![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
|
|
|
|
|
||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Departments of Clinical Pharmacology (K.V., M.A.G.), Pharmacokinetics Dynamics and Metabolism (E.T., F.R.N.), Neurosciences Discovery Biology (H.R., W.E.H.), Clinical Statistics (J.L.F.), and the New Haven Pfizer Clinical Research Unit (I.V.K.), Pfizer Global Research and Development, Groton Connecticut
(Received November 16, 2006; Accepted April 25, 2007)
 |
Abstract |
|---|
 Top Abstract Materials and MethodsResultsDiscussionReferences |
|---|
7 times that in their wild-type counterparts, confirming that impaired CNS distribution was explained by P-gp efflux transport. In contrast, P-gp efflux did not explain the impaired CNS penetration of CP-900,725. Intracerebral microdialysis was used to characterize rat brain extracellular fluid (ECF) distribution. Interestingly, the ECF-to-serum ratio of the P-gp substrate CP-615,003 was 7-fold below the CSF-to-serum ratio, whereas this disequilibrium was not observed for CP-900,725. In a clinical study, steady-state CSF exposures were measured after administration of 100 mg of CP-615,003 b.i.d. The human CSF-to-plasma ratios of CP-615,003 and CP-900,725 were both 10-fold below their ex vivo plasma unbound fractions, confirming impaired human CNS penetration. Preliminary estimates of CNS receptor occupancy from human CSF concentrations were sensitive to assumptions regarding the magnitude of the CSF-ECF gradient for CP-615,003 in humans. In summary, this case provides an example of intersite differences in CNS pharmacokinetics of a P-gp substrate and potential implications for projection of human CNS receptor occupancy of transporter substrates from CSF pharmacokinetic data when direct imaging-based approaches are not feasible.
). However, in the monkey and in humans, oxidative deamination is the primary route of metabolism resulting in the formation of a circulating metabolite CP-900,725 (Shaffer et al., 2005; Fig. 1B), which is pharmacologically active (target receptor binding Ki 1.9 nM).
|
; Shen et al., 2004; Doran et al., 2005), it was necessary to quantitatively and mechanistically evaluate the CNS distribution of CP-615,003 and obtain initial estimates of projected target receptor occupancy at clinical dose(s). In this investigation, we describe the application of a battery of in vitro, in vivo preclinical, and clinical pharmacologic approaches to characterize the CNS disposition and blood-brain barrier transport of CP-615,003 and its active metabolite CP-900,725. A key focus of the studies described in this investigation is to characterize the brain extracellular fluid (ECF)-to-cerebrospinal fluid (CSF) gradient of CP-615,003 in the rat using a combination of intracerebral microdialysis and CSF pharmacokinetic studies and to leverage understanding of this gradient to obtain initial estimates of receptor occupancy from clinical CSF exposures, since direct measurement of human brain ECF exposure is not generally feasible in a drug development setting. Because the ECF/CSF ratio of CP-615,003 in humans is not known, the objective of these preliminary receptor occupancy projections was primarily to explore/illustrate the sensitivity of projected occupancies to assumptions regarding the human CSF-ECF gradient, rather than to report definitive estimates. Given the recent reinforcement in the CNS distribution literature of the heterogeneity of the CNS circulatory system and the appreciation that synchrony between pharmacokinetics in CSF and brain ECF cannot always be assured, especially when there is evidence for transport at the blood-brain and/or blood-CSF barriers (de Lange and Danhof, 2002; Shen et al., 2004), the results of this investigation have important broader implications for preclinical and clinical strategy for the assessment of human CNS penetration of transporter substrates in drug development.
|
Materials and Methods |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
Transport Studies across MDCK and MDR1-MDCK Cell Monolayers. The transepithelial transport of CP-615,003 and CP-900,725 across Madin-Darby canine kidney II (MDCK) cells and Madin-Darby canine kidney II cells expressing the human MDR1 P-glycoprotein transporter (MDR1-MDCK; P. Borst, Netherlands Cancer Institute, Amsterdam, The Netherlands) was studied using a 96-well Transwell transport assay using previously described methods (Chen et al., 2005). In brief, MDCK or MDR1-MDCK cells were seeded onto membranes of 96-multiwell inserts at a density of 2.33 x 105 cells/cm2 and incubated for 4 days at 37°C in MEM Alpha growth medium supplemented with 8.85% fetal bovine serum, 1.77 mM L-glutamine, 88.5 µM nonessential amino acids, and 88.5 units/ml penicillin and streptomycin. For the transport study, solutions of the compound (2 µM concentration; dimethyl sulfoxide solvent concentration of <0.01%) in transport buffer (Hanks' balanced salt solution supplemented with 23.8 mM D-glucose, 19 mM HEPES, 1.2 mM CaCl2, and 0.477 mM MgCl2) were added to the apical (A) or basolateral (B) side (donor compartment), with compound-free transport buffer on the opposite side (receiver compartment). After incubation at 37°C for 5 h, aliquots of samples from the receiver compartment were processed for bioanalysis of concentrations of CP-615,003 or CP-900,725 (see bioanalysis section below for LC/MS/MS methodology). The apparent permeabilities (Papp, cm/s) in each direction (A
B and BA) were calculated as follows (Mahar Doan et al., 2002), where A is the surface area of the membrane inserts (0.0804 cm2), C0 is the initial concentration of the compound applied in the donor compartment (2 µM), and 
Q is the amount (micromoles) of compound transported over time t (5 h = 18,000 s):
 | (1) |
A direction divided by the apparent permeability in the AB direction. Positive and negative controls included in the transport experiment were the strong and weak P-gp substrates quinidine and prazosin, respectively (Polli et al., 2001; Doran et al., 2005), and triprolidine, which is not a substrate for P-gp (Chen et al., 2003a). The 5-h duration of time (t) used in the transport experiments was determined to be within the linear range in preliminary studies.
Whole Brain and CSF Pharmacokinetics in Rats. Male Sprague-Dawley rats (n = 4 per time point per compound) weighing 250 to 300 g, were obtained from Charles River Laboratories (Wilmington, MA). A single oral dose of CP-615,003 (40 mg/kg) in 0.5% aqueous methyl cellulose or a single subcutaneous dose of CP-900,725 (10 mg/kg) in 20% (2-hydroxypropyl)-ß-cyclodextrin was administered to each rat. Rats dosed with CP-615,003 were euthanized in a CO2 chamber at 0.5, 2, 4, 7, 10, 16, and 24 h postdose. Rats dosed with CP-900,725 were euthanized at 0.5, 1, 2, 4, 7, 11, and 24 h postdose. CSF was collected via puncture of the cisterna magna using a 23-gauge needle attached to polyethylene tubing (internal diameter of 50 mm) and syringe. Whole blood was collected by cardiac puncture and centrifuged to prepare serum. Whole brains were collected by decapitation, rinsed in phosphate-buffered saline, and weighed. CSF and whole brains were immediately frozen on dry ice upon collection.
CSF Pharmacokinetics of CP-615,003 in Dogs. Three male beagle dogs with indwelling subarachnoid catheters and subcutaneous access ports were each administered an intravenous bolus injection of CP-615,003 at a dose of 2 mg/kg in saline through the cephalic vein of the foreleg at 0.2 ml/kg. Blood samples (3 ml) were obtained immediately before dosing and at 0.08, 0.17, 0.33, 0.50, 0.75, 1, 1.5, 2, 4, 6, 8, 11, 24, 30, 35, and 48 h postdose, and centrifuged to prepare plasma. CSF samples (approximately 0.25 ml) were obtained before dosing and at 0.08, 0.17, 0.5, 1, 2, 4, 6, 8, 11, and 24 h postdose. Plasma and CSF samples were stored at –20°C until analysis.
Brain Distribution Studies in mdr1 P-gp-Deficient Mice. Male FVB/N (wild-type) and mdr1a/1b (–/–, –/–) mice (n = 5 per genotype per time point per compound) weighing approximately 22 to 27 g, were obtained from Taconic Labs (Germantown, NY). A single subcutaneous dose of CP-615,003 (5 mg/kg) in phosphate-buffered saline or CP-900,725 (5 mg/kg) in 20% (2-hydroxypropyl)-ß-cyclodextrin was administered to each mouse. Mice were euthanized in a CO2 chamber at 0.5, 1, 2.5, and 5 h postdose (CP-615,003 study) or at 0.5, 1.5, and 4 h postdose (CP-900,725 study). Whole blood was collected by cardiac puncture and centrifuged to prepare serum. Whole brains were collected by decapitation, rinsed in phosphate-buffered saline, weighed, and immediately frozen on dry ice.
Brain ECF Pharmacokinetics by Microdialysis in Rats. In vivo microdialysis was performed in awake, freely moving, jugular vein-cannulated male Sprague-Dawley rats weighing 280 to 320 g (n = 3 per compound; Charles River Laboratories). BR-4 microdialysis probes (BAS Inc., West Lafayette, IN) with a 4-mm polyacrylonitrile membrane (30-kDa molecular mass cutoff), were stereotaxically implanted under isoflurane anesthesia in the striatum [coordinates A-P (anteroposterior) +0.7 from bregma, M-L (mediolateral) +3.0 from midline, D-V (dorsoventral) –7.5 from dura; Paxinos and Watson, 1997] and fixed to the skull using bone screws and dental acrylic. The inlet and outlet of the probe were connected with flexible PEEK tubing (inside diameter 0.005 inch; Upchurch Scientific, Oak Harbor, WA) via a dual channel fluid swivel system (Instech Laboratories Inc., Plymouth Meeting, PA), and the probe was perfused overnight at 0.3 µl/min with artificial CSF (147 mM NaCl, 1.3 mM CaCl2, 2.7 mM KCl, 1.0 mM MgCl2) using a CMA/102 microperfusion pump (CMA Microdialysis, Acton, MA). Animals were allowed to recover for 24 h and the probe outlet was connected via flexible PEEK tubing to a CMA/142 fraction collector or Univentor 820 microsampler (Scipro Inc., Sanborn, NY) on the day of the microdialysis ECF pharmacokinetic study. The perfusion rate was increased to 1 µl/min and three basal predrug microdialysate samples were collected. Three rats were each administered a single oral dose of CP-615,003 (40 mg/kg) in 0.5% aqueous methyl cellulose. Dialysate samples (30 µl) were continuously collected, every 30 min, for a period of 8 h and every 60 min (60 µl) for the next 16 h. Blood samples were simultaneously collected at 2, 4, 6, 9, 12, and 24 h postdose to prepare serum. In a second study, three rats were each administered two subcutaneous doses of CP-900,725 (10 mg/kg) in 20% (2-hydroxypropyl)-ß-cyclodextrin, 2.5 h apart (this approach was used because of the relatively short half-life of CP-900,725). Dialysate samples (30 µl) were continuously collected, every 30 min, for a period of 8 h and every 60 min (60 µl) for the next 18 h. Blood samples were simultaneously collected at 15 and 45 min after the first dose and 0.25, 0.5, 1, 2, 4, 6, 8, 10, and 24 h after the second dose for the preparation of serum. The concentrations of CP-615,003 and CP-900,725 in microdialysate samples were divided by their respective mean extraction efficiencies (expressed as fractions) across the dialysis probe to estimate the corresponding concentrations of unbound drug in brain ECF for pharmacokinetic calculations, as described earlier (Stain-Texier et al., 1999).
For both CP-615,003 and CP-900,725, in vitro recovery and delivery experiments were initially performed to verify that the extraction efficiencies across the probe in both directions (recovery and delivery paradigms) were quantitatively similar, to support the use of in vivo delivery (estimated by retrodialysis) as an index of in vivo recovery, as described earlier (Edwards et al., 2002). The in vivo extraction efficiencies of the delivery of CP-615,003 and CP-900,725 across the microdialysis probe were determined using a retrodialysis approach. Microdialysis probes were implanted in the striatum as described earlier and perfused at a flow rate of 1.0 µl/min with artificial CSF containing CP-615,003 or CP-900,725 (n = 4 rats per compound) at a nominal concentration of 5 ng/ml. Microdialysate was continuously collected for 5 h, every 30 min, and stored at –20°C until assay. The extraction efficiency (EED) was calculated from the concentration in the infused solution (C0) and the dialysate concentration at equilibrium (Css) using the following equation (Stain-Texier et al., 1999):
 | (2) |
teq), will likely bias the estimation of ECF concentrations from the measured microdialysate concentrations during the preequilibrium time frame (t = 0to teq).
Serum and Brain Tissue Binding Studies. The binding of CP-615,003 and CP-900,725 in pooled rat or dog serum (at 20–2000 ng/ml CP-615,003 and 10–1000 ng/ml CP-900,725), ex vivo protein binding samples from the clinical study, and 20% rat brain homogenate (100 ng/ml CP-615,003 or CP-900,725) prepared in Dulbecco's phosphate-buffered saline (pH 7.4) were determined by equilibrium dialysis as described previously (Reed-Hagen et al., 1999). Bioanalysis of CP-615,003 and CP-900,725 in dialysate and retentate samples was performed using the LC/MS/MS method described later. The unbound fractions of CP-615,003 or CP-900,725 in undiluted brain tissue (fu,brain) were calculated from their unbound fractions in the homogenate (fu,h) and the homogenate dilution factor (D = 5 for 20% homogenate) as described previously (Kalvass and Maurer, 2002),
 | (3) |
Clinical Study Design. An open-label, nonrandomized study was conducted in healthy adult male and/or female (of non-childbearing potential) subjects between the ages of 18 and 55 years, inclusive. Subjects were admitted to the clinical research unit at least 12 h before dosing on day 1 and were confined to the Clinical Research Unit for a period of 4 days. Six subjects completed the study. Each subject received 100 mg of CP-615,003 (as film-coated tablets) for 3 days (b.i.d. for 2 days and q.d. for 1 day) to achieve pharmacokinetic steady state (based on previous clinical pharmacokinetic experience with CP-615,003). Dosing occurred immediately after breakfast (7:30 AM) and dinner (7:30 PM). After administration of the morning dose of CP-615,003 on day 3, the plasma steady-state pharmacokinetics of CP-615,003 and CP-900,725 over the ensuing 12 h was assessed by serial blood sampling (approximately 7 ml each, to provide approximately 3 ml of plasma) at 1, 2, 3, 4, 6, 8, and 12 h postdosing. Approximately 10 ml of CSF was obtained by lumbar puncture at approximately 4 h postdosing on day 3, simultaneously with the collection of the 4-h plasma pharmacokinetic sample. CSF samples were visually inspected and verified to be clear and free of blood contamination, followed by subsequent confirmation by microscopic analysis of red blood cell concentration in a 10-µl aliquot. At the 4-h time point on day 3, an additional blood sample was collected from each subject for ex vivo measurement of plasma protein binding of CP-615,003 and CP-900,725. Plasma and CSF samples were stored frozen at –20°C until bioanalysis for measurement of concentrations of CP-615,003 and CP-900,725.
Bioanalysis Methodology. Bioanalysis of serum (rat, mouse) or plasma (dog), brain homogenate, CSF, and microdialysate samples for the quantitation of CP-615,003 or CP-900,725 concentrations was performed using a LC/MS/MS method. Brain tissue, homogenized in four mass equivalents of phosphate-buffered saline, and serum samples were subjected to liquid-liquid extraction using the following method. Samples, standards, or quality control samples (100 µl) were spiked with a structurally similar internal standard compound, alkalinized with 50 µl of 0.1 N NaOH, and extracted with 0.7-ml methyl-t-butylether by vortexing for 5 min. After centrifugation, the organic layer was evaporated to dryness under nitrogen at ambient temperature, and the residue was reconstituted in 100 µl of 30% acetonitrile in H2O and subjected to LC/MS/MS. CSF and microdialysate samples were spiked with the internal standard and analyzed without an extraction step.
Human plasma and CSF samples were analyzed after solid phase extraction. Extractions were performed using an Oasis HLB plate (Waters, Milford, MA) reconditioned with 200 µl of MeOH and 200 µl of 40% MeOH in 2% NH4OH. Samples (100 µl) spiked with internal standard (10 µl) and 100 µl of 0.1% NH4OH were loaded onto the extraction plate and washed with 400 µl of 40% MeOH in 2% NH4OH. Samples were eluted with 90% MeOH in 2% acetic acid and evaporated to dryness under nitrogen at ambient temperature. The residue was reconstituted in 50 µl of 10% acetonitrile in H2O and analyzed by LC/MS/MS.
Chromatographic separation of CP-615,003, CP-900,725, and internal standard was performed using a Phenomenex (Torrance, CA) Synergi 4µ Hydro-RP 80Å, 50 x 2.0 mm reverse-phase HPLC column. Separation was achieved isocratically for samples from preclinical studies with the mobile phase consisting of 30:70 acetonitrile/10 mM ammonium formate buffer (pH 3) at a flow rate of 0.3 ml/min. For clinical study samples, the following gradient method was used: 10 to 40% acetonitrile over 0.5 min, 40% acetonitrile for the next 2.5 min, and reequilibration to initial conditions over the next 2 min. A PE Sciex API 3000 mass spectrometer (PerkinElmerSciex Instruments, Boston, MA) equipped with a TurboIonSpray interface was used for detection. Analytes were ionized by electrospray ionization in positive ion mode. The following multiple reaction monitoring transitions were monitored: m/z 374213 (CP-615,003, Rt = 2.05 min), m/z 333162 (CP-900,725, Rt = 2.45 min), and m/z 298106 (internal standard, Rt = 1.88 min). The dynamic range of the assay was 0.25 to 50 ng/ml for preclinical serum, brain, and CSF samples and 0.1 to 10 ng/ml for microdialysate samples. For clinical samples, the dynamic range of the assay was 0.5 to 200 ng/ml and 0.1 to 10 ng/ml for plasma and CSF, respectively. Assay performance was monitored by the inclusion of quality control samples. The intra- and interbatch accuracies and precisions for quality control samples were within 80 to 120% and ±20%, respectively.
Analysis of Brain, CSF, and ECF Pharmacokinetic Data. Composite (average of data from animals contributing to each time point) serum, whole brain, and CSF concentration-time data were analyzed from brain/CSF distribution studies in rats and mice, whereas individual animal data were analyzed from the CSF pharmacokinetic study in dogs and ECF pharmacokinetic microdialysis study in rats, followed by calculation of summary statistics from individual pharmacokinetic parameters. All pharmacokinetic analyses were performed by standard noncompartmental methods using WinNonlin Enterprise Version 3.2 (Pharsight Corporation, Mountain View, CA). The brain-to-serum, CSF-to-serum (or CSF-to-plasma), or ECF-to-serum distribution ratios (Kp) were calculated as ratios of the total exposures (AUC0-inf in the whole brain and CSF distribution studies; AUCteq-inf in the rat brain ECF distribution microdialysis study) in the respective CNS sampling sites divided by the corresponding total exposure estimates in serum (or plasma). The ECF/CSF ratio in rats was estimated as a ratio of the mean ECF/serum ratio from the microdialysis study and the CSF/serum ratio from the CSF distribution study.
Clinical Pharmacokinetic Analyses and Receptor Occupancy Projections. Individual human plasma concentration-time data of CP-615,003 and CP-900,725 on day 3 of CP-615,003 dosing were analyzed to calculate steady-state pharmacokinetic parameters by standard noncompartmental methods using WinNonlin Enterprise Version 3.2 (Pharsight). Maximum concentrations (Cmax) were reported directly from the observed concentration-time profiles. Tmax was defined as the time of the first occurrence of Cmax, and AUC calculations were performed using the log-linear trapezoidal approximation. The human CSF-to-plasma distribution ratios were calculated as the ratio of concentrations in the CSF and plasma at 4 h postdose over the steady-state dosing interval in individual subjects. Human CSF-to-plasma ratios, ex vivo estimates of plasma unbound fraction, and ratio of CSF concentration to calculated unbound plasma concentration (CSF/Cu ratio) were summarized using descriptive statistics (mean and standard deviation). An initial estimate of the human steady-state percentage receptor occupancy after administration of 100 mg of CP-615,003 b.i.d. was projected using the principle of exclusivity as follows, where FROP and FROM are the fractional receptor occupancies resulting from the parent drug (CP-615,003) and its metabolite (CP-900,725), respectively:
 | (4) |
; Shang et al., 2006):
 | (5) |
Statistical Analyses. For the rat brain ECF pharmacokinetic studies and the dog CSF pharmacokinetic study, the geometric mean ECF- or CSF-to-serum AUC ratios were compared with the respective free fractions (i.e., the theoretical expected values in the absence of CNS distributional impairment) using a one-sample t test. p values were calculated using a t distribution with 2 degrees of freedom (one less than the number of animals in these studies). For the clinical CSF distribution study, the geometric mean CSF/Cu ratio was compared with 1.0 (i.e., the theoretical expected value in the absence of CNS distributional impairment) using a one-sample t test. p values were calculated using a t distribution with 5 degrees of freedom (one less than the number of subjects). Statistical analyses were performed using the t test function in R (version 2.3.1; http://www.R-project.org).
|
Results |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
B) across parental MDCK cells were >2 x 10–6 cm/s, indicating at least moderate transcellular permeability. CP-615,003 displayed polarized transport (BA > AB), with the asymmetry enhanced 6-fold in MDR1-expressing MDCK cells, confirming transport by human MDR1 P-glycoprotein. The positive control quinidine showed similar results, with a 6-fold enhancement in polarized efflux transport across MDR1-MDCK cells relative to the parental MDCK cells. In contrast, polarized transport of CP-900,725 was not readily apparent in MDCK or MDR1-MDCK cells, and the results were comparable to the negative control triprolidine, indicating that this metabolite of CP-615,003 is not a substrate for P-gp efflux.
|
Brain and CSF Distribution in Rats. Mean serum and brain tissue concentration-time profiles of CP-615,003 (40 mg/kg p.o.) and CP-900,725 (10 mg/kg s.c.) are displayed in Fig. 2, with the corresponding mean calculated unbound serum and measured CSF concentration-time profiles shown in Fig. 3. Pharmacokinetic parameters characterizing brain and CSF distribution are summarized in Table 2. The shapes of the time course of concentrations of CP-615,003 or CP-900,725 in all three CNS sampling sites (serum, brain, CSF) were generally similar. Inspection of Fig. 3 reveals impairment of CSF distribution of both compounds, with CSF concentrations being generally lower than unbound concentrations in serum.
|
|
|
The brain-to-serum AUC ratio of CP-615,003 was 0.162, and was approximately 13 times below what would be expected from the ratio of free fractions of the compound in rat serum (fu,serum = 0.0613) and brain tissue (fu,brain = 0.0301). Consistent with the observed profiles (Fig. 2), the CSF-to-serum AUC ratio of CP-615,003 (0.0102) was approximately 6 times lower than the serum free fraction, indicating impaired CSF distribution.
The brain-to-serum AUC ratio of CP-900,725 was 0.0824, and was approximately 11 times below what would be expected from the ratio of free fractions of the compound in rat serum (fu,serum = 0.0569) and brain tissue (fu,brain = 0.0630). The CSF-to-serum AUC ratio of CP-900,725 (0.0206) was approximately 3 times lower than the serum free fraction, indicating impaired CSF distribution.
CSF Pharmacokinetics of CP-615,003 in Dogs. The time course of mean CSF concentrations and mean unbound (calculated) plasma concentrations of CP-615,003 after intravenous administration (2 mg/kg) to beagle dogs is shown in Fig. 4, with the pharmacokinetic parameters summarized in Table 3. Maximum concentrations in CSF were achieved in 
10 min postdose. However, the mean concentrations in CSF were approximately an order of magnitude lower than the corresponding mean systemic unbound concentrations, indicating a readily apparent barrier to CSF penetration of CP-615,003 in the dog. CSF-to-serum AUC ratio was 8 times lower than the serum free fraction (fu,serum = 0.267), confirming impaired CSF distribution that was statistically significant (p < 0.01 for comparison of CSF-to-plasma AUC ratio against the unbound fraction).
|
|
Brain Distribution in mdr1 P-gp-Deficient Mice. After administration of a single 5-mg/kg dose of CP-615,003 or CP-900,725, the serum concentration-time profiles were essentially superimposable in mdr1a/1b (–/–) mice and their wild-type counterparts (Fig. 5, A and C). The brain concentrations of CP-615,003 in mdr1a/1b (–/–) mice were, however, clearly higher than in wild-type mice (Fig. 5B), indicating that mdr1 P-gp limits CNS penetration of CP-615,003 in mice. The brain-to-serum AUC ratio of CP-615,003 in mdr1a/1b (–/–) mice was approximately 7 times that in wild-type mice (Table 4). Brain concentrations and the brain-to-serum AUC ratios of CP-900,725 were comparable in mdr1a/1b (–/–) mice and their wild-type counterparts (Fig. 5D; Table 4), indicating that CNS penetration of this metabolite is not influenced by mdr1 P-gp.
|
|
Brain ECF Pharmacokinetics in Rats. The concentration-time profiles of CP-615,003 and CP-900,725 in brain ECF (calculated from the measured microdialysate concentrations and the respective probe recoveries of each compound estimated by in vivo retrodialysis) in three rats and the corresponding mean profiles of unbound serum concentrations are shown in Fig. 6, A (CP-615,003) and B (CP-900,725), with the pharmacokinetic parameters summarized in Table 5. ECF concentrations of CP-615,003 were over an order of magnitude lower than unbound systemic concentrations, indicating impaired CNS distribution. The disposition of CP-615,003 from brain ECF paralleled disposition from systemic circulation and the ECF-to-serum AUC ratio (0.00143 ± 0.00035) was approximately 43 times below the unbound fraction in rat serum (fu,serum = 0.0613), indicating a substantial impairment of distribution to brain ECF that was statistically significant (p < 0.01 for comparison of ECF-to-serum AUC ratio against the unbound fraction). The ECF-to-serum AUC ratio of CP-615,003 is approximately 7 times below the CSF-to-serum AUC ratio determined in the CSF distribution study. For the metabolite CP-900,725, impairment of brain ECF penetration was less severe but nevertheless reached statistical significance (p < 0.05 for comparison of ECF-to-serum AUC ratio against the unbound fraction), with the ECF-to-serum ratio (0.0145 ± 0.0060) being approximately 4 times below the unbound fraction in rat serum (fu,serum = 0.0569) and comparable to the CSF-to-serum ratio determined in the CSF distribution study.
|
|
Human CSF Distribution and Occupancy Projections. Multiple oral doses of 100 mg of CP-615,003 b.i.d. were safe and well tolerated by healthy adult volunteers in the clinical CSF pharmacokinetic study. The steady-state plasma concentration-time profiles of CP-615,003 and its metabolite CP-900,725 over the dosing interval after oral administration of 100 mg of CP-615,003 b.i.d. are shown in Fig. 7A and the pharmacokinetic parameters are summarized in Table 6. Median Tmax values of CP-615,003 and CP-900,725 were 3 h and 4 h, respectively. In all six subjects, CSF concentrations of both CP-615,003 and CP-900,725 were measurable at 4 h postdose. The mean steady-state CSF-to-plasma concentration ratios of CP-615,003 (0.024 ± 0.006) and CP-900,725 (0.0166 ± 0.0047) were approximately 10 times and 9 times below their respective mean ex vivo unbound fractions in simultaneously obtained plasma samples (CP-615,003 fu,plasma of 0.233 ± 0.031; CP-900,725 fu,plasma of 0.148 ± 0.029), indicating impaired human CSF distribution. For both the parent drug and the metabolite, the impairment to CSF distribution was observed in each of the six subjects, with the CSF/Cu ratios for both compounds being significantly different from unity (p < 0.001).
|
|
The projected time course of target receptor occupancy over the steady-state dosing interval is shown in Fig. 7B under the assumptions of CSF concentrations of CP-615,003 being equivalent to receptor-available CNS concentrations, or of being approximately 7-fold higher than brain ECF concentrations (as in the rat). CSF concentrations of CP-900,725 were assumed to reflect brain ECF concentrations for purposes of these projections since the ECF/serum and CSF/serum ratios of CP-900,725 were similar in the rat. When a CSF/ECF ratio of 7 is assumed for CP-615,003, the projected occupancy over the dosing interval ranged from 44% to 73%, whereas the assumption of CSF concentrations being equivalent to ECF concentrations yielded an occupancy range of 75% to 93% over the dosing interval.
|
Discussion |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
; Shang et al., 2006).
In vitro studies identified CP-615,003 to be efficiently transported by MDR1 P-gp, with efflux comparable to the strong P-gp substrate quinidine, whereas its metabolite CP-900,725 [a product of oxidative deamination by monoamine oxidase(s) and/or cytochrome P450 enzymes followed by reduction of the putative aldehyde intermediate to the alcohol via alcohol dehydrogenase] was not a substrate for this transporter. Because CP-900,725 is not a circulating metabolite in rats after CP-615,003 administration (Shaffer et al., 2005), subsequent rodent studies on the CNS distribution of this active human metabolite were performed by direct administration. Consistent with the in vitro findings, mdr1 P-gp significantly limited the distribution of CP-615,003 (but not CP-900,725) in mice, with a 7-fold enhancement of brain distribution in mdr1a/1b (–/–) mice.
Given the increasing recognition of the complexity of anatomical and molecular architecture of the blood-brain and blood-CSF barriers (Rao et al., 1999; Lee et al., 2001; de Lange and Danhof, 2002; de Lange, 2004; Lee and Bendayan, 2004; Shen et al., 2004), we characterized the distribution of CP-615,003 and CP-900,725 to CSF, brain tissue, and brain ECF in a battery of CNS distribution studies in rats. In addition, we explored the sensitivity of projected values of human receptor occupancy from steady-state clinical CSF pharmacokinetic data, to assumptions regarding the magnitude of the CSF-ECF gradient in humans.
CNS penetration of CP-615,003 was severely impaired in rats at all three levels investigated (whole brain tissue, CSF, and striatal ECF), albeit to varying extents (Table 7). Application of the approach described by Maurer and colleagues to interpretation of brain-to-serum distribution ratios of drugs (Kalvass and Maurer, 2002; Maurer et al., 2005) yielded valuable insight into the extent of impairment of brain distribution of CP-615,003 and CP-900,725. The observed brain-to-serum ratio of CP-615,003 in rats was 13-fold below that expected from its relative binding to serum and brain tissue (fu,serum/fu,brain ratio), indicating a significant impairment to CNS penetration, consistent with the observed efflux by P-gp in vitro and in mice in vivo. The CSF-to-serum ratio of CP-615,003 was 6-fold below its serum unbound fraction. Assessment of pharmacokinetics in striatal ECF (a region of specific pharmacologic interest for this compound) indicated that the ECF-to-serum ratio is 43 times lower than the serum unbound fraction of CP-615,003. When coupled with the CSF-to-serum distribution ratio, it can be inferred that CP-615,003 displays a 7-fold CSF-to-ECF gradient in rats. An important implication of this finding is that CSF exposures substantially overestimate receptor-available unbound drug concentrations of CP-615,003 in rat brain ECF. In the endothelial cells of the blood-brain barrier, P-gp expression on the "blood side" limits the distribution of its substrates into brain tissue, consistent with the severe impairment of distribution of CP-615,003 into the ECF space. In contrast, P-gp is expressed on the "CSF side" in the epithelial cells of the choroid plexus and has been described to result in efflux transport of its substrates into (rather than out of) the CSF (Rao et al., 1999; Lee et al., 2001; de Lange and Danhof, 2002; Lee and Bendayan, 2004). Therefore, the observed CSF-to-ECF ratio of greater than unity may be explained by a combination of a strong net sink action at the brain parenchyma due to very efficient efflux transport by P-gp at the blood-brain barrier and the additional contribution of efflux into the CSF by P-gp in the choroid plexus (de Lange, 2004; Shen et al., 2004). These considerations of the anatomic and molecular architecture of the blood-brain and blood-CSF barriers, coupled with the observation of a CSF-ECF exposure gradient for the strong P-gp substrate CP-615,003 but not its active metabolite CP-900,725 (the CNS distribution of which is not limited by P-gp), point to P-gp efflux as the likely mechanism underlying the observed intersite differences in CNS distribution of CP-615,003. Additional mechanistic studies (e.g., CSF and ECF pharmacokinetic assessments in rats after selective pharmacologic inhibition of P-gp) will be needed to provide a confirmatory test of this hypothesis. Furthermore, considering the heterogeneity in factors (physicochemical versus biologic) that determine CNS distribution of drugs, studies similar to those reported in this investigation will need to be performed on other well established P-gp substrates to provide understanding of the relationship of compound-specific characteristics (physicochemical properties, permeability characteristics, transport efficiency, etc.) to intersite differences in CNS distribution of P-gp substrates.
|
The appearance and disposition of CP-615,003 from rat brain tissue and CSF paralleled its systemic pharmacokinetics (Figs. 2A and 3A), and the half-life in rat brain ECF was similar to the serum half-life (Table 5; Fig. 6A). In dogs, although a distributional impairment was readily apparent, it is of interest that CP-615,003 appeared relatively rapidly in CSF (Tmax 10 min after intravenous administration) and disposition half-lives from CSF and systemic circulation were similar (Fig. 4; Table 3). These characteristics across species suggest equally rapid equilibration of drug between CSF and systemic circulation, and between ECF and systemic circulation, which is entirely consistent with similar observations for several other drugs displaying a CSF-to-ECF gradient greater than unity in the compilation by Shen et al. (2004).
Distribution of the metabolite CP-900,725 into the rat CNS was also limited to varying extents, depending on the region of focus (11-fold at the level of whole brain tissue, 3-fold at the level of CSF, and 4-fold at the level of striatal ECF; Table 7). Although the mechanisms underlying the limited CNS distribution of this metabolite are not known, P-gp-mediated transport can be ruled out, as discussed earlier. It is possible that the limitation to CNS penetration of this alcohol metabolite is related to its hydrophilicity (log P = 1.0) relative to the parent drug (log P = 2.5), resulting in a diffusibility index of –0.2 in contrast to the parent drug, which has a diffusibility index of 1.2. A diffusibility index less than zero has been shown to be associated with impaired CSF penetration (Shen et al., 2004). However, the possibility of non-P-gp transport mechanism(s) for this active metabolite at the blood-brain and/or blood-CSF barriers cannot be ruled out and will require additional mechanistic studies using in vitro and/or in vivo models of CNS drug transport. Interestingly, the inferred CSF-to-ECF exposure ratio for CP-900,725 was close to unity, indicating that from a practical standpoint, CSF exposures of this metabolite may represent a reliable surrogate of brain ECF exposures.
As in preclinical species, the CSF distribution of both CP-615,003 and CP-900,725 was impaired in humans, indicating that the preclinical CSF distribution studies provided a reliable indication of the possibility of impaired CSF distribution in humans, consistent with previously reported findings on a wide range of compounds (Shen et al., 2004). The ability to qualitatively pick up impaired CNS penetration of CP-615,003 at the level of CSF across species (rat, dog, human; Table 7) despite a CSF-to-ECF gradient greater than unity indicates that CSF exposure, although not an ideal surrogate of ECF exposure (de Lange and Danhof, 2002; Shen et al., 2004), is nevertheless a practically useful surrogate to permit identification of impaired CNS distribution of P-gp substrates. This conclusion is also supported by findings from other preclinical (Ohe et al., 2003; Doran et al., 2005; Liu et al., 2006) and clinical (Ochs et al., 1980; Khaliq et al., 2000; Zhou et al., 2000) studies of the CSF distribution of P-gp substrates that collectively indicate that CSF exposures of drugs whose CNS penetration is limited by P-gp-mediated efflux at the blood-brain barrier are consistently lower than unbound systemic exposures. This is consistent with the theory that CSF is very much influenced by the events at the blood-brain barrier, whereas the ECF is much less affected by events at the blood-CSF barrier and CSF dynamics (Shen et al., 2004).
The 7-fold CSF-to-ECF gradient observed for CP-615,003 has important implications for quantitative interpretation of human CSF concentrations of CP-615,003 (and potentially other P-gp substrates with properties similar to those of this compound). This is particularly important when direct imaging-based approaches for receptor occupancy measurement are not feasible because of the unavailability of validated imaging ligands for clinical use and/or cost considerations. In the case of CP-615,003, failure to consider the possibility of a CSF-to-ECF gradient resulted in the projection of >75% receptor occupancy over the steady-state dosing interval based on CSF distribution at 100 mg b.i.d. Although the human CSF-to-ECF gradient is not known, the assumption of a CSF/ECF ratio of 7 (as in the rat) resulted in projected occupancies that were <75% over the entire dosing interval (i.e., receptor saturation could not be inferred), indicating that occupancy projections from CSF pharmacokinetic data were sensitive to assumptions regarding the magnitude of the CSF-to-ECF gradient in humans. These observations preclude definitive conclusions regarding the level of receptor occupancy by CP-615,003 in the absence of independent evaluation using direct imaging-based approaches.
With increased understanding of the importance of minimizing P-gp efflux potential during the discovery of neuropsychopharmacologic agents (Mahar Doan et al., 2002; Chen et al., 2003b; Doran et al., 2005), it is unlikely that compounds with a similar CNS distribution profile would be advanced into clinical development for indications requiring CNS distribution for activity. The concepts illustrated in this investigation are nevertheless relevant for modern drug discovery and development as they may be useful for interpreting exposure-response relationships for CNS safety pharmacology endpoints, and for quantitative understanding of the associated therapeutic index of P-gp substrates, across therapeutic areas.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
ABBREVIATIONS: CNS, central nervous system; Kp, partition ratio; MDR1, human multidrug resistance protein 1; mdr1, mouse multidrug resistance protein 1; MDCK, Madin-Darby canine kidney II; A, apical; B, basolateral; ER, efflux ratio; P-gp, P-glycoprotein; CSF, cerebrospinal fluid; ECF, extracellular fluid; LC/MS/MS, liquid chromatography-tandem mass spectrometry; MeOH, methanol; Rt, retention time; FRO, fractional receptor occupancy.
Address correspondence to: Dr. Karthik Venkatakrishnan, Clinical Pharmacology, MS8260-2626, Pfizer Global Research and Development, Eastern Point Rd., Groton, CT 06340. E-mail: karthik.venkatakrishnan{at}pfizer.com
|
References |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  J. Shen, A. M. Carcaboso, K. E. Hubbard, M. Tagen, H. G. Wynn, J. C. Panetta, C. M. Waters, M. A. Elmeliegy, and C. F. Stewart Compartment-Specific Roles of ATP-Binding Cassette Transporters Define Differential Topotecan Distribution in Brain Parenchyma and Cerebrospinal Fluid Cancer Res., July 15, 2009; 69(14): 5885 - 5892. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. L. Shaffer, M. Gunduz, A. D. Vaz, K. Venkatakrishnan, and A. H. Burstein Metabolism and Disposition of a {gamma}-Aminobutyric Acid Type A Receptor Partial Agonist in Humans Drug Metab. Dispos., April 1, 2008; 36(4): 655 - 662. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. W. Angus, J. A. Baker, R. Mason, and I. J. Martin The Potential Influence of CO2, as an Agent for Euthanasia, on the Pharmacokinetics of Basic Compounds in Rodents Drug Metab. Dispos., February 1, 2008; 36(2): 375 - 379. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
 |
 |
|
 |
 |
 |
