PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
Anesthesiology. Author manuscript; available in PMC 2005 November 1.
Published in final edited form as:
PMCID: PMC1249471
NIHMSID: NIHMS2233

Induction Speed Is Not a Determinant of Propofol Pharmacodynamics

Anthony G. Doufas, M.D., Ph.D.,* Maryam Bakhshandeh, M.D., Andrew R. Bjorksten, Ph.D., Steven L. Shafer, M.D.,£ and Daniel I. Sessler, M.D.#

Abstract

Summary

We used individual pharmacodynamic modeling to demonstrate that different sedation endpoints occur at the same effect site propofol concentration, independent of the infusion rate of propofol.

Background

Evidence suggests that the rate at which they are infused may influence plasma-effect site equilibration of intravenous anesthetics. We used 5 different rates of propofol administration to test the hypothesis that different sedation endpoints occur at the same effect site propofol concentration, independent of the infusion rate. We concurrently evaluated the automated responsiveness monitor (ARM) against other sedation measures and the propofol effect site concentration.

Methods

With Human Studies Committee approval, 18 healthy volunteers received 5 consecutive target-controlled propofol infusions. During each infusion the effect site concentration was increased by a rate of 0.1, 0.3, 0.5, 0.7, or 0.9 μg·ml−1·min−1. Bispectral index and ARM were recorded at frequent intervals. The times of syringe drop and loss and recovery of responsiveness were noted. Pharmacokinetic and pharmacodynamic modeling was performed using NONMEM.

Results

Once the correct rate of plasma-effect site equilibration (ke0) was determined for each individual (ke0 = 0.17 min−1, time-to-peak effect = 2.7 min), the effect site concentrations associated with each clinical measure were not affected by the rate of rise of effect site propofol concentration. ARM correlated with all clinical measures of drug effect. Subjects invariably stopped responding to ARM at lower effect site propofol concentrations than those associated with loss of responsiveness.

Conclusions

Population-based pharmacokinetics, combined with real-time electroencephalographic measures of drug effect, may provide a means to individualize pharmacodynamic modeling during target-controlled drug delivery. ARM appears useful as an automated measure of sedation and may provide the basis for automated monitoring and titration of sedation for a propofol delivery system.

Introduction

Although it is generally assumed that the rate of equilibration between the plasma and the site of drug effect is independent of the rate of drug administration, several studies suggest this may not be the case for intravenous anesthetics.14 There may be complex interactions among the rate, dose, and time of anesthetic induction57, as well as physiological factors1,3,4,810, which might influence the rate of plasma-effect site equilibration. If infusion rate alters the time course of plasma-effect site equilibration, this would be a new source of variability that must be understood when designing infusion regimes. We tested the hypothesis that different sedation endpoints occur at the same effect site propofol concentration, independent of the propofol infusion rate. This study was designed using a prototype sedation delivery system for propofol administration.

This study also examined the performance of a an automated responsiveness monitor (ARM) (previously called the automated responsiveness test or “ART”), a novel feedback system for titration of sedative drugs.11,12 We have previously shown that patients stop responding to ARM during moderate sedation and that patients who are otherwise unresponsive invariably do not respond to ARM.11,12 Given the large variability in propofol effect site concentrations at loss of responsiveness, ARM may be a useful device to assess patient sensitivity to propofol during titration, particularly if ARM is integrated into a propofol delivery system. ARM has not been prospectively tested under non-steady state conditions, and thus, was included in this examination of the relationship of propofol infusion rate to measures of propofol drug effect.

Materials and Methods

With approval of the University of California at San Francisco Committee on Human Research (San Francisco, California) and written informed consent, we evaluated 18 healthy volunteers of both sexes. Age was restricted to 18–50 years. Volunteers fasted at least eight hours before the study.

Protocol

The sedation delivery system includes three major elements: 1) standard anesthetic monitoring including arterial pressure, electrocardiogram, end-tidal pCO2, and oxygen saturation; 2) a computer-controlled propofol infusion system; and 3) the Automated Responsiveness Monitor (ARM). The Automated Responsiveness Monitor consists of an earphone positioned in one ear, held in place with a strap, and a handpiece about the size and shape of a small cellular phone strapped into the palm of the dominant hand. Earpiece function is monitored with an online indicator. A thumb button is mounted on the handpiece. A computerized voice asks the participant to press the button at regular intervals. A vibrator built into the hand piece vibrates at the same time. The voice and vibration repeat until the thumb button is pressed (maximum of five requests over a 10-second period). The voice gets louder and the vibrations get more intense with each repetition.

The sedation delivery system monitors and ARM apparatus were applied to the participating volunteers. Electrodes to capture the bispectral index of the electroencephalogram (BIS™ 3.3 algorithm) were applied to the forehead according to the manufacturer’s instructions (Aspect Medical Systems, Inc., Newton, MA, USA). The BIS recording began with a 2 min period of quiet relaxation with the volunteer’s eyes closed.

A 20-g venous catheter was inserted into the non-dominant arm above the wrist for the propofol infusion. A 20-g catheter was inserted at the antecubital fossa on the dominant arm and lactated Ringer’s solution (200 ml) was infused as a bolus. Subsequently, fluid was infused at a rate of 100 ml/h. A 20-g catheter for blood sampling was inserted into the radial artery of the non-dominant hand. Surface warming was used to maintain tympanic membrane temperature between 37.0 and 37.5°C. Volunteers breathed 30% oxygen via a standard anesthesia mask during each trial.

The volunteers were familiarized with the ARM apparatus for 10–15 minutes before the first sedation trial. The volume of the earpiece was adjusted to a level that the volunteer was able to hear easily. We confirmed that the volunteers responded promptly to the ARM during this pre-study period.

We used a target-controlled drug delivery system according to the method of Shafer and Gregg13 to target propofol effect site concentrations using the propofol pharmacokinetics reported by Schnider et al.14 with a half-life of plasma-effect site equilibration of 91 s.15 The performance of the system was previously evaluated under pseudo-steady state conditions.16 The drug delivery system consisted of a Harvard 2 (Harvard Clinical Technology, South Natick, MA) electronic syringe pump and a customized software driver.

The software did not target steady-state effect site propofol concentrations, but instead produced a constant ramp in the effect site concentration. The effect site ramp rates were 0.1, 0.3, 0.5, 0.7, and 0.9 μg·ml−1·min−1. Each ramp began either at 0 (first ramp) or a predicted effect site concentration lower than 0.5 μg/ml. So as to eliminate any potential time- or pre-trial effect site concentration-related effects on the study outcomes, the order of the 5 sedation ramps was randomized within and across the volunteers.

Each ramp was continued until loss of responsiveness (LOR), as defined by an Observer’s Assessment Alertness/Sedation (OAA/S) score equal to 1 (i.e., no response to mild shaking).17 The OAA/S score was initially measured after the first negative response to the ARM prompt and was repeated every 15 sec thereafter, immediately after each ARM assessment (negative or positive). After LOR, the infusion was stopped, and OAA/S was determined every 15 seconds until recovery of responsiveness (OAA/S = 2).

Monitoring continued during the recovery period, for at least 15 minutes after the volunteers regained responsiveness. Once they had been awake for at least 15 minutes and their predicted propofol effect site concentrations had decreased below 0.5 μg/ml, the next ramp in the randomized sequence was started. Figure 1 shows a representative trial with the effect site (Fig. 1A) and plasma (Fig. 1B) propofol concentrations, as well as bispectral index over time (Fig. 1A) at propofol ramp rates of 0.9, 0.5, 0.7, 0.1, and 0.3 μg·ml−1·min−1.

Figure 1
A representative volunteer trial. Graph A shows the propofol effect site concentration and bispectral index during 5 consecutive ramps (0.9, 0.5, 0.7, 0.1, and 0.3 μg·ml−1·min−1). Graph B shows the predicted (solid ...

Measurements

Heart rate, blood pressure, end-tidal pCO2, respiratory rate, arterial oxygen saturation, and BIS values were recorded by an automated data-acquisition system for off-line analysis. End-tidal CO2 was collected through a tight-fitting, hand-held anesthesia mask. The data were captured at 15-second intervals except for the non-invasive blood pressure measurement, which was captured at 2-minute intervals.

The first clinical end-point was the “syringe-drop.” With each ramp, the volunteer held a water-filled 60-ml syringe over the floor with the palm facing down. The time at which the subject released the syringe was recorded.

The second and third clinical end-points were loss and recovery of responsiveness, respectively, which were based upon the responsiveness component of the OAA/S.17 Loss and recovery of responsiveness were defined as the first OAA/S score of 1 (loss − no response to mild shaking) followed by the first OAA/S score of 2 (recovery − response to mild shaking).

Our fourth clinical endpoint was loss of the ability to respond to the ARM prompt. ARM was tested at 15-second intervals during each ramp, as well as during recovery from each propofol infusion.

Arterial blood samples for propofol determination were obtained at each of the four clinical endpoints. Additionally, blood samples were obtained at each 0.5 μg/ml predicted effect site propofol concentration increment during the 0.1, 0.3, and 0.5 μg·ml−1·min−1 ramp trials and at each 1.0 μg/ml during the 0.7 and 0.9 μg·ml−1·min−1 ramp trials. The samples were analyzed using high-performance liquid chromatography assay modified from the method of Plummer.18 This method has a detection limit of 5 μg/L and a coefficient variation of 4.1% at a propofol plasma level of 2 μg/ml.

Data Analysis

Individual demographic and morphometric data were presented in tabular format. Heart rate, non-invasive blood pressure, respiratory rate, end-tidal CO2 and arterial oxygen saturation were averaged within and then across the volunteers and presented for each propofol infusion ramp rate separately.

We used the traditional two-step approach of Sheiner et al.19 to model individual subject effect site concentrations with the measured arterial propofol concentrations and the BIS as a continuous high-resolution measure of drug effect. With this model, pharmacokinetics of the drug are estimated, followed by an estimation of ke0, the rate constant for plasma-effect site equilibration, and the parameters of the concentration vs. response model.

Pharmacokinetics

The parameters of a traditional three-compartment mammillary pharmacokinetic model were fit to the data using NONMEM20 with first-order conditional estimation. The drug infusion regimen recorded by the sedation delivery system every 15 seconds was used as the input to the model. NONMEM estimated post-hoc Bayesian volumes and clearances in each individual, as well as typical values (e.g., geometric means) of the volumes and clearances. Variability in volume and clearance was modeled assuming log-normal inter-individual variability. Residual intra-individual error was assumed to be proportional to the prediction (i.e., a constant coefficient of variation).

Goodness of fit was assessed by examination of plots of predicted vs. measured concentration and calculation of the median performance error and the median absolute performance error.21 First, for each blood sample the performance error (PE) was calculated as:

PE=Cm-CpCp*100,

where Cm and Cp are the measured and predicted plasma propofol concentrations, respectively. Subsequently, the median performance errors (MDPE) and the median absolute performance errors (MDAPE) were calculated for each subject separately. Performance error, MDPE and MDAPE were calculated for each subject twice: first using the original prediction for plasma propofol levels based on the pharmacokinetics reported by Schnider et al.14, and second using the prediction derived from the individual post hoc Bayesian estimates. The performance indices based upon the original and the post-hoc Bayesian estimates were compared using paired t-tests with a level of significance of P < 0.05.

The influence of time and ramp rate on propofol pharmacokinetics was assessed by plotting the measured/predicted propofol concentrations against time and ramp rate.

Pharmacodynamics

The post hoc Bayesian estimates of each volume and clearance term were used to calculate plasma and effect site concentrations. As described by Sheiner et al.,19 the effect site was assumed to be linked to the plasma by a compartment of trivial volume with a first-order equilibration constant of ke0. The shape of the effect site concentration vs. BIS response relationship was assumed to be sigmoidal and described by the logistic relationship:

BIS=E0-EmaxCeγBISCe50,   BISγBIS+CeγBIS,

where E0 is the baseline BIS, Emax is the maximum effect of propofol on BIS, Ce is the propofol concentration at the site of drug effect, and Ce50, BIS is the effect site propofol concentration associated with 50% of the maximum effect. γBIS is the steepness of the concentration vs. response relationship (also termed the “Hill coefficient”). The parameters ke0, Ce50, BIS, E0, Emax, and γBIS were estimated using NONMEM. Inter-individual variability was permitted on ke0 and Ce50, and was assumed to be log-normally distributed. Residual intra-individual error was assumed to be additive. Tpeak was calculated by simulating an intravenous bolus injection and determining the time of peak concentration in the effect site.22

The relationship between ramp rate and Ce50, BIS was modeled assuming a linear relationship between Ce50, BIS and ramp rate. Model selection was based on the improvement in −2 log likelihood, with a reduction in −2 log likelihood of 3.84 considered significant (χ2 < 0.05).

We assessed model performance by plotting the post hoc Bayesian BIS predictions against the measured BIS and looking for systematic misspecification. Inter-subject variability was displayed by plotting curves showing the individual effect site propofol concentration vs. BIS relationships.

The effect site propofol concentration at “syringe drop” was calculated using the post hoc Bayesian estimates of the volumes, clearances, and ke0. The influence of ramp rate on the sedation delivery system and Bayesian prediction of the propofol effect site concentration was evaluated graphically by calculating the mean concentration and 95% confidence bounds at syringe drop at each ramp rate. The confidence bounds were constructed as ± 1.96 SEM.

The effect site propofol concentrations at loss of responsiveness (first failure to respond to mild shaking) and recovery of responsiveness (first response to mild shaking after loss of responsiveness) were calculated using the post hoc Bayesian estimates of the volumes, clearances, and ke0. The influence of ramp rate on loss and recovery of responsiveness using the sedation delivery system and Bayesian prediction was evaluated graphically by calculating the mean concentration and confidence bounds at loss and recovery of responsiveness at each ramp rate. The confidence bounds were constructed as ± 1.96 SEM.

The relationship between the effect site propofol concentrations at loss and recovery of responsiveness was evaluated by plotting the concentration at loss of responsiveness vs. the concentration at recovery of responsiveness vs. the line of identity. Similarly, the relationship between BIS at loss and recovery of responsiveness was evaluated by plotting the BIS at loss of responsiveness vs. BIS at recovery of responsiveness vs. the line of identity.

Automated Responsiveness Monitor (ARM)

Logistic regression was performed with NONMEM to estimate the probability of ARM response as a function of effect site propofol concentration. Each response to ARM was given a score of 1, and each non-response to ARM was given a score of 0. The probability of response to ARM was then calculated as:

P=1-CeγARMCe50,   ARMγARM+CeγARM

If, as defined above, R is the observed response to ARM and P is the probability of response to ARM, then the probability of each observation was defined as

Probability of observation=R×P+(1-R)×(1-P)

The probability of response is the probability that the patient will respond to the stimulus, ranging from 1 when no drug is present to 0 as the propofol concentration approaches infinity. The probability of an observation refers to an individual observation during the study. Since P in the model is the probability of response, then if the patient responded, the probability of that observation is P. However, if the patient did not respond, then the probability of that in the model is 1−P. For example, if no drug is present, then the probability of response is 1, and the probability of non-response is 0. The probability of the observation depends on what the observation was. NONMEM estimated the model parameters to identify the parameter values that maximized the probability of all of the observations.

We also used this model, mutatis mutandis, to estimate the Ce50 for syringe drop (Ce50, syringe), loss of responsiveness (Ce50, LOR), and recovery of responsiveness (Ce50, ROR). We compared the Ce50, ARM against Ce50, BIS, Ce50, syringe, Ce50, LOR, and Ce50, ROR. We also assessed graphically the relationship between the lowest effect site propofol concentration at which each subject became unresponsive to ARM at any of the ramp rates and the average concentration at which “syringe drop,” loss and recovery of response to mild shaking occurred.

Results

Hemodynamic and respiratory physiology was essentially unchanged during the various infusion rates of propofol. Demographic data are shown in Table 1.

Table 1
Demographics of the subjects

Pharmacokinetics

Post hoc Bayesian volumes and clearances as estimated by NONMEM for each individual separately are presented in Table 2. The post hoc Bayesian estimates of volumes and clearances improved the prediction of plasma propofol concentration compared with the original population pharmacokinetics (Table 3). Both pharmacokinetic predictions were unbiased (MDPE < 2%). The accuracy of the Schnider’s pharmacokinetic model was very good (MDAPE of 21%), but the accuracy was significantly greater for the post hoc Bayesian pharmacokinetic parameters (MDAPE of 13%; P < 0.05). Individual post hoc Bayesian estimates of the pharmacokinetic parameters improved the relation between the predicted and measured plasma propofol concentrations (Fig. 2B) compared to the original estimates (Fig. 2A).

Figure 2
Goodness of fit for the original predictions (graph A) and the predictions based on the post-hoc Bayesian estimates (graph B) vs. the measured arterial propofol concentrations.
Table 2
Pharmacokinetic parameters in individual subjects.
Table 3
Performance Characteristics of the target controlled infusion.

Neither time (Fig. 3A and 3C) nor ramp rate (Fig. 3B and 3D) affected the residual errors, which suggested that neither of these covariates influenced the pharmacokinetics of propofol.

Figure 3
The influence of time (graphs A and C) and ramp rate (graphs B and D) on the measured/predicted propofol concentrations. Graphs A and B present the predictions from the original pharmacokinetic model,14 and graphs C and D are the predictions from the ...

Pharmacodynamics

Individual pharmacodynamic results for the 18 volunteers are shown in Table 4. Intersubject variability could only be estimated on ke0 and Ce50, BIS. NONMEM estimated that the BIS response started at 96 (E0), and reached a nadir at 20 (E0 − Emax). The ke0 estimated by NONMEM was 0.17 min−1 ± 30% coefficient of variation (C.V.), which yielded a typical time to peak effect of 2.7 min. Plotting the post hoc Bayesian BIS predictions against the measured BIS did not reveal any major model misspecification (Fig. 4A); however, individual propofol concentration vs. BIS relationships demonstrated considerable inter-subject variability (Fig. 4B). Ramp rate was not a significant covariate of Ce50, BIS.

Figure 4
The goodness of fit for the pharmacodynamic model based on BIS (graph A); Fig. 4B shows the inter-individual variability in the pharmacodynamic model.
Table 4
Pharmacodynamic results in individual subjects.

The predicted effect site concentrations of propofol and BIS values at which syringe drop, loss of responsiveness (LOR), and recovery of responsive (ROC) occurred are given in figures 57, respectively. LOR occurred at 29 ± 6 (mean ± SD), 12 ± 3, 8 ± 2, 7 ± 2, and 5 ± 1 minutes after the start of the propofol infusion at rates of 0.1, 0.3, 0.5, 0.7, and 0.9 μg·ml−1·min−1, respectively. The original sedation delivery system prediction of the propofol effect site concentration at “syringe drop” (Fig. 5A) and “loss of responsiveness” (Fig. 6A) endpoints increased linearly as a function of ramp rate, whereas BIS values at each endpoint were similar across the different ramp rates (Figs. 5C and and6C).6C). Post hoc Bayesian estimation of the effect site concentration using individually predicted ke0 values demonstrated that the end points were reached at the same effect site propofol concentration at each ramp rate (Figs. 5B and and6B).6B). Recovery of responsiveness occurred at the same effect site propofol concentration independent of the ramp rate or the prediction model used (Figs. 7A, B, and C). This was an expected result because the plasma and effect site concentrations were changing more slowly on recovery of responsiveness and thus were in close equilibration and insensitive to errors in ke0.

Figure 5
The relationship of the effect site concentration and BIS at the time of syringe drop to the effect site ramp rate. With the wrong rate constant (ke0) (graph A), it appeared that at higher ramp rates more propofol was required at the effect site to induce ...
Figure 6
Similar to Figure 5, the effect site propofol concentration associated with loss of responsiveness (graph A) appeared to increase with increasing ramp rates. However, when the value of the rate constant (ke0) was individualized, the effect site propofol ...
Figure 7
The effect site propofol concentration associated with recovery of responsiveness was independent of the rate constant (ke0), whether determined with the original (graph A) or individualized values of ke0 (graph B), likely reflecting the slow rate of ...

Prediction of the propofol effect site concentration, using the individual Bayesian estimates of the pharmacokinetic parameters and ke0, showed very tight correlation between loss and recovery of responsiveness (Fig. 8A). This was not the case for the BIS values at loss and recovery of responsiveness (Fig. 8B), reflecting the intrinsic noise of the BIS measurement.

Figure 8
The effect site concentrations at loss of responsiveness were nearly identical to those at recovery of responsiveness (graph A). There was less agreement between the bispectral index values on loss of responsiveness and recovery of responsiveness (graph ...

Automated Responsiveness Monitor (ARM)

Independent of the infusion rate, at low propofol concentrations all the volunteers were able to respond to the ARM prompt, whereas at high concentrations none of them was able to do so (Fig. 9A). In between these is a concentration range where response to ARM was variable. This behavior is captured by the logistic regression model (Fig. 9B) that shows the transition from 100% probability of response to the ARM to 100% probability of no response to the ARM.

Figure 9
Graph A shows every recording of the automated responsiveness monitor (ARM) in every subject. Each horizontal line represents a single subject and the ticks show individual ARM responses. The presence (response) and lack (no response) of ARM response, ...

The effect site concentration at first loss of the ARM (1.49 ± 0.46 μg/ml, mean ± SD) closely correlated with loss or recovery of responsiveness (R2=0.87). This concentration (Fig. 10, thick line) is close to the concentration at which subjects dropped the syringe (Fig. 10, open circles) and was consistently below the concentration at which they lost responsiveness (Fig. 10, thin line) or regained responsiveness (Fig. 10, filled circles). The Ce50, ARM was 1.76 ± 0.60 μg/ml, which is 32% less than the Ce50 LOR, 2.57 ± 0.91 μg/ml. The individual Ce50 values for loss of response to ARM closely correlated with Ce50, BIS (R2=0.73), Ce50, syringe, (R2=0.63), Ce50, LOR (R2=0.91), and Ce50, ROR (R2=0.88).

Figure 10
The propofol effect site concentrations at the first loss of the automated responsiveness monitor (ARM, thick line), “syringe drop” (open circles), loss of responsiveness (thin line), and recovery of responsiveness (filled circles) in ...

Discussion

The main purpose of the study was to determine whether the responses to various clinical measures of sedation are influenced by the rate of rise of propofol effect site concentration. In addition, we specifically assessed Automated Responsiveness Measure (ARM) against several clinical measures of sedation.

In order to test our first hypothesis, we modeled the effect site propofol concentration, using the two-step approach proposed by Sheiner et al.19 Initially, it appeared that the rate of rise of propofol’s effect site concentration influenced both ARM and the other measures of sedation (Figs. 5 and and6,6, top graphs). This was based on the integrated pharmacokinetic/pharmacodynamic model of propofol reported by Schnider et al.14,15 However, we were able to use the bispectral index to revise the pharmacokinetic and pharmacodynamic model for each subject. Based on the post hoc Bayesian pharmacokinetics and individualized values of ke0, we found that the effect site propofol concentration for the clinical measures of drug effect was independent of the rate of rise of the effect site propofol concentration.

Given that the pharmacokinetics reported by Schnider et al.14,15 performed well in this study and that BIS was available to the computer throughout the study, these results suggested that real-time estimation of ke0 using model prediction of pharmacokinetics and electroencephalographic measure of drug effect may be a viable way to individualize target controlled infusions that target the site of drug effect.

We are unable to explain the discrepancy between the value of ke0 (0.17 min−1) and tpeak (2.7 min) in this study and the ke0 (0.46 min−1) and tpeak (1.7 min) reported by Schnider et al.15 and validated by Struys et al.23 The pharmacokinetics reported by Schnider et al.14 performed well in this study, so the basis of the discrepancy was entirely with the electroencephalographic hysteresis. One possibility is that the hysteresis was affected by the mode of administration. While Schnider et al. 23 performed both bolus and an infusion studies, it is likely that most of the information about plasma-effect site equilibration delay came from the bolus data. In contrast, the present study was entirely based on infusions and even the most rapid effect site ramp rate still required 5 minutes to achieve unresponsiveness. It may be that plasma effect site propofol equilibration varies between boluses (e.g., very rapid infusions [2.5 mg/kg over 20 seconds]) and more conventional infusions. This is a readily testable hypothesis. It might also be affected by the time delay in the BIS monitor (about 15 seconds), which we did not include in the model. Schnider et al., have centered the electroencephalographic epoch on the time point of the observation. We did not do this, and it could have modestly affected our estimates.

Early distribution kinetics determined the rate and extent of drug distribution to the brain and other tissues.24 Conventional pharmacokinetic models may overestimate the central volume of distribution because the complexity of intravascular mixing is ignored25 and thus the estimate of central volume of distribution is very dependent on the details of early drug sampling. If a standard mammillary pharmacokinetic model is to be used by a target controlled infusion system (and such models are the only ones presently incorporated in such systems), then the ideal pharmacokinetic profile should be derived from data obtained during and after a brief drug infusion.26 This is consistent with the observation that the most accurate results are with target controlled infusion devices using pharmacokinetic data sets derived from “slow injection” or “continuous infusion” studies.2729 However, both the Schnider pharmacokinetic study14 and the present study are based on infusions and rapid arterial sampling; thus, differences in study design can’t explain the differences in estimates of ke0.

Ludbrook et al. demonstrated that propofol decreases local cerebral blood flow in sheep30 and humans10 in a concentration-dependent manner. This might explain why the ke0 of propofol could change between a bolus, which might acutely decrease cerebral blood flow, and an infusion, where the changes in cerebral blood flow would be attenuated by the lower peak arterial concentration.

Lastly, Kuizenga et al. demonstrated that the addition of a second effect site improves the prediction of the EEG response to a propofol infusion.31 Interestingly, Ke0 estimated with parametric and nonparametric modeling of EEG31,32 and BIS data32 was almost identical (0.16–0.21min−1) to ours, while the median time to loss of responsiveness was 2.8 minutes.31

Interestingly, loss of response to the Automated Responsiveness Monitor (ARM) occurred at approximately the same concentration as a subject’s releasing a filled syringe and reproducibly occurred at a effect site propofol concentrations 15–40% less than those associated with loss of responsiveness (Fig. 10), providing a degree of protection from excessive sedation. Automated Responsiveness Monitor correlates well with other clinical measures of sedation, as well as with effect site propofol concentration. The effect site propofol concentration at which subjects lost response to ARM was independent of the rate of rise of concentrations over the range studied, provided that the correct rate of plasma-effect site equilibration was used.

In conclusion, it may be possible to estimate individual ke0 values from population estimates of pharmacokinetic parameters and real-time measurements of electroencephalographic effect, and thus optimize the pharmacokinetic/pharmacodynamic model in target-controlled infusion systems. For the range of infusions studied, the rate of rise in the effect site concentration does not affect propofol pharmacodynamics. Future studies need to address whether propofol’s rate of blood-brain equilibration is faster following bolus administration than following conventional infusions. Automated responsiveness appears useful as a measure of sedative drug effect.

Acknowledgments

Supported by Scott Laboratories, Ltd. (Lubbock, TX), NIH Grant GM 061655 (Bethesda, MD), the Gheens Foundation (Louisville, KY), the Joseph Drown Foundation (Los Angeles, CA), and the Commonwealth of Kentucky Research Challenge Trust Fund (Louisville, KY). Dr. Shafer is a consultant to Ethicon Endo-Surgery Inc., a Johnson & Johnson Co., which is developing a sedation delivery system. Mallinckrodt Anesthesiology Products, Inc. (St. Louis, MO) donated the thermocouples we used. The authors greatly appreciate the assistance of Randy Hickle, M.D., (CEO, Scott Laboratories, Inc.), Brett Moore, B.S., (Computer Programmer, Scott Laboratories, Inc.), and Jason Derouen, B.S., (Electrical Engineer, Scott Laboratories, Inc.), and Ellie Lekov, M.D., (Research Fellow, University of California at San Francisco). We would like to thank Nancy L. Alsip, Ph.D., (Medical Editor, University of Louisville) for her editorial assistance.

Footnotes

Received from the Outcomes Research™ Institute and Departments of Anesthesiology and Pharmacology, University of Louisville, Louisville, KY; the Department of Anesthesia and Perioperative Care, University of California at San Francisco; the Department of Anesthesia, Stanford University School of Medicine, Stanford, CA; the Palo Alto Veterans Administration Medical Center, Palo Alto, CA; and the Department of Anaesthesia, Royal Melbourne Hospital, Parkville, VIC, Australia.

References

1. Upton RN, Ludbrook GL. A model of the kinetics and dynamics of induction of anaesthesia in sheep: variable estimation for thiopental and comparison with propofol. Br J Anaesth. 1999;82:890–9. [PubMed]
2. Stokes DN, Hutton P. Rate-dependent induction phenomena with propofol: implications for the relative potency of intravenous anesthetics. Anesth Analg. 1991;72:578–83. [PubMed]
3. Rolly G, Versichelen L, Huyghe L, Mungroop H. Effect of speed of injection on induction of anaesthesia using propofol. Br J Anaesth. 1985;57:743–6. [PubMed]
4. Peacock JE, Lewis RP, Reilly CS, Nimmo WS. Effect of different rates of infusion of propofol for induction of anaesthesia in elderly patients. Br J Anaesth. 1990;65:346–52. [PubMed]
5. Jacobs JR, Reves JG. Effect site equilibration time is a determinant of induction dose requirement. Anesth Analg. 1993;76:1–6. [PubMed]
6. Kazama T, Ikeda K, Morita K, Kikura M, Ikeda T, Kurita T, Sato S. Investigation of effective anesthesia induction doses using a wide range of infusion rates with undiluted and diluted propofol. Anesthesiology. 2000;92:1017–28. [PubMed]
7. Gentry WB, Krejcie TC, Henthorn TK, Shanks CA, Howard KA, Gupta DK, Avram MJ. Effect of infusion rate on thiopental dose-response relationships. Assessment of a pharmacokinetic-pharmacodynamic model. Anesthesiology. 1994;81:316–24. [PubMed]
8. Upton RN, Ludbrook GL, Grant C, Martinez AM. Cardiac output is a determinant of the initial concentrations of propofol after short-infusion administration. Anesth Analg. 1999;89:545–52. [PubMed]
9. Kazama T, Ikeda K, Morita K, Ikeda T, Kikura M, Sato S. Relation between initial blood distribution volume and propofol induction dose requirement. Anesthesiology. 2001;94:205–10. [PubMed]
10. Ludbrook GL, Visco E, Lam AM. Propofol: relation between brain concentrations, electroencephalogram, middle cerebral artery blood flow velocity, and cerebral oxygen extraction during induction of anesthesia. Anesthesiology. 2002;97:1363–70. [PubMed]
11. Doufas AG, Bakhshandeh M, Bjorksten AR, Greif R, Sessler DI. Automated responsiveness test (ART) predicts loss of consciousness and adverse physiologic responses during propofol conscious sedation. Anesthesiology. 2001;94:585–92. [PubMed]
12. Doufas AG, Bakhshandeh M, Haugh GS, Bjorksten AR, Greif R, Sessler DI. Automated responsiveness test and bispectral index monitoring during propofol and propofol/N2O sedation. Acta Anaesthesiol Scand. 2003;47:951–957. [PubMed]
13. Shafer SL, Gregg KM. Algorithms to rapidly achieve and maintain stable drug concentrations at the site of drug effect with a computer-controlled infusion pump. J Pharmacokinet Biopharm. 1992;20:147–69. [PubMed]
14. Schnider TW, Minto CF, Gambus PL, Andresen C, Goodale DB, Shafer SL, Youngs EJ. The influence of method of administration and covariates on the pharmacokinetics of propofol in adult volunteers. Anesthesiology. 1998;88:1170–82. [PubMed]
15. Schnider TW, Minto CF, Shafer SL, Gambus PL, Andresen C, Goodale DB, Youngs EJ. The influence of age on propofol pharmacodynamics. Anesthesiology. 1999;90:1502–16. [PubMed]
16. Doufas AG, Bakhshandeh M, Bjorksten AR, Greif R, Sessler DI. A new system to target the effect site during propofol sedation. Acta Anaesthesiol Scand. 2003;47:944–950. [PubMed]
17. Chernik DA, Gillings D, Laine H, Hendler J, Silver JM, Davidson AB, Schwam EM, Siegel JL. Validity and reliability of the Observer’s Assessment of Alertness/Sedation Scale: study with intravenous midazolam. J Clin Psychopharmacol. 1990;10:244–51. [PubMed]
18. Plummer GF. Improved method for the determination of propofol in blood by high- performance liquid chromatography with fluorescence detection. J Chromatogr. 1987;421:171–6. [PubMed]
19. Sheiner LB, Stanski DR, Vozeh S, Miller RD, Ham J. Simultaneous modeling of pharmacokinetics and pharmacodynamics: application to d-tubocurarine. Clin Pharmacol Ther. 1979;25:358–71. [PubMed]
20. Beal SL, Sheiner LB. NONMEM Users Guide, parts I & II. San Francisco, University of California, 1980
21. Varvel JR, Donoho DL, Shafer SL. Measuring the predictive performance of computer-controlled infusion pumps. J Pharmacokinet Biopharm. 1992;20:63–94. [PubMed]
22. Minto CF, Schnider TW, Gregg KM, Henthorn TK, Shafer SL. Using the time of maximum effect site concentration to combine pharmacokinetics and pharmacodynamics. Anesthesiology. 2003;99:324–33. [PubMed]
23. Struys MM, De Smet T, Depoorter B, Versichelen LF, Mortier EP, Dumortier FJ, Shafer SL, Rolly G. Comparison of plasma compartment versus two methods for effect compartment--controlled target-controlled infusion for propofol. Anesthesiology. 2000;92:399–406. [PubMed]
24. Krejcie TC, Avram MJ. What determines anesthetic induction dose? It’s the front-end kinetics, doctor! Anesth Analg. 1999;89:541–4. [PubMed]
25. Chiou WL. Potential pitfalls in the conventional pharmacokinetic studies: effects of the initial mixing of drug in blood and the pulmonary first-pass elimination. J Pharmacokinet Biopharm. 1979;7:527–36. [PubMed]
26. Avram MJ, Krejcie TC. Using front-end kinetics to optimize target-controlled drug infusions. Anesthesiology. 2003;99:1078–86. [PubMed]
27. Barvais L, Cantraine F, D’Hollander A, Coussaert E. Predictive accuracy of continuous alfentanil infusion in volunteers: variability of different pharmacokinetic sets. Anesth Analg. 1993;77:801–10. [PubMed]
28. Shafer SL, Varvel JR, Aziz N, Scott JC. Pharmacokinetics of fentanyl administered by computer-controlled infusion pump. Anesthesiology. 1990;73:1091–102. [PubMed]
29. Vuyk J, Engbers FH, Burm AG, Vletter AA, Bovill JG. Performance of computer-controlled infusion of propofol: an evaluation of five pharmacokinetic parameter sets. Anesth Analg. 1995;81:1275–82. [PubMed]
30. Ludbrook GL, Upton RN, Grant C, Gray EC. Brain and blood concentrations of propofol after rapid intravenous injection in sheep, and their relationships to cerebral effects. Anaesth Intensive Care. 1996;24:445–52. [PubMed]
31. Kuizenga K, Kalkman CJ, Hennis PJ. Quantitative electroencephalographic analysis of the biphasic concentration-effect relationship of propofol in surgical patients during extradural analgesia. Br J Anaesth. 1998;80:725–32. [PubMed]
32. Kuizenga K, Proost JH, Wierda JM, Kalkman CJ. Predictability of processed electroencephalography effects on the basis of pharmacokinetic-pharmacodynamic modeling during repeated propofol infusions in patients with extradural analgesia. Anesthesiology. 2001;95:607–15. [PubMed]