Intraoperative Seizures in Cardiac Surgical Patients Undergoing Deep Hypothermic Circulatory Arrest Monitored with EEG

From January 1995 through February 2000, the records of all patients scheduled to undergo intraoperative encephalographic monitoring at the Hospital of the University of Pennsylvania, Philadelphia, Pennsylvania, were entered into a research database in a protocol approved by the University of Pennsylvania Institutional Review Board. A retrospective review of this database was performed to determine the frequency of electrographic seizures on the scalp electroencephalogram. In most cases, a 16-channel electroencephalogram was recorded with gold cup electrodes placed with the collodion technique according to the standard international 10–20 system of electrode placement. In many patients undergoing surgery for intracerebral aneurysms, the F8, F4, F7, and F3 electrode locations could not be monitored because they would have interfered with the surgical field. In some emergent cases, sterile subdermal electrodes were used, or a reduced number of recording electrodes were used to record the scalp electroencephalogram.

The criteria used to define seizure activity on the electroencephalographic recordings were based on ictal patterns on the electroencephalogram that have been established to correlate with clinical seizures in epileptic patients. 3–4Electroencephalographic seizure patterns included generalized spike and wave activity, repetitive rapid spiking that changed in frequency and location over 20–300 s, or focal or generalized buildup of fast activity that evolved over time into generalized spike discharges. The diagnosis of electroencephalographic seizure was made only if one of the established electroencephalographic seizure patterns evolved continuously in frequency and amplitude over more than 20 s. This definition was used to distinguish electroencephalographic seizure activity from interictal electroencephalographic patterns, such as isolated, randomly occurring focal spikes or sharp waves that can sometimes be observed during general anesthesia. All electroencephalographic recordings were interpreted by a board-certified electroencephalographer.

All thoracic aortic operations (table 1) were performed using cardiopulmonary bypass (CPB) and deep hypothermic circulatory arrest with retrograde cerebral perfusion according to a standardized protocol used at the University of Pennsylvania. 5–6In brief, deliberate hypothermia was induced by keeping the temperature gradient between the blood and heat exchanger in the CPB circuit at 10°C or lower. Arterial blood gas tensions were managed using alpha-stat. Deep hypothermic circulatory arrest was instituted after 3 min of sustained electrocerebral silence by electroencephalography. Retrograde cerebral perfusion was performed with oxygenated blood infused into the superior vena cava at a pressure of 25 mm Hg or less. Rewarming after deep hypothermic circulatory arrest was started gradually after reinstitution of antegrade cerebral perfusion on CPB by keeping the temperature gradient between the blood and heat exchanger 10°C or lower without allowing the nasopharyngeal temperature to exceed 37.5°C.

Intraoperative seizures were identified in 4 of a total of 914 patients who had intraoperative electroencephalographic monitoring (table 1).

The patient was a 62-yr-old man presenting for repair of a descending aortic dissection. The patient had experienced transient confusion after the repair of a previous Stanford type A aortic dissection. He had a history of hypertension and cocaine use but no previous seizures. Versed, fentanyl, and pancuronium were used to induce general anesthesia. Anesthesia was maintained with fentanyl and isoflurane, 0.3–0.5%. The preinduction and postinduction electroencephalographic recordings showed diffuse slowing but no focal or epileptiform abnormalities. Three minutes after initiation of CPB, lidocaine, 200 mg intravenous bolus, was administered into the venous reservoir of the bypass circuit without incident. The patient was cooled to a nasopharyngeal temperature of 14.5°C, during which time the electroencephalogram showed typical changes associated with cooling (fig. 1A) before the appearance of electrocerebral silence. The patient underwent 57 min of hypothermic circulatory arrest with retrograde cerebral perfusion. After reinstitution of CPB and active rewarming, continuous electroencephalographic activity was seen 55 min later. Shortly thereafter (pH, 7.37; arterial carbon dioxide tension (Paco²), 46 mm Hg; arterial oxygen tension (Pao²), 209 mm Hg; isoflurane concentration, 0.3%; nasopharyngeal temperature, 32°C), generalized spikes were seen on the electroencephalogram, followed by generalized seizure activity on the electroencephalogram (fig. 1B). There was no response to 5 mg intravenous versed, but 150 mg intravenous thiopental terminated the seizures. Lidocaine, 200 mg, was administered later in the procedure without recurrence of seizures. No further seizures were seen, despite electroencephalographic monitoring at bedside in the surgical intensive care unit. The patient experienced transient postoperative confusion but eventually recovered. Six hours after the last dose of lidocaine, lidocaine concentration was 1.3 μg/ml.

The patient was a 61-yr-old man with a history of hypertension and smoking who presented for an acute Stanford type A dissection. He underwent graft reconstruction of the ascending aorta, aortic valve resuspension, and hemiarch replacement. Fentanyl, pancuronium, and versed were used to induce general anesthesia. Baseline electroencephalographic recording showed rare left temporal sharp waves. After CPB was initiated, active cooling was begun. Before ventricular fibrillation, 200 mg lidocaine, 2 g MgSO⁴, and 250 mg methylprednisolone were infused into the venous reservoir of the bypass circuit. Within 2 min, at a nasopharyngeal temperature of 31°C and an inhaled isoflurane concentration of 0%, left and right temporal sharp waves appeared. Generalized electrographic seizure activity, which was most prominent over the left temporal region, followed (fig. 2) and then terminated spontaneously, although right and left temporal sharp waves persisted even after the administration of 100 mg intravenous thiopental. The plasma lidocaine concentration 20 min after the bolus was 3.3 μg/ml. After hypothermic circulatory arrest, the electroencephalogram returned to normal, although left temporal sharp waves persisted. There were no postoperative seizures or neurologic deficits at follow up.

The patient was a 50-yr-old man with a history of an anterior wall myocardial infarction, coronary artery disease, congestive heart failure, and pulmonary emboli. Preoperatively, the patient was confused, but neurologic examination was nonfocal. Baseline electroencephalographic recording showed diffuse slowing but no epileptiform abnormalities. The patient underwent pulmonary thromboembolectomy using hypothermic circulatory arrest with retrograde cerebral perfusion. After 6 min of initiating cooling during CPB, 200 mg intravenous lidocaine, along with 1 g intravenous methylprednisolone and 1 g intravenous MgSO⁴, was administered into the venous reservoir. Within 2 min, electrical seizure activity appeared bilaterally and resolved quickly after the administration of 50 mg intravenous thiopental. The seizure occurred during CPB soon after the initiation of deliberate hypothermia at a nasopharyngeal temperature of 31°C with an inhaled isoflurane concentration of 0%. No further seizures were observed. Postoperatively, the patient remained confused.

The patient was a 74-yr-old woman with a large leaking distal aortic arch and a Crawford type II thoracoabdominal aortic aneurysm. Surgery was performed emergently. After induction of general anesthesia, the electroencephalogram was symmetric. CPB with active cooling was initiated. Lidocaine, 200 mg, was administered intravenously into the CPB circuit when the nasopharyngeal temperature was 34.3°C and the inhaled isoflurane concentration was 0%. This was followed by generalized electrographic seizure activity. The seizure activity was terminated with 5 mg intravenous versed. Hypothermic circulatory arrest with retrograde cerebral perfusion was instituted and lasted 40 min at a nasopharyngeal temperature of 11°C. The electroencephalogram recovered normally, and the patient had no postoperative neurologic deficits.

The overall incidence of seizures during surgical procedures in a selected population of patients undergoing operations with electroencephalographic monitoring was 0.4%. All cases of intraoperative electroencephalographic seizure activity occurred in a high-risk subgroup of cardiac surgical patients undergoing thoracic aortic reconstruction with hypothermic circulatory arrest. No intraoperative seizures were observed in the noncardiac surgery patient population.

The case studies of patients with intraoperative seizures suggested two specific risk factors for intraoperative seizures. One factor was the administration of high dose lidocaine boluses into the CPB circuit. In three of the four cases, there was a temporal relation between the administration of lidocaine into the CPB circuit and the onset of seizures. Lidocaine has been a recognized cause of drug-induced seizures and was the cause of approximately 6% of all drug-induced seizures in one study. 7–8In studies of regional anesthesia with local anesthetics, seizures occurred in 0.025–0.1% of patients. 9–10A higher frequency of seizures in the cardiac surgical population may be explained by the rapid transient of lidocaine directly infused into the arterial system. Although the pharmacokinetics of intravenous lidocaine have not been shown to be affected by CPB, direct bolus administration of the drug into the venous reservoir of the CPB circuit may not have allowed time for redistribution, equilibration, or metabolism, resulting in excessively high intraarterial lidocaine concentrations via  the arterial limb of the perfusion circuit. 11Direct exposure of the brain to high arterial lidocaine concentrations may have precipitated the seizures. In fact, one case report showed that as little as 40 mg infused directly in the carotid artery caused electroencephalographic slowing and the appearance of epileptiform discharges. 12However, in the awake patient, as little as 5 mg lidocaine administered into the carotid artery was sufficient to cause a convulsion. 13 

The second important factor predisposing patients to intraoperative seizures was preexisting neurologic injury. At least three of the patients who had intraoperative seizure activity had some evidence of neurologic injury before the seizure. The patient in case 1 had a previous aortic surgical procedure involving deep hypothermic circulatory arrest that was complicated by prolonged confusion. The patient in case 2 had left temporal sharp waves on the baseline electroencephalographic trace. The patient in case 3 had preoperative confusion. In addition to preexisting neurologic injury, the recognized risk of neurologic injury during surgery is high in thoracic aortic operations and pulmonary thromboembolectomies that require hypothermic circulatory arrest. 10–11The presence of intraoperative neurologic injury was unlikely to be the only factor causing intraoperative seizures because seizures were not detected in any of the neurosurgical patients undergoing aneurysm surgery, many of whom had preexisting neurologic injury or were at high risk of having at least transient intraoperative brain injury. The anesthetic technique also may have been a factor because the inhaled anesthetic concentration was low or 0 at the time of the event in the majority of patients who had intraoperative seizures.

In conclusion, the incidence of intraoperative seizures was rare and too infrequent to implicate the postictal state as a cause for postoperative confusion. Bolus administration of lidocaine into the CPB circuit in patients at risk for neurologic injury seemed to explain the majority of intraoperative seizures that were observed. Slow infusion of lidocaine into the CPB circuit or intravenous administration of the drug may decrease the risk of intraoperative seizures.