Pharmacodynamic Comparison

Pharmacodynamic Comparison

Duration of Opioid Antagonism by Nalmefene and Naloxone in the Dog: An Integrated PharmacokineticlPharmacodynamic Comparison PETER VENG-~EDERSEN*',JE...

732KB Sizes 0 Downloads 134 Views

Duration of Opioid Antagonism by Nalmefene and Naloxone in the Dog: An Integrated PharmacokineticlPharmacodynamic Comparison PETER


Received Januarv 20, 1995, from the *Co//eueof Pharmacy, The Universifv of l o w , Iowa City, /A 52242, and 'Ohmeda Pharmaceutical -Accepted for publication June'20, 1995@ Products Division, Inc,, 100 Mountain Avenuk, Murray Hill,. NJ 07974. aliquot prior to radioimmunoassay (93% recovery). For nalmefene and naloxone determinations, a separate 0.5-mL plasma aliquot was extracted using ethyl ether (85 and 87% recoveries, respectively). Following extraction, analytes were incubated with antibodies raised against fentanyl (fentanyl determination) or naltrexone (nalmefene or naloxone determination) and 3H-fentanyl (fentanyl determination) or 3H-naloxone (nalmefene or naloxone determination) tracer. The crossreactivity of the naltrexone antibody with fentanyl at levels of up to 350 ng/mL was >0.1%. Following incubation with the antibody, unbound analytes were removed by selective absorption onto dextrancoated charcoal. Linear assay ranges were 0.1-6.4,0.0625-2.0, and 0.078-5.0 ng/mL for fentanyl, nalmefene, and naloxone, respectively. The lowest concentration in the linear range was the assay limit of quantitation. Inter- and intraday coefficients of variation were <14 and <12% (fentanyl), <17 and <18% (nalmefene)and <13 and 12% (naloxone). Animal C a r e and Husbandry-The study was approved by the Ohmeda Animal Care and Use Committee. Eight male adult beagle dogs ranging in weights from 11.7 to 13.4 kg were purchased from Schering Plough (Lafayette, NJ) and housed in stainless steel cages with rubber-coated wire mesh floors over adsorbent paper. Canned food (Hill's Science Diet Adult Dog Food, Topeka, KS) and water were provided ad libitum. The photoperiod was 6:OO a.m. t o 6:OO p.m., room temperature was 72 f 4 O F , and the relative humidity was 50 f 5%. S t u d y Conduct-On the morning following an overnight fast, the animals were placed in a restraining sling and two angiocatheters (20 gauge x 11/4") were inserted into contralateral cephalic veins to facilitate drug administrations. An additional angiocatheter was inserted into a saphenous vein for blood sample acquisition. The Nalmefene [ 174cyclopropylmethyl)-4,5a-epoxy-6-methyl- inside of the non-tattooed ear was selected as the location for enemorphinan-3,14-diol] is a new opioid antagonist currently placement of the transcutaneous pC0z electrode (TINA TCM3 monitor, Radiometer, Copenhagen, Denmark). Prior to electrode applicaundergoing clinical development. It is structurally and tion, the dog's ear was closely shaved with electric animal clippers pharmacologically similar t o naloxone, which is the only pure and a size 40 blade. After cleansing with soap and water, the ear opioid antagonist clinically available for intravenous use. The was further shaved with a disposable twin-blade razor and towel major disadvantage of naloxone is that it has a relatively short dried. After the TINA plastic application disc was applied, two drops elimination half-life and duration of action compared with the of contact solution were placed within the disc. The previously opioid agonists that it is used to antag~nize.l-~ Because of calibrated sensor electrode was placed on the skin within the its short duration of action relative to clinically used opioid application disk, and then data acquisition commenced. After the agonists, the potential exists for renarc~tization.l-~Several appropriate TINA equilibration time (-20 min), the fentanyl infusion preclinical s t ~ d i e shave ~ , ~reported a longer pharmacodynamic was initiated and administered continuously throughout the study a t a rate of 30 p g k g h with a Harvard Apparatus model 22 infusion duration of action for nalmefene than naloxone; however, it pump (South Natick, MA). Maintenance of angiocatheter patency was is unclear if these studies compared equieffcacious doses of periodically confirmed throughout the study. Approximately 120 min the two antagonists. The purpose of this study is to perform after the start of the fentanyl infusion, either nalmefene (12 pgkglh) an integrated pharmacokinetidpharmacodynamiccomparison or naloxone (48 pgkglh) was administered as constant-rate infusions of the two antagonists and present a kinetic model explaining for 30 min. All doses are expressed as the free base (molecular the longer duration of action of nalmefene. weights for fentanyl, nalmefene, and naloxone free base are 336,339, and 327, respectively). On a third treatment, so as not to interfere with the 2 x 2 crossover design, fentanyl was administered as a 30Experimental Section p g k g h infusion to obtain plasma concentration information in the absence of antagonist administration. Blood samples (-5 mL) were Chemicals-Ampoules of nalmefene (1mg free base per 1mL) or drawn from the saphenous catheter into heparinized tubes just prior naloxone (1 mg free base per 1 mL) were compounded and filled at to and a t 5, 10, 15, 25, 90, 120, 125, 130, 135, 140, 150, 180, 210, 240, Taylor Pharmacal Inc. (Decatur, IL) and further diluted with saline 270, 300, 330, 360, 390, and 420 min during the fentanyl infusion. to yield a sufficient infusion volume. Ampoules of fentanyl (50 ,ug Immediately following blood collection, the samples were stored on free base per 1 mL) were obtained from Janssen Pharmaceutica ice prior to centrifugation for 10 min a t 1200 x g. The resulting (Piscataway, NJ) and used without dilution. plasma samples were stored prior to assay in plastic 5-mL cryogenic Assay Methodology-All samples were analyzed for fentanyl, vials at -60 "C. nalmefene, or naloxone, by sensitive and specific radioimmunoassay D a t a Collection and Reduction-Transcutaneous pC0z was (Harris Laboratories, Inc., Lincoln, NE). For fentanyl determinations, measured with a daily calibrated TINA monitor. The analogue output a n ethyl acetate extraction step was performed on a 0.5-mL plasma from the TINA monitor was sent to a Gould TA-4000 recorder and then routed to a Po-Ne-Mah data acquisition system, where it was digitized and stored. The experimental collection rate was 10 pCO2 Abstract published in Advance ACS Abstracts, August 1, 1995.

Abstract 0 A continuous fentanyl infusion was administered to eight adult, male beagle dogs for a duration of -400 rnin at a rate of 30 pg/ kg/h. The extent of respiratory depression was quantified by continuous, noninvasive, transcutaneous pCO2 recordings. Upon reaching a pseudosteady-state of respiratory depression at -120 min of fentanyl infusion, the animals then received, in a 2x2 crossover fashion separated by -3 weeks, 30-minute equieff icacious infusions of nalmefene (12 ygikgih) or naloxone (48 yglkglh). Multiple venous blood samples were taken throughout the dosing regimen, and the resulting fentanyl, nalmefene, or naloxone plasma concentrations were determined. The concentrationtime data were analyzed by noncompartmental methods and subsequently linked to the pharmacodynamic effect data by a competitive antagonism link model. Separately, the biophase concentrations were linked to the plasma concentration-time profiles through a single-exponential conduction function. The various pharmacokinetic/pharmacodynamic parameters resulting from this semiparametric analysis were analyzed by ANOVA, using a statistical model that considers carryover effects. The results of these analyses indicate that several pharmacokinetic/pharmacodynamic parameters of the two antagonists were comparable. However, nalmefene had a significantly more protracted terminal disposition and a significantly greater persistency in the biophase evaluated over the experimental time frame from 0 to 450 min.

0 1995, American Chemical Society and American Pharmaceutical Association


Journal of Pharmaceutical Sciences / 1I01 Vol. 84, No. 9, September 1995

recordings per minute. The data, up to 4200 raw data points for each treatment corresponding to 420 min of pC02 recordings, were edited for outliers caused by infrequent and temporary sensor dislocations. To facilitate the plotting and the data analysis, the data were reduced to -1000 data points by a four point moving average. StatisticalAnalysis-Eight dogs were randomly assigned to a 2x2 crossover design (two sequences, two periods) with four dogs assigned to treatment sequence 1 (fentanyl naloxone followed by fentanyl nalmefene) and four to treatment sequence 2 (fentanyl nalmefene followed by fentanyl naloxone). The various pharmacokinetic and pharmacodynamic parameters were analyzed according to the following statistical model and ANOVA





where Yijk is the parameter being compared, and S,P , D, and C denote subject, period, drug, and carryover (sequence) effects, respectively, and i, j, k are indices referring to subject no. (i = 1, 2, ..., nk), period f j = 1,2), and sequence (K = 1,2). The Fvalue of the carryover effect was calculated as the mean square ratio MSca&Sinter between carryover and intersubject variability with 1 and nl + n2 - 2 degrees of freedom, respectively. Statistical significance was inferred if p < 0.05 in the ANOVA. A program was developed in FORTRAN that directly implements the described statistical model and provides the probabilities ( p ) associated with the ANOVA F values. The program was validated against the SAS statistical package running in a mainframe environment (SAS Analysis System v6.0, SAS Institute, Cary, NC), using published data table^.^ PharmacokineticAnalysis-A noncompartmental system analysis approach was used to analyze the concentration-time data. The fentanyl plasma concentration ( c f t h ) , resulting from a constant rate infusion (R),is described accordingly by the following convolution expression: = R*[A, exp(-a,t) + A , exp(-a,t)] =

where A and a are the parameters of the biexponential disposition function. The plasma concentrations of the antagonists resulting from the 30-min infusion were similarly described with a biexponential disposition function using the following expression:


c(t) = R -{exp(-a,(t’

where: -

T for t’



= 0 otherwise

The variable t’ is the time since the start of the antagonist infusion and T (T = 30 min) is the length of the antagonist infusion. The A and a, in a common notation, is used to represent the disposition parameters (unit impulse response parameters) for the two antagonists, like the notation used for the agonist in eq 1. The selection of the biexponential disposition function was made based on Akaike’s Information Criterion.* The following pharmacokinetic parameters were calculated from terms derived from the individual disposition functions: Clearance: C1= l/(Az/al +A,/%) Initial Volume of Distribution: V = l/(Al


+ A,)

“Distribution” Half-life: t,,, (1)= (ln(2))/al

1102 / Journal of Pharmaceutical Sciences Vol. 84, No. 9, September 1995


The contributions of the distribution or elimination components of the disposition function to the total area under the curve of concentration versus time (AUC), and inversely, C1, can be fractionally expressed in the following two equations: Distribution Fractional AUC: F, = (A,/a,)/AUC


F, = (Ada,)/AUC


Elimination Fractional AUC:

The tx parameters (x = 5, 25, 50) termed the x% reduction times are defined as the times it takes an initial drug concentration following a bolus injection to decrease to x% of the initial value. These parameters, which are dose independent for drugs exhibiting linear disposition, are obtained by numerically solving eq 9 for tz: x% Reduction Time: A, exp(-a,t,)

+ A 2 exp(-a,t,)=


+ A,)


The mean residence time (MRT) is defined as the mean time a drug molecule spends in the body9 and is calculated using parameters derived from the disposition function as follows:

Mean Residence Time (0-infinity):


AUMC iAUC A,/a, + Ada,

Because the calculation of MRT (0-infinity) and terminal half-life parameters requires a n extrapolation beyond the sampling times, a poorly determined extrapolation will have a marked effect on both parameters. To circumvent extrapolations to time infinity, a truncated MRT parameter is proposed IMRT (tend)]that does not involve extrapolation beyond the last observation time (tend). The mean truncated MRT parameter is accordingly defined by the following expression:



- T),) - exp(-a,t’)}

(t’ - T), = t’

“Elimination” Half-life: t,,, (2) = (ln(2))/a,

(4) (5)

where c ( t ) denotes the disposition function (impulse response function). The last observation was always an effect observation (pC02), which in all cases was not shorter than 450 min. Thus, to standardize the comparison a value of 450 min for tend was used throughout the study. Such a normalization is desirable because MRT(tend)depends on tend. In this study, the net effect being measured (pCO2) depends on the relative concentrations of agonist and antagonist in the biophase and the transduction of these biophase concentrations into a n overall effect. The biexponential disposition parameters (Al, a,, A,, and az), obtained by least-squares fitting to the concentrationtime data, were used to generate the disposition function. The biophase concentrations are related to plasma concentrations through a linear, convolution relationship:

where Cb&) is the biophase concentration of fentanyl (c,) or nalmefend naloxone (Cb), c,(t) is the plasma concentration of fentanyl, nalmefene, or naloxone, and y is the conduction function parameter (commonly referred to as L). The transduction function converts the biophase concentrations of agonist or antagonist into the effect response variable ( p C 0 ~being ) measured, using equations developed for competitive antagonism and modified to include a shape parameter term, N.l0J1The effect of an agonist in the presence of a n antagonist, Eab, (where subscripts a and b designate agonist and antagonist, respectively) is given by the following expression:

of the function representing the integrated pharmacokinetidpharmacodynamic model is optimized in a n iterative fashion by changing the conduction and transduction parameters (7, E m , ECSOa, E C ~ O ~ , Eo, and N) while holding the previously determined disposition function parameters (AI, al,A2 and a 2 1 constant. where E,b is the effect of fentanyl (a)in the presence of nalmefene or naloxone (b),Eo is the baseline effect, EMMis the maximal effect, ca is the fentanyl biophase concentration, Cb is the nalmefene or naloxone biophase concentration, ECSOa is the fentanyl concentration that produces half the Em,, effect in the absence of the antagonist, Ec50b is the nalmefene or naloxone concentration that produces a 50% reduction in the fentanyl effect when the concentration of fentanyl is equal to EC50, and N is the sigmoidal shape parameter. The effect of fentanyl (a) in the absence of antagonist (b) is given by the expression:

Several descriptive parameters can be calculated once the biophase concentration-time profile has been established. The mean transit time of drug molecules in the biophase, BIOMTT, is calculated as:

BIOMTT = l i y


The BIOMTT is the mean time it takes biophase available drug molecules to exit the biophase after entering the biphase. The following expression can be used to avoid extrapolation influence on the calculation of BIOM’IT by considering elimination only in the first 450 min: J450ty exp(-y t)dt


BIOMTT(450) = L 4 5 0 y exp(-y t)dt

The shape of the biophase concentration-time profile can be assessed in a general, nonparametric way by introducing a biophase, centerof-gravity time term (BIOCGT).12 The BIOCGT parameter indicates how prolonged the biophase concentration persists and thus can be used as a measure of the duration of action. This parameter is readily calculated from the conduction and disposition functions as the sum of the mean residence times of the drug molecules (MRT) and the mean transit time at the biophase (BIOMTT):




where MRT has previously been described as follows:


To avoid extrapolation errors the biophase concentration-time profile is evaluated only during the time period of observations (450 min), according to the following expression:

The conduction and transduction functions are mathematically connected to the fxed concentration-time profiles from the pharmacokinetic analysis to form the integrated pharmacokinetic/pharmacodynamic model. The resulting function that is fitted to the effect data from the combined agonist/antagonist administration uses the agonist concentration and antagonist concentration as input variables to generate a n effect response output. Because the two input variables are functions of time, the resulting integrated pharmacokinetidpharmacodynamic model is also a function of time and can be used to predict the effect versus time response. During the simultaneous fitting of the effect data for each dog, the size and shape

Pharmacokinetics-Individual curvefits to the concentration-time data for a representative dog using the appropriate equations just described are shown in Figure 1. The figure is composed of five separate concentration-time profiles and is arranged as follows: ( a )nalmefene concentration-time data during and following the 30-min 12-,ug/kg/h nalmefene infusion (the fitted curve shown is eq 2); ( b ) naloxone concentration-time data during and following the 30-min 48-pg/kg/h naloxone infusion (the fitted curve shown is eq 2); ( c ) fentanyl concentration-time data during a continuous 3O-pgkgh fentanyl infusion in the nalmefene-antagonizedtreatment (the fitted curve shown is eq 1); ( d ) fentanyl concentration-time data during a continuous 30-,ug/kg/h fentanyl infusion in the naloxone-antagonized treatment (the fitted curve shown is eq 1); and ( e )fentanyl concentration-time data from a continuous 30-pg/kg/h infusion of fentanyl administered in the absence of antagonist (the fitted curve shown is eq 1). Inspection of the individual curvefits to the concentrationtime data for nalmefene and naloxone confirms that selection of a biexponential disposition function was appropriate for these data. A summary of the pharmacokinetic parameters derived from the 30-min infusions of nalmefene and naloxone that also includes the parameters for fentanyl is shown in Table 1. Comparable mean values for clearance, volume of distribution, distribution and elimination half-lives, fractional AUCs, t 5 0 , t 2 5 , and MRT (0-infinity) were seen for nalmefene and naloxone (Table 2). For both antagonists, the fractional AUC associated with the “elimination” phase is greater than that associated with the “distribution” phase (F2> Fl). The only discernible difference between the two antagonists appears to lie in the terminal elimination phase, as indicated by the significant differences of the t 5 reduction time parameter and MRT(450) (p = 0.003 and 0.002, respectively; Table 2). Pharmacodynamics-A set of individual curve fits to the effect data for a representative dog using the appropriate integrated pharmacokinetidpharmacodyndcs model is shown in Figure 2. The figure is composed of three effect-time profiles and is arranged as follows: ( a )effect-time data prior to, during, and following the 12-pgkg/h nalmefene infusion (the fitted curve shown is eq 13, where ca is defined by eqs 12 and 1, and c b is defined by eqs 12 and 2); ( b )effect-time data prior to, during, the following the 48-pg/kgh naloxone infusion (the fitted curve shown is eq 13, where ca is defined by eqs 12 and 1, and c b is defined by eqs 12 and 2);( c ) effect-time data from a 3O-pgkgRi fentanyl infusion (the fitted curve shown is eq 14, with ca defined by eqs 12 and 1). Inspection of the individual curvefits produced by the integrated pharmacokinetic/pharmacodynamic model indicates a good correlation between the raw data and model-predicted values. A summary of the parameters obtained by the integrated pharmacokinetidpharacodynamic estimation procedure exemplified in Figure 2 is displayed in Table 3. The ANOVA summary comparison of the derived parameters is shown in Table 4. A difference in the persistence of antagonist biophase concentration, BIOCGT(450),for nalmefene and naloxone was determined, and nalmefene possesses a significantly longer duration ( p = 0.0072). Journal of Pharmaceutical Sciences / 1103 Vol. 84, No. 9, September 1995



+ naloxone




160 Id0 ZdO 2dO 3dO 350 460 460



fentanyl only

(e) l4 1


r J




5 U z 0 U

c 5 LL




Id0 Id0 260 2$0 3dO 3$0 460 450





I60 180 260 ZBO 360 3$0 4 6 1 3 0



Figure 1-Pharmacokinetic curve fits for dog W44. Table 1-Summary of Nalmefene, Naloxone, and Fentanyl Pharmacokinetic Parameters Obtained from a 30-Min Infusion During Concomitant Fentanyl Infusion Nalrnefene (12 Icglkglh)

CI, mUrninlkg V, Ukg tin (11, h fiiz (21, h Fi F2

ko, h fzs, h

k, h MRT, h MRT(450),h

Naloxone (48 Pg/kg/h)

Fentanyl (30 I c g h l w







67.0 4.54 0.342 3.56 0.311 0.689 0.442 1.04 5.37 3.48 1.74

23.5 1.9 0.167 2.19 0.111 0.111 0.172 0.355 1.70 1.82 0.365

72.8 3.29 0.248 3.42 0.414 0.586 0.288 0.627 2.71 3.16 1.38

11.0 1.09 0.133 3.31 0.194 0.194 0.132 0.257 0.790 3.48 0.435

83.1 1.50 12.0 0.207 0.408 0.592 0.250 0.511 2.61 5.05 0.904

41.2 1.01 20.6 0.241 0.453 0.453 0.250 0.502 5.04 16.4 0.862

Discussion Pharmacokinetic Comparison-Whereas the concentration-time data for both antagonists appear well described by a biexponential disposition function, the fentanyl concentration-time data were not equally well described by the biexponential disposition function (Figure 1). The disposition of fentanyl may have been influenced by inherent assay variability, as evidenced in the concentration-time curves following infusions of fentanyl only (Figure le). Alternatively, the disposition may be influenced by the fluctuation of fentanyl plasma concentrations in response to antagonist administration (Figures l c and Id; 120-200 min). The infusion of each antagonist appeared to result in a temporary decrease of plasma fentanyl concentrations. The antagonistinduced effect on plasma fentanyl concentration diminished and fentanyl plasma concentration increased following termination of the nalmefene or naloxone infusion. This effect may be due to the significant hepatic clearance of fentanyl13 that likely is influenced by changes in heptatic blood flow. Opioids in general, and fentanyl specifically, reduce hepatic 1104 / Journal of Pharmaceutical Sciences Vol. 84, No. 9, September 1995

blood flow.14-17 The observed decrease in fentanyl plasma concentrations (i.e., acceleration of fentanyl elimination) resulting from nalmefene and naloxone likely results from their antagonism of the fentanyl-induced reduction in hepatic blood flow. The clearance of nalmefene determined in this study following a 30-min infusion (67 ml/min/kg) agrees well with that previously established for this species following bolus administration (62 ml/min/kg).ls Likewise, the clearance of naloxone determined in this study (73 mumidkg) is also in agreement with that previously reported (70 ml/rnir~/kg).~~ Plasma concentrations at the end of antagonist administration (150 min) are (mean SD) 0.90 0.21 and 4.07 f 1.43 ng/mL for nalmefene and naloxone, respectively, demonstrating a mean plasma concentration ratio of 1:4.5 (nalmefene: naloxone). The only discernible pharmacokinetic difference between the two antagonists is observed in the terminal elimination phase, as indicated by the significant differences of the t 5 reduction time parameter and MRT(450) ( p = 0.003 and 0.002, respectively). The time required for nalmefene plasma concentrations to decrease by 50, 75, or 95% of their maximal value ( t 5 0 , t 2 5 , and t 5 respectively) was greater than that for naloxone (1.5, 1.7-, and 2.0-fold, respectively), with statistical significance achieved a t a point between t 2 5 and t 5 ( p = 0.070 and 0.003, respectively; Table 2). The reduction time parameters avoid extrapolation errors when comparing elimination characteristics, and are particularly useful in situations involving only a single bolus administration. This is relevant in the cases of naloxone and nalmefene, as their typical clinical application involves single-dose bolus administration. The statistical difference in the derived MRT(450) parameter further confirms the elimination-phase differentiation between nalmefene and naloxone ( p = 0.002). The MRT(450) is defined as the mean time a drug molecule resides in the body before being eliminated in the time period from t = 0 to t = 450. The MRT has also been reported in the literature as the center-of-gravity time.20,21The center-of-gravity time is larger the more slowly the disposition function decreases, and thus has similar attributes to half-life. Calculation of MRT-



Table 2-Summary of Nalmefene and Naloxone Pharmacokinetic Parameters (ANOVA)

p Values Variation



Carryover Inter-residual Direct drug Period

0.454 0.257 0.495 0.644

0.379 0.443 0.063 0.991



0.219 0.784 0.325 0.545



0.507 0.180 0.913 0.455








0.014 0.649 0.144 0.086

0.014 0.649 0.144 0.086

0.302 0.813 0.143 0.509

0.489 0.818 0.070 0.457

0.561 0.205 0.003 0.832

0.331 0.108 0.751 0.222

0.514 0.002 0.002 0.010





N 0

u a. I-












70 -


Id0 I d 0 2d0 2dO 3d0 3dO 4d0 420






E 60














160 260 260 3dO 3d0 4dO 4dO












55 -



u L



45 40



Table 3-Parameters Derived from Integrated PharmacokineW Pharmacodynamic Analysis


160 1$0 200 250 300 350 400 450




Figure 2-Pharmacodynamic curve fits for dog W44

(450) avoids errors from extrapolation to time infinity when evaluating the reduction in plasma concentrations while









26.3 0.559 2.81 0.0200 1.07 0.799 4.15 2.83

4.53 0.642 2.01 0.0304 1.63 0.881 2.03 0.630

29.2 2.00 4.31 0.0967 0.450 0.450 3.61 1.83

6.75 2.23 1.86 0.107 0.297 0.296 3.43 0.451

31.5 0.473 1.33 NA 39.6 3.49 45.4 3.98

17.3 0.535 3.i.10

NA 46.6 0.234 46.2 0.219

considering the general shape of the disposition curve in the most relevant time period ( t = 0 to t = 450). The pharmacokinetic analysis did not demonstrate major differences in the plasma profiles of nalmefene and naloxone. However, there was a significantly slower terminal decline in nalmefene plasma concentrations, which was apparent when plasma concentrations declined to <25%of their maximal value. Because fractional AUC parameters revealed that the “elimination”phase was a major contributor to the overall concentration-time profile, the slower terminal decline in plasma concentration appears significant and may explain the observed pharmacodynamic differences between nalmefene and naloxone. Pharmacokineti&harmacodynamic Integration-The integration of pharmacokinetic and pharmacodynamic data allows a comparative assessment of pharmacodynamic parameters generated from fitting the respective equations to pharmacodynamic effect data, using fixed plasma concentrations determined in the pharmacokinetic analysis section. As exemplified by Figures 1 and 2, the pharmacokinetic/pharmacodynamic model fits were generally in good agreement with the raw data, thus demonstrating the appropriateness of the pharmacokinetic/pharmacodynamicmodel employed to generate the resulting parameters. Throughout the study, the imposed fentanyl effects were consistent, as demonstrated by a lack of significant differences of EMAXor ECSOa values in nalmefene or naloxone treatments (Table 4). Comparison of the parameters derived from the biophase profiles indicates no significant differences in N , ECSO~, BIOMTT, BIOMTT(450), or BIOCGT; however, nalmefene displayed nearly two-fold greater values in the duration parameters than did naloxone. There was a significant difference @ = 0.0075) between the two antagonists in the biophase center-of-gravitytime, BIOCGT(450),evaluated over the experimental, nonextrapolated time period (450 min; Table 4). The BIOCGT(450) parameter reflects the shape of the concentration-time profile in the biophase and considers both the intensity and persistence of biophase concentrations without extrapolation beyond the study time period. This persistency of the biophase concentration is ultimately determined by the combination of the plasma kinetics of the drug and how the plasma concentration profile is mapped into the biophase profile, and thus is a useful indicator of pharmacoJournal of Pharmaceutical Sciences / 1105 Voi. 84, No. 9, September 1995

Table 4-Summary of Nalmefene and Naloxone Pharmacokinetic-Pharmacodynamic Parameters (ANOVA) p Values Variation Carryover Inter-residual Direct drug Period








BIOCGT (450)

0.519 0.395 0.278 0.0718

0.929 0.686 0.188 0.761

0.531 0.270 0.0896 0.038

0.851 0.476 0.104 0.353

0.263 0.229 0.225 0.175

0.219 0.0843 0.135 0.076

0.450 0.108 0.600 0.154

0.598 0.0293 0.0075 0.0163

logical duration of action. The data obtained in this analysis agrees with the 1:4 (nalmefenenaloxone) potency ratio determined in preliminary investigations to establish equiefficacious dosings. The effect-site concentrations required to reveal a produce 50% reversal of the effect of fentanyl (ECSO~) ratio of 1:4.8. Furthermore, there is a close agreement between the maximal antagonist effect (-150 min; Figure 2) at the chosen 1:4 dosing. The finding of a longer biophase concentration persistency obtained in the integrated pharmacokinetidpharmacodynamic analysis is consistent with the findings in the pharmacokinetic analysis, where it was determined that nalmefene differed from naloxone mainly by a slower terminal decline in its plasma concentration. This slower terminal decline appears to be mapped into a slower decline in the biophase, resulting in an increased biophase center-of-gravity time. This study appears to be the first report of a comprehensive, integrated, pharmacokinetic/pharmacodynamic evaluation of two opioid antagonists. The difficulty in assessing the duration of action of an antagonist (because it has no intrinsic activity) was overcome by developing the fentanyl-induced, respiratory depression model and using a crossover study design. The duration of action of the antagonist depends on the following three variables: (1)the agonist plasma concentration a t the time of reversal, (2) the duration since the agonist was administered, and (3) the dose of the antagonist. Each of these variables has been controlled with the methodology employed. The constant-rate fentanyl infusion controls variables 1and 2, and the 1:4 equiefficaciousantagonist doses control variable 3. The pharmacodynamic effect of fentanyl, respiratory depression, was continuously monitored with transcutaneous pCOz measurements and provided a continual, immediately responsive, noninvasive measure of a clinically relevant opioid effect that was reversed by antagonist administration. Although reversal of a continual fentanyl infusion is not encountered clinically, the constant infusion maintained the fentanyl plasma concentration over the course of the experiment and allowed a more direct and simple comparison of the durations of action of nalmefene and naloxone. In summary, frequent pharmacodynamic measurements, coupled with frequent venous blood samplings throughout the dosing regimen, have provided the necessary data to describe the dose, plasma concentration, and effect relationship, which has enabled an in-depth kinetic dissection of two compounds whose primary pharmacokinetic parameters (clearance, vol-

1106 / Journal of pharmaceutical Sciences Vol. 84, No. 9, September 1995

ume of distribution, and half-life) were quite similar. A difference in the persistency of biophase concentrations has been detected that provides a mechanistic explanation for a longer duration of action of nalmefene. The two-fold magnitude difference in duration was also consistent with the pharmacokinetic analysis that shows a more protracted terminal disposition phase for nalmefene, as quantified by the tZ5and t 5 reduction time parameters.

References and Notes 1. Hug, C. C., Jr. In Pharmacokinetics ofAnesthesia; Prys-Roberts,

C . ;Hug, C. C., Jr., Eds.; BlackwelI Scientific. Boston, MA, 1984. 2. Bailey, P. L.; Clark, N. J.; Pace, N. L.; Stanley, T. H.; East, K. A,; van Vreeswijk, H.; van de Pol, P.; Clissold, M. A,; Rozendaal, W. Anesth. Analg. 1987, 66, 1109-1114. 3. Fragen, R. J. J . Clin. Anesth. 1992, 4(Supp. 11, 9s-15s. 4. Glass, P. S.; Jhaveri, R. M.; Smith, L. R. Anesth. Analg. 1994, 78, 536-541. 5. Dvson. D. H.: Dohertv. T.: Anderson. G. I.: McDonell. W. N. Vet. Surg. 1990,19, 398:403: 6. van Vugt, D. A.; Webb, M. Y.; Reid, R. L. Neuroendocrinology 1989,49, 275-280. 7. Chow, S . C.; Liu, J . P. Design and Analysis of Bioauailabdity and Bioequiualence Studaes; Marcel Dekker. New York, 1992; pp 64-69. 8. Yamaoka, K.; Nakagawa, T.; Uno, T. J . Pharmacokinet. Biopharm. 1978,6, 165-175. 9. Gibaldi, M.; Perrier, D. In Drugs and the Pharmaceutical Sciences; 2nd edition, Pharmacohinetics; Marcel Dekker. New York, 1982. 10. Gaddum, J . H. Pharmacol. Rev. 1957,9, 211-218. 11. Ariens, E. J.; Simonis, A. M.; van Rossum, J . M. In Molecular Pharmacology, Ariens, E. J., Ed.; Academic New York, 1964, Vol. 1, pp 119-286. 12. Veng-Pedersen, P.; Tillman, L. G. J . Pharm. Sci. 1989, 78,848854. 13. Murphy, M. R.; Olson, W. A,; Hug, C. C., Jr. Anesthesiology 1979, 50, 13-19. 14. K e n , N. D.; Reitan, J. A,; White, D. A.; Wu, C. H.; Eisele, J. H., Jr. Anesth. Analg. 1986, 65, 765-770. 15. Garty, M.; Ben-Zvi, Z.; Hurwitz, A. J . Pharmacol. Exp. Ther. 1985,234, 391-394. 16. Reilly, C . S.; Merrell, J.; Wood, A. J.; Koshakji, R. P.; Wood, M. Br. J . Anaesth. 1988, 60, 791-796. 17. Gelman, S. Can. J . Physiol. Pharmacol. 1987, 65, 1762-1779. 18. Kvalo. L. 'I?.: Wilhelm. J . A. Ohmeda Pharmaceutical Products Division, Inc., Drug Disposition and Product Safety Report # DD-93-053-002, 1994. 19. Pace, N. I..; Parrish, R. G.; Lieberman, M. M.; Wong, K. C.; Blatnick, R. A. J . Pharmacol. Exp. Ther. 1979,208, 254-256. 20. Veng-Pedersen, P. Clin. Pharmacokinet. 1989, 16, 686-694. 21. Veng-Pedersen, P. Clin. Pharmacokinet. 1989, 17, 424-440. I