`
`Potent VEGF blockade causes regression of coopted
`vessels in a model of neuroblastoma
`
`Eugene S. Kim*, Anna Serur*, Jianzhong Huang*, Christina A. Manley†, Kimberly W. McCrudden*, Jason S. Frischer*,
`Samuel Z. Soffer*, Laurence Ring†, Tamara New†, Stephanie Zabski‡, John S. Rudge‡, Jocelyn Holash‡,
`George D. Yancopoulos‡, Jessica J. Kandel*, and Darrell J. Yamashiro*†§
`
`Divisions of *Pediatric Surgery and †Pediatric Oncology, College of Physicians and Surgeons of Columbia University, New York, NY 10032; and ‡Regeneron
`Pharmaceuticals, Incorporated, 777 Old Saw Mill River Road, Tarrytown, NY 10591
`
`Communicated by P. Roy Vagelos, Merck & Co., Inc., Bedminster, NJ, July 5, 2002 (received for review April 19, 2002)
`
`Vascular endothelial growth factor (VEGF) plays a key role in
`human tumor angiogenesis. We compared the effects of inhibitors
`of VEGF with different specificities in a xenograft model of
`neuroblastoma. Cultured human neuroblastoma NGP-GFP cells
`were implanted intrarenally in nude mice. Three anti-VEGF agents
`were tested: an anti-human VEGF165 RNA-based fluoropyrimidine
`aptamer; a monoclonal anti-human VEGF antibody; and VEGF-Trap,
`a composite decoy receptor based on VEGFR-1 and VEGFR-2 fused
`to an Fc segment of IgG1. A wide range of efficacy was observed,
`with high-dose VEGF-Trap causing the greatest inhibition of tumor
`growth (81% compared with controls). We examined tumor an-
`giogenesis and found that early in tumor formation, cooption of
`host vasculature occurs. We postulate that this coopted vascula-
`ture serves as a source of blood supply during the initial phase of
`tumor growth. Subsequently, control tumors undergo vigorous
`growth and remodeling of vascular networks, which results in
`disappearance of the coopted vessels. However, if VEGF function
`is blocked, cooption of host vessels may persist. Persistent coop-
`tion, therefore, may represent a novel mechanism by which neu-
`roblastoma can partly evade antiangiogenic therapy and may
`explain why experimental neuroblastoma is less susceptible to
`VEGF blockade than a parallel model of Wilms tumor. However,
`more effective VEGF blockade, as achieved by high doses of
`VEGF-Trap, can lead to regression of coopted vascular structures.
`These results demonstrate that cooption of host vasculature is an
`early event in tumor formation, and that persistence of this effect
`is related to the degree of blockade of VEGF activity.
`
`It is well established that the growth and metastasis of solid
`
`tumors depend on angiogenesis. Many studies have shown that
`one mechanism by which tumors induce the formation of new
`blood vessels is expression of proangiogenic factors. The most
`commonly implicated factor is vascular endothelial growth factor
`(VEGF), a specific endothelial cell mitogen, permeability, and
`survival factor, overexpressed in virtually all human tumors (1).
`Antagonism of the VEGF pathway results in inhibition of
`angiogenesis and tumor growth in a number of tumor model
`systems (2).
`In previous studies, we have shown that the degree of efficacy
`of VEGF suppression differs markedly in different experimental
`tumors. Despite similar levels of initial VEGF expression, after
`administration of anti-VEGF antibody, both primary tumor
`growth and metastasis were nearly completely eradicated in
`experimental Wilms tumors, whereas in a parallel model of
`neuroblastoma, tumor growth was only moderately affected with
`persistence of metastasis (3, 4). This difference in response to
`antiangiogenic therapy suggests that other mechanisms may
`support perfusion and tumor growth in neuroblastoma.
`Recruitment or cooption of preexisting blood vessels by
`tumors, as previously described by Holash and colleagues, may
`represent one such mechanism (5–7). Early in tumor develop-
`ment, preexisting vasculature is engaged by growing tumor cells.
`Subsequently, the coopted vessels regress. The tumor becomes
`hypoxic, VEGF expression is up-regulated, and neoangiogenesis
`
`is induced. In other studies, vascular cooption in experimental
`glioblastoma after tumors were exposed to VEGF blockade
`suggests that this mechanism may also be invoked by established
`tumors deprived of VEGF (8, 9).
`We have recently examined vascular cooption in a xenograft
`model of neuroblastoma (10). Neuroblastoma tumor cells im-
`planted into the kidney of nude mice result in large vascular
`tumors. Treatment with an anti-VEGF antibody causes partial
`tumor inhibition, decreased angiogenesis, and by angiography
`results in the appearance of novel rounded structures at vessel
`branches, which we had initially termed ‘‘terminal vascular
`bodies’’ (10). In the current study, we demonstrate that these
`structures are in fact renal glomeruli that have been coopted by
`tumor tissue. This cooption is an early event in tumor growth in
`experimental neuroblastoma, because it is present in both con-
`trol xenografts and tumors treated with the anti-VEGF agents
`[monoclonal anti-VEGF antibody (11) and VEGF-Trap, a com-
`posite decoy receptor based on VEGF receptor-1 (VEGFR-1)
`and VEGFR-2 fused to an Fc segment of IgG1 (12)]. In control
`neuroblastomas, this stage is followed by vascular remodeling in
`conjunction with brisk angiogenesis; coopted glomeruli disap-
`pear. However, in tumors treated with either anti-VEGF anti-
`body or low-dose VEGF-Trap, cooption persists, permitting
`perfusion and ongoing tumor growth. Treatment with high-dose
`VEGF-Trap results in stunted tumors, which are virtually avas-
`cular and in which only rare coopted glomeruli are present,
`suggesting that the coopted vasculature has regressed. These
`studies demonstrate that cooption of host vasculature occurs
`early in neuroblastoma tumor growth. Persistent cooption ap-
`pears to depend on an intermediate degree of VEGF blockade
`and may result in the relative resistance of a specific tumor to
`antiangiogenic therapy, which may be overcome by a higher
`degree of VEGF blockade.
`
`Methods
`Cell Line. The human neuroblastoma cell line NGP, which had
`previously been transfected with a retroviral vector containing
`enhanced green fluorescent protein, was maintained in culture
`in 75-cm2 flasks with McCoy’s 5A medium (Mediatech, Fisher
`Scientific). Medium was supplemented with 10% FBS and 1%
`penicillin兾streptomycin (GIBCO). Cells were grown at 37°C in
`5% CO2 until confluent. NGP-GFP cells were harvested by
`trypsinization, counted with trypan blue staining, and washed
`and resuspended in sterile saline solution (phosphate buffered
`saline, GIBCO) at a concentration of 107 cells兾ml.
`
`Animal Model. All experiments were approved by the Institutional
`Animal Care and Use Committee of Columbia University.
`Female NCR nude mice, 4–6 weeks of age (Taconic Farms),
`
`Abbreviations: VEGF, vascular endothelial growth factor; rVEGF, VEGF receptor; LD, low
`dose; HD, high dose; ␣SMA, ␣-smooth muscle actin; TUNEL, terminal deoxyribonucleic
`d-UTP nick end labeling.
`§To whom reprint requests should be addressed. E-mail: dy39@columbia.edu.
`
`www.pnas.org兾cgi兾doi兾10.1073兾pnas.172398399
`
`PNAS 兩 August 20, 2002 兩 vol. 99 兩 no. 17 兩 11399 –11404
`
`CELLTRION - EXHIBIT 1090
`
`
`
`were housed in a barrier facility and acclimated to 12-hr light兾
`dark cycles for at least 1 day before use.
`
`Tumor Implantation. The left flank was prepared in a sterile
`manner after anesthetizing the mice with i.p. ketamine (50
`mg兾kg) and xylazine (5 mg兾kg). An incision was made exposing
`the left kidney, and an inoculum of 106 NGP-GFP tumor cells in
`0.1 ml of PBS was injected with a 25-g needle. The flank muscles
`were closed with a single 4–0 Polysorb suture (US Surgical,
`Norwalk, CT) and the skin closed with staples.
`
`Antiangiogenic Treatment. There were three sets of experiments,
`each with its own set of control tumors: (i) NX1838, a RNA-
`based fluoropyrimidine polyethylene glycol-conjugated aptamer
`targeting human VEGF165 (NeXstar, Boulder, CO) (13, 14) at
`250 g per dose NX1838 (n ⫽ 5) or vehicle control (0.05%
`mouse serum albumin, Sigma, n ⫽ 5) injected i.p. daily; (ii)
`humanized monoclonal anti-VEGF antibody (A4.6.1, Genen-
`tech) (11), at 100 g兾dose (n ⫽ 19) or vehicle control (n ⫽ 21)
`administered i.p. biweekly; and (iii) 100 g per dose [low dose
`(LD), VEGF-Trap LD, n ⫽ 10], 500 g per dose [high dose
`(HD), VEGF-Trap HD, n ⫽ 10] VEGF-Trap (Regeneron
`Pharmaceuticals, Tarrytown, NY) (12), or vehicle control
`(n ⫽ 21) administered i.p. biweekly. Treatment began 3 or 4 days
`after tumor implantation and continued for 5.5 weeks (6-week
`time point). An additional set of mice (four mice for control,
`anti-VEGF antibody, VEGF-Trap LD, and VEGF-Trap HD)
`were examined after 3.5 weeks of treatment (4-week time point).
`
`Fluorescein Angiograms. Fluorescein angiography for demonstra-
`tion of vascular architecture was performed as previously de-
`scribed (9). After anesthetizing the mice with i.p. ketamine (50
`mg兾kg) and xylazine (5 mg兾kg), a sternotomy was performed to
`expose the heart. Five percent fluorescein isothiocyanate–
`dextran (2,000,000 Mr, Sigma) in a volume of 1-cc PBS was
`injected into the left ventricle. After perfusion of the tumor,
`mice were killed.
`
`Harvesting of Specimens. After death of mice, tumors and con-
`tralateral kidneys were removed. One-half of each kidney兾tumor
`was fixed in fresh 4% paraformaldehyde for histology and
`immunohistochemistry at room temperature. The other half was
`flash frozen in liquid nitrogen and stored at –80°C. Paraform-
`aldehyde-fixed specimens were subsequently embedded in par-
`affin blocks.
`
`Immunohistochemistry. ␣-Smooth muscle actin (␣SMA) staining. A
`monoclonal anti-␣SMA actin antibody (Sigma) was diluted
`(1:200), and incubated overnight at 4°C. Specimens were then
`incubated in turn with a 1:400 rabbit anti-mouse biotinylated
`secondary antibody (Zymed). Enhanced horseradish peroxi-
`dase-conjugated streptavidin and a substrate chromogen, AEC
`(3-amino-9-ethyl carbazole), was used to develop a brown–red
`color (HistoStain-Plus kit, Zymed).
`Terminal deoxyribonucleic d-UTP nick end labeling (TUNEL) as-
`say. Mounted paraformaldehyde-fixed specimens underwent
`TUNEL assay for apoptosis by using the In Situ Cell Death Kit
`(Roche Applied Science, Indianapolis) and were examined by
`light microscopy.
`
`In Situ Hybridization. Tissue was initially preserved in 4% para-
`formaldehyde overnight, transferred to 17% sucrose, and em-
`bedded in OCT compound and frozen. Tissue sections were then
`probed with 35S-labeled cRNA with a probe spanning codons
`57–192 of human VEGF as described (6).
`
`Statistical Analysis. Tumor weights were expressed as mean ⫾
`SEM and compared by Kruskal–Wallis analysis.
`
`Inhibition of neuroblastoma tumors by anti-VEGF agents. Neuroblas-
`Fig. 1.
`toma xenografts were treated with the anti-VEGF agents (NX1838, n ⫽ 5;
`anti-VEGF antibody, n ⫽ 19; VEGF-Trap LD, n ⫽ 10; and VEGF-Trap HD, n ⫽ 10),
`with the results expressed as percent of control tumors (control NX1838, n ⫽
`5; control anti-VEGF antibody, n ⫽ 21; control VEGF-Trap LD, n ⫽ 10; and
`control VEGF-Trap HD, n ⫽ 10), with error bars representing SEM for treated
`tumors. Statistical analysis was done by Kruskal–Wallis analysis, with NX1838
`(P ⫽ 0.08), anti-VEGF antibody (P ⫽ 0.12), VEGF-Trap LD (P ⫽ 0.10), and
`VEGF-Trap HD (P ⫽ 0.0009).
`
`Results
`Comparison of Anti-VEGF Reagents in Experimental Neuroblastoma.
`Using a metastasizing murine model of neuroblastoma (3), we
`tested the effects of anti-VEGF reagents with differing targets.
`We evaluated NX1838, an RNA-based fluoropyrimidine poly-
`ethylene glycol-conjugated aptamer, which has an estimated
`dissociation constant (Kd) for human VEGF165 of 200 pM
`(NeXstar) (13, 14);
`the monoclonal anti-VEGF antibody
`(A.4.6.1, Genentech) (11) that specifically binds to human
`VEGF with an affinity of 0.1–10 nM but not murine VEGF; and
`VEGF-Trap, a soluble composite decoy receptor consisting of
`Ig-like domains of VEGFR-1 and VEGFR-2 fused to an Fc
`segment (VEGF-Trap, Regeneron Pharmaceuticals) (12), which
`binds with very high affinity (⬇1 pM) to multiple isoforms of
`VEGF from several species (including human and murine). The
`VEGF-Trap also binds to placental growth factor. Tumor weight
`was evaluated at 6 weeks.
`We found that treatment with NX1838 and anti-VEGF anti-
`body resulted in partial inhibition of tumor growth (Fig. 1, 52 and
`53% of control, respectively, P ⫽ 0.08 and P ⫽ 0.12, respectively).
`We also tested the efficacy of VEGF-Trap at two doses, 100 g
`(VEGF-Trap LD) and 500 g (VEGF-Trap HD). VEGF-Trap
`LD partially suppressed tumor growth (30% of control, P ⫽
`0.10), whereas VEGF-Trap HD displayed the highest degree of
`tumor suppression of the antiangiogenic agents tested (19% of
`control, P ⫽ 0.009) (Fig. 1). These results demonstrate that
`VEGF-Trap at 500 g per dose can significantly suppress tumor
`growth in our xenograft model of neuroblastoma.
`
`Vascular Cooption Is an Early Event in Neuroblastoma Tumor Growth.
`We next performed a detailed comparison of angiogenesis in the
`control tumors with the anti-VEGF antibody and VEGF-Trap-
`exposed tumors (because of the limited supply of NX1838, we
`were unable to conduct a more detailed analysis). We examined
`angiogenesis by fluorescein angiography and ␣SMA immuno-
`staining at 4 and 6 weeks. At 4 weeks, fluorescein perfusion
`outlined rounded structures at the branch points of vessels in
`control, anti-VEGF antibody-treated, VEGF-Trap LD, and
`VEGF-Trap HD tumors (Fig. 2 A–D, respectively; see white
`arrowheads). Closer examination by confocal microscopy sug-
`gested that the structures (about 80 microns in diameter) were
`renal glomeruli, with afferent and efferent blood vessels (Fig. 2
`E–H, respectively). Entrapment of renal glomeruli within tumor
`
`11400 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.172398399
`
`Kim et al.
`
`
`
`MEDICALSCIENCES
`
`Fluorescein angiography. Tumor vasculature was evaluated by fluorescein– dextran angiography at 4 (A–H) and 6 weeks (I–N). At 4 weeks, standard
`Fig. 2.
`fluorescent microscopy (A–D, ⫻10 original magnification) revealed cooption of renal glomeruli in Control (A), anti-VEGF antibody (B), VEGF-Trap LD (C), and
`VEGF-Trap HD (D). Close-up view of coopted glomeruli by confocal microscopy: Control (E), anti-VEGF antibody (F), VEGF-Trap LD (G), and VEGF-Trap HD (H),
`demonstrates connection of glomeruli to afferent and efferent vessels, seen in best in E and G. At 6 weeks, standard fluorescent microscopy (I–L, ⫻4 original
`magnification) demonstrates vascular remodeling with abundant vessels in control tumors (I), but persistent cooption in the anti-VEGF antibody (J) and
`VEGF-Trap LD (K), with glomeruli indicated by arrowheads. In VEGF-Trap HD (L), sparse vasculature and little cooption were seen at 6 weeks. Pseudodepth
`coloring of Control (M) and anti-VEGF antibody (N), demonstrates the abundant vasculature in Control tumors, and the persistent cooption in the anti-VEGF
`antibody-treated tumors. The coopted glomeruli are approximately 80 m in size (Bar ⫽ 100 m.).
`
`tissue was evident after hematoxylin兾eosin staining in control
`(Fig. 3 A and E), anti-VEGF antibody (Fig. 3 B and F),
`VEGF-Trap LD (Fig. 3 C and G), and VEGF-Trap HD (Fig. 3
`D and H). At the interface of tumor–kidney invasion, the tumor
`
`cells replace the renal parenchyma; however, glomeruli are
`preserved, becoming encased in tumor. Red blood cells are
`visible within the glomeruli, indicating that these structures
`remain perfused, despite being surrounded by tumor tissue.
`
`Kim et al.
`
`PNAS 兩 August 20, 2002 兩 vol. 99 兩 no. 17 兩 11401
`
`
`
`Cooption of renal glomeruli at 4 weeks. Early in tumor growth, all tumors demonstrate cooption of renal glomeruli (arrowheads) as seen by
`Fig. 3.
`hematoxylin兾eosin staining (A–D, ⫻20; E–F, ⫻10): Control (A and E), anti-VEGF antibody (B and F), VEGF-Trap LD (C and G), and VEGF-Trap HD (D and H). In control
`(E) and VEGF-Trap HD (H), coopted glomeruli (arrowheads) encased by tumor tissue (t) are seen adjacent to renal tissue (r).
`
`These results demonstrate that cooption of renal vasculature and
`glomeruli occurs early in tumor growth.
`At 6 weeks, control tumor growth was characterized by
`vascular remodeling with abundant new vessels (Fig. 2 I and M).
`We did not detect glomerular cooption in 6-week controls either
`by fluorescein angiography or by hematoxylin兾eosin staining
`(data not shown). In contrast, 6-week xenografts exposed to
`anti-VEGF antibody or VEGF-Trap LD displayed persistent
`vascular cooption both by fluorescein angiography (anti-VEGF
`antibody, Fig. 2 J and N; VEGF-Trap LD, Fig. 2K) and hema-
`toxylin兾eosin staining (data not shown). Confocal microscopy
`demonstrates that the multiple coopted glomeruli are perfused
`by branches originating from long straight vessels (anti-VEGF
`antibody, Fig. 2N). When anti-VEGF antibody and VEGF-Trap
`LD therapy are stopped, these vascular structures are lost within
`3 weeks and replaced by abundant new vessels, suggesting that
`neoangiogenesis had remodeled the originally coopted structure
`(data not shown). In the VEGF-Trap HD group, tumors at 6
`weeks had very sparse vasculature with little glomeruli cooption
`seen by angiography (Fig. 2L). Because fluorescein–dextran
`angiography will demonstrate only perfused glomeruli, we also
`examined sections stained by hematoxylin兾eosin to determine
`
`whether nonperfused glomeruli remained within the tumor
`tissue. By using this method of examination, again only rare
`glomeruli were seen (data not shown), indicating that there was
`near complete regression of the coopted vascular structure that
`was seen at 4 weeks. These results demonstrate that partial or
`incomplete blockade of VEGF can block vascular remodeling
`and elicit persistent vascular cooption in experimental neuro-
`blastoma, whereas more complete VEGF blockade results in
`regression of the coopted vasculature and a more profound
`antitumor effect.
`
`Anti-VEGF Reagents Result in Decreased Vasculature. Staining for
`␣SMA immunopositive cells was performed to further evaluate
`recruitment of perivascular cells (15). Control tumors displayed
`a dense perivascular network at 4 (Fig. 4A) and 6 weeks (Fig. 4E).
`Animals injected with anti-VEGF antibody, VEGF-Trap LD,
`and VEGF-Trap HD developed tumors with markedly de-
`creased ␣SMA-staining vascular components as compared with
`controls at both 4 (Fig. 4 B–D) and 6 weeks (Fig. 4 F–H).
`Similarly, markedly decreased endothelium was also seen by
`immunostaining for the endothelial marker platelet–endothelial
`cell adhesion molecules-1 (data not shown). When injections of
`
`Fig. 4. Decreased vascularity (reflected by diminished recruitment of perivascular cells) is seen by ␣SMA staining. Four (A–D) and 6 weeks (E–H), ⫻10 original
`magnification. Control (A and E) tumors have abundant vasculature and are associated with numerous perivascular cells. There was marked decrease in neoangio-
`genesis with only a few larger-caliber vessels in tumors after injection of anti-VEGF antibody (B and F), VEGF-Trap LD (C and G), and VEGF-Trap HD (D and H).
`
`11402 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.172398399
`
`Kim et al.
`
`
`
`Fig. 5. Apoptosis in renal glomeruli. Apoptosis was evaluated at 4 (A–D) and 6 weeks (E–H) by TUNEL. Control (A and E), anti-VEGF antibody (B and F), VEGF-Trap
`LD (C and G), and VEGF-Trap HD (D and H). Endothelial apoptosis was seen in anti-VEGF antibody (Inset B), VEGF-Trap LD (Inset C), and VEGF-Trap HD (D) but
`rarely in control tumors (A). Both control tumors and VEGF-Trap HD demonstrated apoptosis within glomeruli, suggesting that in tumors where cooption is
`transient, glomeruli undergo early apoptosis. Red blood cells are seen in the glomeruli of anti-VEGF antibody (B) and VEGF-Trap LD (C) tumors, but there is little
`apoptosis within the glomeruli. Apoptosis within glomeruli is seen at a later time point (6 weeks) in the anti-VEGF antibody (F) and VEGF-Trap LD (G) tumors.
`
`MEDICALSCIENCES
`
`mors may serve as a useful surrogate for the effectiveness of
`antiangiogenesis agents.
`
`Discussion
`Our data are consistent with a model in which certain tumors,
`such as neuroblastomas, initially coopt host vasculature and then
`remodel this host vasculature, destroying the initial coopted
`structures. These results further lead to the hypothesis that such
`cooption can be dramatically and differentially affected by the
`degree of VEGF blockade. Partial VEGF blockade, as may be
`achieved by the anti-VEGF antibody A4.6.1, or low-dose VEGF-
`Trap, allows for initial vessel cooption but inhibits later remod-
`eling, resulting in long-term persistence of the coopted vascular
`
`Fig. 6. Anti-VEGF treatment results in up-regulation of VEGF expression. In
`situ hybridization demonstrated a low-level diffuse expression of VEGF in
`control (A) tumors at 6 weeks, but marked up-regulation of VEGF in anti-VEGF
`antibody (B), VEGF-Trap LD (C), and VEGF-Trap HD (D) tumors at the same time
`point. In D, renal tissue is present (r) and contains glomeruli that express VEGF.
`
`anti-VEGF agent were withheld for 2 weeks, 8-week tumors
`demonstrated increased blood vessel development compared
`with their 6-week counterparts (data not shown).
`
`Anti-VEGF Reagents Result in Endothelial and Glomerular Apoptosis.
`To explore the mechanism of decreased angiogenesis seen with
`anti-VEGF treatment, TUNEL assays were performed to detect
`the presence of apoptosis, particularly in the endothelial cells
`and glomeruli. Control tumors had little apoptosis in endothelial
`cells at 4 or 6 weeks (Fig. 5 A and E). In the treated tumors
`(anti-VEGF antibody, VEGF-Trap LD, and VEGF-Trap HD) at
`4 weeks, endothelial cell apoptosis was observed (Fig. 5 B–D). At
`4 weeks, apoptosis was found in coopted glomeruli in the control
`and VEGF-Trap HD tumors but rarely in the VEGF-Trap LD
`or anti-VEGF antibody-exposed tumors. At 6 weeks, however,
`there was marked apoptosis within glomeruli in all three anti-
`VEGF-treated groups (Fig. 5 F–H). The presence of apoptosis
`within glomeruli at 4 weeks is consistent with the absence of
`these structures at 6 weeks in the control and VEGF-Trap HD
`groups and contrasts with the status of coopted glomeruli in
`VEGF-Trap LD and anti-VEGF antibody-treated tumors. Apo-
`ptosis in glomeruli outside of the tumor mass was not observed.
`Our results suggest that very efficient blockade of VEGF with
`VEGF-Trap HD overrides factors tending to preserve coopted
`glomeruli in VEGF-deprived neuroblastoma xenografts, result-
`ing in endothelial apoptosis and involution of these coopted
`vascular structures.
`
`Anti-VEGF Reagents Result in VEGF Up-Regulation. To evaluate
`VEGF expression in these tumors, in situ hybridization was
`performed by using a probe that crossreacts with both mouse and
`human VEGF. At 4 weeks, all xenografts demonstrated a diffuse
`low level of VEGF expression (data not shown). At 6 weeks, the
`control xenografts continued to demonstrate a diffuse low level
`of VEGF expression (Fig. 6A). Treatment of xenografts with the
`anti-VEGF reagents resulted in marked increase in VEGF
`expression in the anti-VEGF antibody, VEGF-Trap LD and
`VEGF-Trap HD tumors (Fig. 6 B–D, respectively). These results
`suggest that the blockade of VEGF and the resultant decrease
`in perfusion, via decreased neoangiogenesis and endothelial
`apoptosis, lead to tumor hypoxia and a strong up-regulation in
`VEGF expression. Hypoxia-induced VEGF expression in tu-
`
`Kim et al.
`
`PNAS 兩 August 20, 2002 兩 vol. 99 兩 no. 17 兩 11403
`
`
`
`structures. The lack of vascular growth results in a poorly
`perfused tumor, which in turn results in hypoxic induction of
`VEGF in the tumor and thus reflects successful blockade of
`angiogenesis. More complete VEGF blockade, as effected in
`these studies by the high-dose VEGF-Trap, results in a modifi-
`cation of the above events. As with partial VEGF inhibition,
`initial vascular cooption occurs and subsequent remodeling is
`also blocked, leaving the coopted vasculature intact in the short
`term. In contrast to partial VEGF inhibition, however, it appears
`that more complete blockade eventually leads to regression of
`the coopted vasculature. In this setting, tumor growth is also
`significantly more inhibited, indicating that persistent coopted
`vessels can indeed support some tumor growth, even in the
`absence of new angiogenesis.
`One possible explanation for this disparate effect of high-dose
`VEGF-Trap, as compared with anti-VEGF antibody or low-dose
`VEGF-Trap, is that at this dose this agent efficiently blocks not
`just the VEGF required for new vasculature to sprout but also
`the low levels of this factor required to support the long-term
`integrity of coopted vasculature. As a result, the coopted struc-
`tures undergo apoptosis. Tumor growth is more effectively
`suppressed than if neoangiogenesis alone were inhibited.
`The ability of high-dose VEGF-Trap to disrupt cooption may
`be a consequence of its higher affinity for human VEGF than
`that of the anti-VEGF antibody A4.6.1, or the prolonged circu-
`lation time of the VEGF-Trap (11, 12). Thus, VEGF-Trap may
`be able to more completely block the human VEGF derived
`
`from the implanted human tumors. Alternatively, because
`VEGF-Trap also binds mouse VEGF (12), which is not recog-
`nized by the anti-VEGF antibody A4.6.1 (13), its greater efficacy
`could be because of blockade of both tumor-derived (human) as
`well as host-derived (murine) VEGF. Finally, the increased
`efficacy of the high-dose VEGF-Trap compared with the VEGF
`antibody may be because of its ability to bind VEGF family
`members other then VEGF A, such as placental growth factor,
`which is known to bind this agent (14). However, we did not test
`either the NX1838 aptamer or anti-VEGF antibody at higher
`doses, so it is possible that increased concentrations of these
`anti-VEGF agents would cause a more complete blockade of
`tumor growth as well as disrupting cooption.
`Determination of the relative contribution of low-level
`VEGF and other VEGF family members to cooption of host
`vessels will depend on the development of new probes to
`individually differentiate these factors. Our results suggest that
`the ability of neuroblastoma to support continued growth via
`cooption of preexisting vasculature represents an important
`mechanism of tumor resistance to antiangiogenic treatment and
`may require adjustment of these emerging therapies to counter
`such responses.
`
`We are grateful to A. Lalla for technical assistance. This work was
`supported by the Pediatric Cancer Foundation (J.J.K. and D.J.Y.), the
`Sorkin Fund, (J.J.K.), and National Cancer Institute Grant
`1R01CA088951-01A1 (D.J.Y.).
`
`1. Leung, D. W., Cachianes, G., Kuang, W. J., Goeddel, D. V. & Ferrara, N.
`(1989) Science 246, 1306–1309.
`2. Ferrara, N. & Alitalo, K. (1999) Nat. Med. 5, 1359–1364.
`3. Rowe, D. H., Huang, J., Li, J., Manley, C., O’Toole, K. M., Stolar, C. J.,
`Yamashiro, D. J. & Kandel, J. J. (2000) J. Pediatr. Surg. 35, 977–981.
`4. Rowe, D. H., Huang, J., Kayton, M. L., Thompson, R., Troxel, A., O’Toole, K. M.,
`Yamashiro, D., Stolar, C. J. & Kandel, J. J. (2000) J. Pediatr. Surg. 35, 30–32.
`5. Holash, J., Wiegand, S. J. & Yancopoulos, G. D. (1999) Oncogene 18,
`5356–5362.
`6. Holash, J., Maisonpierre, P. C., Compton, D., Boland, P., Alexander, C. R.,
`Zagzag, D., Yancopoulos, G. D. & Wiegand, S. J. (1999) Science 284,
`1994–1998.
`7. Zagzag, D., Friedlander, D. R., Margolis, B., Grumet, M., Semenza, G. L.,
`Zhong, H., Simons, J. W., Holash, J., Wiegand, S. J. & Yancopoulos, G. D.
`(2000) Pediatr. Neurosurg. 33, 49–55.
`8. Rubenstein, J. L., Kim, J., Ozawa, T., Zhang, M., Westphal, M., Deen, D. F.
`& Shuman, M. A. (2000) Neoplasia 2, 306–314.
`
`9. Kunkel, P., Ulbricht, U., Bohlen, P., Brockmann, M. A., Fillbrandt, R., Stavrou,
`D., Westphal, M. & Lamszus, K. (2001) Cancer Res. 61, 6624–6628.
`10. Kim, E., Moore, J., Huang, J., Soffer, S., Manley, C. A., O’Toole, K.,
`Middlesworth, W., Stolar, C. J., Kandel, J. J. & Yamashiro, D. J. (2001)
`J. Pediatr. Surg. 36, 287–290.
`11. Presta, L. G., Chen, H., O’Connor, S. J., Chisholm, V., Meng, Y. G., Krummen,
`L., Winkler, M. & Ferrara, N. (1997) Cancer Res. 57, 4593–2599.
`12. Holash, J., Davis, S., Papadopoulos, N., Croll, S. D., Ho, L., Russell, M.,
`Boland, P., Leidich, R., Hylton, D., Burova, E., et al. (2002) Proc. Natl. Acad.
`Sci. USA 99, 11393–11398.
`13. Ruckman, J., Green, L. S., Beeson, J., Waugh, S., Gillette, W. L.,
`Henninger, D. D., Claesson-Welsh, L. & Janjic, N. (1998) J. Biol. Chem.
`273, 20556–20567.
`14. Tucker, C. E., Chen, L., Judkins, M. B., Farmer, J. A., Gill, S. C. & Drolet,
`D. W. (1999) J. Chromatogr. Biomed. Appl. 732, 203–212.
`15. Morikawa, S., Baluk, P., Kaidoh, T., Haskell, A., Jain, R. K. & McDonald, D. M.
`(2002) Am. J. Pathol. 160, 985–1000.
`
`11404 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.172398399
`
`Kim et al.
`
`