Visual Overview
Abstract
Pimasertib (AS703026 or MSC1936369B) is a selective inhibitor of MEK1/2, the mitogen-activated protein kinase (MAPK) signaling pathway, which is often dysregulated in cancer cells. Pimasertib has shown potent preclinical antitumor activity and its clinical activity is being investigated in various tumor types. In this phase I study, the disposition and biotransformation of 14C-radiolabeled pimasertib was investigated in six patients with locally advanced or metastatic solid tumors (NCT01713036). Ultra-performance liquid chromatography–mass spectrometry and radiodetection techniques were used to investigate the profiles and structures of metabolites in plasma, urine, and feces after a single oral dose of 14C-pimasertib. A total of 14 different phase I and II metabolites of 14C-pimasertib were detected, which were principally generated through oxidations and conjugations (direct and indirect); but other reactions included isomerization, N-dealkylation, deamination, and deiodination to form minor metabolites. Two major metabolites (>10% of total drug-related material), M554 and M445, were identified in plasma and urine. In feces, M445 was the primary metabolite with only trace amounts of M554 excreted. All other metabolites, including enantiomers of M445 and pimasertib, were detected to a lesser extent (<5%) in these matrices. M445 was identified as a carboxylic acid of pimasertib. M554 was identified as a novel phosphoethanolamine conjugate on the propanediol moiety of pimasertib by high-resolution mass spectrometry and multiple nuclear magnetic resonance spectroscopy techniques. To our knowledge, a phosphoethanolamine conjugate is a novel metabolite not previously described for a pharmaceutical agent and requires detailed further investigations to understand any implications.
Introduction
The mitogen-activated protein kinase (MAPK) pathway includes the signaling proteins RAS, RAF, MEK, and ERK, which have a key regulatory role in normal cell functions such as proliferation, survival, growth, migration, and differentiation (Santarpia et al., 2012; Akinleye et al., 2013). However, dysregulation of this pathway is implicated in tumor development and mutations in RAS and RAF genes have been identified in several human cancers (Fernandez-Medarde and Santos, 2011; Santarpia et al., 2012). Consequently, the MAPK pathway is a target for potential anticancer agents, and inhibitors of the different signaling proteins and upstream activators of this pathway are in development (Santarpia et al., 2012).
Pimasertib (AS703026 or MSC1936369B) is a selective adenosine triphosphate noncompetitive inhibitor of MEK1/2 (Goutopoulos et al., 2009). Preclinical examination of single-agent pimasertib has confirmed potent antitumor activity in vitro and in vivo in tumor cells derived from multiple myeloma and colorectal cancer harboring aberrant MAPK pathway signaling (Kim et al., 2010; Yoon et al., 2011). Furthermore, pimasertib combined with the BRAF inhibitor PLX4032 has been shown to induce apoptosis in cell lines derived from BRAF-mutated human malignant melanoma (Park et al., 2013).
Pimasertib is currently in clinical development for the treatment of a number of solid and hematologic cancers. As part of the development process for any new pharmaceutical agent, it is vital to fully understand its absorption, distribution, metabolism, and excretion (ADME) properties. Metabolite data can confirm a compound’s suitability for use in humans. It may highlight the requirement for additional in vitro and/or in vivo toxicology investigations and additional in vitro investigations to determine the drug-drug interaction potential of any major metabolites. This understanding is critical because, for example, drug interactions have the potential to impede treatment efficacy, as well as to put patients at risk of adverse events. In addition, potential drug interactions impact on the recommended dose and frequency of a drug, and can highlight certain patient subgroups that may not be suitable for the treatment being investigated. Mass balance studies, which use 14C radioactive labeling in collected plasma and excreta to identify and quantify the main elimination pathways and metabolites of drugs in vivo, are basic trials to characterize the pharmacokinetics (PK) of drugs and, therefore, are also recommended by such guidelines as the US Food and Drug Administration Guidance for Industry, Safety Testing of Drug Metabolites (Center for Drug Evaluation and Research, 2008), the International Conference on Harmonisation (ICH), Non-Clinical Safety Studies for the Conduct of Human Clinical Trials for Pharmaceuticals (International Conference on Harmonisation, 2009), and the European Medicines Agency’s Guideline on the Investigation of Drug Interactions (European Medicines Agency, 2012).
Final mass balance and total radioactivity data as well as early clinical findings have been reported previously from the human mass balance phase I study (NCT 01713036) that investigated the metabolic profile, mass balance, and oral bioavailability of pimasertib in patients with locally advanced or metastatic solid tumors (von Richter et al., 2016). In the current report, we describe in more detail the biotransformation of pimasertib following a single oral dose in the phase I study, and identify the metabolic profile of pimasertib in patients with advanced cancers.
Materials and Methods
Study Drug, Reference Compounds, and Chemicals
Unlabeled pimasertib (N-[(2S)-2,3-dihydroxypropyl]-3-[(2-fluoro-4-iodophenyl)amino] isonicotinamide; MSC1936369B) was supplied as the hydrochloride salt (≥99.3% chemical purity) by Merck Serono, Tiburtina, Italy. Radiolabeled [14C]-pimasertib was supplied as the hydrochloride salt (99.5% and 100% chemical and radiochemical purity, respectively, and specific activity of 2.12 GBq/mmol or 4.52 MBq/mg salt) by Merck KGaA, (Grafing, Germany). The [14C]-label was positioned in a metabolically stable position at the C-12 of pimasertib. Metabolites were named “M”, followed by their molecular weight, e.g., M445. Reference compounds of metabolites included the R-enantiomer of pimasertib (MSC1940796B, 98.9% purity; Merck Serono), MSC2359570A (M445; 98.9% purity; BoaoPharma, Inc., Natick, MA), MSC2486240A (M445 R-enantiomer; 96.0% purity; Boaopharma, Inc.), MSC1919744A (M357; 98.9% purity; Sanmar Specialty Chemicals Ltd., Chennai, India), and MSC2521347A (M554; 99.5% purity; Boaopharma, Inc., Jiangsu, China). All specified chemical and radiochemical purities were determined by high-performance liquid chromatography (HPLC). All other chemicals, solvents and consumables used were obtained from commercial sources and were at least of analytical grade.
Clinical Study Conduct
The clinical study (EudraCT number 2012-002562-12) was conducted by PRA Hungary Ltd. (Budapest, Hungary) in accordance with the Declaration of Helsinki, the ICH Good Clinical Practice (GCP) guideline (ICH Topic E6, GCP), the study protocol, local legislation regarding the rights and welfare of human participants in biomedical research, and applicable regulatory requirements. All participants provided written informed consent. All study-related materials were reviewed and approved by an independent ethics committee. Metabolite analyses of the human samples generated from this study were conducted in accordance with the principles of Good Laboratory Practice at Merck KGaA, Grafing, Germany.
Study Design
An open-label, single-center trial conducted in six male patients [age 18–65 years; Eastern Cooperative Oncology Group (ECOG) performance status ≤1] with locally advanced or metastatic solid tumors [defined by Response Evaluation Criteria In Solid Tumors (RECIST) v1.1] was performed. Full details of the patients, study design, dosing, and execution of the clinical part of this phase I study have previously been reported (von Richter et al., 2016). This report provides details of the identification of [14C]-pimasertib metabolites in biologic samples taken from these patients.
Metabolite profiling and identification was assessed following oral administration of unlabeled pimasertib (60 mg) spiked with [14C]-pimasertib [2.6 MBq (70 μCi)] given as four capsules with 240 ml tap water.
Sample Collection and Storage
Venous blood samples were collected predose and then at regular intervals from 0.5 to 168 hours post–oral 14C-pimasertib administration (predose and postdose samples at 1, 2, 4, 10, and 24 hours were used for metabolite profiling). Urine samples were collected predose and then starting on day 8 at 0–4, 4–8, 8–12, 12–24, 24–48, 48–72, and 72–96 hours post–oral 14C-pimasertib dosing; feces samples were collected 0–12, 12–24, 24–48, 48–72, and 72–96 hours post–oral 14C-pimasertib dosing. On the basis of the 72- to 96-hour samples, collection of urine and feces was extended by further 24-hour collection intervals (up to 288–312 hours) if >1% of the administered radioactive dose was recovered in excreta over 2 consecutive days.
Whole blood (1 ml) was transferred to a polypropylene tube and the remainder was centrifuged (10 minutes, 1500g) to obtain plasma. For each plasma sample, 1 ml was acidified with 20% orthophosphoric acid (100 μl, acidified plasma) and the remainder stored (neutral plasma). Individual fecal samples (per participant per interval) were homogenized (2 minutes with 1- to 2-fold weight equivalent of water) before storage. Whole blood, plasma, urine, and fecal samples were stored in polypropylene tubes at –20°C until analysis for total radioactivity samples and below –70°C for metabolite investigation samples.
Determination of Radioactivity by Liquid Scintillation Counting
Scintillation cocktail was added directly to plasma and urine samples, whereas samples of whole blood were solubilized and then decolored, and fecal samples were dried and combusted before the addition of scintillation cocktail (see Supplemental Methods for full details of sample preparation and internal validation) and analysis by liquid scintillation counting (LSC). Samples were analyzed in duplicate until a statistical error (2σ) of 0.5% was obtained with maximum counting times of 10 minutes (plasma, urine, and feces samples) or 20 minutes (whole blood) on a PerkinElmer Tri-Carb 3100 using low level and normal count modes (PerkinElmer Inc., Waltham, MA) to determine total 14C radioactivity. Analysis of 14C radioactivity was validated over the ranges [lower limit of quantification (LOQ) to upper LOQ] 50 dpm/ml to 20,000 dpm/ml for whole blood, 30 dpm/ml to 80,000 dpm/ml for plasma, 10 dpm/ml to 1,000,000 dpm/ml for urine, and 40 dpm/ml to 1,000,000 dpm/ml for fecal homogenate.
LOQ for extraction recovery was 19–21 dpm for both extracts and oxidized samples of plasma and feces. Recovery of extracted radioactivity was ≥84% and ≥92% in plasma and feces samples, respectively; nonextractable radioactivity recovery was ≤17% and ≤7%, respectively. Take-up efficacy determined for final analyzed extracts before and after evaporation was ≥74%, ≥88%, and ≥84% in plasma, feces, and urine, respectively (see Supplemental Methods for internal validation procedure).
Metabolite Profiling and Identification
Sample Preparation.
Plasma samples were analyzed from each participant at individual time points in neutral and acidified plasma. For urine and feces, each participant’s time-interval samples were pooled to obtain representative urine and feces pools that accounted for >90% of the total excreted radioactivity for each matrix (either urine or feces) and individual. Samples were stored at –80°C until analysis.
Plasma.
Samples underwent a three-step extraction process: 1) to plasma (1 ml), 5 ml of methanol/acetonitrile/water (35+35+30, v/v/v) was added, the sample shaken (up to 30 minutes), cooled (–20°C for 10 minutes) and centrifuged (5000 rpm, 10 minutes), and the superior organic phase transferred to a clean tube; 2) acetonitrile (4 ml) was added to the remaining precipitate, mixed, and centrifuged as previously; the two superior organic phase extracts were combined, shaken (up to 30 minutes) and centrifuged (5000–6000 rpm, 10 minutes), and transferred to a clean tube and dried under a stream of nitrogen; 3) to this combined dried residue, water (1 ml) and n-hexane (5 ml) were added, and the sample was shaken (up to 30 minutes) and centrifuged (5000–6000 rpm, 10 minutes). The n-hexane layer was discarded and the aqueous phase dried under nitrogen. The residue was resuspended in 150 μl of water/methanol/acetonitrile (70+15+15, v/v/v), vortex mixed and sonicated, and analyzed by ultra-performance liquid chromatography–mass spectrometry (UPLC)-MS (50 μl) and LSC (20 μl). Acidified plasma samples were processed as above with the exception that the volumes for the three extractions were: 1) 2.5 ml methanol/acetonitrile/water (35+35+30, v/v/v) was added to 0.5 ml plasma; 2) 2.5 ml acetonitrile; 3) 2.5 ml of n-hexane, with reconstitution in 75 μl water/methanol/acetonitrile (70+15+15, v/v/v).
Urine.
Pooled urine samples (2.5 ml) were dried under nitrogen and the residue resuspended in 500 μl of water/methanol/acetonitrile (70+15+15, v/v/v) before analysis by LSC (20 μl) and UPLC-MS (50 μl).
Feces.
Fecal homogenate (1 g) was processed per the plasma aliquots. However, following the first two extractions (methanol/acetonitrile/water [35+35+30, v/v/v] then acetonitrile) further extractions were performed if the extraction recovery was not sufficient: 2 × 5 ml methanol/acetonitrile/water (35+35+30, v/v/v), 1 × 5 ml methanol, 2 × 5 ml methanol/acetonitrile/water (35+35+30, v/v/v). All extracts were combined and dried under nitrogen and the residue was washed twice with n-hexane (1 ml water plus 5 ml n-hexane). The remaining aqueous phase was dried under nitrogen, reconstituted, and analyzed per the plasma samples.
Glucuronide conjugates in plasma.
Selected plasma samples (100 μl) were dried under a stream of nitrogen, reconstituted (300 μl, 70 mM phosphate buffer pH 7.4), and incubated with β-glucuronidase (50 µl; Sigma-Aldrich, Steinheim, Germany) for 2 hours (37°C, Eppendorf ThermoMixer; Eppendorf, Hamburg, Germany). Acetonitrile (700 μl) was added to stop the reaction; the samples were dried under nitrogen, reconstituted in 100 μl water/methanol acetonitrile/ (70+15+15, v/v/v), and an aliquot (50 μl) was analyzed by UPLC. Likewise, urine samples (2.5 ml) were preincubated (37°C) before the addition of β-glucuronidase (20 μl); samples were then mixed and incubated in a water bath for 2 hours (37°C). The reaction was stopped (5 ml acetonitrile), samples were evaporated to dryness under nitrogen, the residue was reconstituted in 500 μl water/methanol/acetonitrile/ (70+15+15, v/v/v), and an aliquot (50 μl) analyzed by UPLC.
UPLC.
To determine the metabolic profile of [14C]-pimasertib, samples of plasma, urine, and feces were analyzed by UPLC (Waters Acquity system; Waters GmbH, Eschborn, Germany). Reverse-phase separation of pimasertib and its metabolites was performed using a Waters Atlantis T3 column (150 × 3 mm, 3 μm; Waters GmbH) at 35°C with mobile phase of acetonitrile (A): 10 mM ammonium formate, pH 3.5 (B) at a gradient of 5:95, 30:70, 90:10: 5:95 A:B (%) over 50 minutes at a flow rate of 0.8 ml/min. Separation of the two acid enantiomer metabolites, MSC2359570 and MSC2486240, was achieved with a Daicel Chiralpak ZWIX(-) column (150 × 4 mm, 3 μm; Chiral Technologies, Illkirch, France) at 40°C with a mobile phase of methanol/50 mM formic acid plus 50 mM NH4-formate (90:10, v/v) flowing at 1.0 ml/min. For the separation of pimasertib and its enantiomers (MSC1936369 and MSC1940796, respectively) a Daicel Chiralpak AD-3 (100 × 4.6 mm, 3 μm; Chiral Technologies) column was used (at 40°C) with a mobile phase of methanol with 0.1% acetic acid flowing at 0.9 ml/min.
Column recovery of radioactivity from reverse phase and enantioselective separations was determined in 14C-pimasertib-spiked samples of nonextracted plasma, urine, and fecal or pooled samples. Column recovery was ≥94% for all three separation methods.
Microplate Scintillation Counting.
Sample fractions (20 μl) were collected throughout the chromatographic run (every 7 seconds and 2.3 seconds for reverse-phase and enantioselective separations, respectively) into 96-well plates (LumaPlate; PerkinElmer, Rodgau, Germany). After drying under a gentle air stream the radioactivity [counts per minute (cpm)] was determined using a microplate scintillation counter (TopCount NXT; PerkinElmer) with an integration limit of 1% of the total regions of interest.
Mass Spectrometry.
Metabolite identification was performed using a LTQ/Fourier Transform (FT) Orbitrap XL mass spectrometer controlled by XCalibur software version 2.1 (ThermoFisher Scientific, Dreieich, Germany) coupled to a UPLC system (described above). Full-scan ion trap and FT spectra provided the masses of molecular ions of the parent and metabolites. Subsequent tandem mass spectrometry (MS/MS) fragmentation spectra via data-dependent acquisition (DDA) of molecular ions selected via linear ion trap were produced following collision-induced dissociation (CID) (collision energy: 25) using helium as the collision gas. Electrospray ionization (ESI) was used in positive mode with a source voltage of 4.5 kV and a capillary temperature of 350°C.
Following metabolite identification, metabolite confirmation by alignment of the identified metabolites to the radioactive peaks was assessed using a linear ion trap API4000 QTrap mass spectrometer controlled by Analyst software version 1.4.2 (AB Sciex, Framingham, MA) coupled to a UPLC system with simultaneous radio detection through fraction collection and microplate scintillation counting (as described above). The linear ion trap MS ran with ESI in the positive mode and specific multiple-reaction monitoring scanning for each of the identified metabolites. Nitrogen was used as the collision gas.
Structural Elucidation of Metabolite M554
M554 Generation in HepRG Cells.
HepaRG cells from human donors were acclimatized for 4 days prior to incubation (13 T25 flasks containing 7 × 106 cells in Recovery/Tox HepaRG medium; Biopredic International, Saint-Grégoire, France). The culture media was changed (6 ml Maintenance/Metabolism HepaRG medium; ThermoFisher Scientific), and 60 μl stock solution was added [unlabeled pimasertib (5 mM) and 14C-pimasertib (20 μΜ) in ethanol/water 1+1, v/v]. On days 3, 5, and 8 the culture medium was removed and stored (20°C), and replaced with 6 ml of test solution containing unlabeled pimasertib (50 μΜ) and 14C-pimasertib (0.2 μΜ) in Maintenance/Metabolism HepaRG medium. Supernatants (5 ml per flask) from each day were pooled.
Extraction and Purification of M554.
M554 in pooled supernatants was extracted and concentrated by elution through a LiChrolut RP-18 SPE column (40–63 μm, 6 ml, 2 g sorbent; Merck Millipore, Billerica, MA) using methanol (3 × 1 ml). Methanol eluents were dried under a stream of nitrogen and reconstituted in 1 ml of acetonitrile/10 mM ammonium formate buffer pH 3.5 (10+90, v/v). Aliquots (1 μl) were tested by UPLC-MS to ensure that M554 had been extracted, after which the sample underwent a three-step purification process by radio-HPLC: 1) An aliquot (100 μl) of extracted supernatant was separated by HPLC (Agilent Technologies Deutschland GmbH, Böblingen, Germany) with online radiodetection (Flow Scintillation Analyzer Radiomatic 525 TR; PerkinElmer) using a Waters Atlantis T3 column (150 × 3 mm, 3 μm; Waters GmbH) at 35°C with mobile phase of acetonitrile (A):10 mM ammonium formate, pH 3.5 (B) at a gradient of 5:95, 30:70, 90:10: 5:95 A:B (%) over 50 minutes at a flow rate of 0.8 ml/min. A fraction of eluent was taken at a retention time of 26.5 minutes for 4 minutes and evaporated under a stream of nitrogen. This was repeated 9 × 100 μl samples; each fraction was collected and evaporated in the same vial. The final sample was reconstituted in 1 ml of acetonitrile/10 mM ammonium formate, pH 3.5 (10+90, v/v), tested (1 μl for a radiochromatogram and UPLC-MS), and the remainder purified further. 2) Step 1 was repeated, with the exception that fractions were collected manually after the 26.5-minute retention time when prompted by a positive signal at 254 nm using a variable wavelength detector (M554 retention time: ∼28.5 minutes). 3) Step 2 was repeated, with the exception that water replaced ammonium formate buffer in the mobile phase so that no matrix coeluted with M554 and to minimize the salt content of the final sample. The final sample was reconstituted in 2 ml acetonitrile/water (10:90, v/v), tested (10 μl) by LSC (overall yield) and for a radiochromatogram (metabolite profile and purity), and evaporated to dryness with nitrogen for nuclear magnetic resonance spectroscopy (NMR) analysis.
M554 Structure Identification.
Purified M554 fractions were reconstituted in methanol/water (1+1, v/v; 300 μl) and analyzed by UPLC (see reverse-phase method described above) coupled to a LTQ/FT Orbitrap XL mass spectrometer with an ESI interface (ThermoFisher Scientific). The UPLC-MS analytical conditions were as described previously, with online radiodetection and UV detection at 254 nm for the UPLC separation to identify M554 (retention time 26.8 minutes).
Purified and dried M554 fractions for NMR analysis were dissolved in dimethylsulfoxide-D6. These fractions were measured on a Bruker Avance 500 spectrometer with 5 mm CPTCI probe head (Bruker BioSpin GmbH, Rheinstetten, Germany) and measured at 303K to obtain 1H-, 13C-, correlation spectroscopy (COSY), heteronuclear single-quantum correlation spectroscopy (HSQC), heteronuclear multiple-bond correlation spectroscopy (HMBC), and rotating frame nuclear Overhauser effect spectroscopy (ROESY) spectra. 31P- and 1H-31P-HMBC spectra were obtained using a Bruker Avance 400 spectrometer with 5-mm PABBO probe head (Bruker BioSpin GmbH) and measured at 303K.
Calculations
All pharmacokinetic (PK) analyses, including the areas under the curves for total radioactivity, pimasertib, and metabolite (AUCTR, AUCtest, and AUCmetabolite) have been described previously (von Richter et al., 2016). In the current study, the percentage of metabolite in plasma was calculated from: (metabolite peak/total peak area) × 100. In urine and feces, the percentage of administered dose was calculated: (metabolite peak area/total peak area) × % dose excreted cumulativetotal rad, where total peak area is the sum of all radioactive metabolite peak areas. No formal statistical analysis was applied in this study. Descriptive statistics are provided where appropriate.
Results
Six patients with advanced cancers and a median age of 57.5 (40–64) years were enrolled in the study. Patient characteristics, mass balance outcomes, and PK profiles of pimasertib and its main metabolites have been described previously (von Richter et al., 2016). The current report details the structural identification and quantification of pimasertib’s metabolites following oral administration of [14C]-pimasertib to these patients.
Metabolite Profile, Identification, and Quantification
Metabolite Profile.
In total, [14C]-pimasertib was extensively metabolized to 14 different phase I and phase II metabolites, which were assigned the identifiers M305, M357 (MSC1919744), M358, M373, M415, MSC1940796 (R-enantiomer of pimasertib), M445 (MSC2486240), M445 enantiomer (MSC2359570), M447-1, M554 (MSC2521347), M607-1, M607-2, M623-1, and M623-2. Figure 1 shows typical 14C-radiochromatogams from samples of plasma, urine, and feces following metabolite confirmation by UPLC coupled to ion trap MS.
Individual metabolite profiles in plasma from the six patients are shown in Table 1. Two major metabolites were detected in plasma (M554 and M445), which together accounted for nearly half of the total radioactivity exposure in plasma [29.8% and 19.2%, respectively, of the mean total radioactivity exposure (AUC0–24)] (Table 1). Pimasertib and its enantiomer accounted for the next highest exposure (16.5%; sum total for the percentage of AUC0–24), with all other metabolites detected at <2% of the mean total radioactivity exposure (AUC0–24). Notably, the metabolic profiles of pimasertib in these six individuals were variable with some metabolites (M358+M623-2, M607-2, and M447-1) not being detected in some patients (Table 1). Over half of the administered 14C-pimasertib dose was eliminated in urine, with fecal excretion secondary to this. The relative contributions of all the metabolites identified in these different matrices are summarized in Fig. 2. The major metabolites in urine were M445 and M554, which were eliminated to a similar extent [19.1% (range 14.1–23.9%) and 16.8% (range 8.8–22.4%), respectively]. In feces, however, M445 was the main metabolite excreted [18.2% of administered dose (range 7.5–25.6%)], with only trace amounts (0.1%) of M554 detected. With the exception of M358 and M623-2, which coeluted (together 4.5% of dose), all other metabolites were excreted as <3% of the administered dose (0.0–2.7%). Unchanged [14C]-pimasertib was found in nominal amounts in both urine and feces (Fig. 2).
Enantioselective analysis was performed to assess enantiomeric conversion rates for pimasertib and its enantiomer, MSC1940796 (not in plasma owing to insufficient plasma volume), and for M445 and its enantiomer, MSC2486240. Conversion to the other enantiomer was very low for both compounds (Fig. 2). Development of the R-enantiomer of M445 was the only further metabolism observed for M445 (0.9% and 1.6% in urine and feces of the radioactive dose administered). There was no evidence of any further metabolism of M554 (Fig. 2), but enantioselective analysis could not be performed for M554 owing to limited enantiopurity of the reference standard.
Metabolite Identification.
The majority of the 14C-pimasertib dose excreted in urine and feces could be structurally identified (Fig. 2). The principal circulating and excreted entities were a phosphoethanolamine conjugate (M554; MSC2521347) and a carboxylic acid (M445; MSC2359570).
Isomerization of pimasertib produced its R-enantiomer, MSC1940796, whereas the carboxylic acid metabolite M445, its R-enantiomer (M445 enantiomer), and the acid M415 were formed through alcohol oxidation and oxidative cleavage at the propanediol moiety of pimasertib. N-dealkylation and hydrolytic oxidation of the amino-propanediol moiety of pimasertib generated the amide metabolite M357 and the depropanediol acid M358, respectively; however, M358 could also have been generated through oxidative deamination of M357. Further oxidation of the pyridine moiety of M357 produced the pyridine N-oxide metabolite M373. Deiodination of pimasertib yielded the metabolite M305.
Direct conjugation of pimasertib with glucuronic acid generated the metabolites M607-1 and -2. Monohydroxylation or N-oxidation of the (2-fluoro-4-iodophenyl)-pyridin-3-amine moiety produced M447-1, from which further glucuronic acid conjugations formed the metabolites M623-1 and M623-2. However, the exact positions of the oxidations and glucuronidations cannot be confirmed from mass spectral data alone. One unusual metabolite, M554, required further elucidation to determine the final proposed structure of this phosphoethanolamine conjugate.
Liquid Chromatography–Tandem Mass Spectrometry Characterization of the Metabolites
The chemical structures of the different metabolites were proposed on the basis of high-resolution MS and MS/MS data. Structures of M357 (MSC1919744), MSC1940796 (R-enantiomer of pimasertib), M445 (MSC2486240), M445 enantiomer (MSC2359570), and M554 (MSC2521347) were in addition confirmed against reference material. The MS characteristics, proposed sum formulas, key fragment ions, and the mass shift toward pimasertib for each of the metabolites identified are presented in Table 2. The chemical structures of the 14 pimasertib metabolites, their proposed metabolic pathways, and their relative contributions in the different matrices are summarized in Fig. 2. The detailed MS/MS spectra of the metabolites with the associated fragment proposals are shown in Supplemental Fig. 1.
Pimasertib.
The CID product ion FT mass spectrum in positive mode of the test item gave prominent fragment ions at m/z 414, 359, and 214, as well as the pseudomolecular ion at m/z 432. The fragment ion m/z 414 is formed by neutral loss of 18 Da (water) from one of the hydroxyl groups of the pseudomolecular ion. Further loss of 3-amino-propenol (C3H7NO) together with the addition of water formed fragment m/z 359. Further on, the key fragment m/z 214 (8-(2-fluorophenyl)-3,8-diaza-bicyclo[4.2.0]octa-1,3,5-trien-7-one radical cation) was formed by the neutral loss of iodine radical (127 Da) and water loss (18 Da) and concerted intramolecular ring closure.
Metabolite M445.
The CID product ion FT mass spectrum in positive mode of M445 gave prominent fragment ions at m/z 428, 400, 359, and 214, as well as the pseudomolecular ion at m/z 446. The fragment ion m/z 428 was formed by neutral loss of 18 Da (water) from the acid group of the pseudomolecular ion. Further loss of 28 Da (carbon monoxide) formed fragment m/z 400. Further loss of 2-amino-ethenol (C2H5NO) together with the addition of water formed fragment m/z 359, which was already known from the parent compound. Further on, the key fragment m/z 214 (8-(2-fluoro-phenyl)-3,8-diaza-bicyclo[4.2.0]octa-1,3,5-trien-7-one radical cation) was formed by the neutral loss of iodine radical (127 Da) and water loss (18 Da) and concerted intramolecular ring closure.
Metabolite M447-1.
The CID product ion FT mass spectrum in positive mode of M447-1 gave prominent fragment ions at m/z 430, 375, 357, 303, 248, 230 and 202, as well as the pseudomolecular ion at m/z 448. The fragment ion m/z 430 was formed by neutral loss of 18 Da (water) from one of the hydroxyl groups of the pseudomolecular ion. Further loss of 3-amino-propenol (C3H7NO) together with the addition of water formed fragment m/z 375. From fragment m/z 430, concurrent neutral loss of an iodine radical (127 Da) led to fragment m/z 303. Furthermore, the key fragment m/z 214 from the parent compound was here 16 Da greater in mass, indicating hydroxylation or oxidation in the core region. This fragment, m/z 230 (hydroxylated or oxidated 8-(2-fluoro-phenyl)-3,8-diazabicyclo[4.2.0]octa-1,3,5-trien-7-one radical cation), was formed by the neutral loss of iodine radical (127 Da) and water loss (18 Da) and concerted intramolecular ring closure. Further loss of 28 Da (carbon monoxide) thereof formed fragment m/z 202, proposed to be hydroxylated or oxidated 7-(2-fluoro-phenyl)-3,7-diaza-bicyclo[4.1.0]hepta-1,3,5-triene radical cation (also 16 Da greater in mass than the fragment observed at the parent compound). However, the exact position of the hydroxylation or N-oxidation cannot be assessed on the basis of the available mass spectrometric data.
Metabolites M607-1 and-2.
Metabolite M607-1 has a pseudomolecular peak of 176 Da greater mass than the parent drug, indicating a glucuronide conjugate of pimasertib. Metabolite M607-1 has a predominant fragment ion at m/z 432 formed by neutral loss of 176 Da (glucuronic acid) of the pseudomolecular ion. The other fragment ions, m/z 414, m/z 359, and m/z 214, are already known from the parent compound. Fragment m/z 414 is formed by loss of 18 Da (water) from the propanediol moiety. Further loss of 3-amino-propenol (C3H7NO) together with the addition of water forms fragment m/z 359. Furthermore, the key fragment m/z 214 (8-(2-fluoro-phenyl)-3,8-diaza-bicyclo[4.2.0]octa-1,3,5-trien-7-one radical cation) is formed by the neutral loss of iodine radical (127 Da) and water loss (18 Da) and concerted intramolecular ring closure.
Metabolite M623-1 and -2.
Metabolite M623-1 has a pseudomolecular peak that was 194 Da greater in mass than the parent drug pimasertib, indicating oxidation combined with glucuronide conjugation. Metabolite M623-1 has a predominant fragment ion at m/z 448 formed by neutral loss of 176 Da (glucuronic acid) of the pseudomolecular ion. The other fragment ion, m/z 430, was already known from metabolite M447-1 and formed by loss of 18 Da (water) from the propanediol moiety.
Metabolite M305.
The CID product ion FT mass spectrum in positive mode of M305 gave prominent fragment ions at m/z 288, 233, and 215, as well as the pseudomolecular ion at m/z 306. The fragment ion m/z 288 was formed by neutral loss of 18 Da (water) from one of the hydroxyl groups of the pseudomolecular ion. Further loss of 3-amino-propenol (C3H7NO) together with the addition of water formed fragment m/z 233. As there was no iodine in the molecule, the key fragment was 1 Da greater in mass: m/z 215 (8-(2-fluoro-phenyl)-3,8-diaza-bicyclo[4.2.0]octa-1,3,5-trien-7-one cation) and formed by water loss (18 Da) and concerted intramolecular ring closure.
Metabolite M357.
The CID product ion FT mass spectrum in positive mode of M357 gave a prominent fragment ion at m/z 214, as well as the pseudomolecular ion at m/z 358. The fragment ion m/z 214 was formed by neutral loss of 17 Da (ammonia) from the amide group with intramolecular ring closure of the pseudomolecular ion and loss of iodine radical (127 Da). This key fragment m/z 214 is proposed to be the 8-(2-fluoro-phenyl)-3,8-diaza-bicyclo[4.2.0]octa- 1,3,5-trien-7-one radical cation and is already known from the parent compound pimasertib.
Metabolite M358.
The CID product ion FT mass spectrum in positive mode of M358 gave prominent fragment ions at m/z 315 and 214, as well as the pseudomolecular ion at m/z 359. The fragment ion m/z 315 was formed by neutral loss of 44 Da (carbon dioxide) from the acid group of the pseudomolecular ion. Furthermore, the key fragment m/z 214 (8-(2-fluoro-phenyl)-3,8-diaza-bicyclo[4.2.0]octa-1,3,5-trien-7-one radical cation) was formed by the neutral loss of iodine radical (127 Da) and water loss (18 Da) and concerted intramolecular ring closure.
Metabolite M373.
The CID product ion FT mass spectrum in positive mode of M373 gave the prominent fragment ion at m/z 230, as well as the pseudomolecular ion at m/z 374. The fragment ion m/z 230 was formed by neutral loss of 17 Da (NH3) together with the loss of an iodine radical (127 Da) from the pseudomolecular ion.
Metabolite M415.
The CID product ion FT mass spectrum in positive mode of M415 gave the prominent fragment ions at m/z 370 and 214, as well as the pseudomolecular ion at m/z 416. The fragment ion m/z 370 was formed by neutral loss of 46 Da (CO2) from the pseudomolecular ion. Furthermore, the key fragment m/z 214 (8-(2-fluoro-phenyl)-3,8-diaza-bicyclo[4.2.0]octa-1,3,5-trien-7-one radical cation) was formed by the neutral loss of iodine radical (127 Da) and formal loss of 75 Da (glycine) and concerted intramolecular ring closure from the pseudomolecular ion.
Structure Elucidation of the Phosphoethanolamine Conjugate of Pimasertib (M554).
To identify the chemical structure of the novel metabolite M554, sufficient quantities of this metabolite were required for it to be purified for NMR analysis. Initial investigations to generate M554 with microsomes and S9 mix incubations did not produce any M554, and incubations with hepatocytes yielded insufficient quantities (about 10%). Furthermore the yield and purity of the M554 fraction collected from rat urine after pimasertib dosing was too low for meaningful NMR spectra. Therefore, HepaRG cells were used. Incubation of [14C]-pimasertib in HepaRG cells produced the target metabolite M554 in sufficient quantities (about 30% of total drug-related material) for its purification and analysis, in addition to the other known pimasertib metabolites (Fig. 3), as identified in the other matrices. After isolation and purification, M554 was analyzed by high resolution (HR)-MS and HR-MS/MS on the LTQ-Orbitrap MS. The full FT scan of M554 identified a [M+H]+ ion peak at m/z 555.02917 (Fig. 4A). Further fragmentation of this ion peak by CID (MS2 product) produced prominent fragment ions at 537.01971, 511.98798, 493.97745, 428.12579, and 414.01114 detected by HR-MS/MS. A minor fragment at m/z 432.02173 showed the intact scaffold of pimasertib after the loss of phosphoethanolamine from M554 (Fig. 4B). Further fragmentation experiments (MS3 product) provided additional fragments until in total 19 separate predominant fragment ions were identified for M554, which confirmed that this peak is pimasertib-related.
A definite structure for M554 could not be obtained through HR-MS or HR-MS/MS analyses alone; therefore, NMR was applied. The intact scaffold of pimasertib was confirmed from standard two-dimensional NMR spectra, i.e., COSY, HMBC, and ROESY. Additionally, an aminoethylene moiety and one phosphorous atom with a shift characteristic for a phosphoric acid derivative were found. To link the pimasertib scaffold with the additional fragments, a 1H-31P-HMBC was recorded. This revealed correlations of H15 and H16 to the phosphorous atom, thereby unambiguously linking the pimasertib scaffold with the aminoethylene moiety by a phosphoric acid linker. The structure of M554 is shown in Fig. 5 with all relevant NMR assignments. Associated one-dimensional and two-dimensional NMR spectra and the detailed NMR assignments of M554 are shown in the Supplemental Fig. 2. Collectively, HR-MS and NMR analyses confirmed the unusual structure of M554 to be a phosphoethanolamine conjugate of pimasertib.
The proposed phosphoethanolamine structure was synthesized (reference standard, MSC2521347), and the structure of M554 was definitively confirmed following cochromatography of this reference standard with M554 in the human mass balance study (see Supplemental Fig. 3).
Discussion
We have reported the identification and characteristics of metabolites of the MEK1/2 inhibitor pimasertib following oral administration in patients with advanced solid tumors. This metabolite structural analysis is part of a phase I study of novel design that assessed the mass balance, metabolite profile, and oral bioavailability (including absolute bioavailability) of pimasertib in patients with locally advanced or metastatic solid tumors; the full PK outcomes from this study have been published separately (von Richter et al., 2016) and complement the current report describing the metabolic profile and structure of the pimasertib metabolites identified in the same patients.
For the metabolite profiling and structural investigations described, all methods were controlled by assessing important method validation parameters (extraction recovery, take-up efficiency, and column recovery), which were within an acceptable range. The positioning of 14C radiolabel on pimasertib has been shown to be optimal in previous nonclinical studies (Merck KGaA, Darmstadt, Germany, data on file). Following oral administration, [14C]-pimasertib was extensively metabolized in participants with solid tumors; half of the administered dose was excreted in urine and nearly one-third in feces. In total, fourteen different phase I and phase II metabolites were identified in plasma, urine, and feces. In both plasma and urine, two major metabolites of pimasertib were identified, the phosphoethanolamine conjugate M554 and the carboxylic acid M445; in contrast, M445 was the main metabolite found in feces, with only nominal amounts of M554 identified. All other metabolites identified were minor. Unchanged pimasertib was excreted in small amounts (<5% of administered dose) in urine and feces, indicating near complete absorption and extensive metabolism following oral administration. The metabolites formed were principally oxidized species and conjugates of pimasertib, with some minor metabolites formed through isomerism, dealkylation, and deiodination routes. The propranediol moiety was a key target for metabolism, including generation of the major metabolite M445, and pimasertib underwent both direct and indirect conjugation with glucuronic acid. Screening of acidified plasma samples indicated that potentially reactive metabolites such as acyl glucuronides of the acid metabolites were not generated. Enantioselective analysis performed for pimasertib and its enantiomer, MSC1940796, and M445 and its enantiomer, MSC2486240, showed that enantiomeric conversion was low for both compounds and, therefore, of minor importance in the metabolism of pimasertib. This observation was supported by investigations of pimasertib and its enantiomer performed separately using plasma samples taken on day 1 of the current human mass balance study, which confirmed the minor importance of the enantiomeric conversion (Merck KGaA, data on file). However, a full complement of enantioselective analyses could not be performed for pimasertib and its enantiomer in plasma owing to limited plasma volume, nor for M554, as the reference compound, MSC2521347, was of unknown enantiopurity.
In contrast to the profile of pimasertib metabolism, the MEK1/2 inhibitor selumetinib is excreted primarily via feces and principally as glucuronide conjugates, but it also formed an active metabolite, N-desmethyl-selumetinib (Leijen et al., 2011). Similarly, the MEK inhibitor trametinib is also primarily eliminated in the feces (>80%), but unlike either pimasertib or selumetinib, trametinib is only partially metabolized, with the majority of drug circulating in plasma as the parent compound; the portion that is metabolized undergoes deacetylation and mono-oxidation but also some glucuronide conjugation (Novartis Europharm Limited, 2015). These differences in the metabolite profile of some MEK1/2 inhibitors extend to the enzymes implicated in their biotransformation. The enzymes mediating the biotransformation of pimasertib are under investigation, but in vitro investigations have suggested that CYP2C19 and CYP3A4 are involved in the formation of M445 (Merck KGaA, data on file), which is consistent with the cytochrome P450-mediated metabolism (CYP 1A2, 2C19, and 3A4) of selumetinib (Leijen et al., 2011). In contrast, trametinib metabolism is mediated through hydrolytic enzymes rather than cytochrome P450 enzymes (Novartis Europharm Limited, 2015).
Although the exact positioning of some of the oxidation and glucuronidation reactions on the pimasertib metabolites could not be clearly determined with the analytical methods used, the data generated from these structural-elucidation studies provided sufficient evidence to logically deduce the probable generation of all the metabolites identified for pimasertib, as shown in the proposed metabolic pathway. However, we acknowledge that some metabolites, such as M358, could have been generated through different routes, and further evidence is needed to fully understand the potential relationship between these metabolites.
One unusual and major metabolite, M554, was identified as a phosphoethanolamine conjugate of pimasertib. Additional investigations to determine the chemical structure of this metabolite required sufficient generation of the metabolite in vitro. Neither rat urine nor cryopreserved hepatocytes yielded the quantity of M554 required, and M554 was not formed in liver microsomes of different species (human, mouse, rat, dog, monkey) (Merck KGaA, data on file). Therefore, HepaRG cells were used as the tool to generate sufficient quantities of the metabolite. Analysis by HR-MS and MS/MS alone could not confirm the structure of M554, as the integration of the phosphorus was not taken into consideration by the structure-elucidation evaluation process. However, NMR confirmed that M554 is a novel phosphoethanolamine conjugate on the propanediol moiety of pimasertib. Confirmation of this was obtained when the proposed metabolite was synthesized on the basis of the structure proposed (reference compound MSC2521347) and coeluted with M554. It is proposed that this novel metabolite is formed directly from the parent compound, pimasertib.
In the current clinical trial with pimasertib, all of the metabolites identified have been observed previously in in vitro hepatocyte and/or in vivo rat metabolism studies with [14C]-pimasertib (Merck KGaA, data on file). No new or human-specific metabolites were found in this study in patients with solid tumors; however, variations between the metabolite profiles in plasma for the six individuals was noted. Given that the metabolite profile in patients with solid tumors is consistent with the outcomes from in vitro/in vivo studies (Merck KGaA, data on file), it is unlikely that either of the two principal metabolites have any notable adverse effects in patients.
To our knowledge, phosphoethanolamine conjugation has not been described previously for a drug. In a recent publication, a phosphocholine metabolite has been described for a hepatitis C NS5B inhibitor (Zhuo et al., 2016). Furthermore, the formation of a phosphocholine ester has been described for the macrolide everolimus as well (Zollinger et al., 2008). This phosphocholine metabolite could be considered a downstream metabolite from the phosphoethanolamine conjugates, formed via phospholipid N-methyltransferases (Schneider and Vance, 1979). However, as these phosphocholine metabolites of pimasertib were not found in any of our analyzed samples, the direct link to the phosphocholine pathway is missing for pimasertib.
Our hypothesis is that the phosphoethanolamine pathway may be like the Kennedy pathway (Gibellini and Smith, 2010), that is, split into two analogous pathways. Pimasertib could enter this pathway owing to the structural similarity of the propanediol moiety of pimasertib to glycerol, making it closely related to lipid metabolism. However, investigations to determine the key enzyme(s) responsible for the biotransformation of pimasertib to the phosphoethanolamine conjugate, M554, are ongoing, and investigation of why only the phosphoethanolamine conjugate of pimasertib and not the phosphocholine metabolite is formed may be warranted.
The design of this study allowed all key PK parameters to be assessed using both an intravenous microtracer and an oral dose of radiolabeled pimasertib to determine absolute bioavailability and mass balance, as well as profiling and structural identification of metabolites in the relevant patient population.
Acknowledgments
The authors thank the patients, the patients’ families, and the study teams at the participating centers. The authors also thank Dr. Henk Poelman, Dr. Markus Fluck, Ms. Birgit Felden de Neumann, Ms. Martina Daut, Mr. Ingo Noerenberg, and Mr. Thomas Scheller for their contributions to the metabolism studies presented and Dr. Oliver von Richter for his contribution to the mass balance trial overall.
Authorship Contributions
Participated in research design: Scheible, Kraetzer, Johne, Wimmer.
Conducted experiments: Scheible, Kraetzer, Marx, Johne.
Contributed new reagents or analytic tools: Marx.
Performed data analysis: Scheible, Kraetzer, Marx, Johne.
Wrote or contributed to the writing of the manuscript: Scheible, Kraetzer, Marx, Johne, Wimmer.
Footnotes
- Received August 2, 2016.
- Accepted November 18, 2016.
This work was supported by Merck KGaA, Darmstadt, Germany. Medical writing support was provided by Helen Swainston at Bioscript Science, Macclesfield, UK, funded by Merck KGaA, Darmstadt, Germany. Primary laboratory of origin: Merck KGaA, Grafing, Germany.
↵This article has supplemental material available at dmd.aspetjournals.org.
Abbreviations
- AUC
- area under the curve
- CID
- collision-induced dissociation
- COSY
- correlation spectroscopy
- ESI
- electrospray ionization
- FT
- Fourier transform
- HMBC
- heteronuclear multiple-bond correlation spectroscopy
- HPLC
- high-performance liquid chromatography
- HR-MS
- high-resolution mass spectrometry
- ICH
- International Conference on Harmonisation
- LC
- liquid chromatography
- LOQ
- limit of quantification
- LSC
- liquid scintillation counting
- MAPK
- mitogen-activated protein kinase
- MS
- mass spectrometry
- MS/MS
- tandem mass spectrometry
- NMR
- nuclear magnetic resonance spectroscopy
- PK
- pharmacokinetic
- ROESY
- rotating frame nuclear Overhauser effect spectroscopy
- UPLC-MS
- ultra-performance liquid chromatography–mass spectrometry
- Copyright © 2017 by The American Society for Pharmacology and Experimental Therapeutics