Metabolites in safety testing assessment in early clinical development: A case study with a glucokinase activator

Article abstract of DOI:10.1124/dmd.114.060087

The present article summarizes Metabolites in Safety Testing (MIST) studies on a glucokinase activator, N,N-dimethyl-5-((2-methyl-6-((5-methylpyrazin-2-yl)carbamoyl)benzofuran-4-yl)oxy)pyrimidine-2-carboxamide (PF-04937319), which is under development for the treatment of type 2 diametes mellitus. Metabolic profiling in rat, dog, and human hepatocytes revealed that PF-04937319 is metabolized via oxidative (major) and hydrolytic pathways (minor). N-Demethylation to metabolite M1 [N-methyl-5-((2-methyl-6-((5-methylpyrazin-2-yl)carbamoyl)benzofuran-4-yl)oxy)pyrimidine-2-carboxamide] was the major metabolic fate of PF-04937319 in human (but not rat or dog) hepatocytes, and was catalyzed by CYP3A and CYP2C isoforms. Qualitative examination of circulating metabolites in humans at the 100- and 300-mg doses from a 14-day multiple dose study revealed unchanged parent drug and M1 as principal components. Because M1 accounted for 65% of the drug-related material at steady state, an authentic standard was synthesized and used for comparison of steady-state exposures in humans and the 3-month safety studies in rats and dogs at the no-observed-adverse-effect level. Although circulating levels of M1 were very low in beagle dogs and female rats, adequate coverage was obtained in terms of total maximal plasma concentration (～7.7x and 1.8x) and area under the plasma concentration-time curve (AUC; 3.6x and 0.8x AUC) relative to the 100- and 300-mg doses, respectively, in male rats. Examination of primary pharmacology revealed M1 was less potent as a glucokinase activator than the parent drug (compound PF-04937319: EC₅₀ = 0.17 μM; M1: EC₅₀ = 4.69 μM). Furthermore, M1 did not inhibit major human P450 enzymes (IC₅₀ > 30 μM), and was negative in the Salmonella Ames assay, with minimal offtarget pharmacology, based on CEREP broad ligand profiling. Insights gained from this analysis should lead to a more efficient and focused development plan for fulfilling MIST requirements with PF-04937319.

Full text of DOI:10.1124/dmd.114.060087

DMD #60087
Metabolites in Safety Testing (MIST) Assessment in Early Clinical
Development: A Case Study with a Glucokinase Activator
Raman Sharma, John Litchfield, Karen Atkinson, Heather Eng, Neeta B. Amin, William S.
Denney, John C. Pettersen, Theunis C. Goosen, Li Di, Esther Lee, Jeffrey A. Pfefferkorn,
Deepak K. Dalvie, and Amit S. Kalgutkar
Pfizer Inc., Eastern Point Road, Groton, CT (R.S., K. A., H.E., J.C.P., T. C. G., L. D.) San Diego,
CA (D.K.D.) and Cambridge MA (N.B.A., J.L, W.S.D., E.L., J.A.P., A.S.K.)
1

DMD #60087
Running Title: MIST analysis of a glucokinase activator
Address correspondence to: Amit S. Kalgutkar, Pharmacokinetics, Dynamics, and
Metabolism-New Chemical Entities, Pfizer Worldwide Research and Development, 610 Main
Street, Cambridge, MA 02139, USA. Tel: +(617)-551-3336. E-mail:
amit.kalgutkar@pfizer.com
Text Pages (including references): 37
Tables: 3
Figures: 8
References: 28
Abstract: 252
Introduction: 542
Discussion: 1493
2

DMD #60087
Abbreviations used are: T2DM, type 2 diabetes mellitus; GK, glucokinase; PF-04937319, N,N-
dimethyl-5-((2-methyl-6-((5-methylpyrazin-2-yl)carbamoyl)benzofuran-4-yl)oxy)pyrimidine-2-
carboxamide; CYP, cytochrome P450; M1, N-methyl-5-((2-methyl-6-((5-methylpyrazin-2-
yl)carbamoyl)benzofuran-4-yl)oxy)pyrimidine-2-carboxamide; MIST, Metabolites in Safety
Testing; reduced NADPH, β-nicotinamide adenine dinucleotide 2′-phosphate; HLM, human liver
microsomes; DMSO, dimethyl sulfoxide; ISEF, intersystem extrapolation factors; LC-MS/MS,
liquid chromatography tandem mass spectrometry; CID, collision-induced dissociation; m/z,
+
mass to charge ratio; MH , protonated molecular ion; AUC, area under the plasma
concentration-time curve; Cmax, maximal plasma concentration; Tmax, time for occurrence of
C
max; f
u
, unbound fraction; t
R
, retention time; NOAEL no-observed-adverse-effect level; t1/2, half-
life; DDI, drug-drug interaction.
3

DMD #60087
Abstract
The present article summarizes MIST studies on a glucokinase activator PF-04937319, which
is under development for the treatment of T2DM. Metabolic profiling in rat, dog and human
hepatocytes revealed that PF-04937319 is metabolized via oxidative (major) and hydrolytic
pathways (minor). N-Demethylation to metabolite M1 was the major metabolic fate of PF-
0
4937319 in human (but not rat or dog) hepatocytes, and was catalyzed by CYP3A and CYP2C
isoforms. Qualitative examination of circulating metabolites in humans at the 100 mg and 300
mg doses from a 14-day multiple dose study revealed unchanged parent drug and M1 as
principal components. Because M1 accounted for 65% of the drug related material at steady
state, an authentic standard was synthesized and used for comparison of steady state exposures in
human and the 3–month safety studies in rats and dogs at the NOAEL. Although circulating
levels of M1 were very low in beagle dogs and female rats, adequate coverage was obtained in
terms of total Cmax (~7.7x and 1.8x) and AUC (3.6x and 0.8x AUC) relative to the 100 mg and
3
00 mg, doses, respectively in male rats. Examination of primary pharmacology revealed M1
was less potent as a glucokinase activator than the parent drug (compound PF-
4937319:EC50=0.17 M; M1:EC50=4.69 M). Furthermore, M1 did not inhibit major human
CYP enzymes (IC >30 M), was negative in the Salmonella Ames assay, with minimal off-
0
5
0
target pharmacology, based on CEREP broad ligand profiling. Insights gained from this analysis
should lead to a more efficient and focused development plan for fulfilling MIST requirements
with PF-04937319.
4

DMD #60087
Introduction
Type 2 diabetes mellitus (T2DM), characterized by elevated levels of fasting blood glucose, is a
complex disease mainly caused by defects in hepatic glucose metabolism and the failure of
pancreatic -cells to secrete enough insulin to overcome insulin resistance. Glucokinase (GK)
activators offer a novel and promising opportunity for the treatment of T2DM patients
(
Matschinsky et al., 2011; Sarabu et al., 2011; Pfefferkorn, 2013). GK is a member of
hexokinase family of enzymes that are responsible for the phosphorylation of glucose to glucose-
-phosphate (Mithieux, 1996). GK plays a key role in glucose homeostasis by promoting
6
glycogen synthesis in the liver and glucose-sensitive insulin release in -cells (Matschinsky and
Ellerman, 1968; Matschinsky et al., 1998). Activators of GK increase the enzyme’s affinity for
glucose (K ) and also its maximal catalytic rate (Vmax). Consequently, they stimulate insulin
m
biosynthesis and secretion, enhance hepatic glucose uptake, and augment glucose metabolism
and related processes in GK-expressing cells. Early clinical studies of GK activators have
demonstrated improvements in glycemic control, in both healthy volunteers and T2DM patients
(Haynes et al., 2010; Coghlan and Leighton, 2008; Meininger et al., 2011; Sarabu et al., 2012);
however, an elevated risk of hypoglycemia arising from GK activation in the pancreas leading to
inappropriately excessive insulin secretion was also observed. Medicinal chemistry strategies to
overcome the hurdles imposed by the hypoglycemia risk have been recently reviewed
(Pfefferkorn, 2013). One such tactic involves the design of “partial activators” of GK that avoid
reducing the K for glucose to inappropriately low levels, thereby retaining increased
m
dependence of enzymatic activity on physiological glucose concentrations.
5

DMD #60087
N,N-Dimethyl-5-((2-methyl-6-((5-methylpyrazin-2-yl)carbamoyl)benzofuran-4-
yl)oxy)pyrimidine-2-carboxamide (PF-04937319, Figure 1) is a partial activator of GK, which
has advanced to Phase 2 clinical trials for the treatment of T2DM (Pfefferkorn et al., 2011). The
preclinical pharmacokinetics of PF-04937319 in animals were characterized by low to moderate
plasma clearance and elimination half-lives, respectively (Pfefferkorn et al., 2011). The
predicted hepatic clearance using liver microsomes and hepatocytes from rat and dog were found
to be in good agreement with the in vivo blood clearance values in these species implying that
the clearance mechanism of PF-04937319 involves metabolism. Renal and/or biliary excretion
in rats following intravenous administration of PF-04937319 was minimal (< 3% of the
administered dose) (Pfefferkorn et al., 2011), which further supports the hypothesis. First-in-
human pharmacokinetics and tolerability studies in T2DM patients have been recently conducted
with PF-04937319, which provided us with an opportunity to examine circulating human
metabolites. Metabolite scouting studies using steady state plasma samples from the 100 mg and
3
00 mg oral doses of PF-04937319 led to the characterization of an N-demethylated metabolite
(
N-methyl-5-((2-methyl-6-((5-methylpyrazin-2-yl)carbamoyl)benzofuran-4-yl)oxy)pyrimidine-2-
carboxamide, M1). Because the exposure of M1 was significantly greater than 10% of the total
drug-related material based on UV response, it was proposed to be a major human metabolite of
PF-04937319. On the basis of both the U.S. Food and Drug Administration Metabolites in
Safety Testing (MIST) guidance and the later published 2009 ICH guidance M3 (R2) (U.S. Food
and Drug Administration, 2008; European Medicines Agency, 2009), the coverage of M1 in
toxicology species needed to be assessed. The current article presents a summary of the human
6

DMD #60087
metabolite scouting efforts and efforts towards the examination of systemic exposure coverage in
animals and characterization of the on- and off-target pharmacology profile of M1.
Materials and Methods
General Chemicals. PF-04937319 and M1 (chemical purity > 99% by HPLC and NMR) were
synthesized at Pfizer Worldwide Research and Development (Groton, CT). The synthesis of PF-
0
4937319 has been previously published (Pfefferkorn et al., 2011). A synthetic scheme for the
preparation of M1 is depicted in the supplementary materials and methods section
supplementary figure S1). Monobasic and dibasic potassium phosphate buffer, magnesium
(
chloride, and reduced β-nicotinamide adenine dinucleotide 2′-phosphate (NADPH) were
purchased from Sigma-Aldrich (St. Louis, MO). Recombinant CYPs (rCYPs) 1A2, 2C19, 2C9,
2
2
D6, 2E1, 3A4, and 3A5 were purchased from Panvera (Carlsbad, CA), whereas, rCYP2B6,
C8, and pooled male and female human liver microsomes (HLM, n = 50 donors) were
purchased from BD Gentest (Woburn, MA). Cryopreserved rat, dog, and human hepatocytes
were obtained from Celsis (Chicago, IL).
In Vitro Metabolism in Cryopreserved Hepatocytes. Stock solutions of PF-04937319 were
prepared in 10% dimethyl sulfoxide (DMSO) and 90% acetonitrile. The final concentration of
DMSO and acetonitrile in the incubation mixtures were 0.1% and 0.9%, respectively. Williams
E media was prepared by adding 26 mM sodium carbonate and 50 mM HEPES, followed by 0.2
2
µm filtration then 30 min of CO bubbling at 37 °C. This media was used for thawing and
suspension of hepatocytes. Rat, dog, and human hepatocytes were suspended at 0.5 million
viable cells per ml Williams E medium in 24-well polystyrene plates and prewarmed at 37 °C for
7

DMD #60087
3
0 min. Incubations were initiated with the addition of PF-04937319 (final concentration in
incubation = 10 M) and were conducted at 37 °C for 240 min, 75% relative humidity, and 5%
CO . The total incubation volume was 1ꢀml per well. Incubations were terminated by the
2
addition of ice-cold acetonitrile (4 ml) and centrifuged (3000 x g, 15 min). The supernatants
were dried under a steady stream of nitrogen, reconstituted with 25% aqueous acetonitrile (250
l), and analyzed via LC-MS/MS for metabolite formation.
M1 formation in rCYPs and HLM. Initial incubations were conducted in individual rCYPs or
HLM to determine the time and protein concentration associated with the linear range of M1
formation. Linearity was assessed at 0.1–1.5 mg/ml protein from 10–40 min with 60 µM of PF-
0
4937319. rCYPs and liver microsomes were diluted to the desired concentrations in 100 mM
potassium phosphate buffer (pH 7.4) and NADPH, and MgCl were added to final concentrations
2
of 1.3 and 1.0 mM, respectively. As such, PF-04937319 is very slowly metabolized in HLM,
which is reflected in its minimal metabolic turnover (t1/2 > 120 min) in NADPH-supplemented
HLM (Pfefferkorn et al., 2011). The 60 M concentration was simply chosen to allow facile
detection and quantitation of the M1 metabolite in these preliminary studies. For control
incubations, a stock of HLM was placed in a 90 °C water bath for 15 min to heat-inactivate
enzymes. Aliquots of diluted matrices were placed in tubes in a 37 °C dry heat bath and pre-
warmed for 5 min. To initiate the reaction, PF-04937319, prepared at 100-times the final
concentration (6 mM) in 10% DMSO and 90% acetonitrile, was added to the incubations (final
volume = 0.4 ml). Incubations were conducted in duplicate. At each time point, 100 µl of
incubation mixture was removed and transferred to a plate containing 200 µl of acetonitrile with
2
00 ng/ml terfenadine (molecular weight = 472) as internal standard. Standards of M1 were
8

DMD #60087
prepared in acetonitrile and combined with 100 µl heat-inactivated HLM in potassium phosphate
buffer and 200 µl acetonitrile with internal standard. 100 µl of blank acetonitrile was added to
each sample to match the dilution of the standards. Plates were vortexed and centrifuged (10
min, 2000 x g). 150 µl of supernatant was removed and combined with 150 µl water containing
0
.2% formic acid. Samples were analyzed by LC-MS/MS to determine M1 concentration in
each sample. M1 formation rates were calculated by dividing M1 concentration by reaction time
and enzyme concentration.
Enzyme kinetics studies were conducted with the enzymes that produced appreciable
amounts of M1: HLM, rCYP1A2, rCYP2C9, rCYP2C19,rCYP2D6,rCYP3A4, and rCYP3A5.
From the linearity data, a protein concentration and time were selected that were within the linear
range of M1 formation: HLM 0.5 mg/ml, 30 min; rCYP1A2 0.5 mg/ml (53 pmol/ml), 30 min;
rCYP2C9 0.25 mg/ml (61 pmol/ml), 20 min; rCYP2C19 0.25 mg/ml (28 pmol/ml), 20 min;
rCYP2D6 0.5 mg/ml (59 pmol/ml), 30 min; rCYP3A4 0.25 mg/ml (22 pmol/ml), 20 min; and
rCYP3A5 0.5 mg/ml (39 pmol/ml), 30 min. Incubations were carried out as described above, in
triplicate, containing PF-04937319 from 1–300 µM. Rate of product formation was plotted
versus substrate concentration. If product formation was saturated, the GraphPad PRISM
(
LaJolla, CA) Michaelis-Menten model was used to calculate constants K
m
and Vmax; Vmax was
divided by K to calculate intrinsic clearance (CLint). Otherwise, the slope of the initial reaction
m
rate was calculated using Microsoft Excel (Redmond, WA) which is equivalent to CLint. To
determine contribution by each enzyme, CLint was corrected for pmol CYP per mg protein,
fraction unbound in incubation (HLM fu,inc used as a surrogate for fu,inc in rCYPs) and CLint
-
derived intersystem extrapolation factors (ISEF) as previously described in the literature
9

DMD #60087
(
Stringer et al., 2009; Chen et al., 2011). rCYP CLint values were summed and the fraction of
total CLint attributed to each rCYP was converted to % contribution.
Plasma Protein Binding. The binding of PF-04937319 and M1 to rat (5 males/5 females), dog
(
5 males/5 females), and human (3 males and 3 females) plasma proteins was determined using
disposable Rapid Equilibrium Dialysis device with pre-packed inserts (ThermoScientific, West
Palm Beach, FL). Stock solutions of PF-04937319 were prepared in water/acetonitrile (50/50,
v/v), whereas those of M1 were prepared in DMSO. Plasma samples were prepared by mixing
PF-04937319 or M1 to produce final incubation concentrations of 0.55 and 5.0 M for PF-
0
4937319 and 2 M for M1. Aliquots (220 l; n = 4) of spiked plasma were added to the red
side of the equilibrium dialysis device inserts, followed by the addition of buffer (350 l; n = 4)
to the white side. The device was covered with a gas permeable membrane and agitated on an
2
orbital shaker at approximately 450 rpm within a humidified 75% RH 5% CO incubator
operating at 37 °C for 4 h. Control plasma samples (t = 0; 15 l; n = 2) were prepared from the
remaining spiked plasma solution. Each sample was matrix matched with blank buffer (45 l).
The samples were precipitated with an acetonitrile solution (120 l; containing 5% DMSO and
2
00 ng/ml of a proprietary internal standard (molecular weight = 687). Plasma (15 l) and buffer
(
45 l) were sampled from each insert. The samples were matrix matched with appropriate
blank matrix (plasma samples with 45 l of buffer and buffer samples with 15 l of plasma). All
samples for analysis were prepared in a 96-well plate, centrifuged (3000 x g, 10 min, 25 °C), and
the supernatants were analyzed by LC-MS/MS. Nonspecific binding of PF-04937319 and M1 to
the equilibrium dialysis device was minimal based on control equilibrium dialysis experiments
1
0

DMD #60087
replacing plasma with buffer. The percentage of PF-04937319 or M1 bound to plasma proteins
was calculated as 100 – [(concentration of analyte in buffer/concentration of analyte in plasma) x
1
00].
Metabolite Scouting in Humans. The study was conducted in compliance with the
International Conference on Harmonization Good Clinical Practices guidelines, the ethical
principles that have their origin in the Declaration of Helsinki, and the U.S. Food and Drug
Administration regulations for informed consent and protection of subject rights. The clinical
study design with PF-04937319 incorporated randomized, placebo- controlled multiple oral
doses of 10, 30, 50, 100, and 300 mg QD for 14 days in patients with T2DM (NCT01272804).
The study enrolled 12 patients (9 on actives, 3 on placebos) in 5 separate cohorts (total of
approximately 60 patients). After an overnight fast, patients received PF-04937319 (10 and/or
1
00 mg strength tablet) and venous blood samples from each subject were collected predose (10
ml) and at 0.5, 1.5, 3.0, 5.0, 8.0, 12, and 16 h postdose (3 ml/time point) on Days 1 and 14,
respectively, for examining steady state pharmacokinetics of PF-04937319. Blood samples were
also collected from each subject at 2.0, 6.0, and 12 h postdose (2 ml/time point) on Days 1 and
1
4 for metabolite scouting efforts. Drug administration on Days 1 and 14 was following an
overnight fast of at least 8 h, whereas PF-04937319 was administered with a morning meal for
the remaining 12 days (Days 2–13). Blood samples were collected in heparinized tubes. Within
3
0 min after collection, the blood samples were centrifuged at approximately 1700 x g for ~ 10
min at 4 °C to generate plasma. All samples were stored at or below – 20 °C until analysis.
Plasma for metabolite scouting was obtained from subjects in the cohort dosed with 100 mg and
3
00 mg of PF-04937319. Day 1 and Day 14 plasma samples were pooled according to the
1
1

DMD #60087
method of Hamilton et al. (1981). An equal volume (0.5 ml) from each individual subject pool
(
pre- and post-dose) was combined to make a multi-subject pool for both Day 1 and Day 14
samples. Each pool was diluted with 12 ml of ice-cold acetonitrile with 0.1% formic acid and
vortexed for 2–4 min. The samples were subsequently sonicated for 5 min, centrifuged (2000 x
g for 10 min), and the supernatants were transferred to clean tubes and evaporated to dryness in
vacuo under a steady nitrogen stream. The dried samples were then reconstituted in 300 l of
mobile phase (10% acetonitrile, 0.1% formic acid in water) and an aliquot (50 l) was injected
on LC-MS/MS and examined for circulating metabolites.
M1 Coverage in Preclinical Species. All studies were conducted in accordance with the United
States Food and Drug Administration Good Laboratory Practice Regulations, Title 21 of the
United States Code of Federal Regulations Part 58. All procedures in the protocol were in
compliance with applicable animal welfare acts and were approved by the local Institutional
Animal Care and Use Committee. PF-04937319 was formulated in 1.25% (w/v) hydropropyl
cellulose-SL and 0.05% (w/v) docusate sodium sulfate in reverse osmosis water.
Rats. Male (0.3–0.4 kg) and female (0.2–0.3 kg) Sprague Dawley rats (4/sex/dose) were
administered PF-04937319 by oral gavage once daily at doses of 50, 250, and 1000 mg/kg
(
males) and 10, 50, and 1000 mg/kg (females) for 3 months (at least 91 consecutive days).
Blood samples were collected on Days 1 and 90 at 0.5, 2.0, 7.0, and 24 h postdose. Blood
approximately 0.5 ml) was collected from the jugular vein into tubes containing K2EDTA and
centrifuged within 1 h of collection. Plasma was harvested and stored at -10 to -30 °C.
(
1
2

DMD #60087
Dog. Male and female purebred Beagles (11–12 months old, 8.7–11.7 kg males, 6.0–8.7 kg,
females) were administered PF-04937319 (n=3/sex/dose) by oral gavage at doses of 5.0, 50 or
2
50/125/100 mg/kg/day for 3 months. During the study, a reduction in dose from 250 mg/kg/day
to 125 mg/kg/day occurred after 1 week of dosing, due to hypoglycemia observed in some dogs.
On day 17, the glucose level was still critical, and this led to a further lowering of the dose from
125 mg/kg/day to 100 mg/kg/day preceded by 4 days of dosing holiday. Serial blood samples
were collected from all animals at the following time points: 0.5, 2, 7, 12, and 24 hours after
dosing on study Days 1 and 91 for toxicokinetic measurements of PF-04937319 and M1. Blood
(
approximately 1.0 ml) was collected into tubes containing K2EDTA and centrifuged within 1 h
of collection. Plasma was harvested and stored at -10 to -30 °C.
LC-MS/MS Analysis for Quantitation of PF-04937319 and M1. Pharmacokinetics and
toxicokinetics assessment for PF-04937319 and M1 from clinical and 3-month preclinical (rat
and dog) toxicology studies was conducted using a validated LC-MS/MS method (Pfizer data on
file). For in vitro studies, samples were analyzed directly using a AB Sciex (Framingham, MA)
API4000 QTrap triple quadropole LC-MS/MS with Turbo IonSpray Source, CTC Leap
autosampler (Leap Technology, Carrboro, NC), and Shimadzu LC-20AD HPLC system
(
(
Columbia, MD). Samples (10 l) were injected onto a Halo 2.7 µm C18 3.0 x 30 mm column
Advanced Materials Technology, Wilmington, DE) and analytes were eluted using a binary
gradient at a flow rate of 1.2 ml/min. Mobile phase A consisted of 0.1% formic acid in water
and mobile phase B 0.1% formic acid in acetonitrile. Initial conditions (10% B) were held for
0
9
.3 min and then increased to 45% B over 0.3 min, then to 75% B over 1.4 min, increased to
0% B over 0.1 min, held for 0.2 min, and returned to initial conditions for column re-
1
3

DMD #60087
equilibration. Analytes were detected using multiple reaction monitoring mode for the mass-to-
charge (m/z) transitions 433.2 → 324.4 (PF-04937319), 419.2 → 310.1 (M1), and 472.5 → 436.4
(terfenadine, internal standard). M1 standards were fit by least-squares regression of their areas
to a weighted linear equation, from which the unknown concentrations were calculated. The
dynamic range was 1-1000 nM.
LC-MS/MS Conditions for Metabolite Identification. Qualitative assessment of the
metabolism of PF-04937319 in hepatocytes was conducted using a Thermo Finnegan Surveyor
photodiode array plus detector, Thermo Acela pump and a Thermo Acela Autosampler
(
Thermoscientific, West Palm Beach, FL). Chromatography was performed on a Phenomenex
Hydro RP C18 (4.6 mm x 150 mm, 3.5 m) column (Phenomenex, Torrance, CA). The mobile
phase composed of 5 mM ammonium formate buffer with 0.1% formic acid (pH=3) (solvent A)
and acetonitrile (solvent B) at a flow rate of 1 ml/min. The binary gradient was as follows:
solvent A to solvent B ratio was held at 95:5 (v/v) for 3 min and then adjusted to 55:45 (v/v)
from 0 to 35 min, 30:70 (v/v) from 35 to 45 min, and 5:95 (v/v) from 45 to 52 min where it was
held for 3 min and then returned to 95:5 (v/v) for 6 min before next analytical run. Identification
of the metabolites was performed on a Thermo Orbitrap mass spectrometer operating in positive
ion electrospray mode. The spray potential was 4 V and heated capillary was at 275 °C.
Xcalibur software version 2.0 was used to control the HPLC-MS system. Full scan
measurements from 50–1000 amu were collected at 15,000 resolution, with two data-dependent
acquisition scans of the most intense ion in scan event 1 and 2, respectively. Product ion spectra
were acquired at a normalized collision energy of 65 eV with an isolation width of 2 amu. The
dynamic exclusion function was used with a 1 min exclusion duration after 3 successive product
1
4

DMD #60087
ion scans with an early exclusion if the precursor ion falls below a signal to noise ratio of 20.
Metabolites from hepatocyte incubations were identified in the full-scan mode (from m/z 100 to
8
50) by comparing t = 0 samples with t = 240 min samples or through comparison with synthetic
standard(s), and structural information was generated from the collision-induced dissociation
+
(
CID) spectra of the protonated molecular ions (MH ).
Pharmacokinetic/Toxicokinetic Parameters. Pharmacokinetic parameters were determined
using noncompartmental analysis. Maximum plasma concentrations (Cmax) of PF-04937319 or
M1 (following administration of PF-04937319) in plasma were estimated directly from the
experimental data, with Tmax defined as the time of first occurrence of Cmax. The area under the
plasma concentration–time curve (AUC) from t=0 to 24 or 48 h was estimated using the linear
trapezoidal rule.
Results
In Vitro Metabolic Profile. Figure 2 depicts HPLC-UV chromatograms of cryopreserved rat
(
panel A), dog (panel B), and human (panel C) hepatocyte incubations conducted with PF-
4937319 (10 M) at 37 °C for 240 min. A total of 16 metabolites were detected in hepatocytes
0
from preclinical species and human. There were no human-unique metabolites of PF-04937319;
metabolites derived from human hepatocytes were observed in corresponding matrices from rat
+
and/or dog. Table 1 indicates the retention time (t
R
) and molecular weight (MH ) for PF-
0
4937319 and its corresponding metabolites obtained in hepatocyte incubations. Structural
assignments for key diagnostic fragment ions in the CID spectra of PF-04937319 and its
metabolites are also displayed in Table 1.
1
5

DMD #60087
Metabolite Identification. PF-04937319 (t
R
= 22.3 min) displayed an exact mass of 433.1619
+
2
(
MH ). Product ion scans MS (Supplemental Figure S2, panel A) of PF-04937319 yielded
fragment ions with m/z 324.0979 and 342.1084. The fragment ion at m/z 324.0979 represented
amide bond cleavage and loss of the 5-methylpyrazin-2-amine functionality. The fragment ion at
m/z 342.1084 was consistent with a gas phase addition of water to the m/z 324.1084 fragment.
3
MS product ion scans (Supplemental Figure S2, panel B) of the m/z = 324.0979 yielded
additional fragment ions at m/z 296.1029, 281.0555, 253.0607, 225.0657. The origins of these
ions are depicted in Table 1.
+
Metabolite M1 (t
R
= 20.4 min) displayed an exact mass of 419.1462 (MH ), which was 14 Da
lower than the molecular weight of PF-04937319. M1 was observed in hepatocytes from rat,
dog, and human. The CID spectrum of M1 (Supplemental Figure S3) contained a diagnostic
fragment ion at m/z 310.0822, which was 14 Da lower than the fragment ion at m/z 324.0979 in
the CID spectrum of PF-04937319. M1 was identified as the N-demethylated metabolite of PF-
0
4937319 (i.e., N-methyl-5-((2-methyl-6-((5-methylpyrazin-2-yl)carbamoyl)benzofuran-4-
yl)oxy)pyrimidine-2-carboxamide), with demethylation occuring on the N,N-
dimethylcarboxamido motif. The t and mass spectral characteristics of M1 formed in liver
microsomes and hepatocytes were identical to the ones discerned with an authentic standard
R
(
data not shown). M1 was the major metabolite of PF-04937319 in human hepatocytes.
+
Metabolite M2 (t = 19.7 min) displayed an exact mass of 449.1567 (MH ), which was 16 Da
R
greater than parent drug. M2 was observed in hepatocytes from rat and human. M2 underwent
in-source fragmentation to yield fragment ions at m/z 431.1465 and 419.1467 (Supplemental
Figure S4). The ion at m/z 419.1467 and the corresponding fragment ions at m/z 388.1041,
1
6

DMD #60087
2
3
60.1091, 281.0557, which were obtained from the MS fragmentation of the m/z 419.1467
(
Supplemental Figure S4) suggested that hydroxylation occurred on the N,N-
dimethylcarboxamide motif in PF-04937319. M2 was tentatively identified as the carbinolamide
metabolite of PF-04937319 (i.e., N-(hydroxymethyl)-N-methyl-5-(2-methyl-6-((5-
methylpyrazin-2-yl)carbamoyl)benzofuran-4-yloxy)pyrimidine-2-carboxamide.
+
Metabolite M3 (t = 18.9 min) displayed an exact mass of 342.1084 (MH ). M3 was observed
R
2
in hepatocytes from rat, dog, and human. The MS product ion spectrum contained a diagnostic
fragment ion at m/z 324.0980, which gave rise to additional ions at m/z 296.1029, 281.0557,
3
2
53.0607, and 225.0659 in the MS product ion spectra. M3 was identified as 4-(2-
(
dimethylcarbamoyl)pyrimidin-5-yloxy)-2-methylbenzofuran-6-carboxylic acid, which is
derived from hydrolytic cleavage across the amide bond in PF-04937319.
+
Metabolite M4 (t
R
= 18.25 min) displayed an exact mass of 328.0929 (MH ), which is 14 Da
2
lower than the molecular weight of M3. M4 was observed in rat hepatocytes. The MS product
ion spectrum of M4 yielded a diagnostic fragment ion at m/z 310.822, which is 14 Da less than
m/z 324.0980 fragment found in M3. M4 was identified as 2-methyl-4-(2-
(
methylcarbamoyl)pyrimidin-5-yloxy)benzofuran-6-carboxylic acid, which is derived from an N-
demethylation in M3.
+
Metabolite M5 (t
R
= 18.1 min) displayed an exact mass of 449.1568 (MH ), which was 16 Da
2
greater than PF-04937319. M5 was observed in hepatocytes from rat, dog, and human. The MS
product ion spectrum yielded diagnostic fragment ions at m/z 431.1462 and 324.0980. The ion at
m/z 324.0980 (also found in the CID spectrum of PF-04937319) suggested that the site of
1
7

DMD #60087
monohydroxylation was on the methylpyrazine motif. The fragment ion at m/z 431.1462, which
corresponded to loss of a water molecule, suggested that hydroxylation occurred on the benzylic
position (i.e., the methyl group) rather than on the pyrazinyl ring. M5 was tentatively assigned
as 5-((6-((5-(hydroxymethyl)pyrazin-2-yl)carbamoyl)-2-methylbenzofuran-4-yl)oxy)-N,N-
dimethypyrimidine-2-carboxamide.
+
Metabolite M6 (t = 17.8 min) displayed an exact mass of 384.1302 (MH ) and fragment ions
R
2
at m/z 367.1040 and 341.1246 in the MS product ion spectrum (Supplemental Figure S5). M6
was observed in hepatocytes from dog and human. The fragment ion at m/z 341.1246 suggested
that M6 was derived from a modification to the methylpyrazine motif. The fragment ion at m/z
3
67.1040, which represented a loss of ammonia suggested that an oxidative scission of the
pyrazine ring had occurred to yield a stable urea metabolite (i.e., 1-(4-(2-
dimethylcarbamoyl)pyrimidin-5-yloxy)-2-methylbenzofuran-6-carbonyl)urea). A plausible
mechanism that could account for the formation of M6 from PF-04937319 is shown in Figure
A.
(
3
+
R
Metabolite M7 (t = 17.6 min) displayed an exact mass of 358.1040 (MH ). Metabolite M7
was only observed in rat hepatocytes. M7 underwent in-source fragmentation to yield product
ions m/z 340.0932 and m/z 328.0932. The m/z 340.0932 fragment, which is 16 Da higher than
the m/z 324.0980 fragment ion in M3, suggested M7 was a monohydroxylated metabolite of M3.
2
The fragment ions at m/z 297.0506 and 269.0559 in the MS product ion spectrum of the m/z
3
40.0932 fragment suggested modification of the N,N-dimethylcarboxamido motif. M7 was
tentatively identified as the carbinolamide metabolite of M3 (i.e., 4-(2-
1
8

DMD #60087
(
(hydroxymethyl)(methyl)carbamoyl)pyrimidin-5-yl)oxy)-2-methylbenzofuran-6-carboxylic
acid).
+
Metabolite M8 (t
R
= 16.8 min) displayed an exact mass of 449.1568 (MH ). Metabolite M8
2
was observed in hepatocytes from rat, dog, and human. The MS product ion spectrum of M8
contained fragment ions at m/z 431.1463, 358.1035 and 340.0928 (Supplemental Figure S6).
3
MS product ion spectrum of the m/z 340.0928 fragment yielded several additional fragment ions
at m/z 312.0977, 297.0504 and 168.0766, which were consistent with monohydroxylation on the
2
methylbenzofuranyl motif. The m/z 431.1463 fragment in the MS spectrum, representing a
facile loss of water, strongly suggested that hydroxylation was occurring specifically on the 2-
methyl position of the benzofuranyl motif. M8 was tentatively identified as 5-((2-
(
hydroxymethyl)-6-((5-methylpyrazin-2-yl)carbamoyl)benzofuran-4-yl)oxy)-N,N-
dimethylpyrimidine-2-carboxamide. M8 was the major metabolite of PF-04937319 in dog
hepatocytes.
+
Metabolite M9 (t
R
= 16.6 min) displayed an exact mass of 341.1244 (MH ). Metabolite M9
2
was observed in dog and human hepatocytes. The MS product ion spectrum of M9
(
Supplemental Figure S7) contained a diagnostic fragment ion at m/z 324.0982, which is also
found in the CID spectrum of M3. The 1 mass unit difference between M3 and M9 (oxygen vs
nitrogen based on exact mass) suggested M9 was an amide derivative (i.e., 5-((6-carbamoyl-2-
methylbenzofuran-4-yl)oxy)-N,N-dimethylpyrimidine-2-carboxamide). A plausible mechanism
that accounts for the formation of M9 from 1 is shown in Figure 3B.
1
9

DMD #60087
+
Metabolite M10 (t
R
= 16.3 min) displayed an exact mass of 435.1411 (MH ). Metabolite M10
2
was observed in rat hepatocytes. The MS product ion spectrum of M10 displayed diagnostic
fragment ions at m/z 417.1306 and 326.0773. The fragment ions at m/z 417.1306 (which
represented the loss of a water molecule) and m/z 326.0773 (2 mass units greater the fragment
ion at m/z 324) were consistent with hydroxylation of the 2-methylbenzofuranyl motif and N-
3
demethylation. MS product ion spectra of the m/z 326.0773 ion yielded additional fragment
ions at m/z 298.0820, 154.0608, and 136.0502 all of which were consistent with the proposed
modification. M10 was tentatively identified as 5-((2-(hydroxymethyl)-6-((5-methylpyrazin-2-
yl)carbamoyl)benzofuran-4-yl)oxy)-N-methylpyrimidine-2-carboxamide. M10 was the major
metabolite of PF-04937319 in rat hepatocytes.
+
Metabolite M11 (t
R
= 15.8 min) displayed an exact mass of 465.1518 (MH ). Metabolite M11
was observed in rat hepatocytes. M11 underwent in-source fragmentation to yield fragment ions
2
at m/z 447.1408 and 435.1410. The MS product ion spectra of the m/z 447.1408 ion yielded
diagnostic fragment ions at m/z 404.0989, 376.1041, 338.0771, and 297.0507, which were
consistent with hydroxylation of the 2-methylbenzofuran and the N,N-dimethylcarboxamide
groups. M11 was tentatively identified as a N-(hydroxymethyl)-5-((2-(hydroxymethyl)-6-((5-
methylpyrazin-2-yl)carbamoyl)benzofuran-4-yl)oxy)-N-methylpyrimidine-2-carboxamide.
+
Metabolite M12 (t
R
= 15.3 min) displayed an exact mass of 421.1624 (MH ), which was 12 Da
lower than the molecular weight of PF-04937319. Metabolite M12 was observed in dog and
2
human hepatocytes. The MS product ion spectrum of M12 (Supplemental Figure S8) contained
fragment ions at m/z 403.1514, 342.1086, 324.0979 and 298.1189. The ions at m/z 324.0979 and
2
98.1189 suggested a modification on the methylpyrazine group. Based on exact mass, a
2
0

DMD #60087
structure involving oxidative constriction of the pyrazine ring with loss of a carbon atom was
proposed for M12 (i.e., N,N-dimethyl-5-((2-methyl-6-((2-methyl-2H-imidazol-4-
yl)carbamoyl)benzofuran-4-yl)oxy)pyrimidine-2-carboxamide), and a plausible mechanism for
its formation is shown in Figure 3C.
+
Metabolite M13 (t
R
= 15.2 min) displayed an exact mass of 451.1359 (MH ). Metabolite M13
was observed in rat hepatocytes. M13 underwent in-source collision to yield product ions m/z
2
4
33.1255 and 421.1254. The MS product ion spectrum of the m/z 433.1255 fragment showed
diagnostic fragment ions at m/z 404.0987 and 324.0616, which were consistent with
hydroxylations on the 2-methyl benzofuranyl and the N-methylcarboxamido groups in metabolite
M1. M13 was tentatively identified as N-(hydroxymethyl)-5-((2-(hydroxymethyl)-6-((5-
methylpyrazin-2-yl)carbamoyl)benzofuran-4-yl)oxy)pyrimidine-2-carboxamide.
+
Metabolite M14 (t
R
= 15.0 min) displayed an exact mass of 518.1405 (MH ). Metabolite M14
2
was observed in hepatocytes from rat, dog, and human. The MS product spectrum of M14
showed a characteristic glucuronide loss of 176 Da to yield product ion m/z 342.1084, which is
the molecular mass of M3. Therefore, M14 was identified as the acyl glucuronide conjugate of
M3 (i.e., 5-((4-(2-(dimethylcarbamoyl)pyrimidin-5-yloxy)-2-methylbenzofuran-6-carbonyl)oxy)-
3
,4,6-trihydroxy-tetrahydro-2H-pyran-2-carboxylic acid.
+
Metabolite M15 (t
R
= 14.5 min) displayed an exact mass of 358.1033 (MH ). M15 was
2
observed only in rat hepatocytes. The MS product spectrum contained diagnostic fragment ions
at m/z 340.0928, 315.0614, 297.0505, which were consistent with monohydroxylation of the 2-
2
1

DMD #60087
methylbenzofuran motif in the hydrolytic product M3. M15 was tentatively identified as 4-(2-
(
dimethylcarbamoyl)pyrimidin-5-yloxy)-2-(hydroxymethyl)benzofuran-6-carboxylic acid.
+
Metabolite M16 (t
R
= 14.0 min) displayed an exact mass of 344.0873 (MH ). M16 was
2
observed only in rat hepatocytes. The MS product ion spectrum of M16 contained diagnostic
fragment ions at m/z 326.0773 and 297.0519, which were consistent with N-demethylation in
M15. M16 was tentatively identified as 2-(hydroxymethyl)-4-(2-(methylcarbamoyl)pyrimidin-5-
yl)oxy)-benzofuran-6-carboxylic acid.
M1 Formation Rates in HLM and rCYPs. PF-04937319 (60 M) and NADPH (1.3 mM)
were incubated for 10–40 min with HLM (0.1–1.5 mg/ml), heat-inactivated HLM, or rCYPs,
which included rCYP1A2, rCYP2B6, rCYP2C8, rCYP2C9, rCYP2C19, rCYP2D6, rCYP2E1,
rCYP3A4, and rCYP3A5 isozymes, to determine linearity of M1 formation. Linear protein
concentrations and incubation times were selected for each enzyme and enzyme kinetics were
determined at PF-04937319 concentrations of 1.0–300 M. K and V or initial reaction rate
m
max
were used to generate CLint, which was in turn used to calculate percent contribution to
metabolism by each rCYP. ISEF-corrected M1 formation rates were highest in rCYP3A4,
.
-1.
-1
rCYP2C9, and rCYP2C19 incubations (Table 2). The total CLint in rCYP (0.62 ml min mg )
.
-1.
-1
was slightly higher than the CL noted in HLM (0.23 ml min mg ).
int
Circulating Metabolites of PF-04937319 in Humans. Circulating metabolites of PF-04937319
in human were qualitatively examined in the 100 mg and 300 mg dose groups from the multiple
dose study. Plasma samples were pooled according to the method of Hamilton et al. (1981) and
metabolic profile in steady state samples (Day 14) was compared with Day 1 samples. A total of
2
2

DMD #60087
eight metabolites were detected in human plasma from both dose groups. Figure 4 shows a
representative HPLC-UV (= 310 nm) trace of pooled plasma from humans dosed with 300 mg
of PF-04937319. Of the eight metabolites detected, only two, M1 and a minor metabolite
designated as M3a were observable in the UV trace (λ = 310nm), the remainder of the
metabolites (M3, M4, M5, M8, M10, and M14 shown in Table 1) were detectable at trace levels
by MS only. The major components observed in human plasma were M1 and unchanged parent
drug. Based on UV integration, PF-04937319 and M1, respectively, accounted for 40% and 50%
of circulating drug related material at Day 1 and 31% and 65% of drug related material at Day
+
1
4. Metabolite M3a (t = 18.4 min) displayed a MH at 435.1411, which was 2 Da higher than
R
PF-04937319 (see Table 1). The exact mass of M3a was consistent with N-demethylation and
hydroxylation of PF-04937319. The product ion spectrum of M3a contained diagnostic
fragment ions at m/z 417.1302 and 405.1304, which suggested that both metabolic steps occurred
on the N,N-dimethylamide portion in PF-04937319. M3a has been tentatively identified as the
N-desmethyl carbinolamide metabolite of PF-04937319 (i.e., N-(hydroxymethyl)-5-((2-methyl-
6
-((5-methylpyrazin-2-yl)carbamoyl)benzofuran-4-yl)oxy)pyrimidine-2-carboxamide). Overall,
these data suggest that M1 had accumulated to a significant extent at steady state (Day 14)
compared to Day 1.
Comparison of M1 Concentrations in Animals and Humans. Because M1 was present at
levels greater than 10% of circulating drug related material, it became necessary to qualify the
metabolite in preclinical species as per the MIST guidance. Consequently, an authentic standard
of M1 was synthesized, followed by a quantitative analysis of M1 in the plasma samples from
the multiple dose human study and the 3-month rat and dog toxicity studies as a representative
2
3

DMD #60087
case. Table 3 lists the pharmacokinetic parameters of M1 in humans administered with 100 mg
and 300 mg of PF-04937319 once daily for 14 days. For the purposes of comparison, the
toxicokinetic parameters of M1 (on Days 1 and 90/91) in rats and dogs at the no-observed-
adverse-effect level (NOAEL) doses of 1000 mg/kg (male rats)/50 mg/kg (female rats) and 5
mg/kg (male and female dogs) are also shown. Figures 5, 6, and 7 illustrate representative mean
total plasma concentration versus time profile for PF-04937319 and M1 in humans (Days 1 and
1
4 at the 300 mg dose of PF-04937319), rats (Days 1 and 90 at the 50 mg/kg (female) and 1000
mg/kg (male) doses of PF-04937319), and dogs (Days 1 and 90 at the 5 mg/kg doses of PF-
0
4937319), respectively.
Systemic exposures of M1 were assessed by comparison of Cmax and/or AUC on Day 1 and
steady state (i.e., Day 14 for human and Day 90 or 91 for rat and dog). Both total and free
plasma concentrations were examined in the analysis. In humans, systemic exposure to M1
increased in a dose-dependent fashion (100–300 mg) with 1.9–2.8-fold accumulation of M1 over
the period of 14 days. The estimated half-life (t1/2) of M1 in humans dosed with PF-04937319
was ~ 23 h for the two dose strengths. Compared with the steady state levels of M1 observed in
humans at the 100 mg and 300 mg dose groups of PF-04937319, the NOAEL dose of 5 mg/kg in
male or female dogs revealed considerably lower circulating levels of M1 (total or free) at steady
state. PF-04937319 demonstrated a gender difference in rat pharmacokinetics; female rats
exhibited higher systemic exposure that male rats due to a lower plasma clearance, leading to
differences in NOAEL between male and female animals. Examination of total and free steady
state exposures (Day 91) of M1 in female rats dosed with 50 mg/kg of PF-04937319 also did not
reveal sufficient coverage of the steady state M1 levels (total or free) observed in humans.
2
4

DMD #60087
Steady state M1 exposure in humans at the 100 mg dose group of PF-04937319 was covered in
terms of total Cmax (~7.7x) and total AUC (~3.6x) in male rats dosed with 1000 mg/kg of PF-
0
0
4937319 for 90 days. Steady state M1 exposure in humans at the 300 mg dose group of PF-
4937319 was covered on the basis of total Cmax (~ 1.8x) and AUC (0.8x) in male rats dosed
with 1000 mg/kg of PF-04937319 for 90 days.
Plasma Protein Binding. The extent of in vitro binding of PF-04937319 and M1 to plasma
proteins was evaluated by equilibrium dialysis in rat, dog, and human plasma. The mean (±
S.D.) unbound fraction of PF-04937319 at concentrations of 0.55 and 5.5 M were 0.24 ± 0.02
and 0.24 ± 0.01 (rat), 0.37 ± 0.02 and 0.39 ± 0.02 (dog), and 0.30 ± 0.02 and 0.32 ± 0.02
(
human). The mean unbound fraction of M1 (2 M, equivalent to 0.837 g/ml) in rat, dog, and
human was 0.0801, 0.192, and 0.142, respectively. These results indicate that M1 has slightly
higher plasma protein binding than PF-04937319. Both PF-04937319 and M1 were stable in rat,
dog, and human plasma for 4 h at 37 °C at the concentrations used in the plasma free fraction
determination.
Discussion
The current article discloses our initial findings on MIST studies with the glucokinase activator
PF-04937319 that were conducted during Phase 1 clinical studies. Examination of human
pharmacokinetics following 14 days of daily administration of PF-04937319 (10–300 mg)
revealed mean terminal t1/2 values ranging from 6.3–8.5 h, which also were in reasonable
agreement with the predicted t1/2 value using hepatic clearance values from liver microsomes
and/or hepatocytes and volume of distribution from allometric scaling of animal data. Consistent
2
5

DMD #60087
with the observations in preclinical species, less than 1% of the oral dose was excreted
unchanged in urine.
Qualitative examination of the in vitro metabolic fate in rat, dog, and human hepatocytes
revealed that PF-04937319 was metabolized via oxidative and hydrolytic pathways (Figure 8).
Although there were no human unique metabolites, differences were noted with regards to the
biotransformation profile of PF-04937319 across preclinical species and human. N-
demethylation to M1 was the major route of metabolism in human hepatocytes, while M10,
derived from a hydroxylation of the 2-methylbenzofuran group in M1, was the major metabolite
in rat hepatocytes. In the case of dog, monohydroxylation at the 2-methylbenzofuran group in
PF-04937319 resulting in M8 was the dominant metabolic fate. Of much interest was the
formation of the metabolites M6, M9, and M12 that were derived from pyrazine ring scission.
The proposed structures of M6, M9, and M12 shown in Table 1 are consistent with the observed
molecular weight and fragmentation pattern. Postulated mechanisms leading to the formation of
M6, M9, and M12 are depicted in Figure 3. Pyrazine ring oxidation to epoxide 1 and its
subsequent ring opened form 2 were proposed as common intermediates in the formation of M6
and M12, respectively (Figure 3, pathways A and C). In the case of M6, addition of water on the
pyrazine ring in 2 would lead to the diol 3, followed by ring scission to the dihydrooxazole
intermediate 4. The formation of M6 from 4 can occur via a two step process involving
hydrolysis across the carboximidoamide motif in 4 to 5 followed by elimination of 4-
methyloxazole and water. The formation of M12 can be envisioned from a pyrazine ring
scission in 2 to a 2H-imidazole-carbaldehyde intermediate 6. The hemiacetal form of the 2H-
imidazole-carbaldehyde derivative (i.e., 7) could then spontaneously lose carbon dioxide and
2
6

DMD #60087
lead to M12 (Figure 3C). In the case of M9, a two-electron oxidation on the 2-amino-5-
methylpyrazine motif could lead to the corresponding imine-methide intermediate 8 followed by
hydrolysis to the primary carboxamide metabolite M9 via the intermediate carbinolamine 9
(
Figure 3B). The mechanisms depicted in Figure 3 are purely speculative, and will require
additional mechanistic studies as further proof for the proposed pathways. A pragmatic starting
point involving the isolation, purification and NMR characterization of M7, M9, and M12 to
obtain unequivocal proof for the proposed structures is currently ongoing in our laboratories.
Metabolite characterization in early development phase has been the major focus of MIST
discussions and although the strategic approaches and methodologies may vary, there is a clear
consensus that an assessment of circulating metabolite exposure across species at steady state to
underwrite safety is a critical component of this stage (Aurell Holmberg et al., 2014; Haglund et
al., 2014; Nedderman et al., 2011; Yu et al., 2010; Leclercq et al., 2009; Luffer-Atlas, 2008).
The availability of steady state plasma samples from the 14 day study on PF-04937319 in T2DM
patients were an appropriate starting point towards MIST studies. Plasma samples from the 100
mg and 300 mg dose groups were chosen for the analysis, since they represented an efficacious
dose range based on glucose lowering effects observed in the T2DM patients (Pfizer data on
file). The major components in human plasma were unchanged parent drug and M1, an
observation that was consistent with the in vitro metabolic profile in human hepatocytes and
previous literature (Dalvie et al, 2009). Because M1 accounted for 65% of the drug related
material on Day 14, its synthesis was triggered and the authentic standard was used for
quantitative comparison of steady state exposures in plasma from human and the 3–month safety
studies in rats and dogs at the NOAEL. The t of M1 (~ 23 h) was significantly longer than the
1
/2
2
7

DMD #60087
t
1/2 of the parent compound (~ 6.3–8.5 h), and as a result, the systemic exposure of M1 was
higher than the exposure of the parent compound at steady state at both the 100 and 300 mg
doses. In contrast, the circulating levels of M1 were very low in beagle dogs and female rats.
Steady state human exposure of M1 at the 100 mg and the 300 mg dose group, however, was
covered in terms of total Cmax (~7.7x and 1.8x for the 100 mg and 300 mg, doses, respectively) in
male rats. Likewise, male rats provided a 3.6x and 0.8x AUC coverage for M1 at the human
doses of 100 and 300 mg, respectively. Because AUC differences of ≥ 2-fold are generally
considered to be meaningful in toxicokinetic evaluations (International Conference on
Harmonisation, 2012), our present data indicate that M1 is toxicologically qualified at this time.
Of much interest was the identification of the minor circulating metabolite M3a, which was
tentatively assigned as a carbinolamide derivative of M1. The formation of M3a can potentially
occur via a CYP mediated N-demethylation of PF-04937319 to M1 followed by a two-electron
oxidation of M1 to an iminium species, which can generate M3a upon hydration of the iminium
bond. Surprisingly, liver microsomal or hepatocyte incubations of PF-04937319 in rat, dog and
human failed to generate M3a. It is possible that M3a is a very minor metabolic fate of PF-
0
4937319, and the qualitative in vitro metabolism studies using unlabeled PF-04937319 may not
be sensitive enough to detect its presence. In vitro studies are in progress to examine whether the
formation of M3a proceeds via the M1 metabolite. Greater granularity in terms of total body
burden of M1 and M3a will be necessary in subsequent mass balance studies in rats and human,
especially for M3a, which can be labeled as human unique (even though it constitutes only ~4%
of total drug related material). Likewise, further studies will be conducted as necessary to
maintain adequate exposure of at least one animal species to metabolite M1 and particularly
2
8

DMD #60087
M3a. This will ultimately be dependent on the clinically efficacious doses of PF-04937319
utilized in future trials, and the corresponding NOAEL in longer term (> 3 months) toxicity
studies.
The availability of an authentic standard of M1 also facilitated efforts into CYP phenotyping.
Examination of the rates of formation of M1 in incubations of PF-04937319 in a panel of
recombinant human CYP isozymes revealed that M1 is mainly formed by CYP3A4, CYP2C9,
and CYP2C19. Comparison of the plasma protein binding of PF-04937319 and M1 in the
preclinical species of toxicological evaluation (rat and dog) and human revealed that M1 was ~
3
-, 2-, and 2-fold more bound to plasma proteins in rat, dog, and human, respectively. A recent
drug-drug interaction (DDI) guidance from the European Medicines Agency
www.ema.europa.eu/docs/en_GB/document_library/Scientific_guideline/2012/07/WC50012960
.pdf) and the draft DDI guidance from the Food and Drug Administration
www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/ucm29
362.pdf) have proposed that metabolites present at >25% of the parent AUC and >10% of the
(
6
(
2
total drug-related exposure should be investigated in vitro for their DDI potential. Therefore, the
inhibitory potential of M1 against the major human CYP isoforms was examined and compared
with the parent compound. Both PF-04937319 and M1 were devoid of competitive CYP
inhibition against CYP1A2, CYP2C8, CYP2C19, CYP2D6, and CYP3A4 (IC50 values > 30 M)
(Pfizer data on file). PF-04937319 and M1 were also devoid of time- and concentration-
dependent inhibitory effects on CYP3A4.
Additional investigations on the on- and off-target pharmacology of M1 were also initiated.
First and foremost, the pharmacologic activity of PF-04937319 and M1 was compared in the in
2
9

DMD #60087
vitro biochemical activation assay (Pfefferkorn et al. 2011), which revealed that M1 was
significantly less potent as a human GK activator than the parent compound (PF-04937319: EC50
=
0.17 M; M1: EC50 = 4.69 M). Furthermore, at the 300 mg dose group, the steady state
unbound Cmax of M1 in humans was estimated to be 0.54 M, which is significantly lower than
its in vitro EC50 value. Additional assessment of safety revealed that M1 (like the parent
compound) was devoid of mutagenic responses in the Salmonella Ames assay (in the absence or
presence of metabolic activation). Likewise, no off-target activity was discerned upon
evaluation of M1 (at a concentration of 100 M) for broader pharmacological activity in a panel
of receptors, ion channels and enzymes in the CEREP panel
(
http://www.cerep.fr/cerep/users/pages/productservices/industrialization.asp). In conclusion, we
have described a case study on the assessment of the MIST regulatory guidance during the early
development phase of a clinical candidate, and the initial pragmatic steps undertaken to de-
risk/qualify a metabolite that exceeded 10% of total drug related material in humans. Insights
gained in this analysis should lead to a more efficient and focused development plan for fulfilling
MIST requirements with PF-04937319.
3
0