Case reports suggesting both pro- and anticonvulsant effect(s) of propofol have been published in recent years. The effects of sedative doses of propofol on epileptiform activities in patients suffering from intractable partial epilepsy were systematically investigated.
Fourteen patients suffering from complex partial seizures were studied. Electroencephalogram (EEG) was recorded from intracranial electrodes implanted in the hippocampi and temporal neocortex. Propofol was given as a computer-controlled infusion in four steps to achieve target plasma propofol concentrations of 0.3, 0.6, 0.9, and 1.2 micrograms/ml. Each concentration was maintained for 30 min, and steady-state kinetics were confirmed by blood levels drawn at 10th and 30th min at each level. Between the 15th and 30th min of each concentration of propofol, EEG was analyzed for presence of electroencephalographic seizure activity and/or number of interictal spikes (IIS). Effects of propofol on IIS frequency at different electrode sites were compared using a two-way repeated measures analysis of variance. A value of P < 0.05 was considered significant.
None of the patients developed a seizure during the 2 h of propofol infusion. No "false spikes" (spikes developing with propofol infusion in areas where no spikes were seen in baseline EEG) were seen. Although effects of propofol on IIS frequency were highly variable across patients and at different propofol doses in the same patient, there was no statistically significant effect of propofol on any electrode site with any of the doses studied.
We were unable to demonstrate a significant change in epileptiform activity with sedative doses of propofol in patients suffering from complex partial epilepsy.
Address reprint requests to Dr. Samra: Department of Anesthesiology, University of Michigan Medical Center, 1G323 University Hospital, 1500 East Medical Center Drive, Ann Arbor, Michigan 48109-0048.
Key words: Anesthetics, intravenous: propofol. Epilepsy: seizures. Monitoring: electroencephalogram.
PROPOFOL is a relatively new intravenous anesthetic with pharmacokinetics that provide rapid induction and recovery from anesthesia. Initial animal studies suggested that propofol is a pure hypnotic with little or no anti- or proconvulsant activity. More recently, several contradictory reports of both proconvulsant and anticonvulsant events associated with propofol anesthesia in humans have been published. [2-7]Association of propofol anesthesia with involuntary movements without clectroencephalographic (EEG) changes also has been reported. In the clinical reports suggesting an epileptogenic effect, [2,3]propofol was one of the drugs used along with opioid fentanyl or alfentanil. In one report, in four of five cases, the only evidence of seizures was development of tonic-clonic movements and documentation of EEG seizure activity was not available to confirm the supposed epileptic nature of abnormal movements. Thus, nonepilcptogenic extra-pyramidal motor dysfunction and other causes of abnormal movements could not be excluded. Reports suggestive of anticonvulsant effect of propofol [4-7]also are based on anecdotal case reports and suggest the need for prospective clinical studies. in a recent investigation of the specific response of the epileptic focus to propofol anesthesia, the effect of one or two bolus injections with rapidly changing blood levels of propofol on EEG was studied. The current study reports the effect of a range of sedative doses of propofol on interictal spike (IIS) counts both at the site of ictal onset and in nonepileptogenic regions of the brain in patients with intractable partial epilepsy.
The study protocol was approved by the University of Michigan Institutional Review Board. Fourteen consenting adult patients (six men, eight women) diagnosed with intractable partial epilepsy were recruited to participate in this study. Criteria for candidacy for epilepsy surgery at our institution were reported previously. These patients' epilepsy was considered resistant to medical management, and they were declared candidates for surgical excision of their epileptic focus. All patients had bilateral, stereotactically directed recording electrodes implanted under general anesthesia. The sites of electrode placement were determined based on review of data from preoperative interictal and ictal scalp-recorded EEG, magnetic resonance imaging (MRI), and fluorodeoxyglucose positron emission tomography. Table 1shows the distribution of electrode placement sites in the 14 patients. Each patient had multiple electrodes in both hippocampi, with one electrode pair (with 1 cm space between the electrodes) in the anterior and one pair in the middle hip-pocampal region selected for EEG analysis. Twelve patients had subdural electrodes over the lateral temporal neocortex, with one pair (spaced 1 cm apart) selected for EEG analysis. The remaining two patients had other sites selected such that the site of seizure onset was represented among the analyzed electrode pairs. Thus, the assumed site of seizure onset was always included in the set of electrode pairs used for EEG analysis. Position of the electrodes in different parts of the brain was confirmed by MRI, and the selected pairs of electrodes were confirmed to be symmetrically placed.
Postoperatively, EEG was continuously recorded with Telefactor Beehive Unit (200 Dehaven Street, West Conshohocken, PA) and stored on a magnetic tape for later analysis. A video camera simultaneously monitored the patient for clinical manifestations of epilepsy in the form of complex partial seizures with or without secondarily generalized convulsive activity. A minimum of 48 h was allowed for recovery from the effects of general anesthesia before propofol infusion for this study was started. If a patient experienced a seizure during this waiting period, then a minimum of 8 h was allowed before the beginning of the study.
At the beginning of study period, a 30-min epoch of baseline EEG was recorded, and a blood sample for determination of anti-epileptic drug (phenytoin or carbamazepine) and propofol concentrations was drawn immediately before starting the experiment. Computer-controlled propofol infusion was then started using the STANPUMP program.* STANPUMP drives an infusion pump to administer anesthetic drugs according to a three-compartment pharmacokinetic model. STANPUMP (Shafer and Dyck age- and weight-adjusted model) was used to achieve and maintain target plasma propofol concentrations of 0.3, 0.6, 0.9, and 1.2 micro gram/ml in four steps. Each concentration was maintained for 30 min. Blood samples were obtained after 10 and 30 min of infusion at each concentration to ascertain a steady drug level. Level of sedation achieved in each patient was graded using Ramsay sedation score as follows: Awake levels: 1 = patient anxious and agitated or restless or both; 2 = patient cooperative, oriented, and tranquil; 3 = patient responds to commands only. Sleep levels were graded depending upon patient's response to light glabellar tap or loud auditory stimulus: 4 = a brisk response, 5 = a sluggish response, and 6 = no response. Sedation score was noted at the time of obtaining each blood sample for propofol concentration.
Patient's electrocardiogram, heart rate, and hemoglobin oxygen saturation (SpO2) were continuously monitored, and blood pressure was monitored at 5-min intervals using a Spacelabs portable bedside monitor (PC Express model 90308, Spacelabs Medical, Redmond, WA) throughout the propofol infusion. Fifteen-minute epochs of EEG recorded between the 15th and 3Oth min of each propofol concentration were analyzed EEG recordings were examined for evidence of seizure activity and HS. To be counted as a definite IIS, a sharply contoured wave was required to have a duration of at least 200 ms and an amplitude of 2-3 times the maximum amplitude of background alpha and beta frequencies recorded at the particular electrode site. Number of IIS was counted, by a qualified electroencephalographer, at each selected electrode pair over this 15-min period during each dose of propofol. Numeric data for IIS counts were subjected to a two-way repeated measures analysis of variance using STATISTICA, version 4.0 software (StatSoft, Tulsa, OK). Propofol dose and the electrode site were used as the two repeated measures factors, and a value of P < 0.05 was considered statistically significant.
All patients successfully completed the study. There was no clinically observed total ictal epileptiform activity either in the form of localized or generalized electroencephalographic or behavioral seizure during the period of propofol infusion in any patient. Six patients had a seizure the night before the study. Five of these patients also had a seizure within 24 h after completion of the study. In the sixth patient, electrodes were removed soon after completion of the study; therefore, EEG was not subsequently recorded. This patient did not have any clinical evidence of seizure activity for the next 24 h. All patients who experienced no seizures for 24 h before propofol infusion remained free of any seizure activity for at least 24 h after administration of propofol. All patients maintained ventilation and Sp sub O2at the baseline level throughout the study period. Three patients required support of airway by chin lift during deeper levels of sedation. In one patient, we omitted the infusion of 1.2 micro gram/ml target level because he had a sedation score of 6 after 0.9 micro gram/ml target infusion. Blood concentrations (mean plus/minus SD) of propofol achieved at different infusion rates are shown in Figure 1. Overall blood levels achieved were close to target levels, i.e., 0.22, 0.64, 0.97, and 1.32 micro gram/ml at 0.3, 0.6, 0.9, and 1.2 micro gram/ml target plasma concentrations, respectively. Individual variations among patients are indicated by standard deviation bars in figure 1, which shows an overall steady blood concentration of propofol achieved at each dose level between 10th and 30th min. the period of EEG analysis.
Hippocampus was the site of ictal onset in 10 of 14 patients (Table 2), the origin being from the left side in 2, from the right side in 3, and bilateral (seizure originating independently in each hippocampus) in 3 patients. As mentioned above, all of our patients had been given a trial of maximum medical treatment before the implantation of electrodes; therefore, they had been taking either phenytoin or carbamazepine for several months. The blood concentration of antiepileptic drugs was less than the usual therapeutic level in nine, at therapeutic level in one, and greater than therapeutic level in four patients on the day of the experiment (Table 2, Table 4).
Because none of the patients had a seizure, we studied the effect of propofol on 115 frequency in detail. Effects of propofol on IIS counts were highly variable. We were unable to define any trends in change in spike counts on the basis of either the dose of propofol, site of ictal onset, antiepileptic drug level, or the clinical level of sedation achieved with propofol. It was observed that spike counts often increased with 0.3 micro gram/ml target infusion, and decreased at 0.6 and/or 0.9 micro gram/ml, and again increased at 1.2 micro gram/ml dose. This is diagramatically shown in Figure 2, Table 4and Figure 3, which display mean dose response of spike counts from hippocampal (Figure 2) and temporal lobe (Figure 3) electrodes in 14 patients. Table 2, Table 4shows the effect of different doses of propofol on spike counts per 15-min block of EEG in 14 patients. Data for sedation scores, blood levels of propofol and antiepileptic drug (phenytoin or carbamazepine), and site of ictal onset in each patient also are included. Statistical analysis revealed no significant effect of dose of propofol (P = 0.81) or electrode site (P = 0.15), and there was no significant interaction between the propofol dose and electrode location (P = 0.81).
We also analyzed the possible association of variability in seizure frequency and recentness with variable increased, unchanged, or decreased spike frequency during propofol administration (Table 3). Seizure frequency was known precisely during the period of intracranial monitoring, because EEG monitoring was continuous, and all individuals in the study had successful localization of the ictal onset zone, indicating that electrodes were positioned correctly for seizure detection. Continuous EEG monitoring occurred for an average period of 7 days (range 1-19 days) before propofol administration. Frequency of complex partial (with or without secondary generalization) seizures ranged from 0 to 5 seizures per day over the prepropofol monitoring period, with a group mean of 0.77 seizures per day before propofol administration. Inspection of the spike frequency data against the rank order of seizure frequency (Table 3) shows no evidence of an association between prepropofol seizure frequency and IIS frequency changes from baseline to propofol administration. The same conclusion was reached when simple partial seizures were included in the calculation of seizure frequency and when only secondarily generalized seizures were used to calculate seizure frequency. A similar analysis was performed with regard to recentness of seizure, i.e., the interval from the most recent seizure before propofol administration until the initiation of propofol infusion. Again, there was no evidence of a tendency for any particular pattern (increase, no change or decrease) of propofol effect on spike frequency to be associated with more recent or less recent seizures before propofol administration; this was true whether all seizures, only complex partial (with or without secondary generalization) seizures, or only secondarily generalized seizures were used to determine the time of the most recent seizure before propofol administration.
Finally, we examined patterns of spike frequency change from baseline to propofol effect with regard to the site of seizure origin (Table 2, Table 4). The seven individuals with unilateral temporal lobe epilepsy had variably increased, unchanged or decreased spike frequency after propofol administration, as did the five individuals with bilateral temporal lobe epilepsy. Of the two individuals with extratemporal partial epilepsy, one had increased and the other unchanged spiking during propofol administration. However, no clinical or EEG manifestation of seizure activity was seen in any of the patients studied. Most patients had spikes at each of the temporal lobe recording sites, although six patients had one or two of these sites that never demonstrated spikes either with or without propofol administration. For all of the patients, every site that had IIS in the absence of propofol also had IIS in the presence of propofol at some level of propofol. Eleven patients had one or more sites at which at least one level of propofol was associated with cessation of IIS during the 15 min of EEG recordings. All sites at which IIS were recorded during propofol infusion also demonstrated IIS in theabsence of propofol, although often (in eight patients) we had to examine up to 1 h of baseline EEG recordings before propofol infusion because certain sites with IIS during propofol infusion did not have spikes during the 15-min period before propofol administration (the epoch used for statistical analysis). However, these sites did have rare IIS over longer intervals.
The aim of our study was to determine whether propofol, in a dose range used to achieve conscious sedation, has an anti- or proconvulsant effect or a significant effect on interictal epileptiform activity in humans. We studied the patients with intractable partial epilepsy for several reasons: (1) This clinical population is prone to developing convulsions. (2) Conscious sedation is used commonly in this patient population during craniotomy for the resection of epileptic focus. (3) Published reports suggestive of anticonvulsant effect of propofol [4-7]have involved patients suffering from status epilepticus. (4) Ethically, this patient group, with intracranial electrodes already in place, could be studied easily.
Intracranial recording offers several advantages over scalp-recording of EEG : (1) It completely removes myogenic (arising from muscle electrical potentials) and kinesiogenic (movement-related) artifacts. These artifacts cover both slow and fast frequencies and thus can affect the entire range of a scalp-recorded EEG. (2) Electrodes are located closer to the generator source of EEG activity, and therefore, signals are of higher amplitude than scalp-recorded EEG signals. By comparison, scalp-recorded EEG potentials are filtered by the meninges, skull, and scalp tissue. This filtering attenuates fast-frequency waves more than those with slower frequencies. Epileptic activity, interictally and during seizures, is predominantly defined by fast frequencies. Therefore, intracranial EEG recording is a more sensitive method for detecting any epileptic activity. (3) Some intracranial electrodes lie within the hippocampus, the most common site of onset of seizure activity, and are capable of recording highly localized seizure activity that may not be recorded with simultaneous scalp EEG.
Our results suggest that, using the computer-controlled infusion of propofol, we were successful in achieving a steady blood level of propofol during the period of EEG analysis. The target plasma propofol levels were slightly different than those achieved, but the plasma propofol levels between 15th and 30th min of each dose administration were similar (Figure 1). Ranges of sedation in our patients varied from 2 to 6 (Table 2, Table 4), suggesting that we studied the dose range of propofol required to produce various degrees of conscious sedation.
None of the patients in our study experienced either a clinical or electrographic seizure during the duration of propofol administration. All sites at which IIS were recorded during propofol administration also demonstrated IIS in the absence of propofol. Thus, we did not find any evidence that propofol can induce "false" spikes (i.e., IIS occurring at sites that do not produce IIS spontaneously). Pathologic focal IIS are markers of individuals at high risk of developing partial epilepsy. In individuals known to have partial-onset seizures, pathologic focal IIS also are useful markers of sites at which EEG seizure activity may originate. Many patients with partial seizures have more than one IIS focus, but often only one of these spike foci is a site at which EEG seizures begin. Several investigators have found it useful to tailor cortical resection to include "spiking" cortex near the areas where electrographic seizures have been recorded, to optimize the probability of postoperative freedom from seizures. [14,15]The usefulness of mapping IIS to tailor the margins of cortical resection in epilepsy surgery presumably will be reduced in the presence of any drug that induces IIS in cortex not otherwise generating IIS. Our observation of lack of "false spikes" during propofol infusion is in direct contrast to the reported effect of some of the other drugs used for anesthesia and/or awake sedation in epileptics. [16,17]Fiol et al. reported that methohexital, when used in patients with epilepsy undergoing temporal lobectomy, induced seemingly "new" spike foci in 43% of the patients, both in the epileptic focus (temporal lobe) and nonepileptogenic brain. Although they did not specifically state the period for which they had analyzed the preoperative EEG, all of their patients had extensive presurgical evaluations with multiple video and telemetry EEGs with sphenoidal electrodes. In our study, every area of the brain studied that showed IIS during propofol infusion also had shown IIS in EEG within an hour before propofol infusion. Another significant difference between these two studies is that Fiol et al. observed the effect of a bolus dose(s) of methohexital in patients anesthetized with oxygen/nitrous oxide, fentanyl, and isoflurane, whereas our patients did not receive any other sedative or anesthetic drugs. Similarly, Cascino et al., in a retrospective study, reported a significant increase in the mesial temporal lobe mean IIS frequency during alfentanil administration, which in one patient progressed to an EEG seizure. In their study, too, the effect of 50 micro gram/kg alfentanil, given as a bolus injection in patients anesthetized with isoflurane or nitrous oxide and fentanyl was retrospectively studied. Increased epileptogenic activity in epileptic patients also has been reported with etomidate and ketamine. Propofol sedation is known to cause activation of beta frequency in EEG, and masking of epileptiform activity with propofol sedation during epilepsy surgery has been reported. This pharmacologic effect of propofol did not affect the IIS frequency response in our study because the criteria we used to identify IIS required the IIS to have at least twice the amplitude of background alpha and beta activities.
Our results differ from previously reported epileptogenic effects of propofol. [2,3]Hodkinson et al. described activation of epileptogenic foci in three patients with temporal lobe epilepsy after a bolus of 2 mg/kg propofol in patients anesthetized with oxygen/nitrous oxide, isoflurane, and fentanyl. In each case in their study, electrocorticography revealed frequent discharges of spikes and spike-wave complexes 20-30 s after propofol injection and continuing for up to 7 min. The dose of propofol used in their study was greater (induction doses), and an interaction between propofol and other drugs, especially fentanyl, cannot be excluded. Makela et al. reported five cases of seizures associated with propofol anesthesia. EEG showing seizure activity was available in one patient only. Four other cases were based on a retrospective review of case records of patients who were given propofol (along with narcotics) and experienced "seizures" characterized by tonic clonic movements. No EEG to demonstrate epileptiform activity was done. Borgeat et al. have shown that, in children, dystonic movements after an induction dose of propofol can occur without EEG evidence of epileptiform activity, suggesting a subcortical or extrapyramidal origin rather than cortical epileptic activity. Hopkins reported a case of recurrent opisthotonos associated with anesthesia induced with propofol and maintained with nitrous oxide and 1.5% enflurane. No EEG was done in that patient. The exact receptor site of action of propofol in the brain is not known. It has been suggested that, as do other hypnotic-sedative drugs, propofol predominantly acts on GABA receptors. It is possible that dystonia (movements) and opisthotonos observed after propofol administration may result from effects on dopaminergic brain sites. A dopaminergic effect also will explain some of the other properties of propofol, e.g., antiemetic effect and euphoria, which are similar to the pharmacologic properties of butyrophenones (droperidol) known dopamine receptor (D2) blockers and also give rise to extrapyramidal signs of muscle rigidity. Regulation of seizure threshold by dopamine receptors is confusing and reported to be of a biphasic nature. Early appreciation of dopamine's role in epilepsy comes from laboratory investigations with prototype dopamine agonist apomorphine, which acts on both dopamine D1and D2receptors. There is now sufficient evidence suggesting that endogenous dopamine is involved to some degree in the initiation and/or registration in most types of experimental epilepsy. Dopamine usually has an anticonvulsant effect, believed to be mediated by D2receptors, although a proconvulsant action mediated by an as yet undisclosed mechanism is also possible. There is some suggestion that these opposing pro- and anticonvulsant forces originate from separate rostrocaudal sources and are in constant competition with each other. The notion that the dopamine receptors mediating epileptic influences are separate from those regulating spontaneous motor function is at present unproven but remains a possibility. This line of thinking and a possible effect of propofol on dopaminergic receptors and GABA receptors will explain the published anecdotal reports of epileptogenic effect [2,3,9]of propofol. This also will explain the usefulness of propofol in cases of status epilepticus after all other anticonvulsants were ineffective [4-7]and its efficacy in reducing the duration of convulsive episode during electroconvulsive shock therapy. [23-25]Simultaneous administration of narcotics and other anesthetics and larger doses of propofol in prior investigations [2,3,9]might have contributed to an epileptogenic effect of propofol.
We have found evidence that, in individuals with partial epilepsy, propofol sedation alone did not cause seizures. Certain propofol doses within the range of conscious sedation are associated with cessation of spontaneously occurring interictal spikes. However, in a few subjects, there was increase in IIS at some of the electrode sites. Variability in (1) the location of the ictal onset zone, (2) the frequency of seizures before propofol administration, and (3) the interval from the most recent seizure to the propofol infusion have no obvious correlations with variability in propofol-induced changes in spike frequency. It remains possible that one or more of these variables might have a partial correlation with propofol effect on spike frequency, because 14 cases are insufficient for differential analysis of variance. Nonetheless, it seems clear that none of these three well defined variables can explain the variability in propofol effect on spike frequency.
We conclude that propofol sedation can be used for patients with epilepsy undergoing temporal lobectomy under local anesthesia. Propofol has an advantage over two other short-acting drugs, methohexital and alfentanil, in that it does not produce any "false" spikes or electrographic seizures in this setting.
*STANPUMP is available from the author: Steven L. Shafer, M.D., Anesthesiology Service, 112A, PAVAMC, 3801 Miranda Avenue, Palo Alto, California 94304.