learning and confirming with preclinical studies: modeling
TRANSCRIPT
DMD #53926 1
Learning and Confirming with Preclinical Studies: Modeling and
Simulation in the Discovery of GDC-0917, an IAP Antagonist
Harvey Wong, Stephen E. Gould, Nageshwar Budha, Walter C. Darbonne, Edward E.
Kadel III, Hank La, Bruno Alicke, Jason S. Halladay, Rebecca Erickson, Chia Portera,
Anthony W. Tolcher, Jeffery. R. Infante, Michael Mamounas, John A. Flygare, Cornelis
E.C.A. Hop and Wayne J. Fairbrother.
Departments of Drug Metabolism and Pharmacokinetics (HW, HL, JSH, CECAH),
Translational Oncology (SEG, BA), Clinical Pharmacokinetics (NB), Oncology
Biomarkers (WCD, EEK), Exploratory Clinical Development (CP, MM), Safety
Assessment (RE), and Medicinal Chemistry (JAF), Early Discovery Biochemistry (WJF),
Genentech Inc., South San Francisco, CA, USA.
South Texas Accelerated Research Therapeutics, LLC., San Antonio, TX, USA (AWT)
Sarah Cannon Research Institute/Tennessee Oncology, Nashville, TN (JRI)
DMD Fast Forward. Published on September 16, 2013 as doi:10.1124/dmd.113.053926
Copyright 2013 by the American Society for Pharmacology and Experimental Therapeutics.
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 2
RUNNING TITLE: Preclinical Modeling and Simulation of GDC-0917.
Author to whom correspondence should be addressed:
Harvey Wong, PhD Drug Metabolism and Pharmacokinetics Genentech 1 DNA Way, MS 412a South San Francisco, CA, 94080 Phone: 650-225-5739 Fax: 650-467-3487 E-mail: [email protected]
Number of pages: 42
Number of Tables: 4
Number of Figures: 7
Number of References: 25
Number of Words in Abstract: 236
Introduction: 764
Discussion: 1338
List of abbreviations: AUC0-24: area under the concentration-time curve from zero to 24
hours; AUCinf :area under the concentration–time curve from zero to infinity; Cl: plasma
clearance; Cmax: highest observed plasma concentration; F: bioavailability; IV:
intravenous; I: cIAP level expressed as a percentage of the DMSO control; Iinitial: initial
value of I; Imax: the maximum value of I; IC50: GDC-0917 concentration where I is 50%
of Imax; kng: net growth rate constant; K: rate constant describing the anti-tumor effect of
GDC-0917; Kmax: maximum value of K; KC50: GDC-0917 concentration where K is
50% of Kmax; MRT: mean residence time; t1/2: half-life; PO: oral; tmax: time at which Cmax
occurred; Vss: volume of distribution at steady state.
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 3
ABSTRACT
The application of modeling and simulation techniques is increasingly common in
preclinical stages of the drug development process. GDC-0917 [(S)-1-((S)-2-cyclohexyl-
2-((S)-2-(methylamino)propanamido)acetyl)-N-(2-(oxazol-2-yl)-4-phenylthiazol-5-
yl)pyrrolidine-2-carboxamide] is a potent second generation antagonist of inhibitor of
apoptosis (IAP) proteins being developed for the treatment of various cancers. GDC-
0917 has low to moderate clearance in mouse (12.0 ml/min/kg), rat (27.0 mL/min/kg) and
dog (15.3 mL/min/kg), and high clearance in monkey (67.6 mL/min/kg). Accordingly,
oral bioavailability was lowest in monkey compared to other species. Based on our
experience with a prototype molecule with similar structure, in vitro-in vivo extrapolation
was used to predict a moderate clearance (11.5 mL/min/kg) in human. Predicted human
volume of distribution was estimated using simple allometry at 6.69 L/kg. Translational
pharmacokinetic-pharmacodynamic (PK-PD) analysis using results from MDA-MB-231-
X1.1 breast cancer xenograft studies and predicted human pharmacokinetics suggests that
ED50 and ED90 targets can be achieved in humans using acceptable doses (72 mg and 660
mg, respectively), and under an acceptable timeframe. The relationship between GDC-
0917 concentrations and pharmacodynamic response (cIAP1 degradation) was
characterized using an in vitro PBMC immunoassay. Simulations of human GDC-0917
plasma concentration-time profile and cIAP1 degradation at the 5 mg starting dose in the
Phase 1 clinical trial agreed well with observations. This work shows the importance of
leveraging information from prototype molecules, and illustrates how modeling and
simulation can be used to add value to preclinical studies in the early stages of the drug
development process.
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 4
INTRODUCTION
Inhibitor of apoptosis (IAP) proteins are involved in the regulation of programmed cell
death or apoptosis and were first identified in baculoviruses where they serve to suppress
host cell death responses during viral infection (Flygare and Fairbrother 2010,
Varfolomeev and Vucic 2011). IAP proteins suppress apoptosis, in part, by inhibiting
activated cytosolic cysteine/aspartate-specific proteases (caspases) that are critical for its
execution (Flygare and Fairbrother 2010, Varfolomeev and Vucic 2011). The
mammalian family of IAP proteins includes X-linked IAP (XIAP), cellular IAP 1 and 2
(cIAP1 and cIAP2), and melanoma IAP (ML-IAP). Of these, only XIAP directly inhibits
effector caspases-3 and-7, which execute the apoptotic death program. XIAP also
indirectly inhibits effector-caspase activity via inhibition of the initiator caspase-9. ML-
IAP and cIAPs directly antagonize second mitochondria-derived activator of
caspase/direct IAP-binding protein with low pI (Smac/DIABLO), an endogenous IAP
antagonist that acts by blocking XIAP-mediated inhibition of caspases. The cIAP
proteins have an additional role of preventing TNF-α-mediated activation of initiator
caspase-8, thereby stopping downstream activation of effector caspases-3,-7. Finally, the
cIAP proteins positively regulate canonical- and negatively regulate non-canonical-NF-
κB signaling and induction of NF-κB target genes, including TNF-α. Frequently, IAP
proteins are over expressed in cancer cells where they act as regulators of cancer cell
survival and are often indicators of poor prognosis (Vucic and Fairbrother, 2007).
Inhibition of IAP proteins has been shown to sensitize cancer cells to pro-apoptotic anti-
cancer agents. Therefore, the development of IAP antagonists as potential anti-cancer
agents has been of interest.
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 5
Mathematical modeling and simulation, and the “learn-and-confirm paradigm”,
(Sheiner, 1997) have been applied widely in later stages of the drug development process,
largely with data from clinical trials (Chien et al., 2005). The “learn-and-confirm
paradigm” involves the use of mathematical models to integrate information and “learn”
about a drug candidate, but at the same time continuously “confirm” model assumptions
as new information is generated during the course of the drug development process. The
net result of successive iterations of “learning” and “confirming” is an overall better
understanding of a drug candidate’s pharmacological behavior. The application of
modeling and simulation techniques has been rapidly increasing in the preclinical drug
discovery setting as a means to reduce drug candidate attrition due to a lack of clinical
efficacy. In particular translational pharmacokinetic-pharmacodynamic (PK-PD) analysis
has been used to help interpret and translate the results of preclinical studies (Salphati et
al., 2010, Liu et al., 2011, Wong et al., 2012b, Wong et al., 2012c, Yamazaki et al.,
2012). PK-PD modeling involves the use of differential equations to quantitatively
describe drug behavior in the biological system of interest. This type of analysis is
particularly helpful in understanding concentration effect relationships where there are
apparent disconnects in drug concentration and effect related to temporal delays in drug
action. In addition, mechanism-based PK-PD models allow for known species
differences in factors such as pharmacokinetics or in vivo drug potency to be
quantitatively assessed as well as enabling a more robust characterization of
concentration-effect relationships from frequently variable in vivo preclinical efficacy
data. Finally, establishment of PK-PD models enables better integration of available data
and prospective simulations.
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 6
We have previously reported on the biological characteristics, including the
preclinical and clinical pharmacokinetics, of GDC-0152 a potent pan-selective antagonist
of IAP proteins (Figure 1; Flygare et al., 2012). GDC-0917 [(S)-1-((S)-2-cyclohexyl-2-
((S)-2-(methylamino)propanamido)acetyl)-N-(2-(oxazol-2-yl)-4-phenylthiazol-5-
yl)pyrrolidine-2-carboxamide] is a potent second generation pan-selective IAP antagonist
with binding affinities to the BIR3 domains of X-IAP and cIAP1/2, and the BIR domain
of ML-IAP in the low nanomolar range (Ki < 60 nM) using a fluorescence polarization-
based competition assay. The previously undisclosed structure of GDC-0917 is shown in
Figure 1. GDC-0917 caused a decrease in cell viability in MDA-MB-231 human breast
cancer cells while having no effect on normal human mammary epithelial cells. In
contrast to GDC-0152, which was administered intravenously, the intended route of
administration of GDC-0917 is oral. The objective of the current study was to
characterize the preclinical pharmacokinetic properties of GDC-0917. We show how
learnings from a prototype molecule, GDC-0152, can be applied to a second-generation
molecule, GDC-0917. In particular, we show how our prior knowledge of the disposition
characteristics of GDC-0152 was leveraged in order to predict human pharmacokinetics
for GDC-0917. Furthermore, prospective human simulations were performed in order to
determine the number of patient cohorts required to achieve predicted human efficacious
doses as a means to assess the feasibility of the human clinical trial. The current study
serves as an illustration of how mathematical modeling and simulation techniques can be
used to provide an integrated preclinical assessment of a drug candidate and aid in the
translation of preclinical data to the clinic.
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 7
MATERIALS AND METHODS
PRECLINICAL GDC-0917 IN VIVO PHARMACOKINETIC (PK) STUDIES IN
RAT, MOUSE, DOG AND MONKEY
The preclinical in vivo studies described in this manuscript were conducted according to
protocols approved by the IACUC at Genentech (South San Francisco, CA, USA).
Rat PK study: Three male Sprague-Dawley rats (Charles River Laboratories, Inc.,
Hollister, CA, USA) were given an intravenous (IV) dose of 1 mg/kg (free-base
equivalent) of GDC-0917 trifluoacetate salt (Genentech, Inc., South San Francisco, CA,
USA) in phosphate buffered saline, via femoral vein cannulae. An additional three male
rats were given a single oral (PO) dose of 5 mg/kg (free-base equivalent) of GDC-0917
trifluoacetate salt in phosphate buffered saline. At the initiation of the study, the rats
weighed from 284 to 306 g. Blood samples (~0.2 mL per sample) were collected from
each animal via jugular vein cannulae at the following timepoints: predose, 0.033, 0.083,
0.25, 0.5, 1, 2, 4, 8, and 24 hours postdose. All samples were collected into tubes
containing potassium ethylenediaminetetraacetic acid (EDTA) as an anticoagulant.
Blood samples were centrifuged within 30 minutes of collection and plasma was
harvested. Plasma samples were stored at approximately –70oC until analysis for GDC-
0917 concentrations by a liquid chromatographic-tandem mass spectrometric
(LC/MS/MS) assay method.
Mouse PK Study: Twenty-seven female SCID.bg mice (Charles River Laboratories,
Hollister, CA, USA) were given an IV dose of 1 mg/kg of GDC-0917 in 15%
hydroxypropyl-β-cyclodextrin, 20 mM succinic acid in water. An additional 27 female
SCID.bg mice were given a single PO dose of 3 mg/kg of GDC-0917 in 0.5% w/v
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 8
methylcellulose with 0.2% v/v Tween 80. Mice weighed from 17 to 20 g at the initiation
of the study. Approximately 0.2 mL of blood was collected from each mouse by terminal
cardiac puncture. Blood samples (n = 3 mice per timepoint) were taken from
intravenously dosed mice at the following timepoints: predose; 0.033, 0.17, 0.5, 1, 3, 6,
9, and 24 hours postdose. Blood samples were collected from orally dosed mice at the
following timepoints (n = 3 mice per timepoint): predose; 0.083, 0.25, 0.5, 1, 3, 6, 9, and
24 hours postdose. All blood samples were collected into tubes containing EDTA as an
anticoagulant. Blood samples were centrifuged within 30 minutes of collection and
plasma was harvested. Plasma samples were stored at approximately –70oC until
analysis for GDC-0917 by LC/MS/MS.
Dog PK Study: Three male beagle dogs (Covance, Cumberland, VA, USA) were given a
single IV dose of 1 mg/kg of GDC-0917 in 15% hydroxypropyl-β-cyclodextrin, 20 mM
succinic acid in water. An additional three male beagle dogs (Marshall Bioresources,
Beijing, China) were administered a PO dose of 1 mg/kg of GDC-0917 in 0.5% w/v
methylcellulose with 0.2% v/v Tween 80. At the initiation of this study, dogs weighed
from 9.2 to 12.5 kg. Blood samples (~3 mL per sample) were collected from the
peripheral vein of IV and PO animals at the following timepoints: predose, 0.033, 0.083,
0.25, 0.5, 1, 3, 6, 9, and 24 hours post-dose. All blood samples were collected in tubes
containing EDTA. Blood samples were centrifuged and plasma was harvested within 1
hour of collection. Urine was collected from PO dogs into containers on wet ice during
the following intervals: predose and from 0-8 and 8-24 hours postdose. Plasma and
urine samples were stored at approximately −70°C until analysis for GDC-0917 by
LC/MS/MS.
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 9
Monkey PK Study: Three male cynomolgus monkeys (Covance Laboratories, Inc.,
Madison, WI, USA) were given a single IV dose of 1 mg/kg GDC-0917 in 15%
hydroxypropyl-β-cyclodextrin, 20 mM succinic acid in water via a saphenous vein. An
additional three male monkeys were given a PO dose of 2 mg/kg GDC-0917 in 0.5% w/v
methylcellulose with 0.2% v/v Tween 80. At the initiation of this study, monkeys
weighed from 2.1 to 2.6 kg. Blood samples (~1 mL per sample) were collected from the
femoral vein of each animal at the following timepoints after IV dosing: predose, 0.033,
0.083, 0.25, 0.5, 1, 2, 4, 8, 12, and 24 hours postdose. Following PO dosing, blood
samples were collected at the following timepoints: predose, 0.083, 0.25, 0.5, 1, 2, 4, 8,
12, and 24 hours postdose. All blood samples were collected into tubes containing
EDTA. Blood samples were centrifuged within 1 hour of collection and plasma
harvested. Urine was collected from monkeys receiving the IV dose at predose and from
0-8 and 8-24 hours postdose in plastic containers surrounded by wet ice. Plasma and
urine samples were stored at approximately −70°C until analysis for GDC-0917 by
LC/MS/MS.
HUMAN PHARMACOKINETIC STUDY WITH GDC-0917
The human pharmacokinetic parameters were determined using data from a Phase 1 dose
escalation study using a modified continual reassessment method design in cancer
patients. Institutional Review Boards at participating institutions approved the protocol
for the Phase 1 study. Written informed consent was obtained from all trial participants.
Briefly, doses ranging from 5 to 600 mg of GDC-0917 were administered orally as a
single dose to cancer patients. Following a 3 day washout period, GDC-0917 was
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 10
administered daily for 14 days. Plasma samples were collected at predetermined time
points after the first dose (0, 0.5, 1, 2, 3, 6, 9, 24, 48, and 72 hours) and the last dose (0,
0.5, 1, 2, 3, 6, and 24 hours) for characterization of single-dose and steady state
pharmacokinetics. Plasma samples were stored at approximately −70oC until analysis for
GDC-0917 concentrations by LC/MS/MS. Arithmetic mean concentration-time data are
presented for three Caucasian female patients following a single oral 5 mg dose of GDC-
0917. The three patients ranged from 36 to 56 years of age and their weight ranged from
66 to 99 kg.
Blood was collected for pharmacodynamics marker (cIAP1) measurement at
predose and 6 hours postdose following a single oral dose of GDC-0917. cIAP1 protein
concentrations in patient samples were determined using a cIAP1 immunoassay in
peripheral blood mononuclear cells as described below. cIAP1 levels in patient samples
taken at 6 hour postdose are expressed as a percentage of predose cIAP1 levels.
PHARMACOKINETIC ANALYSIS
All pharmacokinetic parameters were calculated by non-compartmental methods as
described in Gibaldi and Perrier (1982) using WinNonlin® version 3.2
(Pharsight Corporation; Mountain View, CA, USA). Parameters are presented as a mean
± standard deviation.
PREDICTION OF VOLUME OF DISTRIBUTION BY ALLOMETRIC SCALING
Simple allometric scaling was used to predict human volume of distribution. In vivo
volume of distribution at steady-state (Vss) estimates from preclinical pharmacokinetic
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 11
studies were fit to the following log transformed form of the allometric equation
(Boxenbaum, 1984) by linear regression:
log(Y ) = log(a)+ b log(W )
where Y is the volume of distribution, a is the coefficient of the allometric equation, W is
the body weight, and b is the allometric exponent. Predicted human volume of
distribution was extrapolated using the fitted line and an assumed body weight of 70 kg.
PREDICTION OF HEPATIC CLEARANCE USING METABOLIC STABILITY
STUDIES IN HEPATOCYTES
The metabolic stability of GDC-0917 (1 µM) was determined in pooled (n = 5-10)
human, rat, and dog cryopreserved primary hepatocytes (CellzDirect Invitrogen
Corporation, Durham, NC, USA). Incubations (final incubation volume 250 µL) were
performed at 37oC under an atmosphere of 95/5 CO2/O2 using 0.5 million cells/mL in
hepatocyte incubation media (In Vitro Technologies, Baltimore, MD, USA). 50 µL
aliquots (cells and media) were taken at 0, 1, 2 and 3 hour following the start of the
incubation and quenched using 100 µL ice cold acetonitrile. The resulting samples were
centrifuged at 3000 g for 10 min. Supernatant (90 µL) was diluted in water (180 µL)
and % remaining of GDC-0917 was assessed by LC/MS/MS. All incubations were
performed in triplicate. Intrinsic clearance was estimated from the % GDC-0917
remaining versus time plot using standard methods described by Obach et al (1997).
Equation 1
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 12
Intrinsic clearance was scaled to hepatic clearance using the “well-stirred” model by
standard methods (Khojasteh et al., 2011).
PREDICTION OF HALF-LIFE
Human half-life (t1/2) was predicted using the following equation:
693.02/1 ×=CL
Vt ss
where CL is the predicted human hepatic clearance based on in vitro-in vivo extrapolation
of hepatocyte metabolic stability studies with GDC-0917 and Vss is the predicted human
volume of distribution using allometry (see above).
ANTI-TUMOR EFFICACY STUDIES IN MDA-MB-231-X1.1 XENOGRAFT
MICE
Briefly, 10 million MDA-MB-231-X1.1 breast adenocarcinoma cells resuspended in
Hank’s Balanced Salt Solution and Matrigel (1:1, v/v) were implanted subcutaneously
into the upper right flank of female SCID.bg mice (Charles River Laboratories, Hollister,
CA, USA). MDA-MB-231-X1.1 cells are MDA-MB-231 cells (originally from ATCC,
Manassas, VA, USA) that were selected for improved in vivo growth rates. When tumor
volumes reached approximately 100-300 mm3, mice were assigned to treatment groups to
get a similar mean tumor size for each treatment group. Treatment groups (n = 5 per
group) were administered once daily oral doses of vehicle (15%
hydroxypropyl-β-cyclodextrin, 20 mM succinic acid in water), 0.08, 0.17, 0.34, 0.68,
Equation 2
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 13
1.36, 2.72, 5.43, 10.87 or 16.30 mg/kg of GDC-0917 for 21 days. Tumor volumes were
measured in two dimensions (length and width) using Ultra Cal IV calipers (Model 54 10
111, Fred V. Fowler Company, Inc., Newton, MA). The following formula was used
with Excel v11.2 (Microsoft Corporation, Redmond, WA) to calculate tumor volume
(TV): TV (mm3) = (length × width2) × 0.5. Tumor sizes and body weights were recorded
twice weekly, and the mice were regularly observed over the course of the study. Mice
were euthanized if their tumor volume exceeded 2000 mm3 or if their body weight
dropped by more than 20% of the starting weight.
A satellite group of female SCID.bg mice were given single oral doses of 0.5 to
20 mg/kg of GDC-0917 in 15% hydroxypropyl-β-cyclodextrin, 20 mM succinic acid in
water to provide an assessment of the oral pharmacokinetics of GDC-0917 to support the
xenograft efficacy study. Blood was collected at 0.08, 0.25, 0.5, 1, 3, 6, 9, and 24 hours
postdose and plasma was harvested within 1 hour of blood collection. Plasma samples
were frozen at approximately −80oC until analysis for GDC-0917 concentrations using
LC/MS/MS.
PHARMACOKINETIC-PHARMACODYNAMIC (PK-PD) MODELING
PK-PD modeling was performed using SAAM II (Saam Institute, University of
Washington, Seattle, WA).
Supporting Mouse Pharmacokinetic Studies for PK-PD Modeling:
Plasma concentration-time data from female SCID.bg mice given 0.5 to 20 mg/kg
GDC-0917 were simultaneously fitted to a one-compartment model with oral absorption.
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 14
The pharmacokinetics of GDC-0917 appeared linear over the dose range tested. Mean
estimates of the absorption rate constant (ka), the elimination rate constant (ke), and the
apparent volume of distribution (V/F) from all dose groups were used to simulate
GDC-0917 concentrations in the modeling of the MDA-MB-231-X1.1 xenograft efficacy
studies.
MDA-MB-231.X1.1 Breast Cancer Xenograft Efficacy Studies:
MDA-MB-231.X1.1 xenograft efficacy studies were fitted to an indirect response model
(Mager et al., 2003) as described by the following differential equation.
d(TV )
dt= kng(TV ) − K(TV ) (Equation 3)
where K = Kmax ×C
KC50 + C
TV (mm3) is defined as the tumor volume, t (hr) is time, kng (hr-1) is the net growth rate
constant, K (hr-1) is the rate constant describing the anti-tumor effect of GDC-0917, Kmax
(hr-1) is the maximum value of K, C (µM) is the concentration of GDC-0917, and KC50
(µM) is the GDC-0917 concentration where K is 50% of Kmax. Concentrations of
GDC-0917 in mice were simulated based upon the pharmacokinetic parameters obtained
from the supporting mouse pharmacokinetics studies described above. Mean tumor
volumes for each dose group were used for fitting. Pharmacodynamic parameters are
presented as the estimate followed by the coefficient of variation (CV) in parentheses.
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 15
cIAP IMMUNOASSAY IN PERIPHERAL BLOOD MONONUCLEAR CELLS
(PBMCs)
The purpose of this study was to establish a cIAP1 immunoassay in peripheral blood
mononuclear cells (PBMCs) for use in the clinical trial. Whole blood was collected in
eight 8 mL cell preparation tubes (CPT; Becton Dickinson, Franklin Lakes, NJ, USA)
containing sodium heparin from each donor. Blood was pooled and transferred to 14 mL
tubes and incubated in the dark at room temperature for ~ 16 hours in the presence of
vehicle (DMSO (dimethylsulfoxide)) or 0.0001 to 10 µM of GDC-0917. Following
incubation, each blood sample was transferred back into a CPT tube and PBMCs were
harvested from blood after centrifugation as per manufacturer’s instructions and washed
in cold phosphate buffered saline. Next, resulting PBMC pellets were lysed in 100 µL 1×
Cell Extraction Buffer (Invitrogen, Grand Island, NY, USA) containing protease
inhibitors (Roche, Basel, Switzerland). Lysates were transferred to 1.5 mL microfuge
tubes and centrifuged at 13,000 × g for ten minutes at 4°C. Total cellular protein
concentration for each PBMC lysate was determined using a micro-bicinchoninic acid
assay (Pierce Biotechnology, Rockford, lL, USA). cIAP1 concentration was determined
using an internally-developed qualified cIAP1 immunoassay described as follows.
The cIAP1 assay involved incubation of the PBMC sample for ~ 16 hours with a
biotinylated goat anti-human cIAP1 polyclonal antibody (R&D Systems, Minneapolis,
MN, USA), and a ruthenium-labeled rat anti-human cIAP1 monoclonal antibody (Enzo
Life Sciences, Farmingdale, NY, USA) in assay diluent. Strepavidin-coated plates were
blocked for non-specific binding using 3 % bovine serum albumin in wash buffer (0.05%
polysorbate-20 in phosphate buffered saline) for ~16 hours at 4°C. The blocked plate was
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 16
washed three times with 200 µL assay wash buffer. The sample/antibody mixture was
incubated with the blocked immunoassay plate for 90 minutes at room temperature. The
assay plate was again washed three times in assay wash buffer prior to addition of a
chemiluminescent substrate reactive with the ruthenium-labeled secondary anti-cIAP1
antibody. cIAP1 concentration in test samples was determined from a four-parameter
regression analysis of a standard curve based on chemiluminescence signal in the
immunoassay. cIAP1 levels in samples are expressed in this manuscript as a percentage
of the DMSO control.
The relationship between GDC-0917 concentrations and cIAP1 decrease was
characterized by estimation of pharmacodynamic parameters obtained by fitting an
inhibitory Imax model to the data using GraphPad Prism V4.02 (GraphPad Software Inc.,
San Diego, CA) based on the following equation.
I = Iinitial − Imax ×C
IC50 +C
⎛
⎝⎜
⎞
⎠⎟ Equation 4
where I (%) is the cIAP1 level expressed as a percentage of the DMSO control, Iinitial (%)
is the initial value of I, Imax (%) is the maximum value of I, IC50 (µM) is the concentration
of GDC-0917 where I is 50% of Imax, and C (µM) is the GDC-0917 concentration.
HUMAN SIMULATIONS
Simulations of human pharmacokinetics were performed using a one-compartment model
with oral absorption. Parameters used for human simulations were derived from
predicted human hepatic CL from hepatocyte metabolic stability studies, and predicted
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 17
human volume of distribution from simple allometry. The absorption rate constant
(1.72 hr-1) used for human simulations was obtained from modeling the dog oral PK
concentration-time data using a one compartment model with oral absorption.
Bioavailability was assumed to be 1-E where E is the hepatic extraction ratio determined
with hepatocyte data using standard methods (Khojasteh et al., 2011).
The predicted human doses required for 50% (ED50) and 90% (ED90) inhibition of
tumor growth were determined using Equation 3 and pharmacodynamic parameters
derived from MDA-MB-231-X1.1 xenograft efficacy studies. Concentrations of GDC-
0917 driving the response for Equation 3 were simulated using a one compartment model
with oral absorption and the predicted human pharmacokinetic parameters described in
the previous paragraph. Simulations were performed to the last day of dosing in the
xenograft study (Day 21) which was the timepoint at which inhibition of tumor growth
was assessed. We chose not to correct for potential differences in plasma protein binding
between the mouse and human in these simulations as definitive plasma protein binding
studies were not yet available when this analysis was performed. In addition, preliminary
plasma protein binding data suggested that unbound fraction could be approximately
three-fold higher in humans (15% unbound) compared to mouse (5% unbound).
Therefore, not correcting for potential species differences in plasma protein binding
would be considered the most conservative estimate of ED50 and ED90 doses in human.
Additional simulations were performed at the starting dose of 5 mg prior to the
start of the Phase 1 clinical trial. These simulations were performed using Equation 4 and
the associated pharmacodynamics parameters estimated from the in vitro cIAP
experiment with GDC-0917 in PBMCs. GDC-0917 concentrations driving the response
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 18
in Equation 4 for this simulation were generated using a one compartment model with
oral absorption and the predicted human PK parameters for GDC-0917. The purpose of
these simulations was to predict both the anticipated GDC-0917 concentration-time
profile, and the degree of pharmacodynamic marker (cIAP1) modulation that would be
observed in the first cohort of patients. Finally, an assessment of the number of cohorts
required to achieve human ED50 and ED90 doses was performed in order to provide an
early assessment of clinical trial feasibility and duration.
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 19
RESULTS
Pharmacokinetics of GDC-0917 in rat, mouse, dog and monkey.
Pharmacokinetics studies were conducted in rat, mouse, dog, and cynomolgus monkey in
order to characterize GDC-0917 pharmacokinetics in preclinical species. Figure 2 shows
semilog plots of GDC-0917 plasma concentration versus time for mouse, rat, dog and
monkey following intravenous and oral administration. Derived pharmacokinetic
parameters for each species are presented in Table 1. Since GDC-0917 blood-plasma
ratio was approximately 1 (0.9-1.0) in all species tested, in vivo plasma clearance was
considered equivalent to hepatic clearance. GDC-0917 has low plasma clearance (~13%
of hepatic blood flow; Davis and Morris, 1993) in mice, at approximately 12 mL/min/kg.
It has moderate plasma clearance (~50% of hepatic blood flow) in rats and dogs, at 27
and 15.3 mL/min/kg, respectively. In monkeys, GDC-0917 has high plasma clearance of
67.6 mL/min/kg (~ 160% of hepatic blood flow). Terminal half-life (t1/2) ranged from
0.825 hour in monkeys to 6.12 hours in dogs. The volume of distribution at steady state
(Vss) was low to moderate in all species evaluated. With the exception of monkeys where
oral bioavailability was 16.8%, oral bioavailability was high, ranging from 77.0% in rats
to 97.0% in dogs. Renal clearance of GDC-0917 was assessed in dog and monkey and
found to be negligible, accounting for <1% and approximately 2% of the plasma
clearance, respectively.
Predicted Human Clearance, Volume of Distribution, and Half-life of GDC-0917
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 20
Hepatocyte metabolic stability studies were conducted in human, rat and dog hepatocytes
in order to predict hepatic clearance in human, and to verify that an in vitro to in vivo
correlation exists in rat and dog. Table 2 presents predicted hepatic clearance based on
hepatocyte studies for rat, dog and human along with the observed in vivo plasma
clearance in rat and dog. Predicted hepatic clearance for rat and dog were within 2-fold
of the observed in vivo plasma clearance for both species. Human hepatic clearance was
predicted to be moderate at 11.5 mL/min/kg and this value was used for subsequent
human predictions and simulations.
Simple allometric scaling was performed to provide a predicted human volume of
distribution of GDC-0917 (Figure 3). Estimates of volume of distribution from mouse,
rat, dog and monkey (Table 1) were used to predict a human volume of distribution of
6.69 L/kg (Figure 3). The predicted t1/2 in human based upon the predicted volume of
distribution and clearance was 6.70 hours.
MDA-MB-231-X1.1 Breast Cancer Xenograft Efficacy Studies
Anti-tumor efficacy studies were performed in mice xenografted with MDA-MB-231-
X1.1 breast cancer in order to both demonstrate and characterize the anti-tumor activity
of GDC-0917. Figure 4A presents a tumor volume versus time profile for a range of
daily oral doses of GDC-0917 (0.08 to 16.3 mg/kg) given for 21 consecutive days.
GDC-0917 anti-tumor activity was dose dependent with slight tumor regression observed
at doses > 5.43 mg/kg. GDC-0917 was well tolerated with all dose groups experiencing
less than an 11% decrease in mean body weight.
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 21
Xenograft PK-PD Modeling and Simulation
Supporting mouse pharmacokinetic studies were conducted to provide PK information
for PK-PD modeling of xenograft efficacy studies. Estimates of GDC-0917
pharmacokinetic parameters in mouse following oral administration of 0.5, 3, 6, and
20 mg/kg in 15% hydroxypropyl-β-cyclodextrin, 20 mM succinic acid in water are as
follows: ka = 17.6 hr-1, ke = 0.19 hr-1, and V/F = 1.74 L/kg. These PK parameters were
used to simulate GDC-0917 concentrations when fitting tumor volume data from the
xenograft efficacy study since it was not possible to collect blood from xenografted mice
that were on study.
Figure 4B compares observed and predicted tumor volumes following
simultaneous fitting of all MDA-MB-231-X1.1 tumor data to an indirect response model
(Equation 3). In general, the indirect response model was able to adequately characterize
the PK-PD relationship between GDC-0917 plasma concentrations and tumor volume
growth inhibition. Estimated in vivo GDC-0917 pharmacodynamic parameters
characterizing MDA-MB-231-X1.1 tumor growth and inhibition are presented in Table 3.
Human Simulations Determining ED50 and ED90 Doses and Clinical Trial Feasibility
Human simulations were performed in order to predict GDC-0917 doses needed to
achieve the ED50 and ED90 endpoints based on MDA-MB-231-X1.1 xenograft efficacy
studies. These simulations were performed using predicted human pharmacokinetics
described above and the pharmacodynamics parameters presented in Table 3. In
addition, the minimum number of cohorts required to reach the predicted ED50 and ED90
was assessed assuming a starting dose of 5 mg (derived from preclinical safety studies).
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 22
Simulation results and predicted human ED50 and ED90 doses are shown in Table 3 and
Figure 5. ED50 and ED90 doses in human were predicted to be 72 mg and 660 mg,
respectively. The predicted human concentration-time profile of the ED50 and ED90
doses and their associated AUC0-24 are presented in Figure 5B and Table 3, respectively.
Based on a tradition phase 1 oncology clinical trial design (Green, et al., 2003), and
assuming no occurrence of dose limiting toxicities, a doubling of the GDC-0917 dose
with each cohort, and linear pharmacokinetics we anticipate to reach ED50 (72 mg) and
ED90 (660 mg) doses by cohorts 5 (80 mg) and cohort 8 (640 mg) (Figure 5C). Overall
these prospective simulations suggest that ED50 and ED90 are achievable in patients with
acceptable doses and within an acceptable number of cohorts.
cIAP1 Immunoassay in PBMCs
A cIAP1 immunoassay in PBMCs was established and tested in vitro prior to the
initiation of human clinical trials. Figure 6A presents percent decreases of cIAP1 relative
to a DMSO control caused by GDC-0917 added to whole blood at concentrations ranging
from 0.0001 to 10 µM. GDC-0917 induced reduction of cIAP1 levels in PBMCs in a
concentration dependent manner showing greater than 80% inhibition at concentrations
greater than 0.1 µM (56.5 ng/mL) (Figure 6A). Following fitting of the cIAP1 and GDC-
0917 concentration data to an inhibitory Imax model (Equation 4), pharmacodynamic
parameters were estimated and are presented in Figure 6B. Since the cIAP1
immunoassay involves the addition of GDC-0917 to whole blood prior to isolation of
PBMCs, these PD parameters reflect PD response to total blood concentrations rather
than unbound.
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 23
Human Simulation of 5 mg Phase I Starting Dose
Simulations were performed prior to the start of the Phase 1 clinical trial in order to
predict both the anticipated GDC-0917 plasma concentration-time profile and the degree
of cIAP1 modulation following the administration of a 5 mg starting dose to patients.
These simulations used the predicted human PK and the PD parameters from the cIAP1
immunoassay in PBMCs (see Figure 6B). As shown in Figure 7A and Table 4, our
prediction of human pharmacokinetics at the 5 mg dose of GDC-0917 agreed well with
previously unpublished observations from the first cohort. Figure 7B presents a
simulated cIAP1 response over a 24 hour period following administration of the 5 mg
starting dose. Predicted cIAP1 levels at 6 hours postdose was approximately 32% of
predose level (Figure 7B). A mean cIAP1 decrease determined from two patient samples
collected at 6 hours postdose was comparable to the prediction, being 39% of predose
levels (Figure 7B).
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 24
DISCUSSION
The preclinical pharmacokinetics and drug metabolism properties of a drug candidate
often serve as an early assessment of its potential to succeed as a drug. However, the
amount of information available in the early stages of the drug development process is
often limited as preclinical studies are, out of necessity, performed in a staged format.
Furthermore, the ability to translate preclinical findings to human continues to remain a
challenge. GDC-0917 is a second-generation IAP antagonist that offers the advantage of
oral dosing over our prototype molecule, GDC-0152, which was administered
intravenously, once every two weeks. Also, GDC-0917 preclinical toxicology studies
enabled continuous daily dosing allowing for a more flexible choice of dosing regimen in
the clinic. Previously, we described the application of a toxicokinetic-toxicodynamic
(TK-TD) analysis of GDC-0152 rat and dog toxicology studies (Wong et al., 2012a). We
learned from this analysis that dogs were much more sensitive to IAP antagonists than
rats, and confirmed a lower sensitivity in humans using clinical data. The described
analysis de-risked preclinical toxicity concerns around IAP proteins as drug targets
thereby enabling the advancement of a second-generation IAP antagonist. The current
work describes the integration of preclinical modeling and simulation and the “learning-
and-confirm” paradigm in the drug discovery process. We describe how we leveraged
our understanding of GDC-0152 to predict human pharmacokinetics for GDC-0917. In
addition, we show how preclinical modeling and simulation was used in the selection of
GDC-0917 as a clinical candidate.
Based on previous work with GDC-0152 (Flygare et al., 2012; Yue et al., 2013),
we anticipated that the structurally similar GDC-0917 would be eliminated mostly by
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 25
hepatic metabolism and that renal elimination of the compound would be limited. The
low estimated renal clearance of GDC-0917 in dog and monkey (Table 1) is consistent
with these expectations. GDC-0152 human clearance was predicted using both simple
allometry and in vitro-in vivo extrapolation (Flygare et al., 2012). Both methods gave
similar estimates of human clearance (9.6 mL/min/kg [allometry] and 10 mL/min/kg [in
vitro-in vivo extrapolation]) and were successful in predicting observed clearance in
patients (9 mL/min/kg). GDC-0152 volume of distribution was predicted using simple
allometry (1.2 L/kg) and again the prediction matched well with clinical observations (0.6
L/kg) being within an acceptable 2-fold of the observed value (Flygare et al., 2012).
Allometric methods for prediction of human clearance assume a relationship
between clearance and body-weight. For compounds where clearance is due to
elimination dictated by physiological processes such as blood flow or filtration,
allometric methods work reasonably well (Lin, 1998). Therefore, for high clearance
compounds where clearance is blood flow-dependent, and compounds that are cleared
primarily via the urinary route, allometry is anticipated to perform reasonably well.
However, the rationale of using allometric scaling for human PK prediction has been
controversial (Bonate and Howard, 2000) and a known shortcoming of using in vivo
allometric methods is when there are known species differences in metabolism. As such,
we consciously decided to use in vitro-in vivo extrapolation of metabolic stability data
from human hepatocytes to predict human clearance for GDC-0917, providing that a
reasonable in vitro to in vivo correlation could be established in preclinical species.
GDC-0917 showed good preclinical pharmacokinetic properties having low to
moderate plasma clearance in all species with the exception of the monkey which
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 26
exhibited high clearance. The two primary preclinical species used to screen
pharmacokinetics during compound optimization were the rat and dog. Therefore, early
on, we performed metabolic stability studies in hepatocytes from these two preclinical
species and human. The in vitro hepatocyte data were used to estimate hepatic clearance
in rat, dog, and human. A good in vitro to in vivo correlation was observed in rat and dog
since the predicted hepatic clearance and the observed in vivo clearance were within two
fold of each other (Table 2). With a good in vitro to in vivo correlation established using
rat and dog, and no evidence of renal elimination being a major route of GDC-0917
elimination (Table 1), human hepatic clearance was predicted to be moderate (11.5
mL/min/kg) using in vitro-in vivo extrapolation of human hepatocyte data. Similar to
GDC-0152, simple allometry was used to predict human volume of distribution (i.e.
6.69 L/kg) (Figure 3). This prediction included the volume of distribution from all
preclinical species, including the monkey, which was gathered at a later point in the
compound evaluation process. Based upon the predicted human clearance and volume of
distribution, GDC-0917 half-life in humans was estimated to be 6.70 hours.
Pharmacokinetic-pharmacodynamic modeling offers a means to improve the
translation of preclinical studies to humans (Wong et al., 2012b, Wong et al., 2012c,
Yamazaki et al., 2012). Tumor xenografted mice serve as a “workhorse” efficacy model
to assess anti-tumor effects of novel agents, and therefore are widely used in oncology
drug discovery (Johnson et al., 2001, Kelland, 2004, Peterson and Houghton, 2004,
Wong et al., 2012b). Our previous investigation showed that a correlation exists between
anti-tumor activity in xenografted/allografted mice and clinical response providing that
two conditions were fulfilled. These conditions were (1) that species differences in
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 27
mouse and human pharmacokinetics had to be normalized, and (2) that the specific
xenograft model used in the assessment had to be relevant to human disease being treated
(Wong et al., 2012b). At the time of our work with GDC-0917, we arbitrarily set ED50
and ED90 as two efficacy endpoints from which we would evaluate the compound in a
breast cancer model. Fitting of the indirect response model (Equation 3) to tumor volume
data from MDA-MB-231-X1.1 xenografted mice allowed us to define the relationship
between GDC-0917 plasma concentrations and anti-tumor effect. This relationship is
characterized by the pharmacodynamic parameters, Kmax (the maximum value of rate
constant describing the anti-tumor effect of GDC-0917) and KC50 (GDC-0917
concentration where K is 50% of Kmax). Based on comparable estimates of Kmax to the net
growth rate constant (kng), the maximum anti-tumor effect of GDC-0917 in MDA-MB-
231-X1.1 xenografted mice is near tumor stasis. This is evident upon examination of the
tumor volume versus time plot shown in Figure 4A. In particular, at the high dose levels,
increases in dose result in very little additional activity beyond tumor stasis.
Pharmacodynamic parameters estimated from xenografted mice were fixed and
simulations were performed to identify ED50 and ED90 doses in human using predicted
human pharmacokinetics (Figure 5, Table 3). The predicted ED50 (72 mg) and ED90
(660 mg) doses for humans were within an acceptable dose load for an oncology
indication. Assuming a starting dose of 5 mg in patients, ED50 and ED90 could be
reached within 5 to 8 cohorts, respectively, (Figure 5C) suggesting that these endpoints
could be achievable within an acceptable timeframe of approximately 1 to 2 years based
on normal enrollment rates (~ 3 months per patient cohort).
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 28
The final activity following the “learning” phase of our preclinical modeling and
simulation exercise was to “confirm” our predictions upon receiving patient data. Based
on the predicted GDC-0917 pharmacokinetics and pharmacodynamics at the 5 mg
starting dose, we anticipated observing some degree of pharmacodynamic biomarker
(cIAP1) modulation in the first patient cohort. As shown in Figure 7, our simulations
were consistent with previously unpublished clinical observations of pharmacokinetics
(Figure 7A) and pharmacodynamics (Figure 7B) in cancer patients given 5 mg of GDC-
0917. These simulations served to increase confidence in our prospective simulations,
and illustrate the value of confirming predictions upon receipt of human data.
Overall, the preclinical pharmacokinetics of GDC-0917 is favorable. The
compound exhibits low to moderate in vivo clearance in all species, other than the
monkey, and has good oral bioavailability. Based on our translational PK-PD analysis,
ED50 and ED90 are achievable in humans at acceptable doses and within a reasonable
timeframe. Our experience with GDC-0917 shows the importance of leveraging learning
from prototype molecules when advancing second-generation molecules. Finally, our
work with GDC-0917 illustrates how implementation of modeling and simulation
techniques, and the “learn and confirm” paradigm, into the early phases of the drug
development process can add value by integrating existing preclinical data and aid in its
translation to the clinic.
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 29
ACKNOWLEDGEMENTS
The authors thank members of the Drug Metabolism and Pharmacokinetics department
and the In Vivo Studies group for their technical assistance. The authors also thank the
patients involved in the clinical studies.
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 30
AUTHORSHIP CONTRIBUTIONS
Participated in research design: Wong, Gould, Budha, Halladay, Alicke, Erickson,
Flygare, Portera, Tolcher, Infante, Mamounas, Flygare, Fairbrother.
Conducted experiments: Wong, La, Darbornne, Kadel, Alicke, Erickson, Tolcher.
Contributed new reagents or analytical tools: Darbonne, Flygare.
Performed data analysis: Wong, Gould, Budha, Darbonne, Kadel, La, Halladay, Alicke,
Tolcher, Infante, Flygare.
Wrote or contributed to the writing of the manuscript: Wong, Darbonne, Alicke, Budha,
Portera, Tolcher, Infante, Hop, Fairbrother.
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 31
REFERENCES
Bonate PL and Howard D (2000) Prospective allometric scaling: does the emperor have
clothes? J Clin Pharmacol 40: 335-340.
Boxenbaum H (1984) Interspecies pharmacokinetic scaling and the evolutionary-
comparative paradigm. Drug Metab Rev 15:1071–1121.
Chien JY, Friedrich S, Heathman MA, de Alwis DP and Sinha V. (2005)
Pharmacokinetics/Pharmacodynamics and the stages of drug development: role of
modeling and simulation. AAPS J. 7:E544-559.
Davies B, and Morris T (1993) Physiological Parameters in Laboratory Animals and
Humans Pharm Res 10: 1093-1095.
Flygare, J.A., and Fairbrother, W.J. (2010). Small-molecule pan-IAP antagonists: a patent
review. Expert Opin. Ther. Pat. 20,251-267.
Flygare JA, Beresini M, Budha N, Chan H, Chan IT, Cheeti S, Cohen F, Deshayes K,
Doerner K, Eckhardt SG, Elliott LO, Feng B, Franklin MC, Frankovitz Reisner S,
Gazzard L, Halladay J, Hymowitz SG, La H, LoRusso P, Maurer B, Murray L, Plise E,
Quan C, Stephan JP, Shin YG, Tom J, Tsui V, Um J, Varfolomeev E, Vucic D, Wagner
AJ, Wallweber HJA, Wang L, Ware J, Wen Z, Wong H, Wong JM, Wong M, Wong S,
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 32
Yu R, Zobel K, and Fairbrother WJ. (2012) The Discovery of a Potent Small-Molecule
Antagonist of Inhibitor of Apoptosis (IAP) Proteins and Clinical Candidate for the
Treatment of Cancer (GDC-0152). J Med Chem, 55:4101-4113, DOI:
10.1021/jm300060k.
Gilbaldi M, Perrier D (1982) Pharmacokinetics, 2nd ed. Marcel Dekker Inc., New York.
Green S, Benedetti J, and Crowley J (2003) Clinical Trials in Oncology, Second Edition,
Chapman & Hall/ CRC, Baton Rouge, Florida.
Johnson JI, Decker S, Zaharevitz D, Rubinstein LV, Venditti JM, Schepartz S,
Kalyandrug S, Christian M, Arbuck S, Hollingshead M and Sausville EA (2001)
Relationships between drug activity in NCI preclinical in vitro and in vivo models and
early clinical trials. Br J Cancer 84:1424-31
Kelland LR (2004) “Of mice and men”: values and liabilities of the athymic nude mouse
model in anticancer drug development. European Journal of Cancer 40:827-836.
Khojasteh SC, Wong H, and Hop CECA (2011) Drug Metabolism and Pharmacokinetics
Quick Guide, Springer, New York.
Lin JH (1995) Species similarities and differences in pharmacokinetics. Drug Metab
Dispos 23: 1008-1021.
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 33
Liu L, Di Paolo J, Barbosa J, Rong H, Reif K and Wong H. (2011) Anti-Arthritis Effect
of a Novel Bruton’s Tyrosine Kinase Inhibitor in Rat Collagen-Induced Arthritis and
Mechanism-Based Pharmacokinetic/Pharmacodynamic Modeling: Relationships between
Inhibition of BTK Phosphorylation and Efficacy. J Pharmacol and Exp Ther, 338: 154-
163. doi: 10.1124/jpet.111.181545.
Obach RS, Baxter JG, Liston TE, Silber BM, Jones BC, MacIntyre F, Rance DJ, Wastall
P (1997) The prediction of human pharmacokinetic parameters from preclinical and in
vitro metabolism data. J Pharmacol Exp Ther 283: 46-58.
Peterson JK and Houghton PJ. (2004) Integrating pharmacology and in vivo cancer
models in preclinical and clinical drug development. European Journal of Cancer
40:837-44
Salphati L, Wong H, Belvin M, Edgar KA, Prior WW, Sampath D and Wallin JJ. (2010)
Pharmacokinetic-Pharmacodynamic Modeling of Tumor Growth Inhibition and
Biomarker Modulation by the Novel PI3K Inhibitor 2-(1H-Indazol-4-yl)-6-(4-
methanesulfonyl-piperazin-1-ylmethyl)-4-morpholin-4-yl-thieno[3,2-d]pyrimidine
(GDC-0941) Drug Metab and Disp, 38: 1436-1442. doi:10.1124/dmd.110.032912.
Sheiner LB (1997) Learning versus confirming in clinical drug development. Clin
Pharmacol Ther. 61:275-291.
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 34
Varfolomeev, E., and Vucic, D. (2011). Inhibitor of apoptosis proteins: fascinating
biology leads to attractive tumor therapeutic targets. Future Oncol. 75, 633-648.
Vucic, D., and Fairbrother, W.J. (2007). The inhibitor of apoptosis proteins as
therapeutic targets in cancer. Clin. Cancer Res. 13, 5995-6000.
Wong H, Budha N, West K, Blackwood B, Ware JA, Yu R, Darbonne WC, Gould SE,
Steigerwalt R, Erickson R, Hop CECA, LoRusso P, Eckhardt SG, Wagner A, Chan IT,
Mamounas M, Flygare J and Fairbrother WJ. (2012a) Dogs are More Sensitive to
Antagonists of Inhibitor of Apoptosis (IAP) Proteins than Rats and Humans: A
Translational Toxicokinetic/Toxicodynamic Analysis. Tox Sci. 130: 205-213. doi:
10.1093/toxsci/kfs235.
Wong H, Choo EF, Alicke B, Ding X, La H, McNamara E, Theil FP, Tibbitts J, Friedman
LS, Hop CECA, and Gould SE. (2012b) Anti-tumor activity of targeted and cytotoxic
agents in murine subcutaneous tumor models correlates with clinical response Clin
Cancer Res, 18: 3846-3855. doi:10.1158/1078-0432.CCR-12-0738.
Wong H, Vernillet L, Peterson A, Ware JA, Lee L, Martini J-F, Yu P, Li C, Del Rosario
G, Choo EF, Hoeflich KP, Shi Y, Aftab BT, Aoyama R, Lam ST, Belvin M, and Prescott
J. (2012c) Bridging the Gap between Preclinical and Clinical Studies Using PK-PD
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 35
Modeling: An Analysis of GDC-0973, a MEK Inhibitor. Clin Cancer Res, 18: 3090-
3099. doi:10.1158/1078-0432.CCR-12-0445.
Yamazaki S, Vicini P, Shen Z, Zou HY, Lee J, Li Q, Christensen JG, Smith BJ and
Shetty B. (2012) Pharmacokinetic/pharmacodynamic modeling of crizotinib for
anaplastic lymphoma kinase inhibition and antitumor efficacy in human tumor xenograft
mouse models. J Pharmacol Exp Ther, 340:549-57.
Yue Q, Mulder T, Rudewicz P, Solon E, Budha N, Ware JA, Lyssikatos J, Hop CECA,
Wong H and Khojasteh SC. (2013) Evaluation of Metabolism and Disposition of GDC-
0152 in Rats Using 14C Labeling Strategy at Two Different Positions: A Novel
Formation of Hippuric Acid from 4-Phenyl-5-amino-1,2,3-thiadiazole. Drug Metab and
Disp, 41:508–517.
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 36
FOOTNOTES
Send reprint requests to: Dr. Harvey Wong, Genentech, Inc., MS# 412a, 1 DNA Way,
South San Francisco, CA 94080. E-mail: [email protected].
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 37
FIGURE LEGENDS
Figure 1 – Structure of GDC-0152 and GDC-0917
Figure 2 – Plasma Concentration-Time Profile of GDC-0917 in (A) Mouse, (B) Rat, (C)
Dog and (D) Monkey.
Figure 3 – Predicted Human Volume of Distribution Using Allometry
Figure 4 – (A) Anti-tumor Effect of GDC-0917 on MDA-MB-231-X1.1 Xenografts.
The indicated doses have units of mg/kg. (B) Observed versus Predicted
Tumor Volumes Following Fitting of Tumor Volume Data to a PK-PD
model. The solid line is the line of unity and the dashed line is the regression
line. The slope of the regression line and correlation coefficient ( r ) are
presented in the plot.
Figure 5 – (A) Predicted Human ED50 (72 mg) and ED90 (660 mg) Using MDA-MB-
231.X1-1 Xenografts. (B) Predicted Human Pharmacokinetics at 72 and
660 mg. (C) Simulated Human Phase I Clinical Trial.
Figure 6 – (A) Plot Showing cIAP1 Response to a Range of GDC-0917 Concentrations
(0.0001 to 10 µM) in PBMCs. (B) Fit of Inhibitory Imax model (Dashed Line)
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
DMD #53926 38
to cIAP1 and GDC-0917 Concentrations. Pharmacodynamic Parameter
Estimates are Presented on the Plot.
Figure 7 – (A) Observed and Predicted GDC-0917 Concentration-Time Profile at a Dose
of 5 mg in Human Patients. (B) Observed and Predicted cIAP1 Modulation in
Patients Given a 5 mg Starting Dose. Observed mean cIAP decrease from
patients at 6 hours postdose was 39% of predose levels. The simulated cIAP
decrease at 6 hours postdose was 32% of predose levels.
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
39
Table 1 Pharmacokinetics (Mean ± SD) of GDC-0917 in the Mouse, Rat, Dog, and Monkey
Parameters Mouse Rat Dog Monkey
IV administration
No. of animals 27
(3/timepoint) 3 3 3
Sex Female Male Male Male
Dose (mg/kg) 1 1 1 1
CL (mL/min/kg) 12.0 27.0 ± 1.91 15.3 ± 0.643 67.6 ± 6.17
t1/2 (hr) 1.47 0.900 ± 0.0862 6.12 ± 4.22 0.825 ± 0.0634
MRT (hr) 1.76 1.12 ± 0.0945 4.44 ± 2.38 1.05 ± 0.112
Vss (L/kg) 1.27 1.81 ± 0.0529 4.09 ± 2.27 4.25 ± 0.352
Oral administration
No. of animals 27 (3/timepoint)
3 3 3
Sex Female Male Male Male
Dose (mg/kg) 3 5a 1 2
Cmax (ng/mL) 884 545 ± 261 254 ± 129 33.6 ± 10.3
tmax (hr) 0.500 2.00 ± 1.73 1.00 ± 0.00 1.33 ± 0.577
AUCinf (ng • hr/mL) 3870 2390 ± 976 1060 ± 606 83.2 ± 12.2
F (%) 93.5 77.0 ± 31.4 97.0 ± 55.6 16.8 ± 2.45
Renal clearance (mL/min/kg)
NA NA 0.0897 ± 0.0980b 1.44 ± 1.72c
AUCinf = area under the concentration–time curve from zero to infinity; CL = plasma clearance; Cmax = highest observed plasma concentration; F = bioavailability; IV = intravenous; MRT = mean residence time; NA = not available; t1/2 = half-life; tmax = time at which Cmax occurred; Vss = volume of distribution at steady state. a free-base equivalent (trifluoroacetate salt of GDC-0917 was administered) b Renal clearance was assessed for PO group only. c Renal clearance was assessed for IV group only.
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
40
Table 2. In Vitro Predicted Hepatic Clearance and Observed In Vivo Clearance of GDC-0917.
Species
Predicted CLhepatic (Hepatocytes) (mL/min/kg)
Observed CL (mL/min/kg)
Rat
18.3
27.0
Dog 11.9
15.3
Human
11.5 _
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
41
Table 3. Pharmacodynamic Parameter Estimates and Predicted Human Doses Based on MDA-MB-231-X1 Breast Cancer Xenograft Efficacy Studies with GDC-0917.
Parameter
Parameter Estimatea
Kmax (hr-1)
3.28 × 10-3 (6.24)
KC50 (µM) 0.114 (20.3)
kng (hr-1)
3.15 × 10-3 (1.84)
Predicted Human Doses and Exposures
Predicted Human ED50 (mg)
72 Predicted ED50 steady-state AUC0-24 (µM×hr) 1.36
Predicted ED50 steady-state AUC0-24 (ng/mL×hr) 7.70 × 102
Predicted Human ED90 (mg) 660 Predicted ED90 steady-state AUC0-24 (µM×hr) 12.5
Predicted ED90 steady-state AUC0-24 (ng/mL×hr) 7.06 × 103
a Fitted parameter presented as estimate followed by the coefficient of variation in parentheses.
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
42
Table 4. Pharmacokinetics of GDC-0917 in Cancer Patients Given a 5 mg Starting Dose
Parameters
Cmax (ng/mL)
tmax (hr)a
AUCinf (ng/mL×hr)
t1/2 (hr)
Observed
4.57 ± 1.67
2 – 6
37.1 ± 25.4
3.98 ± 1.56
Predicted 4.37 2.4 46.4 6.70
a observed tmax is presented as a range
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from
This article has not been copyedited and formatted. The final version may differ from this version.DMD Fast Forward. Published on September 16, 2013 as DOI: 10.1124/dmd.113.053926
at ASPE
T Journals on N
ovember 2, 2021
dmd.aspetjournals.org
Dow
nloaded from