Clinical significance of CYP2C19 polymorphisms on the metabolism and pharmacokinetics of 11β‐hydroxysteroid dehydrogenase type‐1 inhibitor BMS‐823778

Abstract

Aims

BMS‐823778 is an inhibitor of 11β‐hydroxysteroid dehydrogenase type‐1, and thus a potential candidate for Type 2 diabetes treatment. Here, we investigated the metabolism and pharmacokinetics of BMS‐823778 to understand its pharmacokinetic variations in early clinical trials.

Methods

The metabolism of BMS‐823778 was characterized in multiple in vitro assays. Pharmacokinetics were evaluated in healthy volunteers, prescreened as CYP2C19 extensive metabolizers (EM) or poor metabolizers (PM), with a single oral dose of [¹⁴C]BMS‐823778 (10 mg, 80 μCi).

Results

Three metabolites (<5%) were identified in human hepatocytes and liver microsomes (HLM) incubations, including two hydroxylated metabolites (M1 and M2) and one glucuronide conjugate (M3). As the most abundant metabolite, M1 was formed mainly through CYP2C19. M1 formation was also correlated with CYP2C19 activities in genotyped HLM. In humans, urinary excretion of dosed radioactivity was significantly higher in EM (68.8%; 95% confidence interval 61.3%, 76.3%) than in PM (47.0%; 43.5%, 50.6%); only small portions (<2%) were present in faeces or bile from both genotypes. In plasma, BMS‐823778 exposure in PM was significantly (5.3‐fold, P = 0.0097) higher than in EM. Furthermore, total radioactivity exposure was significantly higher (P < 0.01) than BMS‐823778 exposure in all groups, indicating the presence of metabolites. M1 was the only metabolite observed in plasma, and much lower in PM. In urine, the amount of M1 and its oxidative metabolite in EM was 7‐fold of that in PM, while more glucuronide conjugates of BMS‐823778 and M1 were excreted in PM.

Conclusions

CYP2C19 polymorphisms significantly impacted systemic exposure and metabolism pathways of BMS‐823778 in humans.

What is Already Known about this Subject

• BMS‐823778 is a potent inhibitor of 11β‐hydroxysteroid dehydrogenase type‐1 and developed for the treatment of Type 2 diabetes.

• In early clinical trials, the plasma concentration profiles of BMS‐823778 were highly variable among individual subjects.

• Preliminary analysis indicated that BMS‐823778 had a higher exposure in one CYP2C19 poor metabolizer.

What this Study Adds

• In vitro assays demonstrated that CYP2C19 was a major enzyme involved in BMS‐823778 metabolism.

• The plasma exposure of BMS‐823778 was significantly higher in CYP2C19 poor metabolizers than in extensive metabolizers.

• In vitro to in vivo extrapolation under predicted the difference of BMS‐823778 plasma exposure between CYP2C19 poor metabolizers and extensive metabolizers.

Introduction

Glucocorticoids are critical regulators of intermediary metabolism and have dramatic effects on glucose and lipid homeostasis 1. In humans, the major glucocorticoids are the active cortisol and the inactive cortisone. 11β‐hydroxysteroid dehydrogenase type‐1 (11β‐HSD1) converts cortisone to cortisol in metabolic tissues such as liver and adipose, while 11β‐hydroxysteroid dehydrogenase type‐2 (11β‐HSD2) deactivates cortisol to cortisone 2. Increased activity of 11β‐HSD1 in metabolic tissues can increase intracellular levels of active glucocorticoids, and consequently lead to metabolic changes such as insulin resistance and hyperglycaemia, dyslipidaemia and adipose tissue redistribution 3. Therefore, inhibition of 11β‐HSD1 could be a potential therapeutic target for the treatment of Type 2 diabetes. Increased expression of 11β‐HSD1 in adipose tissue of obese and insulin resistant humans has been observed in many studies 4, 5. A Phase 2 study demonstrated a significant 0.6% decrease in hemoglobin A1c together with a significant decrease in total cholesterol in subjects with type 2 diabetes mellitus treated with the 11β‐HSD1 inhibitor INCB13739 (200 mg) for three months 6.

BMS‐823778 is a potent, selective, competitive and fully reversible inhibitor of human 11β‐HSD1 7. Cell free biochemical assays showed that BMS‐823778 was more potent against human 11β‐HSD1 than the animal isoforms, and was 10 000‐fold selective for human 11β‐HSD1 over 11β‐HSD2. Ex vivo analysis of 11β‐HSD1 activity in murine adipose tissue showed a dose‐dependent inhibition 3 h after oral administration of BMS‐823778. Dose‐dependent whole‐body pharmacodynamic inhibition of 11β‐HSD1 was observed in cynomolgus monkeys following single oral doses from 0.1 to 10 mg kg^–1 8.

BMS‐823778 has been investigated in a combined single ascending dose and multiple ascending dose study in healthy male subjects (Study MB121001), in which the plasma exposures of BMS‐823778 were found to be highly variable among individual subjects (unpublished results). A follow‐up study was conducted in healthy Chinese subjects resident in the USA (Study MB121006). It was found that the only subject who had the highest plasma exposure and lowest apparent clearance is a CYP2C19 poor metabolizer (*2/*3; unpublished results). However, the detailed metabolism pathways of BMS‐823778 in humans are still not clear. Therefore, in this study the metabolism profiles of BMS‐823778 were evaluated using in vitro assays, and more importantly the metabolism and pharmacokinetics of BMS‐823778 were investigated in 14 healthy male volunteers who had been prescreened for CYP2C19 genotypes.

Materials and methods

Chemicals and biological reagents

BMS‐823778, M1 (hydroxylated metabolite of BMS‐823778), [¹⁴C]BMS‐823778 (72.5 μCi/mg) and [¹³C₂, D₆]BMS‐823778 were supplied by the Department of Chemistry, Bristol‐Myers Squibb (Princeton, NJ, USA) 9. β‐nicotinamide adenine dinucleotide phosphate‐reduced form (β‐NADPH), uridine 5′‐diphospho‐glucuronic acid (UDPGA) and alamethicin were obtained from Sigma‐Aldrich Co. (St Louis, MO, USA). Cryopreserved human hepatocytes were purchased from Celsis (Baltimore, MD, USA). Pooled human liver microsomes (HLM, 20 subjects), human cDNA‐expressed CYP (CYP1A2, 2A6, 2B6, 2C8, 2C9, 2C18, 2C19, 2D6, 2E1, 2 J2, 3A4, 3A5 or 3A7), human cDNA‐expressed UGT (UGT1A1, 1A3, 1A4, 1A6, 1A7, 1A8, 1A9, 1A10, 2B4, 2B7, 2B15 or 2B17), human cDNA‐expressed FMO (FMO1, FMO3 or FMO5), and CYP2C19 genotyped HLM were purchased from BD Bioscience (Woburn, MA). All other chemicals and solvents were obtained at the highest purity available.

In vitro studies to characterize the metabolism of BMS‐823778

Human hepatocyte incubation

Pooled suspension hepatocytes (2 × 10⁹ cells l^–1, n = 1) from 3 donors were incubated with [¹⁴C]BMS‐823778 (3 μmol l^–1) in KHB buffer for 2 h at 37°C under an oxygen/carbon dioxide (95/5) atmosphere with shaking (100 rpm). Pooled hepatocytes were used to minimize the enzyme activity variability among individual donors. The incubation time and drug concentration were selected based on a preliminary experiment that would yield quantifiable amounts of metabolites, as well as that the assays were under well‐defined experimental conditions 10. At the end, aliquots (0.5 ml) were mixed with an equal volume of acetonitrile (100%) to stop the reaction. The samples were then centrifuged at 1500 × g for 15 min and the supernatant was analysed by high‐performance liquid chromatography (HPLC)–tandem mass spectrometry (MS/MS) to identify and quantify metabolite formation.

HLM and human cDNA‐expressed enzyme incubation

To identify the liver enzymes to metabolize BMS‐823778, [¹⁴C]BMS‐823778 was incubated with HLM and a list of human cDNA‐expressed enzymes (Figure 1). The incubation mixtures contained phosphate buffer (0.1 mol l^–1, pH 7.4), MgCl₂ (10 mmol l^–1), [¹⁴C]BMS‐823778 (3 μmol l^–1) and HLM (4 mg protein ml^–1) or human cDNA‐expressed enzymes (CYP and FMO: 100 pmol ml^–1; UGT: 1 mg ml^–1). NADPH (2 mmol l^–1) was added in assays for Phase I enzyme reactions. UDPGA (2 mmol l^–1) and alamethicin (100 μg ml^–1) were included for glucuronide metabolite formation. The reactions volume was 1 ml and the incubation was continued for 2 h at 37°C with shaking (100 rpm). At the end, methanol (100%, 1 ml) was added to stop the reaction. After centrifugation at 1500 × g for 30 min, the supernatant was subjected to LC–MS/MS analysis.

Figure 1.

Characterization of BMS‐823778 (3 μmol l^–1) metabolism in in vitro assays. (A) metabolism of [¹⁴C]BMS‐823778 in human hepatocytes and HLM with added cofactor NADPH or UDPGA (n = 1); (B) reaction phenotyping of M1 and M2 formation by human cDNA‐expressed CYPs and FMOs (n = 1); (C) reaction phenotyping of M3 formation by human cDNA‐expressed UGTs (n = 1); (D) formation of M1 relative to control (no chemical inhibitor) in HLM incubation in the presence of ketoconazole (1 μmol l^–1, CYP3A4 inhibitor) or benzylnirvanol (1 μmol l^–1, CYP2C19 inhibitor; n = 1 in triplicate and data presented as mean ± SD; ***, P < 0.01 compared to control, and ^###, P < 0.01 compared to CYP3A4 inhibition by one‐way ANOVA with Dunnett posttest)

HLM incubations with chemical inhibitors

BMS‐823778 (3 μmol l^–1) was incubated with pooled HLM (1 mg ml^–1) at 37°C in triplicate in the absence and presence of benzylnirvanol (1 μmol l^–1) or ketoconazole (1 μmol l^–1), which are specific chemical inhibitors for CYP2C19 and CYP3A4/5, respectively 11, 12. Benzylnirvanol was preincubated with HLM in the presence of NADPH for 10 min before adding BMS‐823778, to complete block the CYP2C19 activity 13. The incubation mixture (1 ml) was the same as described above except the addition of chemical inhibitor. The stock solution of both chemical inhibitors were prepared in dimethyl sulfoxide (DMSO). In the incubations without chemical inhibitors (control group), DMSO was added to match the final organic solvent percentage (0.2%) in other reactions, to eliminate the impact of DMSO on enzyme activity. After 2 h incubation, the reaction was stopped with methanol (100%, 1 ml) containing carbamazepine as internal standard (IS). The formation of M1 was determined by LC–MS/MS.

CYP enzyme kinetics for M1 formation

BMS‐823778 at the final concentrations from 0.39 to 200 μmol l^–1 was incubated with CYP2C19 (10 pmol ml^–1), CYP3A4 (30 pmol ml^–1) or CYP3A5 (30 pmol ml^–1). The incubation time was 45 min for CYP2C19 and 30 min for CYP3A4 and CYP3A5 at 37°C. The protein concentration and incubation time of each CYP enzyme were optimized in preliminary experiments that the reactions were in a linear phase under these conditions. The assay was conducted in duplicate with a total incubation volume of 100 μl, including phosphate buffer (0.1 μmol l^–1, pH 7.4), MgCl₂ (10 mmol l^–1) and NADPH (2 mmol l^–1). The reaction was stopped by adding an equal volume of acetonitrile (100%) containing alprazolam (1 μmol l^–1) as IS. Following centrifugation at 6700 × g for 10 min, the supernatant was analysed by LC–MS/MS to quantify M1 formation. M1 reference standard was spiked into the same reaction matrix except that the active CYP recombinant enzyme was replaced with control recombinant CYP (membrane protein isolated from Escherichia coli containing empty expression plasmid) at the final concentrations ranged from 40 nmol l^–1 to 20 μmol l^–1 and processed the same as incubation samples.

Incubation with CYP2C19 genotyped HLM

[¹⁴C]BMS‐823778 (3 μmol l^–1) was incubated with CYP2C19 genotyped HLM (*1/*1, n = 7; *1/*2, n = 5; *1/*3, n = 1; *2/*2, n = 4; *3/*3, n = 1). The activities of CYP2C19 in those HLM were pre‐evaluated by incubation with S‐mephenytoin, as provided by the vendor (BD Bioscience). The incubation mixture and conditions were the same as described in HLM and human cDNA‐expressed enzyme incubation, and the HLM concentration was 1 mg ml^–1.

Pharmacokinetics and metabolism of BMS‐823778 in humans

Study design

This was an open‐label, single‐dose study in healthy male subjects following the study protocol MB121004. This study was conducted in accordance with Good Clinical Practice. The protocol, amendments and subject informed consent were received appropriate approval by the Institutional Review Board/Independent Ethics Committee. The subjects were first prescreened for CYP2C19 genotypes. Briefly, DNA was extracted from EDTA blood samples with a MagNA Pure LC Total Nucleic Acid Isolation kit (Roche Diagnostics, Mannheim, Germany). Purified DNA samples were analysed with DMET microarray assay using the DMET Plus Premier pack kits following manufacture instructions. Genotyping data were generated with Affymetrix GeneChip Command console software and analysed with the DMET Console software. A total of 14 prescreened healthy male subjects, aged between 19 and 34 years, completed the trial in three groups: Group 1, CYP2C19 extensive metabolizer (EM, 7 subjects with CYP2C19 genotype *1/*1; race: six White and one Asian); Group 2, CYP2C19 poor metabolizer (PM, three subjects with CYP2C19 genotype *2/*2; race: two White and one Asian) and Group 3, three subjects with CYP2C19 EM (*1/*1; race: two White and one Black/African American) and one subject with CYP2C19 PM (*2/*2; race: one White). Groups 3 was specifically included in this study to collect bile for the characterization of metabolic profiles of [¹⁴C]BMS‐823778 in biliary excretion. After at least an 8 h overnight fast, each subject received a single oral solution dose of 10 mg [¹⁴C]BMS‐823778 containing 80 μCi of radioactivity. The amount of dosed radioactivity in humans was determined based on dosimetry study in rat (unpublished results).

Safety monitoring

All subjects were closely monitored for adverse events (AEs) throughout the study. Safety was assessed by monitoring of vital signs, physical examinations, electrocardiogram (ECG) and clinical laboratory tests.

Sample collection

Blood samples were collected in all three groups at 0, 0.5, 1, 1.5, 2, 3, 4, 6, 8, 12, 24, 48, 72, 96, 120, 144, 168, 192, 216, 240, 264, 288, 312, 336, 360, 384, 408, 432, 456, 480 and 504 h postdose for BMS‐823788 and total radioactivity (TRA) quantitation. For biotransformation profiling, blood samples were collected at 0, 1, 4, 12, 24, 48 and 96 h postdose in all subjects. Urine and faeces were collected from each subject over intervals throughout the study (0–504 h). Bile samples were collected only in Group 3 during 3–4, 4–6 and 6–8 h periods postdose using suction method 14.

Sample preparation and analysis

Plasma, urine, faeces and bile samples from each subject were processed and the TRA was counted for each individual as reported previously 15. For metabolite profiling, pooled plasma samples were prepared separately from Group 1 (EM) and Group 2 (PM) by mixing an equal volume (0.5 ml) of plasma sample at 1, 4 or 24 h from each subject. Plasma AUC_0–48h pool samples were prepared using Hop method 16. Urine and faecal samples were pooled from Group 1 (EM) or Group 2 (PM) separately, which were prepared by mixing a constant portion (5–10% of weight or volume) of each interval samples from all subjects in the same group collected from 0 to 336 h 15. Similarly, bile samples (3–8 h) from Group 3 were pooled from EM and PM separately. Due to the low radioactivity concentrations, plasma samples beyond 48 h and urine and faecal samples after 336 h were not included in the metabolite profiling analysis. All samples were then extracted and processed for LC–MS/MS analysis similar as reported previously 15, 17.

Analytical analysis

LC–MS/MS quantitation

The quantitation of M1 from in vitro assays using nonradiolabeled BMS‐823778 was conducted with exploratory LC–MS/MS methods. To determine the M1 formation in HLM inhibition assay, samples were injected onto a Waters Atlantis dC18 column (5 μm, 2.1 × 150 mm) connected to Shimadzu LC‐10 AD VP pumps and a Sciex 4000 Q‐trap. The mobile phase consisted of two solvents: A) 0.1% formic acid in water and B) 100% acetonitrile. The gradient started at 5% of solvent B, held at 5% for 2 min, linearly increased to 40% at 30 min, and then increased to 90% at 32 min. The flow rate was 0.3 ml/min. The specific transitions were M1 (m/z 344 → 326) and carbamazepine (237 → 194). Relative formation of M1 were determined based on the peak area ratio of M1 over IS. M1 from enzyme kinetic assays was quantified using a Waters Acquity UPLC and a Sciex 4000 Q‐trap. Samples were eluted using a Waters Acuity UPLC BEH column (1.7 μm, 2.1 × 50 mm) with a flow rate of 0.8 ml/min. The mobile phases were 0.1% formic acid in water with 5 mmol l^–1 ammonium formate (A) and 0.1% formic acid in acetonitrile (B). The gradient started with 0% of solvent B, followed by a linear increase to 70% at 1.1 min and to 95% at 1.11 min. The specific transitions were M1 (m/z 344 → 326) and alprazolam (309 → 281). The amount of M1 were calculated with the standard curve. The lower limit of M1 quantitation was 40 nm, and the ion responses were reliable up to 20 μmol l^–1. In the evaluation of both analytical methods, no carryover was observed between two sample injections. The concentration of BMS‐823778 in human plasma and urine samples was quantified using a validated LC–MS/MS method using [¹³C₂, D₆]BMS‐823778 as reported previously 18.

Metabolite profiling and quantitation

Metabolites from in vitro assays using [¹⁴C]BMS‐823778 and from human study were profiled using a YMC ODS AQ column (5 μm, 4.6 × 150 mm) connected to an Agilent 1100 HPLC (Agilent, Foster City, CA, USA). The mobile phase was: A) 0.1% formic acid in water and B) 100% acetonitrile. For samples from in vitro assays, the gradient was: solvent B started at 0%, then linearly increased to 10% at 5 min, to 70% at 32 min, to 90% at 33 min, held at 90% for 3 min and then decreased to 10% at 37 min. The gradient to elute the ¹⁴C material in human samples was: Solvent B started at 5%, increased linearly to 40% at 60 min, to 80% at 61 min, held at 80% for 4 min, and then decreased to 5% at 66 min. The HPLC flow rate was 1 ml/min in both methods. After sample injection, 75% of HPLC elute was collected into Deepwell LumaPlate‐96 plates at 0.25 min intervals, and the amount of radioactivity was determined using a TopCount analyser 17. The rest was directed into a LTQ ion trap or a LTQ‐Orbitrap mass spectrometer (Thermo, San Jose, CA, USA) for metabolite identification. The amount of BMS‐823778 and metabolites were quantified on the basis of the percentage of the TRA of each peak observed in the entire HPLC radiochromatogram. The structures of metabolites with reference standards available were identified based on their retention times and by comparing the MS fragmentation patterns with reference standards. The structures of the other metabolites were proposed based on their MS fragmentation patterns.

Data analysis

For enzyme kinetics, the maximum formation velocity of M1 (Vmax) and substrate concentration resulting in half Vmax (Km) were calculated by nonlinear regression curving fitting with least squares (ordinary) method using GraphPad Prism 7. The Michaelis–Menten equation (v = Vmax*[S]/(Km + [S])) was applied to determine the kinetic parameters for CYP3A4 and CYP3A5. As inhibition was observed for M1 formation at high concentrations of BMS‐823378 in the CYP2C19 incubation, a modified equation (V = V_max *[S]/(K_m + [S]*(1 + [S]/K_i))), in which Ki represents the inhibition constant, was used. The intrinsic clearance (Vmax/Km) was then normalized to the clearance in mg of HLM, considering the protein abundance of each CYP enzyme. The contribution or fraction metabolized by individual CYP was calculated as a percentage of the total CYP clearance 11. The pharmacokinetic parameters (Cmax, Tmax, AUC_[0–T], AUC_[INF], CLT/F and Vz/F) of BMS‐823788 and TRA in human plasma were derived from a noncompartmental model using Phoenix Winnolin 6.3 (Certara, Princeton, NJ, USA). Statistics was performed using Student t test for two group comparison, one‐way ANOVA with Dunnet post‐test for multiple groups (n ≥ 3), or F‐test for correlation analysis. Differences were considered significant when P < 0.05.

Nomenclature of targets and ligands

Key protein targets and ligands in this article are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY 19, and are permanently archived in the Concise Guide to PHARMACOLOGY 2015/16 20.

Results

In vitro assays

Human hepatocyte and HLM incubations

Following 2‐h incubation with human hepatocytes, 95% of [¹⁴C]BMS‐823778 was unchanged. Three metabolites were observed and identified as the hydroxylated metabolite M1 (3.8%) and M2 (0.62%), and glucuronide metabolite M3 (0.15%; Figure 1A). In HLM incubations, M1 and M2 were observed in the presence of NADPH, and M3 was detected in the presence of UDPGA and alamethicin. Only a small amount the BMS‐823778 (<5%) was metabolized in all incubations.

Enzyme phenotyping

For the formation of M1, CYP2C19 demonstrated the highest turnover rate compared with other CYP and FMO enzymes (Figure 1B). CYP3A4/3A5 were also able to catalyse BMS‐823778 to M1, but M1 formation was smaller as compared to CYP2C19. Other investigated enzymes showed minimal to no activity. For M2 formation, CYP3A4 and CYP3A5 had higher activities than other enzymes. The formation of M3 was mainly catalysed by UGT1A4, with slight contribution by UGT1A3 (Figure 1C).

Fraction metabolized by CYPs

In the presence of CYP3A4/3A5 inhibitor ketoconazole, the formation of M1 in HLM incubated with [¹⁴C]BMS‐823778 was significantly decreased by an average of 32% as compared to control (no inhibitor). A stronger inhibition was observed in the presence of CYP2C19 inhibitor benzylnirvanol, and the formation of M1 was inhibited by 51% (Figure 1D). As CYP2C19, CYP3A4 and CYP3A5 were the major enzymes to catalyse M1 formation, the reaction kinetics were then determined for these three enzymes (Figure S1). The values of Vmax and Km are summarized in Table 1. When considering the protein abundance of each enzyme in HLM, the clearance (Cl_int) of CYP2C19 was similar to that of CYP3A4. The fraction metabolized by each CYP is also presented in Table 1.

Table 1.

Enzyme kinetics of M1 formation from BMS‐823778 in the incubations with human cDNA‐expressed CYP2C19, CYP3A4 or CYP3A5. The experiment was conducted with n = 1 in duplicate, and data are presented as average values

	Vmax (pmol min–1 pmol CYP–1)	Km (μmol l–1)	In vitro intrinsic clearance (μl min–1 pmol CYP–1)	Enzyme abundancea (pmol mg HLM–1)	Clint,expressed P450 (μl mg HLM–1 min–1)	fmP450(%)
CYP2C19	1.48	33.9	0.044	11	0.48	41.4
CYP3A4	0.35	62.8	0.006	93	0.52	44.6
CYP3A5	0.33	34.4	0.010	17	0.16	14.0

^a, the abundances of CYP2C19, CYP3A4 and CYP3A5 were obtained from a previous report 34. fm, fraction metabolized

M1 formation in CYP2C19 genotyped HLM

The formation of M1 was greatly variable in all CYP2C19 genotyped HLM when incubated with 3 μmol l^–1 of [¹⁴C]BMS‐823778. However, the activity to form M1 in HLM with wild‐type (WT) CYP2C19 (*1/*1) was significantly higher (P = 0.022) than that in HLM with heterozygous allelic variants (*1/*2 and *1/*3). Due the intersample variability, the differences between WT and homozygous allelic variants (*2/*2 and *3/*3) were not significant (P = 0.054, Figure 2A). Additionally, M1 formation was not different between heterozygous and homozygous. Overall, the formation of M1 in WT CYP2C19 HLM is 2.1‐fold greater, on average, than that in HLM with CYP2C19 allelic variants. When compared with the CYP2C19 activity characterized by 4′‐hydroxylation of S‐mephenytoin formation rate (provided by vendor), the formation of M1 was well correlated with the reported CYP2C19 activity in the 18 individual lots of HLM (P < 0.001, Figure 2B).

Figure 2.

Metabolism of BMS‐823778 in CYP2C19 genotyped HLM. (A) relative formation of M1 (peak area ratio of M1 over IS) in HLM with predetermined CYP2C19 genotypes (*, P < 0.05 compared to *1/*1 by one‐way ANOVA with Dunnett posttest); (B) correlation of relative formation of M1 with documented CYP2C19 activities in HLM from 18 individual donors (P < 0.001 with F‐test)

Human metabolism and pharmacokinetic study

Safety

A single oral dose of 10 mg [¹⁴C]BMS‐823778 appeared to be safe and generally well tolerated by the healthy male subjects in this study. There were no deaths, serious AEs or discontinuations. Vomiting and dizziness were the most frequent AEs, which were reported in three subjects each. The AEs of vomiting were related to the temporary placement of nasoduodenal tube. There were no apparent clinically meaningful differences between baseline and study discharge with respect to laboratory test, vital signs and ECG intervals. The genotype of CYP2C19 had no impact on the observed AEs.

Pharmacokinetics

As the pharmacokinetic parameters of BMS‐823778 and TRA were not statistically different (P > 0.1) between Group 1 and Group 3 (EM; Table S1), subjects with CYP2C19 EM genotype from all groups were combined together for pharmacokinetic analysis. Similar analysis were also conducted for subjects with CYP2C19 PM. Following a single oral dose of [¹⁴C]BMS‐823778 (10 mg, 80 μCi), the maximum plasma concentrations (Cmax) of BMS‐823778 were reached at 1.3 h and 1.6 h in EM and PM, respectively (Figure 3). However, in PM, the plasma exposure of BMS‐823778 (AUC_[INF]) was significantly increased (5.3‐fold, P = 0.0097) and total clearance (CLt/F) was significantly decreased (5.3‐fold, P < 0.0001) as compared to in EM (Table 2). The Cmax of TRA was delayed and maintained higher than that of BMS‐823778. The exposure ratio of BMS‐823778 over TRA was 25% in EM, but approximately 60% in the PM.

Figure 3.

Plasma concentrations of BMS‐823778 (A) and total radioactivity (TRA, B) in healthy male subjects with CYP2C19 EM (n = 10) or PM (n = 4) following a single oral dose of [¹⁴C]BMS‐823778 (10 mg, 80 μCi). Data are expressed as mean ± SD

Table 2.

Pharmacokinetic parameters of BMS‐823778 and total radioactivity (TRA) following a single oral dose of [¹⁴C]BMS‐823778 (10 mg, 80 μCi) to healthy male subjects with CYP2C19 EM or PM

		Cmax	Tmax	AUC[0–T]	AUC[INF]	CLt/F	Clr	Vz/F	% Urine recovery
		ng ml–1 GEO.MEAN (%CV)	h GEO.MEAN (%CV))	ng*h ml–1 GEO.MEAN (%CV)	ng*h ml–1 GEO.MEAN (%CV)	ml min−1 GEO.MEAN (%CV)	ml min−1 GEO.MEAN (%CV)	ml GEO.MEAN (%CV)	% GEO.MEAN (%CV))
BMS‐823778	EM (n = 10)	62 (23)	1.3 (47)	1851 (31)	1964 (30)	85 (24)	0.16 (48)	1 008 880 (74)	0.18 (59)
	PM (n = 4)	76 (13)	1.6 (55)	9214** (24)	10396** (28)	16*** (25)	0.1 (51)	220498** (14)	0.59* (35)
TRA	EM (n = 10)	93### (20)	6.3# (81)	7391### (20)	7597### (20)	‐	‐	‐	‐
	PM (n = 4)	85## (12)	1.9* (41)	14645* , ### (19)	17204* , ## (24)	‐	‐	‐	‐

^* P < 0.05;

^** P < 0.01; and

^*** P < 0.001 when compared to EM;

^# P < 0.05;

^## P < 0.01; and

^### P < 0.001 when TRA was compared to BMS‐823778 in the same subjects.

Excretion

Since similar observations in urinary and faecal excretion between Group 1 (EM) and Group 3 (EM; Figure S2), results in same genotype were combined for mass balance analysis. The average total recovery of radioactivity in CYP2C19 EM was 69.7% (95% confidence interval: 62.1%, 77.3%) within 504 h after a single oral administration of [¹⁴C]BMS‐823778, in which the majority of the radioactivity was excreted in urine [68.8%, (61.3%, 76.3%)], while 0.9% (0.2%, 1.1%) of dose was in faeces. In PM, only 48.7% (45.2%, 52.3%) of the dosed radioactivity was recovered within 504 h, with 47.0% (43.5%, 50.6%) in urine and 1.7% (0.9%, 2.5%) in faeces. When evaluating BMS‐823778 in urine, the amount of the parent drug was significantly higher in PM than in EM, but the renal clearance was not different (Table 2). Less than 0.1% of dosed radioactivity was presented in 3–8 h bile samples for both CYP2C19 genotypes in Group 3.

Metabolite profiling

In plasma samples, BMS‐823778 was the major drug related component and M1 was the only metabolite detected by radioactivity from EM (Group 1) and PM (Group 2). Radiochromatograms of plasma samples from 1 h, 4 h, 24 h and AUC_0–48h pool from EM and PM are shown in Figures S3 and S4, respectively. In the plasma AUC_0–48h pool samples, the amount of M1 in EM was 2.4‐fold of that in PM. M1 was also the only metabolite found in faecal samples (0–366 h) from PM and EM based on radioactivity peak (Figure S5). However, multiple metabolites were identified in urine and bile samples in both EM and PM (Figures S5 and S6). In urine samples collected up to 336 h, BMS‐823778 counted for <1% of the dosed radioactivity (Table 3). The amounts of M1 recovered in urine were similar between PM and EM. However, higher level of M4 was observed in EM, but more glucuronide metabolites, M3, M5 and M9, were found in PM. The metabolism pathways in EM and PM are proposed in Figure 4.

Table 3.

Distribution of metabolites in pooled plasma, urine, faeces and bile samples from humans with CYP2C19 EM or PM following a single oral dose of [¹⁴C]BMS‐823778 (10 mg, 80 μCi)

	% Distribution of radioactivity (% dose)
	EM				PM
	Plasma (0‐48 h)	Urine (0–336 h)	Faeces (0–336 h)	Bile (3–8 h)	Plasma (0‐48 h)	Urine (0–336 h)	Faeces (0–336 h)	Bile (3–8 h)
BMS‐823778	58.7	0.5 (0.34)	51.9 (0.46)	29.1	86.1	2.1 (0.96)	64.0 (1.29)	34.5
M1	33.3	5.8 (3.93)	33.3 (0.30)	12.4	13.9	8.4 (3.86)	23.7 (0.48)	6.5
M2	ND	ND	ND	ND	ND	ND	ND	ND
M3	ND	4.2 (2.85)	ND	20.8	ND	21.4 (9.82)	ND	19.7
M4	ND	49.9 (33.8)	ND	ND	ND	3.2 (1.47)	ND	ND
M5	ND	ND	ND	11.1	ND	3.1 (1.42)	ND	10.8
M7	ND	2.9 (1.97)	ND	ND	ND	7.1 (3.26)	ND	ND
M8	ND	0.8 (0.54)	ND	7.6	ND	4.1 (1.88)	ND	4.9
M9	ND	14.5 (9.83)	ND	1.1	ND	42.6 (19.6)	ND	6.3
M10	ND	4.9 (3.32)	ND	4.5	ND	ND	ND	1.1
M13	ND	0.7 (0.47)	ND	ND	ND	ND	ND	ND
Total	92	84.2 (57.1)	85.2 (0.76)	86.6	100	92.0 (42.2)	87.7 (1.8)	83.8

ND, not detected by radioactivity

Figure 4.

Proposed metabolism pathways of BMS‐823778 in humans with CYP2C19 EM or PM. Solid arrows represent metabolism pathways for EM and empty arrows represent for PM. The width of the arrows approximately indicates the fraction metabolized

Discussion

Biological matrices prepared from human donors, such as hepatocytes or liver microsomes, are commonly used to study drug metabolism, providing valuable data to predict the drug disposition in humans 21, 22. Here, three metabolites were identified using in vitro assays with M1 being the predominant species. However, within 2 h incubation only a small portion of [¹⁴C]BMS‐823778 (<5%) was metabolized in both hepatocytes and HLM incubations, indicating the slow clearance of BMS‐823778 by liver enzymes. This agrees with the slow elimination observed in human study. The formation of each metabolite was then further evaluated by recombinant liver enzymes to identify the specific enzymes catalysing BMS‐823778, as investigation of the metabolism pathways is critical to predict the drug pharmacokinetics in special populations 23. CYP2C19 was found to be a major enzyme to metabolize BMS‐823778 to M1, and its formation was well correlated with the CYP2C19 activity in genotyped HLM. Therefore, the intersubject variability of BMS‐823778 plasma exposures in early clinical studies could be a result of different CYP2C19 activities among studied individuals.

Genetic variances of liver metabolic enzymes are well known to impact the pharmacokinetics of drugs, leading to altered therapeutic effects or unexpected toxicities 24. CYP2C19 is a metabolic enzyme that has demonstrated high frequency of polymorphisms in humans 25. Multiple genetic variants of CYP2C19 have been identified, in which one defect in exon 5 (CYP2C19 *2) and a premature stop codon in exon 4 (CYP2C19 *3) are the two most commonly observed mutations 26, 27. The clinical significance of CYP2C19 generic variance has been reported for many classes of drugs including proton‐pump inhibitors, anticonvulsants, hypnosedatives and antidepressants 25. For example, the plasma exposure of omeprazole was 5‐fold higher in PM than in EM, interestingly resulting in an improved acid suppression at standard dose in PM 28, 29, 30. However, the adverse effects of mephenytoin and sertraline were reported in CYP2C19 PM due to increased plasma exposure 31, 32.

In this study, 14 subjects were grouped according to their CYP2C19 genotypes. As expected from the in vitro results, the plasma exposure of BMS‐823778 was significantly higher in PM than in EM. Since similar Cmax and Tmax of BMS‐823778 were observed in both PM and EM, the differences of plasma exposures are most likely to be due to the lower rate of elimination in PM. The systemic exposure of TRA was approximately 3‐fold of BMS‐823778 exposure in EM, but the ratio was only 1.6‐fold in PM. These findings suggest that BMS‐823778 was extensively metabolized and a greater portion of TRA in plasma was present as metabolites in EM than in PM. Profiling metabolites in plasma samples demonstrated higher amounts of M1 in EM than in PM. Therefore, genotypic expression of CYP2C19 is an important determinant of BMS‐823778 elimination in humans.

Similar conclusion can also be obtained from excretion results. Minimum dosed radioactivity (<2% of dose) was recovered in faeces in both EM and PM following oral administration, suggesting that BMS‐823778 and M1, the two major drug related components, were extensively absorbed and not mainly eliminated in faeces. The majority of dosed radioactivity was recovered in urine, but only <1% of dose was BMS‐823778. Thus, we can conclude that renal excretion was the major route for TRA elimination but metabolism was necessary for BMS‐823778 to be cleared. More importantly, the metabolism was significantly affected by CYP2C19. Although a trace amount of dosed radioactivity (<0.1%) was recovered in bile from both EM and PM as evaluated in Group 3, the bile collection period (5 h) is short compared to the slow elimination of BMS‐823778. In bile, BMS‐823778, together with its glucuronide metabolites (M3 and M5), contributed 61% and 65%, and M1 contributed 12% and 7% of TRA in EM and PM, respectively. Due to the extensive oral absorption, all the biliary excreted BMS‐823778 and the glucuronide metabolites (converting back to BMS‐823778 in gut) are expected to be reabsorbed. To support this, higher plasma concentrations of BMS‐823778 were observed in several samples when compared to those collected 24 h earlier in the same subject. Therefore, the enterohepatic recirculation, together with the low metabolic clearance, contributed to the slow elimination of BMS‐823778 in humans.

The metabolism profile of BMS‐823778 in humans is more complicated than was observed in HLM and hepatocyte incubations. M1 and M3 were observed both in vitro and in vivo, but M2 was only found in vitro. M2 may also be formed in humans but could be further metabolized by glucuronidation, as proposed in Figure 4. Other in vivo metabolites were not observed in in vitro assays, probably because of low metabolic clearance. The only metabolite observed in plasma was M1, while multiple additional metabolites were present in urine. It looks likely that the amount of these metabolites was too small to be detected by radioactivity in plasma due to slow formation and fast renal clearance. Surprisingly, the amount of M1 was similar in urine between EM and PM, approximately 4% of dose. However, the percentage of M4 relative to the dosed radioactivity in the urine of EM was around 23‐fold of that in PM. Therefore, it is highly likely that the formation of M4 was from M1 and also impacted by the genotype of CYP2C19. M1 seemed more preferably conjugated to form M9 in PM, as a higher amount of M9 was determined in the urine of PM. Therefore, the further metabolism of M1 explains the similar excretion profile of M1 between PM and EM.

As determined in the recombinant enzymes, M1 formation was mainly through CYP2C19, and also by CYP3A4 and CYP3A5. The in vitro intrinsic clearance (Vmax/Km) of CYP2C19 was 7‐fold that of CYP3A4. However, when extrapolating the kinetic clearance to HLM clearance considering the enzyme protein abundance, CYP2C19 was predicted to contribute to 41% of M1 formation. Similar results were also observed in the HLM with CYP2C19 chemical inhibitor and genotyped HLM assay. In the urinary excretion, the total amount of M1, M4 and M9 (originated from M1 metabolism) in EM was 48% of dose, approximately 2‐fold that in PM (25% of dose). Thus, the in vivo determined CYP2C19 contribution to BMS‐823778 metabolism is in a good agreement with the in vitro results. However, BMS‐823778 plasma exposure (AUC_[INF]) in PM was 5.3‐fold of that in EM, suggesting that the apparent fraction of BMS‐823778 metabolized by CYP2C19 is close to 0.8 in vivo. The mechanism for this in vitro–in vivo disconnection is not clear. One possible explanations is that enterohepatic recirculation may contribute to the greater fraction metabolized in vivo. Evidence to support this is the higher ratio of BMS‐823778 secreted in the bile of PM. Additionally, more BMS‐823778 glucuronide metabolites (M3 and M5) were formed in PM, as indirectly supported by the higher level of M3 and M5 in the urine from PM, which were then reabsorbed as BMS‐823778 from gastrointestinal tract into systemic circulation. In consequence, the PM seemed likely to attain improved bioavailability due to the enterohepatic recirculation and therefore a higher plasma exposure.

A glucuronide metabolite (M3) was also identified and formed predominantly through UGT1A4 in vitro, suggesting that UGT1A4 could play a role in the metabolic clearance of BMS‐823778. Genetic variance (*2) of UGT1A4 has been reported to reduce metabolic activity and therefore can influence BMS‐823778 pharmacokinetics 33. In this study, the genotype of UGT1A4 were also characterized in the enrolled subjects, of whom three subjects in CYP2C19 EM and one subject in CYP2C19 PM were with UGT1A4 *2/WT (Data not shown). We found that the three subjects with UGT1A4 *2/WT and CYP2C19 EM had BMS‐823778 systemic exposures within the range of subjects carrying UGT1A4 WT/WT and CYP2C19 EM. However, a single subject with a UGT1A4 *2/WT and CYP2C19 PM had the highest BMS‐823778 systemic exposure. Therefore, outstanding high systemic exposures could happen to BMS‐823778 in subjects with both CYP2C19 and UGT1A4 polymorphisms. However, due to the limited number of subjects and lack of homozygous allelic variants in this study, further study is warranted to fully understand the clinical impact of UGT1A4 polymorphism on BMS‐823778.

Collectively, the in vitro assays were able to predict that CYP2C19 polymorphisms could impact the metabolism profiles of BMS‐823778 in humans. However, the in vitro–in vivo extrapolation under predicted the plasma exposure of BMS‐823778 in PM. This greater than anticipated plasma exposure is probably due to enterohepatic recirculation involving BMS‐823778 and glucuronide conjugates. In conclusion, the activity of CYP2C19 significantly altered the pharmacokinetics and metabolism pathways of BMS‐823778; therefore, the therapeutic effect of BMS‐823778 could be greatly variable in population with high frequency of CYP2C19 polymorphisms.

Competing Interests

There are no competing interests to declare.

Supporting information

Table S1 Pharmacokinetics of BMS‐823778 and total radioactivity in Group 1, 2 and 3 following a single oral dose of [¹⁴C]BMS‐823778 (10 mg, 80 μCi) in healthy male subjects

Figure S1 Nonlinear regression with least squares (ordinary) fitting of M1 formation versus BMS‐823778 concentration in the incubations with human cDNA expressed CYP3A4 (A), CYP3A5 (B) and CYP2C19 (C). n = 1 in duplicate and the average values are presented. The Michaelis–Menten equitation (v = Vmax*[S]/(Km + [S])) was applied to determine the kinetic parameters for CYP3A4 and CYP3A5. As inhibition was observed for M1 formation at high concentrations of BMS‐823378 in CYP2C19 incubation, a modified equation (V = V_max *[S]/(K_m + [S]*(1 + [S]/K_i))), in which Ki represents the inhibition constant, was used

Figure S2 Cumulative urinary (A) and faecal (B) excretion of dosed radioactivity in healthy male subjects in Group 1 (n = 7), Group 2 (n = 3), Group 3 (EM, n = 3) and Group 3 (PM, n = 1) following a single oral dose of [¹⁴C]BMS‐823778 (10 mg, 80 μCi). Data are expressed as mean ± SD

Figure S3 Biotransformation profiles of pooled plasma extracts from humans in Group 1 (EM) after oral administration of [¹⁴C]BMS‐823778 (10 mg, 80 μCi)

Figure S4 Biotransformation profiles of pooled plasma extracts from humans in Group 2 (PM) after oral administration of [¹⁴C]BMS‐823778 (10 mg, 80 μCi)

Figure S5 Biotransformation profiles of pooled urine and faecal homogenates from humans (Group 1 and Group 2) after oral administration of [¹⁴C]BMS‐823778 (10 mg, 80 μCi)

Figure S6 Biotransformation profiles of pooled bile (3–8 h) from humans (Group 3 EM & PM) after oral administration of [¹⁴C]BMS‐823778 (10 mg, 80 μCi)

Cheng, Y. , Wang, L. , Iacono, L. , Zhang, D. , Chen, W. , Gong, J. , Humphreys, W. G. , and Gan, J. (2018) Clinical significance of CYP2C19 polymorphisms on the metabolism and pharmacokinetics of 11β‐hydroxysteroid dehydrogenase type‐1 inhibitor BMS‐823778. Br J Clin Pharmacol, 84: 130–141. doi: 10.1111/bcp.13421.