Volitinib, a potent and highly selective c‐Met inhibitor, effectively blocks c‐Met signaling and growth in c‐MET amplified gastric cancer patient‐derived tumor xenograft models

Abstract

Purpose

To investigate the incidence of cMET gene copy number changes and protein overexpression in Chinese gastric cancer (GC) and to preclinically test the hypothesis that the novel, potent and selective cMET small‐molecule inhibitor volitinib, will deliver potent anti‐tumor activity in cMET‐dysregulated GC patient‐derived tumor xenograft (PDX) models.

Experimental design

A range of assays were used and included; in vitro cell line panel screening and pharmacodynamic (PD) analysis, cMET fluorescence in‐situ hybridization (FISH) and immunohistochemical (IHC) tissue microarray (TMA) analysis of Chinese GC (n = 170), and anti‐tumor efficacy testing and PD analysis of gastric PDX models using volitinib.

Results

The incidence of cMET gene amplification and protein overexpression within Chinese patient GC tumors was 6% and 13%, respectively. Volitinib displayed a highly selective profile across a gastric cell line panel, potently inhibiting cell growth only in those lines with dysregulated cMET (EC50 values 0.6 nM/L–12.5 nM/L). Volitinib treatment led to pharmacodynamic modulation of cMET signaling and potent tumor stasis in 3/3 cMET‐dysregulated GC PDX models, but had negligible activity in a GC control model.

Conclusions

This study provides an assessment of tumor cMET gene copy number changes and protein overexpression incidence in a cohort of Chinese GC patients. To our knowledge, this is the first study to demonstrate anti‐tumor efficacy in a panel of cMET‐dysregulated gastric cancer PDX models, using a novel selective cMET‐inhibitor (volitinib). Thus, the translational science presented here provides strong rationale for the investigation of volitinib as a therapeutic option for patients with GC tumors harboring amplified cMET.

Highlights

• We characterize cMET gene and protein expression in Chinese gastric cancer patients.

• We identify correlations between cMET gene, protein expression and response.

• Volitinib shows potent activity in cMET‐dysregulated cell lines and PDX models.

• Our data highlight the potential for volitinib in cMET‐driven gastric cancers.

1. Introduction

Located on the 7q31 locus, the MET proto‐oncogene encodes a receptor tyrosine kinase which exhibits specificity for a single known high affinity ligand, hepatocyte growth factor (HGF) (Nakamura et al., 1989; Park et al., 1987; Stoker et al., 1987). HGF binding to MET leads to receptor dimerization and auto‐phosphorylation on multiple tyrosine residues within the intracellular kinase domain, resulting in subsequent phosphorylation of the juxtamembrane domain and C‐terminal docking sites (Gherardi et al., 2012). These phosphorylation events enable a striking diversity of cellular responses through activation of multiple downstream effector proteins (such as the adaptor proteins Grb2 and Gab1) (Birchmeier et al., 2003; Weidner et al., 1996), leading to activation of the PI‐3‐kinase, Ras/RAF/MEK/ERK, PLCγ, STAT and FAK signaling pathways. Such signaling allows MET to regulate cell growth, migration, invasion, proliferation and angiogenesis (Humphrey et al., 1995; Matsumoto and Nakamura, 1996).

In normal tissues, MET expression is tightly regulated in cells of epithelial origin (Prat et al., 1991) however, dysregulation of the MET signaling pathway occurs in a wide range of human epithelial cancers including lung, colorectal, breast, pancreatic, ovarian, hepatic and gastric cancers (Di Renzo et al., 1995; Edakuni et al., 2001; Fujita and Sugano, 1997; Humphrey et al., 1995; Inoue et al., 2004; Tsuta et al., 2012). Molecular mechanisms which contribute to this dysregulation are varied and include germline or somatic MET mutation, gene rearrangement, amplification, protein overexpression or changes in ligand‐induced autocrine or paracrine signaling. Indeed, the occurrence of such dysregulating mechanisms often correlates with poor prognosis, as demonstrated for several types of cancer, including gastric (Go et al., 2010; Ichimura et al., 1996; Miyata et al., 2009; Nakajima et al., 1999).

Despite improvements in early diagnosis, surgical techniques and more recently, the uptake of targeted therapies including trastuzumab (Bang et al., 2010), advanced gastric cancer remains the second most common cause of global cancer‐related death with high incidence, relatively poor prognosis and limited treatment options (Jemal et al., 2011). Eastern Asia in particular (notably China, Japan and Korea), suffers from a high incidence of gastric cancer due in part to dietary factors, smoking and the high prevalence of Helicobacter pylori infection (Naylor et al., 2006; Parkin, 2006). Taken together, these data provide a compelling rationale for targeting of the HGF/MET signaling pathway as a therapeutic strategy in multiple tumor types, and especially in gastric cancer of Asian origin.

A number of strategies are being explored to therapeutically inhibit c‐Met activity, including c‐Met or HGF‐specific antibodies and small molecule tyrosine kinase inhibitors. In the latter category, a major challenge to the development of selective ATP‐competitive inhibitors has been the high degree of sequence similarity within the ATP‐binding pockets of canonical protein kinases, and indeed, many current c‐Met targeted agents have relatively promiscuous, mixed pharmacology profiles (recently reviewed in (Scagliotti et al., 2013)). Volitinib represents a novel, potent and highly selective c‐Met small molecule tyrosine kinase inhibitor with favorable preclinical pharmacokinetic and tolerance profiles (Cui et al., 2013; Gu et al., 2013). Volitinib is currently in Phase I clinical trials in China and Australia. A further challenge facing the development of novel agents targeting the cMET signaling pathway concerns the definition of appropriate and accurate biomarker criteria to enable prospective selection of patients. Within gastric cancer specifically, a number of early phase trials have been conducted using cMET tyrosine kinase inhibitors and cMET or HGF‐binding antibodies and unfortunately, despite evidence of clinical responses, none have yet definitively identified robust prospective biomarkers of response (Catenacci et al., 2011; Lennerz et al., 2011; Oliner et al., 2012; Shah et al., 2013). Clinical responses to some of these agents have been documented in patients with tumors harboring cMET gene amplification or cMET protein ‘overexpression’, but consistent data linking scoring criteria to response, or the relationship between cMET gene amplification and protein overexpression, is limited.

In this study we performed a detailed analysis of cMET gene copy number and protein overexpression in a cohort of Chinese gastric cancer patients. We describe one of the first reports of the novel, potent and selective cMET tyrosine kinase inhibitor, volitinib, which was screened across a panel of gastric cancer cell lines and displayed potent anti‐proliferative activity only in cell lines harboring aberrant cMet signaling. More importantly, we established translational significance by demonstrating volitinib anti‐tumor efficacy and pharmacodynamic activity in a panel of cMET‐dysregulated gastric patient‐derived tumor xenograft (PDX) models. In doing so, we provide insight into the relationship between cMET gene amplification and protein expression in gastric cancer and highlight expression thresholds required for preclinical response to volitinib.

2. Materials and methods

2.1. Volitinib

For in vitro studies, volitinib was prepared as a 10 mM DMSO stock solution and diluted in the relevant assay media. For in vivo studies, volitinib was formulated in a 0.5% (v/v) solution of carboxymethylcellulose‐sodium. Animals were given volitinib or vehicle control once daily (qd) by oral gavage.

2.2. Cell culture and in vitro anti‐proliferative cell panel screening

Cell lines were obtained from the American Type Culture Collection (ATCC), Japanese Collection of Research Bioresources (JHSF) or from internal collections and were maintained as described previously (Xie et al., 2013). All cell lines were genetically tested and authenticated using the StemElite ID system kit (Promega, #G9530) and were not cultured for more than 3 months prior to performing the work described here. All cells were maintained in RPMI1640 media and supplemented with 10% fetal bovine serum with glutamine (all Invitrogen). Exceptions included; Hs746t, IM95m and OCUM‐1, which were maintained in DMEM with 10% fetal bovine serum and glutamine. Cell line proliferation was measured after 72 h using either CellTiter 96^® Aqueous One Solution Cell Proliferation Assay Kit (Promega), or by AlamarBlue^® Proliferation Assay (Invitrogen).

2.3. cMET fluorescence in‐situ hybridization (FISH)

The cMET FISH probe was generated internally by directly labeling BAC (CTD‐2270N20) DNA with Spectrum Red (ENZO, Cat # 02N34‐050). The CEP7‐ Spectrum Green probe (Vysis, Cat #32‐132007) for the centromeric region of chromosome 7 was used as internal control. FISH assays were performed on 4 micron dewaxed and dehydrated FFPE sections. The SpotLight Tissue pretreatment Kit (Invitrogen, Cat #00‐8401) was used for pretreatment (boiled in reagent 1 for ∼15 min then coated with reagent 2 for ∼10 min, minor time adjustments were made for individual samples). Sections and probes were codenatured at 80 °C for 5 min and then hybridized at 37 °C for 48 h. After a quick post wash off process (0.3%NP40/1 × SSC at 75.5 °C for 5 min, twice in 2 × SSC at room temperature for 2 min), sections were finally mounted with 0.3 μg/ml DAPI (Vector, Cat #H‐1200), and stored at 4 °C avoiding light for at least 30 min prior to scoring. cMET gene and CEP7 signals were observed using a fluorescence microscope equipped with the appropriate filters allowing visualization of the intense red cMET gene signals, the intense green chromosome 7 centromere signals, and the blue counterstained nuclei. Enumeration of the cMET gene and chromosome 7 was conducted by microscopic examination of 50 tumor nuclei, which yielded a ratio of cMET to CEP7. cMET gene copy number classifications based on FISH assay data were defined as follows; ‘disomy’ – >90% of tumor cells containing 2 copies or less, ‘low trisomy’ – 10–39% of tumor cells containing 3 copies and <10% containing ≥4 copies, ‘high trisomy’ – ≥40% of tumor cells containing 3 copies and <10% containing ≥4 copies, ‘low polysomy’ – 10–39% of tumor cells containing ≥4 copies, ‘high polysomy’ – ≥40% of tumor cells containing ≥4 copies, ‘gene amplification’ – cMET/CEP7 probe ratio of ≥2 or cluster signals in >10% of tumor cells.

2.4. Protein expression analysis

Hs746t cells were treated with volitinib, PF‐4217903 (an unrelated selective cMET inhibitor) or DMSO control for 1 h at 37 °C. Cells were lysed using lysis buffer (Cell Signaling Technology, 9803) and lysates then centrifuged at 12000 rpm for 10 min at 4 °C. Supernatants were mixed with 5 × SDS loading buffer at 1:4 (lysate:SDS) and boiled at 100 °C for 10 min. The mixtures were centrifuged at 12,000 rpm for 5 min and supernatants collected for immunoblotting. Supernatants were loaded into each well and subjected to SDS‐PAGE. Proteins were transferred onto nitrocellulose membranes using the iBlot^® 7‐Minute Blotting System (Invitrogen). The membranes were incubated in blocking solution containing 5% skimmed milk powder plus 0.5% Tween 20 in TBS at room temperature for 1 h. Membranes were then incubated overnight at 4 °C with the appropriate primary antibodies diluted in blocking solution. Membranes were then incubated with the appropriate IRDye^®800CW conjugated secondary antibodies (LI‐COR biosciences) (1/3000 dilution in the blocking solution) at room temperature for 1 h. Western blot bands were detected and visualized using the Odyssey System (LI‐COR Biosciences). Anti‐p‐Met^Y1234/1235 (3129), anti‐p‐Met^Y1349 (3121), anti‐Met (3127), anti‐p‐Akt (4060), anti‐Akt (4691), anti‐p‐Erk1/2 (9106), anti‐Erk (4695), were all purchased from Cell Signaling Technology (Beverly, MA).

2.5. Phospho‐cMET ELISA assay

Harvested xenograft tumor tissue fragments (100 mg each) were processed in 1.0 mL cold lysis buffer (Cell Signaling Technology, Cat#9803) containing 1 mM PMSF (Bio Basic Inc., Lot: LJ0304B409C) by homogenization at 28,000 rpm for 6 s twice (Homogenizers, FA25, Fluko). Tumor lysates were spun down at 12,000 rpm for 15 min twice (Centrifuge: 5417R, Eppendorf), and supernatants transferred to 1.5 mL eppendorf tubes and stored on ice prior to ELISA assay. Total‐Met (R&D Systems, DYC358) and phospho‐Met ELISA assays (R&D Systems, DYC2480) were performed according to the manufacturer's instructions.

2.6. Immunohistochemistry

For IHC detection of total‐cMET, a rabbit monoclonal anti‐total MET antibody (SP44, Ventana Medical Systems, Roche) was used on an automatic immunostainer (Discovery XT, Ventana Medical Systems, Roche, Tucson, AZ, USA). IHC scoring criteria for total cMET protein using the SP44 antibody were as follows; ≥10% of tumor cells showing membrane or cytoplasmic staining with a) strong intensity were scored as 3+ (at 2.5–5× magnification), b) weak to moderate intensity were scored as 2+ (at 10–20× magnification), c) faint intensity were scored as 1+ (at 40× magnification), d) no staining was scored as 0 (at 40× magnification). Herein, cMET protein “overexpression” was defined as IHC scores of 2+ or 3+. For IHC detection of phospho‐cMET, antigen retrieval was performed on formalin‐fixed, paraffin‐embedded tissues for 5 min in high pH EDTA buffer (Dako), followed by washing in running tap water for 5 min 3 μm sections were rinsed in TBST, and incubated with endogenous peroxidase block on a LabVision autostainer for 10 min. Slides were washed twice with TBST and then incubated with rabbit monoclonal anti‐phospho‐cMET (Y1234/1235) antibody (Cell Signaling) for 60 min at room temperature and finally washed twice in TBST. Detection and visualization were performed using goat anti‐rabbit horseradish‐peroxidase polymer (DAKO) and diaminobenzidine (K3468; Dako), respectively. Following IHC study, slides were counterstained and sealed by following standard procedures. Baseline levels of total and phospho‐cMET were scored and interpreted by a qualified pathologist. Quantification of phospho‐cMET was performed using the Aperio imaging system (Leica).

2.7. In vivo anti‐tumor studies

For efficacy studies using the Hs746t cell line, 4 × 10⁶ cells were mixed with 50% matrigel and implanted subcutaneously into nude mice. For efficacy studies with patient‐derived tumor xenograft (PDX) mouse models, a panel of gastric cancer PDX mouse models were established by directly implanting fresh surgical tumor tissues into immunodeficient mice. Briefly, PDX tissue fragments (approximately 15 mm³) were implanted subcutaneously via Trocar needle into female nude mice. Female nude (nu/nu) (8–10‐week‐old) mice (Vital River, Beijing, China) were used for PDX model generation and in vivo efficacy studies. All animal experiments were performed in accordance with the guidelines approved by IACUC. Tumor‐bearing mice with a tumor size range of 150–250 mm³ were randomly divided into vehicle control or treatment groups (7–8 animals per group). Animals were dosed orally by gavage. Subcutaneous tumor size and body weight were measured twice weekly. Tumor volumes were calculated by measuring two perpendicular diameters with calipers. Tumor volumes (V) were calculated using the formula: V = (length × [width]²)/2. The percentage of tumor growth inhibition (%TGI = 1 – [changes of tumor volume in treatment group/changes of tumor volume in control group] × 100) was used for the evaluation of anti‐tumor efficacy. For tumor regression, in which the tumor volume following treatment was smaller than the tumor volume at initial dosing, the following equation was used: regression% = 100 × (T0−Ti)/T0. T0 and Ti are the tumor volume in the same group but measured at different time points. T0 is the tumor volume on the day before first dosing of the compound, and Ti is the tumor volume at the last measurement day after compound treatment. Statistical significance was evaluated using a one‐tailed, two sample t test. P ＜ 0.05 was considered statistically significant.

3. Results

3.1. Potent volitinib in vitro anti‐proliferative activity correlates with dysregulation of cMET expression and pharmacodynamic activity in a gastric cancer cell line panel screen

Volitinib potently inhibits the tyrosine kinase activity of recombinant cMET in vitro with an IC₅₀ value of 4 nmol/L. In vitro drug selectivity was examined against a diverse panel of representative human kinases and >200‐fold selectivity was established versus 267 kinases (<30% inhibition at 1 μM volitinib) (Supplementary Table 1). To assess sensitivity to volitinib, an in vitro anti‐proliferative screen was performed using a comprehensive gastric tumor cell line panel described previously (Xie et al., 2013). Cellular proliferation over a 3 day period was assessed in the presence of drug using a standard metabolism‐based proliferation assay and pEC₅₀ values determined (Figure 1A). Of 22 cell lines screened, 6 showed relative sensitivity to volitinib (EC₅₀ values < 100 nM), and included SNU‐5, SNU‐638, Hs746T, SNU‐620, GTL16 and IM95m (EC₅₀ values of 0.6, 0.8, 1.3, 3.5, 5.6 and 14.7 nM, respectively). In vitro potencies in sensitive cell lines were compared between volitinib and PF‐04217903, a selective cMET inhibitor also undergoing clinical evaluation (Cui et al., 2012) (Supplementary Table 2). In parallel, arrayCGH analysis and cMET immunohistochemical staining of cell pellets was performed to derive cMET gene copy number and protein expression data. Notably, an excellent correlation was observed between high cMET protein expression and volitinib anti‐proliferative sensitivity, with 5 out of the 6 sensitive cell lines (SNU‐5, SNU‐638, Hs746t, SNU‐620 and GTL16) all staining strongly for cMET protein expression (IHC 3+). Interestingly, elevated cMET gene copy number also showed a reasonable correlation with strong cMET protein expression, with 4 of the 5 strongly expressing cell lines harboring an average of 15–41 copies of the cMET gene. In 5 of the 6 cMET IHC 3+ cell lines, protein staining was exclusively localized to the cytoplasm and membrane. For the single remaining IHC 3+ cell line (SNU‐668) and all other cell lines (IHC ≤ 2+), cMET staining was either absent or mainly nuclear and cytoplasmic (representative cMET FISH and IHC images are shown in Supplementary Figure 1). Despite both low cMET gene copy number and negligible protein staining in cell line IM95m (GCN = 3, IHC 0), sensitivity to volitinib was a likely consequence of high level HGF ligand expression, as previously reported by Iwai et al.(Iwai et al., 2003). All other cell lines had low cMET gene copy number (2–5 copies) and moderate/no cMET protein staining (IHC 2+ or less) and displayed minimal responses to volitinib with EC₅₀ values of ∼1 μM or greater.

Figure 1.

Potent volitinib in vitro anti‐proliferative activity correlates with dysregulation of cMET expression and pharmacodynamic activity in a gastric cancer cell line panel screen. A, 22 gastric cancer cell lines were tabulated alongside the volitinib 50% effective inhibitory concentration (EC50), cMET gene copy number (GCN), cMET protein immunohistochemistry score and predominant cellular staining location. Shaded rows denote the most volitinib‐sensitive cell lines. CM: Cytoplasmic/membraneous staining, CN: Cytoplasmic/nuclear staining, None: No staining. B, Hs746t cells were treated with 0.1 μM volitinib or PF‐4217903 (an unrelated selective cMET inhibitor) for 1 h and cell lysates prepared for Western blot analysis using anti‐sera to the proteins denoted. Note – the predicted molecular weight of the mature form of cMET is 145 kDa.

To assess pharmacodynamic modulation of cMET‐signaling using volitinib in vitro, the cMET dysregulated cell line, Hs746t, was chosen for further study. FISH analysis and IHC staining confirmed cMET amplification and protein overexpression compared to a normal control line (AZ521) (Supplementary Figure 1). Pharmacodynamic modulation of cMET‐signaling was assessed following a 1hr incubation with volitinib and subsequent Western blotting with anti‐sera raised against total and phosphorylated forms of cMET, ERK1/2 and AKT, all well established markers of cMET activation and downstream signaling. Clear inhibition of all three phosphorylated markers was observed in Hs746T cell lysates following a 100 nM dose of volitinib (Figure 1B). Thus, volitinib treatment likely exerts an anti‐proliferative effect in ‘oncogene addicted’ cMET‐dysregulated Hs746T cells through inhibition of phospho‐cMET and downstream signaling through ERK and AKT pathways.

3.2. cMET gene amplification and protein overexpression frequently co‐occur in Chinese gastric cancer

Previous data from our lab confirmed a cMET gene amplification incidence of 6.1% and protein overexpression (defined as IHC score ≥ 2+) in 12.3% of Chinese gastric cancer patient tumors (Liu et al., 2014). A full description of clinicopathological parameters and biomarker correlation analyses are also provided within Liu et al. In order to more fully explore the association of a range of cMET gene copy numbers with protein expression, we undertook a more detailed FISH analysis of our samples. This analysis revealed a full spectrum of tumor cMET gene copy number changes, from disomy, low to high trisomy, low to high polysomy through to gene amplification (tumor incidences were 18%, 29%, 6%, 33%, 8% and 6%, respectively) (Table 1). Similarly, more detailed immunohistochemical (IHC) analysis was performed using a specific, validated antibody to the cMET protein however, the availability of sufficient tissue limited this study to 170 samples. Whilst the majority of tumor samples (87%) stained negative or low for cMET protein (IHC score 0 or 1+), a significant proportion (13%) showed moderate to high intensity membranous cMET staining (IHC score 2+ or 3+) (Table 2). Representative cMET FISH and cMET IHC images are shown in Supplementary Figure 2. Analysis of both FISH and IHC score data was performed to address the concordance of cMET gene copy number with protein expression (Table 3). Notably, 10/12 (83%) of cMET amplified tumor samples also showed cMET IHC 3+ positivity and interestingly, a further 3 cases also stained cMET IHC 3+ but were found to be non‐amplified for cMET. This data highlights that the majority of cMET amplified tumors show high level cMET protein expression, but that cMET protein overexpression can also occur in the absence of gene amplification.

Table 1.

cMET gene copy number changes and protein overexpression frequently co‐occur in Chinese gastric cancer. 195 gastric tumor patient samples were analyzed for cMET gene copy number changes using FISH assay.

cMET FISH classification	Patient samples, n (%)
Disomy	35 (18)
Low trisomy	56 (29)
High trisomy	12 (6)
Low polysomy	65 (33)
High polysomy	15 (8)
Gene amplification	12 (6)
Total	195 (100)

Table 2.

170 of the 195 samples were stained for cMET protein expression using immunohistochemistry.

cMET IHC score	Patient samples, n (%)
0	143 (84)
1+	6 (3)
2+	8 (5)
3+	13 (8)
Total	170 (100)

Table 3.

Tabulated data correlating cMET FISH classification with cMET IHC score on a dataset common to 170 gastric tumor samples. Details of FISH and IHC scoring criteria are listed in ‘Materials and methods’.

	cMET FISH classifications
	Disomy	Low trisomy	High trisomy	Low polysomy	High polysomy	Amplification
MET IHC
0, n (%)	29 (96.7)	42 (87.5)	9 (81.2)	50 (89.3)	12 (92.3)	1 (8.3)
1+, n (%)	0 (0)	3 (6.3)	1 (9.1)	2 (3.6)	0 (0)	0 (0)
2+, n (%)	1 (3.3)	3 (6.3)	0 (0)	2 (3.6)	1 (7.7)	1 (8.3)
3+, n (%)	0 (0)	0 (0)	1 (9.1)	2 (3.6)	0 (0)	10 (83.3)
Total, n	30	48	11	56	13	12

3.3. Volitinib is highly efficacious and shows pharmacodynamic modulation of cMET signaling in standard cMET‐dysregulated and patient‐derived tumor xenograft models

To evaluate volitinib preclinical efficacy, we performed an in vivo anti‐tumor efficacy study using the Hs746t xenograft model. Mice bearing established tumors were treated once‐daily, orally for 16 days with 0.3, 1 or 2.5 mg/kg doses of volitinib. Dose‐responsive anti‐tumor efficacy was observed in this study with tumor growth inhibition (TGI) values of 17% (NS), 74% (P < 0.001) and 97% (P < 0.001), obtained using volitinib doses of 0.3, 1 and 2.5 mg/kg, respectively (Figure 2A).

Figure 2.

Volitinib treatment results in anti‐tumor efficacy and pharmacodynamic modulation of cMET signaling in the cMET‐driven xenograft model, Hs746t. A, Volitinib was administered by oral gavage once daily (qd) to nu/nu mice bearing established s.c. Hs746t xenografts at the doses indicated (7–8 mice/treatment group). Tumor volumes are plotted against time. ** – P ≤ 0.001. B Pharmacodynamic modulation of tumor phospho‐cMET was established by harvesting volitinib‐treated Hs746t tumor fragments at the times and doses indicated and performing ELISA assay on tumor lysates. % inhibition values were calculated with the formula, inhibition rate (IR, %) = (1−OD450drug treated/OD450vehicle) × 100%.

Next, we employed an ELISA assay to assess modulation of tumor phospho‐cMET levels following volitinib administration to Hs746T tumor‐bearing mice. Tumors were harvested at 4 and 8 h‐post a single oral dose of volitinib and processed according to the manufacturer's instruction. Dosage of 0.3, 1 and 2.5 mg/kg volitinib resulted in 63%, 81% and 94% reductions in levels of tumor phospho‐cMET respectively at 4 h, and 39%, 76% and 88% at 8 h (Figure 2B). Thus, potent dose‐responsive pharmacodynamic modulation of tumor phospho‐cMET levels correlates with anti‐tumor efficacy in xenograft model Hs746t.

In order to test the hypothesis that use of a selective cMET pharmacological inhibitor could offer therapeutic benefit to GC patients harboring cMET‐dysregulated tumors, we established a panel of 34 gastric cancer primary xenograft (PDX) models derived directly from patient tumor material. Tissue sections from these models were characterized using cMET FISH and IHC analysis. Of these, 3 models (SGC071, SGC184 and SGC141) were found to harbor cMET gene amplification and high‐level cMET protein expression (denoted as “IHC 3+”) (Figure 3A – upper 2 panels). Model SGC070 served as a negative control (low polysomy) with moderate cMET protein expression but undetectable phospho‐cMET levels, indicative of pathway inactivity. These PDX models were characterized and used between passages 4–7 and showed relative homogeneity of both cMET gene copy number and cMET protein expression across multiple intratumoral tissue sections. Models had the following cMET:CEP7 ratios and cMET average gene copy numbers; SGC071 – >2, 4, SGC184 – >10, >20, SGC141 – >10, >50, SGC070 – 1.2, 3.5. Furthermore, these models tested negative for Her2 positivity (defined as Her2 amplified or IHC 3+ using the Herceptest kit) and FGFR2 expression (data not shown). No models were identified with a cMET IHC 3+, non‐amplified cMET profile.

Figure 3.

Volitinib displays antitumor efficacy in short duration efficacy studies using cMET‐dysregulated gastric cancer patient‐derived xenograft models. A, cMET gene and protein expression levels were characterized in all 4 gastric PDX models using FISH and IHC assays (top 2 panels). Models SGC071, SGC184 and SGC141 are all cMET‐amplified and strongly express cMET protein (IHC 3+). Model SGC070 shows cMET low polysomy and variable staining of cMET protein (IHC1+/3+). Pharmacodynamic modulation of phospho‐cMET was assessed by comparing phospho‐cMET IHC staining at baseline (pre‐dose) and 2 h after volitinib dosing (post‐dose) (lower 2 panels). B, Volitinib was administered by oral gavage once daily (qd) to nu/nu mice bearing established patient‐derived gastric tumor fragments at the doses indicated (7–8 mice/treatment group). Tumor volumes are plotted against time. ** – P ≤ 0.001. NS – Not significant (P ≥ 0.05). For FISH images, magnification is ×60. Scale bars represent 10 μm for FISH images and 50 μm for IHC.

Tumor‐bearing mice were randomly grouped and dosed orally, once daily with vehicle, 1 mg/kg or 10 mg/kg volitinib for a period of 14–21 days. Volitinib demonstrated significant anti‐tumor efficacy within all 3 cMET‐amplified PDX models in short term duration in vivo efficacy studies (Figure 3B). At volitinib doses of 1 and 10 mg/kg, TGI values were 52% and 84% (SGC071), 39% and 92% (SGC184), and 73% and 90% (SGC141) – all values were significant (P < 0.001). Within each study, xenograft tissue was removed 2 h following the last dose of 10 mg/kg volitinib and processed for IHC analysis using total and phospho‐cMET antibodies. In each of the volitinib sensitive models, high baseline levels of total cMET (IHC 3+) correlated with activated phospho‐cMET and treatment with volitinib resulted in a clear, qualitative reduction in tumor phospho‐cMET staining (Figure 3A – lower 2 panels).

In contrast, volitinib treatment did not result in significant tumor growth inhibition in the cMET ‘normal’ model, SGC070 (non‐significant TGI values of 18% and 21% at 1 and 10 mg/kg, respectively). Despite moderate expression of total cMET protein (IHC 2+), cMET pathway activation (as determined by phospho‐cMET IHC staining) was undetectable. In all of the anti‐tumor efficacy studies presented here, treatment with volitinib was well tolerated and did not result in significant body weight loss or other obvious side effects, consistent with the preclinical toxicology profile of volitinib (manuscript in preparation).

3.4. Combined volitinib and docetaxel therapy results in efficacy benefit in cMET dysregulated xenograft models

As taxanes are a relatively effective and commonly used class of chemotherapeutics in the clinical treatment of gastric cancer, we sought to perform a comparison of both volitinib mono‐ and combination therapy with docetaxel. The cMET‐dysregulated models, Hs746t (standard xenograft) and SGC184 (PDX), were chosen for this analysis. Mice bearing Hs746t tumors were dosed for 14 days with vehicle, oral once‐daily volitinib (0.6 mg/kg), intravenous once‐weekly docetaxel (3 mg/kg) or a combination of volitinib/docetaxel (same dose and schedule as monotherapy groups). Monotherapy dosing with volitinib or docetaxel resulted in 56% and 81% TGI, respectively, whilst combined dosing achieved tumor regression with 101% TGI (all P values < 0.001) (Figure 4A). Within the SGC184 study, mice were dosed for 22 days with vehicle, oral once‐daily volitinib (10 mg/kg), intravenous once‐weekly docetaxel (10 mg/kg) or a combination of volitinib/docetaxel (same dose and schedule as monotherapy groups). Monotherapy dosing with volitinib or docetaxel resulted in 86% and 101% TGI, respectively, whilst combined dosing achieved tumor regression with 160% TGI (all P values < 0.0001, with the exception of the docetaxel vs docetaxel/volitinib group, P < 0.05) (Figure 4B). In all cases, dosing combinations were well tolerated (maximum 8% body weight loss, data not shown) and combination benefit was achieved in both studies. Thus, both in primary and standard preclinical xenograft models, volitinib can be combined with one of the most commonly used standard of care chemotherapies to generate superior anti‐tumor efficacy at well tolerated doses of each drug.

Figure 4.

Volitinib enhances the efficacy of docetaxel in both standard and patient‐derived cMET‐driven xenograft models. A, Volitinib, docetaxel or a combination of the two were administered according to the administration routes and schedules noted above to nu/nu mice bearing established Hs746t tumors at the doses indicated. Tumor volumes are plotted against time. B, As for A, but using PDX model SGC184. qd, once daily; qw, once weekly; p.o, oral dosing; i.v., intravenous dosing. ** – P ≤ 0.001. * – P ≤ 0.05.

4. Discussion

Strong preclinical evidence exists to demonstrate roles for cMET in the processes of cell growth, survival, invasion and angiogenesis, and dysregulation of cMET pathway signaling has been shown to occur in a variety of human cancers with correlations to poor clinical outcomes and drug resistance. Accordingly, over the past decade the cMET/HGF axis has attracted significant drug discovery and development investment and the early clinical signals obtained using targeted agents such as onartuzumab, rilotumumab and crizotinib, in selected patient populations, are largely positive. Within gastric cancer however, despite a number of studies, there still remains a need to accurately characterize cMET gene copy number and protein overexpression incidence and to establish translational relevance. Within this study, we further characterized cMET gene copy number and protein overexpression within a cohort of 170 Chinese gastric cancers and moreover, used the novel, selective and potent cMET TKI inhibitor, volitinib, to demonstrate in vitro and in vivo correlations between cMET‐dysregulation and drug sensitivity. To our knowledge, we are the first to extend the translational significance of targeting the cMET pathway in GC by employing GC PDX models.

Previous attempts to characterize cMET gene amplification and protein expression in gastric cancers have generated highly variable data (0–23% gene amplification (Hara et al., 1998; Janjigian et al., 2011; Kuniyasu et al., 1992; Nakajima et al., 1999; Tsugawa et al., 1998; Tsujimoto et al., 1997)), and 24–71% IHC positivity (Huang et al., 2001; Lee et al., 2012; Nakajima et al., 1999), likely a consequence of differences in tumor biology, geographical sampling and assay methodologies. This uncertainty led us to generate our own cMET FISH and IHC data on a cohort of Chinese gastric tumor samples (Liu et al., 2014), with further analysis conducted herein. Compared to Southern blot, FISH methodology is technically more standardized, less affected by tissue variables and although relatively labor‐intensive, represents the ‘gold standard’ approach to gene copy number determination. We also employed immunohistochemistry to enable quantitation of protein levels as the routine applicability of this technology in the clinic, together with cost effectiveness and industry expertise, are major advantages. Further analysis of this dataset showed incidences of 8% and 6% cMET high polysomy and gene amplification respectively, and 12% protein overexpression (defined as IHC 2+ or IHC 3+). Therefore, using near identical methodologies and antibody reagents, our data are broadly in agreement with that of Lee et al. who describe incidences of 19% high polysomy/amplification and 24% overexpression (also defined as IHC 2+ or 3+) respectively, in a Korean GC population (Lee et al., 2012).

Following an accurate definition of cMET gene amplification and protein overexpression incidence, translational studies were conducted using volitinib. Volitinib (HMPL‐504), is a potent and highly selective cMET small molecule tyrosine kinase inhibitor with robust pharmacokinetics and an excellent preclinical tolerance profile. As such, volitinib represents a promising clinical agent which ensures pharmacokinetic flexibility with a selectivity and toxicology profile which offers the opportunity to combine with other therapeutic agents, whilst providing the convenience of oral dosing. Gastric cancer cell line panel data confirmed a strong correlation between cMET expression and volitinib anti‐proliferative activity, thus emphasizing the highly selective nature of volitinib. Our data demonstrated potent, dose‐responsive volitinib anti‐tumor efficacy in a cMET‐dysregulated tumor xenograft model, Hs746t, with corresponding pharmacodynamic modulation of phospho‐cMet. In order to overcome the limitations imposed by cell line‐derived xenograft models, we employed a panel of GC PDX models, which by their nature, offer superior patient tumor modeling fidelity in terms of tissue architecture, genetic heterogeneity, stromal interactions and potentially, better predictive value. Importantly, using short term duration in vivo efficacy studies, volitinib demonstrated significant anti‐tumor efficacy only in cMET‐amplified PDX models, thus again, highlighting the selectivity and efficacy of this agent. Interestingly, use of a common dose of 1 mg/kg volitinib across both xenograft and PDX models generally resulted in slightly lower anti‐tumor efficacy in the PDX models. Although this is currently a focus for ongoing investigation, we speculate that the relative heterogeneity of tumor cMET expression could be a plausible explanation for this observation.

To date, a number of early clinical trials have been conducted within the gastric cancer setting using novel agents targeting cMET signaling. Foretinib (GSK1363089), a mixed cMET and VEGFR2 inhibitor was tested using 2 different schedules (intermittent and daily) within a total of 74 patients (Shah et al., 2013). Using RECIST criteria, the best observed response was stable disease (SD), which occurred in 10 patients. 3 patients were identified with tumor cMET amplification (FISH MET:CEP7 ratio ≥ 2) but showed no evidence of response (1 SD, 2 progressive disease (PD)). It is unclear from this study whether lack of response in the cMET amplified tumors was attributable to the drug (eg. insufficient exposure, incorrect scheduling) or the tumor itself (lack of cMET protein expression/heterogeneity or other oncogenic ‘drivers’), or perhaps a combination of the two. Crizotinib (Xalkori), was initially developed as a cMET inhibitor but gained FDA‐approval on the basis of its ALK activity in selected patients with non‐small cell lung cancer. In a phase I expansion trial, 4 gastroesophageal patients with cMET amplified tumors (FISH MET:CEP7 ratio ≥ 2.2 across 50 tumor nuclei) were treated with crizotinib; 2 showing rapid progression, 1 with a confirmed PR (41% reduction) and 1 showing stable disease (Lennerz et al., 2011). Unfortunately again, no corresponding analysis of cMET protein expression in the patient biopsies was performed. Therapeutic antibodies targeting HGF (Rilotumumab) and cMET (Onartuzumab) have also shown promising results in early trials. In a phase II trial enrolling patients with gastric or gastroesophageal junction (GEJ) tumors, rilotumumab treatment combined with chemotherapy demonstrated an improvement in overall survival of around 5 months, compared to chemotherapy alone (Oliner et al., 2012). Importantly, this benefit was only observed in patients with cMET “high” tumors (defined as >50% of cells with IHC ≥1+ cMET staining) and indeed, patients with MET “low” tumors actually appeared to do worse on the rilotumumab/chemotherapy combination compared to chemotherapy/placebo. Interestingly, retrospective analyses of tumor cMET gene copy number, baseline plasma levels or soluble MET did not correlate with overall survival or progression free survival. In the case of onartuzumab, a single case study reported a patient with chemorefractory metastatic gastric cancer to the liver, who was enrolled onto a phase I trial (Catenacci et al., 2011). The primary tumor had high cMET gene polysomy (average cMET gene copy number ≥4 in 60% cells), ‘detectable’ cMET expression by IHC and evidence of autocrine HGF production. The patient received 4 doses of onartuzumab (once every 3 weeks), at which time a complete response (CR) was declared. The CR lasted 2 years at which time, multiple new metastatic foci were detected.

Notably, the majority of these novel agents are now deploying enrollment strategies in later phase gastric trials designed to enrich for patients bearing tumors with high level cMET expression (either IHC‐based, FISH‐based or both), based on preceding data identifying these patient populations as the most likely to benefit from treatment (NCT01324479, NCT01697072, NCT01662869). Thus, the application of an accurate and robust patient stratification approach appears essential to ensuring the delivery of clinical benefit through the use of cMET inhibitory agents. As noted above, although the thresholds of response to these agents are likely to differ (due to differing drug modes of action, exposures and potencies), there is still a lack of consistent data comparing cMET gene expression levels with protein expression and pharmacological response. Analysis of our GC patient tumor samples revealed a positive correlation between cMET gene amplification and protein expression, as 83% (10/12) of the gene amplified tumor samples also scored IHC 3+ for protein expression. Interestingly however, there were an additional 3 IHC 3+ cases which were not found to harbor cMET gene amplification and which therefore suggest that cMET protein expression may also be controlled by mechanisms other than gene copy number increase (eg. transcriptional or post‐translational). Our cell line panel, xenograft and PDX model data largely confirmed this correlation between cMET gene copy elevation and protein expression and extended this to pharmacological response to volitinib in vitro and in vivo. Although we were unable to test volitinib anti‐tumor efficacy in a cMET non‐amplified, IHC 3+ model (none were identified), we speculate that tumors with this profile would be likely to respond to a cMET inhibitor, in a manner analogous to that of HER2 IHC3+ (but not HER2 amplified) gastric cancers which respond to trastuzumab (Bang et al., 2010). Conversely, despite an 8% incidence of cMET high polysomy in tumors within our patient cohort, we showed no significant correlation with protein overexpression (12/13 cMET high polysomy tumors scored IHC 0 and 1/13 scored IHC 2+). Furthermore, although we were unable to identify a cMET high polysomy PDX model, volitinib showed negligible anti‐tumor efficacy in a cMET low polysomy (IHC 2+) model (model SGC070, 18% TGI at 1 mg/kg volitinib, P = ns). Indeed, parallels can be drawn again with the HER2 pathway as breast tumors with increased HER2 gene copy number (as a result of chromosome 17 polysomy) are known to behave as HER2‐negative tumors, both in terms of HER2 protein expression and disease prognosis (Vanden Bempt et al., 2008). Based on these findings above, we therefore speculate that cMET high polysomy gastric tumors would be unlikely to respond to cMET inhibitor therapy due to a lack of protein expression and dependency on cMET pathway signaling. Taken together, the clear correlations observed here between elevated cMET gene copy number, protein expression and preclinical efficacy in response to volitinib, suggest that MET IHC could be a pragmatic primary screening test for patient selection using anti‐cMET therapy.

In conclusion, we have investigated cMET gene copy number and protein overexpression in a cohort of Chinese GC patients using ‘gold standard’ approaches and importantly, have revealed associations between these parameters and responsiveness to the potent and selective cMET inhibitor, volitinib, both in vitro and in vivo. Significantly, our use of GC PDX models strengthens the translational relevance of these findings and highlights the crucial importance of understanding and applying accurate patient selection biomarker criteria with an appropriate diagnostic technology platform. Overall, our data underscore the potential therapeutic utility of volitinib and cMET pathway inhibition as a promising treatment option in appropriately selected gastric cancer patients.

Grant support

This research was conducted and funded by AstraZeneca and Hutchison MediPharma.

Disclosure of potential conflicts of interest

All authors are full‐time employees of AstraZeneca or Hutchison Medi Pharma Ltd.

Supporting information

The following are the supplementary data related to this article:

Supplementary Figure 1 Representative cMET FISH and IHC images of gastric cancer cell lines. Cell pellets were prepared and stained as described in ‘Materials and methods’. For FISH images, CEP7 centromeric probe signals are shown in green and cMET probe signals in red. Magnification is ×60. For IHC, staining was performed using a validated and specific cMET IHC antibody, SP44 (Ventana Medical Systems Roche).

Supplementary Figure 2 Corresponding representative FISH and IHC assay images of patient gastric tumor specimens. For cMET FISH, CEP7 centromeric probe signals are shown in green and cMET probe signals in red. Magnification is ×60. For cMET IHC, staining was performed using a validated and specific cMET IHC antibody, SP44 (Ventana Medical Systems Roche).

Supplementary Table 1 Volitinib exhibits a highly selective enzyme inhibition profile. Volitinib was tested against a panel of 267 diverse human kinase enzymes (listed) and showed <30% inhibition at a concentration of 1 μM. All recombinant human enzymes were tested in reactions using ATP concentrations at or near the Km value for the specific enzyme, using a 33P‐ATP radiolabeling reaction.

Supplementary Table 2

Supplementary data 1.

1.1.

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.molonc.2014.08.015.

Gavine Paul R., Ren Yongxin, Han Lu, Lv Jing, Fan Shiming, Zhang Wei, Xu Wen, Liu Yuan Jie, Zhang Tianwei, Fu Haihua, Yu Yongjuan, Wang Huiying, Xu Shirlian, Zhou Feng, Su Xinying, Yin XiaoLu, Xie Liang, Wang Linfang, Qing Weiguo, Jiao Longxian, Su Weiguo, Wang Q. May, (2015), Volitinib, a potent and highly selective c‐Met inhibitor, effectively blocks c‐Met signaling and growth in c‐MET amplified gastric cancer patient‐derived tumor xenograft models, Molecular Oncology, 9, doi: 10.1016/j.molonc.2014.08.015.