Integrin α_vβ₃-Targeted Radioimmunotherapy of Glioblastoma Multiforme

Glioblastoma multiforme is the most prevalent and malignant primary brain tumor in adults. It is a high-grade WHO class (IV) tumor that usually presents in the sixth or seventh decade of life and carries an average prognosis of 1 year (1). Standard synergistic treatment of microsurgical resection combined with postoperative external radiation and chemotherapy is not effective in significantly extending the lifespan of patients (2, 3). With a high proliferation rate, marked neovascularization, central necrosis, and extensive local invasion into normal brain parenchyma, gliomas have developed resistance to traditional radiation and chemotherapy agents (4). Current therapeutic strategies have shown poor efficacy due to lack of tumor cell specificity, poor tumor penetration, and invasive administration technique.

The discovery of tumor-specific antigens has led cancer research to focus on antibody-based therapeutics. Integrin α_Vβ₃ is a cell adhesion molecule highly expressed on actively angiogenic endothelium and expanding glioblastoma multiforme tumoral cells (5–9), rarely expressed on nascent vascular smooth muscle endothelium (10). It is well known that malignant gliomas require an immense vascular network to support tumorigenesis, parenchymal invasion, and distant metastasis (11). In this process of angiogenesis, integrin α_Vβ₃ directs extracellular matrix adhesion to support the recruitment of endothelial progenitor cells. Additionally, malignant gliomas have been shown to express α_Vβ₃ in great density on their cell surface (7, 12, 13). Thus, integrin α_Vβ₃ is a reliable marker to distinguish malignant from benign cancer phenotype due to its strong correlation with tumor invasiveness, metastasis, and poor clinical outcome (14–20).

Inhibition of α_Vβ₃ by monoclonal antibodies, cyclic RGD peptide antagonists (21), and peptidomimetics (22) has shown efficacy in halting angiogenesis (23) and tumor cell growth (24, 25). Abegrin (MEDI-522 or Vitaxin) is a humanized monoclonal antibody against human integrin α_Vβ₃ currently in phase II clinical trials (26). It has shown high glioma-specific uptake as well as ideal pharmacokinetics and dose optimization in glioblastoma multiforme murine models (12). It is hypothesized that the “triad” of Abegrin activity results in the reduction of tumor growth, angiogenesis, and distant metastasis (27). Thus, this antibody is well suited for targeting glioblastoma multiforme cells in vivo.

Currently, the use of immunotherapy has been augmented with radionuclides to create radioimmunotherapeutics. Such therapeutics have been attempted and shown efficacy in a number of clinical studies. Promising results have been obtained by intracavitary and/or pretargeted radioimmunotherapeutics with ⁹⁰Y in high-grade gliomas (28), craniopharyngiomas (29), and non–Hodgkin's (30) and central nervous system lymphomas (31, 32). ⁹⁰Y is a high-energy pure β-emitter ideal for targeted radiation therapy with a physical t_1/2 of 64 hours and an effective tissue penetration of 5.3 mm (33), generating a sphere of therapeutic efficacy beyond that of bound antibody (34). Thus ⁹⁰Y may be uniquely suited for patients with large, bulky tumor masses with aberrant vascular networks such as malignant gliomas.

The investigation of radioimmunotherapeutics in the treatment of glioblastoma multiforme has largely focused on intracavitary or general pretargeted therapy. Although these therapeutics have shown some improvement in clinical outcome, the lack of tumor specificity increases human myelotoxicity as larger doses are required for therapeutic efficacy (35). Thus, the use of a tumor-specific antibody conjugated to a potent radionuclide may provide the level of radiotherapy and molecular inhibition necessary to treat aggressive brain tumors. The purpose of this study was to evaluate the treatment efficacy of a novel ⁹⁰Y-Abegrin radioimmunotherapeutic in murine xenograft glioblastoma models by use of noninvasive in vivo molecular imaging modalities.

DOTA conjugation and radiolabeling.⁹⁰Y was obtained from Perkin-Elmer and the monoclonal antibody Abegrin was obtained from MedImmune, Inc. The synthesis of DOTA-Abegrin (1,4,7,10-tetra-azadodecane-N,N′,N″,N‴-tetraacetic acid) was reported earlier (12). Briefly, ⁹⁰Y was added to the DOTA-Abegrin conjugates at 25 μg of DOTA-Abegrin per mCi of ⁹⁰Y. The reaction mixture was incubated for 1 h at 40°C with constant shaking. The ⁹⁰Y-DOTA-Abegrin conjugate was purified by PD-10 column using PBS as the eluent. The radioactive fractions containing ⁹⁰Y-DOTA-Abegrin were collected and passed through a 0.2-μm syringe filter for further in vitro and in vivo experiments. A similar procedure was used for ¹¹¹In-DOTA-Abegrin conjugation.

Biodistribution. Twenty nude mice were randomly divided into five groups, each of which had four animals. Each mouse was injected with 15 μCi of ¹¹¹In-DOTA-Abegrin via tail vein. The mice were sacrificed at 4, 24, 72, 120, and 168 h postinjection. Blood, major organs, and tissues were harvested, weighed, and measured in a γ-counter. Organ uptake was calculated as a percentage of injected dose per gram of tissue (%ID/g).

Maximum tolerated dose of ⁹⁰Y-Abegrin. The maximum tolerated dose (MTD) in nontumor-bearing female nude mice was determined by tail vein injection of escalating ⁹⁰Y doses (50, 100, 150, 200, and 300 μCi). Each dose was tested in seven mice. Body weight and health status were determined twice weekly over 35 d. Peripheral blood was collected from the tail vein twice weekly and then tested in a blood analyzer. Samples from mice within a given dose group were pooled, and diluted 1:200 in PBS for RBC counts, 1:100 in 1% (w/v) ammonium oxalate for platelet counts, or 1:20 in 3% (w/v) acetic acid for WBC and neutrophil counts. Because organ damage may not immediately present, animals were monitored for an additional 1-mo period. The MTD was set at the dose below the first dose that caused severe loss of body weight (>20%) or animal death.

U87MG glioblastoma multiforme xenograft model. The U87MG human glioblastoma multiforme cell line was obtained from American Type Culture Collection. U87MG cells were grown in DMEM and cultured in medium supplemented with 10% (v/v) fetal bovine serum (FBS) at 37°C in a humidified atmosphere with 5% CO₂.

Female athymic nude mice were supplied from Harlan at 4 to 5 wk of age. All animal procedures were done according to a protocol approved by Stanford University Institutional Animal Care and Use Committee. The U87MG brain tumor model was established by injection of 5 × 10⁶ cells (in 50 μL PBS) into the s.c. tissue of the left shoulder. Tumor size was monitored via serial caliper measurements every 3 d. Each measurement was taken three times and averaged to provide an approximation of tumor size.

For orthotopic preparation, mice were anesthetized with i.m. ketamine, and an electric drill was used to create a 1-mm diameter burr hole 2 mm lateral and 1 mm anterior to the bregma on the right side. Following dural penetration, a 10 μL Hamilton syringe was stereotactically inserted 2.0 mm from the surface of the brain and a 5-μL suspension of 1 × 10⁶ U87MG cells was delivered over a period of 5 min. The needle was then retracted over 5 min and the scalp was reapproximated using 5-0 vicryl sutures.

Radioimmunotherapeutic dose selection and therapy. To characterize the dose response of ⁹⁰Y-Abegrin, multiple escalating doses were tested in animals bearing s.c. U87MG tumors. Separated into four categories, animals received either saline or 50, 100, or 200 μCi of ⁹⁰Y-Abegrin. Efficacy was assessed via serial tumor size and body weight measurement.

To assess therapeutic efficacy, inoculated mice that approached a uniform tumor size of ∼300 mm³ 4 wk after tumor cell inoculation were chosen for study. At this time, animals were segregated into four groups and injected with a single dose of ⁹⁰Y-Abegrin (n = 10), ⁹⁰Y-IgG (n = 5), Abegrin (n = 5), or saline (n = 9). All four groups of tumor-bearing mice were subjected to small-animal positron emission tomography (PET) imaging as well as serial tumor and body weight measurement.

HT-29 xenograft model. To evaluate the efficacy of ⁹⁰Y-Abegrin in low integrin-expressing cancers, a 50-μL suspension of 5 × 10⁶ HT-29 cells, human colon cancer, was injected s.c. into the upper right flanks of female athymic nude mice. Animals with visible tumor growth were again separated into three groups and received either 200 μCi of ⁹⁰Y-Abegrin, 200 μCi of ⁹⁰Y-IgG, or saline. These animals underwent serial tumor and body weight measurement.

Small-animal PET imaging. PET imaging of tumor-bearing mice was done on a microPET R4 rodent model scanner as previously described (34). Mice were injected with ¹⁸F-FLT (⁹⁰Y-Abegrin: 7 mice; ⁹⁰Y-IgG: 4 mice; Abegrin: 5 mice; control: 5 mice) or ¹⁸F-FDG (⁹⁰Y-Abegrin: 7 mice; ⁹⁰Y-IgG: 4 mice; Abegrin: 5 mice; control: 4 mice) via the tail vein, anesthetized with 2% isoflurane and placed near the center of the field of view where the highest image resolution and sensitivity was available. Assuming a tissue density of 1 g/mL, the counts per milliliter per minute were converted to counts per gram per minute, and then divided by the injected dose (ID) to obtain an imaging region of interest–derived percentage ID per gram of tissue (%ID/g) as previously described (36–38).

Histopathologic processing. Visualization of collagen fibers and cellular architecture was achieved by use of our trichrome staining protocol. Briefly, sections were incubated in Weigert's iron hematoxylin working solution, incubated with Biebrich scarlet-acid fuchsin solution, and differentiated in phosphomolybdic-phosphotungstic acid solution until collagen was not red. Sections were transferred to aniline blue, differentiated in 1% acetic acid, quickly dehydrated through 95% ethyl alcohol, and mounted with resinous medium.

For Ki67 staining, an antigen retrieval kit from Abcam was used according to the manufacturer's specifications. Endogenous peroxidase was blocked with 3% hydrogen peroxide in PBS and incubated with a protein-blocking solution containing PBS (pH 7.5). Samples were incubated with a 1:100 dilution of Ki67, incubated in diaminobenzidine, mounted on glass slides, and coverslipped.

For visualization of endothelial cells, CD31 staining was chosen. Frozen tissue sections (7 μm) were fixed with cold acetone, incubated with a protein-blocking solution, and incubated with a 1:100 dilution of rat monoclonal anti-CD31 antibody (BD Pharmingen) and 1:200 Cy5-conjugated donkey antimouse secondary antibody. Samples were mounted with 4′,6-diamidino-2-phenylindole (DAPI) for staining of cell nuclei.

Immunofluorescent terminal deoxynucleotidyltransferase-mediated biotin-dUTP nick-end labeling (TUNEL) analysis was done by use of a commercially available kit (Roche Applied Science). According to the manufacturer's specifications, samples were fixed with 10% formalin (methanol free) and permeabilized by incubation with 0.2% polyethylene glycol. Reaction buffer was added for incubation in a humidified chamber. The reaction was terminated with 30 mmol/L NaCl and 3 mmol/L sodium citrate (pH 7.2), followed by removal of unincorporated fluorescein-dUTP and staining of nuclei with DAPI.

Fluorescence images were acquired with an Axiovert 200M fluorescence microscope equipped with a DAPI (excitation, HQ 365 nm; emission, HQ 397 nm), Texas Red (excitation, HQ 535 nm; emission, HQ 610 nm), and Cy5.5 (excitation, HQ 665/45 nm; emission, HQ 725/50 nm) filter set. Images were obtained with a thermoelectrically cooled charge-coupled device and analyzed with MetaMorph software.

Volumetric histology was done on animals inoculated with orthotopic U87MG tumors. Animals were sacrificed on posttreatment day 14, and brain tissue was harvested after systemic perfusion with 20 mL cold saline followed by 20 mL of 10% formalin. A combination of sharp and blunt dissection was used to differentiate tumors from benign parenchymal tissue. Isolated tumors underwent volumetric measurement.

Statistical analysis. Quantitative data were expressed as mean ± SD. Statistical analysis was done using a 1-way ANOVA for multiple groups and an unpaired Student's t test. P values <0.05 were considered statistically significant (GraphPad Prism).

Dose escalation and toxicity evaluation. Nontumor-bearing mice were separated into five groups, each of which underwent a single injection of ⁹⁰Y (50, 100, 150, 200, and 300 μCi). Animal body weight, WBC, neutrophil, RBC, HCG, and platelet counts were analyzed after radioisotope injection. Only the 300 μCi group experienced animal death on postinjection days 10 and 15. The remaining 200, 150, 100, and 50 μCi groups had a final weight increase on postinjection day 32 to 136.2 ± 8.3%, 135.7 ± 11.7%, 145.5 ± 10.1%, and 146.0 ± 12.0% of the original body weight on day 0, respectively (Supplementary Fig. S1A). The WBC count of each dosage group was tracked over 22 days (Fig. 1A). Animals in the 300 μCi group experienced a reduction of WBC to 15.9 ± 1.3% of their original value. All other groups maintained a varying degree of increasing WBC with the 200 μCi group to 146.3 ± 36.9%, the 150 μCi group to 131 ± 27.08%, the 100 μCi group to 206.3 ± 60.6%, and the 50 μCi group to 199.1 ± 38.8%. Final absolute neutrophil counts were 0.55 ± 0.071 in the 300 μCi (day 15), 5.87 ± 6.40 in the 200 μCi, 5.49 ± 2.44 in the 150 μCi, 7.38 ± 3.19 in the 100 μCi, and 5.12 ± 2.33 in the 50 μCi group (Fig. 1B). RBC counts of the 300 μCi group were reduced to 50.1 ± 12.4%, the 200 μCi group to 92.7 ± 4.5%, the 150 μCi group to 95.6 ± 6.5%, the 100 μCi group to 95.1 ± 2.3%, and the 50 μCi group to 103.6 ± 11.2% of their original value (Fig. 1C). Platelet counts were also followed and revealed reduced counts with increasing dosage. Final counts 22 days postinjection were 13.6 ± 9.7% for the 300 μCi, 76.0 ± 35.2% for the 200 μCi, 86.6 ± 34.3% for the 150 μCi, 89.4 ± 24.6% for the 100 μCi, and 94.3 ± 37.1% for the 50 μCi group (Fig. 1D). HGC also showed a similar trend with the 300, 200, 150, 100, and 50 μCi groups having values of 51.9 ± 12.9%, 97.4 ± 6.8%, 100.2 ± 8.3%, 96.7 ± 2.4%, and 105.7 ± 10.1%, respectively (Supplementary Fig. S1B).

Animals in the 150, 200, and 300 μCi groups experienced a reduction in WBC, neutrophil, and platelet counts on day 11, which may represent an initial burden on the bone marrow system. Counts continued to increase, however, with some returning to physiologic levels. Overall, animals that received 300 μCi experienced a reduction in all hematologic counts and had a higher rate of mortality.

Biodistribution of ¹¹¹In-Abegrin. Animals injected with ¹¹¹In-Abegrin showed rapid and high uptake in liver and spleen to 24.81 ± 1.31 and 14.06 ± 1.79 %ID/g at 4 hours postinjection with slow washout to 7.99 ± 1.26 and 6.30 ± 3.0 %ID/g at 168 hours postinjection, respectively (Supplementary Fig. S2). ¹¹¹In-Abegrin was cleared rapidly from the blood 31.31 ± 7.16, 13.69 ± 1.08, 4.60 ± 2.61, 1.36 ± 0.70, and 1.17 ± 1.74 %ID/g at 4, 24, 72, 120, and 168 hours, respectively. The route of excretion was hepatic and splenic, as evidenced by reduced renal concentrations of 8.39 ± 1.28 and 1.14 ± 0.49 %ID/g at 4 and 168 hours, respectively, consistent with our previous finding of ⁶⁴Cu-Abegrin (12). Nontarget organs such as gastrointestinal, heart, lungs, muscle, and bone all maintained low levels of ¹¹¹In-Abegrin.

Dose response efficacy. To evaluate the efficacy of tumor regression with respect to the dose of radioimmunotherapeutics, animals underwent treatment with a single injection of saline or 50, 100, or 200 μCi of ⁹⁰Y-Abegrin. Initially, all groups began with an average tumor size between 130 and 240 mm³ (Fig. 2A). A statistically significant difference was observed on posttreatment day 14 among the four groups (P < 0.05). Specifically, final fractional tumor volumes (V_final/V_initial) of 4.74 in the control, 5.03 in the 50 μCi group, 2.04 in the 100 μCi group, and 0.43 in the 200 μCi group were obtained. Animal weight was also tracked over the experimental groups; no statistically significant difference was revealed on post treatment day 14 (Fig. 2B).

Anti–glioblastoma multiforme efficacy. The growth inhibition of experimental groups was monitored via serial caliper measurements. Initially, all groups began with an average tumor size between 150 and 300 mm³ (Fig. 2C). ⁹⁰Y-Abegrin treatment animals maintained a statistically significant reduction in tumor size beginning on posttreatment day 2 and eventually showed partial tumor regression with a final fractional tumor volume (V_final/V_initial) of 0.69. All other groups showed increased final fractional tumor volumes of 3.76 for ⁹⁰Y-IgG, 5.43 for Abegrin, and 5.72 for control animals (P < 0.0001). Further, animal body weight continued to increase posttreatment, with no statistically significant difference between ⁹⁰Y-Abegrin and other animals (Fig. 2D).

Animals bearing orthotopic tumors were divided into three experimental groups and received either 100 μCi of ⁹⁰Y-Abegrin, 100 μCi of ⁹⁰Y-IgG, or saline injection. Volumetric histology revealed a final tumor volume of 15.56 ±.49, 26.72 ± 8.68, and 26.94 ± 3.46 mm³ in the three groups, respectively (Fig. 3A). Only a reduction in the ⁹⁰Y-Abegrin group was of statistical significance (P < 0.005). Animal body weight did not reveal significant differences (Fig. 3B).

Integrin α_Vβ₃ specificity. To assess the specificity of integrin targeting in producing tumor reduction, a low integrin-expressing cancer cell line was chosen. HT29 cells, a colorectal carcinoma, injected s.c. were randomized into three groups. Animals received 100 μCi of ⁹⁰Y-Abegrin, 100 μCi of ⁹⁰Y-IgG, or saline injection. They obtained final fractional tumor volumes of 4.35, 3.67, 4.59, respectively (Fig. 3C). Both final tumor sizes and animal body weight did not significantly differ among the three groups (P > 0.05; Fig. 3D).

PET imaging. Small-animal PET imaging was used to characterize the in vivo treatment efficacy. ¹⁸F-FDG and ¹⁸F-FLT radiotracers were used to quantify tumor cell metabolic activity and DNA synthesis, respectively. ¹⁸F-FDG imaging revealed a small but statistically significant reduction in both ⁹⁰Y-Abegrin (6.34 ± 0.42 %ID/g) and Abegrin (6.63 ± 0.29 %ID/g) signal intensity when compared with saline control (7.62 ± 0.25 %ID/g; P < 0.001), indicating a reduction in cell proliferation and metabolic activity (Fig. 4A). ¹⁸F-FLT imaging showed tumor accumulation values 2 hours post injection of 1.73 ± 0.1 %ID/g in ⁹⁰Y-Abegrin, 2.94 ± 1.5 %ID/g in ⁹⁰Y-IgG, 2.39 ± 0.28 %ID/g in Abegrin, and 3.04 ± 1.28 %ID/g in control, reflecting reduced DNA synthesis in the ⁹⁰Y-Abegrin group (P < 0.001; Fig. 4B). Representative sagittal images are shown in Fig. 5A and B.

Ex vivo validation of ⁹⁰Y-Abegrin radioimmunotherapeutics. Frozen sections were recovered for trichrome, Ki67, CD31/DAPI, and TUNEL staining (Fig. 6). In ⁹⁰Y-Abegrin treatment mice, trichrome analysis revealed increased extracellular matrix deposition, tumor cell necrosis with diffuse interstitial fibrosis, fibrin stranding, and a significant presence of degenerated peripheral vasculature, suggesting tumor regression. ⁹⁰Y-IgG treatment mice showed piecemeal necrosis and hypereosinophilia with significantly greater neutrophil infiltration than ⁹⁰Y-Abegrin. Both control and Abegrin groups showed mitotically active tumor cells, abnormal and irregular microvasculature, and marginating neutrophils. The vasculature surrounding the necrotic center displayed cavernous capillaries, likely generated to support tumor expansion.

Ki67 analysis showed reduced proliferative index in the treated mice (4.79 ± 0.42%), whereas the ⁹⁰Y-IgG, Abegrin, and control animals had significantly higher proliferative indices of 8.9 ± 1.2%, 13.4 ± 2.1%, and 16.5 ± 0.7% (P < 0.005), respectively.

Quantitative TUNEL-positive cell analysis revealed increased DNA fragmentation and apoptosis in ⁹⁰Y-Abegrin animals when compared with all other groups. The ⁹⁰Y-Abegrin animals had significantly higher percentage of TUNEL-positive cells than the ⁹⁰Y-IgG, Abegrin, and control groups (P < 0.0001). CD31 and DAPI staining showed increased vascular fragmentation, dysmorphic vessel structure, and reduced endothelial presence in ⁹⁰Y-Abegrin animals.

This study shows that ⁹⁰Y-Abegrin, a humanized monoclonal antibody against human integrin α_Vβ₃, exhibits high integrin α_Vβ₃ specificity and therapeutic efficacy against human glioblastoma multiforme tumors. The discovery and clinical translation of a nontoxic, targeted, noninvasive radioimmunotherapeutic for glioblastoma multiforme offers new hope for survival and improved quality of life to patients diagnosed with malignant brain tumors.

We have previously shown the optimization of the DOTA/Abegrin ratio for in vivo targeting and maximized tumor uptake (12). The conjugate with the most DOTA residues per Abegrin was chosen for conjugation as it increases the specific activity of the tracer as well as the absolute uptake within the tumor. Further, DOTA conjugation of Abegrin does not deteriorate the receptor affinity of the antibody (data not shown). Thus DOTA-Abegrin compounds are ideal for radionuclide conjugation and targeting of α_Vβ₃.

Currently, there are no clinical or preclinical trials that evaluate the MTD and associated toxicities of ⁹⁰Y-Abegrin. Human studies that evaluated the toxicity of ⁹⁰Y conjugated to other targeting probes found neutropenia and thrombocytopenia to be common side effects (39–41). Our goal was to develop a one-time dose therapeutic capable of exerting long-term radiation to isolated areas of malignancy. Of the doses chosen for evaluation, 300 μCi resulted in profound animal mortality within 10 to 15 days of administration. Doses of 200 μCi and less did not produce animal death or any statistical difference in body weight, WBC, neutrophil, RBC, platelet, and HGC counts during the observation period (Fig. 1). Subphysiologic levels of WBC, neutrophils, and platelets were shown on posttreatment day 11 in many of the experimental groups. Whereas bone marrow toxicity is a definite concern, the subsequent increase of these values may indicate an acute toxic burden that gradually improves without permanent localization of radioactivity. Furthermore, dose response analysis showed maximal antitumor efficacy in the 200 μCi group without animal mortality (Fig. 2A and B). Given these results, 200 μCi was chosen as our target radionuclide dose.

The advantage of highly selective tumor localization is low radiation burden to nontarget tissues. Vital organ systems such as the hepatic, renal, hematologic, intestinal, bone marrow, and disease-free brain are spared of devastating side effects and toxicity, as seen with ⁹⁰Y-Abegrin. Previously, to achieve therapeutic levels of radiation within tumor tissue, large, potentially toxic, doses of systemic therapy had to be administered (42). To determine the biodistribution of ⁹⁰Y-Abegrin, we chose ¹¹¹In-Abgerin as an imaging surrogate. The biodistribution of ¹¹¹In and ⁹⁰Y-labeled antibodies, proteins, and peptides has been shown to be biologically equivalent with respect to their uptake in tumors and other major organs (43). Consistent with previous reports of Abegrin biodistribution, our therapeutic shows little accumulation in areas of critical importance, such as bone marrow and circulating blood (12). The highest levels of antibody were found in liver and spleen, suggesting a spleno-hepatic clearance mechanism likely due to the high molecular weight of the antibody (∼150 kDa; Supplementary Fig. S2; ref. 12). Nontarget organs such as the kidney and the intestine maintained low antibody concentration, reducing the likelihood of their function in clearance. Further, serum t_1/2 of ⁹⁰Y-Abegrin was found to be 12 to 24 hours, shorter than the t_1/2 of Abegrin in humans.

Monoclonal antibody–based therapeutics offer a number of advantages that make them more amenable to clinical translation. Specifically, radionuclide conjugated antibodies exert their antitumor effect by localizing continuous decaying low-dose-rate irradiation to the area of malignancy. Because ⁹⁰Y β-particle emission can penetrate up to 5 mm of tissue, this therapeutic can destroy antigen-negative tumor cells that have no specific antibody localized to their cell surface (44). This crossfire effect proves tremendously valuable in the case of glioblastoma multiforme, a heterogeneous tumor that lacks organized vascular structure. Ultimately this may translate into lower dosage requirement, reduced toxicity, noninvasive administration and thus a much better quality of life.

With the above parameters in mind, ⁹⁰Y-Abegrin is ideally suited for anti–glioblastoma multiforme therapy. In our study, partial tumor regression was successfully achieved by a one-time dose of 200 μCi of ⁹⁰Y-Abegrin. Animals in the treatment group reached a final fractional tumor volume of 0.68 with stable weight gain and no increase in toxicity or mortality (Fig. 2C and D). Those animals in the ⁹⁰Y-IgG and Abegrin groups had a much larger final fractional tumor volume, with no significant difference from control animals. The antiangiogenic effects of ⁹⁰Y-Abegrin resulted in reduced CD31⁺ vascular endothelial cells, whereas control animals maintained increased vascular caliber and density (Fig. 6). Similarly, trichrome analysis showed a significant presence of degenerated peripheral vasculature in treated animals, suggesting tumor regression and vascular collapse. Control and Abegrin groups, however, showed mitotically active tumor cells and cavernous capillaries depicting active angiogenesis.

To evaluate the necessity of integrin targeting in inducing tumor regression, HT-29, low integrin-expressing, human colon adenocarinoma tumors were also examined. Previous studies have shown low cross reactivity of Abegrin with murine α_vβ₃,suggesting a low probability of antibody localization to xenograft integrin-negative tumors (12). It was our suspicion that due to the low expression of integrin by HT-29, binding of Abegrin is very unlikely and thus therapeutic efficacy would not be evident. Final fractional tumor volumes revealed no significant difference among the three groups, suggesting lack of therapeutic efficacy (Fig. 3C and D).

Noninvasive ¹⁸F-FDG and ¹⁸F-FLT small-animal PET imaging was used to quantify the rate of cell apoptosis and proliferation. Treatment animals showed a statistically significant reduction in ¹⁸F-FLT signal intensity, indicating decreased cellular proliferation and tumor expansion (Figs. 4A and 5A). Consistent with these findings, ex vivo Ki67 histology revealed a reduced proliferative index in treatment animals, indicating low levels of active mitogenesis (Fig. 6). As seen in our study, the strong correlation of ¹⁸F-FLT and Ki67 findings have been previously reported as prognostic indicators for malignant brain tumors (45).

In vivo molecular imaging techniques provide real-time functional information of tumor behavior. The ability to noninvasively characterize tumor integrin expression, vascular permeability, and proliferation will allow for more accurate patient selection, monitoring, and therapeutic intervention. In the case of glioblastoma multiforme, ¹⁸F-FDG has shown poor specificity in characterizing size, functional activity, and heterogeneity of primary and recurrent brain tumors due to the high basal glucose metabolic rate of normal brain tissue. Specifically, ¹⁸F-FDG uptake of low-grade tumors is similar to that of normal white matter, and high-grade tumor uptake may approximate normal gray matter, decreasing the sensitivity of tumor detection (45). FLT-based small-animal PET imaging is more sensitive for intracranial malignancies due to its inability to cross the blood-brain barrier (BBB) and thus reduce nonspecific uptake (46). However, this functionally results in increased FLT signal intensity only in areas of BBB breakdown, reducing our capacity to discern among other possible malignant or infectious processes (47).

To more closely simulate clinical conditions, animals underwent orthotopic implantation of U87MG tumor cells. Although this remains a xenograft model, our intention was to more closely replicate the neurologic deficits produced by an enlarging intracranial mass and evaluate the efficacy of ⁹⁰Y-Abegrin in penetrating the BBB. In the context of malignant brain neoplasms, it is often assumed that BBB breakdown has occurred, but therapeutic penetration is still of chief concern. If considered as adjuvant therapy to supplement surgical resection or radiosurgical treatment, physical trauma induced by these modalities will often violate the integrity of the BBB. Our analysis revealed statistically significant tumor reduction in those animals receiving ⁹⁰Y-Abegrin, suggesting either breakdown or the ability to permeate through the BBB (Fig. 3A and B).

The mechanism by which ⁹⁰Y-Abegrin exerts its therapeutic effect is likely multifactorial. The radioactive decay of ⁹⁰Y-Abegrin is one of the most powerful β-emitting radioimmunotherapeutics tested, at 939 keV per disintegration. This allows for deep tumor penetration and effective radiation-induced cytotoxicity to cancerous cells while sparing nontarget organs and parenchymal brain tissue (48). In contrast to β-particles, α-particle emission has a shorter path length, is more focused, less diffuse, and lacks deep tumor penetration (49). In the case of malignancies that develop an organized vascular network, α-emission may prove less toxic to normal tissue, but it is likely that α-particles would not penetrate the devascularized and bulky areas of glioblastoma multiformes. Similarly, γ-ray photon emission of ¹⁷⁷Lu and ¹¹¹In, although useful for imaging, deposit little energy within surrounding tissue, thereby exerting futile cytotoxic effect (50). Clinical studies have shown safety and some efficacy in the use of ⁹⁰Y as adjuvant therapy for malignant brain tumors (51–53), but further phase II clinical trials are needed to establish true therapeutic potential. In preclinical studies ⁹⁰Y-labeled antibodies induced glioma regression and prolonged survival, but few antibodies possess both tumor and vascular targeting function (54).

Currently in phase II clinical trials, Abegrin is advantageous in that in addition to targeting endothelial cells of highly vascular tumors it also targets α_Vβ₃ expressed on nascent glioma cells, further localizing therapy. To date, no preclinical or clinical study has examined the role of Abegrin therapy in high-grade gliomas, but a number of groups have investigated RGD peptide antagonists against α_Vβ₃. Preclinical in vivo trials with IS20I (55) and cyclic RGD peptides (13, 56) have all shown some degree of glioma growth inhibition, but assessment of toxicity, recurrence, and functional mechanism have not been reported. In our study, Abegrin treatment alone did not induce tumor regression, apoptosis, or vascular injury. One possible reason may be the “dose-limited” antitumor activity whereby high-dose Abegrin therapy does not induce the same tumor regression observed at low doses (27). Thus, the majority of cytotoxicity seen with ⁹⁰Y-Abegrin may be attributed to radiation-induced damage rather than α_Vβ₃ signaling inhibition. RGD peptide antagonists have shown efficacy in increasing radiation sensitivity of glioma cells by suppressing integrin expression. It is hypothesized that radiation-induced cell damage initiates up-regulation of α_Vβ₃ expression and Akt signaling as a defense mechanism to stabilize injured vasculature, promoting radiation resistance (5). Thus, ⁹⁰Y-Abegrin likely acts synergistically to thwart host repair mechanisms and induce both vascular and tumor cell death.

The xenograft model examined in the present study may not be a true approximation of clinical utility. Anatomically, s.c. tumor implantation circumvents the restrictions of a BBB, which may account for the high uptake of ⁹⁰Y-Abegrin observed within tumor tissue, making the value of FLT versus FDG imaging difficult to derive. Similarly, the degree of radiation experienced by parenchymal brain tissue and resulting neurologic deficits due to mass effect and radiation injury could not be assessed; further survival-based orthotopic studies would help clarify these concerns.

To fully translate human dosimetry of ⁹⁰Y-Abegrin, large animal studies with sequential neurologic evaluation would prove valuable. Abegrin has been shown to cross-react with both hamster and primate α_Vβ₃, making these animal models crucial for clinical translation. Although the current study did evaluate MTD, further examination of minimum dose necessary would reduce toxicity and aid in determining human dosimetry. Activity fractionation combined with multiple-dose regimens may reduce nontarget organ radiation and avoid systemic side effects. Current treatment for glioblastoma multiforme involves gross tumor resection, adjuvant stereotactic radiosurgery, chemotherapy, and aggressive imaging surveillance. Major considerations with such a strategy include high tumor recurrence rate, systemic chemotoxicity, radiation-induced parenchymal brain damage, and refractory treatment resistance (2). ⁹⁰Y-Abegrin would likely play a role as synergistic neoadjuvant therapy reducing tumor size, parenchymal invasion, and vascularity before tumor resection. This may influence the safety of surgical resection by reducing the risk of hemorrhage, maintaining clear tumor margins, and possibly reducing postoperative neurologic deficit. With the ability to noninvasively image tumors for integrin expression, we may potentially screen patients for therapeutic candidacy, increasing the likelihood of efficacy and also providing prognostic value.

No potential conflicts of interest were disclosed.

We thank Drs. Fredrick T. Chin, Murugesan Subbarayan, and David W. Dick for the production of ¹⁸F-FDG and ¹⁸F-FLT, and Beth Hoyte for her contribution to preparation of the figures.