Advertisement
Research ArticleArticle

Lack of Exposure in a First-in-Man Study Due to Aldehyde Oxidase Metabolism: Investigated by Use of 14C-microdose, Humanized Mice, Monkey Pharmacokinetics, and In Vitro Methods

Klaus Gjervig Jensen, Anne-Marie Jacobsen, Christoffer Bundgaard, Dorrit Østergaard Nilausen, Zia Thale, Gamini Chandrasena and Martin Jørgensen
Drug Metabolism and Disposition January 2017, 45 (1) 68-75; DOI: https://doi.org/10.1124/dmd.116.072793

Abstract

Inclusion of a microdose of 14C-labeled drug in the first-in-man study of new investigational drugs and subsequent analysis by accelerator mass spectrometry has become an integrated part of drug development at Lundbeck. It has been found to be highly informative with regard to investigations of the routes and rates of excretion of the drug and the human metabolite profiles according to metabolites in safety testing guidance and also when additional metabolism-related issues needed to be addressed. In the first-in-man study with the NCE Lu AF09535, contrary to anticipated, surprisingly low exposure was observed when measuring the parent compound using conventional bioanalysis. Parallel accelerator mass spectrometry analysis revealed that the low exposure was almost exclusively attributable to extensive metabolism. The metabolism observed in humans was mediated via a human specific metabolic pathway, whereas an equivalent extent of metabolism was not observed in preclinical species. In vitro, incubation studies in human liver cytosol revealed involvement of aldehyde oxidase (AO) in the biotransformation of Lu AF09535. In vivo, substantially lower plasma exposure of Lu AF09535 was observed in chimeric mice with humanized livers compared with control animals. In addition, Lu AF09535 exhibited very low oral bioavailability in monkeys despite relatively low clearance after intravenous administration in contrast to the pharmacokinetics in rats and dogs, both showing low clearance and high bioavailability. The in vitro and in vivo methods applied were proved useful for identifying and evaluating AO-dependent metabolism. Different strategies to integrate these methods for prediction of in vivo human clearance of AO substrates were evaluated.

Introduction

Lu AF09535 is a high-affinity negative allosteric modulator at the human metabotropic glutamate 5 receptor and was chosen for advancement in to first-in-man (FIM) studies. The candidate was selected from a series of adamantyl diamide derivatives with aryl or heteroaryl substituents at R1 and R2 positions as described by Li et al. (2013). During the selection process of Lu AF09535 in the drug discovery phase, aldehyde oxidase (AO)-dependent metabolism was identified to be involved in the metabolism of other structurally related compounds in the series of potential candidates. However, initially, the human liver microsome and hepatocyte intrinsic clearance was found to be low for Lu AF09535 as did in vitro and in vivo clearances in the preclinical species rat, mouse, and dog, with high oral bioavailability of 60 to 70% (data not shown). Various in vitro-in vivo extrapolation approaches including allometric scaling, predicted low clearance in humans and, together with the other favorable characteristics including a desirable pharmacological profile, Lu AF09535 was advanced for FIM studies. Here, the compound exhibited very low plasma exposure with the first measurable area under the curve (AUC) detected only after a single oral dose of 75 mg, raising numerous questions, such as the correct strength of dosing solution used, effects of dosing formulation, potential for precipitation, bioanalysis errors, etc.

Since the Metabolites in Safety Testing (MIST) guidelines were issued in 2008 and 2009, Lundbeck has implemented a strategy for addressing metabolite safety by investigations of metabolism by inclusion of a microdose of 14C-labeled drug in the FIM study with subsequent analysis by accelerator mass spectrometry (AMS) (Lappin and Seymour, 2010; Lappin et al., 2012). The main objectives of this type of studies are to determine the routes and rates of excretion of the drug-related material (mass balance) and to investigate the human metabolite profiles according to the MIST guidelines. In addition, microdosing studies have proven to be very informative, not only to support MIST guidance but also to address complex metabolism-related issues in humans. One example, presented in this publication, describes the elucidation of the mechanisms behind the low exposure of Lu AF09535 in humans.

Several avenues were pursued to understand the failure of the initial prediction of human clearance of Lu AF09535 and to test other scaling methods for their utility in future screening cascades to avoid repetition of the observed unanticipated extensive metabolism. To elucidate the mechanism behind the extensive human metabolism of Lu AF09535, it was investigated-whether AO played a major role in its biotransformation. The importance of non-CYP-mediated metabolism in drug discovery has been the focus of numerous studies during the recent years. Examples of drugs where the human exposure was poorly predicted before the first human dose due to metabolism by pathways other than the CYP enzymes were reported recently (Diamond et al., 2010; Akabane et al., 2011; Zhang et al., 2011). AO is a molybdoflavoprotein that catalyzes the metabolism of not only aldehydes but also nitrogenous heterocycles (Garattini et al., 2008; Pryde et al., 2010). AO metabolism is known to be particularly difficult to predict due to a number of reasons. Because the absorption distribution metabolism excretion disciplines within drug discovery traditionally has focused on CYP metabolism to achieve low clearance drugs, most of the existing methods for prediction of human pharmacokinetics have been largely established based on CYP metabolism profiles. Furthermore, rat and dog, which are typical species being used for allometric scaling to predict human exposure, both seem to differ from humans in AO-mediated metabolism, both having low or no expression of the enzyme (Pryde et al., 2010). Also, large strain differences in AO expression have been reported among rats (Sugihara et al., 1995). AO activity generally seems to be high in monkeys (Kitamura et al., 2001) and humans with a widespread distribution of AO in human hepatic and extrahepatic tissues (Moriwaki et al., 2001; Pryde et al., 2010).

The literature contains a range of approaches to improve the prediction of human exposure of compounds that are shown to be AO substrates. As a follow up to a clinical microdosing study with Lu AF09535, we investigated several experimental scaling approaches. First, the yardstick approach described by Zientek et al. (2010) in which a range of known AO substrates are categorized into low, medium, or high clearance compounds in vivo, thereby providing a “calibration range” for in vitro intrinsic clearance data obtained in relevant cell media. With this approach, compounds can be evaluated in vitro in S9 fraction or cytosol media and the in vivo clearance can be qualitatively scaled for evaluation. Second, the metabolism of Lu AF09535 in chimeric PXB-mice with humanized livers was investigated. PXB-mice have a liver replacement index of 70% or more and have been shown to have human-like responses to drug metabolism (Tateno et al., 2004; Sanoh et al., 2012a) and contain human AO enzyme as well (Kitamura et al., 2008). It is hypothesized that metabolism in humanized mice may combine the activity of all human liver enzymes (including CYPs and AO) and transporters and thereby provide a more complete picture of human metabolism than could be achieved by using in vitro assays or conventional in vivo animal models. Third, the pharmacokinetics of Lu AF09535 was assessed in cynomolgus monkeys, because this species carries AO and thereby stands as a potentially more predictive species for human on AO-mediated drug metabolism. In this work, various in vitro and in vivo approaches were assessed and applied to elucidate the role of AO during the extensive metabolism of Lu AF09535. Furthermore, different strategies were evaluated to integrate the methods for predicting human metabolic clearance of AO substrates.

Materials and Methods

14C Microdosing Study in Healthy Male Volunteers

The clinical study was given favorable ethical approval by Medisch Ethische ToetsingsCommissie, Assen, as the Independent Ethics Committee, and the Centrale Commissie Mensgebonden Onderzoek (CCMO) as Competent Authority for clinical trials in the Netherlands (NL39565.056.12, Eudract number 2011-002925-23).

All participants were male healthy subjects recruited through advertisements in local newspapers by QPS Netherlands B.V., Groningen, the Netherlands. All participants underwent a physical and neurologic examination including electrocardiogram, hematologic screen, urinalysis, and assessment of vital signs and relevant medical history. Negative urinalysis confirmed no recreational substance use. Participants tested negative for alcohol with breathalyzers.

Lu AF09535 was administered to the participants in a randomized, double-blind, sequential-group, placebo-controlled, single ascending oral dose study in healthy young men. Lu AF09535 was administered to participants in cohorts of 9 (Lu AF09535:PLB, 6:3) in a sequential design (cohort 1, 3 mg; cohort 2, 15 mg; cohort 3+4, 75 mg). Collection of samples includes 6 Lu AF09535 treated subjects in cohort 1–3, and 4 Lu AF09535 treated in cohort 4. In cohort 4, low dose (263 nCi) of radiolabeled compound, [14C]-Lu AF09535 was coadministered with 75 mg of Lu AF09535. Plasma, urine, and feces samples were collected from all subjects (4 subjects received Lu AF09535). Plasma samples were taken at 17 time points up to 96 hours postdose, whereas excreta were collected quantitatively in 24-hour time intervals up to 96 hours postdose. Fecal samples were homogenized in sterile water (1:2 w/v). Study samples were pooled across subjects to produce a single sample pool for each time point in each matrix. All samples were graphitized and submitted for AMS analysis for total drug-related material using a National Electrostatics Corporation 1.5 SDH Compact AMS system. For metabolite profiling, study samples were subsequently pooled across time points to produce a single pooled sample per matrix. The pooled samples represented all Lu AF09535-treated subjects across the following time intervals: plasma 0.5–36 hours, urine 0–48 hours, and fecal homogenate 0–72 hours postdose. Plasma samples were pooled according to the method proposed by Hamilton et al. (1981), whereas urine and faces were pooled using a fixed percentage of the total volume collected.

Metabolite profiling of the pooled samples were obtained by HPLC fractionation using a Shimadzu Prominence HPLC system equipped with a Phenomenex Synergy Polar-RP column (4.6 × 250 mm, 4 μm) and a fraction collector Shimadzu, Columbia, MD; Phenomenex, Macclesfield, Cheshire, UK. A gradient 67 minute system was run with mobile phases consisting of 20 mM ammonium formate (pH = 7) and acetonitrile. Injection volume was 100 μl and flow rate 1 ml/min. Fractions were collected at the rate of 1 fraction per minute for a total of 59 fractions. Each fraction was submitted for AMS analysis. In parallel, selected plasma, urine and feces pools were analyzed for metabolite identification by liquid chromatography tandem mass spectroscopy (LC-MS/MS) using the same HPLC method as outlined above on a Finnegan TSQ Quantum Ultra mass spectrometer coupled to a Thermo Accela HPLC system with an ARC Radio-LC system (Thermo Fisher, Waltham, MA). Human metabolites were characterized with regards to molecular mass, and, when possible, structural information was assigned to the metabolites.

Metabolism in Human Liver Cytosol

Human liver cytosol fractions were obtained from Tebu-bio, (Copenhagen, Denmark). The batch contained a pool of liver cytosol from 50 donors (mixed sex) and the initial protein content was 10 mg/ml. [14C]-Lu AF09535 1 µM was incubated in human liver cytosol with and without AO inhibitor (menadione 10 µM or raloxifene 1 µM) to investigate the metabolism of Lu AF09535 by AO. In parallel, four compounds known to be substrates of AO (zoniporide, O6-benzylguanine, carbazeran and zaleplon obtained from Sigma-Aldrich (St. Louis, MA) were incubated as comparators to rank the extent of metabolism by AO. Lu AF09535 and the comparators were incubated in duplicates in human liver cytosol mix with final assay containing 1 mg/ml protein, 4 mM MgCl2, 4 mM NADPH and 100 mM sodium phosphate buffer pH 7.4 in purified water. Test concentration was 1 µM for all compounds, and incubation temperature was 37°C. Incubation time was chosen to be 30 minutes, at which time approximately 80% of Lu AF09535 had been metabolized in the incubation conditions used in this study. The cytosol-mix was preincubated for ∼5 minutes under continuous shaking in a CO2 incubator [95%/5% (air/CO2)] and, when relevant, an inhibitor was added. After preincubation, Lu AF09535 or comparators were added to the cytosol mix. T0 samples were immediately terminated using stopping reagent (MeCN), whereas the remaining samples were incubated 30 minutes (T30) under continuous shaking in the CO2 incubator at 37°C and then terminated using stopping reagent (MeCN). All samples were centrifuged and the supernatant analyzed using LC-MS/MS (MRM transitions Lu AF09535 407 > 270; Zoniporide 321 > 262; O6-Benzylguanine 242 > 91; Carbazeran 361 > 218) with online radio detection (method described above). Samples were analyzed for Lu AF09535 and selected metabolites.

Simcyp Model for AF09535

The following data were used to build the Simcyp (ver. 11.01) compound file for Lu AF09535: molecular weight 406 g/mol, monoprotic base pKa 0.5, LogP 2.4, fraction unbound in human plasma 0.147, and blood/plasma ratio 1.0. A first-order absorption kinetics was assumed with the following absorption parameters: fraction unbound in gut of 0.5, Madin-Darby canine kidney permeability 38.4 × 10−6 cm/s with a reference value in the same system for propranolol of 23 × 10−6 cm/s. The minimal physiologic based pharmacokinetic model was used with a volume of distribution at steady state of 1.27 l/kg using prediction method 1 (Simcyp). For elimination, an intrinsic clearance of 42.6 µl/min/106 hepatocytes was entered in the whole organ metabolic clearance option. The fraction unbound in the hepatocytes is not known, but as a substitute a value of 0.856 was used, originating from the build-in calculator of unbound fraction in microsomal incubations from physicochemical properties using 0.5 mg/ml microsomal protein.

Pharmacokinetics in Humanized Mice

The pharmacokinetics of Lu AF09535 was assessed in chimeric mice with humanized livers (PXB mice) obtained from PhoenixBio Co, Ltd, Higashi-Hiroshima City, Japan, and severe combined immunodeficiency (SCID) mouse obtained from Charles Rivers Laboratories, Inc, (Yokohama, Japan). The PXB-mouse strain was generated using urokinase-type plasminogen activator (uPA+/+)/ SCID mice repopulated with human hepatocytes (h-PXB mice). The study included a total of four male PXB mice with an estimated liver replacement index of 70% or more (based on blood concentration of human serum albumin). Four male SCID mice were included in the pharmacokinetic study as controls. All animals were 10–14 weeks of age at the time of the study and had a body weight of ≥15.6 g. Each animal was given a single oral dose of 25 mg/kg Lu AF09535 (in 20% hydroxy propyl-β-cyclodextrin). Blood samples (50 μl) were drawn at 10, 30, 60, and 120 minutes postdose from two animals in each group and at 20, 45, 90, and 180 minutes postdose from the two other animals (n = 2 per time point) to provide a full plasma time profile up to 3 hours postdose. The individual blood samples were transferred to K2EDTA tubes and centrifuged at 1000 g in 4°C for 10 minutes and 20 μl plasma was transferred to a microtube. Before bioanalysis, plasma samples drawn from two animals at each time point were pooled. Samples were analyzed for Lu AF09535 and its metabolites by UPLC-MS/MS as described below.

Pharmacokinetics in Monkeys

The pharmacokinetics of Lu AF09535 was assessed in male Cynomolgus monkeys (Macaca fascicularis, 3–7 kg, Charles River Laboratory, Harlow, UK) after intravenous and oral dosing in a cross over design (n = 3). Animals were fasted overnight before each dose administration and offered standard laboratory diet 1 hour post dose administration and mains quality tap water was available ad libitum. At the first day animals received a slow (30 seconds) intravenous bolus dose of 0.25 mg/kg via the tail vein followed by a 2 ml saline flush. Blood samples (0.5 ml) were taken at 5, 15, and 30 minutes and 1, 2, 4, 6, 8, and 24 hours. After 1 week of washout, the animals received an oral dose of 0.5 mg/kg via gastric gavage, and blood samples were collected at 20 and 40 minutes and 1, 2, 3, 4, 6, 8, and 24 hours. For both intravenous and oral administrations, Lu AF09535 was dosed in solution using 10% hydroxy propyl-β-cyclodextrin, pH 4 in a volume of 1 ml/kg. All samples were withdrawn from the femoral vein and collected into chilled, EDTA-coated tubes, centrifuged for 10 minutes (3000 rcf, 4°C), and the resulting plasma was stored at −80°C until analysis. The procedures involving the care or use of animals in this study were approved by the animal welfare committee at the CRO before the initiation of the study, and all animal procedures were carried out in compliance with EC Directive 86/609/ EEC. Plasma concentrations of Lu AF09535 were measured by LC-MS/MS as described below.

Bioanalysis

Plasma samples from pharmacokinetic (PK) studies in mice, monkeys, and humans were analyzed using a UPLC-MS/MS system consisting of a Waters Acquity UPLC coupled to a Xevo TQS mass spectrometer (Milford, MA), operated by MassLynx software version 4.1. The mass spectrometer electrospray ion source was operated in the positive ion mode and detection of the ions was performed in the multiple reaction monitoring (MRM) mode. A deuterium-labeled analog of Lu AF09535 was used as internal standard. The Acquity UPLC system was equipped with an Acquity UPLC BEH C18, 50 × 2.1 mm, 1.7 µm analytical column and the column temperature 60°C. Mobile phases were A: 0.1% formic acid/acetonitrile (95:5); B: 0.1% formic acid/acetonitrile (5:95). The system was run in gradient mode (0–1 minute 40% B, 1–1.5 minutes 100% B, 1.5–2 minutes 40% B) at a flow rate of 0.6 ml/min and a total run time of 2 minutes. For analysis of hydroxy metabolites in plasma from humanized mice, a slightly modified gradient was applied.

Calibration standards and quality control samples were prepared using a Hamilton Microlab STARlet liquid handling robot (Hamilton Robotics, Bonaduz, Switzerland). Plasma samples (calibration standards, quality controls, and study samples) were protein precipitated by addition of 250 µl acetonitrile to 50 µl plasma followed by thorough mixing and centrifugation. One hundred microliters supernatant was transferred to another 96-well plate and added 300 µl mobile phase A. The final sample (7.5 µl) was injected onto the UPLC-MS/MS system. Calibration range was 0.200–200 ng/ml.

Data Analysis

All pharmacokinetic parameters were calculated using WinNonlin 5.2 (Pharsight Corporation, St. Louis, MO), using compartmental modeling for intravenous dosing and noncompartmental analysis for the oral data. Normalization of the radioactive dose (263 nCi) to the total dose of 75 mg per subject [total drug related material (DRM) given as nanogram equivalent per milliliter was calculated from the radioactive concentration determined by AMS analysis (dpm/ml) and a specific activity of Lu AF09535 of 130 ng/dpm]:

Embedded Image

Scaling of Aldehyde Oxidase Metabolism in Human Liver Cytosol

The turnover of Lu AF09535 and comparators in human liver cytosol were calculated as the percentage of compound remaining after 30 minutes incubation using the analytical peak areas from LC-MS/MS chromatograms (comparators) or radiochromatograms (Lu AF09535) for 0 minutes incubation (Acompound,T0) and 30 minutes incubation (Acompound,T30), respectively.

Percent turnover is calculated using the following equation (where compound = Lu AF09535, Carbazeran, Zaleplon, Zoniporiden, or O6-benzylguanine):Embedded ImageThe turnover of Lu AF09535 was compared with the AO metabolized comparator compounds (Zaleplon, Zoniporide, O6-benzylguanine, and Carbazeran) with low to high turnover to rank the compound.

Single Species Allometric Scaling of Lu AF09535 in Humanized Mouse

The single species allometric scaling approach proposed by Sanoh et al. (2012b) was applied for prediction of the human clearance of Lu AF09535, using the pharmacokinetic parameters obtained from the PK study in humanized mice (PXB-mice). Sanoh et al. (2012b) investigated the predictability of hepatic intrinsic clearance of 13 clinical drugs, by “single species allometric scaling (SSS)” using data from intravenous administration of 13 drugs to PXB mice. The 13 drugs represented substrates of a wide range of drug metabolism enzymes (including aldehyde oxidase) and covers human in vivo clearance ranging from 0.055 to 118 ml/min⋅kg. Direct correlation of human in vivo total clearance (CLhuman) to in vivo total clearance of PXB mice (CLPXB mice) was found to be relatively poor. However, when applying a scaling exponent (e), a value of 0.840 achieved a good correlation with 85% of the compounds within threefold of actual values for total CLhuman. The equation below is used for SSS, where BWhuman, is set to 70 kg and BWPXB mice is 0.02 kg. For the evaluation of the scaling in the current experiments, oral CL (CL/F) was applied in both mice and humans:

Embedded Image

Results

14C Microdosing Study in Healthy Male Volunteers

In the first-in-man study, the plasma exposure of Lu AF09535 was negligible in the three initial cohorts with doses of 3, 15, and 75 mg, respectively, reaching a maximum plasma concentration of less than 4 ng/ml in subjects dosed 75 mg, which was substantially lower than predicted.

In the microdose cohort, a radioactive dose of [14C]-Lu AF09535 corresponding to 263 nCi was administered to four subjects together with a single dose of 75 mg Lu AF09535. Total radioactivity for drug-related material (DRM) in plasma, urine, and feces samples was measured for the full time profile by AMS. The plasma time profile for total DRM (sum of Lu AF09535 and its metabolites) is shown together with the corresponding data from conventional bioanalysis of Lu AF09535 in equivalent plasma samples in Fig. 1. For total DRM, the maximum plasma concentration was reached at 1.5 hours postdosing followed by a relatively slow decline (T1/2 = 13.5 hours), and at 96 hours postdose, considerable amounts of DRM was still detectable (approximately 10 ng Eq/ml).

Fig. 1.
Fig. 1.

Total Lu AF09535 drug-related material (DRM) and corresponding exposure of Lu AF09535 in plasma collected from healthy male subjects (n = 4) after a single oral administration of 75 mg free base (263 nCi) in the 14C microdosing study in humans.

When normalizing the radioactive dose to the total dose of 75 mg per subject, an exposure (AUC0–96 hours) of 22.100 hours×ng Eq/ml in plasma were estimated for total DRM and Cmax was 1500 ng/ml. In comparison, data obtained by conventional bioanalysis of Lu AF09535, an exposure (AUC0–96 hours) of 1.0 hour×ng/ml (CL/F = 75,000 l/h) were calculated for parent compound and Cmax was 1.7 ng/ml (illustrated in Fig. 1). These data show that only negligible amount of the circulating DRM was parent compound (Lu AF09535) and that the parent compound was subject to extensive (first pass) metabolism because high amounts of circulating DRM appeared to consist of Lu AF09535 metabolites. Furthermore, total radioactivity for drug related material in urine and feces was measured for the full-time profile up to 96 hours, and results are shown in Table 1. Approximately 10% of DRM is excreted in urine and 80% excreted in feces. The majority (40%) of DRM was excreted in feces in the 24- to 48-hour time interval. Metabolite profiling in human plasma, urine and feces are shown in Table 2. Results are given as percent of total radioactivity in individual matrices (pooled samples) and as percent of total dose based on 10% recovery in urine and 70% recovery in feces. Lu AF09535 was metabolized to a total of six metabolites, with overall similar metabolite profiles in all three matrices. Virtually no radioactivity was recovered as unchanged Lu AF09535 in neither urine nor feces, indicating that the compound was fully absorbed and the low human exposure was caused by extensive first-pass metabolism, potentially including gut metabolism.

View this table:
TABLE 1

Mean cumulative recovery of total drug related material in pooled urine and fecal homogenate collected from healthy male subjects (n = 4) after a single oral administration of 75 mg free base (263 nCi) of [14C]-Lu AF09535

View this table:
TABLE 2

Lu AF09535 Metabolite profiling in human plasma, urine, and feces (pooled samples) given as % total radioactivity in individual matrices

Numbers in brackets are % total dose based on 10% recovery in urine and 70% recovery in feces

A schematic overview of the metabolic pathway of Lu AF09535 is shown in Fig. 2. All metabolites correspond to mono-, di-, or trihydroxylations of the parent compound and their glucuronide conjugates. The structure of one mono-hydroxymetabolite (M1) was verified using a synthetic standard to be an oxidation in the pyrimidine moiety. For all other metabolites the structures have not been verified. The overall dominant metabolite was a dihydroxy metabolite, which was mainly excreted in feces and corresponds to approximately 48% of the total radioactive dose administered to subjects. The mass spectrometry fragmentation pattern indicates that one hydroxylation is in the pyrimidine moiety and one in the pyrazine moiety, respectively, corresponding well to typical AO-mediated metabolism (Alfaro and Jones, 2008)

Fig. 2.
Fig. 2.

Suggested metabolic pathway for Lu AF09535.

Metabolism in Human Liver Cytosol

Hydroxylation of the pyrimidine and pyrazine moiety could be indicative of aldehyde oxidase metabolism (Pryde et al., 2010). Incubations of Lu AF09535 in human liver cytosol with and without AO-specific inhibitors (menadione and raloxifene) showed that stability of Lu AF09535 after 30 minutes was 33% without inhibitor and 64% and 55% with addition of menadione and raloxifene, respectively. The reason for the incomplete inhibition is unknown but could result from suboptimal incubations conditions such as too high substrate concentration (Km of Lu AF09535 for AO is not known), likewise metabolism by other unidentified enzymes in the cytosol preparation cannot be excluded. Although complete inhibition of Lu AF09535 metabolism was not shown, this data confirmed that AO metabolism was an important pathway in human liver cytosol.

A study was performed to achieve a scaling of the extent of AO-mediated metabolism of Lu AF09535 according to the “yardstick” approach represented in a publication by Zientek et al. (2010), in which an attempt has been made to provide a “ranking order” for 11 AO substrates, ranking from low to very high clearance using in vivo clearance data. For this study Carbazeran, O6-benzylguanine, Zoniporide, and Zaleplon were chosen as AO substrates to cover medium to very high AO metabolism. Possible concentration and/or time dependency on the rates and extent of metabolism for any of the compounds was not investigated in this study; however, results obtained showed that the order for clearance of the four AO substrates correlated to the results by Zientek et al. (2010), i.e., Carbazeran > O6-benzylguanine > Zoniporide > Zaleplon.

In Table 3 results are shown for percent turnover obtained in this study for Lu AF09535 and the four AO substrates. Lu AF09535 was metabolized much faster than O6-benzylguanine, Zoniporide, and Zaleplon, and almost as fast as Carbazeran, and is thereby categorized as a high level substrate of AO. Table 3 also shows the in vivo clearance data for the four AO substrates, including AO intrinsic clearance ≈13,000 ml/min × kg for Carbazeran. The results of this study indicate that the in vivo clearance of Lu AF09535 may be in the same order of magnitude as Carbazeran and thereby confirm that the observed extensive human in vivo clearance of Lu AF09535 could be largely attributed to AO metabolism. The metabolic pathway of Lu AF09535 in human liver cytosol corresponded to the pathways observed in vivo in human, i.e., primarily via M1 (monohydroxy metabolite) to a dihydroxy metabolite, suggesting that this study in human liver cytosol is a good predictor of in vivo human metabolism.

View this table:
TABLE 3

Stability of Lu AF09535 and four AO substrates after 30-min incubation in human liver cytosol (%) and in vivo intrinsic clearance data for the AO substrates

Pharmacokinetics in Humanized Mice

It was investigated whether the metabolism of Lu AF09535 in chimeric mice with humanized livers (PXB mice) was predictive of the observed extensive human metabolism. Humanized mice contain the human type AO enzyme and it was hypothesized that metabolism in humanized mice may combine the activity of all human liver enzymes and transporters (Sanoh et al., 2012a) and thereby provide a more complete picture of human metabolism than achieved in in vitro assays or conventional in vivo animal models.

The results from analysis of Lu AF09535 and selected metabolites (mono- and dihydroxy) in plasma samples from PXB (humanized) and SCID (control) mice are shown in Table 4 and plasma time profiles in Fig. 3. Lu AF09535 was very rapidly absorbed in both humanized and control mice, with Tmax at 10 minutes (the first sampling time point). In control mice, Lu AF09535 reached a relatively high Cmax of 4000 ng/ml, whereas a maximum plasma concentration of 1190 ng/ml was reached in the humanized mice. At the last sampling time point (3 hours), the concentration of Lu AF09535 in humanized mice had declined to less than 1% of Cmax, whereas in the control mice the concentration was almost 10% of Cmax (388 ng/ml). Exposure calculated as AUC0–3 hours showed that Lu AF09535 was approximately 8 times higher in control mice (AUC0–3 hours = 4310 hours×ng/ml) compared with humanized mice (AUC0–3 hours = 537 hours×ng/ml). Furthermore, analysis of metabolites generally showed a significantly higher exposure of M1 and M2 in humanized mice compared with parent, indicating a significantly higher rate of metabolism in the humanized mice compared with control SCID mice. Particularly the exposure of M2a (dihydroxy metabolite that is oxidized in both the pyrimidine and pyrazine moiety of the molecule) in PXB mice differed significantly from control mice as only negligible concentrations were detected in SCID. This metabolism pathway corresponds well with the pathway observed in human and with aldehyde oxidase metabolism. This supports that the humanized mice model is a valuable tool for prediction of human metabolism.

View this table:
TABLE 4

Plasma exposures (AUCs) of Lu AF09535 and its metabolites in humanized (PXB) and control mice (SCID) after oral dosing of 25 mg/kg

Fig. 3.
Fig. 3.

Plasma time profiles for Lu AF09535 in humanized mice (PXB) and control mice (SCID) (n = 2).

An attempt was made to predict the human clearance of Lu AF09535, using the clearance data obtained from the chimeric mice with humanized liver (PXB mice) and single species allometric scaling (Sanoh et al., 2012b). The oral clearance of Lu AF09535 in PXB mice was 15.5 ml/min and when applying the SSS exponent of 0.840 an in vivo total oral clearance in human (CL/FL) was predicted to be 14.7 l/min or 883 l/hour. This result obviously predicts a very high clearance in humans, however, not in the same order of magnitude as the actual oral clearance observed in the FIM study (CL/F ∼75,000 l/hour).

Pharmacokinetics in Monkeys

Individual plasma-concentration time courses of Lu AF09535 after intravenous and oral administration in three Cynomolgus monkeys are shown in Fig. 4. After intravenous dosing of 0.25 mg/kg, a two-compartment model with first-order elimination best described the plasma-concentration time course data. Lu AF09535 displayed low average systemic clearance of 1.2 ± 0.07 l/hour/kg relative to the liver blood flow (2.6 l/hour/kg) and a low volume of distribution of 0.9 ± 0.08 l/kg. After oral administration of 0.5 mg/kg, plasma concentrations were only detectable in one subject (Fig. 4) with a Cmax of 13 ng/ml. For this subject, plasma concentrations fell below the detection limit 2 hours post drug administration, and its oral bioavailability was calculated to 5% from the respective extrapolated plasma AUCs after oral and intravenous dosing. For the two other subjects, oral bioavailability was even lower, and AUC could not be calculated because of lack of exposure after oral route of administration in these animals.

Fig. 4.
Fig. 4.

Individual plasma-concentration-time profiles of Lu AF09535 after intravenous (●) and oral (○) administration in three Cynomolgus monkeys. Two subjects in the oral treatment leg displayed exposures lower than the detection limit (1 ng/ml).

Discussion

The low exposure of Lu AF09535 observed in the FIM study obviously raised numerous questions regarding the reasons for these findings, e.g., incorrect dose strength of oral solution administered to the participants, vehicle effects, potential drug precipitation, lack of absorption, bioanalytical errors, etc., which were not answerable from the available data. The 14C-microdose part of the FIM study provided the following valuable answers. Lu AF09535 appeared to be well absorbed as negligible parent compound was detected in feces (<1% excreted as unchanged compound in feces). In plasma, AMS analysis revealed distinctly higher exposure of total drug-related material than what could be accounted for by the corresponding conventional bioanalysis of parent compound. Hence, it was clear that significant amount of drug-related material was absorbed but metabolized before reaching systemic circulation. Furthermore, as approximately 80% of drug-related material was recovered in excreta within 96 hours, the compound did not appear to extensively accumulate in nonplasma compartments of the body. Metabolite profiling showed similar profiles in plasma, urine, and feces. Lu AF09535 was also not detected in plasma of 14C-microdose study, confirming results obtained by conventional bioanalysis. Metabolite characterization in all three matrices detected metabolites corresponded to mono-, di-, and trihydroxylations, and the positions of oxidation generally corresponded to biotransformation products of AO-mediated metabolism.

As part of the pre-FIM discovery/development package, metabolic stability is measured in vitro in liver microsomes and hepatocytes from rats, dogs, and humans used for prediction of human clearance. Initially all in vitro data showed very low clearance in agreement with the in vivo animal PK investigations in rats and dogs (data not shown). After the lack of exposure observed in the FIM studies, the human hepatocyte intrinsic clearance assay was repeated using a new pool of hepatocytes. In that study, Lu AF09535 was found to be metabolically unstable with a high intrinsic clearance (7.6 l/min relative to the liver blood flow of 1.4 l/min). The reason for the differences in metabolic stability between studies has not been clarified, but it could be speculated that hepatocyte isolation procedures and storage, handling in the laboratory, polymorphism, etc. or all could have contributed to such interstudy differences (Hutzler et al., 2012; Zou et al., 2012; Hutzler et al., 2014). Importantly however, using the high human hepatocyte intrinsic clearance as input for predicting human oral clearance, via the physiologically based pharmacokinetic simulation program Simcyp, a high oral clearance of 95 l/h was predicted, while still almost 800-fold lower than the observed human oral clearance of approximately 75,000 l/hour. Thus, high human hepatocyte intrinsic clearance alone was far from sufficient to predict the observed human clearance. Although human hepatocytes are known to harbor AO, large variability among donors exists, complicating the predictions of human pharmacokinetics of AO metabolized drug candidates. Even with appropriate controls and characterization of the hepatocyte cultures used, scaling of clearance from intrinsic clearance data in human hepatocytes using conventional methods, as the ones being used successfully for CYP metabolized compounds, may lead to considerable under predictions as the current example clearly shows. During the drug candidate selection phase, a relative high oral clearance of 95 l/hour would often cause concern for the developability of the compound, triggering a search for lower clearance compounds with similar pharmacological and safety profile. However, for AO metabolized compounds, even predictions of medium clearance (estimated from hepatocyte intrinsic clearance) could prove highly inadequate, because the gap between predicted and actual clearance may be dramatic. Appropriate scaling factors, knowledge of functional polymorphism, and inclusion of extrahepatic contribution to the total clearance of AO metabolized compounds are clearly needed to improve the clearance predictions for such compounds.

Aldehyde oxidase involvement was confirmed by incubations in human liver cytosol using the AO inhibitors raloxifene and menadione, showing pronounced inhibition of metabolism. Involvement of AO metabolism was consistent with the observed high clearance of Lu AF09535 in humans and low clearance and high bioavailability in rat, mouse, and dog, where AO activity is known to be low or nonexisting compared with human (Pryde et al., 2010). The stability in human blood was also investigated, as Lu AF09535 contains two amide bonds prone to potential hydrolysis, but the compound was shown to be stable for at least 2 hours. In addition, the potential cleavage products from likely hydrolysis were not detected in plasma, urine, or feces using synthetic standards of the cleavage products, i.e., no radioactive peaks coeluted (data not shown).

In vitro-in vivo scaling approaches using human liver microsomes and hepatocytes have proven reasonably successful for CYP-metabolized compounds, whereas AO substrates have proven to be much more challenging and somewhat arbitrary scaling factors of 8–15 have been applied with some success (Zientek et al., 2010; Akabane et al., 2012). In the present case study, even higher scaling factors are needed to bridge the predicted versus observed difference. However in vitro-in vivo scaling using tissue from hepatic origin only may not be sufficient to account for AO metabolism, because AO is also expressed in extrahepatic tissue (Moriwaki et al., 2001). Currently, the lack of knowledge of species differences in catalytic activity of AO, expression levels in various tissues, and how to scale these intrinsic clearances from extrahepatic tissues constitutes a great challenge for obtaining accurate predictions of human clearance of highly metabolized AO substrates. The yardstick approach proposed by Zientek et al. (2010) has become a widely adapted method for prediction of in vivo human metabolism because it appears to provide a reliable categorization of AO substrates into low, medium, and high clearance compounds. The results obtained in this study also clearly showed that Lu AF09535 ranked among the (very) high clearance substrates. Lu AF09535 was metabolized at a rate comparable to the highest clearance comparator drug Carbazeran [Cl/F ∼8100 l/h (Zientek et al., 2010)] and although the actual observed in vivo clearance was not adequately predicted, an application of the yardstick approach before the FIM study would unquestionably have been highly informative.

With regard to in vivo animal models, humanized mice (PXB-mice) may provide a more complete representation of the in vivo human metabolism because the humanized mouse offers the functionalities of the human liver. However, although this mice model expresses human liver AO, it still expresses the mouse AO extrahepatically. The pharmacokinetic study in humanized mice with Lu AF09535 showed substantially higher exposure and less conversion to metabolites in control mice compared with humanized mice, indicating increased rates of metabolism in the humanized mice. Prediction of in vivo human clearance using single species allometric scaling (SSS) from these mice as proposed by Sanoh et al. (2012b) predicted Lu AF09535 to be a high clearance compound with a predicted human oral clearance of approximately 900 l/h. Although high clearance was predicted, it was yet far from the actual oral clearance observed in the FIM study (75,000 l/h), which may indicate that extrahepatic metabolism are also involved in the clearance of Lu AF09535. In monkeys, a large presystemic clearance component was evident for Lu AF09535 because the oral bioavailability was very low despite its moderate systemic clearance after intravenous administration. Hence, assuming no limitations in the gastrointestinal absorption, the monkey may serve as a confirmatory species for pharmacokinetic studies of AO substrates and thus prove useful for qualitative evaluation of potential AO substrates, assuming similar expression and substrate specificity between human and monkey in general.

Specifically for Lu AF09535, none of the scaling methods used here were able to predict at least within twofold of the extreme clearance measured in healthy subjects. Thus, we might have overlooked additional metabolizing enzyme(s) or missed a specific human gut bacterial-based metabolism, forming the same phase one metabolites, but so far we have not investigated this. As previously described, the conventional approach for prediction of human clearance using in vitro data obtained from human hepatocytes and/or other human subcellular fractions such as cytosol S9 or using in vivo pharmacokinetic data from preclinical species (rat and dog) does not provide adequate prediction for AO substrates. This issue has led to several cases where the drug development process of a new chemical entity has been terminated due to unexpected low human exposure (Pryde et al., 2010; Akabane et al., 2011; Zhang et al., 2011). This includes the development of Lu AF09535 as presented in this publication. To evaluate if it would be beneficial to include additional preclinical testing of human metabolism before the FIM study, we therefore evaluated different in vitro and in vivo approaches for prediction of AO metabolism in human. Among in vitro methods, the yardstick approach using human liver cytosol and human hepatocytes with documented catalytic AO activity seems to be the best option for screening large numbers of compounds for AO-dependent metabolism. The two in vivo methods, humanized mouse and the cynomolgus monkey, clearly identified the potential for low exposure found in humans and should be considered for compounds suspected of being significantly metabolized by AO.

Authorship Contributions

Participated in research design: Jensen, Jacobsen, Bundgaard, Nilausen, Thale, Chandrasena, and Jørgensen.

Conducted experiments: Jacobsen, Bundgaard, Nilausen, and Chandrasena.

Performed data analysis: Jensen, Jacobsen, Bundgaard, Nilausen, Chandrasena, and Jørgensen.

Wrote or contributed to the writing of the manuscript: Jensen, Jacobsen, Bundgaard, Nilausen, Thale, Chandrasena, and Jørgensen.

Contributed new reagents or analytic tools: Thale.

Footnotes

Abbreviations

AMS
accelerator mass spectrometry
AO
aldehyde oxidase
AUC
area under the curve
CYP
cytochrome P450
DRM
drug-related material
FIM
first-in-man
LC-MS/MS
liquid chromatography tandem mass spectroscopy
MIST
metabolites in safety testing
PK
pharmacokinetic
PXB mouse
Phoenix Bio chimeric mouse with humanized liver
SCID
severe combined immunodeficiency
SSS
single species allometric scaling

References

  1. ↵
    1. Akabane T,
    2. Tanaka K,
    3. Irie M,
    4. Terashita S, and
    5. Teramura T
    (2011) Case report of extensive metabolism by aldehyde oxidase in humans: pharmacokinetics and metabolite profile of FK3453 in rats, dogs, and humans. Xenobiotica 41:372–384.
    OpenUrlCrossRefPubMed
  2. ↵
    1. Akabane T,
    2. Gerst N,
    3. Masters JN, and
    4. Tamura K
    (2012) A quantitative approach to hepatic clearance prediction of metabolism by aldehyde oxidase using custom pooled hepatocytes. Xenobiotica 42:863–871.
    OpenUrlCrossRefPubMed
  3. ↵
    1. Alfaro JF and
    2. Jones JP
    (2008) Studies on the mechanism of aldehyde oxidase and xanthine oxidase. J Org Chem 73:9469–9472.
    OpenUrlCrossRefPubMed
  4. ↵
    1. Diamond S,
    2. Boer J,
    3. Maduskuie TP Jr.,
    4. Falahatpisheh N,
    5. Li Y, and
    6. Yeleswaram S
    (2010) Species-specific metabolism of SGX523 by aldehyde oxidase and the toxicological implications. Drug Metab Dispos 38:1277–1285.
    OpenUrlAbstract/FREE Full Text
  5. ↵
    1. Garattini E,
    2. Fratelli M, and
    3. Terao M
    (2008) Mammalian aldehyde oxidases: genetics, evolution and biochemistry. Cell Mol Life Sci 65:1019–1048.
    OpenUrlCrossRefPubMed
  6. ↵
    1. Hamilton RA,
    2. Garnett WR, and
    3. Kline BJ
    (1981) Determination of mean valproic acid serum level by assay of a single pooled sample. Clin Pharmacol Ther 29:408–413.
    OpenUrlCrossRefPubMed
  7. ↵
    1. Hutzler JM,
    2. Yang YS,
    3. Albaugh D,
    4. Fullenwider CL,
    5. Schmenk J, and
    6. Fisher MB
    (2012) Characterization of aldehyde oxidase enzyme activity in cryopreserved human hepatocytes. Drug Metab Dispos 40:267–275.
    OpenUrlAbstract/FREE Full Text
  8. ↵
    1. Hutzler JM,
    2. Yang YS,
    3. Brown C,
    4. Heyward S, and
    5. Moeller T
    (2014) Aldehyde oxidase activity in donor-matched fresh and cryopreserved human hepatocytes and assessment of variability in 75 donors. Drug Metab Dispos 42:1090–1097.
    OpenUrlAbstract/FREE Full Text
  9. ↵
    1. Kitamura S,
    2. Nakatani K,
    3. Ohashi K,
    4. Sugihara K,
    5. Hosokawa R,
    6. Akagawa Y, and
    7. Ohta S
    (2001) Extremely high drug-reductase activity based on aldehyde oxidase in monkey liver. Biol Pharm Bull 24:856–859.
    OpenUrlCrossRefPubMed
  10. ↵
    1. Kitamura S,
    2. Nitta K,
    3. Tayama Y,
    4. Tanoue C,
    5. Sugihara K,
    6. Inoue T,
    7. Horie T, and
    8. Ohta S
    (2008) Aldehyde oxidase-catalyzed metabolism of N1-methylnicotinamide in vivo and in vitro in chimeric mice with humanized liver. Drug Metab Dispos 36:1202–1205.
    OpenUrlAbstract/FREE Full Text
  11. ↵
    1. Lappin G,
    2. Seymour M,
    3. Gross G,
    4. Jørgensen M,
    5. Kall M, and
    6. Kværnø L
    (2012) Meeting the MIST regulations: human metabolism in Phase I using AMS and a tiered bioanalytical approach. Bioanalysis 4:407–416.
    OpenUrlCrossRefPubMed
  12. ↵
    1. Lappin G and
    2. Seymour M
    (2010) Addressing metabolite safety during first-in-man studies using ¹⁴C-labeled drug and accelerator mass spectrometry. Bioanalysis 2:1315–1324.
    OpenUrlCrossRefPubMed
  13. ↵
    1. Li G,
    2. Jørgensen M, and
    3. Campbell BM
    (2013) Metabotropic glutamate receptor 5-negative allosteric modulators for the treatment of psychiatric and neurological disorders (2009-July 2013). Pharm Pat Anal 2:767–802.
    OpenUrl
  14. ↵
    1. Moriwaki Y,
    2. Yamamoto T,
    3. Takahashi S,
    4. Tsutsumi Z, and
    5. Hada T
    (2001) Widespread cellular distribution of aldehyde oxidase in human tissues found by immunohistochemistry staining. Histol Histopathol 16:745–753.
    OpenUrlPubMed
  15. ↵
    1. Pryde DC,
    2. Dalvie D,
    3. Hu Q,
    4. Jones P,
    5. Obach RS, and
    6. Tran TD
    (2010) Aldehyde oxidase: an enzyme of emerging importance in drug discovery. J Med Chem 53:8441–8460.
    OpenUrlCrossRefPubMed
  16. ↵
    1. Sanoh S,
    2. Horiguchi A,
    3. Sugihara K,
    4. Kotake Y,
    5. Tayama Y,
    6. Ohshita H,
    7. Tateno C,
    8. Horie T,
    9. Kitamura S, and
    10. Ohta S
    (2012a) Prediction of in vivo hepatic clearance and half-life of drug candidates in human using chimeric mice with humanized liver. Drug Metab Dispos 40:322–328.
    OpenUrlAbstract/FREE Full Text
  17. ↵
    1. Sanoh S,
    2. Naritomi Y,
    3. Sato K,
    4. Kawamura A,
    5. Horiguchi A,
    6. Sugihara K,
    7. Tateno C,
    8. Horie T,
    9. Kitamura S, and
    10. Ohta S
    (2012b) Prediction of human pharmacokinetics by allometric scaling using using chimeric mice with humanized liver. 12th European ISSX Meeting; 2012 Jun 17–21.
  18. ↵
    1. Sugihara,
    2. K,
    3. Kitamura,
    4. S,
    5. Tatsumi, and
    6. K
    (1995) Strain differences of liver aldehyde oxidase activity in rats. Biochem Mol Biol Int 37:861–869.
    OpenUrlCrossRefPubMed
  19. ↵
    1. Tateno C,
    2. Yoshizane Y,
    3. Saito N,
    4. Kataoka M,
    5. Utoh R,
    6. Yamasaki C,
    7. Tachibana A,
    8. Soeno Y,
    9. Asahina K,
    10. Hino H,
    11. et al.
    (2004) Near completely humanized liver in mice shows human-type metabolic responses to drugs. Am J Pathol 165:901–912.
    OpenUrlCrossRefPubMed
  20. ↵
    1. Zhang X,
    2. Liu HH,
    3. Weller P,
    4. Zheng M,
    5. Tao W,
    6. Wang J,
    7. Liao G,
    8. Monshouwer M, and
    9. Peltz G
    (2011) In silico and in vitro pharmacogenetics: aldehyde oxidase rapidly metabolizes a p38 kinase inhibitor. Pharmacogenomics J 11:15–24.
    OpenUrlAbstract/FREE Full Text
  21. ↵
    1. Zientek M,
    2. Jiang Y,
    3. Youdim K, and
    4. Obach RS
    (2010) In vitro-in vivo correlation for intrinsic clearance for drugs metabolized by human aldehyde oxidase. Drug Metab Dispos 38:1322–1327.
    OpenUrlPubMed
  22. ↵
    1. Zou P,
    2. Yu Y,
    3. Zheng N,
    4. Yang Y,
    5. Paholak HJ,
    6. Yu LX, and
    7. Sun D
    (2012) Applications of human pharmacokinetic prediction in first-in-human dose estimation. AAPS J 14:262–281.
    OpenUrlPubMed
View Abstract
Back to top

In this issue

Download PDF
Article Alerts
Email Article
Citation Tools
Research ArticleArticle

Lack of Exposure in a FIM Study Due to Aldehyde Oxidase

Klaus Gjervig Jensen, Anne-Marie Jacobsen, Christoffer Bundgaard, Dorrit Østergaard Nilausen, Zia Thale, Gamini Chandrasena and Martin Jørgensen
Drug Metabolism and Disposition January 1, 2017, 45 (1) 68-75; DOI: https://doi.org/10.1124/dmd.116.072793
del.icio.us logo Digg logo Reddit logo Twitter logo CiteULike logo Facebook logo Google logo Mendeley logo

Jump to section