Drug-drug interaction of atazanavir on UGT1A1-mediated glucuronidation of molidustat in human.

Molidustat is an oral inhibitor of hypoxia-inducible factor (HIF) prolyl-hydroxylase enhancing the erythropoietin (EPO) response to HIF; it is in clinical development for the treatment of anaemia related to chronic kidney disease. The predominant role of glucuronidation for molidustat clearance (formation of N-glucuronide metabolite M1) and subsequent renal excretion was confirmed in a human mass balance study, with about 85% of the drug being excreted as M1 in urine. The inhibitory effects of 176 drugs and xenobiotics from various compound classes on the UGT-mediated glucuronidation of molidustat in human liver microsomes (HLMs) were investigated. Based on preclinical findings, glucuronidation of molidustat was predominantly mediated by the 5'-diphospho-glucuronosyltransferase (UGT) isoform UGT1A1. Therefore, atazanavir, which is a potent inhibitor of UGT1A1, was chosen for the evaluation of pharmacokinetics and EPO release following a single oral dose of 25 mg molidustat. Molidustat exposure increased about twofold upon coadministration with atazanavir when considering area under plasma concentration-time curve from zero to infinity (AUC) and maximum plasma concentration (Cmax ). Baseline-corrected increase of EPO was 14% and 34% for Cmax and AUC (calculated over 24 hours), respectively. Coadministration of molidustat and atazanavir was well tolerated.

INTRODUCTION

Anaemia is a common complication of chronic kidney disease (CKD). This is partly caused by impairment of physiological hypoxia detection at the renal level and the corresponding release of hypoxia‐induced factors (HIFs). , In the presence of oxygen, HIF‐α subunits are tagged by proline hydroxylation for deactivation by HIF proline hydroxylase, , which reduces erythropoietin (EPO) expression. Molidustat has been shown to inhibit HIF‐PH (EC 1.14.11.29) in the nanomolar concentration range, resulting in sustained physiological transcription of the EPO gene. This, in turn, leads to increased EPO release and increased haemoglobin concentrations in patients with CKD, as established in the phase 2 DIALOGUE trials.

Clinical intervention in renal anaemia via erythropoiesis‐stimulating agents (ESAs), the standard treatment in CKD‐related anaemia since 1989, has well‐known disadvantages. These include the parenteral administration of the protein drugs, primary/secondary resistance (in ~10%‐20% of patients) and potential cardiovascular side effects (eg increased risk of cardiovascular disease and death). , , , Therefore, the stabilization of HIFs, which are the primary physiological signal for hypoxia‐induced increase in erythropoiesis, is a new and promising clinical pathway for the management of CKD‐induced anaemia. , , , ,

The molidustat pharmacokinetic (PK) profile, including the metabolic fate and routes of excretion, has been studied in humans following oral administration. Molidustat was almost completely absorbed from the gastrointestinal tract and had an absolute oral bioavailability of about 60%. Molidustat is subsequently cleared extensively via conjugation, resulting in the N‐glucuronide metabolite M1; this metabolite undergoes renal elimination, with about 85% of the administered dose being recovered as M1 in urine. Only a minor fraction of the dose is excreted as molidustat via urine (~4%) or faeces (~6%). , Based on the high relevance of glucuronidation for the molidustat PK profile, the uridine 5'‐diphospho‐glucuronosyltransferase (UGT) isoforms involved had to be identified and subsequently screened for potential drug‐drug interactions (DDIs) with co‐medications or food. Based on these findings, atazanavir was chosen as a model compound to be tested in a clinical DDI study evaluating the influence of a strong UGT1A1 inhibitor on the PK and pharmacodynamic (PD) profiles of molidustat. In the following paper, the results from the DDI study for molidustat with repeated doses of atazanavir are reported. The data from UGT isoform identification and in vitro screening for potential interaction partners, which guided the selection of atazanavir as co‐medication for the clinical study, are also presented.

METHODS

The study was conducted in accordance with the Basic & Clinical Pharmacology & Toxicology policy for experimental and clinical studies.

Identification of relevant UGT isoforms

Isoform‐expressing microsomes

In a first step, 12 human UGTs expressed in insect cells (ie UGT1A1, 1A3, 1A4, 1A6, 1A7, 1A8, 1A9, 1A10, 2B4, 2B7, 2B15, 2B17; Supersomes, BD Gentest, Woburn, MA, USA), as well as microsomes from non‐transfected control insect cells, were screened for the catalytic activity for the formation of the N‐glucuronide of molidustat (M1), the major human metabolite. For this, [14C]molidustat was incubated at final concentrations of 1.0 µmol L−1 and 10 µmol L−1, with 0.5 mg mL−1 microsomal protein in the presence of alamethicin (50 µg mg−1 protein), after 15 minutes of preincubation on ice. Incubations (N = 1) were performed in 100 mmol L−1 potassium phosphate buffer (pH 7.4), containing 5 mmol L−1 magnesium chloride. After addition of molidustat and saccharolactone (5 mmol L−1), the solution was pre‐warmed to 37°C for 5 minutes, and the reaction was started subsequently by adding uridine 5'‐diphosphoglucuronic acid (UDPGA) at a final concentration of 5 mmol L−1. After 30 minutes or 60 minutes at 37°C, the reaction was terminated by addition of 5% phosphoric acid/acetonitrile (1:10) and cooled on ice. After centrifugation, supernatants were injected onto a high‐performance liquid chromatographic (HPLC) system (HP1100, Agilent) with off‐line radioactivity detection (Wallac 1450 MicroBeta, PerkinElmer/Wallac) using Ultima Flo AP scintillation fluid (PerkinElmer/Canberra Packard). For incubations with low molidustat concentration (1.0 µmol L−1), samples were concentrated under reduced pressure prior to HPLC analysis. Chemicals were purchased from Fluka or Sigma, or Merck, if not mentioned otherwise.

Isoform‐specific inhibitor experiments

In a second step, microsomes from the human liver (Cytonet), intestine or kidney (XenoTech) were incubated (N = 2, arithmetic means presented) with molidustat (4.9 µmol L−1) under essentially the same conditions as described above at a final protein concentration of 0.5 mg mL−1. Isoform‐specific inhibitors of UGTs were added before initiating the reaction with UDPGA; these inhibitors were 5 µmol L−1 atazanavir (inhibitor of UGT1A1), 10 µmol L−1 hecogenin (inhibitor of UGT1A4,), 5 µmol L−1 niflumic acid (inhibitor of UGT1A9) , or 10 µmol L−1 mefenamic acid (inhibitor of UGT2B7, at higher concentrations also UGT1A9).

Kinetic parameters

In order to determine enzyme kinetic parameters of the glucuronidation reaction in human liver microsomes, [14C]molidustat was incubated at concentrations of 2.4, 4.9, 9.9, 24.9, and 49.1 µmol L−1 (N = 2, Table S1). The experimental conditions were identical as described for isoform‐specific inhibitor experiments. For the determination enzyme kinetic parameters of the glucuronidation reaction in recombinant UGT1A1 and UGT1A9, [14C]BAY 85‐3934 was incubated at a concentration of 2.4, 4.9, 9.9, 24.9, 49.1 and 96.6 µmol L−1 (N = 2, Tables S2 and S3). The experimental conditions were identical as described for isoform‐expressing microsomes. Enzyme kinetic fits were performed by non‐linear regression analysis using SigmaPlot software (version 11). For human liver microsomes and UGT1A1, the data were fit to the Hill equation, and for UGT1A9, the data were fit to the conventional Michaelis‐Menten equation.

Correlation analysis

To further elaborate the contribution of UGT1A1 and UGT1A9 to the glucuronidation of molidustat, molidustat was incubated with 23 individual human liver microsomal preparations and one sample of pooled human liver microsomes (HLMs), which had been characterized in parallel for their enzymatic activity for two UGT isoform‐specific substrates (ie 17β‐estradiol to 17β‐estradiol 3‐glucuronide [UGT1A1] and propofol to β‐D‐glucuronide [UGT1A9]). Incubations (N = 2, arithmetic means presented) were performed as described above for microsomal experiments with UGT isoform‐specific inhibitors, using molidustat (5 µmol L−1), 17β‐estradiol (20 µmol L−1) or propofol (100 µmol L−1), , 0.3 mg mL−1 protein and 15‐30‐minute incubation times. Samples were analysed for M1, 17β‐estradiol 3‐glucuronide or propofol glucuronide via HPLC mass spectrometry (HP1100 or HP1290, Agilent, and AB SCIEX API3200 or API 4000, AB Sciex Pte. Ltd.). M1‐d8 (in‐house synthesis) or naphthol glucuronide (Sigma‐Aldrich) served as the internal standards. Velocity of molidustat glucuronidation was correlated to respective isoform‐specific activity for UGT1A1 (17β‐estradiol 3‐glucuronidation) and UGT1A9 (propofol glucuronidation) using linear regression analysis.

Screening for potential DDI interactions

Incubations (N = 1; 30 minutes) with pooled HLMs (0.3 mg mL−1) were performed as described above, with molidustat (5 µmol L−1) (total volume: 200 µL) in the presence of six different concentrations for 176 xenobiotics (obtained from commercial suppliers; see Table 1 and Table S4) to determine the inhibitory potency and, if applicable, half maximal inhibitory concentration (IC50) values for inhibition of M1 formation from molidustat. Incubations without inhibitor served as controls and in each run of 14 putative inhibitors atazanavir served as positive control. Experiments were performed on a Genesis Workstation (Tecan) or a Microlab STARlet (Hamilton Robotics). Reactions were terminated by addition of acetonitrile/formic acid (10:1) containing internal standard M1‐d8. After centrifugation, supernatants were analysed by liquid chromatography‐tandem mass spectrometry (LC‐MS/MS) for quantification of M1 as described above. Linear interpolation was used to determine IC50 values. If less than 50% inhibition was observed, data were not extrapolated. To identify potentially relevant interaction partners for molidustat glucuronidation (clearance), the predicted increase in molidustat area under plasma concentration‐time curve from zero to infinity (AUC) in the presence of the respective co‐medication (AUCR) was calculated according to scientific , and regulatory proposals , using the following formula.

TABLE 1

Inhibitory potential (IC50) of various compounds on molidustat glucuronidation in pooled human liver microsomes and estimated increase in molidustat AUC

Inhibitor	IC50,	AUCRinlet,	AUCRinlet,
	µmol L−1	ka 0.1 min−1	ka 0.01 min−1
Erlotinib	0.13	7.26	4.54
Nilotinib	0.17	5.10	2.03
Sorafenib	0.32	3.01	1.74
Atazanavir	0.42	6.29	3.44
	0.42	2.91 a	2.44 b
Saquinavir	0.47	3.18	1.32
Montelukast	0.49	1.04	1.02
Ritonavir	0.60	1.83	1.38
Tacrolimus	0.70	1.01	1.01
Itraconazole	1.2	1.03	1.00
Lopinavir	1.3	1.52	1.19
Dasatinib	1.4	1.13	1.02
Mifepristone	1.4	2.06	1.16
Axitinib	1.5	1.01	1.00
Nelfinavir	1.5	1.72	1.12
Bilirubin	1.5	n.c.	n.c.
Tranilast	1.7	4.37	3.65
Nifedipine	1.7	1.04	1.01
Gefitinib	1.8	2.61	1.22
Curcumin	2.2	n.c.	n.c.
Genistein	2.3	n.c.	n.c.
Raloxifene	2.3	1.18	1.02
Lapatinib	2.5	1.58	1.07
Atorvastatin	2.6	1.06	1.01
Telmisartan	2.8	1.02	1.00
Carvedilol	2.9	1.03	1.01
Indinavir	3.0	5.59	2.61
Cisapride	4.1	1.01	1.00
Domperidone	4.3	1.15	1.02
Rifampicin	4.3	4.15	2.26
Everolimus	4.5	1.20	1.02
Cyclosporin	4.9	1.24	1.04
(‐)‐Epigallocatechin	5.1	n.c.	n.c.
Quercetin	5.5	n.c.	n.c.
Ketoconazole	6.1	1.04	1.01
Glibenclamide	6.3	1.00	1.00
Daidzein	6.8	n.c.	n.c.
Febuxostat	7.2	1.03	1.02
Efavirenz	8.0	1.05	1.01
Imatinib	9.4	1.25	1.05
Clotrimazole	10.2	2.90	1.26
Docetaxel	10.5	1.08	1.02
Lovastatin	10.5	1.05	1.01
Celecoxib	10.6	1.09	1.01
Rosiglitazone	10.7	1.00	1.00
17a‐Ethynylestradiol	10.8	1.00	1.00
Paclitaxel	10.8	1.21	1.03
Midazolam	11.1	1.01	1.00
Nefazodone	11.4	1.02	1.01
Naringenin	11.6	n.c.	n.c.
Mevastatin	13.2	n.c.	n.c.
Fluvastatin	13.4	1.00	1.00
Verapamil	14.1	1.23	1.03
Niflumic acid	14.6	1.77	1.22
Bergamottin	14.8	n.c.	n.c.
Pitavastatin	14.8	1.00	1.00
Simvastatin	15.1	1.02	1.00
Mefenamic acid	16.0	2.57	1.57
Omeprazole	17.1	1.02	1.00
Droperidol	17.3	n.c.	n.c.
Loperamide	17.9	1.00	1.00
Sunitinib	18.7	1.02	1.00
Miconazole	19.3	n.c.	n.c.
Pantoprazole	19.7	1.01	1.01
Mibefradil	20.2	1.00	1.00
Astemizole	23.5	1.00	1.00
Sulfinpyrazone	31.4	1.04	1.03
Diazepam	34.8	1.00	1.00
Lansoprazole	36.8	1.00	1.00
Sulindac	37.5	1.55	1.09
Diclofenac	38.2	1.00	1.00
Doxorubicin	40.4	1.01	1.01
Androsterone	44.0	n.c.	n.c.
Sildenafil	46.2	1.00	1.00
Amiodarone	49.7	1.00	1.00
Diflunisal	75.6	1.10	1.09
Gemfibrozil glucuronide	96.1	n.c.	n.c.

AUCRinlet (k 0.1 min−1) is the increase based on hepatic inlet concentration, assuming a default k of 0.1 min−1, and is calculated as , in which f is the unbound fraction in plasma; [I]max is the maximal total (free and bound) inhibitor concentration in plasma at steady state; F is the fraction absorbed after oral administration (as worst‐case scenario set to 1 if no data available); k is the first‐order absorption rate constant in vivo, assuming a default value of 0.1 min−1 as suggested by guidelines and a less conservative value of 0.01 min−1; Qh is the hepatic blood flow (ie 97 L/h per 70 kg body weight) , ; and R is the blood‐to‐plasma concentration ratio (set to 1 if no data available).

AUC, area under plasma concentration‐time curve from zero to infinity; AUCR, area under plasma concentration‐time curve from zero to infinity in the presence of the respective co‐medication; IC50, half maximal inhibitory concentration; n.c., not calculated. Ki = IC50, since the concentration of molidustat (5 µmol L−1) investigated was far less than the estimated Km (90 µmol L−1) in human liver microsomes.

Based on f = 0.7, considering clearance via UGT1A1 only and k value of 0.108 min−1.

Based on f = 0.7, considering clearance via UGT1A1 only and k value of 0.024 min−1.

Fraction metabolized via total glucuronidation (f) was ~0.9, based on results from the mass balance study (upper limit of sum of N‐glucuronide in urine [85%] and molidustat in faeces [~6%], possibly resulting from hydrolysis in gastrointestinal tract from biliary‐excreted M1) without differentiating between UGT1A1 and UGT1A9 in M1 formation. [I] represents the in vivo concentration of the inhibitor available to inhibit glucuronidation of molidustat. The value for [I] should ideally represent the concentration of inhibitor at the site of the UGT isoform in the liver cells because, based on the absolute bioavailability study, the liver is by far the predominant site of molidustat glucuronidation. According to guidelines from the European Medicines Agency and the US Food and Drug Administration, , the maximum unbound concentration at the inlet to the liver ([I]inlet, u) was used (for calculation see Table S5). f is the unbound fraction in plasma; [I]max is the maximal total (free and bound) inhibitor concentration in plasma at steady state; F is the fraction absorbed after oral administration (as worst‐case scenario set to 1 if no data available); k is the first‐order absorption rate constant in vivo, assuming a default value of 0.1 min−1, as suggested by guidelines, and a less conservative value of 0.01 min−1; Q is the hepatic blood flow (ie 97 L h−1 per 70 kg body weight) , ; and R is the blood‐to‐plasma concentration ratio (set to 1 if no data available). Based on K = IC50/(1 + [I]/K), K ~IC50 since the glucuronidation of molidustat was performed at a substrate concentration of 5 μmol L−1, which is far less than the estimated K value of the glucuronidation of molidustat in human liver microsomes (K = 90 µmol L−1). Therefore, the obtained IC50 values were used for the calculations.

Clinical DDI study design

Study conduct

Ethics

All participants provided written informed consent prior to entry to the study, which was conducted in accordance with the latest revised version of the Declaration of Helsinki and the International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use guidelines for Good Clinical Practice. The protocol was reviewed and approved by the Ethics of the North‐Rhine Medical Council (Düsseldorf, Germany) and by the Federal Institute for Drugs and Medical Devices (BfArM) before study initiation (EudraCT number: 2015‐000109‐37).

Study design

This was a single‐centre, open‐label, fixed‐sequence, non‐placebo‐controlled study in healthy, male participants. In the first period, a single immediate‐release (IR) tablet of 25 mg molidustat was given with 240 mL of water 2 hours after a standardized, light breakfast consisting of a bread roll, 20 g butter, 25 g jam and 250 mL coffee substitute or fruit tea. Following a washout of at least 72 hours, in the second period, the participants received 400 mg atazanavir (2 × 200 mg Reyataz, Bristol‐Myers Squibb) about 30 minutes after a light breakfast for 5 days, and, on the last day, a single IR tablet of 25 mg molidustat was given 1.5 hours after intake of atazanavir.

Study participants

The participants had to be healthy, male individuals who were 18‐45 years of age, with a body mass index of at least 18 kg m−2 and no higher than 29.9 kg m−2. They could not have any known acute or chronic diseases (especially cardiovascular, renal or hepatic), hypersensitivity to study drugs or excipients, or other severe allergies, and they must not have received acute or chronic medications (including St. John's Wort or grapefruit) in the past 7 days before study medication (especially medications that are contraindicated with atazanavir or drugs that increase gastric pH) or have participated in other clinical studies in the past 3 months or during the current study. Participants who regularly smoked more than 25 cigarettes or consumed more than 20 g of alcohol a day were also excluded.

Safety evaluations

The safety and tolerability parameters included body temperature, blood pressure, heart rate, 12‐lead electrocardiogram (ECG), standard laboratory evaluation of blood and urine parameters, as well as reporting of adverse events (AEs).

PK sampling for molidustat (both treatments)

Blood sampling was performed at 0, 0.25, 0.5, 0.75, 1, 1.5, 2, 3, 4, 6, 8, 10, 12, 24, 28, 32, 36 and 48 hours after the dose for characterization of molidustat PK.

PK sampling for atazanavir (second treatment only)

For characterization of atazanavir PK, blood samples were obtained on all 5 study days of the second period prior to administration of atazanavir, as well as on the day of molidustat coadministration 0.5, 1, 1.5, 2, 2.5, 3, 4.5, 5.5, 7.5, 9.5, 13.5, 24, 29.5, 37.5 and 49.5 hours after administration of atazanavir. All samples were stored below −18°C (molidustat) or −80°C (atazanavir) and analysed within 120 days or 39 days after sampling, respectively.

PD sampling

Blood samples for determination of EPO concentrations were collected prior to molidustat administration and 4, 6, 8, 10, 12 and 24 hours after molidustat dosing.

Analytical methods

Molidustat and metabolite M1

Plasma concentrations were measured after protein precipitation with acetonitrile/water/trifluoroacetic acid using a validated LC‐MS/MS method for the concentration range of 0.2‐200 µg mL−1, as described before. For molidustat, quality control (QC) samples in the concentration range of 0.500‐160 μg L−1 and dilution QC samples of 4000 μg L−1 (not required in all batches) were determined with an accuracy of 102%‐108% and a precision of 3.11%‐4.29%. For M1, QC samples in the concentration range of 0.500‐160 μg L−1 and dilution QC samples of 4000 μg L−1 (not required in all batches) were determined with an accuracy of 96.1%‐101% and a precision of 2.56%‐6.82%.

Atazanavir

Quantitative analysis of atazanavir in plasma was performed using a fully validated, triple quadrupole LC‐MS/MS assay ranging from 10 µg mL−1 to 10 000 µg mL−1. Following automated protein precipitation and addition of the internal standard (atazanavir‐d5), an aliquot of 50 µL was injected into the liquid chromatographic system with detection in positive TurboIonSpray mode. QC samples in the concentration range of 30‐7500 μg L−1 were determined with an accuracy of 98.0%‐103% and a precision of 1.43%‐2.88%.

EPO

The serum concentrations of EPO were measured by a solid‐phase, two‐site, one‐cycle chemiluminescent enzyme immunometric assay (IMMULITE, Siemens Healthcare Diagnostics), as outlined previously.

Pharmacokinetic evaluation

Non‐compartmental analysis of plasma concentration‐time data was performed using WinNonlin (Version 6.2.1, Certara) and SAS version 9.3 (SAS Institute). Primary PK parameters were maximum plasma concentration (C max) and AUC of molidustat; additionally, clearance (molidustat only), terminal half‐life, time to C max (t max) and area under plasma concentration‐time curve from zero to the time of the last quantifiable concentration (AUC[0‐t last]) were determined for both molidustat and M1.

Pharmacodynamic evaluation

The concentration of EPO over time was measured. C max, t max and AUC(0‐24) of EPO were calculated to describe the PD effect of molidustat. Moreover, C max of EPO was also calculated based on the ratio and the difference to baseline, and AUC(0‐24) based on the difference to baseline.

Sample size—statistical analysis

For determination of sample size, a priori knowledge on relevant variability of target parameters for molidustat was obtained from previous clinical studies. Assuming a coefficient of variation (CV) for intraindividual variability of less than 20% for AUC and less than 52% for C max, a data set of at least 10 participants available for evaluation would result in a point estimate that deviates from the true ratio 1.16‐fold or less (upper bound of 90% range for expected DDI ratio estimate) for AUC and 1.42‐fold for C max. To account for potential drop‐outs, it was planned to randomize approximately 15 participants.

The safety population included all participants receiving at least one dose; the populations for PK and PD analyses included all participants completing both treatments with valid data sets for PK and PD samples, respectively. PK and PD data were summarized by descriptive statistics.

Moreover, PK parameters (AUC, AUC(0‐tlast) and C max of molidustat and M1, as well as AUC(0‐24) and C max of EPO data (absolute concentrations and ratio to baseline for C max and difference to baseline for AUC), were analysed using analysis of variance (ANOVA). Assuming log normally distributed data, the ANOVAs were performed on log‐transformed data. Point estimates for the DDI ratios (molidustat + atazanavir)/molidustat, derived from differences of least‐squares means, as well as exploratory 90% confidence intervals, were calculated. No confirmatory statistical analysis was intended.

RESULTS

In vitro studies

Identification of relevant UGT isoform(s)

Of the human UGT isoforms tested, only UGT1A1 and UGT1A9 had considerable catalytic activity for the formation of M1; molidustat velocities for M1 formation were 13.0 and 5.99 pmol/(mg recombinant microsomal protein × min) at a concentration of 1 µmol L−1, and 115 and 39.1 pmol/(mg recombinant microsomal protein × min) at a concentration of 10 µmol L−1, respectively. Isoforms UGT1A3, 1A7 and 1A8 revealed low activities of 0.79, 1.22 and 0.82 pmol/(mg recombinant microsomal protein × min) at a concentration of 1 µmol L−1 and of 5.9, 5.8 and 6.8 pmol/(mg recombinant microsomal protein × min) at a concentration of 10 µmol L−1, respectively, while the remaining isoforms (ie UGT1A4, 1A6, 1A10, 2B4, 2B7, 2B15, 2B17) lacked measurable activity for glucuronidation of molidustat. To assess the relative contribution of UGT isoforms to metabolic activity in microsomal preparations, the inhibitory effects of the isoform‐specific inhibitors atazanavir (UGT1A1) and niflumic acid (UGT1A9) were studied in human microsomes prepared from the liver, intestine and kidney. Without an inhibitor, the catalytic activity for glucuronidation was highest in liver microsomes (~55 pmol M1/[mg liver microsomal protein × min]) compared with intestinal microsomes (~23 pmol M1/[mg intestinal microsomal protein × min]) or kidney microsomes (~12 pmol M1/[mg kidney microsomal protein × min]; data not shown). The effect of both inhibitors was most pronounced in liver microsomes and was considerably higher for 5 µmol L−1 atazanavir (~70% reduction in formation of M1) than for 5 µmol L−1 niflumic acid (~25% reduction), as shown in Table 2. In addition, 10 µmol L−1 hecogenin and 10 µmol L−1 mefenamic acid displayed none or very little activity in liver, intestine and kidney microsomes (Table 2).

TABLE 2

Influence of different standard UGT inhibitors on formation of molidustat glucuronide M1 in pooled human microsomes prepared from different tissues (n = 2 each)

Inhibitor/UGT isoform	M1 formation in % of control in microsomal tissue
	Intestine	Liver	Kidney
Control	100	100	100
5 µmol L−1 atazanavir/UGT1A1	24.8	31.7	98.8
10 µmol L−1 hecogenin/UGT1A4	106.0	104.7	102.4
5 µmol L−1 niflumic acid/UGT1A9	91.0	76.3	19.5
10 µmol L−1 mefenamic acid/UGT2B7(+UGT1A9)	87.0	91.5	70.6

‐, not applicable; UGT, uridine 5'‐diphospho‐glucuronosyltransferase.

The concentration dependence of M‐1 formation was investigated with human liver microsomes (Table S1) and with UGT1A1 (Table S2) as well as with UGT1A9 (Table S3). Determination of apparent K or S50 and V max values was not possible as the affinity of molidustat seems to be low towards enzyme preparations tested, and V max values could not be reached up to a substrate concentration of 96 µmol L−1. The estimated S50 values for human liver microsomes and UGT1A1 are 90 μmol L−1 and about 200 μmol L−1, respectively. The estimated K value for UGT1A9 is about 200 µmol L−1. To further clarify the relative contribution of UGT1A1 and UGT1A9 to molidustat (5 µmol L−1) glucuronidation activity, single‐donor HLMs (N = 23) and one batch of pooled HLMs were characterized for isotype‐specific glucuronidation capacity, using 20 µmol L−1 17β‐estradiol (17β‐estradiol 3‐glucuronidation, UGT1A1) and 100 µmol L−1 propofol (propofol glucuronidation, UGT1A9) as model substrates (Table S4). Subsequently, enzymatic activities for the two isoforms were correlated by linear regression analysis to the enzymatic activity for formation of M1, as depicted in Figure 1. A strong correlation was observed for M1 formation and UGT1A1 activity (17β‐estradiol 3‐glucuronidation), with an r 2 of 0.902; for propofol glucuronidation reflecting UGT1A9 activity, the correlation was much less pronounced (r 2: 0.511). Based on these results, it was concluded that hepatic UGT1A1 was the predominant enzyme responsible for M1 formation in humans, with minor involvement of UGT1A9.

FIGURE 1

Correlation analysis (least‐squares linear regression) for glucuronidation velocity of molidustat versus the isotype‐specific substrates (A) 17β‐estradiol for UGT1A1 and (B) propofol for UGT1A9, as observed in 23 individual (closed symbols) or one pooled human liver microsomes (open symbols). Note. UGT, 5'‐diphospho‐glucuronosyltransferase

Screening for potential DDIs

The inhibitory effects of 176 drugs and xenobiotics from various compound classes (eg anticancer drugs, analgesics, antivirals, antibiotics, antifungals) on the UGT‐mediated glucuronidation of molidustat in HLMs were investigated and assessed based on IC50 values. Results obtained are presented in Table 1 and Table S6, including the respective IC50 values and calculated impact on change in AUC of molidustat (AUCR) using different approximations (if applicable). The most efficient inhibitors, with IC50 values in the nanomolar concentration range, were the kinase inhibitors erlotinib (0.13 µmol L−1), nilotinib (0.17 µmol L−1) and sorafenib (0.32 µmol L−1), but the well‐known UGT1A1 inhibitor atazanavir also exhibited high potency (0.42 µmol L−1), followed by saquinavir (0.47 µmol L−1), montelukast (0.49 µmol L−1), ritonavir (0.60 µmol L−1) and tacrolimus (0.70 µmol L−1). A number of compounds exhibited half‐maximum inhibitory activity in the range of 1‐10 µmol L−1, such as itraconazole (IC50: 1.2 µmol L−1), bilirubin (1.5 µmol L−1), additional tyrosine kinase inhibitors (axitinib, dasatinib, gefitinib, imatinib, lapatinib), human immunodeficiency virus (HIV) protease inhibitors (indinavir, lopinavir, nelfinavir), and atorvastatin, carvedilol, cisapride, curcumin, cyclosporin, daidzein, domperidone, epigallocatechin, everolimus, febuxostat, genistein, glyburide (glibenclamide), ketoconazole, mifepristone, nifedipine, quercetin, raloxifene, rifampicin, telmisartan efavirenz and tranilast.

An in vitro‐in vivo scaling approach was used to estimate the AUCR for molidustat based on a conservative f of 0.9 (glucuronidation of molidustat without differentiation via specific UGTs), using for the perpetrator drugs unbound inlet concentration with default ka values of 0.1 min−1 and 0.01 min−1 (Table 1). Since the glucuronidation of molidustat was performed at a substrate concentration of 5 μmol L−1, which is far less than the estimated K value of the glucuronidation of molidustat in human liver microsomes (90 µmol L−1) K equals to IC50 (K = IC50/(1+[I]/K). Therefore, the IC50 values obtained were used for the calculations.

Using this approach, an increase in AUC of more than fivefold was predicted for the kinase inhibitors erlotinib and nilotinib, and the antiretrovirals atazanavir and indinavir. It is worth noting that none of the substances tested were predicted to be strong inhibitors of molidustat glucuronidation when using a lower k value of 0.01 min‐1 corresponding to a slower absorption rate.

For most of the compounds tested (about 130), no risk of potential DDI in vivo was assumed, based on an increase in molidustat AUC of less than 25% (AUCR < 1.25) and/or a lack of relevant inhibitory activity detected (Table 1 and Table S6). This group of test compounds included widely used drugs such as analgesics (acetaminophen, diclofenac, ibuprofen), cardiovascular drugs (eg atorvastatin, simvastatin, fluvastatin, lovastatin, rosuvastatin, candesartan, metoprolol, carvedilol, verapamil, nifedipine, clopidogrel, ticlopidine, warfarin), proton pump inhibitors (omeprazole, lansoprazole), oral antidiabetics (rosiglitazone, glyburide, tolbutamide) or potential concomitant immunosuppressant medications (eg cyclosporine, tacrolimus, everolimus, dexamethasone).

For atazanavir, the effect on molidustat AUC was calculated for glucuronidation via UGT1A1 only, given that the inhibitory potency on UGT1A9 was much less pronounced. The f value of approximately 0.7 via UGT1A1 was estimated based on the total fraction excreted via glucuronidation (f: 0.9), and the remaining activity in the HLMs incubated with high concentrations of atazanavir (~70% inhibition). Furthermore, using reported k values of 0.024 min−1 for slow absorbers and 0.108 min−1 for fast absorbers resulted in a predicted increase in molidustat AUC of 2.4‐ and 2.9‐fold, respectively. Based on these findings, the in vivo DDI study was planned to verify the in vitro predictions for molidustat with concomitant atazanavir dosing.

Clinical DDI study

Participants

In total, 29 participants were screened, of which 14 passed screening and were treated according to protocol and completed the study; however, after finishing both treatments, one participant was found to be participating in a second clinical trial in parallel and was, therefore, excluded from PK, PD and per‐protocol (PP) study populations because effects of other study medication could not be ruled out. All 13 participants in the PK, PD and PP analysis set were white men with a mean age of 33.1 years (range: 20‐45 years), BMI of 24.9 kg m−2 (19.8‐29.3 kg m−2), height of 182 cm (168‐200 cm) and weight of 82.5 kg (63.5‐101 kg).

Pharmacokinetic results

Plasma concentration‐time profiles for molidustat and M1 (geometric means and geometric standard deviation [SD]) are given in Figure 2. For the parent compound, mean concentrations were higher with concomitant atazanavir than with molidustat alone, but the overall shape of the concentration‐time profile was similar; this indicates lack of relevant effect of atazanavir on the terminal elimination characteristics. The stick plot for AUC of molidustat alone or with concomitant atazanavir (Figure 3) illustrates the general trend for a substantial increase in exposure in all participants with concomitant atazanavir. Descriptive statistical summaries of PK parameters obtained for molidustat and M1 are given in Table 3, including the results from exploratory ANOVA for AUC, AUC(0‐t last) and C max of molidustat. Results confirmed a significant increase of approximately 100% in exposure for molidustat, as point estimates for treatment ratios were 2.07 (90% confidence interval [CI]: 1.87‐2.29), 2.08 (1.88‐2.29) and 2.07 (1.53‐2.79) for AUC, AUC(0‐t last) and C max, respectively. At the same time, AUC and C max for M1 were slightly reduced (AUCR: 0.85 and 0.80, respectively). No effect of atazanavir on t max or terminal half‐life was observed for both analytes (Figure 2).

FIGURE 2

Concentration‐time profile for (A) molidustat in plasma, (B) primary glucuronide metabolite M1 in plasma and (C) endogenous erythropoietin in serum, following a single, oral, immediate‐release dose of 25 mg molidustat alone (part 1, filled symbols) or after 4‐day pre‐treatment and concomitant intake of 400 mg atazanavir qd (part 2, open symbols) to healthy males (geometric mean and SD, n = 13). Note. SD, standard deviation; qd, once daily

FIGURE 3

Stick plot for AUC of molidustat following a single, oral, immediate‐release dose of molidustat 25 mg alone (part 1) or after 4‐day pre‐treatment and concomitant intake of 400 mg atazanavir qd (part 2) to healthy males (individual data and geometric mean [filled circles/bold], n = 13). Note. AUC, area under plasma concentration‐time curve from zero to infinity; qd, once daily

TABLE 3

Pharmacokinetic parameters of molidustat (single dose, 25 mg IR tablet) and its glucuronide (M1) in healthy males after administration alone and after pre‐/co‐treatment with atazanavir 400 mg qd for 5 days (geometric mean/CV%), including point estimates and 90% CIs for ratios between treatments (selected parameters)

Parameter	Unit	Molidustat (n = 13)	Molidustat + atazanavir (n = 13)	Point estimate parameter ratio a	90% CI
					Lower limit	Upper limit
Molidustat
AUC	µg*h L−1	393/25.3	813/22.1	2.0665	1.8674	2.2869
AUC(0‐tlast)	µg*h/L	390/25.2	810/22.1	2.0766	1.8792	2.2947
CL/F	L h−1	63.6/25.3	30.8/22.1
C max	µg L−1	215/38.2	444/38.1	2.0689	1.5343	2.7898
t 1/2	H	6.26/46.1	6.89/54.7
t max b	H	0.75 (0.25‐2.0)	0.75 (0.25‐3.0)
Molidustat glucuronide (M1)
AUC	µg*h L−1	2650/13.1	2240/12.3	0.8472	0.8173	0.8781
AUC(0‐tlast)	µg*h L−1	2630/13.3	2230/12.3	0.8470	0.8168	0.8783
C max	µg L−1	576/21.4	461/12.1	0.8002	0.7426	0.8623
t 1/2	H	9.86/53.9	11.1/37.2
t max b	H	2.0 (1.0‐3.0)	2.0 (1.5‐4.0)

AUC, area under plasma concentration‐time curve from zero to infinity; AUC(0‐tlast), area under the plasma concentration‐time curve from zero to the time of the last quantifiable concentration; CI, confidence interval; CL/F, total body clearance of drug calculated after extravascular administration (eg apparent oral clearance); C max, maximum plasma concentration; CV%, coefficient of variation; qd, once daily; t 1/2, apparent terminal half‐life; t max, time to maximum plasma concentration.

(Molidustat + atazanavir)/molidustat.

Median and range.

PK parameters obtained for atazanavir following the 400 mg once daily (qd) dose received 30 minutes after a light breakfast on day 5 are given in Table S7. Trough concentrations (24 hours after administration) obtained on days 2 and 3 before molidustat administration indicated an increase but, from day 4 onwards, concentrations were comparable, indicating that steady state was reached. Moreover, plasma concentrations observed in the present study were similar to the exposure reported previously. ,

Pharmacodynamic results

Baseline values of EPO concentration were similar for both treatments at 8.0 IU L−1 and 8.2 IU L−1 (geometric mean), respectively, and corresponding CVs were 44.6% and 43.7%, respectively. Although the overall shape of the concentration‐time profiles (geometric mean and geometric SD) for EPO was similar for both treatments, the mean EPO concentrations after dosing were slightly higher in the concomitant treatment with atazanavir compared with molidustat alone (Figure 2C). Geometric mean EPO concentrations increased by 6 hours after administration in both treatments and reached mean peak concentrations at 10 hours after dosing for the treatment molidustat alone (18.9 IU L−1) and 6 hours after dosing for the treatment molidustat + atazanavir (22.6 IU L−1). For both treatments, mean EPO concentrations returned close to baseline values at 24 hours after dosing.

A slightly higher EPO C max was observed in concomitant treatment with atazanavir (23.9 IU L−1) compared with molidustat alone (20.5 IU L−1) and for the ratio to baseline‐corrected factors were 2.91 (CV: 22.6%) compared to 2.56 (CV: 23.8%), respectively (Table 4). The ANOVA resulted in treatment ratios for absolute concentrations and ratio to baseline‐corrected values of 1.17 (90% CI: 1.04‐1.31) and 1.14 (90% CI: 1.00‐1.30), respectively, indicating a minor increase of EPO response after treatment with atazanavir. Regarding EPO AUC(0‐24), also a slightly higher mean value was observed, 423 IU L−1*h compared to 365 IU L−1*h (Table 4), and ANOVA resulted in point estimates for treatment ratios of 1.16 (90% CI: 1.06‐1.27) for absolute concentration data and 1.34 (90% CI: 1.12‐1.61) for baseline‐corrected data.

TABLE 4

Erythropoietin absolute concentrations, as well as baseline‐corrected concentrations, after intake of a single oral immediate‐release tablet dose of 25 mg molidustat in healthy males alone and after pre‐ and co‐treatment with 400 mg atazanavir qd

Parameter	N	Geometric mean/CV%	Arithmetic mean ± SD	Range
C max
Absolute EPO concentration, IU L−1
Molidustat	13	20.5/33.5	21.5 ± 6.23	11.5‐31.1
Molidustat + atazanavir	13	23.9/41.8	25.8 ± 11.3	13.2‐55.8
Difference to baseline, IU L−1
Molidustat	13	‐	12.8 ± 4.02	5.60‐18.6
Molidustat + atazanavir	13	‐	16.9 ± 8.28	8.25‐37.9
Ratio to baseline
Molidustat	13	2.56/23.8	2.62 ± 0.578	1.59‐3.34
Molidustat + atazanavir	13	2.91/22.6	2.98 ± 0.715	2.08‐4.80
AUC(0‐24)
Absolute EPO concentration, IU L−1*h
Molidustat	13	365/35.8	385 ± 121	197‐578
Molidustat + atazanavir	13	423/37.0	448 ± 161	223‐827
Difference to baseline, IU L−1
Molidustat	13	164/45.7	177 ± 67.1	61.6‐293
Molidustat + atazanavir	13	220/38.2	234 ± 88.3	124‐396
t max
Time, h
Molidustat	13	10.0 a	‐	6.00‐12.0
Molidustat + atazanavir	13	8.00 a	‐	4.0‐12.0

‐, not determined; AUC0‐24, area under the plasma concentration‐time curve from zero to 24 hours; C max, maximum plasma concentration; %CV, coefficient of variation; EPO, erythropoietin; qd, once daily; SD, standard deviation; t max, time to maximum plasma concentration.

Median.

Safety results

Safety evaluation was performed on the safety analysis set that included all 14 participants treated with molidustat. The study drugs were well tolerated; all treatment‐emergent AEs observed are summarized in Table 5. No severe or serious AEs related to molidustat or the combination with atazanavir occurred. Treatment‐emergent AEs were only observed for the combination of molidustat and atazanavir, for example, the well‐known effect of atazanavir on hepatic conjugation of bile acids (hyperbilirubinemia), which was observed in three participants. All AEs resolved by the end of the observation period without therapy being necessary.

TABLE 5

Number of participants experiencing treatment‐emergent adverse events after intake of a single oral dose of 25 mg molidustat with or without pre‐ and co‐treatment with atazanavir (400 mg qd for 5 days, n = 14)

MedDRA primary system organ class preferred term	Molidustat (n = 14)	Molidustat + Atazanavir (n = 14)
Participants with ≥ 1 TEAE, n (%)	0	4 (28.6)
Gastrointestinal disorders
Abdominal pain	0	1 (7.1)
Hepatobiliary disorders
Jaundice	0	3 (21.4) a
Nervous system disorders
Headache	0	1 (7.1)

MedDRA, Medical Dictionary for Regulatory Activities; qd, once daily; TEAE, treatment‐emergent adverse event.

Considered by investigator to be related to atazanavir.

DISCUSSION

The mass balance study for molidustat confirmed that about 85% of an oral dose is cleared as N‐glucuronide M1 in urine, with minor excretion of unchanged molidustat (~4% via urine; ~6% via faeces). Based on the current in vitro data, including the correlation analysis, UGT1A1 is considered to be the major metabolizing enzyme of molidustat to yield M1, whereas conjugation via UGT1A9 is of minor importance. Comparable correlation analysis with isotype‐specific substrates used for identification of relevant UGT isoforms has been used in the past for characterization of biotransformation of xenobiotics.

An extensive in vitro DDI screening using HLMs revealed several potential inhibitors of the N‐glucuronidation of molidustat via UGT1A1. Like our results, the high inhibitory potency of tyrosine kinase inhibitors, including erlotinib, sorafenib, lapatinib, axitinib and imatinib, has been reported for UGT1A1. , , , Furthermore, clinically relevant DDIs with drugs predominantly cleared via UGT1A1 were predicted previously for erlotinib based on AUC ratios.

Owing to the in vitro results indicating potential DDIs mediated by UGT1A1, the effects of inhibitory co‐medication on the PK and PD of molidustat were evaluated. The predominant role of hepatic UGT1A1 found for molidustat is in line with the established characteristics of glucuronidation of xenobiotics, although intestinal and renal involvement appears to be limited. Atazanavir was chosen as a relevant UGT1A1 inhibitor for the clinical DDI study based on its established use for UGT1A1 DDI studies and the prediction from in vitro data that AUC for molidustat could increase approximately by a factor of 2‐ to 3‐fold.

In the clinical DDI study, pre‐ and co‐treatment with atazanavir resulted in a twofold increase of AUC for molidustat, supporting the assumptions made from the in vitro‐in vivo correlation regarding predominant clearance of molidustat via UGT1A1‐mediated conjugation. However, the exposure to the major metabolite M1 was still approximately 85% in the presence of atazanavir, indicating that UGT1A1 activity might be reduced in the presence of atazanavir, but not completely inhibited.

Given that the concentration‐time profile of molidustat was shifted upwards in a parallel fashion and terminal half‐life of molidustat was only marginally increased in the presence of the UGT1A1 inhibitor atazanavir (6.9 hours vs 6.3 hours), a relevant change in systemic clearance is unlikely. The absolute bioavailability of molidustat following IR tablet administration has been reported to be 59%. Therefore, the increase in exposure observed with atazanavir is likely due to increased bioavailability rather than to a decrease in systemic clearance. Several publications have indicated that glucuronidation in the liver occurs mainly in the perivenous hepatocytes and less frequently in the periportal hepatocytes. , , Considering the low UGT activity in the periportal zone and the high atazanavir and molidustat concentrations coming from the portal vein, it can be assumed that only the UGTs in the periportal zone can be relevantly inhibited by atazanavir. However, the activity of the UGTs in the perivenous hepatocytes is less affected. This might explain the influence of atazanavir on pre‐systemic clearance and the lack of any effect on systemic clearance.

Historically, there has been only a minor fraction of marketed drugs relying on UGTs as the predominant/exclusive clearance pathway; therefore, DDIs via UGT have been of limited clinical relevance. However, recent data indicate that some new drugs are significantly cleared via UGTs. For danirixin, it has been found that approximately 90% of the drug is excreted as glucuronides, with 80% being recovered as directly conjugated parent compound without having undergone oxidative metabolism. Likewise, dolutegravir is mainly cleared via UGT1A1‐mediated glucuronidation, which was sensitive to atazanavir co‐medication with an effect comparable to that described here on molidustat PK at a dose of 400 mg qd.

Recently, it could be shown for atazanavir and raltegravir that, in healthy participants and patients, this combination at therapeutic doses resulted in increased exposure to raltegravir (eg ~70% in mean AUC for single‐dose raltegravir 100 mg when given with multiple doses of atazanavir 400 mg), with a clear correlation of exposure for both drugs (ie high AUC for atazanavir resulted in about twofold higher exposure of raltegravir). , , , For the DDI study reported here, no correlation of atazanavir AUC with exposure of molidustat could be found, which might indicate sufficient exposure to atazanavir to achieve maximum inhibitory effect on molidustat glucuronidation over the range of variability observed in the population of healthy, male participants included.

In addition, in this study, the consequence of increases in molidustat plasma concentrations on the PD effect of EPO increase was investigated. Notably, after a single oral dose of 25 mg molidustat the increase of molidustat Cmax and AUC by about a factor of 2 due to atazanavir co‐treatment had only minor influence on the EPO response. The concomitant intake of atazanavir increased EPO Cmax by 17% and AUC(0‐24) by 16% when compared to molidustat given alone. Coadministration of molidustat and atazanavir was well tolerated.

In the clinical setting, molidustat treatment starts from low dose and is intended for individual titration based on haemoglobin response, and such minor differences in EPO response might not translate into clinically meaningful influences on haemoglobin response. Therefore, it is thought that atazanavir can be co‐administered with molidustat.

CONCLUSIONS

The predominant clearance pathway for molidustat in humans following an oral dosing is N‐glucuronidation via UGT1A1 and, to a lesser extent, via UGT1A9. The resulting metabolite M1 subsequently undergoes renal elimination, representing the major pathway for excretion; only negligible amounts of parent compound are cleared unchanged via urine or faeces. In total, 176 possible interaction partners were screened in vitro to identify potential inhibitors of the major clearance pathway of molidustat. Many important drugs lacked significant influence on molidustat exposure based on an in vitro‐in vivo scaling approach using inhibitory potency for molidustat glucuronidation. UGT‐mediated molidustat glucuronidation is potently inhibited in vitro, especially by tyrosine kinase inhibitors, as well as by HIV protease inhibitors and by tranilast. Given that atazanavir, an established UGT1A1 inhibitor, was identified as a potent inhibitor in vitro, the influence of repeated atazanavir dosing on the PK and PD of molidustat was studied in healthy participants. The observed AUC ratio of 2 for molidustat with and without inhibition of UGT1A1 by atazanavir was close to the predicted 2‐ to 3‐fold increase from in vitro‐in vivo correlation, assuming a fm value of 0.7. Comparable effects were observed for C max of molidustat (twofold increase), although only a mild reduction in systemic exposure to the glucuronide M1 was found (−15%). The effects of UGT inhibition on EPO turnover were much less pronounced than on the PK of molidustat, confirming the exposure‐response relationship observed in other studies. , Molidustat administered concomitantly with atazanavir was well tolerated. Given that molidustat dosing will follow a haemoglobin‐guided individual dose titration, molidustat may be administered together with atazanavir.

It is considered unlikely that other UGT inhibitors will result in a more pronounced increase in molidustat exposure, since atazanavir was identified as one of the most potent ones in vitro.

CONFLICT OF INTEREST

DvdM, MG, NJ, AK, SS, MR and SL are employees of Bayer AG, and KY is an employee of Bayer Yakuhin. In addition, MG, DvdM, SS and SL have stock in Bayer AG, but are not paid in stock or stock options.

Supplementary Material

Click here for additional data file.