Secondary Logo

Journal Logo

Original Article

The role of somatosensory evoked potentials in detecting cerebral ischaemia during carotid endarterectomy: An assessment of its validity under regional anaesthesia

Fielmuth, S.*,†; Uhlig, T.¶,‡

Author Information
European Journal of Anaesthesiology: August 2008 - Volume 25 - Issue 8 - p 648-656
doi: 10.1017/S0265021508003967



The most effective method of preventing cerebral ischaemia during carotid endarterectomy (CEA) is selective shunt insertion during the clamp phase [1]. Since carotid surgery is performed mostly under general anaesthesia [2], the state of perfusion is assessed by monitoring haemodynamic and electrophysiological parameters.

Today there is no consensus about the most appropriate method of detecting cerebral ischaemia. However, somatosensory evoked potentials (SSEP) monitoring of the median nerve has been extensively used during CEA. It is a technique with high reliability, which allows the definition of the critical perfusion threshold in numerous studies [3-5].

Until now, the SSEP-generated shunt criteria published in the literature diverge substantially, are largely arbitrary and in some cases contradictory [6-14]. Therefore it is not possible to determine the most reliable shunt criteria. This study was designed to compare SSEP tracings with the objective clinical-neurological examination to evaluate the reliability of the different SSEP-generated shunt criteria.


After Institutional Review Board approval and informed consent, 102 patients who underwent elective CEA using cervical plexus blockade were prospectively studied. The primary end-point of this trial was the assessment of the sensitivity and specificity of the SSEP-generated shunt criteria, described in the literature as reduction of the amplitude of the primary cortical response at N20/P25 with a critical threshold at 50%, the complete loss of the primary cortical response at N20/P25, the prolongation of central conduction time (CCT) by more than 20%, and the combination of conductive velocity and neuronal recruitment in the need-to-shunt index (NSI), with a cut-off at >0.5. (Fig. 1) Secondary end-points were overall morbidity/mortality, shunt frequency and the rate of neurological, cardiac, anaesthetic and surgical complications.

Figure 1.
Figure 1.:
Somatosensory evoked potentials' parameters investigated as threshold of impaired cerebral blood flow.

On the scheduled day of operation, oral premedication was given consisting of midazolam 7.5 mg unless contraindicated by coexisting diseases. No further sedation was used during the course of the operation.

Preoperatively, the patients received a deep cervical plexus block with 30 mL 0.75% ropivacaine and a superficial cervical plexus block with 10 mL 0.75% ropivacaine [15].

The criteria used to change to the general anaesthesia technique during the procedure were:

  • loss of consciousness that persisted after insertion of the shunt;
  • perioperative generalized epileptic seizures;
  • an uncooperative patient;
  • intraoperative complications with massive blood loss.

The median nerve SSEP was established immediately after cervical plexus block. The volar aspect of the contralateral wrist was stimulated using a current 2 mA above the motor threshold with square wave pulses of 0.1 s duration at a frequency of 5.3 Hz ± 10%. The signal was recorded via platinum needle electrodes (Fa. Nicolet, Madison, WI, USA) at C3′, C4′ and Cv2 referred to Fpz (10/20 system) with grounding at the contralateral Erb point and with electrode impedance less than 5 kOhm. Analysis was carried out using the Sentinel Vers. 4.0 (Fa. Axon Systems) Fast-Fourier-Analysis Amplifier set at 4000×, with an artefact level less than 50 μV. 500 Sweeps were used for averaging and acquired in a time window of 3–43 ms. If the recording of the electrophysiological parameters fulfilled the technical specifications (recording impedance less than 5 kOhm, artefact level less than 50 μV), the investigation was conducted and documented.

Electrocardiograph, non-invasive blood pressure and SPO2 were continuously monitored during the entire procedure. No interventions to raise blood pressure and thus cerebral perfusion pressure were performed as long as the patients' neurological status remained stable. Haematocrit was held between 30% and 35% throughout the procedure and in the first three postoperative days. The patients received an O2 flow of 4 L min−1 via a face mask. The rate and type of breathing were recorded graphically via the capnograph of a standard anaesthetic machine. Anticoagulant medications reported in the past medical history were replaced preoperatively by enoxaparin. Patients received 5000 IU of heparin prior to the common carotid artery clamp being applied. Postoperatively, thromboembolic prophylaxis was provided by low-dose heparin adding aspirin 100 mg from the second postoperative day.

Phases of potential deterioration of cerebral perfusion were defined on the basis of haemodynamic changes during the surgical procedure. The following time points were used:

  • Preoperative phase;
  • Positioning (threat of vertebrobasilar insufficiency);
  • Dissection (threat of mobilization of atheromatous plaques);
  • Clamping of the internal carotid artery (detection of insufficiencies of the circulation);
  • 5, 10, 15 min after clamping (exhaustion of collateral and dilatation reserves);
  • Declamping (embolization of thrombotic material, ischaemia from intimal flaps).

The SSEP parameters and the defined parameters of mental and neurologic state were measured at each of these periods. SSEP tracings were obtained both from the ipsilateral and contralateral parietal areas as well as from the cervical region referred to a midfrontal electrode (International 10-20 System). The current latencies for N13, N20 and P25 in ms together with the current peak-to-baseline amplitudes for N13 and the current peak-to-peak amplitudes for N20/P25 in μV were read from the tracings. We recorded the degree of motor and sensory activity as well as vigilance based on a scale from grade 1 to grade 3 in terms of the alterations of the initial state. The neurological monitoring scale was developed for our own use in assessing clinical deteriorations during CEA. The clinical-neurological monitoring is summarized in Table 1.

Table 1
Table 1:
Clinical neurological monitoring.

During the clamping phase, questions were put to the patients with regard to their state of orientation, possible paraesthesias and subjective well-being in order to judge their vigilance and sensory parameters. The motor parameters were observed by patients honking the toy horn with the contralateral hand. When the results were unchanged, the operation was continued without insertion of a shunt. In cases of reduced vigilance or disturbances in sensory or motor parameters occurring immediately after clamping, a shunt was inserted. In such cases, the operation was continued after the neurological status stabilized. When symptoms were delayed, shunt insertion depended on the progress of the operation. In all cases, the indication for shunting was established exclusively on the basis of clinical criteria. Changes in SSEP parameters did not affect the surgical strategy, as the surgeon was blinded to the neurophysiology. When neurological and haemodynamic indicators were stable in a normal postoperative course, the patients were transferred to a postanaesthesia care unit or a normal ward. Before hospital discharge, a neurologist conducted a final examination using the definitions of the Guidelines for Carotid Endarterectomy of the Stroke Council of the American Heart Association and the Severe Adverse Effect criteria of NASCET [16,17]. This examination was usually scheduled for the tenth postoperative day.

Statistical analysis

The calculation of recorded shunt parameter, as suggested in previous studies by different groups (amplitude reduction N20/P25 >50%[8], amplitude loss of N20/P25 [6], CCT prolongation >20% [12,13], NSI >0.5 [14]), is shown in Table 2. These parameters were correlated chronologically with the corresponding clinical-neurological examination. When the respective threshold value of SSEP-generated shunt criteria was exceeded, but clinical status was not affected, this difference was defined as false positive for the corresponding shunt parameters. When a shunt was required as a consequence of a neurological deterioration, whereas the respective threshold value was not exceeded, the result was rated as false negative. Any delay in electrophysiological response to clinical worsening of more than 2 min was evaluated as a false negative. Reciever operater characterisic curve analysis was used to assess the area under the curve for each parameter investigated. For determining the cut-off values, the ordinal rating scale was used. As the investigated thresholds of impaired cerebral blood flow were clearly defined as a definitive value to be exceeded or not, the derived cut-off values could only be definitely negative or definitely positive.

Table 2
Table 2:
Calculations of measured parameters.


We studied 102 patients in this prospective study. In 11 patients continuous averaging of the signal was impossible due to recording impedances outside the acceptable range or very poor signal-to-noise ratios. In the remaining 91 patients SSEP tracings of both the ipsilateral and contralateral hemisphere were clearly identifiable and judged as baseline. Patient characteristics are presented in Table 3.

Table 3
Table 3:
Patient characteristics (n = 91).

In two patients, 2 mL of 0.75% ropivacaine was accidentally injected into the vertebral artery beneath the deep cervical plexus block despite negative aspiration. Both patients showed the same cerebral symptoms of epileptic seizures, which were suppressed by 5 mg of midazolam. Both patients received oxygen mask ventilation in the subsequent postictal phase of 10-min duration. These patients were transferred to the recovery room with sufficient spontaneous breathing and adequate responses. The intra-arterial injection did not result in any neurological deficiency in either patient. These two patients were included later. It was not necessary to switch from regional anaesthesia to general anaesthesia for any of the 102 patients investigated.

A total of 71 patients were neurologically unaffected during the entire operation, showing good tolerance of cross-clamping. Twenty patients required a temporary shunt insertion to bypass the cross-clamping of the common carotid artery and internal carotid artery, demonstrating a shunt frequency of 22% by clinical indication. The common cause (in 14 patients) was immediate loss of consciousness (within 1 min after clamping) and in six patients it was paralysis of the contralateral hand (motor score 1 : 2 patients, motor score 0 : 4 patients).

Of the 91 patients investigated, 89 left the operating theatre without a neurological deficiency. Two patients who showed a contralateral paresis (pathophysiologically, ischaemic damage can be inferred) recovered completely in the postoperative period.

Secondary end-points

A postoperatively acquired deficiency was seen in six patients. Four patients recovered spontaneously within 24 h. One patient developed a complete hemiparesis 12 h after the operation. The pathophysiological cause was arterio-arterial embolism from the area of surgery proven by a computed tomography angiography (CTA) scan, showing a complete hemispheric infarction, leading to a general brain oedema. The patient deteriorated rapidly, brain-protective interventions were ineffective and the patient died. One further patient suffered from progressive stroke induced by postoperative embolism with hemiparesis that was most pronounced in the arm, proven as a postoperative occlusive event by a CTA scan.

The rate of hospital discharge was 90 out of 91 patients, of whom six were referred to a geriatric rehabilitation facility (not for neurological complications) and one patient was referred for neurological rehabilitation. The overall hospital morbidity and mortality rate was 2.0%.

Validity of SSEP-generated shunt criteria

Amplitude reduction of the N20/ P25-waveform >50% [7,8,10,13] as a marker of critical reduction of cerebral blood flow showed a false-positive indication for shunting in 31 points (34%), whereas 11 cases (13%) of clinically evident reduced CBF were not detected by this parameter. Even the definition of critically reduced CBF after the total loss of N20/ P25 [3,4,6] led to a false alarm in 10 cases and 15 of the clinically relevant CBF-reduction episodes were undetected.

A prolongation of CCT [10,12,13] by greater than 20%, defining the threshold of critical CBF, was exceeded in 55 points, although mental state remained unchanged. In 12 of the cases CCT did not exceed the critical threshold despite significant neurological deterioration. The NSI [14] exceeded the threshold of 0.5 in 11 patients with a clinically uneventful operational course. In 14 patients with clinically reduced CBF, the NSI did not exceed 0.5. None of the current literature-suggested SSEP-generated shunt criteria reflect the mental state confidently and sufficiently as shown in Figures 2 and 3, Tables 4 and 5.

Figure 2.
Figure 2.:
Reciever-operater characterisic curves of the investigated shunt criteria.
Figure 3.
Figure 3.:
Number of patients with false-negative and false-positive shunt indication.
Table 4
Table 4:
Validity of investigated SSEP parameters.
Table 5
Table 5:
Results of ROC curve analysis.


The reduction of perioperative morbidity and mortality rate in carotid surgery can be achieved effectively only by assessment of perioperative cerebral perfusion and by the use of intraluminal shunts according to the indication. A basic prerequisite for this is separation of the patient groups with sufficient and insufficient collateral blood flow. Routine use of an intraluminal shunt entails the danger of embolization of plaques or the creation of intimal flaps with apposition thrombi. Selective shunt indication has thus become widely accepted. Therefore it is necessary to improve the intraoperative monitoring for cerebral ischaemia.

The methodological basis of current neuromonitoring is a model developed by Shapiro [18] in 1978. This model correlates neuronal integrity and electrical activity to the CBF. Using this approach to interpret changes in SSEP tracings as a perfusion-dependent neuronal metabolic situation, different parameters and values have been established to detect a relevant reduction in the CBF. To date, most of the studies using SSEP monitoring in carotid surgery have highlighted three important characteristics:

  • SSEP exhibits very low susceptibility to technical interference in the operating room;
  • The reliable interpretation of measured parameters allows precise definition of threshold values, marking the onset of critical reduction of CBF and thus definitive response;
  • With regard to the sensitivity and specificity of these studies, SSEP monitoring is a reliable method of determining cerebral function during CEA.

In reviewing the reported studies, a substantial variety of parameters and thresholds related to these parameters were used to detect critical deteriorations of CBF. The thresholds seem to be arbitrary, as defining an exactly 50% reduction of the primary cortical complex N20/P25 (Fig. 1a) as critical and in the case of amplitude reduction vs. total loss of primary cortical response, are even contradictory (Fig. 1b). As every study claims high reliability of the results presented, there remains a simple question related to practical considerations: Which one of the suggested parameters is the most reliable to be used for detecting cerebral ischaemia?

Answering this question requires an anaesthesiological technique, which allows direct correlation between clinical parameters and neurophysiological data. Trials using general anaesthesia are inappropriate.

Surgical data revealed a rate of shunt insertion of 22%. Initially this seems to be disappointing in comparison to shunt frequencies of less than 10% in the trials performed under general anaesthesia. However, after taking into consideration the rate of false-negative results from each of the investigated parameters in our study, it is obvious that the rate of undetected critical reductions of CBF might be more important under general anaesthesia.

On the other hand, it has to be mentioned that probably not every neurological deterioration with carotid clamping results in a new neurological deficit. This may be more a matter of clamping duration and therefore of ischaemic alteration despite the evident malfunction during the clamping. Compared with other studies, this study has different results.

Markand and colleagues [11] investigated 28 patients under regional anaesthesia to evaluate SSEP and its usefulness in CEA and found a sufficient correlation to clinical examination in a very small number of cases. Sbarigia and colleagues [19] investigated 50 patients under regional anaesthesia in a similar approach and defined the critical CBF reduction at an amplitude reduction of N20/P25 after clamping of the internal carotid artery at 30% with a specificity of 100% and a sensitivity of 89%. We have demonstrated a much lower sensitivity and specificity at 50% reduction. This may be based on differences in the patients studied. We saw 70% of our patients with neurological symptoms, 30% of them in an unstable neurological state for urgent surgical treatment. These patients may be more compromised by CBF reduction compared to patients with asymptomatic findings. Regarding the number of patients involved in the preliminary report of the study by Sbarigia and colleagues, only one patient missed by their shunt criterion led to a significant reduced sensitivity. One has to wait for the final results, as a larger number might show a different validity.

Dinkel and colleagues demonstrated that only the amplitude loss (Fig. 1a) indicates an endangered cerebral perfusion. This SSEP criterion could be evaluated as having 100% sensitivity and specificity [4,6]. We cannot confirm the diagnostic efficacy of the marker ‘loss of amplitude'. Our results revealed a sensitivity of only 85% and a specificity of 89% (Fig. 3).

With a reduction of CBF to less than 20 mL 100 g−1 min−1, Shapiro and colleagues always observed loss of the primary cortical response of the median nerve SSEP defining the CBF threshold from which the functional metabolism is reduced to a mere substrate metabolism. This CBF is associated with loss of vigilance, motor and sensory activity. In lieu, we observed 10 patients who showed no changes in mental performance during the entire clamping phase, but showed a total loss of the primary cortical complex at N20/P25, which persisted for several minutes. Technical disorders have been excluded. Therefore this finding does not support the thesis that complete loss of the N20/P25-waveform was associated with a dramatic reduction from functional to a stage of substrate metabolism.

In 14 cases there was immediate loss of consciousness during trial clamping. Consciousness returned with insertion of a shunt. Complete loss of primary cortical response, alerting the deterioration of CBF, was not detected. An immediate loss of consciousness with maintained electrical responsiveness can be interpreted as a rapid deterioration of functional metabolism whereas substrate metabolism is maintained in the perfusion range around 20 mL 100 g−1 min−1. This loss of consciousness occurs briefly without signs of electrical silence. Explaining this observation because shunt insertion had been performed so quickly, that an adequate electrophysiological reaction simply could not develop, entails the questionable delay of detection and its importance for neurological damage.

Taking into account the mathematical basis for acquisition of SSEP-tracings, the Fast-Fourier transformation of at least 250 traces could be considered as a systematic disadvantage in the timely detection of impaired CBF.

Furthermore, we saw differences in the current interpretation of the CCT. Pozzessere and colleagues [12] and Russ and colleagues [13] suggested the CCT as a reliable correlate for critical reduction of the CBF. Even the correlation of CCT with the observed clinical examination was poor in our patients, revealing a sensitivity of 87% and a specificity of 40%.

The NSI, as recommended by Fava and colleagues [14], combines the parameters of amplitude and latency of the primary cortical complex N20/P25 as a pathophysiological consequence of the impairment of neuronal recruitment, the nerve impulse conduction velocity or both in relation to the CBF. With the definition of the NSI of more than 0.5 as an endangered CBF, they were able to define the marker with a sensitivity of 100% at a specificity of 100%. We observed 11 patients who fulfilled Fava's shunt indication with a calculated value >0.5, not showing changes of vigilance, motor activity and sensory activity. In 14 patients, the parameter remained lower than the necessary threshold value of 0.5 during clamping despite rapid neurological deterioration.

The gold standard in monitoring cerebral function is an awake patient, who can provide all the information about perfusion-dependent abilities and its clinical relevance. Compared with clinical assessment, the shortcomings of SSEP monitoring were apparent in our study as its strict limitation in monitoring cerebral functions to sensory inputs. It failed to detect six patients with clamping-correlated paresis of the contralateral hand or facial palsy. Similar observations have been reported in the literature. Haupt and colleagues [20] reported on a patient who was operated on for symptomatic carotid stenosis under general anaesthesia and was monitored with median nerve SSEP. Although the neurophysiological data remained unchanged (no reduction of the amplitude N20/P25, no increase of CCT), the patient showed a contralateral hemiparesis, aphasia and apraxia immediately after the operation. Erasmi and colleagues [21] reported two false-negative results.

It should be stressed that the most common cause of cerebral infarction is postoperative arterial embolism, not ischaemic perfusion during surgery. This was evident in our patients who developed neurological deficits postoperatively, remarkably all of them presenting neurological intolerance to cross-clamping of the CCA. This finding is supported by Mayer and colleagues [22], who identified patients intolerant to cross-clamping the CCA being at higher risk for postoperative stroke and mortality. The use of SSEP-generated shunt criteria for monitoring carotid surgery continues to be controversial, as shown by the large variability of the investigation results described in the literature [23-25].

Studies of larger numbers of patients like the ongoing GALA trial might clarify whether there are significant differences in the rate of neurological and cardiac complications between patients receiving general and local anaesthesia. The cerebroprotective potential of general anaesthesia in carotid surgery remains speculative even today, since there is no scientific data on the pharmacological benefit or measures improving perfusion. For the moment there remains only the conclusion from a meta-analysis of 15 studies on SSEP monitoring in carotid surgery by Wöber and Zeitlhofer [26]. This reveals that selective shunting on the basis of intraoperative SSEP monitoring does not improve the postoperative neurological prognosis compared to routine use of a shunt or dispensing with a shunt.


Support was solely provided from departmental resources, no special financial support was granted.


1. Gumerlock MK, Neuwelt EA. Carotid endarterectomy: to shunt or not to shunt. Stroke 1988; 19: 1485–1490.
2. Cheng MA, Theard MA, Tempelhoff R. Anaesthesia for carotid endarterectomy: a survey. J Neurosurg Anaesthesiol 1997; 9: 211–216.
3. Beese U, Langer H, Lang W, Dinkel M. Comparison of near-infrared spectroscopy and somatosensory evoked potentials for the detection of cerebral ischemia during carotid endarterectomy. Stroke 1998; 29: 2032–2037.
4. Dinkel M. Stellenwert des EP-monitorings für die Gefäβchirurgie. Anaesthesiol Intensivmed Notfallmed Schmerzther 1997; 32 (Suppl. 2): 215–219.
5. Kochs E. Zerebrales monitoring. Anaesthesiol Intensivmed Notfallmed Schmerzther 1991; 26: 363–374.
6. Dinkel M, Kamp H-D, Schweiger H. Somatosensorisch evozierte potentiale in der karotischirurgie. Anaesthesist 1991; 40: 72–78.
7. Gigli GL, Caramia M, Marciani MG, Zarola F, Lavaroni F, Rossini PM. Monitoring of subcortical and cortical somatosensory evoked potentials during carotid endarterectomy: comparison with stump pressure levels. Electroencephalogr Clin Neurophysiol 1987; 68: 424–432.
8. Haupt WF, Horsch S. Evoked potential monitoring in carotid surgery: a review of 994 cases. Neurology (Minneap.) 1992; 42: 835–838.
9. Kearse LA Jr, Brown EN, McPeck K. Somatosensory evoked potentials sensitivity relative to electroencephalography for cerebral ischemia during carotid endarterectomy. Stroke 1992; 23: 498–505.
10. Lam AM, Manninen PH, Ferguson GG, Nantau W. Monitoring electrophysiological function during carotid endarterectomy: a comparison of somatosensory evoked potentials and conventional electroencephalogram. Anaesthesiology 1991; 75: 15–21.
11. Markand ON, Warren CH, Moorthy SS, Dilley R. Monitoring of somatosensory evoked responses during carotid endarterectomy. Arch Neurol 1984; 41: 375–378.
12. Pozzessere G, Valle E, Santaro A, Delfini R. Prognostic value of early somatosensory evoked potentials during carotid surgery: relationship with electroencephalogram, stump pressure and clinical outcome. Acta Neurochir (Wien) 1987; 89: 28–33.
13. Russ W, Fraedrich G, Hehrlein FW, Hempelmann G. Intraoperative somatosensory evoked potentials as a prognostic factor of neurologic state after carotid endarterectomy. Thorac Cardiovasc Surg 1985; 33: 392–396.
14. Fava E, Bortolani E, Ducati A, Shieppati M. Role of SEP in identifying patients requiring temporary shunt during carotid endarterectomy. Electroenceph Clin Neurophysiol 1992; 84: 426–432.
15. Moore DC. Regional block. A handbook for use in the clinical practice of medicine and surgery, 4th edn. Springfield, IL: Charles C. Thomas, 1978: 112–22.
16. Moore WS, Barnett HJM, Beebe HG et al. Guidelines for carotid endarterectomy: a multidisciplinary consensus statement from the Ad Hoc Committee, American Heart Association. Circulation 1995; 91: 566–579.
17. North American Symptomatic Carotid Endarterectomy Trila Collaborators. Beneficial effect of carotid endarterectomy in symptomatic patients with high-grade carotid disease. New Engl J Med 1991; 325: 445–453.
18. Shapiro HM, Aidinis SJ. Neurosurgical anesthesia. Surg Clin North Am 1975; 55: 913–928.
19. Sbarigia E, Shioppa A, Misuraca M. Somatosensory evoked potentials versus locoregional anaesthesia in the monitoring of cerebral function during carotid artery surgery: preliminary results of a prospective study. Eur J Vasc Endovasc Surg 2001; 21: 413–416.
20. Haupt WF, Erasmi-Körber H, Lanfermann H. Intraoperative recording of parietal SEP can miss hemodynamic infarction during carotid endarterectomy: a case study. Electroencephalogr Clin Neurophysiol 1994; 92: 86–88.
21. Erasmi H, Walter M, Löhr G. Läβt sich die sicherheit des SEP-monitorings während der carotis-chirurgie verbessern? Langenbecks Arch Chir 1997; 114 (Suppl. II): 1293–1294.
22. Mayer RC, Bingley J, Westcott MJ et al. Intraoperative neurological changes in 1665 regional anaesthetic carotid endarterectomies predicts postoperative stroke. ANZ J Surg 2007; 77: 49–53.
23. Ackerstaff RG, van de Vlassakker JW. Monitoring of brain function during carotid endarterectomy: an analysis of contemporary methods. J Cardiothorac Vasc Anesth 1998; 12: 341–347.
24. De Vleeschauer P, Horsch S, Matamaros R. Monitoring of somatosensory evoked potentials in carotid surgery: results, usefulness and limitations of the method. Ann Vasc Surg 1988; 2: 63.
25. Holland NR. Subcortical strokes from intracranial aneurysm surgery: implications for intraoperative neuromonitoring. J Clin Neurophysiol 1997; 15: 439–446.
26. Wöber C, Zeitlhofer J, Asenbaum S et al. Monitoring of median nerve somatosensory evoked potentials in carotid surgery. J Clin Neurophysiol 1998; 15: 429–438.


© 2008 European Society of Anaesthesiology