The Vascular Endothelial Growth Factor Receptor Tyrosine Kinase

Home , Angiogenesis inhibitor, Pazopanib, Vatalanib

Published OnlineFirst June 22, 2010; DOI: 10.1158/1078-0432.CCR-10-0232 Published OnlineFirst on June 22, 2010 as 10.1158/1078-0432.CCR-10-0232

Cancer Therapy: Preclinical Clinical Cancer Research The Vascular Endothelial Growth Factor Receptor Tyrosine Kinase Inhibitors Vatalanib and Pazopanib Potently Induce Apoptosis in Chronic Lymphocytic Leukemia Cells In vitro and In vivo

Julian Paesler, Iris Gehrke, Rajesh Kumar Gandhirajan, Alexandra Filipovich, Magdalena Hertweck, Felix Erdfelder, Sabrina Uhrmacher, Simon Jonas Poll-Wolbeck, Michael Hallek, and Karl-Anton Kreuzer

Abstract Purpose: There is evidence that vascular endothelial growth factor (VEGF) is a critical microenviron- mental factor that exerts angiogenesis-independent effects on the survival of chronic lymphocytic leukemia (CLL) cells. Vatalanib and pazopanib are potent orally available VEGF receptor tyrosine kinase inhibitors. We investigated the efficacy and selectivity of both compounds in CLL cells, simulated potential combination with conventional cytostatics, and tested the effect of both substances on CLL-like tumor xenografts. Experimental Design: Primary CLL and normal peripheral blood cells were tested for viability after incubation with varying concentrations of both inhibitors. Further, phosphorylation status of VEGF receptor on treatment, caspase activation, and poly(ADP-ribose) polymerase cleavage were assessed. Com- binations of each inhibitor with fludarabine, vincristine, and doxorubicin were analyzed for possible synergistic effects in vitro. For in vivo testing, mice grafted with the CLL-like cell line JVM-3 were treated orally with each inhibitor. Results: Vatalanib and pazopanib decreased phosphorylation of the VEGF receptor, along with induction of apoptosis in CLL cells in clinically achievable concentrations. Healthy B cells were only mildly affected. Immunoblots showed downregulation of the antiapoptotic proteins XIAP and MCL1, whereas poly(ADP-ribose) polymerase cleavage was increased. Combinations with conventional cytostatic agents resulted in synergistic effects. Treatment of xenografted mice with 100 mg/kg of body weight for 21 days resulted in tumor inhibition rates of 76% (vatalanib) and 77% (pazopanib). In two mice, a total tumor eradication could be observed. No gross systemic toxicity occurred. Conclusion: We conclude that VEGF inhibition is a promising new therapeutic approach in CLL. Va- talanib and pazopanib seem to be effective and safe candidates to be further evaluated for this purpose. Clin Cancer Res; 16(13); 3390–8. ©2010 AACR.

Chronic lymphocytic leukemia (CLL) is the most fre- and CD23+ B cells with poor self-regeneration (1). Rather, quent type of leukemia in adults in North America and their accumulation results from a defect in the initiation of Western Europe. Biologically, CLL is characterized by an their regular cell death through apoptosis (2). Accumula- accumulation of mature but dysfunctional CD5+, CD19+, ting neoplastic CLL cells replace other healthy blood com- ponents, which results in symptoms of this disease. CLL Authors' Affiliation: Department I of Internal Medicine, University at patients show a very diverse clinical course. The reasons Cologne, Cologne, Germany may be differences in genetic predisposition or molecular Note: Supplementary data for this article are available at Clinical Cancer causes, and continue to be a subject of scientific research. Research Online (http://clincancerres.aacrjournals.org/). The molecular background of the resistance of CLL cells to Parts of the results were presented at the 51st annual meeting of the apoptosis, the so-called apoptotic block, is a major focus American Society of Hematology, New Orleans, LA, December 5 to 8, 2009. of this research as it may reveal target structures for specific therapies. J. Paesler and I. Gehrke contributed equally to this work. Vascular endothelial growth factor-A (VEGF-A) has been Corresponding Author: Karl-Anton Kreuzer, Department I of Internal Medicine, University at Cologne, Kerpener Strasse 62, 50937 Cologne, discussed to play an angiogenic-independent role in CLL Germany. Phone: 49-221-478-97382; Fax: 49-221-478-97383; cells by preventing apoptosis in these cells. VEGF-A, com- E-mail: [email protected]. monly referred to as VEGF, belongs to the superfamily of doi: 10.1158/1078-0432.CCR-10-0232 platelet-derived growth factors and exhibits several ©2010 American Association for Cancer Research. isoforms through alternative splicing, with VEGF121,

3390 Clin Cancer Res; 16(13) July 1, 2010

VEGF RTKIs in CLL

tions showed a positive correlation between the level Translational Relevance of VEGF receptors on the cell surface and shorter survival in CLL patients (18) there is a single study reporting Basic research has shown that vascular endothelial the duration of progression-free survival tending to be growth factor (VEGF) signaling is active in chronic shorter in patients with low intracellular VEGF levels; lymphocytic leukemia (CLL) and is likely to play a crit- however, this was not statistically significant (19). It ical role in CLL cell-microenvironment interactions in could further be shown in biopsies that the density of this disease. Two new VEGF receptor tyrosine kinase vessels in the bone marrow of CLL patients was greater inhibitors, vatalanib and pazopanib, have been identi- than that in healthy individuals, and vessel density in fied and proven to be safe in humans. To translate bone marrow was positively correlated with disease pro- these facts into a possible clinical application, we show gression (20). There is also evidence for angiogenesis- that these substances possess a direct cytotoxic effect independent effects of VEGF in CLL. Thus, it was shown on CLL cells, which is mediated through regular apo- that exogenic recombinant VEGF165 can reduce the rate ptotic signaling. Low concentrations of each drug com- of spontaneous as well as chemotherapy-induced bined with conventional cytostatics like fludarabine apoptosis in CLL cells in vitro (21). Simultaneously, a resulted in an additive effect. Further, we show that significant upregulation of antiapoptotic proteins such both compounds lead to a substantial tumor reduction as myeloid cell leukemia-1 (MCL1) and X-linked inhib- in a CLL-like xenograft nude mouse model, whereas no itor of apoptosis (XIAP) was found on VEGF stimulation gross systemic toxicity could be observed. Our findings (21). SU11657, a receptor tyrosine kinase inhibitor may therefore serve as a rationale to test vatalanib and (RTKI), was found to reduce the phosphorylation of pazopanib, either alone or combined with convention- VEGF-R1 and VEGFR-R2 as well as trigger apoptosis al cytostatics, in human CLL. in CLL cells in vitro (21, 22). However, the authors speculated that this compound does not have sufficient activity in vivo to work in clinical trials for CLL patients (21). VEGF165, VEGF186, VEGF189, and VEGF206 being the most Vatalanib (an anilino phthalazine) and pazopanib (an common ones (3–5). VEGF is regarded as the most impor- indazolyl pyramidine) are potent inhibitors of VEGF-R. tant proangiogenic factor; however, its biological func- Both substances inhibit the VEGF receptor tyrosine kinase tions range from the regulation of embryonic vascular in the submicromolar range (IC50,VEGF-R1,54and development to the enhancement of vascular permeability 10 nmol/L, respectively; VEGF-R2, 39 and 30 nmol/L, (6, 7). Further, with respect to a nonangiogenic role, VEGF respectively). At higher concentrations, they also inhibit was shown to be a substantial factor for the survival of he- other tyrosine kinases such as platelet-derived growth fac- matopoietic stem cells (8). tor and c-Kit (9, 23). For pazopanib, a steady-state concen- Thus far, three VEGF receptors were described: VEGF- tration of ≥40 μmol/L is required for optimal in vivo R1 or fms-like receptor tyrosine kinase (Flt-1), VEGF-R2 activity against VEGF-R2. This in vivo-in vitro discrepancy or fetal liver kinase-1 (Flk-1), or kinase-insert domain can be attributed at least in part to a high protein binding containing receptor (KDR) and VEGF-R3 or Flt-4. VEGF of the substances (24). Vatalanib has a half-life of receptors homodimerize on ligand binding, which results 5.3 hours. However, its area under the curve (AUC) and in transphosphorylation of their intracellular kinase half-life are reduced along with treatment duration (AUC domains and subsequently to activation of signaling −50% on day 14), most likely through enzyme induction pathways (9). Under physiologic conditions, VEGF recep- (9). The half-life of pazopanib (35 hours) is markedly tors are mainly located on vascular, lymphatic, and endo- longer (9). thelial cells, but have also been shown in several Vatalanib is currently being tested in various clinical nonendothelial cells. VEGF receptors were further de- studies for malignant diseases. In a phase III trial, pazopa- tected on several cancer cells, such as breast, colon, and nib showed promising results in metastatic renal cell car- pancreatic cancer cells. Under functional aspects, VEGF- cinoma and is currently being tested in a number of phase R1 is mainly known for cell differentiation and migra- II trials. Side effects observed with either substance were tion; VEGF-R2 regulates cell proliferation, growth, and generally mild, including nausea and vomiting, diarrhea, survival; whereas VEGF-R3 is associated with lymphan- hypertension, weakness, and dizziness (9, 25). In mouse giogenesis (2, 10–14). models, with daily doses of 100 mg/kg of body weight, CLL cells spontaneously secrete VEGF, primarily the chronic serum levels of ≥10 μmol/L vatalanib were isoforms VEGF165 and VEGF121 (15, 16). Further, they achieved and were well tolerated (26). Both compounds constitutively express activated VEGF-R1 and VEGF-R2 have the advantage of being orally available, strongly (16). Several correlations have been described between increasing their applicability. VEGF levels in serum or plasma and CLL progression Based on the above-mentioned findings, we speculated or disease stage. In early stages, a positive association that vatalanib and pazopanib can induce apoptosis in was found between VEGF levels in serum and disease CLL cells at clinically achievable concentrations. We progression (17). Although the majority of investiga- further intended to investigate both compounds as

www.aacrjournals.org Clin Cancer Res; 16(13) July 1, 2010 3391

Paesler et al.

additional drugs for combination with conventional marked Annexin V, which binds to phosphatidylserine cytostatics. and with propidium iodide (PI). The cells were then classified in a flow cytometer (BD FACS Canto). Living Materials and Methods cells are Annexin V FITC/PI double negative; early apoptotic cells are Annexin-FITC positive/PI negative, due to Patient samples and purification of lymphocytes exposure of phosphatidylserine on their cell surfaces; and All patients had been informed about the use of their Annexin V/PI double-positive cells are considered dead blood for the purposes of the study and had provided cells because they are no longer capable of extruding their consent. Patients were either untreated or did not PI. As in previous time course experiments, we observed receive treatment for the last 3 months before blood a maximum apoptosis rate 24 hours after we performed sampling. The study was conducted in accordance with subsequent investigations using this time point. the Declaration of Helsinki (6th revision, Seoul/South Korea, 2008) and was approved by the local ethics com- Immunoblots mittee. B cells were purified by negative selection using Cell extracts of treated primary CLL cells were pre- RossetteSep (StemCell Technologies, Inc.). This cocktail pared using the M-PER mammalian protein extraction contains antibodies directed against surface markers of reagent (Pierce Thermo Scientific), including Complete non–B cells (CD2, CD3, CD16, CD36, CD56, CD66b, Protease Inhibitor (Roche Diagnostics) and PhosStop and glycophorinA). The antibodies cross-link non–B Phosphatase Inhibitor (Roche Diagnostics) cocktail ta- cells to multiple red blood cells, which are pelleted blets. Protein concentration was determined according along with free red blood cells during density gradient to Bradford's method, using Roti Nanoquant (Carl Roth centrifugation. CLL or healthy B cells were assimilated GmbH). For denaturation of proteins, lysates were in RPMI 1640 (Biochrom AG) with 20% fetal calf serum boiled at 90°C for 10 minutes in lithium dodecyl (FCS), 1% HEPES buffer (Invitrogen GmbH), and sulfate sample buffer (Invitrogen). The proteins were 1% penicillin/streptomycin (Biochrom AG) at 37°C, separated on NuPage Novex 4% to 12% Bis-TRIS gels, 5% CO2 in a humidified atmosphere at a density of transferred to a nitrocellulose membrane, and incubated 5×105/mL. with the primary antibodies [β-actin, XIAP, cleaved poly(ADP-ribose) polymerase (PARP), BD Bioscience; Cell lines MCL1, Santa Cruz Biotechnologies]. Horseradish perox- We obtained fresh JVM-3 cells from the German collec- idase (HRP)–marked immunoglobulins from DAKO tion of microorganisms and cell cultures (DSMZ). Due to Company were used as secondary antibodies. Antibody broad testing at the DSMZ before shipment, no further cell binding was detected using a chemiluminiscence line verification was done in our laboratory. These cells reagent (ECL Western blotting detection reagent, were cultured in RPMI 1640 (Biochrom AG) with Amersham). 20% FCS, 1% HEPES buffer (Gibco Invitrogen), and 1% penicillin/streptomycin (Biochrom AG) at 37°C, VEGF receptor phosphorylation status 5% CO2 in a humidified atmosphere. Purified CLL cells were incubated in RPMI for 24 hours with 10 μmol/L of vatalanib or pazopanib. Af- VEGF RTKIs and chemotherapeutic agents ter 24 hours, cells were fixed in 4% ultrapure formalde- Vatalanib (PTK787) was provided by BayerSchering, hyde (Polysciences, Inc.) for 20 minutes. After washing Inc., and pazopanib (GW786034B) was provided by twice with PBS, 0.5% FCS cells were permeabilized in GlaxoSmithKline, Inc. For in vitro experiments, vatalanib 2 mL of ice-cold methanol (Roth GmbH). Fixed and and pazopanib were dissolved at a concentration of permeabilized cells were washed twice and incubated 10 mmol/L in DMSO (Roth GmbH + Co KG). Conven- for 1 hour with a primary antibody against phospho- tional chemotherapeutic agents were also dissolved in VEGF-R2 (Tyr951; Cell Signaling Technology Inc.). After DMSO to buffered solutions with 1 mmol/L fludarabine, repeated washing steps, cells were incubated for 1 hour 10 μmol/L vincristine, and 10 mmol/L doxorubicin. For with a goat anti-mouse IgG-A488 antibody (Invitrogen). in vivo experiments, both substances were suspended in After washing, fluorescence intensity was measured an aqueous 2% carboxymethylcellulose solution (Sigma through flow cytometry on a FACSCanto (BD Biosci- Aldrich) with 0.1% polysorbate 20 (Caesar & Loretz ence). Flow cytometry data were analyzed with Cyflogic GmbH). The pure aqueous solution served as vehicle software (CyFlo Ltd.). control. Pan-caspase inhibitor Apoptosis assay CLL cells were treated for 24 hours either with 100 μmol/L Purified CLL cells were incubated in RPMI for different vatalanib or pazopanib alone or in combination with time periods and with different concentrations of vatala- 50 μmol/L of the pan-caspase inhibitor Z-VAD-FMK (Pro- nib and pazopanib. Samples were treated according to mega) added 30 minutes before the RTKIs. The percentage the manufacturer's protocol (Apoptosis Detection Kit, of Annexin-FITC/PI double-negative cells was determined BD Pharmingen) with fluorescein isothiocyanate (FITC)- through flow cytometry.

3392 Clin Cancer Res; 16(13) July 1, 2010 Clinical Cancer Research

VEGF RTKIs in CLL

Xenograft mouse model Results Four-week-old female BALB/c nu/nu mice (Charles River Laboratories, Inc.) were irradiated with 3 Gy and Vatalanib and pazopanib induce dose-dependent and inoculated s.c. with 1 × 107 JVM-3 cells in the posterior selective apoptosis in CLL cells flank. When the tumor volume reached about 100 mm3, To test the effect of vatalanib and pazopanib on the survival mice were divided into three groups of 10 mice each. Ani- of CLL cells, CLL cells were incubated for 24 hours with each mals were treated with 100 mg/kg of body weight vatala- substance at concentrations of 1 to 200 μmol/L. The percentage nib, pazopanib, or aqueous 2% carboxymethylcellulose of living cells was determined using flow cytometry. A dose- solution with 0.1% polysorbate 20 (vehicle control) per dependent reduction in the number of living CLL cells was ob- day orally through a gastric tube. Tumor volume (TV) served in all samples. The determined 50% lethal concentration was determined every 2nd day with a caliper using the for- (LC50) was 48.4 μmol/L for vatalanib and 32.7 μmol/L for mula TV = 1/2 × length (mm) × height (mm) × breadth pazopanib. To determine the selectivity of the substances toward (mm). Mice were inspected visually for occurrence of ob- CLL cells, B cells of healthy donors (n = 5) and CLL cells were vious side effects. As part of the inspection of their general similarly treated with vatalanib or pazopanib at concentrations condition, mice were weighed once every week. After the of 10, 50, and 100 μmol/L for 24 hours. Although both last treatment, the liver, kidneys, and spleen were removed RTKI had a significant cell death–inducing effect on CLL cells, for the purpose of investigation. normal B cells were significantly less affected (Fig. 1B). Even at a concentration of 100 μmol/L, 80.5 ± 4.5% of B cells were alive, Statistical analysis whereasthatwasthecaseforonly35.1±4.3%ofCLLcells(P ≤ Statistical analysis was done using the program Graph- 0.0001). Treatment with pazopanib yielded similar results: Only Pad Prism (GraphPad Software). If not mentioned other- 26.5 ± 3.4% of CLL cells survived 24-hour treatment with wise, unpaired two-sided Student's t test was used. 100 μmol/L pazopanib; however, for healthy B cells, the values Differences of P < 0.05 were considered to be significant. were still as high as 89.1 ± 2.2% (P ≤ 0.0001).

Fig. 1. Vatalanib and pazopanib induce dose-dependent and selective apoptosis in CLL cells. A, primary CLL samples were cultured in the presence of 1, 10, 25, 50, 75, 100, and 200 μmol/L vatalanib (left) or pazopanib (right) for 24 hours. Survival was measured via Annexin/PI–based fluorescence-activated cell sorting (FACS) assay. The determined LC50 of vatalanib is 48.4 μmol/L while that of pazopanib is 32.7 μmol/L (n = 24 and n = 19, respectively). B, primary CLL cells (n = 24 for vatalanib/n = 19 for pazopanib) and healthy B cells (n = 5) were treated for 24 hours with 10, 50, and 100 μmol/L vatalanib (left) and pazopanib (right). White columns, CLL samples; black columns, healthy B cells. Survival was assessed by Annexin/PI–based FACS assay. Data are shown as mean ± SEM. P values indicate significance and were calculated using Student's t test. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001.

www.aacrjournals.org Clin Cancer Res; 16(13) July 1, 2010 3393

Paesler et al.

of CLL cells (75.8% survival in DMSO treated control), caspase inhibition protected CLL cells from vatalanib- mediated cell death, as seen by a survival rate of 65.6%. The same effect could be observed with pazopanib, where RTKI treatment alone lead to 17.4% surviving CLL cells, whereas pretreatment with V-VAD-FMK significantly increased CLL cell survival (54.2%).

Combination of VEGF RTKI with conventional cytostatics shows synergistic effects To investigate potential synergistic effects, CLL cells were treated for 24 hours with lower doses of vatalanib or pazopanib (10 μmol/L) alone or in combination with classic chemotherapeutic agents such as fludarabine (10 μmol/L), Fig. 2. Phosphorylation of the VEGF receptor (VEGFR) decreases after μ μ RTKI treatment. CLL cells were treated for 24 hours with 10 μmol/L vincristine (0.1 mol/L), and doxorubicin (10 mol/L). of vatalanib and pazopanib, and the phosphorylation status of VEGFR-2 The combinations of vatalanib with fludarabine, vincris- (Tyr951) was determined by flow cytometry. This representative sample tine, and doxorubicin as well as the combination of demonstrates the decrease in phosphorylated VEGF receptor after pazopanib with fludarabine were associated with treatment with both RTKIs (n = 5). marked and significantly higher rates of apoptosis (Fig. 5). VEGF receptor phosphorylation status decreases after Combined treatment of vatalanib together with cyto- RTKI treatment statics decreased the survival rate an average of 26.2% To check whether vatalanib and pazopanib actually act compared with fludarabine, vincristine, and doxorubicin through inhibition of VEGF signaling, we determined the alone (Fig. 5A). phosphorylation status of the VEGF receptor in CLL sam- The analogous experiment with pazopanib showed sig- ples after 24 hours of treatment with 10 μmol/L vatalanib nificant additive effects when pazopanib was combined and pazopanib compared with a DMSO-treated control group (n = 5). Figure 2 shows a clear shift toward a decreased phosphorylation of VEGF-R2 (Tyr951) in the treated samples. In concordance to previously published data, healthy B cells do not exhibit VEGF receptors (data not shown, ref. 24).

VEGF RTKI induce downregulation of antiapoptotic proteins To further determine the mechanism of apoptosis induction by the VEGF RTKIs, CLL cells were incubated for 24 hours with vatalanib (Fig. 3A) or pazopanib (Fig. 3B) at concentrations ranging between 0.1 and 100 μmol/L. Further, untreated and DMSO (1%)–treated controls were added. Lysates were analyzed through immunoblotting for apoptotic markers. β-Actin served as loading control. Va- talanib and pazopanib induce concentration-dependent PARP cleavage along with reduced levels of the antiapoptotic proteins XIAP and MCL1.

Vatalanib- and pazopanib-mediated apoptosis is caspase dependent To determine the role of caspases in vatalanib- and pazopanib-induced apoptosis, we incubated CLL cells with Fig. 3. both substances at high concentrations of 100 μmol/L Vatalanib and pazopanib induce concentration-dependent cleavage of PARP and downregulation of antiapoptotic proteins. alone or in combination with the pan-caspase inhibitor Whole-cell lysates were analyzed after 24-hour incubation of CLL cells Z-VAD-FMK at 50 μmol/L (n = 3) for 24 hours. Pretreat- with 0.1, 1, 10, and 100 μmol/L vatalanib (A) or pazopanib (B) by ment with the pan-caspase inhibitor resulted in a signifi- immunoblotting. Untreated samples (UTC) and samples treated with 1% cantly reduced apoptosis rate on RTKI treatment. In Fig. 4, DMSO served as control. The blots illustrate the concentration- dependent decrease of the antiapoptotic proteins XIAP and MCL1 and a representative sample is depicted. Although vatalanib the increase of cleaved PARP in CLL cells after exposure to the RTKIs. treatment alone resulted in a survival rate of only 6.3% One representative blot out of three experiments is shown.

3394 Clin Cancer Res; 16(13) July 1, 2010 Clinical Cancer Research

VEGF RTKIs in CLL

Fig. 4. Vatalanib and pazopanib act caspase dependently. Primary CLL cells were cultured for 24 hours with 100 μmol/L of vatalanib or pazopanib in the absence or presence of the pan caspase inhibitor Z-VAD-FMK (50 μmol/L). Survival was measured using Annexin V–FITC/PI staining and flow cytometry. Annexin/PI double- negative cells (Q3) were considered alive, Annexin-positive/ PI-negative cells (Q4) were considered apoptotic, and Annexin/PI double-positive cells (Q2) were considered dead. Cell death induced by vatalanib and pazopanib was significantly reduced when Z-VAD-FMK was present (right column). One representative experiment out of three is shown.

with fludarabine, with a survival rate of 75.2% after pazo- earlier. Before treatment, mice were split into three groups panib treatment alone and 81.3% after fludarabine treat- with the following mean tumor volumes: 112.0 ± 15.8 mm3 ment alone, and only 52.7% survival after combined in the control group, 112.9 ± 13.3 mm3 in the group to be treatment (P = 0.042; Fig. 5B). Vincristine and doxirubicin- treated with vatalanib, and 113.3 ± 18.7 mm3 in the pazo- induced cell death were slightly increased when pazopanib panib group. Mice were treated daily through gastric tube. was added: 53.0 ± 9.3% survival after vincristine alone As early as day 12, mean tumor volume was significantly versus 35.2 ± 5.5% survival after combined treatment with reduced in the pazopanib-treated group compared with vincristine and pazopanib; 59.1 ± 8.0% survival after the vehicle-treated group, with mean tumor volumes of doxorubicin alone versus 45.0 ± 7.4% survival after com- 446.4 ± 144.5 mm3 versus 663.9 ± 180.1 mm3 (P = bined treatment with doxorubicin and pazopanib. Never- 0.0098), respectively. On day 14, the tumor volume was al- theless, no statistical significance could be achieved by so significantly reduced in the group treated with vatalanib this combinations. with a mean tumor volume of 684.9 ± 167.6 mm3 compared with 1,003.1 ± 226.8 mm3 (P = 0.027) in control- Vatalanib and pazopanib inhibit tumor growth in a treated animals (Fig. 6). xenograft mouse model After completion of the treatment period of 21 days, the Both substances were further tested in the JVM-3 xeno- mean tumor volume was 2,458.3 ± 782.5 mm3 in the graft nude mouse model. Mice were generated as explained control group, 671.8 ± 208.7 mm3 (P = 0.002) in the group

www.aacrjournals.org Clin Cancer Res; 16(13) July 1, 2010 3395

Paesler et al.

treated with vatalanib, and 645.0 ± 252.3 mm3 (P = 0.002) After 3 weeks of treatment, two animals of the vatalanib- in the group treated with pazopanib. During the treatment treated group did not show a tumor anymore. period of 21 days mice showed no signs of pain, disability, Analysis of the body weight of mice revealed no signi- or lethargy. Tumor inhibition rates were calculated accord- ficant differences between the control-treated group and ing to the following formula (27): % inhibition = 1 − the animals treated with vatalanib or pazopanib. All ani- (mean tumor growth in the treated group/mean tumor mals showed a constant increase of body weight of growth in the control group). ∼12.5% during the 3 weeks of treatment (Supplementary The tumor inhibition rate was 36% for vatalanib and Table 1). 49% for pazopanib as early as day 14 (Supplementary Table S1). At the last day of treatment (day 21), tumor in- Discussion hibition rates were increased to 76% in the vatalanib- treated group and 77% in the pazopanib-treated group. It has previously been shown that exogenous VEGF can substantially reduce the spontaneous apoptosis rate in CLL cells (22). In our studies we registered that VEGF originated from the bone marrow-derived stromal cell line HS5 improved CLL cell survival in culture. This effect could be reversed by adding a VEGF-neutralizing antibody.1 Furthermore, it was found that model substances with an inhibitory effect on VEGF receptor tyrosine kinases could trigger apoptosis in CLL cells in vitro (22). To what extent these compounds are tolerated by healthy B cells remains unknown. For a potential therapeutic approach, however, the selectivity of the substances to tumor cells is of crucial importance. In our in vitro experiments, we were able to show that the VEGF RTKIs vatalanib and pazopanib, whose pharmacokinetics in solid tumors have been well described, trigger apoptosis in CLL cells in a concentration-dependent manner. Healthy B cells were much less influenced by these orally available drugs that are highly selective for VEGF-R1 and VEGF-R2. This is in agreement with the results achieved by Podar et al.: Healthy peripheral blood mononuclear cells tolerated the treatment with pazopanib at concentrations of up to 20 μg/mL (≈43 μmol/L) much better than multiple myeloma cell lines (25). The fact that the effect of the substances on cell survival is dependent on the induction of apoptosis and is caspase dependent was shown by (a) flow cytometric Annexin/PI staining, (b) immunoblots for PARP cleavage as a direct target of caspase 3, and (c) the effect of the pan-caspase inhibitor Z-VAD-FMK. Downregulation of antiapoptotic proteins such as XIAP and MCL1 after treatment with VEGF RTKIs observed here is in agreement with the observations of other authors who used multikinase inhibitors (21, 22). The published literature provides examples of clinically achievable serum concentrations of both VEGF RTKIs used here: ≥40 μmol/L for pazopanib after once-daily administration of ≥800 mg (28) and 15.8 ± 9.5 μmol/L for vata- Fig. 5. Coincubation of CLL samples with the RTKIs and lanib after once-daily administration of 1,000 mg (29). chemotherapeutics showed an additive effect. Primary CLL cells were From our observations, comparable concentrations were incubated with 10 μmol/L of vatalanib (n = 7) or pazopanib (n = 5), 10 μmol/L fludarabine, 0.100 μmol/L vincristine, 10 μmol/L doxorubicin sufficient to markedly reduce the percentage of living alone, or in the indicated combinations for 24 hours. Survival was CLL cells, whereas healthy B cells were significantly less af- – assessed by Annexin V FITC/PI staining and flow cytometry. Additive fected in this concentration range. These data and the LC50 effects were achieved when vatalanib was combined with either of the values obtained by flow cytometry could be confirmed by chemotherapeutics (A), whereas pazopanib only showed a significant additive effect when combined with fludarabine (B). P values indicate significance and were calculated using Student's t test; ***, P ≤ 0.001; *, P ≤ 0.05; n.s., not significant. 1 Gehrke et al., submitted for publication.

3396 Clin Cancer Res; 16(13) July 1, 2010 Clinical Cancer Research

VEGF RTKIs in CLL

Fig. 6. Vatalanib and pazopanib treatment impedes the growth of JVM-3 xenografts. Vatalanib or pazopanib were given at 100 mg/kg of body weight orally to immunocompromised mice engrafted with the human chronic B-cell leukemia–like cell line JVM-3 (n = 10). For the tumor xenografts, both substances were suspended in an aqueous 2% carboxymethylcellulose solution with 0.1% polysorbate 20. The pure aqueous solution was used as vehicle control. Vatalanib significantly reduced tumor volume starting from day 14, while tumor size was significantly decreased by pazopanib treatment compared with vehicle-treated control as early as day 12. Tumor inhibition (IR) on the last day of treatment was 76% for vatalanib and 77% for pazopanib. Error bars, SEM. P values indicate significance and were calculated using Wilcoxon's matched-paired test; **, P ≤ 0.01; *, P ≤ 0.05.

another ATP-based cell viability assay (ATP Glow, Pro- ease, and due to a lacking alternative are regarded as a mega; data not shown). In multiple myeloma cell lines, model at least related to CLL (27, 30). We tested both in- Podar et al. (25) achieved comparable LC50 values of 5 hibitors, vatalanib and pazopanib, in these xenograft to 15 μg/mL (≈10–30 μmol/L) for pazopanib. Antipro- mouse models. After oral administration of vatalanib liferative effects were registered for vatalanib in multiple and pazopanib, the mean tumor volume was reduced to myeloma cell lines somehow earlier in concentrations only 26% and 27% of the tumor volume in the control ranging from 1 to 5 μmol/L (25). group after 21 days of continuous treatment, respectively. Thus far in the treatment of solid tumors, the use of According to the formula (28) % inhibition = 1 − (mean VEGF RTKIs alone was moderately successful. Thus, com- tumor group in the treated group/mean tumor growth in bination therapies with classic chemotherapeutic agents the control group), this equals a tumor inhibition rate of have been established in this field. It is assumed that VEGF 76% for vatalanib and 77% for pazopanib. Similar inhi- RTKIs remodel the vascular supply of tumors and thus bition rates were achieved with pazopanib, administered make them more accessible to chemotherapeutic agents. twice daily at 100 mg/kg of body weight for 21 days in According to our hypothesis, VEGF RTKIs exert an angio- xenografts of other tumor cell lines: 82%, 64%, and genesis-independent effect in CLL; however, in a study fo- 53% for HT29 colon, A375P melanoma, and PC3 pros- cusing on the potential effect of pazopanib for the tate, respectively, as shown by Kumar et al. (28). In the treatment of multiple myeloma, Podar et al. showed that same study, tumor inhibition rates of 64%, 95%, and combining pazopanib with chemotherapeutic agents may 99% could be achieved by once-daily administration of potentiate the effects in hematologic diseases as well (25). 100 mg/kg of body weight pazopanib for 21 days in We registered a synergistic effect when vatalanib was ad- Caki-2 renal, BT474 breast, H322 lung xenografts, respec- ministered with classic chemotherapeutic agents such as tively. Thus, the inhibition rates we registered after once- fludarabine, vincristine, and doxorubicin. Pazopanib daily administration of the drugs were in the mid range. It showed a statistically significant synergistic effect in com- has to be highlighted that in two animals, complete tu- bination with fludarabine. mor eradication could be observed. However, because To confirm the in vitro data in an in vivo model, the CLL usually presents as leukemic rather than as a tumor CLL-like cell line JVM-3 was xenografted into irradiated disease, it cannot be excluded that in addition to the di- nude mice. The JVM-3 cell line was originally derived rect cytotoxic effects an antiangiogenic mechanism might from a patient with B-prolymphocytic leukemia and have contributed to the treatment response. Certainly, it might therefore at first not be considered a convincing would also be desirable to combine VEGF RTKI with con- CLL mouse model. However, these cells exhibit a Matutes ventional cytostatics in our animal model. However, due score of 3, which, although not sufficient for the diagnosis to the i.p. toxicity of vincristine and doxorubicin, this can- of CLL, suggests a certain biological similarity to this dis- not be done.

www.aacrjournals.org Clin Cancer Res; 16(13) July 1, 2010 3397

Paesler et al.

In conclusion, vatalanib and pazopanib were well Disclosure of Potential Conflicts of Interest tolerated in our CLL xenograft mouse model. Both compounds exerted a selective effect in vitro and in vivo,and No potential conflicts of interest were disclosed. justify the expectation of a good risk-benefit ratio in humans as well. The availability of oral pharmaceutical Acknowledgments

forms and the long terminal half-life of the drugs are pro- We thank BayerSchering and GlaxoSmithKline for providing us with the bably associated with good compliance and permit eco- RTKIs vatalanib and pazopanib, respectively, and Samir Tawadros for nomic outpatient treatment. technical assistance with the mouse experiments. The costs of publication of this article were defrayed in part by the Summing up, our observations provide evidence that payment of page charges. This article must therefore be hereby marked selective inhibition of the VEGF signaling pathway by advertisement in accordance with 18 U.S.C. Section 1734 solely to vatalanib or pazopanib, either alone or in combination indicate this fact. with existing and future therapies, could be a promising Received 01/27/2010; revised 03/25/2010; accepted 04/26/2010; approach for further improvement of CLL therapy. published OnlineFirst 06/22/2010.

References 1. Dameshek W. Chronic lymphocytic leukemia—an accumulative dis- levels of vascular endothelial growth factor predict risk of progres- ease of immunologically incompetent lymphocytes. Blood 1967;29: sion in early B-cell chronic lymphocytic leukaemia. Br J Haematol 566–84. 1999;107:605–10. 2. Shalaby F, Rossant J, Yamaguchi TP, et al. Failure of blood-island 18. Ferrajoli A, Manshouri T, Estrov Z, et al. High levels of vascular formation and vasculogenesis in Flk-1-deficient mice. Nature 1995; endothelial growth factor receptor-2 correlate with shortened sur- 376:62–6. vival in chronic lymphocytic leukemia. Clin Cancer Res 2001;7: 3. Houck KA, Ferrara N, Winer J, Cachianes G, Li B, Leung DW. The 795–9. vascular endothelial growth factor family: identification of a fourth 19. Aguayo A, O'Brien S, Keating M, et al. Clinical relevance of intracel- molecular species and characterization of alternative splicing of lular vascular endothelial growth factor levels in B-cell chronic lym- RNA. Mol Endocrinol;5:1806–14. phocytic leukemia. Blood 2000;96:768–70. 4. Maglione D, Guerriero V, Viglietto G, li-Bovi P, Persico MG. Isolation of 20. Kini AR, Kay NE, Peterson LC. Increased bone marrow angiogenesis a human placenta cDNA coding for a protein related to the vascular in B cell chronic lymphocytic leukemia. Leukemia 2000;14:1414–8. permeability factor. Proc Natl Acad Sci U S A 1991;88:9267–71. 21. Lee YK, Bone ND, Strege AK, Shanafelt TD, Jelinek DF, Kay NE. 5. Leung DW, Cachianes G, Kuang WJ, Goeddel DV, Ferrara N. Vascu- VEGF receptor phosphorylation status and apoptosis is modulated lar endothelial growth factor is a secreted angiogenic mitogen. Sci- by a green tea component, epigallocatechin-3-gallate (EGCG), in ence 1989;246:1306–9. B-cell chronic lymphocytic leukemia. Blood 2004;104:788–94. 6. Bikfalvi ABR. Recent advances in angiogenesis, anti-angiogenesis 22. Lee YK, Shanafelt TD, Bone ND, Strege AK, Jelinek DF, Kay NE. and vascular targeting. Trends Pharmacol Sci 2002;23:576–82. VEGF receptors on chronic lymphocytic leukemia (CLL) B cells inter- 7. Gerber HP, Dixit V, Ferrara N. Vascular endothelial growth factor in- act with STAT 1 and 3: implication for apoptosis resistance. Leuke- duces expression of the antiapoptotic proteins Bcl-2 and A1 in vas- mia 2005;19:513–23. cular endothelial cells. J Biol Chem 1998;273:13313–6. 23. Lin B, Podar K, Gupta D, et al. The vascular endothelial growth factor 8. Gerber HP, Malik AK, Solar GP, et al. VEGF regulates haematopoietic receptor tyrosine kinase inhibitor PTK787/ZK222584 inhibits growth stem cell survival by an internal autocrine loop mechanism. Nature and migration of multiple myeloma cells in the bone marrow micro- 2002;417:954–8. environment. Cancer Res 2002;62:5019–26. 9. Mross K. Inhibition of the VEGF receptor system with tyrosine kinase 24. Sonpavde G, Hutson TE, Sternberg CN. Pazopanib, a potent orally inhibitors. Angiogenesis inhibition in oncology. Pharm Unserer Zeit administered small-molecule multitargeted tyrosine kinase inhibitor 2008;37:404–15. for renal cell carcinoma. Expert Opin Investig Drugs 2008;17:253–61. 10. Hattori K, Heissig B, Wu Y, et al. Placental growth factor reconsti- 25. Podar K, Tonon G, Sattler M, et al. The small-molecule VEGF recep- tutes hematopoiesis by recruiting VEGFR1(+) stem cells from tor inhibitor pazopanib (GW786034B) targets both tumor and endo- bone-marrow microenvironment. Nat Med 2002;8:841–9. thelial cells in multiple myeloma. Proc Natl Acad Sci U S A 2006;103: 11. Fong GH, Rossant J, Gertsenstein M, Breitman ML. Role of the Flt-1 19478–83. receptor tyrosine kinase in regulating the assembly of vascular endo- 26. Wood JM, Bold G, Buchdunger E, et al. PTK787/ZK 222584, a novel thelium. Nature 1995;376:66–70. and potent inhibitor of vascular endothelial growth factor receptor 12. Lohela M, Bry M, Tammela T, Alitalo K. VEGFs and receptors in- tyrosine kinases, impairs vascular endothelial growth factor-induced volved in angiogenesis versus lymphangiogenesis. Curr Opin Cell responses and tumor growth after oral administration. Cancer Res Biol 2009;21:154–65. 2000;60:2178–89. 13. Seetharam L, Gotoh N, Maru Y, Neufeld G, Yamaguchi S, Shibuya M. 27. Loisel S, Le Ster K, Quintin-Roue I, et al. Establishment of a novel A unique signal transduction from FLT tyrosine kinase, a receptor for human B-CLL-like xenograft model in nude mouse. Leuk Res vascular endothelial growth factor VEGF. Oncogene 1995;10:135–47. 2005;29:1347–52. 14. Vaisman N, Gospodarowicz D, Neufeld G. Characterization of the re- 28. Kumar R, Knick VB, Rudolph SK, et al. Pharmacokinetic-pharmaco- ceptors for vascular endothelial growth factor. J Biol Chem 1990; dynamic correlation from mouse to human with pazopanib, a 265:19461–6. multikinase angiogenesis inhibitor with potent antitumor and anti- 15. Chen H, Treweeke AT, West DC, et al. In vitro and in vivo production angiogenic activity. Mol Cancer Ther 2007;6:2012–21. of vascular endothelial growth factor by chronic lymphocytic leuke- 29. Jost LM, Gschwind HP, Jalava T, et al. Metabolism and disposition mia cells. Blood 2000;96:3181–7. of vatalanib (PTK787/ZK-222584) in cancer patients. Drug Metab 16. Kay NE, Bone ND, Tschumper RC, et al. B-CLL cells are capable of Dispos 2006;34:1817–28. synthesis and secretion of both pro- and anti-angiogenic molecules. 30. Matutes E, Owusu Ankomah K, Morilla R, et al. The immunological Leukemia 2002;16:911–9. profile of B-cell disorders and proposal of a scoring system for the 17. Molica S, Vitelli G, Levato D, Gandolfo GM, Liso V. Increased serum diagnosis of CLL. Leukemia 1994;8:1640–5.

3398 Clin Cancer Res; 16(13) July 1, 2010 Clinical Cancer Research

The Vascular Endothelial Growth Factor Receptor Tyrosine Kinase Inhibitors Vatalanib and Pazopanib Potently Induce Apoptosis in Chronic Lymphocytic Leukemia Cells In vitro and In vivo

Julian Paesler, Iris Gehrke, Rajesh Kumar Gandhirajan, et al.

Clin Cancer Res Published OnlineFirst June 22, 2010.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-10-0232

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2010/08/25/1078-0432.CCR-10-0232.DC1

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/early/2010/06/18/1078-0432.CCR-10-0232. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.