Pharmacodynamics of 2-{4-[(1E)-1-(Hydroxyimino)-2,3-dihydro- 1H-inden-5-yl]-3-(pyridine-4-yl)-1H-pyrazol-1-yl}ethan-1-ol (GDC-0879), a Potent and Selective B-Raf Kinase Inhibitor: Understanding Relationships between Systemic Concentrations, Phosphorylated Mitogen-Activated Protein Kinase Kinase 1 Inhibition, and Efficacy

Harvey Wong, Marcia Belvin, Sylvia Herter, Klaus P. Hoeflich, Lesley J. Murray, Leo Wong, and Edna F. Choo
Departments of Drug Metabolism and Pharmacokinetics (H.W., E.F.C.), Translational Oncology (K.P.H., L.J.M., L.W.), and Cell Signaling Pathways (M.B., S.H.), Genentech Inc., South San Francisco, California
Received October 31, 2008; accepted January 14, 2009

ABSTRACT

The Raf/mitogen-activated protein kinase kinase (MEK)/extra- cellular signal-regulated kinase signaling pathway is involved in cellular responses relevant to tumorigenesis, including cell proliferation, invasion, survival, and angiogenesis. 2-{4-[(1E)-1- (Hydroxyimino)-2,3-dihydro-1H-inden-5-yl]-3-(pyridine-4-yl)- 1H-pyrazol-1-yl}ethan-1-ol (GDC-0879) is a novel, potent, and selective B-Raf inhibitor. The objective of this study was to characterize the relationship between GDC-0879 plasma con- centrations and tumor growth inhibition in A375 melanoma and Colo205 colon cancer xenografts and to understand the phar- macodynamic (PD) marker response requirements [phosphor- ylated (p)MEK1 inhibition] associated with tumor growth inhibi- tion in A375 xenografts. Estimates of GDC-0879 plasma concentrations required for tumor stasis obtained from fitting tumor data to indirect response models were comparable, at 4.48 and 3.27 tiM for A375 and Colo205 xenografts, respec-
tively. This was consistent with comparable in vitro potency of GDC-0879 in both cell lines. The relationship between GDC-0879 plasma concentrations and pMEK1 inhibition in the tumor was characterized in A375 xenografts after oral doses of 35, 50, and 100 mg/kg. Fitting pMEK1 inhibition to an indirect response model provided an IC50 estimate of 3.06 tiM. pMEK1 inhibition was further linked to A375 tumor volume data from nine different GDC-0879 dosing regimens using an integrated pharmacoki- netic-PD model. A simulated PD marker response curve plot of K (rate constant describing tumor growth inhibition) versus pMEK1 inhibition generated using pharmacodynamic parameters esti- mated from this model, showed a steep pMEK1 inhibition- response curve consistent with an estimated Hill coefficient of ti8. A threshold of ti40% pMEK1 inhibition is required for tumor growth inhibition, and a minimum of ti60% pMEK1 inhibition is required for stasis in A375 xenografts treated with GDC-0879.

The Raf/MEK/ERK signaling pathway is involved in cellu- lar responses relevant to tumorigenesis, including cell prolif- eration, invasion, survival, and angiogenesis (Sebolt-Leopold and Herrera, 2004; Gollob et al., 2006; Schreck and Rapp,

2006; Zebisch and Troppmair, 2006; Madhunapantula and Robertson, 2008). Currently, three Raf kinase isoforms have been identified and are referred to as A-Raf, B-Raf, and C-Raf (also known as Raf-1) (Madhunapantula and Robertson, 2008). Frequent activating mutations in B-Raf have been

The work in this article was presented at the 15th North American Regional ISSX Meeting; 2008 October; San Diego, CA; and the IXth World Conference on Clinical Pharmacology and Therapeutics; 2008 July; Quebec, Canada.
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
doi:10.1124/jpet.108.148189.
observed in several tumor types, including malignant mela- noma (Davies et al., 2002) and colorectal carcinoma (Yuen et al., 2002). The majority of these mutations are in exon 15, which results in a V600E amino acid substitution, leading to constitutive kinase activation (Mercer and Pritchard, 2003).

ABBREVIATIONS: MEK, mitogen-activated protein kinase kinase; ERK, extracellular signal-regulated kinase; PK, pharmacokinetic; PD, pharma- codynamic; GDC-0879, 2-{4-[(1E)-1-(hydroxyimino)-2,3-dihydro-1H-inden-5-yl]-3-(pyridine-4-yl)-1H-pyrazol-1-yl}ethan-1-ol; MCT, 0.5% methyl- cellulose/0.2% Tween 80; p, phosphorylated; TV, tumor volume; CV, coefficient of variation; %I, percentage of inhibition of pMEK1; S6K1, S6 kinase 1.
360

The association of the Raf/MEK/ERK pathway with human cancers has made it an attractive pathway to target using small molecule inhibitors. Sorafenib (Nexavar, BAY-43-9006) has been reported to inhibit the Raf/MEK/ERK signaling pathway. However, sorafenib is a multikinase inhibitor and its clinical antitumor activity is probably a result of its ac- tivity against proangiogenic receptor tyrosine kinases, such as vascular endothelial growth factor receptor and platelet- derived growth factor receptor (Li et al., 2007). Other re- ported Raf inhibitors include PLX4032, ZM336372, AZ628, AAL881, LBT613, SB-590885, and RAF-265 (Khazak et al., 2007; Li et al., 2007). To date, little or no detailed analysis of the pharmacokinetic-pharmacodynamic (PK-PD) relation- ships between systemic exposure of a Raf inhibitor and its effects on downstream markers of pathway modulation have been published. Reported investigations of PK-PD relation- ships of Raf inhibitors have been largely limited to simple demonstrations of dose-dependent inhibition of Raf/MEK/
ERK pathway in xenograft models (Wilhelm et al., 2004).
Preclinical PK-PD modeling can play an important role in the drug discovery and development process by providing an integrated understanding of relationships between com- pound plasma concentrations, PD marker response, and ef- ficacy. Throughout the characterization of a compound, these data may exist as separate entities and understanding rela- tionships between 1) concentration and efficacy, 2) concen- tration and PD marker response, and 3) PD marker response and efficacy are often difficult because of biological delays between each measured response. PK-PD modeling provides estimates of pharmacodynamic parameters that describe the specific relationships of interest.
GDC-0879 (Fig. 1) is a novel potent, selective B-Raf inhib- itor in various in vitro and cell-based assays with an IC50 estimate of 0.13 nM against purified B-Raf V600E enzyme and a cellular pERK IC50 of 63 nM in the MALME-3M cell line (Hansen et al., 2008). When screened against a panel of 140 kinases at 1 tiM, GDC-0879 showed expected activity only against C-Raf (K. P. Hoeflich, unpublished data). The objective of the current study was to characterize the in vitro and in vivo potency, and the pharmacodynamic properties of GDC-0879 in Colo205 and A375 tumor cell lines, both of which carry the V600E mutation. The relationship among systemic exposure of GDC-0879, phosphorylated MEK1 (pMEK1) inhibition, and tumor growth inhibition was also explored in more detail in the A375 model.

Materials and Methods
GDC-0879 was provided by Array BioPharma (Boulder, CO). Sol- vents used for analysis were of analytical or high-performance liquid chromatography grade (Thermo Fisher Scientific, Waltham, MA). All other reagents or material used in this study were purchased from

Sigma-Aldrich (St. Louis, MO) unless otherwise stated. The Genen- tech Institutional Animal Care and Use Committee approved all procedures in animals.

Pharmacokinetic Studies in Mice
Female athymic nu/nu mice (weighing 25–28 g) (Charles River Lab- oratories, Hollister, CA) were administered oral doses of 15, 25, 50, 100, and 200 mg/kg GDC-0879 in 0.5% methylcellulose/0.2% Tween 80 (MCT). Blood samples (ti1 ml) were collected at 0.5, 1, 2, 4, 8, and 24 h after dose via cardiac puncture (terminal collection) into tubes contain- ing K2EDTA anticoagulant. Immediately upon collection, the blood was mixed with K2EDTA and stored on ice. Within 30 min, blood samples were centrifuged at approximately 1000 to 1500g for 5 min at 4°C, and plasma was harvested. The plasma samples were stored at ti80°C until analysis. Concentrations of GDC-0879 were determined by liquid chro- matography-tandem mass spectrometry. The dynamic range of the assay was 0.005 to 10 tiM.

In Vitro Studies in A375and Colo205 Cell Lines
GDC-0879 in vitro IC50 estimates for pMEK inhibition were deter- mined using A375 and Colo205 cells. In brief, A375 or Colo205 cells were incubated with a range of GDC-0879 concentrations (from 0.5 nM to 6.75 tiM) for 25 min. Cells were lysed, and the lysates were subjected to centrifugation at 16,100g for 30 min, and the level of total protein was determined using the Bradford method (Bradford, 1976). Enzyme- linked immunosorbent assay kits were used to determine pMEK1 and total MEK1 protein levels in a 96-well format (Tago Biosource Interna- tional, Camarillo, CA). Samples were analyzed in duplicate at 20 tig of protein per well according to the protocol of the supplier. The optical densities obtained at 450 nm were converted to units per milliliter (for pMEK1) or nanograms per milliliter (for total MEK1) using a standard curve determined with recombinant pMEK1 or MEK1. The pMEK1/
total MEK1 ratios were then calculated as units per nanogram. The IC50 estimates for pMEK1 inhibition were estimated by nonlinear re- gression using GraphPad Prism version 4.02 (GraphPad Software Inc., San Diego, CA).

Xenograft (A375 and Colo205) Efficacy Studies
Ten million human melanoma A375 cells or 5 million human colorectal carcinoma Colo205 cells were resuspended in phosphate- buffered saline, mixed 1:1 (v/v) with Matrigel (BD Biosciences, San Jose, CA), and implanted subcutaneously into the right flank of naive female athymic nu/nu mice. Tumors were monitored until they reached a mean volume of 200 to 300 mm3.
Tumor sizes and body weights were recorded twice weekly, and the mice were regularly observed over the course of the study. Mice were promptly euthanized if their tumor volume exceeded 2000 mm3 or if their body weight dropped by more than 20% of the starting weight. Ten mice were randomly assigned to each group based on mean tumor volume. Mean tumor volume across all six groups was 200 mm3 at the start of dosing. 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 version 11.2 (Microsoft, Red- mond, WA) to calculate tumor volume (TV): TV (cubic millimeters) ti (length ti width2) ti 0.5.
All test materials were administered by oral gavage for 19 to 22 days. Animals in vehicle groups received 100 til of MCT. Treatment groups (n ti 10 mice/group) received oral doses of GDC-0879 in 100 til of MCT. The treatment groups for A375 xenograft efficacy experi- ments included in this study were as follows: experiment 1: vehicle, 15, 25, and 50 mg/kg once daily; experiment 2: vehicle, 50, 100, and 200 mg/kg once daily and 50 mg/kg b.i.d.; experiment 3: vehicle and 100 mg/kg once daily, 100 mg/kg every other day, 100 mg/kg every

Fig. 1. Chemical structure of GDC-0879. third day, and 100 mg/kg every week.

Treatment groups for Colo205 xenograft efficacy experiments in- cluded in this study were as follows: experiment 4: vehicle, 25, 35, and 50 mg/kg once daily and 35 mg/kg b.i.d.; and experiment 5, vehicle, 50 and 100 mg/kg once daily and 50 mg/kg b.i.d.
Mean tumor volume and S.E.M. were calculated using JMP soft- ware, version 5.1.2 (SAS Institute, Cary, NC). Statistical comparison of tumor volume of treatment groups to vehicle groups was per- formed using the Dunnett’s t test (Dunnett, 1955) with JMP software (SAS Institute, Cary, NC). End of study tumor volumes of each treatment group was compared specifically with end of study tumor volumes for the corresponding vehicle group.

pMEK1 Inhibition Studies in A375 Xenografts
Female athymic nu/nu mice were implanted with tumor cells as described above for the A375 xenograft efficacy experiments. Mean tumor volume across all groups was 215 mm3. Mice were randomly assigned to four treatment groups (vehicle, 35, 50, and 100 mg/kg) based on mean tumor volume. Animals received a single oral dose of vehicle (MCT; n ti 4) or 35 (n ti 20), 50 (n ti 20), or 100 mg/kg (n ti 20) of GDC-0879 in 100 til of MCT. Terminal blood and tumor samples (blood via cardiac puncture under general anesthesia or retro-orbital collection, and tumors) were collected from the 35, 50, and 100 mg/kg treatments groups at 1, 2, 4, 8, and 12 h after dose (n ti 4/time point). Samples were collected only at the 8-h time point for the vehicle group (n ti 4). Plasma was collected, stored, and

TV (cubic millimeters) is defined as the tumor volume, t (hours) is time, kng (hoursti1) is the net growth rate constant, K (hoursti 1) is the rate constant describing the tumor reductive effects of GDC-0879, Kmax (hoursti 1) is the maximal value of K, C (micromolar) is the concentration of GDC-0879, n is the Hill coefficient, and KC50 (mi- cromolar) is the GDC-0879 concentration where K is 50% Kmax. Concentrations of GDC-0879 in mice were simulated based upon the pharmacokinetic parameters obtained from the pharmacokinetics studies. Mean tumor volumes for each dose group were used for fitting. All dose groups from the same tumor type were fit simulta- neously. Pharmacodynamic parameters are presented as the esti- mate followed by the coefficient of variation (CV) in parentheses.
The dose required for 50% inhibition of tumor growth (ED50) for A375 and Colo205 xenograft studies was calculated by fixing phar- macodynamic parameter estimates and simulating doses required for 50% inhibition of the vehicle group’s tumor growth at the mean end of study day. Mean end of study day was used because experi- ments used for PK-PD modeling had differing durations. Concentra- tion required for tumor stasis (Cstasis) was calculated as the concen- tration where the K is equal to kng.
pMEK1 Inhibition Studies in A375 Xenografts. The PK-PD relationship of GDC-0879 plasma concentrations to pMEK1 inhibi- tion was characterized using an indirect response model (Mager et al., 2003) in which GDC-0879 concentrations inhibit the formation of pMEK1 as described by the following equation:

GDC-0879 concentrations were assessed as described above for the pharmacokinetic study. Excised tumor tissues were flash-frozen and stored at ti80°C until analysis.
d(pMEK1)
dt

ti kin

1 ti
C IC50tiC

ti kout(pMEK1)

(2)

The frozen tumors were pulverized on dry ice using a small Bess- man tissue pulverizer (Spectrum Laboratories, Inc., Rancho Dominguez, CA). Cell-free protein extracts were prepared at 4°C for analysis of the pMEK1 and total MEK1 status. The lysis buffer (10 ml) contained 1ti cell lysis buffer (Cell Signaling Technology Inc., Danvers, MA), 1 mM phenylmethylsulfonyl fluoride (Sigma-Aldrich, St. Louis, MO), and one tablet of Complete Mini protease inhibitor cocktail (Roche Diagnostics, Indianapolis, IN). Tissue lysates were subjected to centrifugation at 16,100g for 30 min, and the level of total protein was determined using the Bradford method (Bradford, 1976). ELISA kits were used to determine pMEK1 and total MEK1 protein levels in 96-well format (Tago Biosource International). Sam- ples were analyzed in duplicate at 75 tig of protein per well according to the protocol of the supplier. The optical densities obtained at 450 nm were converted to units per milliliter (for pMEK1) or nanograms per milliliter (for total MEK1) using a standard curve determined with recombinant pMEK1 or MEK1. The pMEK1/total MEK1 ratios were then calculated as units per nanogram.

PK-PD Modeling
PK-PD modeling was performed using SAAM II (Saam Institute, University of Washington, Seattle, WA).
Pharmacokinetic Studies. Plasma concentration-time data from female nu/nu mice were fitted to a one compartment model with oral absorption. The pharmacokinetics of GDC-0879 seemed linear over the dose range tested. Mean estimates of the absorption rate constant (ka),
pMEK1 (units per nanogram) is defined as the pMEK1/total MEK1 ratio (see “pMEK1 Inhibition Studies in A375 Xenografts” under Materials and Methods), t (hours) is the time, kin (units per nanogram per hour) is the formation rate of pMEK1, IC50 (micro- molar) is the GDC-0879 concentration where there is 50% in- hibition of pMEK1, and kout (hoursti 1) is the rate constant describ- ing the loss of pMEK1. At homeostasis, kin ti kout (pMEK1); therefore, kout was replaced by kin/(pMEK1)initial where (pMEK1)initial is the control/predose value of pMEK1. In addition to an indirect re- sponse model described above, the pMEK1 data were fit to an effect compartment model (Mager et al., 2003; data not shown). The re- sulting IC50 estimate was virtually identical to the values estimated from indirect response model. However, fitting statistics, such as the Akaike information criterion and the Schwarz-Bayesian information criterion, had a smaller value for the indirect response model, sug- gesting a better model fit. Pharmacodynamic parameters are pre- sented as the estimate followed by the CV in parentheses.
Integrated PK-PD Efficacy Model in A375 Xenografts. To un- derstand the relationship between GDC-0879 plasma concentrations, pMEK1 inhibition, and tumor growth inhibition, the PK-PD model used to characterize the relationship between GDC-0879 concentrations and pMEK1 inhibition described above was combined with the model used to fit the xenograft efficacy studies. In brief, the PK-PD model describ- ing pMEK inhibition was used to simulate pMEK1 inhibition for all A375 xenograft dose groups. The following equation was used in the fitting process to relate pMEK1 inhibition to tumor volume:

the elimination rate constant (ke), and the apparent volume of distri- bution from all dose groups were used to simulate GDC-0879 concen- trations in the modeling of the xenograft efficacy studies.
d(TV)
dt

ti kng(TV) ti K(TV)

(3)

Xenograft (A375 and Colo205) Efficacy Studies. Xenograft efficacy studies were fitted to an indirect response model (Mager et
where

al., 2003) as described by the following differential equation.
d(TV)
ti kng(TV) ti K(TV)
dt
where

(1)
Kmax ti ti %Iti n
K ti
Kti %Iti n50 ti ti %Iti n
%I is defined as the percentage of inhibition of pMEK1 and was calculated as follows:

K ti
Kmax ti Cn KCn50 ti Cn

%I ti
(pMEK1)initial ti (pMEK1)
(pMEK1)initial

ti 100%

K(%I)50 is defined as the %I where K is 50% Kmax. Concentrations of GDC-0879 in mice were simulated based upon the pharmacokinetic parameters obtained from the pharmacokinetic studies. Mean tumor volumes for each dose group were used for fitting. All dose groups from A375 xenograft experiments were fit simultaneously. Pharma- codynamic parameters are presented as the estimate followed by the CV in parentheses. pMEK1 inhibition required for tumor stasis was calculated using the estimated pharmacodynamic parameters as the
%pMEK1 inhibition where the K is equal to kng.
Results
GDC-0879 had comparable in vitro potency in A375 mela- noma and Colo205 colorectal carcinoma cell lines, both of which are V600E B-Raf mutant, with IC50 estimates of pMEK1 inhibition of 59 and 29 nM, respectively.
The pharmacokinetic parameters of GDC-0879 after oral administration of 15, 25, 50, 100, and 200 mg/kg in MCT in mice were estimated as follows: ka ti 8.20 hti1, ke ti 0.59 hti1, and apparent volume of distribution ti 6.19 l/kg. These esti- mated parameters were used to simulate GDC-0879 plasma concentrations when fitting tumor volume data from xeno- graft efficacy studies because it was not possible to collect blood from xenograft mice that were on study.
Tumor growth inhibition curves after a range of oral doses and schedules of GDC-0879 in A375 xenograft tumor-bearing

mice are shown in Fig. 2, A to C. All of the doses tested resulted in a statistically significant tumor growth inhibition (p ti 0.05) compared with their respective vehicle control groups, with the exception of the 15 and 25 mg/kg once daily treatment groups from experiment 1 (Fig. 2A) and the 100 mg/kg group every third day and 100 mg/kg once a week treatment groups from experiment 3 (Fig. 2C). Overall, GDC- 0879 showed dose-dependent inhibition of tumor growth in A375 xenografts. The effect of GDC-0879 on tumor growth after daily oral dosing did not seem to increase appreciably compared with the 100 mg/kg once daily dose to the 200 mg/kg once daily dose, suggesting saturation of effect (Fig. 2B). Figure 2D compares observed and predicted tumor vol- umes after simultaneous fitting of all A375 tumor data to a simple indirect response model (eq. 1). In general, a simple indirect response model was able to adequately characterize the PK-PD relationship between GDC-0879 plasma con- centrations and tumor volume growth inhibition. Esti- mated in vivo GDC-0879 pharmacodynamic parameters characterizing A375 tumor growth and inhibition are pre- sented in Table 1.
Tumor growth inhibition curves following a range of oral doses of GDC-0879 in Colo205 xenograft mice are shown in Fig. 3, A and B. All of the doses tested resulted in a statisti-

Fig. 2. A375 melanoma xenograft experiments 1 (A), 2 (B), and 3 (C). Tumor volumes are presented as mean ti S.E. Observed versus predicted A375 xenograft mean tumor volumes after fitting to an indirect response model (D).

TABLE 1
Summary of pharmacodynamic parameters estimated from A375 and Colo205 xenograft studies

eters characterizing Colo205 tumor growth and inhibition are presented in Table 1. Overall, the in vivo potency of GDC-0879 seemed to be similar in A375 and Colo205 xeno-

Parametera

kng (hti 1) Kmax (hti 1) KC50 (tiM)
A375 Melanoma Xenografts

0.0035 (1.1) 0.030 (12.0) 8.80 (16.0)
Colo205 Colorectal Carcinoma Xenografts
0.0039 (1.8) 0.028 (26.1)
19.98 (38.9)
grafts with comparable EC50, ED50, and Cstasis values. GDC-0879 plasma concentrations and corresponding
pMEK1/ MEK1 ratios for single oral doses of 35, 50, and 100 mg/kg were determined and are presented in Fig. 4, A and B,

n 3.02 (29.1) 1.00 (fixed)
ED50 (mg/kg)b 28 32
Cstasis (tiM) 4.48 3.27
aFitted parameters are expressed as estimate followed by the coefficient of variation in parentheses.
bED50 doses were estimated by simulation as described under Materials and Methods.

cally significant tumor growth inhibition (p ti 0.05) compared with their respective vehicle control groups. Similar to obser- vations with A375 xenografts, GDC-879 showed dose-depen- dent inhibition of tumor growth in Colo205 xenografts after oral administration over a broad dose range. As seen in Fig. 3C, a simple indirect response model (eq. 1) was able to adequately characterize the PK-PD relationship between GDC-0879 plasma concentrations and tumor growth inhibi- tion. Estimated in vivo GDC-0879 pharmacodynamic param-
respectively. Administration of GDC-0879 at all of the tested doses resulted in effective inhibition of the Raf signaling pathway, as measured by pMEK1 levels. The inhibition was profound at 1 and 2 h after dose, especially at the 50 and 100 mg/kg dose levels. Moderate pMEK1 inhibition was observed at 4 to 8 h after dose and the pMEK1 signal was recovered within 12 h. The relationship of GDC-0879 plasma concen- trations to tumor pMEK1 inhibition in A375 xenografts was characterized using an indirect response model (eq. 2) in which GDC-0879 inhibits the production rate of pMEK1 in a saturable manner. As seen in Fig. 4C, the model adequately characterizes the relationship between GDC-0879 concentra- tion and pMEK1 inhibition. Estimated pharmacodynamic parameters are shown in Fig. 4C.
The relationship between inhibition of the Raf signaling

Fig. 3. Colo205 colorectal carcinoma xenograft experiments 4 (A) and 5 (B). Tumor volumes are presented as mean ti S.E. Observed versus predicted Colo205 mean tumor volumes after fitting to an indirect response model (C).

Fig. 4. GDC-0879 plasma concentrations in A375 xenografts after oral doses of 35, 50 and 100 mg/kg (A) and corresponding pMEK1/MEK1 ratios (B). Observed versus predicted pMEK1/MEK1 ratio following fitting to an indirect response model (C).

pathway and efficacy was explored in more detail in A375 xenografts by using an integrated PK-PD efficacy model (Fig. 5A). In brief, the indirect response model relating GDC-0879 plasma concentrations to pMEK1 levels described above was used to simulate pMEK1 inhibition for all doses used in A375 xenograft efficacy experiments (experiments 1–3). Tumor volumes from all A375 xenograft experiments were fit simul- taneously to an indirect response model relating pMEK1 inhibition to tumor growth inhibition (eq. 3). Figure 5B is a plot of the observed versus predicted tumor volumes of the resulting fit along with estimated pharmacodynamic param- eters. Using these parameters, a plot of the relationship between Raf pathway knockdown (i.e., pMEK1 inhibition) and efficacy (i.e., K, rate constant describing the tumor growth inhibition effects of GDC-0879) was generated and is shown in Fig. 5C. Based upon this plot and the associated pharmacodynamic parameters used in its generation, ti40% pMEK1 inhibition seems to be required to obtain tumor growth inhibition in A375 xenografts, and approximately 60% pMEK1 inhibition, at a minimum, is needed for stasis.

Discussion
The Raf/MEK/ERK pathway is a highly conserved signal- ing pathway that plays a central role in cell proliferation and survival in eukaryotes (McCubrey et al., 2007). Raf kinases
are a key component of this pathway and are activated via a complex process involving phosphorylation after recruitment to plasma membranes and binding to Ras, an oncogene that is mutated in 30% of all cancers (Friday and Adjei, 2008). Activated Raf proteins directly phosphorylate multiple serine residues of MEK1 and MEK2, resulting in their activation. Both MEK1 and -2 act on ERK protein kinases, which have multiple and diverse targets that are involved in the regula- tion of several cellular processes such as cell proliferation, survival, mitosis, and migration (Friday and Adjei, 2008). Three Raf kinase isoforms have been identified and are re- ferred to as A-Raf, B-Raf, and C-Raf (also known as Raf-1) (Madhunapantula and Robertson, 2008). In comparison with other Raf isoforms, mutations in B-Raf are by far the most common, being found in approximately 50 to 70% of melano- mas, 30% of papillary thyroid cancer, and 10 to 15% of colo- rectal and ovarian cancers, making this one of the most frequently mutated genes in human cancers (Davies et al., 2002; Li et al., 2007). The majority of B-Raf mutations are in exon 15, which results in a V600E amino acid substitution, leading to constitutive kinase activation (Mercer and Prit- chard, 2003). As such, B-Raf represents an extremely attrac- tive target for the development of anticancer therapies.
GDC-0879 represents a novel potent and selective B-Raf inhibitor that is being evaluated as a potential antitumor agent. The current study shows that GDC-0879 exhibits po-

Fig. 5. Integrated PK-PD efficacy model linking GDC-0879 plasma concentration, pMEK1 inhibition, and tumor growth inhibition in A375 xenografts (A). Observed versus predicted mean tumor volumes after fitting to integrated PK-PD model and estimated pharmacodynamic parameters (B). Relationship of %pMEK1 inhibition to tumor growth inhibition constant [K] (C).

tent inhibition of Raf/MEK/ERK signaling pathway in V600E B-Raf mutant cell lines with low cellular pMEK1 inhibition IC50 estimates of 59 and 29 nM in A375 melanoma and Colo205 colorectal carcinoma cells, respectively. The compa- rable in vitro potency of GDC-0879 in these two cell lines translates to similar in vivo potency in the corresponding xenograft models (Table 1). GDC-0879 showed no evidence of activating apoptotic pathways in vitro in either cell line (data not shown); thus, the observed effects seem to be largely attributed to its impact on cell proliferation.
Indirect response models are useful in the characterization of PD effects where the monitored PD response is a down- stream event that results from a specific interaction of an agent with a receptor or enzyme (Mager et al., 2003). This scenario is analogous with the mode of action of GDC-0879, which inhibits B-Raf kinase activity leading to the inhibition of downstream MEK phosphorylation. The ability of GDC- 0879 to inhibit the phosphorylation of MEK1 in a concentra- tion-dependent manner was well characterized using an in- direct response model (Fig. 4). The estimated in vivo IC50 value of 3.06 tiM in plasma is approximately 50-fold higher than in vitro IC50 estimates obtained from studies using A375 melanoma cells (i.e., 59 nM). Observed differences in the in vivo versus in vitro IC50 of pMEK1 inhibition is prob- ably because of a combined effect of plasma protein binding and tumor disposition characteristics of GDC-0879. GDC- 0879 is approximately 78% protein bound in mouse plasma, and the volume of distribution was reported to be low, sug- gesting that the compound does not readily partition to pe- ripheral tissues (E. F. Choo, F. Feng, B. Liederer, E. Plise, N. Randolph, and Y. Ran, unpublished data).
The PK-PD modeling performed in the current study suggests that plasma GDC-0879 concentrations of 3 to 4 ti M are necessary for tumor stasis in both Colo205 and A375 xenografts (Table 1). More importantly, we were able to quantify the relationship between pathway modulation (pMEK1 inhibition) and tumor growth inhibition through
the use of an integrated PK-PD model (Fig. 5). A “PD marker-response curve” showing the relationship between pMEK1 inhibition and K (a measure of tumor growth in- hibition effect of GDC-0879) was sigmoidal in nature, re- quiring a Hill coefficient of ti 8 to describe it, suggesting a steep PD marker-response curve. GDC-0879 required a threshold of ti 40% pMEK1 inhibition beyond which a rapid increase in K was observed. This observed “switch-like” behavior is consistent with reports of ultrasensitive stim- ulus-response curves observed for mitogen-activated pro- tein kinase cascades (Huang and Ferrell, 1996). Overall, these data suggest that a significant degree of pathway modulation (ti 40% pMEK1 inhibition) is required to trig- ger tumor growth inhibition in A375 xenografts. Our ob- servations are consistent with anecdotal data from the use of MEK inhibitors in the clinic where it has been suggested that ti 60% pERK knockdown may be required for tumor growth inhibition (Tan et al., 2007; European Organiza- tion for Research and Treatment of Cancer Conference).
Similar approaches have been used to describe the inhibi- tory effects of anticancer agents on tumor growth. In most cases, concentration-tumor growth inhibition, and concentra- tion-PD marker response relationships have been modeled with separate PK-PD models. For example, with the cMet kinase inhibitor PF02341066, an effect compartment model (also referred to as a link model) was used to describe the PK-PD relationship between plasma concentration and cMet phosphorylation (Yamazaki et al., 2008). A separate indirect response model was used to relate PF02341066 plasma con- centrations with tumor growth inhibition. Based upon simi- lar estimates of EC50 (213 ng/ml) from tumor modeling and EC90 (167 ng/ml) from cMet phosphorylation modeling, it was concluded that ti90% inhibition of PD marker response was required to significantly inhibit tumor growth by ti50% (Yamazaki et al., 2008). With everolimus, an mammalian target of rapamycin inhibitor, an effect compartment model (direct-link model) was used to describe the relationship be-

tween plasma concentration and inhibition of S6 kinase 1 activity (S6K1), a marker of mammalian target of rapamycin signaling in pancreatic tumor-bearing rats. After scaling of rat pharmacokinetics to human, this PK-PD model was used to accurately describe S6K1 inhibition-time profiles in pa- tients and was later used to predict the PD marker response with different dosing regimens (Tanaka et al., 2008). The relationship between S6K1 inhibition and antitumor activity for everolimus was not quantitatively defined, and the PK-PD model was used largely to compare the effects of different dosing regimens on S6K1 inhibition.
For GDC-0879, an improved understanding of PD marker requirements associated with antitumor activity was enabled by the PD marker response curve simulated from pharmaco- dynamic parameters estimated using the described inte- grated PK-PD model (Fig. 5C). A PD marker-response curve is particularly valuable in cases where the monitored PD response is upstream of the antitumor effect. In these situa- tions, potential biological time delays between PD marker- response and antitumor effect can complicate understanding of the relationship between these two events. Similar ap- proaches utilizing integrated PK-PD models have been ap- plied to elegantly describe the action of corticosteroids on multiple PD markers, and in turn their effect on disease progression (e.g., paw swelling/edema and bone mineral den- sity) in a rat model of rheumatoid arthritis (Earp et al., 2008a,b). In addition, an integrated PK-PD model has been used to examine the action of antihyperglycemic agents on disease progression of type II diabetes mellitus (de Winter et al., 2006). However, to our knowledge, use of an integrated PK-PD model and generation of a PD marker-response curve correlating PD marker response to antitumor activity as described here has not been reported previously.
There are several assumptions and caveats associated with preclinical PK-PD modeling of anticancer therapeutics. Mouse xenograft models serve as the most common preclin- ical in vivo efficacy model in the evaluation of anti-cancer therapies (Kelland, 2004; Teicher, 2006). A primary assump- tion of preclinical PK-PD modeling using xenografts is that concentrations required for PD marker response and antitu- mor activity translate directly from mouse xenograft models to humans. This extrapolation requires that the distribution of drug is similar, this is despite reported differences in tumor vasculature and transport in xenograft versus human tumors (Jang et al., 2003). Noted differences in growth rate of human and xenograft tumors also complicate the interpreta- tion of xenograft data. Keeping the mentioned assumptions and caveats in mind, the present PK-PD study of a novel Raf inhibitor may help facilitate the design of future clinical trials and establish pharmacokinetic and pharmacodynamic endpoints through an improved understanding of drug con- centrations required for tumor stasis and of pathway sup- pression required for antitumor efficacy.

Acknowledgments
We thank Jonas Grina, Joshua D. Hansen, and the Chemistry Department of Array BioPharma for the synthesis and supply of GDC-0879 and colleagues in the Translational Oncology, Drug Me- tabolism and Pharmacokinetics, and the In Vivo Studies Group for contributions in generating data for this study.

References
Bradford MM (1976) A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal Biochem 72:248–254.
Davies H, Bignell GR, Cox C, Stephens P, Edkins S, Clegg S, Teague J, Woffendin H, Garnett MJ, Bottomley W, et al. (2002) Mutations of the BRAF gene in human cancer. Nature 417:949–954.
de Winter W, DeJongh J, Post T, Ploeger B, Urquhart R, Moules I, Eckland D, and Danhof M (2006) A mechanism-based disease progression model for comparison of long-term effects of pioglitazone, metformin and gliclazide on disease processes underlying type 2 diabetes mellitus. J Pharmacokinet Pharmacodyn 33:313–343.
Dunnett C (1955) A multiple comparison procedure for comparing several treat- ments with a control. J Am Stat Assoc 50:1096–1121.
Earp JC, Dubois DC, Molano DS, Pyszczynski NA, Almon RR, and Jusko WJ (2008a) Modeling corticosteroid effects in a rat model of rheumatoid arthritis II: mecha- nistic pharmacodynamic model for dexamethasone effects in Lewis rats with collagen-induced arthritis. J Pharmacol Exp Ther 326:546–554.
Earp JC, Dubois DC, Molano DS, Pyszczynski NA, Keller CE, Almon RR, and Jusko WJ (2008b) Modeling corticosteroid effects in a rat model of rheumatoid arthritis I: mechanistic disease progression model for the time course of collagen-induced arthritis in Lewis rats. J Pharmacol Exp Ther 326:532–545.
Friday BB and Adjei AA (2008) Advances in targeting the Ras/Raf/MEK/Erk mito- gen-activated protein kinase cascade with MEK inhibitors for cancer therapy. Clin Cancer Res 14:342–346.
Gollob JA, Wilhelm S, Carter C, and Kelley SL (2006) Role of Raf kinase in cancer: therapeutic potential of targeting the Raf/MEK/ERK signal transduction pathway. Semin Oncol 33:392–406.
Hansen JD, Grina J, Newhouse B, Welch M, Topalov G, Littman N, Callejo M, Gloor S, Martinson M, Laird E, et al. (2008) Potent and selective pyrazole-based inhib- itors of B-Raf kinase. Bioorg Med Chem Lett 18:4692–4695.
Huang CY and Ferrell JE Jr (1996) Ultrasensitivity in the mitogen-activated protein kinase cascade. Proc Natl Acad Sci U S A 93:10078–10083.GDC-0879
Jang SH, Wientjes MG, Lu D, and Au JL (2003) Drug delivery and transport to solid tumors. Pharm Res 20:1337–1350.
Kelland LR (2004) Of mice and men: values and liabilities of the athymic nude mouse model in anticancer drug development. Eur J Cancer 40:827–836.
Khazak V, Astsaturov I, Serebriiskii IG, and Golemis EA (2007) Selective Raf inhibition in cancer therapy. Expert Opin Ther Targets 11:1587–1609.
Li N, Batt D, and Warmuth M (2007) B-Raf kinase inhibitors for cancer treatment. Curr Opin Investig Drugs 8:452–456.
Madhunapantula SV and Robertson GP (2008) Is B-Raf a good therapeutic target for melanoma and other malignancies? Cancer Res 68:5–8.
Mager DE, Wyska E, and Jusko WJ (2003) Diversity of mechanism-based pharma- codynamic models. Drug Metab Dispos 31:510–518.
McCubrey JA, Steelman LS, Chappell WH, Abrams SL, Wong EW, Chang F, Leh- mann B, Terrian DM, Milella M, Tafuri A, et al. (2007) Roles of the Raf/MEK/ERK pathway in cell growth, malignant transformation and drug resistance. Biochim Biophys Acta 1773:1263–1284.
Mercer KE and Pritchard CA (2003) Raf proteins and cancer: B-Raf is identified as a mutational target. Biochim Biophys Acta 1653:25–40.
Schreck R and Rapp UR (2006) Raf kinases: Oncogenesis and drug discovery. Int J Cancer
Sebolt-Leopold JS and Herrera R (2004) Targeting the mitogen-activated protein kinase cascade to treat cancer. Nat Rev Cancer 4:937–947.
Tan W, DePrimo SE, Krishnamurthi SS, Rinehart JJ, Nabell LM, Nickens D, Bentivegna S, Wilner KD, et al. (2007) Pharmacokinetic (PK) and pharmacody- namic (PD) results of a phase I study of PD-0325901, a second generation oral MEK inhibitor, in patients with advanced cancer. American Association of Cancer Research-National Institute of Cancer-European Organization for Research and Treatment of Cancer Conference; 2007 October, San Francisco, CA.
Tanaka C, O’Reilly T, Kovarik JM, Shand N, Hazell K, Judson I, Raymond E, Zumstein-Mecker S, Stephan C, Boulay A, et al. (2008) Identifying optimal biologic doses of everolimus (RAD001) in patients with cancer based on the modeling of preclinical and clinical pharmacokinetic and pharmacodynamic data. J Clin Oncol 26:1596–1602.
Teicher BA (2006) Tumor models for efficacy determination. Mol Cancer Ther 5:2435–2443.
Wilhelm SM, Carter C, Tang L, Wilkie D, McNabola A, Rong H, Chen C, Zhang X, Vincent P, McHugh M, et al. (2004) BAY 43–9006 exhibits broad spectrum oral antitumor activity and targets the RAF/MEK/ERK pathway and receptor tyrosine kinases involved in tumor progression and angiogenesis. Cancer Res 64:7099– 7109.
Yamazaki S, Skaptason J, Romero D, Lee JH, Zou HY, Christensen JG, Koup JR, Smith BJ, and Koudriakova T (2008) Pharmacokinetic-pharmacodynamic model- ing of biomarker response and tumor growth inhibition to an orally available cMet kinase inhibitor in human tumor xenograft mouse models. Drug Metab Dispos 36:1267–1274.
Yuen ST, Davies H, Chan TL, Ho JW, Bignell GR, Cox C, Stephens P, Edkins S, Tsui WW, Chan AS, et al. (2002) Similarity of the phenotypic patterns associated with BRAF and KRAS mutations in colorectal neoplasia. Cancer Res 62:6451–6455.
Zebisch A and Troppmair J (2006) Back to the roots: the remarkable RAF oncogene story. Cell Mol Life Sci 63:1314–1330.

Address correspondence to: Dr. Edna F. Choo, Drug Metabolism and Phar- macokinetics, Genentech, Inc., 1 DNA Way, MS 412a, South San Francisco, CA 94080. E-mail: [email protected]