`DRUG METABOLISM AND DISPOSITION
`Copyright © 2006 by The American Society for Pharmacology and Experimental Therapeutics
`DMD 34:1480–1487, 2006
`
`Vol. 34, No. 9
`9001/3131990
`Printed in U.S.A.
`
`Physiologically Based Pharmacokinetic Modeling of FTY720
`(2-Amino-2[2-(-4-octylphenyl)ethyl]propane-1,3-diol hydrochloride)
`in Rats After Oral and Intravenous Doses
`
`Guy M. L. Meno-Tetang,1 Hongshan Li, Suzette Mis, Nancy Pyszczynski, Peter Heining,
`Philip Lowe, and William J. Jusko
`
`Novartis Pharma AG, Basel, Switzerland (G.M.L.M.-T., P.H., P.L.); and Department of Pharmaceutical Sciences, School of
`Pharmacy and Pharmaceutical Sciences, State University of New York, Buffalo, New York (H.L., S.M., N.P., W.J.J.)
`
`Received December 16, 2005; accepted May 31, 2006
`
`ABSTRACT:
`
`FTY720 (2-amino-2[2-(-4-octylphenyl)ethyl]propane-1,3-diol hy-
`drochloride) is a new sphingosine-1-phosphate receptor agonist
`being developed for multiple sclerosis and prevention of solid
`organ transplant rejection. A physiologically based pharmacoki-
`netic model was developed to predict the concentration of FTY720
`in various organs of the body. Single oral and intravenous doses of
`FTY720 were administered to male Wistar rats, with blood and
`tissue sampling over 360 h analyzed by liquid chromatography/
`tandem mass spectrometry. A well stirred model (perfusion rate-
`limited) described FTY720 kinetics in heart, lungs, spleen, muscle,
`kidneys, bone, and liver, with a permeability rate-limited model
`being required for brain, thymus, and lymph nodes. Tissue-to-
`blood partition coefficients (RT) ranged from 4.69 (muscle) to 41.4
`(lungs). In lymph nodes and spleen, major sites for FTY720-induced
`changes in sequestration of lymphocytes, RT values were 22.9 and
`
`34.7, respectively. Permeability-surface area products for brain,
`thymus, and lymph nodes were 39.3, 122, and 176 ml/min. Intrinsic
`hepatic clearance was 23,145 l/h/kg for the free drug in blood (fub
`0.000333); systemic clearance was 0.748 l/h/kg and terminal half-
`life was 23.4 h. The fraction orally absorbed was 71%. The model
`characterized well FTY720 disposition for this extensive dosing
`and tissue collection study in the rat. On scaling the model to dogs
`and humans, good agreement was found between the actual and
`predicted blood concentration-time profiles. More importantly,
`brain concentrations in dogs were well predicted from those of the
`rat. In absolute terms, the predictions were slightly lower than
`observed values, just under a 1.5-fold deviation, but the model
`accurately predicted the terminal elimination of FTY720 from the
`brain.
`
`FTY720 (2-amino-2[2-(-4-octylphenyl)ethyl]propane-1,3-diol hy-
`drochloride) is a new sphingosine-1-phosphate receptor agonist that is
`being developed for prevention of solid organ transplant rejection
`(Napoli, 2000). FTY720 exerts its immunomodulatory actions by
`affecting lymphocyte production (Yagi et al., 2000),
`trafficking
`(Chiba et al., 1998; Brinkmann et al., 2000, 2001), infiltration (Yana-
`gawa et al., 2000), and apoptosis (Enosawa et al., 1996; Bohler et al.,
`2000; Nagahara et al., 2000). The maximum effects of FTY720 on
`these immune responses are achieved at doses smaller than those
`producing effects against graft rejection (Yanagawa et al., 1998).
`
`This work was supported by Novartis Pharma AG and in part by Grant GM
`24211 from the National Institutes of Health. The liquid chromatograph/tandem
`mass spectrometer was obtained through a shared instrumentation grant
`(S10RR14573) from the National Center for Research Resources, National Insti-
`tutes of Health.
`1 Current affiliation: Serono International S.A, Geneva, Switzerland.
`Article, publication date, and citation information can be found at
`http://dmd.aspetjournals.org.
`doi:10.1124/dmd.105.009001.
`
`Regulation of gene expression may also account for pharmacological
`and toxicological effects of FTY720: for instance, a 26-week phar-
`macology study in rats using gene chips showed that, at doses of 0.3
`and 1.5 mg/kg/day, genes of B and T lymphocyte markers in blood
`(CD79 and CD3) were down-regulated (Novartis Pharma AG, internal
`communication).
`The elimination of FTY720 from the body occurs mainly via
`metabolism. Two primary pathways metabolize FTY720: 1) phos-
`phorylation at one of its two hydroxy groups (yielding FTY720-P)
`and 2) hydroxylation at the terminal methyl group (M12) (Novartis
`Pharma AG, internal communication). The blood clearance values
`for FTY720 in dogs, monkeys, and humans are 0.0617, 0.113,
`and 0.0433 ml/h/kg, respectively. The elucidation of the mecha-
`nism of action and pharmacodynamics of FTY720 on immune cells
`will necessitate the characterization of its disposition not only in
`blood but also in target organs such as lymph nodes, spleen, and
`thymus.
`The objective of this work was to develop a physiologically based
`pharmacokinetic (PBPK) model in rats to characterize the kinetics of
`
`ABBREVIATIONS: FTY720, 2-amino-2[2-(-4-octylphenyl)ethyl]propane-1,3-diol hydrochloride; PBPK, physiologically based pharmacokinetic;
`Y32919, 2-amino-2-[2-(4-octyloxyphenyl)ethyl]propane-1,3-diol hydrochloride; HPLC, high-performance liquid chromatography; LN,
`lymph
`nodes; AUC, area under the concentration-time curve; RT, tissue-to-blood partition coefficient; PST, permeability-surface area product; CLint,
`intrinsic clearance.
`
`1480
`
`Apotex v. Novartis
`IPR2017-00854
`NOVARTIS 2055
`
`
`
`FTY720 PBPK MODEL IN RAT, DOG AND HUMAN
`
`1481
`
`FTY720 both in major organs and in lymphatic tissues such as spleen,
`thymus, and lymph nodes.
`
`Materials and Methods
`
`Chemicals. Analytical grade FTY720 and Y32919 (internal standard;
`2-amino-2-[2-(4-octyloxyphenyl)ethyl]propane-1,3-diol hydrochloride) were
`supplied by Novartis Pharma AG (Basel, Switzerland). HPLC-grade dichlo-
`romethane and water were purchased from Burdick and Jackson (Muskegon,
`MI). Methanol optima grade and glacial acetic acid HPLC grade were pur-
`chased from Fisher Scientific (Fairlawn, NJ). tert-Butyl-methyl ether HPLC
`grade was purchased from Sigma Aldrich (Milwaukee, WI) and ammonium
`acetate microselect was purchased from Fluka (Milwaukee, WI).
`Animals. Normal male Wistar rats, 80 to 90 days old, with body weights
`ranging from 300 to 375 g, were used in this study (Harlan Sprague-Dawley
`Inc., Indianapolis, IN). The animals were housed in pairs in stainless steel
`cages in a controlled environment with a 12-h light/dark cycle. Filtered tap
`water and food were available ad libitum. Before the dosing day, animals were
`kept in the animal facility for at least 7 days for acclimatization.
`Dosing and Sampling. Single i.v. bolus doses of FTY720 (0.3, 2, and 4
`mg/kg) were given to rats via penal vein injection. For each group, three rats
`were sacrificed by aortic exsanguination under ketamine/xylazine anesthesia
`just before dosing and 0.5, 1, 3, 6, 12, 24, 48, 72, 120, 176, 240, and 360 h
`postdose. Arterial blood was collected, and lungs, heart, brain, kidneys, thy-
`mus, spleen, liver, muscle, fat, inguinal lymph nodes (LN), mesenteric LN,
`axillary LN, cervical LN, popliteal LN, and Peyer’s patches were dissected.
`Tissue collection was performed only after the 0.3 and 2 mg/kg doses. Samples
`collected at each time point from each of the three animals were pooled in
`equal volumes, homogenized in physiological buffer (4 times tissue weight, pH
`7.4), and stored at ⫺20°C.
`Single oral doses of FTY720 (2.8 and 7.5 mg/kg) were also administered to
`rats. Serial blood sampling was performed for venous blood. For the dose of
`2.8 mg/kg, spleen samples were collected at 1, 8, 12, 24, 48, 72, and 120 h.
`After administration of 7.5 mg/kg, brain, liver, lungs, muscle, and heart tissue
`samples were collected at 8 and 72 h.
`Eighteen male beagle dogs (3 per time point) aged 7 months and weighing
`7.6 to 10.1 kg received single oral doses of 10 mg/kg. At necropsy, on days 1,
`3, 7, 14, 21, and 28 after drug administration, the frontal lobe of the brain was
`collected.
`Clinical Study. Healthy male volunteers aged between 20 and 39 years,
`weight 68 to 95 kg, were given 1-mg doses of FTY720. Venous blood samples
`were obtained before FTY720 administration and then 1, 2, 4, 6, 8, 12, 16, 20,
`24, 36, 48, 72, 96, 120, and 168 h postdose. The study was carried out in
`compliance with the protocol and according to Good Clinical Practice with full
`informed consent according to the Declaration of Helsinki. Further details are
`available in a prior publication (Kovarik et al., 2004).
`Bioanalytical Instruments and FTY720 Assay. The liquid chromatogra-
`phy/tandem mass spectrometry system was a PE Sciex API 3000 with heated
`nebulizer interface (Applied Biosystems, Foster City, CA) together with Agi-
`lent 1100 pumps, detector, and autosampler (Agilent Technologies, Palo Alto,
`CA). The HPLC column was a 3.5 Symmetry Shield RP8, 4.6 ⫻ 50 mm
`(Waters Corp., Milford, MA).
`After thawing and rehomogenization of samples, FTY720 was extracted as
`follows. Aliquots (0.1 ml) of calibration standards, quality controls, or un-
`knowns were added to 0.1 ml of 100 ng/ml Y32919 in methanol, 0.5 ml of 0.1
`N sodium hydroxide solution added, followed by 6 ml of tert-butyl-methyl
`ether and dichloromethane (75:25 v/v). After shaking for 45 min followed by
`centrifugation for 10 min at 2000g, the organic phase was transferred and
`evaporated under N2, reconstituted with 150 l of HPLC eluent, sonicated for
`5 min, mixed, and centrifuged for 5 min at 15,000g, and then injected into the
`liquid chromatograph/tandem mass spectrometer.
`The chromatography system was eluted with 70% methanol plus 30% 0.02
`M ammonium acetate at 1 ml/min. The parent molecular ion for FTY720 is
`308.3 Da (⫹1⫹ proton) with a daughter ion of 255.3 m/z. Y32919 has a parent
`of 324.4 Da and daughter of 271.4 m/z. The integrity of the original compounds
`was checked using the mass spectrometer, and there was no carryover into
`either the FTY720 or the Y32919 detection channels. The lower limit of
`quantitation was 0.5 ng/ml (11.9% CV) and the upper limit 2000 ng/ml (2.33%
`CV). The recovery was 86.1% at 30 ng/ml and 95.8% at 1250 ng/ml. The
`
`intraday and interday CV% ranges were 4.15 to 5.94% and 5.01 to 5.96%,
`respectively.
`Partition Coefficients. The AUC (area under the concentration-time
`curves) of drug in blood and tissues was calculated by
`
`AUC ⫽冘
`
`n⫺1
`
`i⫽0
`
`关共Ci⫹1 ⫹ Ci兲 䡠 共ti⫹1 ⫺ ti兲/2兴 ⫹
`
`Cn
`z
`
`(1)
`
`where Ci is the drug concentration at t ⫽ ti and Z is the terminal slope of the
`
`concentration-time curve. The tissue-to-blood partition coefficient (RT) for
`noneliminating tissues was obtained from
`
`RT ⫽
`
`AUCT
`AUCV
`
`⬇ AUCT
`AUCA
`
`(2)
`
`where AUCT, AUCV, and AUCart are derived from drug concentration versus
`time profiles for tissues and venous and arterial blood. This equation was used
`to calculate RT for lungs, heart, brain, kidneys, thymus, spleen, muscle, and
`lymph nodes for the 0.3 and 2 mg/kg i.v. doses.
`The volume of distribution at steady state (VB,ss) was calculated as
`
`VB,ss ⫽ 冘
`
`non-elim
`
`关RT/B 䡠 VT兴 ⫹ RH/B 䡠 Vliver 䡠 共1 ⫺ Eliver兲 ⫹ Vblood
`
`(3)
`
`Tissue
`
`where VT represents specific tissue volumes and Eliver is the hepatic extraction
`ratio.
`Compartmental Analysis. Arterial blood concentration-time data for
`FTY720 after i.v. administration of 0.3 and 2 mg/kg doses were analyzed by
`compartmental analysis using the computer software WinNonlin ver. 4.0
`(Pharsight Corporation, Mountain View, CA).
`Model Development. Local models were developed with blood and organ
`concentration-time data obtained after single i.v. doses of 0.3 and 2 mg/kg.
`Data from each tissue were first considered individually. The choice of final
`structural models was based upon inspection of quality of fits to the data and
`the Akaike information criterion. Parameter values obtained with local models
`were then used as initial estimates for the whole body model. The fitted
`parameters were tissue-to-blood partition coefficients (RT), permeability-sur-
`face area products (PST), and liver intrinsic clearance (CLint). The first-order
`absorption rate constant (kabs) and fraction of drug absorbed (Fabs) were
`determined by adapting the i.v. PBPK model to oral data obtained after the
`single doses of 2.8 and 7.5 mg/kg. For the oral PBPK model, all parameters
`(RT, PST, and CLint) were maintained as constant. Only parameters kabs and
`Fabs were optimized. To verify the linearity of FTY720 kinetics with respect to
`dose, blood and tissue data obtained after the single 4 mg/kg i.v. dose were
`superimposed on the corresponding simulated curve based upon the parameters
`determined with the 0.3 and 2 mg/kg doses. Single i.v. bolus doses were
`regarded as a 7.5-s i.v. infusion, and drug was assumed to distribute evenly in
`the whole vein pool once injected. Based on metabolism and clearance data
`(Novartis Pharma AG, internal communication), the liver was considered as
`the sole eliminating organ.
`Local Models. FTY720 concentration-time data after i.v. doses were fitted
`
`to
`
`Cart ⫽ A ⫻ e⫺␣⫻t ⫹ B ⫻ e⫺⫻t ⫹ C ⫻ e⫺␥⫻t
`
`(4)
`
`Equations for well stirred organs (heart, lungs, muscle, spleen, liver, and
`kidneys) were
`
`dXT
`dt
`
`⫽ QT ⫻
`
`Xart
`VT
`
`⫺ QT ⫻
`
`XT
`共RT ⫻ VT兲
`
`(5)
`
`Equations for permeability rate-limited transport (brain, thymus, lymph
`nodes) were:
`
`Blood compartment
`
`dXT1
`dt
`
`⫽ QT ⫻
`
`Xart
`VT1
`
`⫺ QT ⫻
`
`XT1
`VT1
`
`⫺ PST ⫻ fub ⫻
`
`⫹ PST ⫻冉fub
`
`RT
`
`冊 ⫻
`
`XT2
`VT1
`
`XT1
`VT1
`
`(6)
`
`
`
`1482
`
`MENO-TETANG ET AL.
`
`TABLE 1
`Standard physiologic parameters for rats weighing 250 g
`
`Organ
`
`Lungs
`Braina,b
`Heart
`Stomach
`Gut
`Pancreas
`Spleen
`Kidneys
`Thymusa,b
`
`Mass
`
`g
`1.0
`1.7
`0.8
`1.1
`10
`1.3
`0.6
`2.3
`0.7
`
`Q
`
`ml/min
`43.0
`1.33
`3.92
`1.13
`7.52
`0.52
`0.63
`9.23
`0.3
`
`Organ
`
`Muscle
`Lymph nodesa,b,c
`Liver
`Skin
`Bone
`Fat
`Arterial blood
`Venous blood
`
`Mass
`
`g or ml
`122
`0.9
`10.3
`40
`15.8
`10
`5.6
`11.3
`
`Q
`
`ml/min
`7.5
`0.16
`2.0
`5.83
`2.53
`0.4
`
`a Fractional volume of vascular space: brain, 0.014; thymus and lymph nodes, 0.009.
`b Fractional volume of interstitial space: brain, 0.188; thymus and lymph nodes, 0.150.
`c Sum of mesenteric, axillary, inguinal, cervical, Peyer’s patches, and popliteal lymph nodes.
`
`Interstitial and intracellular compartments
`
`dXT2
`dt
`
`⫽ PST ⫻ fub ⫻
`
`XT1
`VT2
`
`⫺ PST ⫻冉fub
`
`RT
`
`冊 ⫻
`
`XT2
`VT2
`
`(7)
`
`where QT is organ blood flow rate, VT is organ volume, Xart is input from
`arterial blood, PST is permeability-surface area product, RT is tissue-to-blood
`partition coefficient, fwT is fractional volume of vascular and/or interstitial
`space, VT1 is volume of blood ⫹ interstitial fluid (VT1
`⫽ fwT
`⫻ VT), VT2 is
`⫽ VT
`⫺ VT1), and fub is free fraction of
`volume of intracellular space (VT2
`FTY720 in blood.
`Organ volumes (Vi) were scaled from published data for a 250-g rat
`(Bernareggi and Rowland, 1991; Davies and Morris, 1993) using
`
`V2 ⫽
`
`BW2 䡠 V1
`BW1
`
`(8)
`
`where BW1 is the rat standard body weight (250 g) and BW2 is the actual body
`weight of animals used in the current study (mean 362 g). Blood flow of tissues
`except thymus and lymph nodes was similarly scaled using
`
`Q2 ⫽
`
`BW2 䡠 Q1
`BW1
`
`(9)
`
`Blood flow for thymus and lymph nodes was scaled from Cahill and Trnka
`(1978). All physiologic parameters used for this analysis are listed in Table 1.
`Drug concentrations for each tissue were converted to ng/ml from ng/g using
`tissue density.
`Model Equations for Whole Body Model. The whole body PBPK model
`for FTY720 is depicted in Fig. 1. It depicts the body as composed of 13 tissue
`compartments and 2 blood compartments (arterial and venous pools) with the
`lungs closing the loop. Drug is placed in the venous compartment. All equa-
`tions were solved simultaneously using the maximum likelihood estimator in
`ADAPT II.4 (D’Argenio and Schumitzky, 1997). Although FTY720 concen-
`trations in bone, gastrointestinal tract, and skin were not collected, these organs
`were included as lumped compartments representing the gastrointestinal
`tract—splanchnic compartment—and “rest-of-body”. It was assumed that
`these compartments obey the distribution characteristics of a well stirred
`model.
`Interspecies Scale-Up. The PBPK model developed for rat was scaled-up
`to dog and humans. Physiological values (organ volume and blood flow rate)
`were taken from the literature. Unbound equilibrium distribution ratio (RTu) in
`dogs and humans for each tissue was calculated from RT values obtained with
`the rat model and corresponding fub values for each species (fub values were
`0.000241 in dogs and 0.000361 in humans). The PST values for brain, thymus,
`and lymph nodes for dog and human were predicted from the measurement in
`rat by use of an allometric equation
`
`PST ⫽ A共M兲B
`
`(10)
`
`where M is organ mass (or weight) and A and B are allometric coefficients. The
`B value of 0.67 was fixed, assuming that permeability of tissue cellular
`membrane and organ structure is geometrically similar among mammals.
`
`FIG. 1. Whole body PBPK model for FTY720. Subscripts are: A, arterial blood; V,
`venous blood; b, brain; h, heart; li, liver; sp, spleen; lu, lungs; g, gut-stomach-
`pancreas; kd, kidney; f, fat; sk, skin; m, muscle; th, thymus; ln, lymph nodes; and
`bn, bone.
`
`To predict the blood concentration versus time profile for FTY720 in dogs,
`a first-order absorption rate constant (kabs) and bioavailability values were
`taken from the results of a compartmental analysis. Because no in vitro data
`were available for the biotransformation of FTY720 in dogs, the intrinsic
`clearance in dogs (CLintD) was calculated by nonlinear regression analysis with
`the whole body PBPK model, keeping all distribution parameters (RT and PST)
`identical to those from the rat after correction for binding in blood.
`To predict the concentration versus time profile for FTY720 in human
`blood, the first-order absorption rate constant (kabs) was taken from the results
`of compartmental analysis. Because no i.v. formulation for FTY720 is avail-
`able for humans, we assumed that humans had the same bioavailability (F) as
`rats. The intrinsic clearance in humans (CLint,H) value was calculated in two
`ways. First, CLint,H was predicted from the product of the human in vitro
`clearance determined with liver microsomes and the in vivo to in vitro CLint
`ratio from rats. Second, when human FTY720 concentration data became
`available, they were used to improve the estimate of human CLint,H, keeping all
`distribution parameters (RT and PST) identical to those from the rat after
`correction for binding in blood.
`
`Results
`Blood Profiles. Figures 2 and 3 depict blood concentration-time
`profiles for FTY720 after i.v. and oral administration. It appears from
`the graphs that the model captured the venous blood concentration-
`time profiles relatively well for the 0.3, 2, and 4 mg/kg i.v. doses. In
`contrast, arterial concentration-time profiles were somewhat overes-
`timated. Venous blood concentration-time profiles for the 2.8 and 7.5
`mg/kg oral doses were also well captured by the PBPK model. The
`model yielded a liver intrinsic clearance (CLint) value based upon free
`FTY720 concentrations in blood of 23,145 l/h/kg. The corresponding
`systemic clearance value, according to the well stirred model, was
`⫽ 0.000333). After oral administration, the first-
`0.748 l/h/kg (fub
`
`
`
`FTY720 PBPK MODEL IN RAT, DOG AND HUMAN
`
`1483
`
`FIG. 2. Time course of FTY720 concentrations in rat venous blood after the indicated i.v. and oral doses. Symbols are experimental values and lines are PBPK model
`predictions.
`
`FIG. 3. Time course of measured (symbols) and predicted (lines) FTY720 concentrations in rat arterial blood and lymph nodes.
`
`order absorption rate constant was 0.052 h⫺1 and fraction of drug
`absorbed was 0.45.
`Well Stirred Organs. Concentration versus time profiles for heart,
`lungs, liver, kidneys, muscle, fat, and spleen captured by the PBPK
`model are depicted in Figs. 4 to 7. Concentration-time curves of these
`organs declined in parallel with that of venous and arterial blood.
`Tissue-to-blood partition coefficients (RT) calculated by noncompart-
`mental analysis and those obtained with the PBPK model are reported
`in Tables 2 and 3. Model-defined RT values were obtained with
`relatively good precision, with CV% on parameters generally ⬍20%.
`In most organs, the deviation between the noncompartmental and
`PBPK-defined RT values never exceeded 3-fold. Among these well
`stirred organs, FTY720 distributed most extensively into lungs and
`liver. The extent of distribution into fat and skeletal muscle was
`moderate.
`Permeability-Limited Organs. The concentration-time curves for
`
`FTY720 in lymph nodes, brain, and thymus are shown in Figs. 3, 4,
`and 6. The model captured the observed data fairly well, with the
`exception of thymus, for which an overprediction of tissue concen-
`tration levels was noticed. Nevertheless, the model still captured the
`trend in the data as well as the terminal slope of the drug in that organ.
`To develop the organ model for lymph nodes, it was assumed that
`fractional volumes of vascular and interstitial spaces were identical to
`those of thymus because the two organs belong to the immune system.
`The PBPK model captured fairly well the features of the data. Tissue-
`to-blood partition coefficient values were 22.9 for lymph nodes, 27.1
`for brain, and 15.7 for thymus. The corresponding PST values were
`176, 39.3, and 122 ml/min (Table 3).
`Other Organs. Organs that were not sampled, such as bone and
`skin, were lumped into a single compartment labeled “rest of body”.
`A nonsampled splanchnic compartment comprising stomach and in-
`testinal tract was also incorporated to account for the flow of blood
`
`
`
`1484
`
`MENO-TETANG ET AL.
`
`FIG. 4. Time course of measured (symbols) and predicted (lines) FTY720 concen-
`trations in rat heart, lungs, and brain after i.v. and oral administration.
`
`FIG. 5. Time course of measured (symbols) and predicted (lines) FTY720 concen-
`trations in rat liver, kidneys, and muscle after i.v. and oral administration.
`
`through the liver. The model-predicted RT values were 50.9 for the
`rest-of-body and 11.1 for the splanchnic compartment. Being unob-
`served, these tissue-to-blood partition coefficients may not reflect the
`real extent of FTY720 distribution in these organs.
`Prediction of FTY720 Kinetics in Dogs and Humans. Figure 8
`depicts FTY720 blood concentration-time profile in the dog after
`i.v. (Fig. 8, top) and oral (Fig. 8, middle) administration. The
`PBPK model was able to capture the experimental data reasonably
`well. Despite an overestimation of Cmax after oral administration,
`the model described the terminal slope with good accuracy.
`FTY720 concentrations in dog brain were simulated for a single
`oral dose of 10 mg/kg. We used the appropriate dog physiology
`and the extrapolated organ distribution parameters from the whole
`body PBPK model. As shown in Fig. 8 (bottom), the predictions
`were slightly lower than the observed values, by just under 1.5-
`fold, but the model predicted accurately the slope of decay of
`FTY720.
`Figure 9a shows the FTY720 concentration-time profile in human
`blood after a single oral dose of 1 mg, as predicted from the rat PBPK
`model with the in vitro intrinsic clearance determined from human
`liver microsomes. Although the prediction was not perfect, it was
`within interindividual variability. It was noted that the predicted Cmax
`value for FTY720 was different from the observed values. Figure 9b
`shows the FTY720 concentrations in human blood obtained after
`optimizing the intrinsic clearance to fit the human data. A great
`improvement was observed and the observed concentration data were
`better described by the PBPK model.
`
`Discussion
`The present analysis characterizes the concentration-time profiles
`for FTY720 in venous and arterial blood and in an extensive array of
`organs after i.v. and oral administration to rats. Gaining insight into
`the pharmacokinetic behavior of a drug in different organs is highly
`desirable, especially in therapeutic target organs or those for which the
`compound may be potentially harmful. We were able to analyze
`FTY720 concentration versus time profiles in the typical major or-
`gans, but also in other rather unusual ones, viz. lymph nodes, spleen,
`and thymus. FTY720 is an immunomodulator that exerts its pharma-
`cological action by sequestering lymphocytes in secondary immune
`organs. Thus, being able to characterize the behavior of the drug in
`pharmacological targets such as spleen and lymph nodes constitutes a
`unique achievement.
`There are several advantages of developing PBPK models. Two
`obvious advantages are the possibility to investigate, in animals,
`organs that could otherwise never be assessed in humans, and the
`possibility to extrapolate the model to higher species by adjusting
`organ volumes and perfusion rates, and adjusting for differences in
`metabolism, plasma protein binding, and blood cell distribution be-
`tween species.
`The results of the investigation showed that FTY720 distributed
`extensively into various organs. More importantly, distribution and
`clearance appeared to be linear in the dose range of 0.3 to 4 mg/kg.
`The model-predicted RT values were similar to those calculated by
`noncompartmental analysis. On average, a 3-fold deviation was ob-
`served between the two analyses (Table 2 versus Table 3).
`
`
`
`FTY720 PBPK MODEL IN RAT, DOG AND HUMAN
`
`1485
`
`FIG. 6. Time course of measured (symbols) and predicted (lines) FTY720 concen-
`trations in rat fat and thymus after i.v. and oral administration.
`
`Brain, lymph nodes, and thymus concentration-time data could be
`described by the permeability-limited model only. Although this
`model seems natural for brain because of the blood-brain barrier to
`xenobiotics, the applicability to lymph nodes and thymus is less clear.
`To our knowledge, specific cell membrane resistance to drug transport
`into thymus or lymph nodes has not been documented. The PST values
`for lymph nodes, brain, and thymus translate to transfer clearance (i.e.,
`product of PST and fub) values of 0.058, 0.0131, and 0.048 ml/min,
`respectively. These transfer clearances are all smaller than the corre-
`sponding organ blood flows (Table 1). Thus, modeling FTY720
`kinetics in these tissues using a permeability-limited model was
`warranted.
`There were some discrepancies between tissue binding parameters
`obtained with local and whole body models. Several reasons may
`explain these differences. First, each organ in the body is, in fact, in
`equilibrium with the blood leaving the organ, rather than with the
`measured venous blood as assumed in the local models. This discrep-
`ancy is even more pronounced for the lungs, which are in equilibrium
`with the arterial blood. Second, in the determination of local organ
`parameters, we used observed venous blood concentration, whereas
`the whole body model takes into account the model-predicted blood
`concentration. As can be seen in Fig. 2, the predicted venous blood
`concentrations after the bolus administration are higher than the
`observed data, especially for the lower dose of 0.3 mg/kg.
`The volume of distribution at steady state estimated with eq. 3
`yielded a value of 8.61 l/kg. This figure is close to the one obtained
`by a standard compartmental analysis (13.7 l/kg, Table 4). The intrin-
`sic clearance estimated by the current model was 23,145 l/h/kg. This
`value seems very large. However, because FTY720 is highly bound to
`
`FIG. 7. Time course of measured (symbols) and predicted (lines) FTY720 concen-
`trations in rat spleen after i.v. and oral administration.
`
`TABLE 2
`Tissue-to-blood partition coefficient (RT) of FTY720 calculated from AUC ratios
`
`Organ
`
`Lungs
`Heart
`Brain
`Kidneys
`Thymus
`Spleen
`Liver
`Muscle
`Lymph nodesa
`
`Intravenous Doses
`
`0.3 mg/kg
`
`2 mg/kg
`
`89.9
`27.7
`53.4
`53.7
`19.6
`67.4
`83.0
`9.4
`28.6
`
`68.2
`17.4
`49.2
`35.8
`19.6
`62.1
`47.0
`10.5
`32.6
`
`a Pool of mesenteric, axillary, inguinal, cervical, and popliteal lymph nodes.
`
`plasma proteins (Novartis Pharma AG, internal communication) and
`distributes extensively into red blood cells, this intrinsic clearance
`corresponded to a systemic clearance value of 0.748 l/h/kg. This value
`was not too different from the average systemic clearance obtained by
`compartmental analysis (0.497 l/h/kg, Table 4). Due to strong binding
`to blood components, extensive distribution, and a relatively low
`systemic clearance, FTY720 has a half-life of about 20 h in rats
`(Table 4).
`In the development of the PBPK model, it was assumed that
`FTY720 is eliminated from the body by metabolic biotransformation
`occurring mostly in the liver. This assumption was based on the fact
`that no parent drug could be detected in urine samples. OneOne comp
`compli-li-
`
`compoundound underundergoesgoes revers
`
`
`
`ation inin FTY720FTY720 metabolismmetabolism isis thatthat thethe comp
`revers--
`ccation
`ibliblee metabmetabolismolism (Novart
`
`
`ication). In-In-arma AG,AG, internainternall commun
`(Novartisis PhPharma
`communication).
`
`
`
`
`terconversionion bbetweenetween FTY720FTY720 ananddiitsts phosphorylated metametabolitebolite
`terconvers
`phosphorylated
`
`
`
`FTY720-P) iiss catacatalyzedlyzed bbyytytypep 1e 1aandnd 22sphingosine kikinases.nases. ThThese
`((FTY720-P)
`sphingosine
`
`
`
`1486
`
`MENO-TETANG ET AL.
`
`TABLE 3
`Physiologic pharmacokinetic parameters for FTY720 in rats, dogs, and humans using the whole body PBPK model
`Parameters for dogs and humans were derived from those of rats using their respective drug free fractions in blood.
`
`Species
`
`Parameters
`
`BW (kg)
`fub
`CLint (l/min/mg protein)-microsomes
`a (l/h/kg)
`CLint
`Fabs
`kabs (h⫺1)
`Rlungs
`Rbrain
`Rheart
`Rspleen
`Rmuscle
`Rthymus
`Rkidneys
`RLN
`Rliver
`Rfat
`RGIT
`Rrest-of-body
`PSbrain (ml/min)
`PSthymus (ml/min)
`PSLN (ml/min)
`
`b
`
`Rat
`
`0.362
`0.000333
`88
`2315
`0.71
`0.0007
`41.4
`27.1
`13.8
`34.7
`4.69
`15.8
`22.3
`22.9
`34.9
`0.647
`11.1
`50.9
`39.3
`122
`176
`
`CV%
`
`(12.2)
`(18.5)
`(19.3)
`(12.5)
`(11.9)
`(15.0)
`(10.5)
`(14.1)
`(11.72)
`(9.3)
`(12.9)
`(14.6)
`(9.84)
`(54.4)
`(8.3)
`(12.1)
`(11.3)
`(16.9)
`
`Dog
`
`11.65
`0.000241
`N.A.
`1018
`0.6
`0.00843
`29.9
`19.5
`9.96
`25.0
`3.39
`11.4
`16.1
`16.5
`25.2
`0.47
`8.01
`36.8
`379
`572
`820
`
`Human
`
`70
`0.000361
`33
`202
`0.5
`0.000743
`44.8
`29.3
`14.9
`37.5
`5.08
`17.1
`24.1
`24.8
`37.8
`0.70
`12.0
`55.1
`3613
`1517
`2173
`
`N.A., not applicable.
`a Based on free fraction in blood.
`b Value for liver here is equal to Rli 䡠 (1 ⫺ CLH,int/Qhp) rather than Rli.
`
`FIG. 8. Time course of measured (symbols) and predicted (lines) FTY720 concen-
`trations in blood of dogs after oral (top) or i.v. (middle) dosing. The bottom panel
`shows predictions of brain concentrations.
`
`FIG. 9. Time course of measured (symbols) and predicted (lines) FTY720 concentra-
`tions in blood of humans. a represents predictions using in vitro microsomal CLint, and
`b represents predictions from whole body PBPK after optimization of CLint.
`
`
`
`FTY720 PBPK MODEL IN RAT, DOG AND HUMAN
`
`1487
`
`TABLE 4
`Pharmacokinetic parameters for FTY720 in blood after single i.v. doses in rats
`calculated by compartmental analysis
`
`Parameter
`
`t1/2 (h)
`CL (l/h/kg)
`Vss (l/kg)
`
`Dose
`
`Mean
`
`S.D.
`
`0.3 mg/kg
`
`2 mg/kg
`
`4 mg/kg
`
`18.5
`0.341
`5.61
`
`30.7
`0.462
`16.2
`
`20.9
`0.687
`19.4
`
`23.4
`0.497
`13.7
`
`6.43
`0.176
`7.22
`
`
`enzymes are found in numerous tissues including liver, eye, lung, andenzymes are found in numerous tissues including liver, eye, lung, and
`
`muscle. The characterization of reversible systems can be achievedmuscle.
`only after collecting concentration-time data of parent and metabolite
`after separate i.v. administration of parent and metabolite (Cheng and
`Jusko, 1993). No metabolite concentration data were measured in the
`Therefore, the present PBPK model may only be an
`present study. Therefore, the present PBPK model may only be an
`approximation of FTY720 kinetics in blood and organs because
`approximation of FTY720 kinetics in blood and organs because
`reversible metabolic processes were not taken into account.
`reversible metabolic processes were not taken into account. Interest in
`FTY720-P is high; it is the pharmacologically active moiety. How-
`ever, with an equilibrium between FTY720 and FTY720-P, RT and
`PST will be apparent parameters incorporating both partition-diffusion
`to and from tissues, together with reversible metabolic interconver-
`sion. The rapidity of the achievement of an equilibrium between FTY
`and FTY-P may explain why a model of the parent works well for
`characterizing the pharmacodynamics of lymphocyte trafficking in
`monkeys (Li et al., 2002).
`The PBPK model was able to mimic blood concentration-time
`profiles in dogs after oral and i.v. administration when optimizing
`only the intrinsic clearance and maintaining the tissue binding param-
`eters from the rat. In addition, we were able to test the model
`prediction of brain concentrations in dogs. Although the predictions
`were slightly lower than the observed values, the model predicted
`accurately the slope of decay of FTY720 in dog brain. This may allow
`the determinat