic Carotid Endarterectomy Trial.
shunted. Notably, no patients in this series had significant EEG changes during the operation (Table II). The perioperative stroke and neurologic event rate was 0.0% (0/181). There was no operative mortality. The overall morbidity rate was 2.8%, including 4 (2.2%) patients with cardiac events (cardiac arrhythmia and myocardial infarct) and 1 (0.55%) patient returned to the operating room because of bleeding or hematoma. Postoperatively, 1 patient was found to have a distal clamp injury on a 1-month ultrasound scan; it was asymptomatic but was treated with a postoperative stent. Clinical factors, including male sex, age 80 years, hypertension, diabetes, coronary artery disease, and history of tobacco use were not significant predictors for shunt placement or neurophysiologic changes using logistic regression analysis. All patients had intraoperative precarotid clamping manipulation of blood pressure to 20% baseline, so this could not be evaluated as a variable. The noticeable clinical factor was that while only 17% (31/181) of all the CEAs performed were for symptomatic disease, 54% (6/11) of the total patients that had neurophysiologic changes were patients with symptomatic disease. Four of these 6 patients required a shunt, because the other 2 were successfully treated with additional elevation of their blood pressure and marked by a return to normalcy of their neurophysiologic parameters. The presence of symptomatic disease was a predictor of both change in neurophysiologic monitoring (odds ratio [OR]: 7.93; 95% confidence interval [CI]: 1.27e49.63; P ¼ 0.0269) and the need for a shunt (OR: 6.96; 95% CI: 1.97e24.55; P ¼ 0.0026). In this study, contralateral internal carotid artery (ICA) disease was not a significant predictor of
neurophysiologic change (OR: 1.6; 95% CI: 0.19e 13.77; P ¼ 0.685). In total, 5.5% of our population had contralateral disease. Only 1 of 11 (9%) patients with neurophysiologic changes had documented untreated significant contralateral disease before the operation. This single patient did require a shunt. It should be noted that the vast majority of the patients operated upon were asymptomatic (82.4%), and this population at our center all have >80% carotid stenosis. In addition, because of the low shunt rates in this seriesdand relatively low shunt rates in all selective shunt studiesdit was determined that a study designed to show superiority of 1 monitoring method over the other methods in selecting the appropriate patients to shunt (80% power; 5% significance level) would require between 3,000e5,000 patients.
DISCUSSION Various multicenter trials have shown CEA to be beneficial in symptomatic and asymptomatic carotid artery stenosis in select patients1e4; however, the method of intraoperative cerebral monitoring remains variable. When performed at centers of excellence, studies have suggested that selective shunting with routine monitoring, including EEG, carotid stump pressure measurement, and awake patients with local anesthesia, are all associated with a low perioperative stroke rate and morbidity.6,7 As we are aware, most carotid-related strokes, and certainly most perioperative strokes, are related to thromboembolic phenomena during carotid surgery.19e21 Many surgeons believe that CEA performed without a carotid shunt eliminates the potential risk of plaque dislodgement or distal vessel injury caused during shunt placement and allows for better visualization of the distal endarterectomy end point. Nonetheless, it must be noted that there are centers that practice routine carotid shunting with good results.22 Intraoperative EEG monitoring has been an established method to detect cerebral ischemia during carotid surgery, and various studies have shown that CEA performed with routine EEG monitoring and selective shunt placement is safe and associated with a relatively low rate of perioperative stroke.10e 15 This had been the practice in our group until 2004. At that point in time, through collaboration with our neurophysiology department, we initiated a protocol of SEP, MEP, and EEG monitoring during CEA. The addition of SEP monitoring theoretically allows for the increased detection of cortical and subcortical ischemia. While other groups have written about this protocol, we hoped to compare our
Vol. 28, No. 3, April 2014
EEG, SEP, and MEP during CEA 669
Table II. Characteristics of patients with neurophysiologic changes Age (yr)
Sex
Symptoms
Contralateral stenosis
MEP
SEP
MEP and SEP
Shunt
75 78 67 80
M M F F
No No No Yes
Yes No No Yes
Yes Yes Yes Yes
Yes No No Yes
Yes No No Yes
66 66 71 79 77 57 77
M F F M F M F
Yes (amaurosis fugax/TIA) Yes (L arm weakness) No Yes (speech problems, L hand weakness) No No Yes (L-sided weakness) Yes (TIA) Yes (L eye blindness) No Yes (L eye blindness)
LCEA (8 yrs ago) No RCEA (8 months ago) No No No No
No Yes No Yes Yes Yes Yes
Yes Yes Yes Yes Yes Yes Yes
No Yes No No Yes Yes Yes
No No No No Yes Yes Yes
F, Female; L, left; LCEA, left carotid endarterectomy; M, male; MEP, motor evoked potential; RCEA, right carotid endarterectomy; SEP, somatosensory evoked potential; TIA, transient ischemic attack.
need for shunt and to identify possible predictors of neurophysiologic change.17,18 Not surprisingly, we found that symptomatic disease was the only predictor of a need for shunting. Perioperative morbidity and mortality in our series was comparable with other series of CEA1e4; however, our overall shunt rate of 2.76% was very low. Other studies have found shunt rates as high as 15% with routine intraoperative EEG monitoring to direct selective shunt placement.14,22,23 Most surgeons would argue that the criterion standard for the true need for shunting would be the awake patient. Rates for carotid shunting in the awake patient have been reported to be in the 4e 6% range in patients with 80% contralateral disease. Therefore, in our experience, most patients with contralateral stenosis tolerated carotid crossclamping without neurophysiologic changes. Of the 5 patients who had bilateral staged CEAs in our series, 1 had MEP/SEP changes during their first operation. Others have found that patients with moderate ipsilateral ICA stenosis (range: 50e79%)
were more likely to be associated with EEG changes during clamping compared with those with severe ipsilateral ICA stenosis (range: 80e99%).23 While this study was not designed to look at this phenomenon, it is consistent with our previous observations. Our hypothesis, as well as the hypotheses of others, is that this reflects the greater relative loss of ipsilateral flow as a result of clamp placement in the moderate stenosis group. That being said, the only predictor of neurophysiologic change in this study was the presence of symptomatic disease. While there has been an association between symptomatic disease and the need for shunting,23 others have noted lower stump pressures in patients with symptomatic disease, but this has not been correlated with the actual need for shunting.26 As previously noted, symptomatic patients may have lower grades of stenosis than those that are asymptomatic, because symptomatic disease is predominantly secondary to distal embolic disease rather than flow-occluding lesions. Therefore, patients that have 80% carotid stenosis may be better preconditioned to tolerate the diminished flow of carotid clamping. One group has previously used SEP to decide to shunt in >1,000 operations. In that group, moderate to severe changes were noted within the first 3 minutes after carotid clamping, and in a subset of that series a shunt was used in 16.1% of 205 consecutive cases.27e29 This same group has used EEG and SEP in combination and noted that in general there was a close association between EEG and SEP. In their series, EEG changes were noted to occur 15e 30 seconds earlier than SEP changes in cerebral hypoperfusion. However, a review of the data available in the literature found that SEP has a sensitivity and specificity of 0.52 and 0.98 for detecting a
670 Alcantara et al.
neurologic event, which is comparable to the sensitivity and specificity of digital EEG (0.58 and 0.98, respectively). The sensitivity and specificity of analog EEG is much lower (0.27 and 0.87, respectively).18 In our institution’s experience, the interpretation of SEP/MEP is less subjective than EEG, which also could contribute to the discrepancy in our findings. We hypothesized that the added specificity of SEP and MEP would allow us to decrease our shunt rate even further. However, contrary to previous studies using multiple modes of neuromonitoring, there was not a significant correlation of EEG and SEP changes in the 181 operations performed, because none of our patients had EEG changes. This might be the result of the CSA EEG algorithm, because the processed EEG data could potentially mask significant analog EEG changes.30 This may also reflect the fact that these patients experienced primarily subcortical ischemia, because this is one of the theoretical advantages of SEP monitoring over EEG. When EEG, SEP, and MEP are combined, one would expect the sensitivity to detect intraoperative ischemia to increase; however, there is no large cohort study focused on combination SEP, MEP, and EEG monitoring. Recently, a multicenter study reported a subset of patients with postoperative neurologic deficits who had intraoperative MEP changes without significant SEP changes.31 The fact that we saw no neurologic sequelae in our patient cohort limits our ability to derive statistically significant conclusions regarding perioperative stroke prevention, and may be more a reflection on patient selection. Our cohort consisted mainly of asymptomatic patients with >80% stenosis. As stated previously, patients with higher levels of stenosis are likely to be preconditioned to tolerate diminished cerebral blood flow. Interestingly, given the low neurologic deficit rate recorded in large multicenter trials for CEA, we have estimated that it would take 3,000e5,000 patients to determine clinical superiority of one monitoring technique over another. We note that MEP and SEP appear to be effective monitoring modalities for CEA as well as other intracerebral vascular surgeries.32 Combining these modalities with EEG allows for a more complete assessment of cerebral ischemia. In the event of a true intraoperative stroke, all 3 of these modalities would eventually show significant changes. However, when one modality shows significant irreversible changes, a shunt must be placed. In the literature, one subject that is underemphasized is that SEP and MEP are indicative of specific cerebral functions, whereas EEG changes can only be attributed
Annals of Vascular Surgery
to a generalized region. In addition, CSA-based EEG monitoring can be prone to show changes even with steady anesthesia,33 although we did not witness this phenomenon in our cohort. It must be noted that SEP/MEP monitoring the way we perform it is quite labor intensive. We have a neurophysiologist physician monitoring the SEP, MEP, and EEG during the entire case. In this time of cost concerns, one must consider the additional cost burden. The cost effectiveness of SEP monitoring has been addressed in cervical spine surgery; in that series, it added 6% to the cost of the procedure but was looked at as a cost savings because it was cited to prevent neurologic complications.34 Given our low complication rate and the minimal cost impact of shunting, reduction in the shunt rate alone cannot be evaluated as a variable associated with a cost savings. While it is true that we had good outcomes, these cannot be attributed alone to the addition of SEP and MEP monitoring, which appears to add 8% to the cost of the procedure. Despite this, we feel that the benefit of having a less subjective modality of measuring brain ischemia, and the ability to identify the approximate territory involved is worth the added time and cost.
CONCLUSION CEA performed in conjunction with routine EEG, MEP, and SEP monitoring with selective shunt placement is safe and is associated with excellent patient outcomes at our institution. MEP, SEP, and EEG evaluate cerebral response to changes in perfusion; therefore, they are all theoretically good tools for intraoperative detection of patients that require shunting. By using triple modality monitoring, our group hoped to observe synergy resulting in increased specificity and greater precision in selecting patients for carotid shunting. Although a very low shunt rate was shown in our series, it cannot be directly attributed to an increased specificity resulting from the addition of SEP and MEP to EEG, because no patients had a postoperative deficit. In addition, as more data are collected, we will begin to better understand the relationship between each modality and how each can be used most effectively. With increased understanding, a targeted system of monitoring using 1 of the systems discussed here will hopefully be implemented based on the individual patient’s disease and risk profile. For example, the incidence of significant neurophysiologic changes and the need for shunt placement is significantly influenced by a patient’s symptomatic status. Presumably, missed neurophysiologic changes in
Vol. 28, No. 3, April 2014
this subset would have the greatest potential consequence for neurologic complications. It could be argued that triple monitoring in this scenario offers the physician the best chance at detecting neurophysiologic changes and the subsequent prevention of complications. The added benefit of using all 3 modalities remains unclear, and in today’s costconscious world of health care, our results do not justify implementing this particular technique of neuromonitoring across the board. Nonetheless, in centers with the appropriate expertise and infrastructure in place, our data suggest that this technique can be safely implemented to optimize patient outcome.
EEG, SEP, and MEP during CEA 671
15.
16.
17.
18.
19.
REFERENCES 1. North American Symptomatic Carotid Endarterectomy Trial Collaborators. Beneficial effect of carotid endarterectomy in symptomatic patients with high-grade carotid stenosis. N Engl J Med 1991;325:445e53. 2. Barnett HJM, Taylor DM, Eliasziw M, et al. Benefit of carotid endarterectomy in patients with symptomatic moderate to severe stenosis. N Engl J Med 1998;1415e25. 3. Executive Committee for the Asymptomatic Carotid Atherosclerosis Study. Endarterectomy for asymptomatic carotid artery stenosis. JAMA 1995;273:1421e8. 4. Halliday A, Mansfield A, Marro J, et al. Prevention of disabling and fatal strokes by successful carotid endarterectomy in patients without recent neurological symptoms: randomised controlled trial. Lancet 2004;363:1491e502. 5. Hobson RW, Mackey WC, Ascher E, et al. Management of atherosclerotic carotid artery disease: clinical practice guidelines of the Society of Vascular Surgery. J Vasc Surg 2008;48:480e6. 6. Woodworth GF, McGirt MJ, Than KD, et al. Selective versus routine intraoperative shunting during carotid endarterectomy: a multivariate outcome analysis. Neurosurgery 2007;1170e6. 7. Blackshear WM, DiCarlo V, Seifert KB, et al. Advantages of continuous electroencephalographic monitoring during carotid artery surgery. J Cardiovasc Surg 1986;27:146e53. 8. Shah DM, Darling RC 3rd, Chang BB, et al. Carotid endarterectomy in awake patients: its safety, acceptability, and outcome. J Vasc Surg 1994;19:1015e9. 9. Bekker AY, Basile J, Gold M, et al. Dexmedetomidine for awake carotid endarterectomy: Efficacy, Hemodynamic profile, and side effects. J Neurosurg Anesthesiol 2004;16: 126e35. 10. Faught E. Loss of high frequency EEG activities as a sensitive indicator of cerebral ischemia. Electroencephalogr Clin Neurophysiol 1984;58:119. 11. McGrail KM. Intraoperative use of electroencephalography as an assessment of cerebral blood flow. Neurosurg Clin North Am 1996;7:685e92. 12. Harada RN, Comerota AJ, Good GM, et al. Stump pressure, electroencephalographic changes, and the contralateral carotid artery: another look at selective shunting. Am J Surg 1995;170:148e53. 13. Salvian AJ, Taylor DC, Hsing YN, et al. Selective shunting with EEG monitoring is safer than routine shunting for carotid endarterectomy. Cardiovasc Surg 1997;5:481e5. 14. Schneider JR, Droste JS, Schindler N, et al. Carotid endarterectomy with routine electroencephalography and selective
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
shunting: Influence of contralateral internal carotid artery occlusion and utility in prevention of perioperative strokes. J Vasc Surg 2002;35:1114e22. Cho I, Smullens SN, Streletz LJ, et al. The value of intraoperative EEG monitoring during carotid endarterectomy. Ann Neurol 1986;20:508e12. Plestis KA, Loubser P, Mizrachi EM, et al. Continuous electroencephalography monitoring and selective shunting reduces neurological morbidity rates in carotid endarterectomy. J Vasc Surg 1997;25:620e8. Moritz S, Kasparak P, Arlt M, et al. Accuracy of cerebral monitoring in detecting cerebral ischemia during carotid endarterectomy. Anesthesiology 2007;107:563e9. Florence G, Guerit JM, Gueguen B. Electroencephalography (EEG) and somatosensory evoked potentials (SEP) to prevent cerebral ischaemia in the operating room. Neuophysiol Clin 2004;34:17e32. Jacobowitz GR, Rockman CB, Lamparello PJ, et al. Causes of perioperative stroke after carotid endarterectomy: special considerations in symptomatic patients. Ann Vasc Surg 2001;15:19e24. Smith JL, Evans DH, Fan L, et al. Interpretation of embolic phenomenon during carotid endarterectomy. Stroke 1995; 26:2281e4. de Borst GJ, Moll FL, van de Pavoordt HD, et al. Stroke from carotid endarterectomy: when and how to reduce perioperative stroke? Eur J Vasc Surg 2001;21:484e9. Rerkasem K, Rothwell PM. Routine or selective carotid artery shunting for carotid endarterectomy (and different methods of monitoring in selective shunting). Cochrane Database Syst Rev 2009;4:CD000190. Tan TW, Garcia-Toca M, Marcaccio EJ, et al. Predictors of shunt during carotid endarterectomy with routine electroencephalography monitoring. J Vasc Surg 2009;49: 1374e8. Karmeli R, Lubezky N, Halak M, et al. Carotid endarterectomy in awake patients with contralateral carotid artery occlusion. Cardiovasc Surg 2001;9:334e8. Bekker A, Gold M, Ahmed R, et al. Dexmedetomidine does not increase the incidence of intracarotid shunting in patients undergoing awake carotid endarterectomy. Anesth Analg 2006;103:955e8. Chiriano J, Abou-Zamzam AM, Nquyen K, et al. Preoperative carotid duplex findings predict carotid stump pressures during endarterectomy in symptomatic but not asymptomatic patients. Ann Vasc Surg 2010;24:1038e44. Guerit JM, Witdoeckt C, de Tourtchaninoff M, et al. Somatosensory evoked potential monitoring in carotid surgery. I. Relationships between qualitative SEP alterations and intraoperative events. Electroencephalogr Clin Neurophysiol 1997;104:459e69. Witdoeckt C, Ghariani S, Guerit JM. Somatosensory evoked potential monitoring in carotid surgery. II. Comparison between qualitative and quantitative scoring systems. Electroenchephalogr Clin Neurophysiol 1997;104:328e32. Kearse LA, Brown EN, McPeck K. Somatosensory evoked potential sensitivity relative to electroencephalography for cerebral ischemia during carotid endarterectomy. Stroke 1992;23:498e505. Kearse LA Jr, Martin D, McPeck K, et al. Computer-derived density spectral array in detection of mild analog electroencephalographic ischemic pattern changes during carotid endarterectomy. J Neurosurg 1993;78:884e90. Malcharek MJ, Ulkatan S, Marin o V, et al. Intraoperative monitoring of carotid endarterectomy by transcranial motor
672 Alcantara et al.
evoked potential: a multicenter study of 600 patients. Clin Neurophysiol 2013;124:1025e30. 32. Lopez J, Chang S, Steinberg G. The use of electrophysiological monitoring in the intraoperative management of intracranial aneurysms. J Neurol Neurosurg Psychiatry 1999; 66:189e96.
Annals of Vascular Surgery
33. Heyer E, Adams D, Moses C, et al. Erroneous conclusion from processed electroencephalogram with changing anesthetic depth. Anesthesiology 2000;92:603e7. 34. Ayoub C, Zreik T, Saway R, et al. Significance and costeffectiveness of somatosensory evoked potential monitoring in cervical spine surgery. Neurol India 2010;58:424e8.
