Preclinical activity of ABT-869, a multitargeted receptor tyrosine kinase inhibitor

ABT-869 is a structurally novel, receptor tyrosine kinase (RTK) inhibitor that is a potent inhibitor of members of the vascular endothelial growth factor (VEGF) and platelet-derived growth factor (PDGF) receptor families (e.g., KDR IC₅₀ = 4 nmol/L) but has much less activity (IC₅₀s > 1 μmol/L) against unrelated RTKs, soluble tyrosine kinases, or serine/threonine kinases. The inhibition profile of ABT-869 is evident in cellular assays of RTK phosphorylation (IC₅₀ = 2, 4, and 7 nmol/L for PDGFR-β, KDR, and CSF-1R, respectively) and VEGF-stimulated proliferation (IC₅₀ = 0.2 nmol/L for human endothelial cells). ABT-869 is not a general antiproliferative agent because, in most cancer cells, >1,000-fold higher concentrations of ABT-869 are required for inhibition of proliferation. However, ABT-869 exhibits potent antiproliferative and apoptotic effects on cancer cells whose proliferation is dependent on mutant kinases, such as FLT3. In vivo ABT-869 is effective orally in the mechanism-based murine models of VEGF-induced uterine edema (ED₅₀ = 0.5 mg/kg) and corneal angiogenesis (>50% inhibition, 15 mg/kg). In tumor growth studies, ABT-869 exhibits efficacy in human fibrosarcoma and breast, colon, and small cell lung carcinoma xenograft models (ED₅₀ = 1.5–5 mg/kg, twice daily) and is also effective (>50% inhibition) in orthotopic breast and glioma models. Reduction in tumor size and tumor regression was observed in epidermoid carcinoma and leukemia xenograft models, respectively. In combination, ABT-869 produced at least additive effects when given with cytotoxic therapies. Based on pharmacokinetic analysis from tumor growth studies, efficacy correlated more strongly with time over a threshold value (cellular KDR IC₅₀ corrected for plasma protein binding = 0.08 μg/mL, ≥7 hours) than with plasma area under the curve or C_max. These results support clinical assessment of ABT-869 as a therapeutic agent for cancer. [Mol Cancer Ther 2006;5(4):995–1006]

The reversible phosphorylation of proteins is one of the primary biochemical mechanisms mediating eukaryotic cell signaling. This reaction is catalyzed by protein kinases that transfer the γ-phosphate group of ATP to hydroxyl groups on target proteins (1). Five hundred eighteen such enzymes exist in the human genome, of which ∼90 selectively catalyze the phosphorylation of tyrosine hydroxyl groups (2, 3). Cytosolic tyrosine kinases are intracellular, whereas receptor tyrosine kinases (RTK) possess both extracellular and intracellular domains and function as membrane spanning cell surface receptors. As such, RTKs mediate the cellular responses to environmental signals and facilitate a broad range of cellular processes, including proliferation, migration, and survival.

RTK signaling pathways are normally highly regulated. However, their overactivation has been shown to promote the growth, survival, and metastasis of cancer cells (4). In particular, members of the vascular endothelial growth factor (VEGF) and platelet-derived growth factor (PDGF) receptor families of RTKs promote tumor progression through a variety of mechanisms. VEGF RTKs KDR, FLT1, and FLT4 (VEGFR-2, VEGFR-1, and VEGFR-3) contribute to tumor progression through their ability to mediate tumor angiogenesis and lymphangiogenesis and to enhance vascular permeability. Results from clinical studies with Avastin have recently validated inhibition of VEGF signaling as a cancer target (5). RTKs from the PDGF family (PDGF-α, PDGF-β, FLT3, KIT, and CSF-1R) play a role in tumor angiogenesis and modify the tumor microenvironment in a fashion beneficial to tumor cell growth and metastasis. Additionally, mutations of FLT3 and KIT that result in constitutive kinase activity are directly associated with the proliferation of acute myeloid leukemia (AML) blast cells and gastrointestinal stromal tumor cells, respectively (6–13).

Although VEGF and PDGF RTKs are compelling cancer targets in and of themselves, preclinical studies indicate that greater antiangiogenic activity can be achieved by targeting these mechanisms simultaneously. Tumors are capable of secreting multiple angiogenic factors, and they depend on these factors at different stages of progression (14–20). For example, VEGF causes a large increase in blood vessel formation; yet, these vessels are immature and leaky. The formation of thicker, more stable vessels requires encapsulation by pericytes that is driven by PDGFR-β signaling (21, 22). Pericyte coverage is impaired by PDGFR-β inhibition, and microvessels lacking pericyte coverage regress upon deprivation of VEGF signaling (23, 24). Synergistic antiangiogenic effects may, therefore, be possible by targeting not only endothelial cells but also the supporting perivascular cells (24, 25).

The observation that the tumor microenvironment actively participates in tumor progression provides another reason to pursue agents capable of blocking multiple VEGF and PDGF receptor family members (26–28). Tumor-associated fibroblasts that express PDGFR-β are functionally distinct from normal fibroblasts and are thought to foster tumor cell invasion and metastasis through the release of growth factors and proteases (29). FLT1 and CSF-1R regulate the production and activation of tumor-associated macrophages that hasten the breakdown of the extracellular matrix by triggering gelatinase A expression (30, 31). KIT-expressing inflammatory mast cells have been implicated as accessory cells in tumor angiogenesis (32, 33). Simultaneous inhibition of these RTKs may, therefore, make the tumor microenvironment less hospitable for tumor growth. In addition, multitargeted VEGF/PDGF RTK inhibitors can also directly inhibit cancer cells whose proliferation is reliant on a mutant or overexpressed kinase, such as mutations in FLT3 observed in AML patients and KIT receptors in gastrointestinal stromal tumor and mast cell tumors (34–38).

The compelling nature of RTKs as anticancer targets provided the impetus for a search for potent and novel RTK inhibitors. The culmination of this search, ABT-869, is a novel multitargeted inhibitor of VEGF and PDGF receptor family members (Fig. 1). The present communication describes the in vitro and in vivo properties of ABT-869 in a preclinical setting that support future clinical studies in cancer therapy.

The chemical structures and synthesis of the kinase inhibitors N-[4-(3-amino-1H-indazol-4-yl)phenyl]-N1-(2-fluoro-5-methylphenyl) urea (ABT-869; ref. 39), 6-[2-(methylcarbamoyl)phenylsulfanyl]-3-E-[2-(pyridin-2-yl)ethenyl]indazole (AG013736; ref. 40), 4-amino-5-fluoro-3-[6-(4-methyl-1-piperazinyl)-1H-benzimidazol-2-yl]-2(1H)-quinolinone (CHIR258; ref. 41), N-(3-trifluoromethyl-4-chlorophenyl)-N′-(4-(2-methylcarbamoyl-pyridin-4-yl)oxyphenyl)urea (BAY 43-9006; ref. 42), 5-[5-fluoro-2-oxo-1,2-dihydroindol-(3Z)-ylidenemethyl]-2,4-dimethyl-1H-pyrrole-3-carboxylic acid-(2-DEAE)amide (SU11248; ref. 43), and 1-(4-chloroanilino)-4-(4-pyridylmethyl) (PTK787; ref. 44) have been reported. These chemical entities were synthesized at Abbott for comparison studies and are designated in this article using the abbreviations previously used for these compounds.

Potencies (IC₅₀ values) were determined by assays of active kinase domains cloned and expressed in baculovirus using the FastBac baculovirus expression system (Life Technologies, Gaithersburg, MD) or obtained commercially. For tyrosine kinase assays, a biotinylated peptide substrate containing a single tyrosine was used with 1 mmol/L ATP, an Eu-cryptate–labeled anti-phosphotyrosine antibody (PT66), and Strepavidin-APC in a homogeneous time-resolved fluorescence assay (45). Serine/threonine kinases were assayed using 5 μmol/L ATP, [³³P]ATP, and a biotinylated peptide substrate with peptide capture and incorporation of ³³P determined using a SA-Flashplate (Perkin-Elmer, Boston, MA; ref. 46). Each inhibitor was assayed at multiple concentrations prepared by serial dilution of a DMSO stock solution of the compound. The concentration resulting in 50% inhibition of activity was calculated using nonlinear regression analysis of the concentration response data.

An ELISA-based screen in 3T3 murine fibroblasts engineered to express human KDR was used to evaluate KDR receptor phosphorylation and is described in detail elsewhere (47). A stimulatory autocrine loop in an angiopoietin-expressing 3T3 fibroblast line engineered to express human TIE2 was used to quantify inhibition of TIE2, and NIH3T3 cells engineered to express a KDR:KIT chimera (KDR extracellular domain fused to KIT transmembrane and intracellular domains) were used to quantify inhibition of KIT. In brief, autocrine or stimulated (50 ng/mL VEGF₁₆₅, R&D Systems, Minneapolis, MN) cells were incubated with inhibitor for 20 minutes. Lysates were prepared and added to plates precoated with capture antibody (receptor specific, coated into 96-well plates overnight followed by blocking with 3% nonfat milk in PBS) and incubated for 2 hours. Phosphorylated receptor was detected using anti-phosphotyrosine 4G10-biotin (UBI, Charlottesville, VA) and streptavidin-horseradish peroxidase (HRP; UBI). The HRP reaction was initiated with K-Blue substrate (Neogene Corp., Lexington, KY), quenched with 1 mol/L phosphoric acid and measured at 450 nm on a SpectraMax plate reader. For CSF-1R phosphorylation, NIH3T3 cells engineered to express human CSF-1R were stimulated for 10 minutes with M-CSF (200 ng/mL), lysed, and processed for Western blot by immunoprecipitation with anti-CSF1R antibody (Calbiochem, La Jolla, CA) with detection after SDS-PAGE using anti-phosphotyrosine 4G10 and anti-mouse-HRP antibody. For PDGFR-β, lysates from PDGF-BB stimulated NIH3T3 cells (50 ng/mL for 5 minutes) were immunoprecipitated with anti-PDGFR-β antibody (SC-432, Santa Cruz Biotechnology, Santa Cruz, CA). Phosphorylated PDGFR-β was detected after SDS-PAGE with anti-phosphotyrosine 4G10-HRP 16-105, (Upstate Biotechnology, Charlottesville, VA). For KDR phosphorylation in cell lysates obtained from human umbilical artery endothelial cells (HUAEC; Cambrex Corp., East Rutherford, NJ), tumors, or lungs, the KDR antigen was concentrated by immunoprecipitation using an anti-VEGFR2 antibody (SC-6251, Santa Cruz Biotechnology or AF-357, R&D Systems). Phosphorylated KDR was visualized on Western blots using anti-phosphotyrosine 4G10-HRP (07-374, Upstate) and total KDR with anti-Flk-1 antibodies (SC-315 or SC-6251, Santa Cruz Biotechnology). Antigens were visualized by enhanced chemiluminescence with the Pierce Dura SuperSignal substrate (Pierce, Rockford, IL). The phosphotyrosine and total receptor bands were digitized by UN-SCAN-IT software (Silk Scientific, Orem, UT) for quantification. The anti-phosphotyrosine bands were normalized using the total receptor bands, and the percent inhibition was calculated at each concentration of inhibitor.

HUAEC were plated into 96-well plates at 2,500 per well and incubated with serum-free medium for 24 hours. Drug and VEGF (final, 10 ng/mL) were added and incubated for 72 hours in serum-free medium. For carcinoma cell lines, 2,500 per well were plated overnight in full growth medium. Drug was added to the cells in full growth medium and incubated for 72 hours. For leukemia cells, generally 50,000 per well were plated in full growth medium, drug added, and incubated for 72 hours. The effects on proliferation were determined by addition of Alamar Blue (final solution, 10%), incubation for 4 hours at 37°C in a CO₂ incubator, and analysis in a fluorescence plate reader (544 nm, excitation: 590 nm, emission). For VEGF-stimulated growth, percent inhibition of proliferation was determined using the difference between VEGF-stimulated cells and unstimulated cells as the control, and IC₅₀ values were determined by nonlinear regression analysis of the concentration response data.

Seventy-two-hour drug and vehicle-treated MV4-11 cells were fixed with 80% ethanol, washed with PBS, and then incubated with propidium iodide (50 μg/mL). DNA content was determined by fluorescence cell analysis using a FACSCalibur (BD Biosciences, San Jose, CA) flow cytometer, and cell cycle distribution was analyzed with CellQuest software. Apoptotic population (sub-G₀-G₁) was confirmed by Annexin V staining using the Annexin V-FITC Apoptosis Detection kit (BD PharMingen, San Diego, CA) according to the supplier's instructions.

Estradiol-induced edema in mice was measured following a procedure originally described in rat (48). Female BALB/c mice >12 weeks old were treated with 10 units of pregnant mare's serum gonadotropin at 72 and 24 hours before compound administration (0.1 mL/mouse i.p.). ABT-869 was orally given (n = 6) at various times before a 0.1 mL/mouse i.p. injection of 5 mg/mL 17β-estradiol. Mice were sacrificed 3 hours after estradiol injection. Uteri were removed, cleaned of fat and connective tissue, and weighed. Mean increase in weight above sham controls was computed and used to calculate percent inhibition of vehicle-treated control.

A hydron-sucralfate pellet, containing either 30 ng of basic fibroblast growth factor (bFGF; right cornea) or 150 ng VEGF (left cornea), was inserted into a corneal pocket made by surgical incision ∼0.7 mm from the limbus of eyes of CF1 mice (Charles River Labs, Wilmington, MA). On day 5 (right cornea with bFGF) and again on day 7 (left cornea with VEGF pellet), a magnified corneal image was obtained by using a digital camera attached to a slit lamp biomicroscope. Data acquisition and storage were achieved with Leica imaging software. Statistical significance was evaluated with a two-tailed t test.

VEGF-induced KDR phosphorylation in lung was assessed essentially as described (49). ABT-869 or vehicle was given by oral gavage to mice (three per group). At 1 or 3.5 hours after dosing, 3 μg VEGF was given via tail vein, and the animals were sacrificed 2 minutes later. Lungs were quickly resected and stored in liquid nitrogen until processed. Frozen tissue (175 mg/mL) was homogenized (PowerGen 35, Fisher Scientific, Pittsburgh, PA) for 30 seconds in radioimmunoprecipitation assay buffer [50 mmol/L Tris-HCl (pH 7.4), 1% IGEPAL, 150 mmol/L NaCl, 1 mmol/L EDTA, and 0.25% sodium deoxycholate] containing protease inhibitors (Complete, Roche Diagnostics, Indianapolis, IN) and NaF (1 mmol/L). After 1 hour at 4°C, the were lysates cleared by centrifugation at 15,000 rpm in a refrigerated microcentrifuge for 15 minutes. The supernatants were stored at −20 C until analysis of KDR phosphorylation as described above by Western blot analysis.

To assess duration of inhibition, lungs from mice bearing an HT1080 tumor in the flank (∼1 g) were harvested 1 to 6 hours after treatment with ABT-869 and homogenized in radioimmunoprecipitation assay buffer, and the resulting extract was assayed for phosphorylated KDR using the ELISA as described in Receptor Phosphorylation.

Cell lines were obtained from the American Type Culture Collection (Manassas, VA). For flank tumor xenograft studies, cells were suspended in PBS, mixed with an equal volume of Matrigel (phenol red free; BD Biosciences), and inoculated into the flank of mice of the indicated strain. For orthotopic models, cells were inoculated into either the mammary fat pad (breast) of mice or intracranially (glioma) into the right cerebral hemisphere 3 mm in depth and 2.5 mm lateral, 2 mm anterior to the bregma in Fisher rats. The number of cells inoculated were 0.2 million (H526, DLD-1, MDA-231, MDA-435LM, and HCT-116), 0.5 million (HT1080, A431, and 9L), and 7.5 million (MV4-11). MX-1 cells were inoculated as a 1:10 dilution of brei obtained from 0.25 g of tumor propagated in severe combined immunodeficient mice. At the designated time after inoculation, tumor-bearing animals were divided into groups (n = 10), and administration of vehicle (2% ethanol, 5% Tween 80, 20% PEG400, 73% saline) or inhibitor at the indicted dose was initiated. Tumor growth in the flank was assessed by measuring tumor size with calipers and calculating size using the formula (L × W²/2). Tumor volume for the orthotopic glioma model was determined using magnetic resonance imaging. Inhibition of solid tumor growth on a specific day or tumor size was represented as mean volume of the test drug group over the mean volume of the untreated (control) group (T/C) or as percent inhibition of control [(1 − T/C) × 100].

Mice with established MDA-231 tumors (0.2–0.3 g) were given ABT-869 (12.5 mg/kg, twice daily or bid) or vehicle for 3.5 days. Tumors were harvested 4 hours after final treatment, fixed in Streck Tissue Fixative, and paraffin embedded. Five-micrometer sections were incubated with monoclonal rat anti-mouse CD31 antibody (BD PharMingen) followed by biotinylated rabbit anti-rat immunoglobulins (DAKO, Carpinteria, CA), then by StreptABC complex/HRP (DAKO). Mosaic pictures of whole tissue section stained for CD31 were taken with Zeiss Axioplan2 Image Microscope. The density of microvessels outlined by CD31-positive endothelial cells in tumors was analyzed with AxioVision4.1 image measuring and analyzing program (Carl Zeiss MicroImaging, Inc., Thornwood, NY).

At the terminus of tumor growth studies, each test group (n = 10) was dosed with drug, and blood was withdrawn over a 12-hour period. Plasma (100 μL) obtained from the blood samples was diluted with 1 volume of 10% methanol, 25 μL of 0.25 mmol/L Na₂CO₃ and extracted twice with 3 volumes of ethyl acetate. After removing the ethyl acetate under nitrogen, the residue was reconstituted in methanol and subjected to high-performance liquid chromatography using UV detection (225 nm) on a YMC ODS-A column (Waters Corp., Milford, MA) eluted with acetonitrile/10 mmol/L sodium acetate (pH 5.5; 43:57, v/v). The detection limit for ABT-869 was ∼20 nmol/L with an extraction efficiency of 68%.

The potency of ABT-869 for inhibiting targeted RTKs is given in Table 1. For comparison, potency against a set of nonrelated RTKs and serine/threonine kinases is also provided. ABT-869 exhibits IC₅₀ values that range from 4 nmol/L (KDR) to 190 nmol/L (FLT4) for members of the VEGF and PDGF receptor families. The molecule is also active against TIE2 and, to a lesser extent, RET, but is much less active (IC₅₀ > 10 μmol/L) against other nonrelated tyrosine kinases, such as steroid receptor coactivator and epidermal growth factor receptor. ABT-869 was also generally not effective against serine/threonine kinases. Of those evaluated, only SGK was significantly inhibited at concentrations of ABT-869 below 1 μmol/L (IC₅₀ = 0.94 μmol/L). This activity was measured at low ATP (5 μmol/L compared with 1 mmol/L for tyrosine kinase assays). Because ABT-869 has been shown to be an ATP-binding site competitive inhibitor (47), any activity against serine/threonine kinases observed in these studies is most likely an overestimation of activity that would be observed in the high-ATP cellular environment. As is discussed later (see Comparison to Other Tyrosine Kinase Inhibitors), this pattern of selectivity toward VEGF and PDGF RTK families remains evident after assessment against a broader range of kinases.

The potency of ABT-869 against tyrosine kinase enzyme preparations is reflected in receptor-mediated responses to VEGF and PDGF at the cellular level. Phosphorylation of KDR induced by VEGF was inhibited by ABT-869 with an IC₅₀ of 4 nmol/L in 3T3 murine fibroblasts engineered to express human KDR. (Table 2). A similar potency for inhibition of receptor autophosphorylation was seen with ABT-869 when HUAECs were used as the target cell. Because of the low abundance of KDR in these cells, an ELISA format could not be used, and the analysis was done using immunoprecipitation and Western blot techniques. As shown in Fig. 2A, ABT-869 inhibited VEGF-stimulated phosphorylation of KDR completely at 10 nmol/L and by ∼70% at 3 nmol/L (IC₅₀ ∼ 2 nmol/L). Taken together, these results clearly show that ABT-869 is a potent inhibitor of KDR phosphorylation in engineered cell lines and primary endothelial cells.

The enzyme-inhibitory activity of ABT-869 against other tyrosine kinases was also evident in cellular assays. ABT-869 showed significant activity against ligand-induced PDGFR-β, KIT, and CSF-1R phosphorylation, with IC₅₀s of 2, 31, and 10 nmol/L, respectively (Table 2). Indeed, in the case of PDGFR-β, ABT-869 was substantially more potent in the cellular assay than in the enzyme assay (2 versus 66 nmol/L). In contrast, cellular TIE2 autophosphorylation was less sensitive to ABT-869 than other family members and required a concentration of 3,500 nmol/L to reach 50% inhibition of receptor activity. The reasons for these discrepancies are not known, but they may reflect the artificiality of the isolated kinase assays (e.g., truncated kinase domains, synthetic substrates, and high ATP concentration). Consequently, cellular potency may be more predictive of in vivo activity.

Receptor phosphorylation in the VEGF pathway is an important mitogenic signal for endothelial cells (50). As expected, ABT-869 inhibited VEGF-stimulated HUAEC proliferation (Table 3), although the IC₅₀ was 5- to 10-fold lower than would be predicted by activity in the cellular phosphorylation and enzyme inhibition assays. In contrast, with a panel of tumor cells grown in serum-containing media (Table 3), ABT-869 had only weak activity (IC₅₀ > 1 μmol/L) against cells whose proliferation is not driven by a VEGF or PDGF tyrosine kinase, such as HT-29 and MDA-435 carcinoma cells. Antiproliferative effects in these cellular assays is observed at >1,000-fold higher ABT-869 concentrations than in the VEGF-stimulated HUAEC assay. ABT-869 did potently inhibit (IC₅₀ = 4 nmol/L; Table 3) the proliferation of MV4-11 leukemia cells that are known to express a constitutively active form of FLT3 (FLT3-ITD; ref. 51).

It is worth noting that the cellular potency of ABT-869 is affected by serum protein. The presence of 50% mouse or human plasma increased the IC₅₀ value for inhibiting KDR phosphorylation by a magnitude (60- and 85-fold; Table 2) that closely approximates the free fraction predicted by the protein binding of ABT-869 in serum (98.2% and 99.0%, respectively). These results suggest that potency determined in “protein-free” settings, such as those used to assess VEGF-stimulated HUAEC proliferation, may be shifted as much as 85 fold in the presence of physiologically relevant protein concentrations. Thus, the IC₅₀ value for HUAEC proliferation could shift to as high as 20 nmol/L in the presence of serum, a concentration still well below that necessary to inhibit the proliferation of non–VEGF-dependent tumor cells (Table 3).

The antiproliferative effect of ABT-869 was associated with changes in the cell cycle profile and survival of MV4-11 cells evident after Annexin V/propidium iodide flow analysis. Exposure to ABT-869 for 72 hours caused a decrease in S and G₂-M phases with a corresponding increase in the sub-G₀-G₁ apoptotic population (Fig. 2B). An increase in apoptosis to a similar magnitude was also observed by measuring Annexin staining at 24 and 48 hours (7% and 12%; data not shown). The induction of apoptosis coupled with the antiproliferative effect of ABT-869 are in keeping with the strong survival signal provided by mutated FLT3 in AML (51).

Murine lung expresses a high level of KDR and as such provides an accessible target organ to monitor the in vivo effects of receptor kinase inhibitors on ligand-mediated receptor activation (49). Inhibition of KDR phosphorylation was assessed using immunoprecipitation/Western blot techniques with lung tissue obtained from mice dosed with ABT-869 before administration of VEGF. As is shown in Fig. 3A, oral administration of ABT-869 resulted in inhibition of KDR phosphorylation in lung tissue. When assessed 1 hour after dosing, a dose of 0.3 mg/kg provided complete inhibition (Fig. 3A, left). When the sampling time was extended to 3.5 hours after dose, a period more reflective of the duration of significant inhibition observed in a VEGF response model after an efficacious dose of ABT-869 (see below), a dose of 3 mg/kg was required for complete inhibition of receptor phosphorylation (Fig. 3A, right). These results show that ABT-869 inhibits ligand-induced phosphorylation of a target receptor, and that a dose level of at least 3 mg/kg is required for complete inhibition for >3.5 hours. Increased vascular permeability, a hallmark of VEGF-induced responses in the uterus, serves as an in vivo indication of ligand activation of KDR in surrogate tissue (48, 52). As shown in Fig. 3B, ABT-869 given orally inhibited the edema response in a dose-dependent manner with a potency (ED₅₀ = 0.5 mg/kg) that agreed well with the potency for inhibiting receptor phosphorylation of ABT-869 in lung. The acute response to estradiol challenge was also used to evaluate duration of inhibition of a VEGF-mediated response following a single dose of ABT-869. The dose chosen for these studies (5 mg/kg) is sufficient (when given bid) to produce 60% to 70% inhibition in most of the tumor growth models discussed below. As shown in Fig. 3C, administration of this dose resulted in >50% inhibition of the VEGF response for 3 to 4 hours. This corresponds to an inhibition of VEGF-mediated responses for ∼6 to 8 hours in a 24-hour period with the compound given bid.

ABT-869 was also effective in an in vivo model of growth factor–induced angiogenesis. When given daily for 5 days (bFGF) or 7 days (VEGF), ABT-869 (7.5 and 15 mg/kg bid) significantly (P < 0.001) inhibited both bFGF- and VEGF-induced increases in vessel density in the cornea (Fig. 3D). A similar pattern was observed when angiogenesis was quantified by vessel length (data not shown). The efficacy for inhibition of angiogenesis induced by VEGF is consistent with the potency for inhibiting KDR phosphorylation in lung. However, ABT-869, in spite of a lack of significant potency for inhibiting FGF receptor (Table 1), also inhibited bFGF-induced angiogenesis. This activity is most likely due to the role of VEGF and PDGF family member kinases in FGF signaling and has been reported for other small molecule inhibitors (53).

ABT-869 has been evaluated in flank xenograft models using human tumor cell lines that represent a broad range of tumor types, including a highly angiogenic fibrosarcoma (HT1080; refs. 54, 55), a small cell lung carcinoma known to express KIT (H526; ref. 56), colon carcinoma (DLD-1), and breast carcinoma (MX-1; ref. 57). Treatment with ABT-869 inhibited tumor growth in each of these models in a dose-dependent manner (Fig. 4A-D), although the potency for robust inhibition (ED₇₅) varied from a low of 4.5 mg/kg bid (colon carcinoma) to a high of 12 mg/kg bid (small cell lung carcinoma). Within this group, the fibrosarcoma (7.5 mg/kg bid) and breast carcinoma (6 mg/kg bid) were intermediate in sensitivity to ABT-869 treatment.

Two additional human tumor cell lines (A431, human epidermoid carcinoma and MV4-11, AML) have been used to evaluate ABT-869 in a flank xenograft setting. In both cases, reduction in tumor size was observed after treatment with dose levels of ABT-869 equal to or less than used in the studies described above. As shown in Fig. 4E, treatment of mice with small A431 tumors (∼0.25 g) resulted in suppression of tumor growth. Treatment of mice with large established tumors (∼1.25 g, day 21) resulted in a decrease (∼25%) in tumor size followed by prolonged tumor stasis. Removing treatment resulted in tumor growth at a rate similar to that in vehicle-treated mice (day 28). Resumption of treatment halted tumor growth and reduced tumor size (day 42).

Tumor regression (>50% reduction in tumor size) in response to treatment with ABT-869 was clearly evident from results with a flank xenograft model using human myeloid leukemia cell line MV4-11 (Fig. 4F). Treatment of mice with established tumors (≈1 g) resulted in rapid tumor regression at doses ≥1.5 mg/kg bid. Larger tumors (>2 g) were also subject to regression upon treatment with ABT-869. Growth inhibition with subsequent reduction in tumor size was observed at doses as low as 0.5 mg/kg bid.

ABT-869 has been evaluated in an orthotopic setting with two breast carcinoma cell lines: MDA-231 (epithelial) and MDA-435LM (ductal). As monotherapy, ABT-869 exhibited efficacy in the orthotopic setting (Fig. 5A and B) comparable with that observed in the flank models described above. In combination, ABT-869 augmented the effects seen with paclitaxel in both of the orthotopic models. ABT-869 was also effective in a rat orthotopic glioma model. Administration of ABT-869 (1.5, 5, and 10 mg/kg bid) to rats bearing brain tumors derived from rat 9L cells resulted in dose-dependent inhibition of tumor growth (Fig. 5C). The efficacy at the highest dose (67% inhibition) was similar to that achieved in the breast orthotopic model and illustrates the effectiveness of ABT-869 in a nonmurine species.

In parallel studies with s.c. tumors derived from the MDA-231 breast carcinoma cell line immunohistologic examination was used to examine the effect of treatment with ABT-869 on tumor vascularization. Analysis of tumors resected after 3.5 days of treatment with ABT-869 (12.5 mg/kg bid) revealed a decrease in the density of CD31-positive vessels compared with vehicle treated (0.16% versus 0.49%, P = 0.02; Fig. 5D). The reduction in microvasculure density is consistent with an antiangiogenic mechanism of action of ABT-869.

In an effort to relate efficacy to drug exposure rather than dose, the concentration of ABT-869 was assayed in plasma samples obtained over a 12-hour period after the last dose at the completion of the efficacy studies. The relationship between efficacy and plasma concentration (C_max) or exposure (24-hour area under the curve or AUC_{24 hours}) across a variety of tumor models is illustrated in Fig. 6. These results highlight the variability in sensitivity to treatment with ABT-869. Thus, for a given exposure (e.g., 2.7 μg·hour/mL), efficacy ranges from 100% (highly responsive, A431) to 75% (moderately responsive, HT1080) to 35% inhibition (relatively resistant, MDA-435). Thus, choosing a value for efficacious exposure, whether in terms of C_max or AUC_{24 hours}, is by necessity model dependent. The moderately sensitive HT1080 fibrosarcoma model was selected as a representative, or benchmark, model for defining pharmacokinetic targets for robust tumor growth efficacy (75% inhibition). As shown in Fig. 6A, the target C_max and AUC_{24 hours} for ABT-869 are 0.4 μg/mL and 2.7 μg·hour/mL, respectively.

Terminal plasma drug concentration data presented in Fig. 6B indicate that continuous inhibition of the target RTKs is not required for activity. The dose from the illustrated study with the HT1080 model provided ABT-869 plasma concentrations that exceed the IC₅₀ for inhibition of cellular KDR in the presence of plasma (0.24 μmol/L) for only ∼4 of the 12 hours in the bid dosing cycle, although this dose resulted in a 69% reduction in tumor growth. The extent of time above the plasma IC₅₀ agrees well with the duration of >50% inhibition of a functional response to VEGF (uterine edema; Fig. 3C) and receptor phosphorylation (Fig. 6B).

As is shown in Fig. 6A for both C_max and AUC_{24 hours}, splitting the total daily dose into two doses (bid: 77%) was statistically (P < 0.01) more effective than the same dose given once daily (qd, 67%). The observation that increasing dosing frequency at least maintains efficacy in the HT1080 model implies that longer but not necessarily complete coverage of the target may be preferred. Therefore, assessment of time over threshold is likely to be an important predictor of efficacy. The relationship between duration of exposure and efficacy of ABT-869 in the HT1080 model is illustrated in Fig. 6C. This relationship suggests that ABT-869 exposure exceeding threshold values for 7 of 24 hours is sufficient for robust (>75%) tumor growth inhibition.

To gain an understanding of the potential clinical merit of ABT-869, the inhibition profile of ABT-869 against a panel of kinases was compared with the profiles of five previously described inhibitors that have undergone clinical development (58–60). Hierarchical clustering of enzyme inhibition potency (Fig. 7A) reveals that ABT-869 provides complete coverage (K_i ≤ 10 nmol/L) of the targeted kinases (VEGF and PDGF RTKs). By comparison, the relatively selective inhibitors PTK787 and BAY 43-9006 are only partially effective against the VEGF and PDGF family of kinases. AG013736, CHIR 258, and SU 11248 are active against most of the targeted kinases. However, these inhibitors exhibit more activity against nontargeted kinases than does ABT-869.

The in vivo activities of the reference kinase inhibitors were assessed in the VEGF-mediated uterine edema model. As is illustrated in Fig. 7B, ABT-869 is more potent than the other inhibitors in blocking VEGF-induced edema.

ABT-869 is a novel, ATP-competitive inhibitor of VEGF and PDGF RTKs that lacks significant activity against representative cytosolic tyrosine kinases and serine/threonine kinases. The selectivity profile of ABT-869 differs from several previously described inhibitors (e.g., PTK787, ZD6474, CP-547,632, and AG013736) and antibodies (Avastin) that are more restricted to KDR and VEGF (58–60). Simultaneous inhibition by ABT-869 of the VEGF and PDGF RTKs, which mediate tumor progression by multiple mechanisms, may result in greater antitumor efficacy and provides the potential to treat a broader range of human cancers than more selective agents. The validity of a multitargeted approach is supported by recent reports of promising clinical results in two renal cell carcinoma studies with SU11248, a molecule that exhibits a similar VEGF and PDGF inhibitory profile to ABT-869. Partial responses were achieved in 40% of patients given SU11248 (61). However, toxicity, primarily grade 1 or 2 fatigue, diarrhea, and/or nausea, necessitates dosing holidays of 2 weeks in each 6-week cycle. These adverse effects may be a consequence of uninterrupted RTK inhibition due to the compound's 40-hour half-life (62, 63) or, alternatively, may reflect activity against off-target kinases exhibited by SU11248 and other kinase inhibitors (Fig. 7A).

Consistent with the putative mode of action defined with isolated enzyme preparations, ABT-869 exhibited potent activity (IC₅₀s = 2–30 nmol/L) for inhibiting KDR, PDGFR-β, KIT, and CSF-1R phosphorylation in intact cells. However, although effective against most members of the KDR and PDGF family of RTKs that were evaluated using intact cell assays, ABT-869 was only marginally active (IC₅₀ = 3.5 μmol/L) in inhibiting phosphorylation of TIE2 in a cellular setting in spite of exhibiting submicromolar potency (170 nmol/L) in the isolated enzyme assay. Unlike the other cellular constructs used in these studies, the TIE2 construct relied on an autocrine loop for activation and may therefore not accurately reflect the sensitivity of a ligand-activated kinase to ABT-869. Further studies will be necessary to understand the basis of this discrepancy between TIE2 potency in the isolated enzyme and cellular settings.

Other examples of mechanism-based cellular activity include the inhibition of KDR-dependent proliferation of endothelial cells and the FLT3-dependent proliferation of MV4-11 leukemia cells. ABT-869 causes a decrease in the phosphorylation of both FLT3 and the downstream signaling protein signal transducers and activators of transcription 5 at the same concentrations in which inhibition of proliferation of MV4-11 cells is observed.¹

Interestingly, in contrast, inhibition of VEGF-induced proliferation of endothelial cells occurs at a concentration 10-fold lower than inhibition of KDR phosphorylation. This raises the possibility that additional kinases targeted by ABT-869 may play a role in VEGF signaling in endothelial cells. This interesting possibility notwithstanding, the in vivo potency for inhibiting KDR phosphorylation and a VEGF functional response (edema) are comparable, which further supports the mode of action of ABT-869 as inhibition of VEGF receptor activation. These observations do not preclude a contribution to efficacy due to inhibition of other growth factor receptors, particularly PDGFR-β, which plays an important role in maintaining tumor vasculature (64). Further studies will be necessary to assess the contribution of other target receptors to the activity of ABT-869. In any case, the potent and direct effects of ABT-869 on VEGF signaling suggest the possibility of observing mechanism-based responses early in the clinical development of ABT-869.

ABT-869 showed efficacy in a broad spectrum of xenograft tumor growth models, including human fibrosarcoma and breast, colon, and small cell lung carcinomas. ABT-869 has also been shown to be effective in an orthotopic model of prostate cancer in rat (data not shown). Although effective in all models tested, certain tumor types seemed more sensitive to treatment with ABT-869 than others. Treatment of mice bearing xenografts derived from epidermoid carcinoma or AML cells with ABT-869 resulted in a reduction in tumor size, whereas the small cell lung carcinoma model (H526) was somewhat less sensitive to treatment (60% inhibition of tumor growth, 5 mg/kg bid).

The antiangiogenic effects of ABT-869 presumably contribute to its antitumor efficacy, particularly in models that use highly vascular tumors, such as MX-1 and HT1080, that are known to express angiogenic factors (54, 55, 57). In that regard, immunohistochemical evidence showing a decrease in tumor vasculature as a result of treatment with ABT-869 was obtained in the breast carcinoma model. Nonetheless, the exquisite sensitivity to ABT-869 in vivo of other tumor cells, such as MV4-11-derived tumors, may be due, at least in part, to the potent antiproliferative activity of ABT-869 for tumor cells expressing mutated receptor kinases that are constitutively active (e.g., FLT3-ITD). Indeed, ABT-869 has been shown to inhibit FLT3 phosphorylation, increase apoptotic cells, and decrease proliferation of MV4-11 cells in vivo, supporting a direct effect of ABT-869 on these cells.¹ However, overexpression of wild-type target kinases that require ligand activation is not sufficient to confer high sensitivity because the H526 cells used in the small cell lung carcinoma model in the current studies (only moderately sensitive to ABT-869) are known to express high levels of the targeted kinase KIT (56). This suggests that ligand-independent activation is key to kinase-dependent tumor growth.

Based on plasma drug exposure data from the preclinical murine experiments, pharmacokinetic targets in the mouse were identified as AUC_{24 hours} ≥ 2.7 μg·h/mL or plasma concentrations above the threshold value of 0.08 μg/mL for ≥7 hours. The duration of exposure required for antitumor efficacy agrees well with duration of inhibition achieved in vivo at the receptor and functional level (>50% for 3–4 hours/12-hour dosing period) and implies that continuous suppression of the target receptor is not necessary for robust antitumor activity. These results are in contrast to previous reports with kinase inhibitors, indicating that constant target suppression is necessary for antitumor efficacy (49). The close relationship between the kinetics for receptor inhibition in vivo and plasma concentration suggests that the receptor interaction with ABT-869 is readily reversible.

In summary, the preclinical characteristics of ABT-869 suggest that this molecule may offer distinct advantages in the realm of kinase inhibitors in cancer therapy. It is a multitargeted inhibitor that is focused on the VEGF and PDGF family of RTKs and lacks activity against off-target kinases. This profile, coupled with the ability to provide intermittent inhibition of target kinases and still maintain antitumor efficacy, may provide a unique clinical advantage by presenting an opportunity to avoid adverse reactions that are associated with long-term exposure to kinase inhibitors without relying on dosing holidays.