Professional Documents
Culture Documents
Using Front-End Kinetics To Optimize Target-Controlled Drug Infusions
Using Front-End Kinetics To Optimize Target-Controlled Drug Infusions
Anesthesiology 2003; 99:1078 86 2003 American Society of Anesthesiologists, Inc. Lippincott Williams & Wilkins, Inc.
Background: The mode of drug administration, blood sam- can be derived from data obtained during and after a brief drug
pling schedule, and sampling site affect the pharmacokinetic infusion.
model derived. The present study tested the hypothesis that
three-compartment pharmacokinetic model parameters de-
rived from arterial drug concentrations obtained after rapid
EARLY drug distribution kinetics (front-end kinetics) de-
intravenous administration can be used to design a target-con- termine the rate and extent of both drug distribution to
trolled drug infusion (TCI) that deviates minimally from the the brain and its dilution by distribution to indifferent
target. tissues.1 Both cardiac output and its peripheral distribu-
Methods: Arterial thiopental concentration data obtained tion are important determinants of the early drug con-
from the moment of injection in a previous study of five dogs
were used. Three three-compartment models were constructed,
centrationversustime relation of intravenously admin-
one based on early concentrations classically obtained at 1, 2, istered drugs and interindividual variability in response
and 3 min; another using all concentrations obtained beginning to rapidly acting intravenous anesthetics. The purpose of
with the thiopental recirculation peak; and the last with the the current study is to illustrate the importance of accu-
initial distribution volume (VC) fixed to the sum of VC and rate characterization of front-end kinetics to the optimal
the nondistributive volume of the recirculatory model from the
earlier study. Using these models, TCIs were designed that
design of target-controlled drug infusions.
would maintain 20 g/ml thiopental concentrations in VC for Traditional pharmacokinetic models (fig. 1) are based
60 min if simulated with the models used in their design. Drug on the simplifying assumption that intravenously admin-
concentrations resulting from these TCIs were then simulated istered drugs mix instantaneously and completely within
using recirculatory model kinetics, and prediction errors were an initial distribution volume (central volume, VC) that
evaluated.
includes, at a minimum, intravascular space.2 In reality,
Results: Models with VCs estimated from intermittent or fre-
quent early blood concentrations overestimated not only VC but the volume of distribution of a drug expands with a time
also the volume and clearance of the rapidly equilibrating tis- course dependent on the physiologic environment and
sues, and their TCIs significantly overshot the target. With VC the chemical characteristics of the drug.35 As a result,
fixed to recirculatory model parameters, drug distribution was the earlier one obtains blood samples after rapid intra-
described in a manner consistent with that of the recirculatory
venous drug administration, and the smaller will be the
model, and the TCI deviated minimally from the target. A sim-
ilar three-compartment model was derived from data obtained estimate of VC.6,7 Nonetheless, conventional pharmaco-
from a simulation of a 2-min infusion using recirculatory ki- kinetic models overestimate VC because they ignore the
netic parameters. complexity of intravascular mixing.8 When pharmacoki-
Conclusions: Because three-compartment models based on netic models in which VC is overestimated are used to
drug concentration histories obtained after rapid intravenous
design target-controlled intravenous drug infusions, drug
administration do not characterize VC accurately, TCIs based on
them produce concentrations exceeding the target. A model concentrations not only greatly exceed the target con-
capable of producing TCIs deviating minimally from the target centration in the first minutes after commencing the
infusion but may also significantly exceed it long after
starting the infusion.9
We have developed a recirculatory multicompartmen-
This article has an accompanying editorial view. Please see:
Egan TD, Shafer SL: Target controlled infusions for intrave- tal pharmacokinetic model that describes drug disposi-
nous anesthetics: Surfing USA not! ANESTHESIOLOGY 2003; tion from the moment of rapid intravenous injection (fig.
99:1039 41. 2).10 12 This model addresses Chious concerns about
traditional mammillary multicompartmental analysis.13,14
In the fit of the recirculatory model to the data, the
* Associate Professor of Anesthesiology, Professor of Anesthesiology. concentration at time zero is zero, and there is a delay
Received from the Department of Anesthesiology, Feinberg School of Medi- between the time drug is administered and the time drug
cine, Northwestern University, Chicago, Illinois. Submitted for publication No-
vember 1, 2002. Accepted for publication April 20, 2003. Supported in part by appears at the sampling site. The model fits the early
grant No. GM43776 from the National Institutes of Health, Bethesda, Maryland. arterial drug concentrations of samples obtained fre-
Presented in part at the Annual Meeting of the American Society of Anesthesiol-
ogists, Orlando, Florida, October 1316, 2002, and the Annual Meeting of the quently soon after rapid intravenous input that resemble
American Society for Clinical Pharmacology and Therapeutics, Washington, DC, the drug concentration profiles resulting from a zero-
April 25, 2003.
Address reprint requests to Dr. Avram: Department of Anesthesiology, Fein-
order intravenous infusion. Pulmonary drug uptake is an
berg School of Medicine, Northwestern University, 303 East Chicago Avenue, integral part of the model.15 Intravascular mixing is char-
Ward 13-199, Chicago, Illinois 60611-3008. Address electronic mail to:
mja190@northwestern.edu. Individual article reprints may be purchased through
acterized by the recirculatory model, as is the role of
the Journal Web site, www.anesthesiology.org. cardiac output in drug distribution. Finally, arterialve-
Grasela et al.19 as modified in our laboratory.10 Plasma Arterial ICG and thiopental concentrationversus
thiopental concentrations were measured within 24 h of time data before evidence of recirculation (i.e., first-pass
sample collection using a high-performance liquid chro- data) were weighted uniformly and fit, independently, to
matographic technique developed in our laboratory.20 the sum of two Erlang distribution functions using
To interpret intercompartmental clearances in relation TableCurve2D (version 3.0; SPSS, Chicago, IL) on a Pen-
to blood flow, the recirculatory models were con- tium-based personal computer to reflect the heterogene-
structed on the basis of whole blood ICG and thiopental ity in the distribution of transit times in the pulmonary
concentrations. circulation and the pulmonary tissue distribution of thio-
pental during this time.15 The thiopental pulmonary tissue
Recirculatory Pharmacokinetic Model volume (VT-P) is the difference between the thiopental
The pharmacokinetic modeling method (fig. 2) has central volume (thiopental mean transit time cardiac
been described in detail previously.11,12 It is based on output) and the central intravascular volume determined
the approach described by Jacquez21 for obtaining infor- by ICG (ICG mean transit time cardiac output).
mation from outflow concentration histories, the so- In subsequent pharmacokinetic analysis, these descrip-
called inverse problem. Thiopental distribution was an- tions of the central circulation were incorporated as
alyzed as the convolution of its intravascular behavior, parallel linear chains or delay elements into independent
determined by the pharmacokinetics of concomitantly recirculatory models for the individual markers using
administered ICG, and tissue distribution kinetics.11 SAAM II (SAAM Institute, Seattle, WA) implemented on a
Pentium-based personal computer.15,16 The concentra-
tiontime data were weighted, assuming a proportional
variance model, in proportion to the inverse of the
square of the observed value.22 Systematic deviations of
the observed data from the calculated values were
sought using the one-tailed one-sample runs test, with
P 0.05, corrected for multiple applications of the runs
test, as the criterion for rejection of the null hypothesis.
Model misspecification was sought by visual inspection
of the measured and predicted marker concentration
versustime relations.
In general, peripheral drug distribution can be lumped
into identifiable (i.e., mathematically distinct) volumes
(V) and clearances (Cl) (fig. 2): nondistributive periph-
eral pathways (VND and ClND), rapidly (fast) equilibrating
tissues (VT-F and ClT-F), and slowly equilibrating tissues
(VT-S and ClT-S). The single identifiable nondistributive
peripheral pathway in the thiopental model (VND and
ClND), determined by the recirculation peak, represents
Fig. 2. The general model for the recirculatory pharmacokinet- blood flow that quickly returns the drug to the central
ics of indocyanine green and thiopental.11 Cardiac output (CO) circulation after minimal apparent tissue distribu-
flows through the central circulation, which is defined by the
delay elements (VC). All delay elements are represented gener- tion.11,12 In the thiopental model, the parallel rapidly
ically by rectangles surrounding four compartments, although and slowly equilibrating tissues are the fast (VF and ClF)
the number of compartments needed in a delay varied between and slow (VS and ClS) compartments of traditional three-
2 and 30. The pulmonary tissue volume (VT-P), a subset of VC, is
calculated for thiopental by subtracting the VC of indocyanine compartment pharmacokinetic models, respectively,
green from that of thiopental. Beyond the central circulation, whereas the central circulation and nondistributive pe-
CO distributes to numerous circulatory and tissue pathways ripheral pathway(s) are detailed representations of the
which lump, on the basis of their blood volume to flow ratios or
tissue volume to distribution clearance ratios (mean transit ideal central volume (VC) of the traditional multicom-
times), into the volumes (V) and clearances (Cl) of the nondis- partmental model (fig. 1).23 Because of the direct corre-
tributive peripheral pathway (VND, ClND) and the fast (VT-F, ClT-F) spondence between the recirculatory model and com-
and slow (VT-S, ClT-S) tissue volume groups. The elimination
clearances (ClEs) are modeled from the arterial sampling site partmental models, elimination clearance (ClE) was
without being associated with any particular peripheral circuit. modeled from the arterial (sampling) compartment to
The parallel rapidly and slowly equilibrating tissues are the fast enable comparison of these results with previous ones.
(VF and ClF) and slow (VS and ClS) compartments of traditional
three-compartment pharmacokinetic models, respectively,
whereas the central circulation and nondistributive peripheral Three-compartment Kinetic Models of Data after
pathway(s) are detailed representations of the ideal central Rapid Intravenous Drug Administration
volume (VC) of the traditional multicompartmental mammillary
model (fig. 1). The dotted ellipse surrounds the components of The ability of traditional compartmental models to
the ideal central volume of a three-compartment model. characterize early drug disposition after rapid intrave-
nous drug administration was tested by comparing the and flow fractions to demonstrate the superiority of
parameters of three different three-compartment phar- infusion-derived pharmacokinetic parameters for com-
macokinetic models of thiopental disposition with those puter-controlled drug infusions. Therefore, we investi-
of the recirculatory model. Each dogs thiopental con- gated the possibility that a three-compartment model
centrationversustime data were fit to three-compart- derived from a short-term infusion study would describe
ment pharmacokinetic models using SAAM II. The first peripheral drug distribution in a manner similar to that
three-compartment model (model 1) was fit to data of the recirculatory model and could be used to design
collected at 1, 2, and 3 min and all data collected sub- target-controlled drug infusions with resultant concen-
sequently to emulate a traditional intense sampling trations that deviate minimally from the target.
postbolus sampling schedule. The second three-com- We thus evaluated a fourth three-compartment model
partment model (model 2) was fit to all data collected that was based on data obtainable during and after a brief
beginning with the thiopental recirculation peak (fig. 3), thiopental infusion (model 4). Because infusion-based
the highest concentration observed after first pass, to thiopental concentrationversustime data were not ob-
obtain the smallest VC it is possible to derive from the tained as a part of the study on which the current report
postbolus data. The central circulation and nondistribu- is based,17 data were generated from simulations of
tive peripheral pathway(s) of the recirculatory model are 2-min, 30-mg/min thiopental infusions using each dogs
detailed representations of the ideal central volume of recirculatory pharmacokinetic model parameters. Thirty-
the traditional multicompartmental model.23 Therefore, two predicted arterial thiopental concentrations were
the third three-compartment model (model 3) had its VC obtained every half minute during the 2 min infusion; at
fixed to the sum of the corresponding thiopental recir- 3, 3.5, 4, 5, and 6 min; every 2 min to 20 min; every 5
culatory model VC and VND and was fit to all data col- min to 30 min; every 10 min to 60 min; every 15 min to
lected beginning with the recirculation peak. 90 min; every half hour to 3 h; and every hour to 10 h.
Normally distributed random error with a mean error of
Simulations of Target-controlled Drug Infusions 0% and an SD of 5% was introduced to the data using the
The descriptions of early drug disposition by the three- random number generation function in Excel (Microsoft,
compartment models were further evaluated by testing Seattle, WA). The data were then fit to three-compart-
their ability to design error-free target-controlled drug ment pharmacokinetic models using SAAM II. From
infusions. Each dogs three-compartment pharmacoki- these models, target-controlled drug infusions were de-
netic parameters were used to design infusions that veloped and simulated in the manner described above.
would maintain a constant VC thiopental concentration
of 20 g/ml for 60 min if simulated with the models used
Statistical Analysis
to design them (the BET dosing regimen).24 The thio-
All pharmacokinetic variables were tested by both the
pental concentrations resulting from infusions based on
Kolmogorov-Smirnov test for a normally distributed pop-
each dogs three-compartment models were then simu-
ulation and the Levene median test for equal variance
lated using the respective dogs recirculatory model. The
(SigmaStat Statistical Software; SPSS). Data passing
prediction errors of these simulations were evaluated by
these tests were treated as interval and were represented
calculating the area between the predicted concentra-
as mean and SD. These data were then compared among
tion history and the target concentration (20 g/ml) for
models using a one-way repeated-measures analysis of
the duration of the infusion (infusion AUC).
variance with post hoc analysis by Tukey multiple com-
parison test.
Three-compartment Model Fit to Infusion and
Postinfusion Data
The above results clearly indicated that the only three-
compartment model based on drug concentration histo- Results
ries obtained after rapid intravenous drug administration
that differed from the recirculatory model only in its The thiopental pharmacokinetic parameters described
description of VC and could be used to produce targeted by the recirculatory model and the four three-compart-
drug infusions that deviate minimally from the target ment models are listed in table 1; the recirculatory model
concentration was the one based retrospectively on the parameters have been reported previously.17 The blood
recirculatory model (model 3). Chiou,13,14 in his indict- thiopental concentrationversustime relations of the
ment of traditional mammillary multicompartmental various data sets were well characterized by the models
analysis, suggested that better estimates of VC might be (fig. 3). Differences in the fits of the models to the data
obtained from studies in which the drug of interest was are most apparent in the first 2 min after rapid intrave-
administered by a short-term intravenous infusion. Wada nous drug administration (fig. 3) but are not apparent
and Ward25 used a recirculatory model of alfentanil dis- when the fit to the entire concentration history is exam-
position based on literature values for tissue volumes ined (fig. 3, inset).
Recirculatory, based on all bolus 1.11 0.70 14.26 37.05 53.12 1.18 1.29 0.36 0.17 3.00** 7.6
data from 0 min on (0.28) (0.34) (7.42) (10.76) (14.40) (0.47) (0.49) (0.12) (0.04) (0.89) (4.8)
Model 1: Three-compartment, 3.37 NA 15.85 38.58 57.80 NA 2.03 0.40 0.17 2.60 162.0
based on 1, 2, 3 min and all (1.06) (7.79) (10.80) (16.16) (0.67) (0.14) (0.05) (0.59) (64.3)
subsequent bolus data
Model 2: Three-compartment, 3.04 NA 15.71 38.54 57.29# NA 1.81 0.38 0.17 2.36 124.6
based on all bolus data from (1.34) (7.79) (10.78) (15.94) (0.68) (0.13) (0.05) (0.60) (37.8)
recirculatory peak on
Model 3: Three-compartment, 1.80 NA 13.69 38.23 53.73 NA 1.41 0.39 0.17 1.96 17.3
based on bolus data, VC fixed (0.60) (7.34) (10.25) (14.70) (0.53) (0.13) (0.05) (0.47) (5.9)
to recirculatory model
VC VND
Model 4: Three-compartment, 1.47 NA 13.70 36.89 52.05 NA 1.28 0.39 0.17 1.83 3.79
based on simulated 2-min (0.44) (8.12) (10.42) (14.11) (0.45) (0.15) (0.04) (0.39) (21.04)
infusion data
Three-compartment Models Fit to Postbolus Data of the recirculatory model, which represents 39% of Cl
(Models 1 and 2) of that model (i.e., 39% of cardiac output).
Many of the parameters of the three-compartment The effect of the differences in the parameters of the
models fit to postbolus arterial drug concentration data three-compartment models fit to the postbolus data col-
collected either according to a traditional intense sam- lected according to either a traditional sampling sched-
pling schedule (model 1) or beginning with the recir- ule (model 1) or beginning with the recirculation peak
culation peak (model 2) differed significantly from those (model 2) from those of the recirculatory model are
of the recirculatory model (table 1). Of the compartmen-
tal volumes, only the VSs of the three-compartment mod-
els did not differ from those of the recirculatory model.
The VCs of the three-compartment models were more
than two and one half times the size of that of the
recirculatory model, whereas the VFs were more than
10% larger than that of the recirculatory model; as a
result, the volumes of distribution at steady state (VSSs)
were approximately 8% larger than that of the recircu-
latory model. Although the ClSs and ClEs of these models
did not differ significantly from those of the recirculatory
model, the ClFs of these three-compartment models
were nearly 60% and 40% larger than those of the recir-
culatory model, respectively. Because the ClFs of these
models exceeded that of the recirculatory model, the Fig. 3. Arterial blood thiopental concentration histories for the
sum of all clearances (Cl) of the traditional sampling first 5.0 min (illustrating the first- and second-pass peaks) after
rapid right atrial injection of 100 mg in a dog anesthetized with
model (model 1) was just 14% less than Cl of the 1.5% halothane in oxygen. Arterial blood thiopental concentra-
recirculatory model (i.e., cardiac output), whereas Cl tion histories for 10 h after injection are illustrated in the inset.
of the recirculation peak model (model 2) was only 22% The symbols represent measured drug concentrations, and the
lines represent fits of the recirculatory model and the three-
less than Cl of the recirculatory model, despite the fact compartment models to the data, which are superimposed on
that neither three-compartment model included the ClND this time scale. VC initial distribution volume.
sample after rapid intravenous drug administration; the on which to base target-controlled drug infusions would
smaller will be the back-extrapolated concentration at be one in which data are collected during and after a
time zero and the larger the estimate of VC. Such an brief drug infusion.
analysis ignores the information available from the first- The ability of a target-controlled drug infusion to pro-
pass concentration-versus-time relation, including the duce stable drug concentrations at or near the target
facts that the concentration at time zero is really zero, concentration depends on the pharmacokinetic data set
that there is a temporal lag between the time of intrave- used in designing the infusion.27 Several authors have
nous drug administration and the time drug appears at an compared the performance of infusion regimens based
arterial sampling site, that there may be significant drug on different pharmacokinetic parameter sets. For exam-
uptake by the lung and washout from it, and that mixing ple, Raemer et al.28 compared alfentanil infusions based
even within what might be considered the true VC is not on population pharmacokinetics derived from data from
instantaneous.13,14 several studies with those based on pharmacokinetics
Nonetheless, as the current study has demonstrated, it derived in a smaller study without the benefit of popu-
is possible for a traditional three-compartment model to lation kinetic analysis. The authors reported that the
describe drug distribution in a manner consistent with population kinetics based infusions had median abso-
that of the recirculatory model if one circumvents the lute performance errors that exceeded 50% and were
assumptions underlying the estimation of VC after rapid especially inaccurate immediately after the infusion tar-
intravenous drug administration. In the case of the cur- get changed, whereas parameters derived from less so-
rent study, that was first done by fixing VC to the sum of phisticated pharmacokinetic analysis produced infusions
VC and VND from the recirculatory model (model 3). having median absolute performance errors less than
With a reasonable estimate of VC, the three-compart- 20%. The authors could not explain the differences in
ment model fit to the data obtained after rapid intrave- the performance of the infusions based on these two
nous drug administration described peripheral drug dis- data sets but noted that the kinetics used to design the
tribution in a manner that was nearly identical to its better-performing infusion were derived in a study in
description by the recirculatory model (table 1), and the which alfentanil was administered over several minutes
target-controlled drug infusion based on it produced and had a much smaller VC. The parameters for the
concentrations that deviated minimally from the target poorly performing infusion were derived from a study in
(fig. 4). which alfentanil was administered by rapid intravenous
Given that a three-compartment model is capable of infusion. Barvais et al.29 similarly observed that, of the
describing drug disposition in a manner consistent with alfentanil infusions derived from nine different pharma-
that of a recirculatory model, the question arose as to cokinetic data sets, those with the best performance
how to conduct an experiment that would allow one to (i.e., median absolute performance errors less than 50%)
arrive at such a model without first deriving the recircu- were those derived from studies in which the drug was
latory model. Chiou suggested that better estimates of VC administered by slow injection or continuous infusion.
might be obtained from studies in which the drug of Similar observations have been made for fentanyl,30 lido-
interest is administered by a short-term intravenous in- caine,31 and propofol.32
fusion because fitting a three-compartment model to It might seem from the above that the pharmacokinet-
data obtained during and after an infusion would avoid ics reported for a given drug depend on the mode of
many of the erroneous assumptions made when fitting drug administration (e.g., bolus vs. infusion). An impor-
data obtained after rapid intravenous drug administra- tant point of the current study is that the pharmacoki-
tion.13,14 Wada and Wards alfentanil simulations con- netics of a drug are independent of the mode of admin-
firmed the wisdom of such an approach.25 Therefore, istration, but, in the absence of saturation effects, the
we generated such data by simulating drug concentra- reported pharmacokinetics depend on the method of
tions during a 2-min thiopental infusion and for the study. If the front-end kinetics are estimated with rea-
subsequent 10 h using the recirculatory pharmacoki- sonable accuracy (e.g., by a recirculatory model after
netic models. The three-compartment model fit to these bolus drug administration or by a three-compartment
data (model 4) had the smallest VCs of any of the three- model during and after drug administration by infusion),
compartment models derived in the current study and, they can be used to predict plasma drug concentrations
like the model based on bolus data in which VC was during and after any method of administration.
fixed to the sum of VC and VND of the recirculatory It was not the purpose of the current study to identify
model (model 3), described peripheral drug distribution an ideal blood sampling protocol for a study of drug
in a manner that was not different from that of the disposition during and after a brief drug infusion. None-
recirculatory model (table 1). Target-controlled drug in- theless, general recommendations may be made on the
fusions based on model 4 deviated minimally from the basis of the current results. The blood sampling protocol
target (fig. 4). Therefore, this study suggests that the should begin soon after commencing and soon after
optimal design of a study to obtain pharmacokinetic data stopping the drug infusion, but not so soon that samples
of Bioengineering Emeritus, University of Washington, Seattle, Washington), for lidocaine based on analysis of arterial and mixed venous data from dogs. J
helpful advice on data weighting. Phamacokin Biopharm 1997; 25:169 90
17. Avram MJ, Krejcie TC, Henthorn TK: The concordance of early antipyrine
and thiopental distribution kinetics. J Pharmacol Exp Ther 2002; 302:594 600
18. Garner D, Laks MM: New implanted chronic catheter device for determin-
References ing blood pressure and cardiac output in conscious dogs. Am J Physiol 1985;
249:H861 4
1. Krejcie TC, Avram MJ: What determines the anesthetic induction dose? Its 19. Grasela DM, Rocci ML Jr, Vlasses PH: Experimental impact of assay-
the front-end kinetics, Doctor! Anesth Analg 1999; 89:541 4
dependent differences in plasma indocyanine green concentration determina-
2. Atkinson AJ Jr, Ruo TI, Frederiksen MC: Physiological basis of multicom-
tions. J Pharmacokinet Biopharm 1987; 15:60113
partmental models of drug distribution. Trends Pharmacol Sci 1991; 12:96 101
20. Avram MJ, Krejcie TC: Determination of sodium pentobarbital and either
3. Niazi S: Volume of distribution as a function of time. J Pharm Sci 1976;
65:452 4 sodium methohexital or sodium thiopental in plasma by high performance liquid
4. Colburn WE: A time-dependent volume of distribution term used to de- chromatography with ultraviolet detection. J Chromatogr 1987; 414:484 91
scribe linear concentration-time profiles. J Pharmacokinet Biopharm 1983; 11: 21. Jacquez JA: Compartmental Analysis in Biology and Medicine, 3rd edition.
389 400 Ann Arbor, BioMedware, 1996
5. Avram MJ, Henthorn TK, Shanks CA, Krejcie TC: The initial rate of change 22. Foster DM: Developing and testing integrated multicompartment models
in distribution volume is the sum of intercompartmental clearances. J Pharm Sci to describe a single-input multiple-output study using the SAAM II software
1986; 75:919 20 system. Adv Exp Med Biol 1998; 445:59 78
6. Chiou WL: Potential effect of early blood sampling schedule on calculated 23. Krejcie TC, Henthorn TK, Shanks CA, Avram MJ: A recirculatory pharma-
pharmacokinetic parameters of drugs after intravenous administration. J Pharm cokinetic model describing the circulatory mixing, tissue distribution and elim-
Sci 1980; 69:8679 ination of antipyrine in dogs. J Pharmacol Exp Ther 1994; 269:609 16
7. Weiss M: Modelling of initial distribution of drugs following intravenous 24. Schwilden H, Schttler J, Stoekel H: Pharmacokinetics as applied to total
bolus injection. Eur J Clin Pharmacol 1983; 24:121 6 intravenous anesthesia: Theoretical considerations. Anaesthesia 1983; 38(suppl):
8. Chiou WL: Potential pitfalls in the conventional pharmacokinetic studies: 512
Effects of the initial mixing of drug in blood and the pulmonary first-pass 25. Wada DR, Ward DS: The hybrid model: A new pharmacokinetic model
elimination. J Pharmacokinet Biopharm 1979; 7:52736 for computer-controlled infusion pumps. IEE Trans Biomed Eng 1994; 41:
9. Wada DR, Drover DR, Lemmens HJM: Determination of the distribution 134 42
volume that can be used to calculate the intravenous loading dose. Clin Pharma- 26. Schttler J, Ihmsen H: Population pharmacokinetics of propofol: A multi-
cokinet 1998; 35:17 center study. ANESTHESIOLOGY 2000; 92:72738
10. Henthorn TK, Avram MJ, Krejcie TC, Shanks CA, Asada A, Kaczynski DA: 27. Van den Nieuwenhuyzen MC, Engbers FH, Vuyk J, Burm AG: Target-
Minimal compartmental model of circulatory mixing of indocyanine green. Am J controlled infusion systems: Role in anaesthesia and analgesia. Clin Pharmacoki-
Physiol 1992; 262:H90310 net 2000; 38:18190
11. Krejcie TC, Henthorn TK, Niemann CU, Klein C, Gupta DK, Gentry WB, 28. Raemer DB, Buschman A, Varvel JR, Philip BK, Johnson MD, Stein DA,
Shanks CA, Avram MJ: Recirculatory pharmacokinetic models of blood, extracel- Shafer SL: The prospective use of population pharmacokinetics in a computer-
lular fluid and total body water administered concomitantly. J Pharmacol Exp
driven infusion system for alfentanil. ANESTHESIOLOGY 1990; 73:66 72
Ther 1996; 278:1050 7
29. Barvais L, Canraine F, DHollander A, Coussaert E: Predictive accuracy of
12. Avram MJ, Krejcie TC, Niemann CU, Klein C, Gentry WB, Shanks CA,
continuous alfentanil infusion in volunteers: Variability of different pharmacoki-
Henthorn TK: The effect of halothane on the recirculatory pharmacokinetics of
physiologic markers. ANESTHESIOLOGY 1997; 87:138193 netic sets. Anesth Analg 1993; 77:80110
13. Chiou WL: The phenomenon and rationale of marked dependence of drug 30. Shafer SL, Varvel JR, Aziz N, Scott JC: Pharmacokinetics of fentanyl admin-
concentration on blood sampling site: Implications in pharmacokinetics, toxicol- istered by computer-controlled infusion pump. ANESTHESIOLOGY 1990; 10911102
ogy and therapeutics: I. Clin Pharmacokinet 1989; 17:17599 31. Schnider TW, Gaeta R, Brose W, Minto CF, Gregg KM, Shafer SL: Deriva-
14. Chiou WL: The phenomenon and rationale of marked dependence of drug tion and cross-validation of pharmacokinetic parameters for computer-controlled
concentration on blood sampling site: Implications in pharmacokinetics, toxicol- infusion of lidocaine in pain therapy. ANESTHESIOLOGY 1996; 84:104350
ogy and therapeutics: II. Clin Pharmacokinet 1989; 17:27590 32. Vuyk J, Engbers FHM, Burm AGL, Vletter AA, Bovill JG: Performance of
15. Krejcie TC, Jacquez JA, Avram MJ, Niemann CU, Shanks CA, Henthorn TK: computer-controlled infusion of propofol: An evaluation of five pharmacokinetic
Use of parallel Erlang density functions to analyze first-pass pulmonary uptake of parameter sets. Anesth Analg 1995; 81:1275 82
multiple indicators in dogs. J Phamacokin Biopharm 1996; 24:569 88 33. Schnider TW, Minto CF, Gambus PL, Andresen C, Goodale DB, Shafer SL,
16. Krejcie TC, Avram MJ, Gentry WB, Niemann CU, Jankowski MP, Henthorn Youngs EJ: The influence of method of administration on the pharmacokinetics
TK: A recirculatory model of the pulmonary uptake and pharmacokinetics of of propofol in adult volunteers. ANESTHESIOLOGY 1998; 88:1170 82