Pharmacokinetics and Pharmacodynamics of Propofol Microemulsion and Lipid Emulsion after an Intravenous Bolus and Variable Rate Infusion

After obtaining the approval of the institutional review board of Asan Medical Center (Seoul, Korea) and written informed consent, 31 volunteers participated in this study. The volunteers who had medical problems, abnormal laboratory findings with clinical significance, or evidence of pregnancy were excluded. The subjects were stratified into three age groups (19–40, 41–64, and 65–79 yr), and each group included 5 male and 5 female volunteers.

The study was designed as a randomized, open-label, two-period, crossover phase I clinical trial. Each subject received both propofol formulations in a crossover fashion separated by a 7-day washout period, and the order of the drug administration was randomized.

Subjects fasted for 6 h before study drug administration. An 18-gauge angiocatheter was placed in a vein of the antecubital area. A second angiocatheter was placed in the contralateral radial artery for frequent blood sampling. Subjects were monitored with electrocardiography, pulse oximetry, end-tidal carbon dioxide concentration, and invasive blood pressure measurement (Datex-Ohmeda S/5; Planar Systems, Inc., Beaverton, OR) and Bispectral Index (BIS) (Aspect 2000; Aspect Medical Systems, Inc., Newton, MA). The smoothing rate for the measurement of BIS was set at 15 s. With baseline measurements, all of these data were recorded continually up to 180 min after administration of intravenous bolus dose. Subjects were hospitalized for 24 h after study drug administration.

Subjects received an initial intravenous bolus of propofol 2.0 mg/kg over 20 s. Time to loss of consciousness (LOC) was determined every 5 s by the loss of response to verbal command. After observing the lowest BIS value, propofol was infused at variable rates ranging from 0 to 12 mg · kg⁻¹· h⁻¹for 60 min to produce various BIS values approximately from 20 to 80 (fig. 2). Time to lowest BIS after an initial bolus was determined from the individual data file of BIS, in which BIS values were updated every 5 s. Time to recovery of consciousness (ROC) after discontinuation of infusion was determined by the recovery of response to verbal command.

Samples were collected in ethylenediaminetetraacetic acid (EDTA) tube and centrifuged for 10 min at 3,500 rpm. Plasma was stored at −70°C until assay.

1. Arterial blood samples (4 ml) were taken at preset intervals: 0, 1.5, 2, 2.5, 3, 3.5, 4, 6, 8, 10, 15, 20, 30, 40, 50, 60, 70, 90, 120, 150 and 180, 240, 300, 600, 720, and 1,200 min after bolus dose of propofol.

2. In addition, arterial samples were drawn when LOC and ROC were observed.

Propofol was isolated from human plasma by extraction using pretreatment with deproteinization and was determined by high-performance liquid chromatography with fluorescence detection. Plasma proteins were precipitated with acetonitrile. The supernatants were analyzed by high-performance liquid chromatography using a Capcell Pak C18 UG120 column and a mixture of methanol and 0.1% trifluoroacetic acid in water (75:25, vol/vol) as a mobile phase. The components of the column effluent were monitored by a fluorometric detector with excitation and emission wavelengths set at 276 and 310 nm, respectively. The lower limit of quantification of propofol was 8 ng/ml. The calibration curve was linear over the range of 8–25,000 ng/ml, with the coefficients of determination (R  ²) greater than 0.999 for all cases. Intraassay precision values were less than 1.5%. Interassay within-day and between-day precision values were less than 2.5% and 9.3%, respectively. Intraassay accuracy values were 100.8–103.2% of the nominal value. Interassay accuracy values were 98.5–103.7% of the nominal value.

Pharmacokinetic parameters were calculated by noncompartmental methods (WinNonlin Professional 4.1; Pharsight Corporation, Mountain View, CA). The area under the curve from the time of administration to the last measured concentration (AUC^last) was calculated by linear trapezoidal integration (linear interpolation). The area under the curve from administration to infinity (AUC^inf) was calculated as the sum of AUC^last+ C^last/λ^Z, with C^lastbeing the last measured concentration and λ^Zbeing the apparent terminal rate constant estimated by unweighted linear regression for the linear portion of the terminal log concentration–time curve. The maximal concentration (C^max) after an intravenous bolus of study drugs was determined from the observed data. Summary statistics were determined for each parameter.

Analysis of variance was performed with a linear mixed effects model that contained effects for sequence, subject nested within sequence, period, and formulation for logarithmically transformed data. The effect of subject was treated as random effect, and all other effects were treated as fixed effects. In addition, P  values were provided using F statistics (P < 0.05 indicated statistical significance). Ninety percent confidence intervals were constructed for the ratio of geometric means between microemulsion and lipid emulsion for AUC^last, AUC^inf, and C^max. It was concluded that the two formulations are bioequivalent if these 90% confidence intervals fall within the range of 80–125%.∥∥

Time to LOC, BIS at LOC, measured plasma concentration of propofol at LOC, and time to lowest BIS after an intravenous bolus were recorded. After discontinuation of infusion, time to ROC, BIS at ROC, and measured plasma concentration of propofol at ROC were recorded. Time elapse from LOC to lowest BIS was also recorded. Summary statistics were determined for each parameter. Also, we calculated 90% confidence intervals for the difference of these observations between microemulsion and lipid emulsion of propofol, based on an analysis of variance with a linear mixed effects model. A 90% confidence interval was constructed for the difference of arithmetic mean between microemulsion and lipid emulsion for each pharmacodynamic parameter (WinNonlin Professional 4.1).

Pharmacokinetic models were fitted using ADVAN 11 subroutine and the first-order conditional estimation procedure of NONMEM® V level 1.1, (GloboMax LLC, Ellicott City, MD). Pharmacodynamic models were fitted using ADVAN6 subroutine and first-order estimation procedure due to fitting failure of the first-order conditional estimation procedure of NONMEM®. A diagonal matrix was estimated for the different distributions of η’s.

The models were evaluated using statistical and graphical methods. The minimal value of the objective function (equal to minus twice the log likelihood) provided by NONMEM® was used as the goodness-of-fit characteristic to discriminate between hierarchical models using the log likelihood ratio test.8A P  value of 0.05, representing a decrease in objective function value of 3.84 points, was considered statistically significant (chi-square distribution, degrees of freedom = 1). The S-plus (MathSoft Inc., Seattle, WA)–based model-building aid Xpose 3.1 was used for graphical model diagnosis.9The covariates analyzed were age, sex, weight, height, body surface area,10and lean body mass.11Covariates significantly influencing the model were added cumulatively until the best description of the data was obtained. A backward elimination step was then performed by fixing the coefficient of each covariate, in turn, to zero. To compare the distribution of η’s to the normal distribution, we calculated the correlation of the points in the normal probability plot, in which the null hypothesis of the test is that the data come from a normal distribution.12,13 

The median weighted residual and median absolute weighted residual for the final pharmacokinetic models and the individual mean of weighted residuals for the pharmacodynamic models were calculated to examine the quality of prediction for the population. Weighted residual was calculated as (measured − predicted)/predicted.

Based on an analysis of variance with a linear mixed effects model, 90% confidence intervals were constructed for the difference of arithmetic means between microemulsion and lipid emulsion for calculated effect site concentration of propofol at the time of LOC and ROC, and for individually predicted pharmacokinetic parameters of propofol (WinNonlin Professional 4.1).

All of the observations in noncompartmental pharmacodynamic analysis were tested for the difference across any significant categorical covariate in the final pharmacodynamic models, using a t  test or Mann–Whitney rank sum test as appropriate (SigmaStat for Windows version 3.10; Systat Software, Inc., Point Richmond, CA). An intravenous bolus of propofol 2 mg/kg for both formulations was simulated to calculate the concentrations of plasma, effect site, and rapidly equilibrating and slowly equilibrating compartments over time.

The interindividual random variability on each of the model parameters (V¹, k¹⁰, k¹², k²¹, k¹³, k³¹) was modeled using a log-normal model. An additive plus proportional model was used for the residual random variability.

The pharmacodynamics were described using an effect compartment model in which k^e0, a first-order elimination rate constant characterizing effect site equilibration, is used to estimate the apparent effect site concentrations.14The relation of BIS with propofol effect site concentration was analyzed using a sigmoid E^maxmodel15:

where Effect is BIS, E⁰is the baseline measurement when no drug is present, E^maxis the maximum possible drug effect, Ce is the calculated effect site concentration of propofol, Ce⁵⁰is the effect site concentration associated with 50% maximal drug effect, and γ is the steepness of the concentration-versus -response relation.

Interindividual random variability of E^max, Ce⁵⁰, and k^e0was modeled using a log-normal model. Interindividual random variability of E⁰was modeled using an additive model, and no interindividual variability of γ was assumed. Residual random variability was modeled using an additive error model.

Using the observation of LOC and ROC, every calculated effect site concentration of propofol in the final pharmacodynamic model for BIS was joined to 0 (awake) or 1 (sleep). The relation between the probability of LOC and the effect site concentration of propofol was analyzed using a sigmoid E^maxmodel:

where Ce is the calculated effect site concentration of propofol, Ce⁵⁰is the effect site concentration associated with 50% probability of LOC, and γ is the steepness of the concentration-versus -response relation.

The likelihood, L, for the observed response, R (awake = 0, Sleep = 1) is described by the following equation:

where Prob is the probability of LOC.

Model parameters were estimated using the option “LIKELIHOOD LAPLACE METHOD = conditional” of NONMEM®. Interindividual random variability of Ce⁵⁰and γ was modeled using a log-normal model.

Safety profile of study drugs was evaluated by monitoring vital signs, pulse oximetry, end-tidal carbon dioxide concentration, treatment frequency of vasoactive drugs, and adverse events. Clinical laboratory tests were performed within 2 weeks before the administration of the first study drug and 24 h after administration of the second study drug.

Ephedrine or atropine was given if needed to maintain systolic pressure above 80 mmHg and heart rate above 50 beats/min. Each subject was preoxygenated with 100% oxygen, and the lungs were manually ventilated with 100% oxygen via  facemask, to maintain an end-tidal carbon dioxide concentration of 35–45 mmHg. Blood pressure, heart rate, body temperature, and respiratory rate were manually measured at 185, 190, 195, 200, and 210 min (postanesthetic recovery room) and 8 and 24 h (ward) after administration of the bolus dose. Safety data were analyzed using the McNemar test or repeated-measures analysis of variance as appropriate (P < 0.05 indicated statistical significance).

Thirty-one volunteers (15 male and 16 female) were recruited, and 1 female subject declined to receive microemulsion formulation after receiving lipid emulsion formulation as a first study drug. This subject was replaced and was not included in the noncompartmental pharmacokinetic and pharmacodynamic analyses, but was included in the population analysis of pharmacokinetics and pharmacodynamics and the analysis of safety profile. Weight, height, and age of the volunteers were 59 ± 10 kg, 163 ± 10 cm, and 47 ± 19 yr, respectively (mean ± SD). The rate and duration of infusion were 8.7 ± 1.6 mg/min and 61.2 ± 2.4 min for lipid emulsion and 8.4 ± 1.8 mg/min and 62.1 ± 4.9 min for microemulsion, respectively. The total amount of propofol administered was 530 ± 96 mg for lipid emulsion and 523 ± 137 mg for microemulsion. None of these showed significant differences between lipid emulsion and microemulsion of propofol.

In all subjects, at least 80% of the total AUC was covered by measured concentrations. The arithmetic means of the AUC indicate differences with respect to the extent of absorption of the formulations of approximately 1.5%. The intersubject variability was similar for both formulations (table 1). The analyses of variance did not indicate any differences between formulations for either sequence or period effects. The 90% confidence intervals for the difference of log-transformed AUC and C^maxare included in the acceptance range for bioequivalence (80–125%).

A variety of pharmacodynamic endpoints and propofol plasma concentrations at LOC and ROC for both study drugs are found in table 2. When the 90% confidence intervals include 100%, no significant differences between the two formulations are shown for those parameters.

The pharmacokinetics of both study drugs were best described by a three-compartment model. The results of the final pharmacokinetic models of both study drugs are summarized in table 3.

The final model of lipid emulsion included age as a significant covariate for k¹⁰. The following equation describes the final model for lipid emulsion:

The final model resulted in an improvement in the objective function (7.32; P = 0.0068) compared with the basic model. All of the η’s except η for k¹⁰, k¹², k³¹of the final pharmacokinetic model were normally distributed. From age 19 to 79 yr, the typical values of Cl¹of lipid emulsion propofol decreased by approximately 22.5%.

Sex and age were significant covariates for k²¹and k¹³in the final pharmacokinetic model of microemulsion, respectively. The typical value of k²¹was 0.118 for females and 0.0804 for males. The following equation describes k¹³in the final model for microemulsion:

The final model resulted in an improvement in the objective function (18.23; P = 0.0001). All of the η’s of the final pharmacokinetic model were normally distributed. The typical value of V²of microemulsion was 27.8 l for males and 18.9 l for females. From age 19 to 79 yr, the typical values of V³and Cl³of microemulsion increased by approximately 47.5%.

Plots for predicted versus  measured concentrations of propofol for both formulations are shown in figure 3. Ninety percent confidence intervals for the difference of individually predicted pharmacokinetic parameters between microemulsion and lipid emulsion are found in table 4.

The estimates of population parameters of the final pharmacodynamic models for lipid emulsion or microemulsion of propofol are found in table 5. Plots for predicted versus  observed BIS in the subjects with the lowest and highest absolute values of the individual mean of weighted residuals for lipid emulsion and microemulsion formulation of propofol are shown in figure 4.

Sex was a significant covariate for Ce⁵⁰in the final model of lipid emulsion. The typical value of Ce⁵⁰was 2.62 μg/ml for females and 1.89 μg/ml for males. Objective function value was decreased by 49.4 (P < 0.0001), compared with the basic model. The η’s of k^e0and E^maxin the final model of lipid emulsion were normally distributed.

In the final model of microemulsion, sex was a significant covariate for k^e0and Ce⁵⁰. The typical value of k^e0was 0.167 for females and 0.0725 for males. The typical value of Ce⁵⁰was 2.23 μg/ml for females and 1.9 μg/ml for males. Objective function value was decreased by 234.2 (P < 0.0001), compared with the basic model. The η’s of k^e0, E⁰, and E^maxin the final model of microemulsion were normally distributed.

Simulation of an intravenous bolus of propofol 2 mg/kg for both formulations is presented in figure 5. For microemulsion, simulation was performed in males and females (k^e0= 0.0725 for males and 0.167 for females), but only in males for lipid emulsion (k^e0= 0.1220).

Because sex was a common significant covariate in the final pharmacodynamic models of both formulations, all of the observations in table 2were tested for the difference between females and males. For lipid emulsion of propofol, the total amount of propofol administered during study period did not show any significant difference between females (mean ± SD: 513 ± 107 mg) and males (562 ± 85 mg), but the time to ROC after discontinuation of infusion was significantly shorter in females (median, 25–75%: 5.9, 3.8–7.1 min) than in males (9.4, 7.1–14.0 min) (P = 0.011).

For microemulsion formulation, the time to LOC after an intravenous bolus was significantly shorter in females (0.3, 0.2–0.4 min) than in males (0.4, 0.3–0.7 min) (P = 0.010). Although the total amount of propofol administered during the study period was significantly larger in females (596 ± 152 mg) than in males (450 ± 67 mg), time to ROC after discontinuation of infusion was significantly shorter in females (4.5 ± 3.3 min) than in males (14.1 ± 8.3 min) (P < 0.001).

The relation between the probability of LOC and effect site concentration of propofol for lipid emulsion and microemulsion formulation and LOC pharmacodynamic parameter estimates are shown in figure 6.

The calculated effect site concentrations of propofol at the time of LOC (mean ± SD) were 1.7 ± 0.6 μg/ml for lipid emulsion and 1.7 ± 0.4 μg/ml for microemulsion of propofol. The ratio based on an analysis of variance with a linear mixed effects model was 99.3%. The 90% confidence interval for the difference between microemulsion and lipid emulsion was 89.8–108.9%. The calculated effect site concentrations of propofol at the time of ROC (mean ± SD) were 2.2 ± 0.7 μg/ml for lipid emulsion and 2.2 ± 0.8 μg/ml for microemulsion of propofol. The ratio based on an analysis of variance with a linear mixed effects model was 96.5%. The 90% confidence interval for the difference between microemulsion and lipid emulsion was 83.8–109.8%.

Overall, a total of 50 adverse events were reported in 31 subjects during study period. Eighteen (58.1%) of 31 subjects experienced adverse events after lipid emulsion dosing, and 14 (46.7%) of 30 subjects experienced adverse events after microemulsion dosing.

The adverse events after lipid emulsion dosing were mild (n = 17), moderate (n = 7), and severe (n = 1). The adverse events after microemulsion dosing were mild (n = 18), moderate (n = 5), and severe (n = 2). No serious adverse event occurred for either formulation. The adverse events after lipid emulsion dosing had probable (n = 20), possible (n = 1), unlikely (n = 1), and no (n = 3) causal relation to the study drug. The adverse events after microemulsion dosing had probable (n = 24) and unlikely (n = 1) causal relation to the study drug. All of the adverse events completely resolved without sequelae. The frequency and severity of the adverse events and the causal relation of adverse events to study drugs did not show significant differences between lipid emulsion and microemulsion. The number of subjects per each adverse event, including the frequency of concurrent treatments, is found in table 6.

Arterial oxygen saturation was maintained at or above 95% throughout study period in all subjects. Blood pressure, heart rate, body temperature, respiratory rate, and clinical laboratory tests did not show significant differences between lipid emulsion and microemulsion.

Aquafol and Diprivan® were bioequivalent. Regardless of formulations, time to peak effect after an intravenous bolus, which was assessed by BIS, was 1.5 min. Morey et al.  3reported that significantly greater doses of propofol were required to induce anesthesia with propofol microemulsions irrespective of surfactant concentration or type than with propofol macroemulsion. However, the clinical characteristics at induction and recovery, including bolus and cumulative doses, were not different between microemulsion and lipid emulsion in this study.

Individually predicted V³and hence V^dssof microemulsion were significantly smaller than those of lipid emulsion. Individually predicted half-lives were shorter in microemulsion than in lipid emulsion. These findings suggest that propofol in microemulsion is less extensively distributed to peripheral tissues and is more rapidly disposed than propofol in lipid emulsion. We speculated that the former may be attributed to a relatively lower tissue/blood partition coefficient of microemulsion system as a propofol vehicle, and the latter may be explained by higher concentration of free propofol in aqueous phase, faster spontaneous destabilization of the microemulsion nanodroplets to release propofol, and relatively smaller size of the microemulsion nanodroplets.

The metabolic clearance of lipid emulsion was decreased with increasing age, which is the same finding as in previous studies.16,17Unlike lipid emulsion, the effects of age on the pharmacokinetic characteristics of microemulsion were equal increases of V³and Cl³with increasing age. Hence, age is not a factor influencing the rate of change in the amount of drug in the slowly equilibrating compartment.

The lack of weight being a significant covariate in this study is probably due to the small weight distribution in our population, which is a classic problem among volunteer studies. In general, a weight-independent model is difficult to apply in a target-controlled infusion system, and the clinical population of patients mostly requires a weight-dependent model.

For lipid emulsion, faster recovery of consciousness in females may be explained by the higher typical value of Ce⁵⁰, which suggests that females may be pharmacodynamically more resistant to the effect of propofol on the central nervous system than males. Gan et al.  18found that females awoke significantly faster after discontinuation of the infusion of lipid emulsion formulation. Ward et al.  19concluded that pharmacokinetic differences were primarily responsible for sex differences. In this study, we did not find any pharmacokinetic characteristic of lipid emulsion propofol to explain faster emergence for females.

Slower onset of sleep in males receiving microemulsion formulation may be partly explained by the higher volume of rapidly equilibrating compartment of microemulsion in males than in females, which in turn causes a lower peak effect site concentration of propofol. The rate of decrease in the effect site concentration of propofol is higher in females receiving microemulsion than in males receiving microemulsion (fig. 5). The effect site concentrations of propofol after approximately 17 min after an intravenous bolus are lower in females receiving microemulsion than in males receiving microemulsion. These findings are attributed to the higher k^e0of microemulsion in females, which may partly explain faster emergence in females. For microemulsion, the higher typical value of Ce⁵⁰in females should be regarded as a factor for faster sleep and emergence, which is the same finding as in lipid emulsion. Females seem to be more resistant to the effect of propofol on the central nervous system regardless of formulations. For microemulsion formulation, a combination of pharmacokinetic and pharmacodynamic factors is responsible for the sex difference.

In a previous study, the plasma concentration necessary for 50% of the participants to be awake at steady state condition was 1.68 μg/ml, and an age effect on the steady state plasma C⁵⁰of propofol for being asleep was observed.20In this study, the effect site concentration of propofol associated with 50% probability of LOC was 1.73 μg/ml for lipid emulsion and 1.67 μg/ml for microemulsion, but an age or formulation effect was not observed. There was little difference of the concentrations (plasma or effect site) of propofol associated with 50% probability of loss of consciousness between steady state and non–steady state conditions.

Based on information on toxicologic data provided by BASF, LD⁵⁰of polyethylene glycol 660 hydroxystearate is 3.16 g/kg for male mice and 5 g/kg for female mice. Tetrahydrofurfuryl alcohol polyethylene glycol is used as a solvent in parenteral pharmaceutical formulations and is generally regarded as a nontoxic and nonirritant material. Our previous preclinical studies of tetrahydrofurfuryl alcohol polyethylene glycol revealed that the approximate lethal dose in rats was more than 4 g/kg, no-observed-effect level, and approximate lethal doses in beagle dogs were 1 and 5 g/kg, respectively (June 28, 2003, single intravenous dose toxicity study of tetrahydrofurfuryl alcohol polyethylene glycol in rats, study No. B03069, intravenous dose-escalation study of tetrahydrofurfuryl alcohol polyethylene glycol in beagle dogs, study No. B03071, Sun-Hee Kim, Ph.D., and Zai-Zhi Huang, Ph.D., Biotoxtech Co., Ltd., Ochang, Korea). Nine men with liver cirrhosis and nine other hospital patients with normal liver function were each injected intravenously with approximately 6.5 g tetrahydrofurfuryl alcohol polyethylene glycol. No overt adverse events were reported in any of the subjects.21There are no available data or literature to clearly indicate no observed effect level of these materials in humans, which should be carefully studied in other well-designed clinical trials. However, we found that the adverse events caused by microemulsion were not different from those by lipid emulsion with respect to type, incidence, and severity of adverse events. Along with the evidence of bioequivalence and the pharmacodynamic endpoints in table 2, this suggests that microemulsion is therapeutically equivalent with lipid emulsion.##

In conclusion, the clinical characteristics at induction and recovery did not show significant differences between Aquafol (microemulsion) and Diprivan® (lipid emulsion). Aquafol and Diprivan® were bioequivalent. Population pharmacokinetic and pharmacodynamic modeling revealed a variety of differences between the two formulations. Aquafol showed a safety profile similar to that of Diprivan®.

The authors thank Sook-Kyung Seo, M.P.H. (Unit Manager, Medical Records Department, Asan Medical Center, Seoul, Korea), and Yoo-Mi Kim, M.P.H. (Researcher, Department of Health Care Industry, Korea Health Industry Development Institute, Seoul, Korea), for the preparation of data used in this study. They also thank Yun-Ah Kim, B.S. (Researcher, Clinical Research Center, Asan Medical Center), for the measurement of propofol concentration.