Accuracy of Cerebral Monitoring in Detecting Cerebral Ischemia during Carotid Endarterectomy

Abstract

Background This study compares the accuracy of cerebral monitoring systems in detecting cerebral ischemia during carotid endarterectomy. Methods The authors compared transcranial Doppler sonography (TCD), near-infrared spectroscopy (NIRS), stump pressure (SP) measurement, and somatosensory evoked potentials (SEP) in 48 patients undergoing carotid surgery during regional anesthesia. Cerebral ischemia was assumed when neurologic deterioration occurred. During clamping, the minimum mean middle cerebral artery velocity (TCD(min)), its percentage change (TCD%), the minimum regional saturation of oxygen (NIRS(min)), its percentage change (NIRS%), the mean SP, and the changes of SEP amplitude were recorded. To analyze the corresponding sensitivity and specificity of each parameter, the authors performed receiver operating characteristic analysis. Results Neurologic deterioration occurred in 12 patients. SP and NIRS were successfully performed in all patients. TCD monitoring was not possible in 10 (21%); SEP was not possible in 2 patients (4%). All parameters provided the ability to distinguish between ischemic and nonischemic patients. TCD% and NIRS% showed significantly better discrimination than TCD(min) and NIRS(min) (P < 0.05). The highest area under the curve (AUC) was found for TCD% (AUC = 0.973), but there was no significant difference compared with NIRS% (AUC = 0.905) and SP (AUC = 0.925). The lowest AUC was found for SEP (AUC = 0.749), which was significantly lower than that for TCD%, NIRS%, and SP. Conclusions TCD%, NIRS%, and SP measurement provide similar accuracy for the detection of cerebral ischemia during carotid surgery. Lower accuracy was found for SEP monitoring. Because of the high rate of technical difficulties (21%), TCD monitoring was the least practical of the investigated monitoring devices.