Abstract
(R)-N-(3-(6-(4-(1,4-Dimethyl-3-oxopiperazin-2-yl)phenylamino)-4-methyl-5-oxo-4,5-dihydropyrazin-2-yl)-2-methylphenyl)-4,5,6,7-tetrahydrobenzo[b]thiophene-2-carboxamide (GDC-0834) is a potent and selective inhibitor of Bruton's tyrosine kinase (BTK), investigated as a potential treatment for rheumatoid arthritis. In vitro metabolite identification studies in hepatocytes revealed predominant formation of an inactive metabolite (M1) via amide hydrolysis in human. The formation of M1 appeared to be NADPH-independent in human liver microsomes. M1 was found in only minor to moderate quantities in plasma from preclinical species dosed with GDC-0834. Human clearance predictions using various methodologies resulted in estimates ranging from low to high. In addition, GDC-0834 exhibited low clearance in PXB chimeric mice with humanized liver. Uncertainty in human pharmacokinetic prediction and high interest in a BTK inhibitor for clinical evaluation prompted an investigational new drug strategy, in which GDC-0834 was rapidly advanced to a single-dose human clinical trial. GDC-0834 plasma concentrations in humans were below the limit of quantitation (<1 ng/ml) in most samples from the cohorts dosed orally at 35 and 105 mg. In contrast, substantial plasma concentrations of M1 were observed. In human plasma and urine, only M1 and its sequential metabolites were identified. The formation kinetics of M1 was evaluated in rat, dog, monkey, and human liver microsomes in the absence of NADPH. The maximum rate of M1 formation (Vmax) was substantially higher in human compared with that in other species. In contrast, the Michaelis-Menten constant (Km) was comparable among species. Intrinsic clearance (Vmax/Km) of GDC-0834 from M1 formation in human was 23- to 169-fold higher than observed in rat, dog, and monkey.
Introduction
Bruton's tyrosine kinase (BTK) is a member of the Tec family of nonreceptor tyrosine kinases, which are expressed in all cells of hematopoietic lineage except plasma cells, natural killer cells, and T lymphocytes (Satterthwaite and Witte, 2000; Brunner et al., 2005). BTK is activated by phosphatidylinositol 3-kinase-dependent plasma membrane recruitment and phosphorylation on tyrosine 551 by the Src family kinase Lyn. Once activated, BTK induces phospholipase Cγ2- and Ca2+-dependent signaling, leading to the activation of nuclear factor κB- and nuclear factor of activated T cell-dependent pathways (Niiro and Clark, 2002). In B cells, BTK is important for B cell antigen receptor-, CD40- and Toll-like receptor 4-mediated activation and proliferation (Khan et al., 1995). Furthermore, BTK plays a role in B cell antigen processing and presentation (Sharma et al., 2009). Of importance, BTK is also essential in Fcγ receptor-mediated inflammatory cytokine production (tumor necrosis factor-α, interleukin-1β, and interleukin-6) in monocytes/macrophages and therefore can contribute to immune complex-induced disease (Di Paolo et al., 2011).
The critical roles of BTK in the development, differentiation, and proliferation of B-lineage cells have been well documented (Pan, 2008). The recent discovery of selective inhibitors for BTK has provided convincing evidence that BTK is an attractive target for the treatment of rheumatoid arthritis (RA) and B cell-related diseases, such as lupus, lymphoma, and leukemia (Pan et al., 2007; Uckun et al., 2007; Di Paolo et al., 2011; Liu et al., 2011). However, there are currently no approved specific BTK inhibitors on the market. (R)-N-(3-(6-(4-(1,4-Dimethyl-3-oxopiperazin-2-yl)phenylamino)-4-methyl-5-oxo-4,5-dihydropyrazin-2-yl)-2-methylphenyl)-4,5,6,7-tetrahydrobenzo[b] thiophene-2-carboxamide (GDC-0834) is a highly selective, reversible, ATP competitive small molecule inhibitor of BTK being developed as a therapeutic agent for RA (Liu et al., 2011). In B cells, this compound effectively blocks B-cell antigen receptor-and CD40-mediated activation and proliferation. In monocytes, it potently inhibits immune complex-mediated inflammatory cytokine production, including tumor necrosis factor-α, which may also contribute to disease pathogenesis in RA.
Preliminary in vitro metabolite identification studies of GDC-0834 suggested significant differences in the major route of metabolism in human compared with that in other preclinical species. In human, an aniline metabolite (M1) formed via an amide hydrolysis reaction appeared to be more predominant than in other preclinical species tested. The objective of this study was to characterize the species differences in the metabolism of GDC-0834 in mouse, rat, dog, monkey, and human using both in vitro and in vivo methodologies. An additional objective was to characterize GDC-0834 in a novel humanized liver mouse (PXB) model.
Materials and Methods
Materials and Animals.
GDC-0834 (Fig. 1A) and its aniline metabolite M1 (Fig. 1B) were provided by Genentech (South San Francisco, CA) and CGI Pharmaceuticals Inc. (Branford, CT). The preclinical in vivo studies described herein were conducted according to protocols approved by the institutional animal care and use committee at Genentech, SRI International (Menlo Park, CA), Covance Laboratories (Madison, WI), and PhoenixBio (Higashi-Hiroshima, Japan). High-performance liquid chromatography-grade solvents (acetonitrile, methanol, and water) were purchased from EMD Chemicals, Inc. (Gibbstown, NJ). Formic acid was purchased from Mallinckrodt Baker, Inc. (Phillipsburg, NJ).
In Vivo Preclinical Pharmacokinetics.
GDC-0834 was dosed at 1 mg/kg intravenously and 5 mg/kg orally in female CD-1 mice (n = 24/route, 3/time point), male Sprague-Dawley rats (n = 3/route) (Charles River Laboratories, Hollister, CA), male beagle dogs (n = 3/route), and male cynomolgus monkeys (n = 3/route) (Covance, Cumberland, VA). For all studies involving oral administration, animals were fasted overnight until 4 h postdose. Water was supplied ad libitum. The blood samples were collected up to 24 h postdose into tubes containing EDTA as an anticoagulant. Blood samples were centrifuged within 30 min of collection, and plasma was harvested. Plasma samples were stored at approximately −70°C until analysis for both GDC-0834 and its aniline metabolite M1 (oral groups only for M1) by liquid chromatography-tandem mass spectrometry (LC-MS/MS, see details under Bioanalysis of GDC-0834 and M1 in Preclinical and Clinical Studies). Pharmacokinetic (PK) analysis was performed on either mean plasma concentration-time data (mouse) or on the individual plasma concentration-time data (rat, dog, and monkey). PK parameters were estimated by noncompartmental methods using WinNonlin Enterprise (version 5.2; Pharsight, Mountain View, CA). The IV-Bolus input model (model 201) was used for the intravenous GDC-0834 dose group, and the Extravascular input model (model 200) was used for the oral GDC-0834 dose group and M1 data.
In Vivo PK in PXB Mice with Humanized Livers.
PXB-mice (PhoenixBio) were produced by transplanting human hepatocytes (BD Biosciences, Woburn, MA) into the spleens of 2- to 4-week-old male uPA(+/+)/severe combined immunodeficiency disease (SCID) mice under diethyl ether anesthesia. After transplantation, mice with a blood human albumin concentration of 7.0 mg/ml or more were selected as PXB-mice for use. GDC-0834 was dosed at 1 mg/kg intravenously and 5 mg/kg orally in both male PXB and control SCID (Charles River Japan, Kanagawa, Japan) mice. In the intravenous groups, one animal per group was sacrificed for sampling at 0.033, 0.167, 0.5, 1, 3, and 6 h postdose. In the oral groups, one animal per group was sacrificed for sampling at 0.083, 0.25, 0.5, 1, 3, and 6 h postdose. A minimum of 300 μl of blood was collected from each animal via the heart into sodium heparin-coated syringes after which the animals were sacrificed by cardiac puncture and exsanguination. Blood samples were stored on ice until centrifugation within 30 min to obtain plasma. GDC-0834 and M1 metabolite plasma concentrations were assessed using LC-MS/MS. PK parameters were determined by noncompartmental methods as described above.
Liver Microsome and Hepatocyte Metabolic Stability Assays.
Metabolic stability assays in liver microsomes were conducted with pooled female CD-1 mouse (n = >10), male Sprague-Dawley rat (n = 20), male beagle dog (n = 3), male cynomolgus monkey (n = 4) (Celsis In Vitro Technologies, Baltimore, MD), and mixed male and female human (n = 15) (CellzDirect, Durham, NC) liver microsomal incubations. The general assay conditions are as follows. Incubation mixtures consisted of liver microsomes (0.5 mg of microsomal protein/ml) and GDC-0834 (1.0 μM) with or without NADPH (1.0 mM) in the potassium phosphate buffer (100 mM; pH 7.4) with a final incubation volume of 0.25 ml. Reactions were initiated by the addition of NADPH or buffer and shaken in a water bath open to the air at 37°C. At times 0, 20, 40, and 60 min, aliquots (50 μl) were removed and added to termination mixtures (100 μl) containing acetonitrile and an internal standard from an in-house project. The samples were then centrifuged for 10 min at 2000g. The supernatant (90 μl) was removed, combined with 180 μl of water, and analyzed by LC-MS/MS.
Metabolic stability assays in hepatocytes were conducted using cryopreserved pooled female CD-1 mouse (n >10), male Sprague-Dawley rat (n >10), male beagle dog (n = 3), male cynomolgus monkey (n = 3), and mixed male and female human (n = 10) hepatocytes (CellzDirect). Vials of hepatocytes were thawed rapidly in a water bath set at 37°C and then were diluted with Dulbecco's modified Eagle's medium (DMEM), pH 7.4. Cells were isolated by centrifugation, pooled, and resuspended in DMEM at 1.0 million viable cells/ml. Membrane integrity of the cells was assessed by trypan blue exclusion. GDC-0834 was dissolved in dimethyl sulfoxide at a final concentration of 10 mM. This GDC-0834 stock was diluted further to 2 μM in DMEM (125 μl) before the addition of an equal volume of the 106 cells/ml cell suspension. GDC-0834 (final incubation concentration of 1.0 μM with 0.1% dimethyl sulfoxide) and cells (final concentration of 0.5 × 106 cells/ml) were incubated at 37°C in a 95% air/5% CO2 atmosphere for 3 h. Aliquots (50 μl) were removed at 0, 1, 2, and 3 h and added to termination mixtures (100 μl) containing acetonitrile and an internal standard from an in-house project. The samples were then centrifuged for 10 min at 2000g. The supernatant (90 μl) was removed, combined with 180 μl of water, and analyzed by LC-MS/MS.
In Vitro Metabolite Identification in Hepatocytes.
Metabolite identification studies were performed using cryopreserved pooled female CD-1 mouse (n >10), male Sprague-Dawley rat (n >10), male beagle dog (n = 3), male cynomolgus monkey (n = 3), and mixed male and female human (n = 10) hepatocytes (CellzDirect). In brief, GDC-0834 (10 μM) hepatocyte incubations were conducted in 96-well plates containing approximately 125,000 cells/well in an incubator at 37°C with 5% CO2 and 95% humidity. Samples from the incubations (100 μl) were taken at 0 and 3 h time points. Reactions were quenched with 200 μl of acetonitrile. Metabolite identification was performed using LC-MS/MS, which consisted of a LC-10AD pump (Shimadzu Corporation, Columbia, MD), a HTS PAL autosampler (CTC Analytics, Carrboro, NC), and a QTRAP 4000 mass spectrometer equipped with a TurboIonSpray (Applied Biosystems, Foster City, CA). Data-dependent ion, product ion, precursor ion, neutral loss, and multiple-reaction monitoring scans in positive ion electrospray mode were used to characterize GDC-0834 and its metabolites. The UV traces obtained were used to quantify the amount of metabolites formed in hepatocytes relative to GDC-0834 at 0 h.
Formation Kinetics of M1 Metabolite.
Enzyme kinetics studies of M1 formation were performed (MDS Pharma Services, Bothell, WA) using pooled liver microsomes from mixed male and female human, male cynomolgus monkey, male beagle dog, and male Sprague-Dawley rat. In brief, GDC-0834 concentrations of 0.156, 0.625, 2.5, 5, 10, 20, 40, 80, and 160 μM were incubated at 37°C in duplicate with microsomes in 75 mM phosphate buffer without NADPH cofactors. The microsomal protein concentrations in the incubations were 0.25 mg/ml for human and 3.0 mg/ml for the other species. Optimized incubation times in human, rat, monkey, and dog were 15, 40, 90, and 120 min, respectively, based on conditions of linear formation of M1 from a pilot study (data not shown). The assay was initiated by the addition of substrate (GDC-0834) and terminated by protein precipitation through the addition of acetonitrile containing 0.2 μM internal standard (metoprolol). The formation rate of M1 was measured by LC-MS/MS. Estimation of the maximum rate of M1 formation (Vmax) and the Michaelis-Menten constant (Km) was performed by nonlinear regression analysis (GraphPad Prism version 4.03; GraphPad Software Inc., La Jolla, CA) involving fitting of the Michaelis-Menten equation to the M1 formation rate (nanomoles per minute per milligram of protein) versus GDC-0834 concentration (micromolar) data. Study results were also plotted in Lineweaver-Burk (1/v versus 1/S) and Eadie-Hofstee (v versus v/S) plots to assess the presence of more than one enzyme system.
Human Pharmacokinetics Prediction Based on Allometry.
Predictions of GDC-0834 clearance (CL) in humans were made using simple allometric scaling (eq. 1) and allometric scaling corrected with MLP according to the rule of exponents (Mahmood and Balian, 1996) (eqs. 2 and 3). The volume of distribution at steady state (Vss) was predicted by simple allometric scaling (eq. 4) where a is the allometric coefficient, b is the allometric exponent, W is the body weight in kilograms, BW is the brain weight in kilograms, and MLP is the maximum lifespan potential in years.
Clearance Prediction Based on In Vitro/In Vivo Extrapolation.
IVIVE was used as another method for human (and animal) clearance prediction. On the basis of the “well stirred” model (Pang and Rowland, 1977), the following equation (eq. 5) was used for the clearance estimation, where fu is the GDC-0834 free fraction in blood, Clint is the intrinsic metabolic clearance based on the in vitro t1/2 method introduced by Obach et al. (1997) from either in vitro liver microsome (LM) or hepatocyte metabolic stability assays, and Q is the hepatic blood flow.
Phase 1 Clinical Trial of GDC-0834.
A clinical double-blind randomized, placebo-controlled, single oral dose study was performed at Covance Laboratories (Madison, WI). A suspension of GDC-0834 was administered orally after at least an 8-h fast or a placebo (microcrystalline cellulose) in suspension. Two cohorts of healthy volunteers, based on a 4:2 (cohort A) and 5:1 (cohort B) treatment allocation (GDC-0834/placebo), were given single oral doses of GDC-0834 at 35 and 105 mg, respectively. Blood samples were collected into tubes containing EDTA as an anticoagulant at predose and 0.5, 1, 2, 4, 6, 12, 24, and 48 h postdose. All subjects remained at the contract research organization for 48 h postdose to evaluate safety and returned on day 8 (7 days after dosing) for safety evaluation. The decision to escalate to the next dose level was based on a safety review of clinical and laboratory data on all subjects in the previous cohort through day 3.
Metabolite Identification in Human Urine and Plasma.
Urine samples from cohort B (105 mg) from predose and 4 h, respectively, were pooled (n = 5). An aliquot (1000 μl) of pooled urine was quenched with 1000 μl of acetonitrile followed by vortexing for 5 min and centrifuging for 10 min at 3400 rpm. Plasma samples from 1 to 6 h from cohort B were pooled (n = 5). An aliquot (1 ml) of pooled plasma was quenched with 3 ml of acetonitrile followed by vortexing for 5 min and centrifuging for 10 min. The supernatants were dried down at room temperature with a CentriVap DNA Concentrator (Labconco, Kansas City, MO) and reconstituted with 2:1 H2O-MeOH. Metabolite identification was performed on reconstituted samples using LC-MS/MS.
LC-MS/MS analysis was conducted using an LTQ XL mass spectrometer coupled to an ultra high-pressure liquid chromatograph (Accela Pump), an Accela autosampler, and an Accela PDA detector (Thermo Fisher Scientific, Waltham, MA). The mass spectrometric conditions were set as follows: positive mode electrospray ionization, capillary voltage at 17 V, source voltage at 4.5 kV, tube lens voltage at 105 V, capillary temperature at 350°C, sheath gas flow at 74 units, auxiliary gas flow at 6 units, sweep gas flow at 19 units using full-scan and MSn scan modes. The high-performance liquid chromatography column used was a Thermo Hypersil Gold C18 (100 × 2.1 mm, 1.9 μm; Thermo Fisher Scientific). The solvent system consisted of solvent A (0.1% formic acid in water) and solvent B (0.1% formic acid in acetonitrile). Solvent B was delivered initially at 5%, held for 1 min, and increased to 95% via a 25-min gradient and then decreased back to 5% at 26 min, at which time it was retained for 4 min at a flow rate of 400 μl/min with a total run time of 30 min.
Bioanalysis of GDC-0834 and M1 in Preclinical and Clinical Studies.
The preclinical samples were prepared for analysis by first aliquoting 25 μl of plasma into a 96-well plate followed by the addition of 25 μl of internal standard solution and 150 μl of acetonitrile to each sample for protein precipitation. The samples were vortexed and centrifuged for 5 min at 10,000 rpm. Then 50-μl aliquots of the resulting supernatant were mixed with 300 μl of water, and 10 μl was injected onto the LC-MS/MS system for analysis. For the clinical samples, a method validated according to the U.S. Food and Drug Administration Guidance for Industry on Bioanalytical Method Validation (http://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/ucm070107.pdf) was used, in which the dynamic range of the assay ranged from 1.0 to 500.0 ng/ml using a 50-μl aliquot of human plasma for measurement of GDC-0834 and its M1 metabolite (Y. G. Shin, S. A. Jones, S. C. Murakami, L. Liu, H. Wong, M. H. Buonarati, and C. E. C. A. Hop, manuscript in preparation).
Concentrations of GDC-0834 and its M1 metabolite in plasma samples from preclinical species were determined by a nonvalidated assay using a TSQ quantum liquid chromatography-triple quadrupole mass spectrometer connected to a CTC PAL (Leap Technologies, Chapel Hill, NC) autosampler. GDC-0834 and M1 were separated using an ACE 5 phenyl reverse-phase (100 × 2.1 mm, 5 μm) column (Advanced Chromatography Technologies; Aberdeen, UK) at room temperature and an LC-10AD pump with a Shimadzu SCL-10A controller (Shimadzu Corporation). The mobile phase consisted of mobile phase A (water with 0.1% formic acid) and mobile phase B (acetonitrile with 0.1% formic acid). GDC-0834 and M1 metabolite were eluted using an initial condition of 20% mobile phase B followed by a linear gradient to 80% mobile phase B over 5 min with a flow rate of 0.5 ml/min. The mass spectrometer was in positive ion mode using an electrospray interface at 400°C with nitrogen serving as both the nebulizing and heating gas. GDC-0834 and its M1 metabolite were analyzed in the multiple reaction monitoring mode using the following transitions: m/z 597.4 → 127.1 and m/z 433.4 → 127.1, respectively. The calibration curves for GDC-0834 and its M1 metabolite were prepared by plotting the appropriate peak area ratios of analyte/internal standard against the known concentrations of GDC-0834 or M1 metabolite in plasma using a linear regression with 1/x2 weighing. Concentrations of GDC-0834 and its M1 metabolite in samples were determined by interpolation from their respective standard curves. The dynamic range was 5 to 10,000 nM for both parent and metabolite in the preclinical studies. A run was deemed acceptable when quality control samples as well as calibration curve samples were within ±25% of the nominal concentration, except at the lowest level of quantification sample for which ±30% was accepted.
Results
Preclinical Results.
Pharmacokinetics of GDC-0834 in mouse, rat, dog, and monkey.
Relevant preclinical PK parameters of GDC-0834 are presented in Table 1. The in vivo plasma CL is low in rats (4.7 ml · min−1 · kg−1), moderate in mice and dogs (43 and 14 ml · min−1 · kg−1, respectively), and high in monkeys (33 ml · min−1 · kg−1) (Table 1). Vss is low to moderate (0.25–1.5 l/kg) in all species evaluated. The terminal half-life (t1/2) ranges from 0.158 h in mouse to 1.09 h in rat. Oral bioavailability (F) in the mouse, rat, dog, and monkey is 18, 16, 32, and 16%, respectively (Table 1).
Pharmacokinetics of GDC-0834 and M1 metabolite in PXB and SCID mouse.
Relevant preclinical PK parameters of GDC-0834 in PXB and SCID mice are presented in Table 2. In general, the pharmacokinetics of GDC-0834 was comparable between PXB and SCID mice. Estimates of plasma CL were 22 and 15 ml · min−1 · kg−1 in PXB and SCID mice, respectively (Table 2). F in PXB and SCID mice was comparable (35% versus 31%) (Table 2). Despite these similarities in GDC-0834 PK parameters, PXB mice show much higher metabolite (M1)/parent ratios than SCID mice for both the intravenous (0.495 versus 0.0112) and oral (0.741 versus 0.0927) groups (Table 2).
GDC-0834 metabolic stability in liver microsomes and hepatocytes.
The predicted hepatic CL values in LM with NADPH were high in all species evaluated (Table 3). The elimination of GDC-0834 in humans appeared to be due to an NADPH-independent pathway because the predicted hepatic CL was similar with and without NADPH in human liver microsomes (Table 3). The contribution of this pathway appeared to be less prominent in other preclinical species on the basis of the liver microsomes without NADPH data. In hepatocyte stability, clearances are predicted as high in human and monkey and moderate in rat, mouse, and dog (Table 3). The in vivo preclinical clearances are also listed in the Table 3 for comparison.
In vitro and in vivo metabolite identification studies.
The in vitro hepatocyte metabolism of GDC-0834 is moderate to extensive in all species tested (Table 4). A significant metabolic pathway for GDC-0834 in mouse, rat, dog, monkey, and human is amide hydrolysis to form M1, and it is most pronounced in human hepatocytes. The extent of metabolism to M1 varied according to species, being 22, 8.6, 12, 3.6, and 62% in mouse, rat, dog, monkey, and human hepatocytes, respectively, based on the peak areas of UV peaks relative to GDC-0834 at the 0-h time point. Other minor metabolic pathways observed include a combination of oxidation, demethylation, and desaturation of GDC-0834 or metabolite. In cynomolgus monkeys, N-demethylation and oxidation are comparable to amide hydrolysis. There were no human-specific metabolites detected in the in vitro study (Table 4; Fig. 2).
The amide hydrolysis of GDC-0834 was characterized in vivo by measuring concentrations of the aniline metabolite, M1 (Fig. 1B), in plasma from the PK studies. Consistent with the in vitro hepatocyte data (Table 4; Fig. 2), M1 was found in only minor (rat and monkey) to moderate (dog) quantities in plasma from preclinical species dosed orally with GDC-0834 (Fig. 3). The metabolite (M1)/parent ratios of exposures were 1.5%, 26%, and negligible, respectively, in rats, dogs, and monkeys.
Human PK Prediction.
Predicted human CL of GDC-0834.
Allometric scaling was performed to provide a CL prediction for GDC-0834 in humans using in vivo CL estimates from mouse, rat, dog, and monkey PK studies. Predicted clearance in humans was moderate (13 ml · min−1 · kg−1) on the basis of simple allometric scaling (Fig. 4A). Because the allometric exponent (b) is 0.9 (0.7 < b < 1), the rule of exponents (Mahmood and Balian, 1996) was applied. On the basis of the rule of exponents, the clearance was estimated by allometric scaling corrected using MLP. This resulted in a low predicted clearance of 5.4 ml · min−1 · kg−1 (Fig. 4B).
In addition to CL predictions using allometric methods, predictions of human CL were made using in vitro methods (Table 3). Human liver microsomes and hepatocyte metabolic stability assays predicted high clearance in humans (∼19 ml · min−1 · kg−1, which is close to the human hepatic blood flow).
Predicted human Vss of GDC-0834.
Allometric scaling was performed to provide a predicted Vss of GDC-0834 in humans using estimates of Vss from mouse, rat, dog, and monkey PK studies. Predicted Vss in humans was estimated to be 1.4 l/kg (Fig. 4C).
Clinical Results.
Clinical human pharmacokinetics of GDC-0834 and M1.
In all subjects dosed with GDC-0834, substantial concentrations of M1 were observed in plasma, whereas levels of GDC-0834 were below the limit of quantitation (<1 ng/ml) in most plasma samples. At 35 mg, the mean highest observed plasma concentration (Cmax) and AUC of M1 are 142 ± 48.1 ng/ml and 837 ± 185 h · ng/ml, respectively (Fig. 5A); at 105 mg, the mean Cmax and AUC of M1 are 390 ± 120 ng/ml and 2448 ± 921 h · ng/ml, respectively (Fig. 5B). The mean Cmax and AUC of M1 are approximately linear between these two doses. The clinical development of GDC-0834 was terminated on the basis of the above data.
In vivo metabolite identification studies with GDC-0834.
In vivo metabolite identification of plasma and urine from the healthy volunteers dosed with GDC-0834 showed only three metabolites: M1 (amide hydrolysis) and its sequential metabolites M9 (N-demethylation) and M10 (oxidation) (Fig. 6). On the basis of structure assignment by mass spectrometry, M9 and M10 both appear to be metabolites of M1, suggesting that amide hydrolysis was the major metabolic pathway in vivo in humans.
Formation Kinetics of M1.
Results of the enzyme kinetics study of the formation of M1 in rat, dog, monkey, and human LM without NADPH are presented in Fig. 7. Vmax, the maximum rate of M1 formation, was substantially higher (21- to 60-fold) in human (0.60 nmol/min/mg protein) compared with that in rat (0.028 nmol/min/mg protein), monkey (0.025 nmol/min/mg protein), and dog (0.010 nmol/min/mg protein) (Fig. 7). In contrast, the Michaelis-Menten constant (Km) was similar among species spanning only an approximate 3-fold range (22–63 μM) (Fig. 7). Overall, the estimated intrinsic clearance (Vmax/Km) of GDC-0834 from M1 formation was 23- to 169-fold higher in human than that in rat, dog, and monkey. The Lineweaver-Burk and Eadie-Hofstee plots for all species studied did not suggest the involvement of more than a single enzyme (data not shown).
Discussion
Species differences in PK are common and result from differences in physiological processes influencing the absorption, distribution, metabolism, and excretion of xenobiotics (Lin, 1995; Toutain et al., 2010). One of the major causes of species-dependent PK is species differences in metabolism. It is well known that qualitative and quantitative differences in metabolism exist between humans and other species. Because P450 enzymes play an important role in the metabolism of many xenobiotics, advances in the understanding of species differences in P450 structure, substrate specificity, and expression have provided much information on the causes of observed differences between humans and preclinical species (Lin, 1995). Comparisons of P450 activities in liver from preclinical species such as mouse, rat, rabbit, dog, and micropig show clear differences in activity compared with that in human liver (Bogaards et al., 2000). In fact, despite the genetic similarity between nonhuman primates and humans, in vitro and in vivo studies examining P450 activity in nonhuman primates also indicate differences (Bogaards et al., 2000; Wong et al., 2004). Despite these known differences in the qualitative and quantitative aspects of animal metabolism, preclinical pharmacokinetic studies still serve as an important means to assess human pharmacokinetics and safety of drug candidates before clinical studies in human.
The prediction of the human pharmacokinetics of drug candidates is particularly challenging (Beaumont and Smith, 2009). Much literature exists on this particular topic, and methodologies available for prediction of human pharmacokinetics are largely either allometric type methods that rely on in vivo data or IVIVE methods that rely heavily on data from in vitro studies (Hosea et al., 2009). The identification of a universal method for the prediction of human pharmacokinetics remains a significant challenge (Hosea, 2011). Allometric methods for prediction of human clearance assume a relationship between clearance and body weight. These methods work well for compounds for which clearance is due to elimination dictated by physiological processes such as blood flow or filtration (Lin, 1995). Therefore, for high-clearance compounds, for which clearance is blood flow-dependent, and for compounds that are cleared primarily via the urinary route, allometry should perform reasonably well. However, the rationale of using allometric scaling for human PK prediction has been controversial (Bonate and Howard, 2000) and a known shortcoming of using in vivo allometric methods is when there are known species differences in metabolism.
In contrast to allometry, IVIVE methods using in vitro data derived from liver fractions apply only to compounds eliminated primarily by the liver. Broad experience with IVIVE methods exists largely for compounds metabolized by P450s (Obach et al., 1997). Literature for IVIVE of compounds metabolized by non-P450 metabolic pathways is much more limited, and correlations are not well established for these pathways. As illustrated by the published experience with UDP-glucuronosyltransferases (Miners et al., 2006), IVIVE of substrates of non-P450 enzymes is more problematic. More specific to GDC-0834, less is known about the enzymes involved in amide hydrolysis. Hydrolysis reactions are not as common as P450-mediated metabolic reactions; however, they play an important role in the metabolism of xenobiotics (Testa and Mayer, 2003). There are a variety of hydrolytic enzymes, including carboxylesterases, cholinesterases, organophosphatases, and amidases/peptidases that hydrolyze compounds containing functional groups such as esters, thioesters, amides, and epoxides. Of these enzymes, cholinesterases and aminopeptidases are probably the most efficient in hydrolyzing the amide bonds of drugs (Uetrecht and Trager, 2007). In addition, carboxylesterases are often active hydrolases in human small intestine and liver and show large species differences in metabolism (Taketani et al., 2007). In preliminary investigations, we performed experiments using hydrolase inhibitors, such as bis-4-nitrophenylphosphate, EDTA, and eserine, for class A and B and other hydrolases, and purified hydrolase enzymes but were unsuccessful in identifying the hydrolytic enzymes responsible for the metabolism of GDC-0834 (data not shown). Finally, to the best of our knowledge, there is no literature on human clearance prediction methods for compounds such as GDC-0834, which appear to be primarily metabolized by amide hydrolysis.
GDC-0834 human clearance predictions using allometric and in vitro methodologies resulted in a wide range of values (5.4–19 ml · min−1 · kg−1). As discussed, the metabolism of GDC-0834 was species-dependent in terms of both rate and metabolic pathway. As such, we had low confidence in our GDC-0834 human clearance predictions estimated using allometric methods. In addition, there appeared to be a lack of IVIVE in preclinical species with predictions of clearance using microsomes being higher than observed in vivo clearance in rat, mouse, and dog. IVIVE improved when hepatocyte data were used; however, predictions of clearance in the rat by IVIVE were still higher than those observed in vivo. Furthermore, our overall understanding of amide hydrolysis enzymes is considerably less than that for P450s, and there is no literature on the IVIVE of compounds metabolized by amide hydrolysis. Given these challenges, our confidence in the IVIVE clearance prediction was equally low. In several studies, a chimeric mouse model with a humanized liver, designated as the PXB mouse, has been developed and characterized for drug discovery (Tateno et al., 2004; Kikuchi et al., 2010). We attempted to retrospectively evaluate this model to determine whether it could have been used as an in vivo tool to elucidate the pharmacokinetic behavior of GDC-0834 in humans. It should be recognized that the utility of PXB mice as a tool for understanding drug metabolism in humans has been largely studied using compounds metabolized primarily by P450 (Kikuchi et al., 2010). Qualitatively, our studies with PXB mice show a higher level of the amide hydrolysis metabolite consistent with in vitro data from human liver microsomes and hepatocytes. However, quantitatively, there appeared to be a significant difference between GDC-0834 clearance in PXB mice (low clearance) and what was observed in humans. Moreover, the clearance was similar between PXB and control SCID mice. It is clear that additional tools are required to better understand the IVIVE of compounds metabolized by amide hydrolysis.
Additional in vitro enzyme kinetics studies were conducted to understand the species differences in the amide hydrolysis of GDC-0834. The results of these studies show that human liver has a much greater capacity (i.e., higher Vmax) for amide hydrolysis of GDC-0834, suggesting a higher expression/activity of the enzymes involved in the amide hydrolysis reaction in human. These findings were consistent with observations from the human pharmacokinetic study, in which there was little to no parent compound detected in human plasma after oral dosing (Fig. 5). Observed metabolites in human plasma and urine samples from this study were all sequential metabolites of the primary amide hydrolysis metabolite, M1 (Fig. 6). In addition, incubations using intestinal S9 fraction suggested that GDC-0834 exhibited little to no intestinal metabolism, and GDC-0834 was found to be chemically stable in simulated gastric fluid with and without pepsin over a period of 2 h (J. Halladay and J. Lubach, personal communication). Furthermore, a blood stability assay indicated that GDC-0834 was stable for 3 h in human blood. Taken together, the in vitro and in vivo data suggest that in humans absorbed GDC-0834 is converted primarily to the M1 metabolite via a first-pass effect by the liver. In a recent study, species differences in amide hydrolysis have been reported for an agonist of the bile acid Takeda G-protein-coupled receptor 5 (TGR5) (Eng et al., 2010). The TGR5 agonist being investigated behaved differently from GDC-0834 and underwent amide bond cleavage in both rat and mouse plasma. In contrast with the findings in rodents, the TGR5 agonist was resistant to hydrolytic cleavage in both dog and human plasma (Eng et al., 2010). Our understanding of species differences in amide hydrolysis will continue to grow as more examples of this metabolic route are observed and reported.
RA is an autoimmune disease that causes chronic inflammation of the joints, the tissues that surround the joints, and other organs in the body. The current standard of care for RA is methotrexate, which is not without its shortcomings, including a number of potentially serious adverse effects, such as hematopoietic suppression, hepatotoxicity, and pulmonary toxicity (Stamp et al., 2006). On the basis of the properties of current therapies, there is a clear need for newer and safer therapies for RA. GDC-0834 is a potent and selective BTK inhibitor with an acceptable preclinical toxicity profile and demonstrated preclinical efficacy in the rat CIA model (Liu et al., 2011). The major liability for GDC-0834 was the low confidence in our human clearance predictions. Approaches to elucidate human pharmacokinetics more rapidly with reduced cost include the use of microdosing studies (Lappin, 2010). In our case, we conducted single-dose toxicology studies to support the submission of a single-dose investigational new drug (IND) application. A single-dose IND application enabled rapid assessment of the human pharmacokinetics of GDC-0834, resulting in the termination of GDC-0834 under a shorter time frame and with a reduced cost than if we chose to evaluate the compound under a traditional IND application. Finally, the results from in vitro studies with liver microsomes and hepatocytes appeared to correlate with the in vivo situation in humans, suggesting an in vitro to in vivo correlation for amide hydrolysis. Information gathered using GDC-0834 greatly helped in our design of molecules that are void of this liability.
Authorship Contributions
Participated in research design: Liu, Halladay, Shin, S. Wong, Coraggio, La, Baumgardner, Le, Gopaul, Boggs, Kuebler, Davis, Liao, Lubach, Deese, Sowell, Currie, Young, Khojasteh, Hop, and H. Wong.
Conducted experiments: S. Wong, Coraggio, La, Baumgardner, Le, Gopaul, Davis, Lubach, Deese, Sowell, and Currie.
Performed data analysis: Liu, Halladay, S. Wong, Coraggio, La, Baumgardner, Le, Gopaul, Boggs, Kuebler, and H. Wong.
Wrote or contributed to the writing of the manuscript: Liu, Halladay, Shin, Lubach, Currie, Young, Khojasteh, Hop, and H. Wong.
Acknowledgments
We thank the BTK team members at Genentech and CGI Pharmaceuticals Inc., and the staff in the Department of DMPK for their contributions in generating the data for this manuscript.
Footnotes
Article, publication date, and citation information can be found at http://dmd.aspetjournals.org.
doi:10.1124/dmd.111.040840.
-
ABBREVIATIONS:
- BTK
- Bruton's tyrosine kinase
- RA
- rheumatoid arthritis
- GDC-0834
- (R)-N-(3-(6-(4-(1,4-dimethyl-3-oxopiperazin-2-yl)phenylamino)-4-methyl-5-oxo-4,5-dihydropyrazin-2-yl)-2-methylphenyl)-4,5,6,7-tetrahydrobenzo[b]thiophene-2-carboxamide
- PK
- pharmacokinetic(s)
- SCID
- severe combined immunodeficiency disease
- LC
- liquid chromatography
- MS/MS
- tandem mass spectrometry
- DMEM
- Dulbecco's modified Eagle's medium
- MLP
- maximum lifespan potential
- IVIVE
- in vitro/in vivo extrapolation
- LM
- liver microsome(s)
- AUC
- area under the concentration-time curve
- P450
- cytochrome P450
- IND
- investigational new drug.
- Received May 23, 2011.
- Accepted July 6, 2011.
- Copyright © 2011 by The American Society for Pharmacology and Experimental Therapeutics