Abstract
Factors determining the pharmacokinetics of 2-chloro-N-(4-chloro-3-(pyridine-2-yl)phenyl)-4-(methylsulfonyl)benzamide (GDC-0449) were investigated using preclinical studies and physiologically based pharmacokinetic (PBPK) modeling. Multiple-dose studies where dogs were given twice-daily oral doses of either 7.5 or 25 mg/kg GDC-0449 showed less than dose-proportional increases in exposure on day 1. At steady state, exposures were comparable between the two dose groups. Oral administration of activated charcoal to dogs receiving oral or intravenous GDC-0449 (25 mg) showed a more rapid decrease in plasma concentrations, suggesting that the concentration gradient driving intestinal membrane permeation was reversible. The biliary clearance of GDC-0449 in dogs was low (0.04 ml/min/kg) and did not account for the majority of the estimated systemic clearance (∼19% of systemic clearance). Likewise, in vitro studies using sandwich-cultured human hepatocytes showed negligible biliary excretion. The effect of particle size on oral absorption was shown in a single-dose study where 150 mg of GDC-0449 of two particle sizes was administered. An oral PBPK model was used to investigate mechanisms determining the oral pharmacokinetics of GDC-0449. The overall oral absorption of GDC-0449 appears to depend on the interplay between the dissolution and intestinal membrane permeation processes. A unique feature of GDC-0449 distinguishing it from other Biopharmaceutical Classification System II compounds was that incorporation of the effects of solubility rate-limited absorption and nonsink permeation on the intestinal membrane permeation process was necessary to describe its pharmacokinetic behavior.
The hedgehog (Hh) signaling pathway regulates proliferation and differentiation during embryogenesis. Hh ligands bind to Patched (PTCH1), a transmembrane protein on target cells. In the absence of Hh, the role of PTCH1 is to inhibit the activity of Smoothened (SMO), a seven-transmembrane protein that serves as the signaling component of the pathway. Binding of Hh proteins to PTCH1 relieves this inhibition and initiates activation of SMO. The increase in SMO activity causes increases in activated forms of Gli, transcriptional factors that serve to regulate the expression of Hh target genes. Activation of the Hh pathway has been implicated in a number of cancers (Scales and de Sauvage, 2009). Mutations in the Hh receptor components, PTCH1 or SMO, result in constitutive pathway activation and have been identified in basal cell carcinoma (Hahn et al., 1996; Johnson et al., 1996) and medulloblastoma (Pietsch et al., 1997; Raffel et al., 1997; Vorechovský et al., 1997). It has also been observed that aberrant Hh ligand production can contribute to the growth of other tumor types, such as colorectal and pancreatic cancer (Yauch et al., 2008), prostate cancer (Fan et al., 2004), and B cell lymphoma (Dierks et al., 2007), through paracrine activation of the Hh pathway. Paracrine signaling typically involves ligand expressed on the cancer cells signaling adjacent stromal components, in the case of solid tumors, or signaling from stromally produced Hh ligand to cancer cells, in the case of hematopoietic cancers. The growing scientific data associating Hh signaling pathway activation with certain types of cancers have made this pathway an attractive target for the development of selective small molecule inhibitors.
2-Chloro-N-(4-chloro-3-(pyridin-2-yl)phenyl)-4-(methylsulfonyl)benzamide (GDC-0449) (Fig. 1) is a small molecule inhibitor of the Hh signaling pathway currently being developed at Genentech, Inc. (South San Francisco, CA). It inhibits Hh signaling with IC50s of 13 and 2.8 nM in Hh-responsive cell lines derived from mouse (10T1/2) and human embryonic palatal mesenchyme cells, respectively. Hh signaling is blocked by GDC-0449 through binding to and inhibiting SMO (Yauch et al., 2009). We previously described the preclinical absorption, distribution, metabolism, and excretion properties of GDC-0449 (Wong et al., 2009). The compound exhibited low plasma clearance in mouse, rat, and dog and moderate clearance in the monkey. These in vivo observations were consistent with in vitro metabolic stability studies performed using hepatocytes. More recently, the oral pharmacokinetics of GDC-0449 in humans has been described (Von Hoff et al., 2009; Ding et al., 2010). The clinical pharmacokinetics was characterized by remarkably high plasma exposures suggestive of a low systemic clearance. Of the preclinical species tested, the dog exhibits characteristics most similar to humans, having the lowest plasma clearance and longest t1/2 in vivo, and GDC-0449 shows virtually no turnover after a 3-h incubation with dog and human hepatocytes (Wong et al., 2009). Detailed studies aimed at understanding the disposition of new chemical entities in humans are often difficult to perform because of obvious ethical restrictions. Here, we describe the results of the preclinical studies and physiologically based pharmacokinetic (PBPK) modeling using the dog to provide insight into understanding the pharmacokinetic characteristics of GDC-0449.
Materials and Methods
In Vivo Pharmacokinetic Studies in Dogs.
Bile duct-cannulated dog study.
This study was performed to assess the role of biliary clearance on the in vivo disposition of GDC-0449 in dogs. At study initiation, dogs used weighed from 7.9 to 9.9 kg. Intact (n = 4) and bile duct-cannulated (n = 4) male beagle dogs (Wuxi Pharmatech Co. Ltd., Shanghai, China) were given a single 25-mg i.v. bolus dose of GDC-0449 (Genentech, Inc.) in 80% v/v polyethylene glycol 400 in water. Animals were not fasted before dosing. Blood samples (approximately 0.5 ml/sample) were collected from a peripheral vessel into tubes containing potassium EDTA (K2EDTA) at the following time points: predose and 0.033, 0.083, 0.25, 0.5, 1, 2, 4, 8, 12, 24, 48, 72, 96, 120, 144, and 168 h postdose. Samples were centrifuged within 1 h of collection. Plasma was collected and kept frozen on dry ice before storage at approximately −70°C. Bile was collected from bile duct-cannulated dogs at the following time intervals: predose, between 0 and 8 h postdose, between 8 and 24 h postdose, between 24 and 48 h postdose, and between 48 and 72 h postdose. Bile was collected into containers cooled by wet ice, and volumes for different intervals were measured and recorded. All the bile samples were stored at approximately −80°C. GDC-0449 in plasma and bile was measured using a liquid chromatography/tandem mass spectrometric method (LC/MS/MS) (Wong et al., 2009).
Activated charcoal studies.
The purpose of this study was to examine the effect of oral administration of activated charcoal on the oral and intravenous pharmacokinetics of GDC-0449 in dogs. At study initiation, dogs used weighed from 8.3 to 11.8 kg. Two groups of male beagle dogs (n = 4/group) (Covance Research Products, Princeton, NJ) received a single oral dose of GDC-0449 as a 25-mg capsule. A third group of dogs (n = 4) received a single 25-mg i.v. dose of GDC-0449 in 80% polyethylene glycol 400 via a cephalic vein. One group of orally dosed animals and all the intravenously dosed animals received a 2-g/kg dose of activated charcoal (Sigma-Aldrich, St. Louis, MO) as a slurry via oral gavage at approximately 24, 27, 31, 36, 48, 60, 72, 84, and 96 h after GDC-0449 administration. All the animals were fasted overnight before dosing until approximately 4 h after the GDC-0449 dose. At approximately 30 min before GDC-0449 administration, all the animals received a single intramuscular injection of 0.024 ml/kg pentagastrin (6 μg/kg) in the left thigh. Blood samples (approximately 3 ml/sample) were collected from the jugular vein of each animal at the following time points: predose and 0.083, 0.25, 0.5, 1, 2, 4, 8, 12, 24, 72, 96, 120, and 168 h postdose. All the blood samples were collected into tubes containing K2EDTA and then chilled on ice until centrifugation. Samples were centrifuged within 1 h of collection. Plasma was collected and kept frozen on dry ice before storage at approximately −70°C. Concentrations of GDC-0449 in plasma were quantitated by LC/MS/MS (Wong et al., 2009).
Multiple-dose pharmacokinetics of GDC-0449.
Multiple-dose oral pharmacokinetics of GDC-0449 was obtained from toxicology studies with GDC-0449. Male beagle dogs (Covance Research Products) were given twice-daily oral doses of 7.5 mg/kg (15 mg/kg/day) or 25 mg/kg (50 mg/kg/day) of GDC-0449, respectively, for 91 days. The two daily doses were administered 6 h apart on each dosing day. At the initiation of the study, dogs were at least 8 months old and weighed 7.0 to 10.7 kg. Blood samples were collected for a 24-h period starting after the administration of the first dose on days 1, 44, and 90 at predose (30 min before the first daily dose) and at 1, 3, 5.75, 6.5, 10, and 24 h postdose. Blood samples (approximately 1 ml each) were collected from a jugular vein into tubes containing K2EDTA and then chilled until centrifugation. Samples were centrifuged within 1 h of collection. Plasma was collected and stored frozen at approximately −60 to −80°C. Concentrations of GDC-0449 in plasma were quantitated using a validated LC/MS/MS method similar to one described previously (Wong et al., 2009). Results from days 1 and 44 are presented in this article because steady state was achieved by day 44.
Oral formulation comparison study.
The purpose of this study was to evaluate the impact of alterations in compound particle size on oral exposure of GDC-0449. At the initiation of this study, dogs weighed from 12.7 to 15.6 kg. Male beagle dogs (Covance Research Products) were given an oral 150-mg dose of GDC-0449 as a single 150-mg capsule (n = 6; smaller particle size; d(50) 21 μm) or as 25- and 125-mg capsules (n = 6; larger particle size; d(50) 120 μm). Approximately 30 min before GDC-0449 administration, animals in both groups received a single 6-μg/kg intramuscular dose of pentagastrin (Sigma-Aldrich) at 0.024 ml/kg. Animals were fasted overnight before dosing through approximately 4 h postdose. Blood samples (approximately 3 ml/sample) were collected from the jugular vein of each animal at the following time points: predose and 0.083, 0.25, 0.5, 1, 2, 4, 8, 12, 24, 72, 96, 120, and 168 h postdose. All the blood samples were collected into tubes containing K2EDTA and then chilled until centrifugation. Samples were centrifuged within 1 h of collection. Plasma was collected and kept frozen on dry ice before storage at approximately −70°C. Concentrations of GDC-0449 in plasma were quantitated by LC/MS/MS (Wong et al., 2009).
Pharmacokinetic Data Analysis.
All the pharmacokinetic parameters were calculated by noncompartmental methods as described by Gilbaldi and Perrier (1982). Pharmacokinetic parameters (aside from Tmax) are reported as the mean ± S.D. Tmax is presented as the median along with the observed range in parentheses.
PBPK Modeling.
A nine-intestinal segment oral absorption PBPK model was constructed in ModelMaker version 4.0 (Cherwell Scientific Ltd., Oxford, UK) based on a modified version of the advanced compartmental absorption and transit model (Agoram et al., 2001). The configuration of the described model is shown in Fig. 2. The first segment represents the stomach, followed by seven small intestine (SI) segments and the colon segment. Two compartments were assigned for each of the nine segments representing the gastrointestinal (GI) tract and were designated to contain either solid or dissolved drug. Rate constants describing gastric emptying (KS), SI compartmental transit (KT), and colon emptying (KC) were set at 4/h, 3.85/h, and 0.0833/h, respectively, to reflect the GI transit times presented for the Beagle Model in GastroPlus (Simulations Plus, Inc., Lancaster, CA). Specific pH for the GI compartments were set at pH values presented for the Beagle Model in GastroPlus as follows: SI1, pH 6.2; SI2, pH 6.2; SI3, pH 6.2; SI4, pH 6.4; SI5, pH 6.6; SI6, pH 6.68; SI7, pH 6.75; colon, pH 6.45. The exception was the stomach compartment, which was set to be similar to human gastric pH (pH 1.2) because our studies involved either pentagastrin-pretreated dogs or fed dogs, which have stomach pH similar to humans (Sagawa et al., 2009). Solubility of the stomach compartment was set at 0.49 mg/ml, the solubility of GDC-0449 in simulated gastric fluid at the stomach pH. The volume in the stomach compartment was set at 450 ml. Aside from the stomach, the volume of the lumen for each GI compartment was calculated using the following equation: Vlumen = πLR2, where L is the length of the intestinal section and R is the radius. A radius of 0.5 cm was used for all the SI compartments, and 2 cm was used for the colon. The following lengths (in centimeters) were used for each GI compartment as per the Beagle Model in GastroPlus: SI1, 44.76; SI2, 32.98; SI3, 24.3; SI4, 17.9; SI5, 13.19; SI6, 9.72; SI7, 7.16; colon, 6.25.
Dissolution process.
Dissolution rate of solid drug in the oral absorption PBPK model was governed by the following equation based on the Noyes-Whitney equation (eq. 1): where Xsolid is the amount of undissolved GDC-0449 in the solid compartments, D is the GDC-0449 diffusion coefficient (default = 10−4 cm2/min), ρ is the drug particle density (GDC-0449 = 1.34 g/cm3), r is the particle radius (adjusted based on GDC-0449 form dosed: 20 μm for the multiple-dose pharmacokinetics study and 21 or 120 μm for the oral formulation comparison study), h is the diffusion layer thickness (if r < 30 μm, h = r; if r > 30 μm, h = 30 μm; Sugano et al., 2007), Csolubility_IF is the measured solubility in simulated intestinal fluid (FASSIF; GDC-0449 = 3 μg/ml at pH from 6.0–6.8), and CGItract is the concentration of dissolved GDC-0449 in the GI tract segment. Based on the equation above, dissolution of GDC-0449 was assumed to be driven by the difference between the FASSIF solubility and the concentration of dissolved GDC-0449 in the GI segment.
Intestinal membrane permeation process.
The rate of membrane permeation of dissolved GDC-0449 from the GI tract into the systemic circulation in the PBPK model was governed by eq. 2: where XsolutionGItract is the dissolved drug in the GI tract segment, Papp is the permeability coefficient of the drug [60.1 × 10−6 cm/s based on experimental data from Caco-2 cells; additional note: efflux ratio in Caco-2 cells is 0.79 (Genentech, Inc.)], A is the surface total area available for absorption; fmono is the fraction of GDC-0449 that is not in micelles (equal to solubility in buffer divided by solubility in FASSIF), and Cunbound_sys is the unbound concentration of GDC-0449 in the systemic circulation. An unbound fraction of 1.50% (determined ex vivo by equilibrium dialysis in 12 dogs) was used to determine unbound concentrations in simulations. For the oral formulation comparison study, the actual measured mean unbound fraction from animals from that study was used (2.77%).
If fmono × CGItract > CsolBuffer (solubility in buffer), then the rate of permeation would be governed by the following eq. 3:
Only the free monomer (fmono × CGItract; GDC-0449 not associated with micelles in FASSIF) was assumed to be available for diffusion across the intestinal membrane. The upper limit of free monomer concentration was set as the GDC-0449 aqueous solubility associated with the pH of the intestinal segment of interest. Aqueous solubility of GDC-0449 was determined twice, giving values of 0.1 and 0.4 μg/ml from pH 6.5 to 7.0 (Genentech, Inc.). Both estimates of aqueous solubility produced pharmacokinetic profiles with similar pharmacokinetic characteristics (data not shown). GDC-0449 aqueous solubility was set at 0.4 μg/ml for all the simulations presented because this estimate was the most conservative for showing nonsink permeation conditions. The total surface area available for absorption (A) was calculated as (2πRL) × total amplification factor. Mammals have been shown to have a total surface area amplification factor as a result of villi and microvilli of approximately 450 (Ferraris et al., 1989). An additional amplification factor of 3 was incorporated because dog is believed to have a higher permeability than humans (Parrott et al., 2009). Absorption was assumed to only occur in the SI.
For simulations performed under sink conditions only (SINK), equations describing membrane permeation (eqs. 2 and 3) were modified as follows (see eqs. 4 and 5): If fmono × CGItract > CsolBuffer
For simulations where the effect of solubility rate-limited absorption (SRLA) on the permeation process was removed from the PBPK model, permeation was governed solely by eq. 2 (or eq. 4 where SINK conditions also apply), even under conditions where fmono × CGItract > CsolBuffer.
The maximal permeation extraction ratio (MPER), a measure of the maximum extraction of compound from the intestinal lumen into the systemic circulation under nonsink conditions, is defined as follows (eg. 6):
Based on the equation described above, under sink conditions, MPER would range from 0.9 to 1. Nonsink conditions for permeation would occur at MPER of <0.9.
Systemic circulation.
The elimination rate constant (ke) from the systemic circulation in the PBPK model was set at 0.00997/h based on a t1/2 of 69.5 h estimated from the bile duct-cannulated dog study (described above) during the first 72 h postdose. Because bile was collected during this period, this t1/2 was assumed to be entirely the result of systemic elimination of the compound with no influence from enterohepatic recycled compound. This t1/2 is virtually identical to the t1/2 estimated for bile duct-cannulated dogs using plasma-concentration data from the full duration of the study (Table 1).
In Vitro Study Using Sandwich-Cultured Human Hepatocytes.
B-CLEAR-HU (Qualyst, Inc., Research Triangle Park, NC) sandwich-cultured fresh primary human hepatocytes six-well plates were used to investigate the hepatobiliary disposition of GDC-0449. In brief, the experiment involved preincubating sandwich-cultured hepatocytes in Hanks' buffered salt solution with (+) or without (−) Ca2+ for approximately 10 min. Ca2+ free buffer has been shown to disrupt the tight junction between the canalicular lumen and the extracellular space, causing substrate that is excreted into bile canalicular networks to diffuse back into the incubation media (Liu et al., 1999). At the end of the preincubation period, GDC-0449 (5 or 50 μM) in Hanks' buffered salt solution (+) Ca2+ was added and incubated for approximately 20 min. At the end of the incubation, an aliquot of incubation buffer was taken for measurement of GDC-0449 by LC/MS/MS. The remaining buffer was aspirated; cells were washed three times with ice-cold (+) Ca2+ buffer; and 1 ml of acetonitrile containing 0.188 μM deuterated internal standard was added to each well and mixed for at least 20 min. The resulting lysate was collected, and protein (Lowry et al., 1951) and GDC-0449 concentrations were assessed. GDC-0449 uptake was normalized by protein concentrations. Incubations were performed in three donors and in triplicate. The biliary excretion index (BEI) and in vitro biliary clearance (CLbiliary) were calculated as described by the following eqs. 7 and 8 (Liu et al., 1999): where Uptake(+)Ca2+ and Uptake(−)Ca2+ are the cumulative uptake of GDC-0449 over the 20-min incubation period for hepatocyte cultures preincubated in buffer with and without Ca2+, respectively. AUCin vitro is the area under the concentration-time curve of the sandwich culture incubations calculated by multiplying the average concentration of GDC-0449 at the start and end of the incubation by the duration of the incubation (i.e., 20 min). Taurocholate and digoxin were run as positive controls (data not shown).
Results
Bile Duct-Cannulated Dog Study.
Figure 3 is concentration-time profile of GDC-0449 after an intravenous dose of 25 mg administered to intact and bile duct-cannulated dogs. The estimated pharmacokinetic parameters are presented in Table 1. The mean plasma clearance (CL) of GDC-0449 in both intact and bile duct-cannulated dogs was very low, approximately 0.4 and 0.6% of hepatic blood flow, respectively. The mean volume of distribution at steady state (Vss) in intact and bile duct-cannulated dogs was approximately 1.6 and 1.9 times, respectively, of total body water volume (Davies and Morris, 1993). Half-life (t1/2) was long, approximately 80 h in intact animals and approximately 68 h in bile duct-cannulated dogs. The mean CLbiliary of GDC-0449 in bile duct-cannulated dogs was very low, approximately 0.04 ml/min/kg, and was approximately 19% of CL in bile duct-cannulated dogs.
Activated Charcoal Studies.
Figure 4 presents a plot of the mean plasma concentration-time profile for dogs that received a 25-mg i.v. or p.o. dose of GDC-0449, with or without administration of activated charcoal. GDC-0449 concentrations at the start and end of charcoal administration (24 and 96 h, respectively) and the area under the concentration-time profile up to 168 h postdose (AUC0–168) are presented in Table 2. The mean GDC-0449 concentration at 96 h postdose for dogs given 25 mg of GDC-0449 orally (0.137 μM) was approximately 21.0% of the concentration observed at the 24-h postdose time point (0.650 μM). In contrast, oral administration of activated charcoal, from 24 to 96 h postdose, to dogs given either an oral or intravenous dose of 25 mg of GDC-0449 resulted in mean GDC-0449 concentrations at 96 h (0.004 and 0.018 μM, respectively) that were approximately 0.5% of the concentrations observed at the 24-h postdose time point (0.847 and 4.49 μM, respectively). Oral administration of activated charcoal to dogs given a 25-mg oral dose of GDC-0449 resulted in a modest reduction of AUC0–168, causing a decrease by approximately 50% (see Table 2). This was because a significant portion of the AUC0–168 originates from the time before the administration of charcoal (0–24 h postdose).
Intravenous data from intact dogs from the bile duct-cannulated dog study (described above) are also presented in Table 2 and Fig. 4 for comparative purposes. Compared with the dogs given 25 mg i.v. of GDC-0449 with oral activated charcoal, both the mean GDC-0449 concentration at 96 h postdose and the AUC0–168 were higher in the intact dogs from the bile duct-cannulated dog study.
Multiple-Dose Pharmacokinetics of GDC-0449.
Estimates of oral pharmacokinetic parameters for GDC-0449 on days 1 and 44 in male dogs given either 7.5 mg/kg b.i.d. or 25 mg/kg b.i.d. are presented in Table 3. Corresponding mean plasma concentration-time profiles are presented in Fig. 5, A and B. Increases in oral exposure were less than dose-proportional on day 1. An approximate 3-fold increase in dose (7.5 mg/kg b.i.d. to 25 mg/kg b.i.d.) resulted in only an approximate 2-fold increase in area under the plasma concentration-time profile from time 0 to 24 h postdose (AUC0–24) and Cmax (Table 3). GDC-0449 plasma concentrations on day 1 showed no signs of decline by 24 h for both dose groups (Fig. 5A). Finally, an expected secondary increase in GDC-0449 plasma concentrations was observed on day 1 after the administration of the second daily dose at 6 h. In contrast to day 1, AUC0–24 and Cmax estimates were very similar for both dose groups on day 44 (Table 3). GDC-0449 plasma concentration-time profiles on day 44 were unusually flat, showing little increase in GDC-0449 plasma concentration after the administration of the second daily dose. Continuous dosing of 7.5 mg/kg b.i.d. for 44 days resulted in accumulation of GDC-0449 with Cmax and AUC0–24 estimates being approximately 5- and 6-fold higher, respectively, on day 44 compared with day 1. The extent of accumulation was less for the 25-mg/kg b.i.d. dose group, with Cmax and AUC0–24 being approximately 2- and 3-fold higher, respectively, on day 44 compared with day 1.
Predicted concentrations from simulations using the oral PBPK model show an approximately 2-fold increase in exposure with dose on day 1 (Fig. 5A). At steady state (Fig. 5B), this difference in exposure was almost nonexistent and is consistent with observed data. Figure 5, C and D, shows the simulated MPER for days 1 and 44. The simulations suggest nonsink conditions exist for the oral permeation of GDC-0449 in dogs. In particular, nonsink permeation conditions were more prominent at steady state (day 44), where the MPER is <0.5 throughout the entire 24-h postdose for both dose levels.
Additional simulations were performed to assess the mechanistic influence of SRLA and nonsink permeation on the less than dose-proportional increases in exposure observed for GDC-0449 in dogs. Modifications were made to the equations describing the intestinal membrane permeation process of the full PBPK model (as described under Materials and Methods) by removing SRLA conditions (NO SRLA), removing nonsink permeation conditions (SINK), and removing both nonsink permeation and SRLA conditions (SINK and NO SRLA). Little effect was observed for the 7.5-mg/kg b.i.d. dose on day 1 (Fig. 6A) and at steady state (day 44) (Fig. 7A). In contrast, for the 25-mg/kg b.i.d. dose, incorporation of SRLA into the PBPK model appeared necessary to better simulate the plasma profile on day 1 (Fig. 6B). Removal of SRLA from the model (“NO SRLA” and “SINK and NO SRLA” scenarios) resulted in a doubling of the average day 1 concentrations of GDC-0449, causing a larger deviation from the observed concentrations. Removal of either or both SRLA and nonsink membrane permeation conditions caused ∼2- to 3.5-fold increases in the simulated average GDC-0449 steady-state (day 44) concentrations (Fig. 7B). The full model incorporating both conditions appeared to provide the best simulation of the observed day 44 data.
Oral Formulation Comparison Study.
The mean GDC-0449 plasma concentration-time profile in dogs from the oral formulation comparison study is shown in Fig. 8 along with the simulated plasma concentration-time profiles generated using the oral PBPK model. The effect of particle size on the oral pharmacokinetics of GDC-0449 in male dogs is presented in Table 4. Estimates of AUC0–168 and Cmax were both approximately 3-fold higher when particle size was reduced for GDC-0449. The median Tmax was comparable between the two groups, 4.0 and 3.0 h. However, the Tmax appeared more variable for the dogs given GDC-0449 as larger particles. As seen in Fig. 8, the PBPK model was able to nicely capture the observed particle size effect.
A simulation study using the oral PBPK model was performed to better understand the impact of changing dose and dose regimen on the observed particle size effect. For this purpose, simulations were performed at total doses of 600 mg (600 mg once daily or 150 mg q.i.d.) (Fig. 9). The doses were chosen such that steady concentrations achieved in the simulation approached relevant concentrations that have been observed clinically. The effect of particle size on oral absorption was predicted to decrease with increasing dose comparing the particle size effect after a single 150-mg dose (Fig. 8) with the simulation of a single dose of 600 mg (Fig. 9A). The effect of administering the 600-mg total dose over four 150-mg doses (i.e., 150 mg q.i.d.) was predicted to increase the effect of particle size (Fig. 9, A and C). The effect of particle size on oral absorption was predicted to be minimal at steady state regardless of the dose regimen (Fig. 9, B and D).
Sandwich-Cultured Human Hepatocytes Studies.
No biliary excretion was observed for GDC-0449 (at 5 and 50 μM) from sandwich-cultured human hepatocytes. Accordingly, estimates of BEI and CLbiliary for GDC-0449 were negligible, suggesting that biliary elimination of GDC-0449 is either very low or absent.
Discussion
The oral absorption of drugs depends on the sequential processes of dissolution and intestinal membrane permeation (Fig. 10). The overall absorption characteristics are a result of the interaction between these two processes. The dissolution rate, described commonly by the Noyes-Whitney equation (eq. 1), is influenced by compound characteristics, such as solubility in intestinal fluid, and more controllable properties, such as particle size. The driving force for dissolution is the gradient between intestinal fluid solubility and compound concentration in the intestinal compartment of interest. After dissolution in the intestinal lumen, drug in solution exists as free molecules or is solubilized by bile micelles (Sugano et al., 2007). Solubilization in micelles can act to enhance intestinal membrane permeation by assisting the movement across the unstirred water layer adjacent to the intestinal membrane (Amidon et al., 1982). Because micelle-bound molecules are not available for membrane permeation, those carried across the unstirred water layer by micelles must be released before crossing the intestinal membrane. Thus, the driving force for membrane permeation is the concentration gradient between the free molecules in the intestinal lumen and the unbound molecules in the systemic circulation as described by eqs. 2 and 3. For most drugs, the free molecule concentration in the lumen is much higher than the unbound concentrations in the systemic circulation such that membrane permeation occurs under “sink” conditions (see eqs. 4 and 5). In these cases, the rate-limiting step for permeation is movement across the unstirred water layer (Sugano et al., 2007).
Inhibition of the Hh signaling pathway and its potential implications on the treatment of certain cancers has resulted in the synthesis of small molecule inhibitors of this pathway (Borzillo and Lippa, 2005; Scales and de Sauvage, 2009). We previously reported the preclinical absorption, distribution, metabolism, and excretion properties of GDC-0449, a small molecule inhibitor of the Hh signaling pathway currently in Phase II clinical trials (Wong et al., 2009). GDC-0449 is extremely stable in dog and human hepatocytes, displaying minimal turnover in 3-h hepatocyte incubations. The systemic clearance of GDC-0449 is exceptionally low in dogs, ∼1% of hepatic blood flow (Davies and Morris, 1993), and this was reflected in the long terminal half-life. The systemic clearance in intact and bile duct-cannulated dogs in the current study is similarly low, ∼0.4 and 0.6% of hepatic blood flow, respectively. Half-life in intact and bile duct-cannulated dogs was ∼80 and 68 h, respectively, and was similar (within 2-fold biological variability) to the previously reported t1/2 of ∼42 h (Wong et al., 2009). The low systemic clearance and the associated long t1/2 in dogs contribute to high steady-state concentrations of GDC-0449 observed in the current study. In terms of the Biopharmaceutical Classification System (BCS), GDC-0449 is considered a BCS class II compound (Dahan et al., 2009) based on its very low solubility and high Caco-2 permeability. The oral absorption of BCS class II compounds is usually limited by particle size effects or intestinal fluid solubility on the dissolution process. Membrane permeation is usually not considered rate-limiting for BCS class II compounds (Takano et al., 2008).
PBPK modeling is a useful tool to understand the interplay between different biological processes and their overall effect on the disposition of a particular molecule over time (Theil et al., 2003). The complexity of the oral absorption process has made the use of physiological modeling widespread. In particular, models based on the compartmental absorption transit model described by Yu and Amidon (1999) have been used extensively (Agoram et al., 2001; Parrott et al., 2009). In the current article, we use a compartmental absorption transit model that divides the GI tract into 9 segments where solid and dissolved material are represented as two separate compartments (18 compartments total; Fig. 2), and dissolution and permeation processes are consistent with the scenario described in Fig. 10. The described oral PBPK model was used to investigate the factors influencing the oral absorption of GDC-0449.
The pharmaceutical and biological characteristics of GDC-0449 make it uniquely distinct from most BCS class II compounds. As discussed, for BCS class II compounds, membrane permeation is usually not considered rate-limiting. The permeation process is often governed by a form of eq. 4, which implies that permeation occurs under sink conditions (Sugano et al., 2007). The permeation equation used in the current model (eq. 2) is based on Fick's Law of diffusion, where the driving force is the concentration gradient between the free molecules in the intestinal lumen (i.e., molecules not associated with micelles) and the unbound molecules in the systemic circulation. Equation 3 “caps” the free molecule concentration at the aqueous solubility of GDC-0449 and serves to describe a solubility rate-limited scenario for the membrane permeation process. A combination of low compound solubility along with high steady-state unbound concentrations act in unison to reduce the concentration gradient that serves as the driving force for membrane permeation. Based on the in vivo properties of GDC-0449, the use of these equations that allow for nonlinear membrane permeation is appropriate.
Our simulations of the multiple-dose studies show that inclusion of SRLA in the permeation process is necessary to describe the steady-state oral pharmacokinetics of GDC-0449 at the doses administered. More unique is the nonsink permeation characteristics that are also required to describe steady-state concentrations. At steady state, the unbound concentrations are such that significant nonsink permeation can exist, which is illustrated in our simulations presented in Fig. 5D. As mentioned above, the permeation process is driven by the concentration gradient between the free molecules in the intestinal lumen and the unbound molecules in the systemic circulation. The MPER is calculated based on the maximum possible concentration gradient at a given plasma concentration because the upper limit of free concentration in the intestinal lumen (i.e., aqueous solubility) is used in its calculation. Regardless, the MPER ratio is well under 0.9 during the entire 24-h interval at steady state, suggesting significant nonsink permeation working to decrease the movement of GDC-0449 across the intestinal membrane into the systemic circulation at both doses tested. Our studies with charcoal clearly show that the permeation across the intestinal membrane depends on a concentration gradient that can be reversed with the application of charcoal, making a nonsink permeation situation entirely plausible.
The oral formulation comparison study shows a particle size effect on the oral absorption of GDC-0449. This is consistent with previous observations for BCS class II compounds, such as danazol, griseofulvin, and aprepitant (Takano et al., 2008). Results of our simulation study using the described oral PBPK model suggest that increasing dose and/or applying multiple doses of GDC-0449 to steady state reduces the effect of particle size on oral absorption. The predicted reduction in the particle size effect can be attributed to a shift from a situation where absorption is rate limited by dissolution to a situation where absorption is rate-limited by solubility and/or the intestinal membrane permeation process where changes in particle size have minimal impact (Sugano et al., 2007).
Biliary clearance presents additional complexity in that compounds eliminated via the bile are reabsorbed and can have a significant impact on the shape of the plasma concentration-time profile. We investigated the contribution of biliary clearance to the disposition of GDC-0449 in vivo using bile duct-cannulated dogs and in vitro using sandwich-cultured human hepatocytes. Both studies suggested that biliary clearance does not appear to be a primary determinant of the disposition of GDC-0449 after oral administration in dogs or humans. In addition, biliary clearance was incorporated into the oral PBPK model and tested (data not shown). The inclusion of biliary clearance had minimal to no effect on simulations generated by the model; thus, biliary clearance was subsequently removed from the final model presented.
The clinical pharmacokinetics of GDC-0449 has also been recently described previously (Von Hoff et al., 2009). The compound showed high steady-state plasma concentrations with a median concentration of approximately 16.1 μM (interquartile range, 13.7–21.6) and an apparent lack of dose dependence from 150 to 540 mg. Data from healthy volunteers show that the t1/2 of GDC-0449 in humans is very long, approximately 12 days (Ding et al., 2010). Based on the pharmacokinetic characteristics, as well as systemic concentrations observed in humans thus far, the factors influencing oral absorption of GDC-0449 in dogs may play a role in humans.
Because most drugs are delivered via the oral route, an understanding of the factors influencing the rate and extent of oral absorption is important during the drug development process. The dissolution of GDC-0449 depends on properties such as particle size and compound solubility. Solubility also influences the permeation process; however; other biological characteristics, including metabolic intrinsic clearance (a controlling factor of in vivo unbound concentrations), also contribute. As described above, a unique feature of GDC-0449 distinguishing it from other BCS II compounds was that incorporation of both the effects of SRLA and nonsink permeation on the intestinal membrane permeation process was necessary to describe its pharmacokinetic behavior. Overall, the oral disposition of GDC-0449 depends on the interplay between the dissolution and permeation processes. Finally, PBPK modeling is an invaluable tool to understand complex dynamic processes such as oral absorption.
Acknowledgments.
We thank Dr. Malcolm Rowland for insightful discussions. We also thank Drs. Richard Graham, Bert Lum, Karin Jorga, Mingxin Qian, Stephen E. Gould, Jennifer Low, Minli Xie, Jin Jin, and Lichuan Liu for discussions on this topic. We thank April Yang and Dr. Jae Chang for assistance with protein binding measurements. All the authors are employees of Genentech, Inc. GDC-0449 was discovered by Genentech, Inc. under a collaboration agreement with Curis, Inc.
Footnotes
Article, publication date, and citation information can be found at http://dmd.aspetjournals.org.
doi:10.1124/dmd.110.032680.
-
ABBREVIATIONS:
- Hh
- hedgehog
- PTCH1
- Patched
- SMO
- Smoothened
- GDC-0449
- 2-chloro-N-(4-chloro-3-(pyridin-2-yl)phenyl)-4-(methylsulfonyl)benzamide
- PBPK
- physiologically based pharmacokinetic
- K2EDTA
- potassium EDTA
- LC/MS/MS
- liquid chromatography/tandem mass spectrometry
- SI
- small intestine
- GI
- gastrointestinal
- FASSIF
- fasted state simulated intestinal fluid
- SRLA
- solubility rate-limited absorption
- MPER
- maximum permeation extraction ratio
- BEI
- biliary excretion index
- CLbiliary
- biliary clearance
- AUC
- area under the concentration-time curve
- CL
- plasma clearance
- Vss
- volume of distribution at steady state
- AUC0–168
- area under the plasma concentration-time profile from time 0 to 168 h postdose
- AUC0–24
- area under the plasma concentration-time profile from time 0 to 24 h postdose
- BCS
- Biopharmaceutical Classification System.
- Received February 10, 2010.
- Accepted April 20, 2010.
- Copyright © 2010 by The American Society for Pharmacology and Experimental Therapeutics