Loss of consciousness (LOC) and immobility to surgical incision seem to be mediated at different levels of the central nervous system. Pharmacologic studies of hypnotic agents have previously focused on combinations of either volatile or intravenous anesthetics. This study examined the combination of inhaled sevoflurane and intravenous propofol at these two clinically relevant anesthetic end points.
Thirty-six elective surgical patients were initially enrolled. Conditions approximating steady state were obtained for sevoflurane and target-controlled propofol infusions. Patients were sequentially evaluated for LOC (loud voice plus mild prodding) and immobility to surgical incision. The study was designed using the Dixon up-down method.
The observed propofol effect target with 50% response plus sevoflurane (0.46% end-tidal concentration) was 1.2 microg/ml (95% confidence interval, 1.1-1.3 microg/ml). It was not significantly different from that predicted (1.5 microg/ml; 95% confidence interval, 1.2-1.7 microg/ml) by simple additivity. The effective plasma concentration of propofol that suppressed movement to skin incision in 50% of patients was 5.4 microg/ml (95% confidence interval, 4.8-6.0 microg/ml) plus sevoflurane (0.86%) and was not significantly different from that predicted by additivity (5.4 microg/ml; 95% confidence interval, 4.8-5.9 microg/ml). Both analyses had adequate power (90%) to detect a significant change (+/-19 to 25%) from predicted value. Repeated-measures analysis of variance identified a Bispectral Index value of 70 as the break point between those who responded at LOC or did not.
Propofol and sevoflurane interact in a simple additive manner to produce LOC and immobility to surgical incision, suggesting a common mechanism or a single site of action. These clinical observations are consistent with a single site of interaction at the gamma-aminobutyric acid type A receptor.
ANESTHETIC requirement for surgery has historically been defined in terms of the minimum alveolar concentration of a volatile anesthetic (MAC) that will suppress movement to skin incision in 50% of patients.1The movement response to skin incision (nociception) has been shown to be mediated at the spinal cord level and is not necessarily related to hypnosis or amnesia mediated at higher cortical centers in the brain.2–4It has been proposed that different neuronal systems mediate the diverse actions of general anesthetics and that specific neurotransmitter-receptor subtypes can be linked to defined parts of the pharmacologic spectrum of these drugs.5,6Because the loss of consciousness (LOC) and the immobility response to surgical incision (measured by MAC) seem to be mediated at different levels in the central nervous system, this clinical study investigated the interaction between propofol and sevoflurane for both LOC and MAC.
The dose requirements for propofol or sevoflurane used alone have been well validated in human subjects. Sevoflurane MAC is 1.71%.7,8The effective plasma concentration of propofol that will suppress movement to skin incision in 50% of patients (Cp50INC) is 15.2 μg/ml.9The end-tidal concentration of sevoflurane that will produce LOC in 50% of patients (LOC50sevo) is 0.89%, and the equivalent effect site concentration of propofol for loss of consciousness (Cp50LOC) is 2.81 μg/ml.10
The lack of a defined molecular target and selective antagonists has limited previous pharmacologic investigations of hypnotic/anesthetic mechanisms to interactions between two volatile anesthetics11–15or between intravenous agents.16–21Eger et al. 11combined halothane and isoflurane and determined that the effects of the 2 agents on MAC were additive. The interaction of volatile and intravenous anesthetics has been studied in rats.22Thiopental was shown to reduce halothane requirements in a log-linear manner (not simple additivity). Drug interaction studies on sedation have focused on midazolam–propofol interactions.17,20,23
Propofol, etomidate, and midazolam have been proposed to produce sedation and immobilization through interaction with the γ-aminobutyric acid type A receptor.5,24,25Although volatile anesthetics are thought to produce immobilization through a different mechanism,26,27some in vivo evidence suggests that the γ-aminobutyric acid type A receptor may also play a role in humans.28Because the sites of action of anesthetics in producing LOC and immobility seem to be different and various putative sites of anesthetic agents have been identified, defining the volatile/intravenous anesthetic interaction for these agents at two clinically relevant end points may provide another tool to help further define the mechanism of action of anesthesia. A companion article describes the interaction of the same two agents at the γ-aminobutyric acid type A receptor in vitro .29
Materials and Methods
After Institutional Review Board (Emory University, Atlanta, Georgia) approval and written informed consent were obtained, patients with American Society of Anesthesiologists physical status of I or II, aged 18–55 yr, presenting for elective surgery, and requiring a skin incision of more than 1 inch were enrolled. Ineligibility criteria included serious hearing impairment (inability to comprehend speech at normal tone of voice), chronic substance abuse, weight greater than 150% of ideal body weight, and neurologic or seizure disorders. Patients who were not fluent in English or had used opioids within 8 h before surgery were also excluded.
Patients were recruited during preoperative preparation before surgery. No preoperative sedation or analgesics were administered. In the operating room, standard anesthesia monitors were placed, including an electrocardiograph, pulse oximeter, noninvasive blood pressure, Bispectral Index of the electroencephalogram (BIS), and end-tidal gas (ETsevoand ETco2) monitors. The Bispectral Index A2000 monitor® (version 3.3; Aspect Medical Systems, Inc., Natick, MA) uses three adhesive electrodes to the forehead (single channel: Fp1–Fpz). BIS, raw electroencephalogram, electromyogram (70- to 110-Hz band referenced to 100 μV in decibels), electrode impedance (ohms), and signal quality index (a global estimate of data quality from 0 to 100) were automatically recorded to a computer at 5-s intervals. BIS and other electroencephalographic variables are an average of recordings taken for 1 min before administration of sedation scale or MAC determination. Hemodynamic variables and end-tidal gas measurements were recorded manually. No other intraoperative medications were administered during the study.
Loss of Consciousness
Sevoflurane (tightly sealed mask via semiclosed circuit) was administered to achieve an end-tidal concentration of 0.45% (half LOC50sevo).10Propofol target-controlled infusion used STANPUMP,§programmed with the pharmacokinetic parameters of Schnider,30and a Harvard syringe pump (South Natick, MA) to a preset “effect site” concentration. The initial propofol target (effect site) concentration used was 1.4 μg/ml.10After 10 min of equilibration at the target concentration (approximately 3 distribution time constants), the modified Observer’s Assessment of Alertness/Sedation (OAA/S)31was administered. LOC was defined as an OAA/S score less than 2 (loss of responsiveness to voice command [open eyes] with light tapping on the shoulder or mild shaking). This arousal response or opening the eyes was considered positive, whereas nonspecific (nonpurposeful) motor responses were considered negative.
After administration of the OAA/S, the concentration of sevoflurane was increased to 0.85% (half MACsevo),8and the propofol target-controlled infusion effect site target was increased to 7.6 μg/ml (half Cp50INC).9The patient’s trachea was intubated after administration of succinylcholine (1.5 mg/kg). The lungs were ventilated to maintain an end-tidal carbon dioxide concentration between 32 and 35 mmHg. After at least a 10-min equilibration at the target end-tidal anesthetic concentration and return of neuromuscular function (train-of-four), skin incision was performed. Movement was defined as gross movement of limbs, head, or body. Coughing, chewing, and swallowing were not considered movement.
The target concentrations of sevoflurane for OAA/S and for incision were not altered during the study. However, the propofol target-controlled infusion targets used for each subsequent patient (Cp50LOCand Cp50INC) were determined by the response of the previous patient. Target propofol effect site concentrations were increased by 10% if the previous patient responded and decreased by 10% if the previous patient did not respond. Responses to OAA/S and skin incision were assumed to be independent. The concentrations at which 50% of patients lost consciousness (Cp50LOC) and 50% moved to skin incision (Cp50INC) were determined by the up–down method of Dixon32by averaging independent crossover pairs for each response. A crossover pair represents a unique set of sequential patients in which the first patient responded and the next patient did not or vice versa . Investigators were not blinded to the sevoflurane or propofol dose.
Validation of Propofol Concentrations to Suppress Consciousness and Movement
Our original data for propofol alone Cp50LOC10and Cp50INC9were obtained using the Duke Computer–Assisted, Continuous–Infusion (CACI) program. This software is no longer available. After Institutional Review Board approval was obtained, an additional 22 patients were recruited to provide five independent, crossover pairs of propofol alone using the STANPUMP algorithm to validate the intercept of propofol alone for isobolographic analysis for Cp50LOCand Cp50INC.23,33
Frequency data (sex) were analyzed with chi-square test. The Kruskal-Wallis test was used to detect significant differences in ranked categorical variables such as American Society of Anesthesiologists physical status. Isobolographic analysis33,34and two-way Student t test were used to compare sevoflurane–propofol patient data to predicted values (simple additivity = half of propofol alone Cp50LOCor Cp50INC). Multifactorial, repeated-measures analysis of variance with post hoc Bonferroni test was used to evaluate the significant relations between variable and cofactors (e.g. , sex, age, weight) using a standard statistical software package (SPSS version 11; SPSS Inc., Chicago, IL). P < 0.05 was considered statistically significant.
Thirteen LOC/Cp50LOCand 10 MAC/Cp50INCcrossover pairs (12 male and 21 female) were obtained from the 36 patients enrolled in the initial study. Three patients had data collected but participated in neither crossover pair. The average age (mean ± SD) of patients in crossover pairs was 40 ± 9 yr. Average height and weight were 168 ± 13 cm and 74 ± 15 kg, respectively. Average starting noninvasive blood pressure and heart rate were 131 ± 17/72 ± 12 mmHg (systolic/diastolic) and 70 ± 14 beats/min. Preinduction pulse oximetry (100 ± 0%), end-tidal carbon dioxide (34 ± 5 mmHg), and Bispectral Index (96 ± 3) were recorded.
Figure 1shows the LOC50of sevoflurane–propofol for 13 crossover pairs. Our experimental method examined only a single point on the predicted line of additivity at a single ETsevo(0.46%; 95% confidence interval [CI], 0.45–0.47%). Visually, this result is within the CI of the line of simple additivity (null hypothesis) connecting propofol alone (Cp50LOC) to sevoflurane alone (ETsevo) values on the isobologram.33The observed Cp50LOCfor propofol at this sevoflurane concentration was 1.2 μg/ml (95% CI, 1.1–1.3 μg/ml) and was not significantly different from the predicted value of 1.5 μg/ml (95% CI, 1.2–1.7 μg/ml) by either isobolographic analysis or two-tailed t test. This analysis had a 90% power to detect a 19% change in propofol dose from predicted.
The isobologram of the corresponding ten MAC/Cp50INCcrossover pairs is shown in figure 2. Based on the study of an additional 22 patients (5 crossover pairs), the propofol-alone intercept (Cp50INC) for the isobologram was 10.7 μg/ml (95% CI, 10.2–11.3 μg/ml). The sevoflurane-alone MAC was 1.71% (95% CI, 1.61–1.82%).8The observed Cp50INCfor propofol was 5.4 μg/ml (95% CI, 4.8–6.0 μg/ml) at an ETsevoof 0.86% (95% CI, 0.85–0.87%). The observed combined propofol–sevoflurane Cp50was not significantly different than that predicted by simple additivity (5.4 μg/ml; 95% CI, 4.8–5.9 μg/ml) by either isobologram or two-tailed t test. This analysis had a 90% power to detect a 25% change in propofol dose from predicted.
Responder or nonresponder status was identified in the repeated-measures analysis of variance as a significant covariate for BIS observations. Although there was no significant change in BIS noted before or after OAA/S administration (data not shown ), mean starting BIS values were significantly higher in patients before test administration in responders (76 ± 12) (OAA/S score ≥ 2, awake) compared with nonresponders (65 ± 9) (table 1). No other hemodynamic or electrophysiologic covariable was statistically significant.
There is reason to believe that LOC and movement responses during anesthesia are mediated separately (via brain and spinal cord, respectively). There are clear data from animal experiments to support these contentions. Rampil et al. 2,3demonstrated that either removing the forebrain or spinal cord transection does not affect MAC. Antognini and Schwartz4demonstrated that when only the brain is perfused with volatile anesthetic, the MAC is increased nearly threefold. Therefore, it seems that the brain is not significantly involved in mediating the movement response to noxious stimuli after inhalational anesthesia. Recently, Stabernack et al. 35suggested that immobility produced by an intravenous hypnotic agent (thiopental) in rats seems to be mediated primarily by supraspinal actions, representing a different mechanism of action than that of volatile agents.
This experiment was designed to identify the midpoint on the response surface at a single drug combination of sevoflurane and propofol for LOC and MAC/Cp50INCunder conditions approaching steady state. Defining drug–drug interactions depends on precise measurement of the individual drug response and variance (intercept ± 95% CI of isobologram). CP50LOCwith propofol alone (2.9 μg/ml; 95% CI, 2.7–3.1 μg/ml) using STANPUMP target-controlled infusion was identical to the results of Kodaka et al. 10using the CACI9system. The definition of LOC in this study (OAA/S score < 2, no response to mild prodding) was different than many previous reports (OAA/S score < 3, no response to loud voice)9,36,37that reported propofol Cp50LOCbetween 2.1 and 3.3 μg/ml.10Our results show simple additivity between sevoflurane and propofol for LOC with no statistical difference between the predicted and observed target propofol effect site concentration at the target sevoflurane concentration. This study had an adequate power (90%) to exclude a type II error and to identify a small change (19%) in propofol effect site concentration. Kodaka et al. 10identified a small statistically significant difference between BIS values at LOC between propofol alone (74 ± 13) and sevoflurane alone (62 ± 11). BIS at LOC50for the sevoflurane–propofol crossover pairs in this study (71 ± 16) was not significantly different from that for propofol-alone (69 ± 13) controls and was similar to values reported by Kodaka et al. 10Additive drug interactions suggest a common mechanism or a single site of action, whereas synergistic or antagonistic responses suggest separate sites or competing mechanisms of action.33,38
Cp50INCfor surgical immobility with propofol alone (10.7 μg/ml; 95% CI, 10.2–11.3 μg/ml) obtained in the additional patients enrolled in this study was less than that reported by Smith et al. 9(15.2; 95% CI, 7.6–22.8) and had much less variability. No significant statistical difference between the predicted and observed propofol effect site target concentration with sevoflurane was found. Our results are consistent with simple additivity between sevoflurane and propofol for suppression of movement to incision. Adequate power (90%) was achieved to exclude a type II error and to identify a 25% change in propofol effect site concentration. The average BIS values before MAC were not significantly different between the propofol alone group and for the sevoflurane–propofol combination demonstrating equivalent hypnotic states. No data are available in the literature for BIS at sevoflurane MAC. Poor signal quality, high electrode impedance, or high electromyographic activity did not influence BIS values (data not shown).
It has technically been difficult to study the actions of intravenous anesthetics on end points such as LOC and immobility to painful stimuli because of the difficulty in approximating steady state (equilibrium) drug concentrations at the effect site. In this study, first, we targeted effect site concentrations rather than plasma concentrations. Plasma drug levels are not in equilibrium with brain (LOC) and spinal cord (MAC/Cp50INC) concentrations, making it difficult to control intraindividual and interindividual variables. Second, we allowed sufficient time to elapse (a minimum of 10 min or three distribution half-lives for propofol after predicted effect site drug concentration was reached) to approximate steady state conditions. Third, BIS monitoring directly measures drug effect in the cerebral cortex and provides a method to define equivalent hypnotic states. The BIS values for LOC and skin incision for the drug combination are similar to the available drug-alone values, and no evidence of hypnotic synergism between agents was found.
Issues with the design and interpretation of MAC studies have been raised by several authors.39,40The quantal study design used here identifies a population mean effective dose (ED50) based on a single observation per patient (surgical incision). This design has been criticized for ignoring interpatient differences, producing expected errors of 10%, systematically underestimating MAC, and has been characterized as a “giant rat” study.40,41Fully defining the interaction of two drugs requires construction of a response surface based on analysis of multiple drug combinations using a continuous measure of response.23,42It has been estimated that at least 10 observations per patient are required to obtain an unbiased estimate of the Hill coefficient with the population approach.42Lu et al. 41demonstrated that examining quantal responses alone will never establish the concentration-versus -response curve. In this study, we chose to maximize the power of the study by using at least 10 crossover observations and to examine only the midpoint of the response surface. It should be noted that the design of this study and other studies using the up–down methodology does not permit blinding and is a limitation of these types of studies.
Although commonly used by clinicians to titrate hypnosis, hemodynamic changes to surgical stimulation are neither sensitive nor specific for awareness or recall during general anesthesia and are poor predictors of both voluntary and involuntary reflexes.43–45Before stimulation, no hemodynamic variable was significantly different between patients who responded or did not respond to OAA/S or surgical incision. However, average BIS before OAA/S in nonresponders (LOC or OAA/S score < 2) was significantly lower than that of responders (table 1), suggesting that a BIS value of 70 or higher was predictive of consciousness (response to loud voice with mild prodding). No difference was found in prestimulus BIS values between patients who moved or did not move to surgical incision.
This study has shown that propofol and sevoflurane produce sedation (LOC) and immobility to surgical stimulation (MAC/Cp50INC) that are predictable from their individual potencies (simple additivity) for each distinct pharmacologic end point. No suggestion of synergistic action between the two anesthetics was found for either response, suggesting that both drugs operate through a common mechanism. This study cannot specifically determine whether the two anesthetics interact at a single site, different drug thresholds at a single site, multiple sites on the same receptor, or on separate receptors producing these two responses. These results are also consistent with the companion article demonstrating additivity at the γ-aminobutyric acid type A receptor for these two agents in vitro at clinically relevant concentrations.29