Intraoperative assessment of cerebral ischaemia during carotid surgery

This study aimed to evaluate the reliability of somatosensory evoked potentials (SEPs) in identifying clamping ischaemia during carotid surgery under general anaesthesia.

We reviewed the records of 196 consecutive carotid endarterectomies (CEA) performed under general anaesthesia, out of 1550 patients operated on between 1975 and 1993. SEPs were monitored after contralateral median nerve stimulation at the wrist in all patients. Moreover they received an intravenous bolus of 2500 IU of heparin and the stump pressure was measured. A completion angiography was performed in all patients. An intraluminal shunt was inserted when the amplitude of the N20-P25 SEPs complex decreased by more than 50% of pre-clamping value (pathologic SEPs) or when a preoperative CT-scan showed an ischaemic area in the contralateral hemisphere. In some patients the shunt was not inserted because of technical difficulties or because the pathological SEP complex decrease developed when the suture of the arteriotomy was almost complete.

As the preclamping amplitude (Ab) of N20-P25 was extremely variable, ranging from 0.9 and 7.5 microV, we adopted the ratio: deltaA=(At-Ab)/Ab (At=amplitude measured during clamping every 2 min at time t). Mean decrease of deltaA measured in the whole group ranged between 15% and 20%, but mean deltaA values in patients with pathological SEPs were around 60%. Even patients with positive CT-scans or with a back-pressure lower than 50 mmHg developed a mean deltaA decrease of about 20%. A pathological SEP was present in 50 patients, but developed in only 12 during the clamping trial. An intraluminal shunt was inserted in 22 cases; the other patients received pharmacological treatment with anaesthetic or vasoactive drugs. Among the patients with pathological SEPs, 3 presented postoperative symptoms related to carotid cross-clamping; only 1 of these was shunted but without SEPs normalisation. No patients with normal SEPs developed a postoperative neurological deficit, giving a reliability of 86.7%, a sensitivity of 100% and a specificity of 86.5%.