Cilengitide Targeting of α_vβ₃ Integrin Receptor Synergizes with Radioimmunotherapy to Increase Efficacy and Apoptosis in Breast Cancer Xenografts¹

Novel and synergistic therapeutic combinations are required for the treatment of metastatic breast cancer, which is currently incurable with standard multimodality therapy (1). High incidence of p53 mutations and bcl-2 protein overexpression in breast cancer increase resistance to chemotherapy and radiotherapy (2, 3, 4, 5, 6, 7, 8, 9). Systemic, tumor-targeted RIT³ has the potential to target tissue specifically and to deliver cancer-specific cytotoxic antibodies to widespread metastatic foci (10). However, studies in a human breast cancer xenograft model demonstrate that RIT, as a single agent, typically does not cure the tumors (1, 11). Tumor penetration of radiolabeled antibodies may be nonuniform and may not be sufficient in all regions of the tumor to provide cure (12, 13). The combination of RIT with other therapeutic modalities is currently being used (14), but the additional chemotherapy or external radiotherapy increases the risk of bone marrow toxicity (10, 11), the major dose-limiting factor in RIT (12).

Antiangiogenic agents may provide an alternative to increase efficacy of RIT without increasing toxicity. These agents target genetically normal endothelial cells that proliferate at a much higher rate during tumor angiogenesis compared with very low endothelial turnover rates in normal tissues (15). Antiangiogenic agents have been shown to increase therapeutic efficacy in conjunction with other chemotherapeutic agents and when used in combination with external radiotherapy (14). The α_vβ₃ integrin receptor, which binds several ligands via an RGD sequence (16), is expressed in normal vasculature (17) but is highly expressed in growing tumor vasculature (17, 18, 19), making it a potential target for antiangiogenic agents (19, 20, 21). High expression and activation of the α_vβ₃ integrin has also been correlated with the more metastatic and invasive breast tumors (22, 23). Inhibition of α_vβ₃ activity by mAb and cyclic RGD peptides has been shown to induce endothelial apoptosis (19), inhibit angiogenesis (18, 20), and increase endothelial monolayer permeability (24). The inhibition of α_vβ₃ activity has been associated with decreased tumor growth in breast cancer xenografts and melanoma xenografts (18, 25, 26). Synergy of cyclic RGD peptide with antibody interleukin-2 fusion protein has resulted in increased efficacy of therapy in murine models of melanoma, colon carcinoma, and neuroblastoma (27). Selective tumor uptake has been demonstrated with radiolabeled cyclic RGD peptides (28, 29).

Our laboratory recently demonstrated up to 50% increased tumor uptake of ⁹⁰Y-labeled DOTA-peptide-ChL6 mAb after treatment with the cyclic RGD peptide EMD 270179 cyclo arg-gly-asp-D-Phe-1-amino cyclohexane carboxylic acid (RGDf-ACHA) in nude mice bearing HBT 3477 breast cancer xenografts (30). On the basis of these initial results and results demonstrating efficacy of cyclic RGD peptide in some other tumor models (25, 27), we hypothesized that Cilengitide, combined with RIT, could act synergistically in inhibiting tumor growth in the HBT 3477 breast cancer tumor model. We tested this hypothesis by treating HBT 3477 tumor bearing mice with Cilengitide and RIT. Cilengitide [cyclo (Arg-Gly-Asp-D-Phe-[Nme]-Val)] has even higher affinity for the α_vβ₃ receptor than does the RGD peptide EMD 270179 (31), and has demonstrated efficacy in melanoma tumor models as a single agent (25). Our results indicate that the combination of the antiangiogenic agent Cilengitide with RIT significantly increased efficacy and tumor and endothelial cell apoptosis without apparent increases in toxicity.

Carrier-free yttrium-90 (⁹⁰Y; Pacific Northwest National Laboratory, Richland, WA, or New England Nuclear, Boston, MA) was purchased as chloride in 0.05 m HCl. ChL6, a human-mouse antibody chimera (Bristol-Myers Squibb Pharmaceutical Research Institute, Seattle, WA), reacts with an integral membrane glycoprotein highly expressed on human breast, colon, ovary, and lung carcinomas (32, 33, 34, 35). Cilengitide [cyclo Arg-Gly-Asp-d-Phe-(N-methyl)-Val; EMD 121974] is an antagonist selective for α_vβ₃ and a_vβ₅ integrins, with IC₅₀ values in the low nanomolar range for isolated α_vβ₃ integrins and in the low micromolar range for α_vβ₃-expressing M21 melanoma cells (30). Peptide synthesis and characterization were performed as described previously (31).

HBT 3477, a human breast adenocarcinoma cell line, was obtained from Bristol-Myers Squibb Pharmaceutical Research Institute. Greater than 70% of HBT 3477 cells stain intensely with L6 (36). In HBT 3477 cells, bcl-2 is expressed and p53 is mutant, with a nonsense mutation in exon 10, resulting in a deletion in the region of the p53 protein that functions in tetramerization and in detection of double-stranded DNA breaks (37). HBT 3477 cells express functional α_vβ₅ integrin, but not α_vβ₃ integrin, because attachment to vitronectin is blocked by α_vβ₅-specific P1F6 antibody but is not blocked by α_vβ₃-specific LM609 antibody. Cilengitide blocks the attachment of HBT 3477 cells to vitronectin with an IC₅₀ of ∼5 μm.⁴

ChL6 was conjugated to DOTA and radiolabeled with ⁹⁰Y as described previously (11, 38, 39), with ≥80% efficiency to prepare ⁹⁰Y-labeled DOTA-peptide-ChL6. ⁹⁰Y-labeled DOTA-peptide-ChL6 was examined for structural and functional integrity by molecular sieving high-performance liquid chromatography (HPLC), CAE, and HBT 3477 cell-binding RIA (38). HPLC and CAE indicated that >90% of ⁹⁰Y-labeled DOTA-peptide-ChL6 was in monomeric form with less than 4% high molecular weight species as determined by CAE. Immunoreactive binding to live cells indicated a >92% reactivity. It was given as a single dose of 200, 230, or 260 μCi ⁹⁰Y-labeled DOTA-peptide-ChL6. Two hundred sixty μCi was chosen for the high-dose RIT studies, because previous studies in our laboratory had demonstrated efficacy in the HBT3477 tumor model at this dose level, typically with 100% response of tumors with few cures and deaths, making this the practical maximum tolerated dose (11). For the low-dose RIT studies, 200 μCi was chosen based on a previous study in which 230 μCi ⁹⁰Y-labeled DOTA-peptide-ChL6 resulted in few cures or complete responses but mostly partial responses (11).

Female (7–10-week old), athymic BALB/c nu/nu mice (Harlan Sprague Dawley, Inc., Frederick, MD) were maintained according to University of California animal care guidelines. HBT 3477 cells (3.0 × 10⁶), harvested in log phase, were injected s.c. into one side of the abdomen for therapy studies (except where noted) and into both sides for immunopathology studies. Injection of RIT was by tail vein and Cilengitide was delivered by i.p. injection. “Day 0” was designated as the time of RIT injection or, for first RGD injection, for RGD-only group (40). Mice were killed by cervical dislocation for immunopathology studies at the times indicated, when tumor burden exceeded allowed limits, or at 84 days for therapy studies. The 84-day period was used because it allows sufficient time for accurate characterization of HBT 3477 tumor response to RIT and had been the standard time period used for the other RIT HBT3477 therapy studies in this laboratory to which these studies were compared (1, 11).

Groups consisted of mice receiving no treatment (24 mice, 14 mice bearing two tumors each and 10 mice bearing one tumor); unlabeled ChL6 antibody (315 μg; 8 mice bearing two tumors each); and Cilengitide, given in six doses of 250 μg on days 0, 2, 4, 6, 8, and 10 (18 mice, each with one tumor).

Groups consisted of mice receiving RIT as a single agent [260 μCi ⁹⁰Y-labeled DOTA-peptide-ChL6, (39 mice, 15 bearing two tumors each and 24 bearing one tumor)]; and RIT (260 μCi ⁹⁰Y-labeled DOTA-peptide-ChL6) combined with Cilengitide in six doses of 250 μg, starting on day 0, 1 h prior to RIT, followed by five more doses on days 2, 4, 6, 8, and 10 (42 mice, all with one tumor each). The Cilengitide dose of 250 μg was chosen based on data showing the biological effect on tumor cells in mice with no toxicity at a similar dose delivered by osmotic pump over 24 h (27), or delivered i.p. as single doses 3–5 times/week (25).

Groups consisted of mice receiving RIT as a single agent [200–230 μCi ⁹⁰Y-labeled DOTA-peptide-ChL6, (28 mice, including 9 bearing two tumors each from a previous study, which received 230 μCi ⁹⁰Y-labeled DOTA-peptide-ChL6, and 19 mice bearing one tumor each, treated with 200 μCi ⁹⁰Y-labeled DOTA-peptide-ChL6)]; and RIT (200 μCi ⁹⁰Y-labeled DOTA-peptide-ChL6) combined with Cilengitide in six doses of 250 μg starting on day 0, 1 h prior to RIT, followed by five more doses on days 2, 4, 6, 8, and 10 (30 mice, all with one tumor each).

Tumors were measured with calipers in three orthogonal diameters three times a week. Tumor volume was calculated using the formula for hemiellipsoids (41). Initial tumor volume was defined as the volume on the day before treatment. Tumors that completely regressed were considered to have a volume of zero. Tumor responses were categorized as follows: cure (C), tumor disappeared and did not regrow by the end of the study (84 days); complete regression (CR), tumor disappeared for at least 7 days, but later regrew; partial regression (PR), tumor volume decreased by 50% or more for at least 7 days but then regrew; nonresponsive (NR), tumor volume decreased less than 50%. For mice bearing two tumors with differing responses, tumor response was described according to both tumor responses. Mice dying prior to 30 days from toxicity were excluded from tumor response results.

Weights and blood counts were measured two to three times per week for 12 weeks postinjection or until death. Blood samples were collected from tail veins using 2-μl microcapillary pipettes. Samples from mice within a dose group were pooled, and diluted 1:200 in PBS [PBS, 0.9% saline/10 mm sodium phosphate (pH 7.6)] 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 counts (42).

Unless otherwise noted, groups consisted of two mice, each bearing two tumors, for a total of four tumors analyzed at each time point. The groups consisted of mice receiving no treatment (four mice, seven tumors); 250 μg Cilengitide, given as single dose followed by sacrifice at 2 h, 6 h, and 1–5 days after peptide injection; RIT only (260 μCi ⁹⁰Y-labeled DOTA-peptide-ChL6) followed by sacrifice at 2 h, 6 h, and 1–6 days (three mice, five tumors at 5 days); and RIT (260 μCi ⁹⁰Y-labeled DOTA-peptide-ChL6) combined with Cilengitide (250 μg) given 1 h prior to RIT, and repeated every other day through 10 days, followed by sacrifice at 2 h, 6 h, and 1–6 days after RIT. The tumors were removed, cut in half, frozen in optimal cutting temperature (O.C.T.) medium, and stored at −70°C until sectioning (10-μm sections). All of the time points were evaluated for apoptosis by TUNEL analysis (43), and selected time points (untreated, 1, 5, and 6 days) were assessed for differences in proliferation rate (Ki67) and microvessel density (CD31).

Tumors were cut into 10-μm sections onto Fisher superplus slides (Fisher, Pittsburgh, PA), air-dried for 1 h and frozen at −70°C until TUNEL analysis with ApopTag Red kit (rhodamine used as label, Intergen, Purchase, NY) following the manufacturer’s instructions subsequent to fixation for 10 min in 1% paraformadehyde. After TUNEL, the slides were rinsed and incubated overnight at 4°C with a rat antimouse mAb against CD31 at 1:100 dilution (PharMingen, San Diego, CA) to identify endothelial cells. Slides were rinsed and incubated for 1 h with an antirat antibody linked to FITC (1:50 dilution; PharMingen). Slides were rinsed, dipped briefly in DAPI (0.2 μg/ml) for background nuclear stain, rinsed again, and mounted, followed by storage in the dark at 4°C until quantitation.

An Olympus microscope equipped with a Chroma Pinkle Filter Set (Chroma, Brattleboro, VT) with excitation filters for UV, FITC and Rhodamine and dual/triple bandpass filters to allow simultaneous viewing of multiple wavelengths, was used to quantify six randomly chosen ×600 fields (150,000 μm²/field) in nonnecrotic regions of each section. Fields were chosen to cover the entire viewing area using a DAPI label, which typically included 300–350 cells. Total apoptosis was determined by the average number of positive nuclei per field for each tumor, whereas endothelial apoptosis was determined using the same fields with a dual bandpass filter to count cells labeled by both FITC (CD31) and rhodamine (TUNEL). Fields were chosen from apparently nonnecrotic areas of tumor sections, because HBT 3477 xenografts typically grow quickly in nude mice, doubling their volumes in 6 days (11), which results in central necrosis of untreated tumors. Because TUNEL may label necrotic cells, although less intensely (44, 45), this strategy was chosen over a completely random process.

Ten-μm sections of tumors from untreated and mice treated with RGD, RIT, and CMRIT, and killed at 1, 5, and 6 days after treatment, were fixed for 10 min in ice-cold acetone, rinsed in PBS, and briefly incubated in methanol with 0.6% H₂O₂ (5 min). After rinsing in PBS, sections were blocked for 10 min with 10% goat serum and 1% BSA in PBS. Mouse anti-Ki67 mAb (PharMingen, clone B56) was applied in blocking solution (6.25 μg/ml) and the slides were incubated at room temperature for 2 h, followed by rinsing in PBS. Goat antimouse rhodamine-labeled or goat antimouse Cy-3-labeled antibody was applied (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA; 1:100) and the slides were incubated for 1 h at room temperature. After a rinsing in PBS, the sections were incubated for 1 h at room temperature with rat antimouse CD31 antibody (PharMingen, 1:100), followed by a rinsing in PBS and a subsequent 1-h incubation with goat antirat FITC-labeled antibody (PharMingen, 1:50). After rinsing in PBS, the slides were counterstained with DAPI (0.4 μg/ml) and mounted in Biomeda gel mount (Fisher) under coverslips. Quantitation of Ki67 was performed using an Olympus microscope at ×1000 magnification for an assessment of proliferation. The mean total number of Ki67-positive cells/field in a tumor was determined by counts from six fields per tumor chosen randomly by DAPI stain. Microvessel density was determined by counting the number of CD31-stained vessels per random field at ×400 magnification. Any endothelial cell or cell cluster that was positive for CD31 and that was separate from an adjacent cluster was counted as one microvessel (46). Six randomly chosen fields per tumor section were used to establish an average for each tumor (47). Average microvessel density for a treatment group was determined by averaging the values from four tumors/group.

Tumors were cut in half, frozen in O.C.T. medium (Tissue Tek; Miles, Inc., Elkhart, IN) and stored at –70°C until sectioning. Ten-μm sections were air-dried and frozen at −70°C until stained. Sections were then fixed in ice-cold acetone for 10 min, rinsed in PBS, and blocked in 10% goat serum in PBS for 30 min. Hamster antimouse CD61 (β3) mAb (PharMingen) was applied at 10 μg/ml and the slides were incubated for 3 h at room temperature. After rinsing in PBS, antihamster rhodamine-linked antibody (Jackson ImmunoResearch Laboratories, Inc.) was applied (1:50), followed by a 1-h incubation at room temperature. After rinsing in PBS, rat antimouse CD31 antibody was applied (1:100; PharMingen) and the slides were incubated for 1 h at room temperature. After PBS rinsing, antirat FITC-linked antibody (1:50; PharMingen) was applied, and the slides were incubated for 1 h at room temperature and rinsed in PBS. The slides were dipped in DAPI (0.2 μg/ml) and mounted with Biomeda gel mount (Fisher) under coverslips. Coexpression of β₃ and CD31 was observed with an Olympus microscope equipped with a Chroma Pinkle Filter Set (Chroma, Brattleboro, VT) with excitation filters for UV, FITC, and rhodamine, and dual/triple bandpass filters to allow simultaneous viewing of multiple wavelengths.

Statistical analysis of mortality data for RIT and CMRIT mice was performed using a Fisher exact test with StatExact software to determine whether mortality was different. Statistical analysis of therapy data were done using a Cochran Mantel Haenszel test to evaluate the effect of the RIT dose on outcome for RIT and CMRIT groups. A comparison of cure rates compared with all of the other responses for RIT versus CMRIT groups was done by the Fisher exact test, in the best tumor response used for 2-tumored animals. Statistical differences between immunopathology groups at the different time points were assessed by ANOVA (Fisher protected least significant difference PLSD) using STATview software as appropriate, with P < 0.05 considered significant.

Most of the tumors in untreated mice, mice receiving Cilengitide alone, and mice receiving ChL6-unlabeled antibody grew without interruption. No apparent effect of Cilengitide (6 × 250 μg doses total, given every other day) was observed on tumor growth, resulting in no cures in 18 mice tested. Two mice receiving unlabeled ChL6 antibody in addition to two untreated mice experienced spontaneous regression of their tumors, leading to a 8% (3/50) cure rate for non-RIT mice (Table 1; Fig. 1). Mice receiving RIT were treated with 260 μCi (high dose) or 200–230 μCi (low dose) ⁹⁰Y-labeled DOTA-peptide-ChL6, by itself or in combination with six doses of Cilengitide. Tumor responses in mice dying from toxicity prior to 30 days after RIT were excluded from efficacy assessment. High-dose RIT alone resulted in four cures in 26 mice (15%), whereas low-dose RIT alone resulted in five cures in 20 mice (25%). CMRIT with low-dose RIT resulted in 8 cures in 22 mice (p = 0.514), whereas CMRIT with high-dose RIT resulted in 10 cures in 19 mice (p = 0.011). Statistical analysis indicated that there was not a difference in outcome of therapy based on RIT dose adjusted for RIT or CMRIT. As the outcome was not altered by the dose (P > 0.8), outcomes for RIT were compared with outcomes for CMRIT. Our results show that CMRIT resulted in significantly more cures (44% cure rate) than RIT (20% cure rate; P = 0.020), consistent with increased efficacy of CMRIT over RIT alone.

No increased mortality or toxicity was induced by Cilengitide (250 μg, six doses over 10 days) alone compared with untreated mice [no mice died from toxicity in any of the groups that did not receive RIT (untreated, unlabeled ChL6, or Cilengitide)]. Cilengitide-treated and untreated mice displayed similar weight changes and RBC, WBC, and platelet levels. Mice treated with either high- or low-dose RIT and mice treated with the combination of RIT and Cilengitide demonstrated decreased weight, RBC, WBC, and platelet counts compared with mice receiving no RIT, but the combination of RIT with Cilengitide did not depress these values beyond those observed with RIT alone (Fig. 2, A–D). At the high dose of RIT, mortality was higher than in untreated mice in both RIT [13 (33%) of 39 mice] and CMRIT [23 (55%) of 42 mice] groups, but mortality was not statistically increased by CMRIT (Fisher exact test, P = 0.0736). Likewise, mortality was increased in low-dose RIT and CMRIT groups over untreated mice, but the mortality of CMRIT [8 (27%) of 30 mice] was not increased above RIT (8 (29%) of 28 mice; Fisher’s exact test, P = 1.000]. When low- and high-dose mortality results are combined, CMRIT does not result in increased mortality over RIT (Fisher’s exact test, P = 0.1652). These results indicate that Cilengitide, administered alone or in combination with RIT, does not significantly increase toxicity.

Apoptosis was assessed by the TUNEL method (43) in combination with CD31 staining to identify endothelial cells (Fig. 3). The number of TUNEL-positive tumor and endothelial cells (total cells) were averaged per tumor from six random fields (×600) of nonnecrotic tissue, chosen such that cells covered the entire area of the field, with an average of approximately 350 cells/field. Untreated tumors had an average of 9 ± 1.0 positive cells/field (2.6%). A single dose of Cilengitide significantly increased apoptosis 1 day after treatment [16.2 ± 1.89 (4.6%)], but this level of apoptosis subsequently decreased. Compared with untreated tumors, RIT alone resulted in significantly increased apoptosis (as noted on Fig. 4), with the greatest number of apoptotic cells seen at 6 days [21.4 ± 2.9 (6.1%)]. Apoptosis after CMRIT was higher than in all of the other treatment groups at all time points except at 6 days [significantly increased apoptosis noted (Fig. 4)]. Apoptosis after CMRIT peaked at 5 days after RIT [30.7 ± 2.0 cells/field (8.8%)], with a lower peak at 1 day after CMRIT [21.9 ± 2.6 (6.3%)]. These two peaks of apoptosis are consistent with two “waves” of apoptosis occurring. However, the difference in total cell apoptosis occurring in CMRIT tumors compared with RIT tumors is additive and, thus, suggests that other mechanisms may also affect efficacy.

Because RGD peptide had previously been shown to induce apoptosis in vascular endothelial cells (19), we first compared endothelial apoptosis in Cilengitide-treated (single dose) tumors with untreated tumors. Significantly increased apoptosis was observed 1 day (2.5 ± 0.5) and 5 days (2.4 ± 0.7) after RGD treatment compared with untreated tumors (0.6 ± 0.1). However, these differences in apoptosis between RGD-treated and -untreated tumors were not reflected by differences in growth or tumor volume even with multiple (6) doses of Cilengitide. RIT-alone also resulted in significantly increased endothelial apoptosis compared with untreated tumors at early time points (Fig. 4). CMRIT was associated with two peaks of endothelial cell apoptosis, at 1 day and 5 days after RIT, with the highest levels at 1 day (3.9 ± 1.2 cells/field). (Fig. 4). The average level of endothelial apoptosis in CMRIT tumors was higher than all of the other groups at all times, except at 3 days and 6 days. However, increased endothelial apoptosis did not appear to precede total-cell apoptosis in CMRIT tumors, with peaks occurring both at 1 day and at 5 days.

The average number of Ki67-positive cells was determined for untreated tumors and for tumors at 1, 5, and 6 days (RGD-alone was not done for 6 days). Ki67 antibody recognizes protein present in the nuclei of proliferating cells at all of the active stages of cell cycle but does not recognize protein found in cells at G₀ (48, 49). Our results indicate that RIT-alone or RGD-alone significantly decreased proliferation rates of cells active in cell cycle at 5 days compared with rates in untreated mice (Table 2). CMRIT similarly decreased proliferation compared with untreated mice and resulted in significantly decreased proliferation at 6 days compared with RIT-alone (Fig. 5).

Microvessel density for each tumor was determined by averaging the total number of noncontiguous, CD31-stained regions in six randomly chosen fields from one section of each tumor from untreated mice, RGD-treated, and RIT-and-CMRIT-treated mice at key time points after treatment (Table 2). Significantly decreased microvessel density compared with untreated mice was observed 6 days after RIT in mice receiving either RIT alone or RIT combined with Cilengitide. Increased endothelial apoptosis (above that with RIT) observed during day 1 and at day 5 did not appear to be associated with measurable differences in microvessel density and did not precede total-cell apoptosis. These data indicate that increased endothelial apoptosis associated with CMRIT may contribute to, but is not likely to be the only mechanism affecting, the therapeutic outcome. In addition, the data through day 6 do not indicate that decreased microvessel counts explain the difference in therapeutic outcome between RIT and CMRIT.

Immunohistochemistry with anti-β₃ antibody reactive with mouse integrin (CD61) demonstrated limited labeling of selected regions of HBT 3477 tumors in nude mice (Fig. 6). β₃-labeled regions included areas appearing as ring-like vascular structures, which were colabeled by an antibody recognizing CD31, which is consistent with α_vβ₃ expression in the blood vessels supplying these tumors. Substantially less expression of β₃ was observed than of CD31, which would be expected if only a small proportion of endothelial cells labeled by CD31 were neovascular.

Our results demonstrate that increased efficacy of treatment without increased toxicity can be achieved when RIT is combined with the α_vβ₃-specific Cilengitide, compared with single-modality RIT or Cilengitide therapy. CMRIT with high-dose RIT resulted in 53% cures compared with 15% cures for RIT alone, and 0% cures for Cilengitide alone, whereas combined-dose levels resulted in 44% cures compared with 20% cures for RIT alone. Although there were increased deaths associated with CMRIT at the higher-dose level, mortalities for CMRIT and RIT were not statistically different. These results indicate that significantly better outcome of therapy is associated with CMRIT compared with therapy with single agent, without an accompanying statistical increase in toxicity.

Synergy of radioimmunotherapy has been demonstrated with other antiangiogenic or chemotherapeutic agents in several tumor models (1, 12, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61). The combination of paclitaxel and radioimmunotherapy has proved effective in increasing cures or inhibiting tumor growth in breast (1, 58), prostate (59), and lymphoma (60) cancer models. mAb to epidermal growth factor receptor in combination with radioimmunotherapy also increased cures in the HBT 3477 breast cancer model (57) and in a squamous cell carcinoma model (56). Chemotherapeutic agents such as gemcitabine (52), topotecan (54), hypoxic cytotoxic agents (55), and combinations of chemotherapeutic agents (51) have demonstrated improved control of tumor growth over RIT-alone. The antiangiogenic agent 2-methoxyestradiol combined with an anticolorectal mAb improved efficacy over RIT-alone in mice bearing human colon cancer xenografts (50). Combined RIT with the antivascular agent combretastatin also improved therapeutic outcomes in colorectal xenografts (53). The combination of antivascular RIT delivered prior to tumor-targeting RIT demonstrated more homogeneous tumor uptake of tumor RIT (61). A number of other combined modality studies using external beam radiation with antiangiogenic agents have shown increased efficacy over radiation alone, including studies with angiostatin (62, 63), endostatin (64), TNP-470 (65, 66), anti-VEGF mAb (67, 68); anti-VEGFR-2 mAb (69), anti-VEGFR soluble receptors (70), VEGFR inhibitor PTK787/ZK222548 (71), and the antivascular agent NM-3 (72). The particular appeal of antiangiogenic agents is their potential to target a nonresistant population of cells without increased toxicity, making antiangiogenic agents promising for chronic therapy to prevent tumor cell recovery and metastasis (14).

The α_vβ₃ integrin receptor, targeted by Cilengitide, when bound to agonists such as vitronectin, fibronectin, and osteopontin, suppresses p53 and the bax cell death pathway (73). RGD cyclic peptides and mAbs that target and block the α_vβ₃ receptor induce apoptosis in endothelial cells, inhibit angiogenesis, and block tumor growth (18, 19, 20, 74). Cilengitide can inhibit the growth of medulloblastoma and glioblastoma brain tumors in mice (75), as well as melanoma tumors in mice (25) through its blockade of the α_vβ₃ receptor. It is currently in clinical trials (Phase I/II anaplastic glioma and Phase I Kaposi’s sarcoma)⁵ (76) with only mild side effects, including nausea, anorexia, fatigue, and malaise, with no bone marrow suppression and no dose-limiting toxicity reached when given as a twice-weekly continuous infusion of up to 850 mg/m² (76).

Because the combination of Cilengitide with RIT increased the percentage of mice cured of an aggressive, p53-mutant human breast cancer (37), we considered potential mechanisms that could account for this increase. Previous data demonstrated that 250 μg of RGD peptide (EMD 270179), provided 1 h before RIT, increased the uptake of RIT by HBT 3477 tumors up to 50% (30), which led to a consideration of whether increased tumor uptake of RIT could have resulted in increased cures. However, in a previous study of this HBT 3477 model using ⁹⁰Y-labeled DOTA-peptide-ChL6 at doses ranging from 110 μCi to 330 μCi, we found that increased cure rates did not follow increased injected doses above 260 μCi ⁹⁰Y-labeled DOTA-peptide-ChL6. These results strongly suggest that increased uptake of RIT associated with Cilengitide is not the major factor responsible for the increased cures.

As another mechanism, we investigated the vascular contribution to CMRIT by assessing endothelial apoptosis and its time course compared with tumor-cell apoptosis. If Cilengitide induced endothelial apoptosis and subsequently induced tumor cell apoptosis after the loss of endothelial cells, we would expect to observe increased endothelial apoptosis occurring prior to increased total cell apoptosis (77, 78). Cilengitide in combination with RIT elevated both endothelial and total-cell apoptosis levels above that observed with RIT at almost all time points, with significantly increased levels at 1 and 5 days. In addition, a single dose of Cilengitide significantly increased endothelial apoptosis at the same time points compared with untreated tumors. Although we did not detect a clear pattern of endothelial apoptosis preceding total cell apoptosis, there was a persistent elevation of endothelial apoptosis in CMRIT compared with RIT. However the effect of this difference was not reflected directly by differences in microvessel density at the time points assessed. It is possible that Cilengitide affected the quality of the microvascular organization, which would not have been reflected by the microvessel density (68). This is consistent with the decrease in proliferation observed at 6 days in CMRIT compared with RIT tumors, the latest time point evaluated, when the number of microvessels had decreased in both RIT and CMRIT tumors.

Other possible differences between CMRIT and RIT could be related to indirect inhibitory effects of Cilengitide. Radiation has been shown to induce the accumulation and activation of β₃ integrin on tumor blood vessels within 1–4 h of irradiation (79, 80). The increase is associated with platelet accumulation in the lumen of irradiated tumor vessels (79, 80). Platelet adhesion to endothelial cells has been inhibited by blockade with antibody to α_vβ₃ integrin (81), and platelets potentially contribute to tumor angiogenesis through growth factors contained in α-granules (82, 83). Cilengitide has been shown to have an IC₅₀ of 420 nm for isolated α_2bβ₃ receptor, compared with 5 nm for α_vβ₃ (30). It is possible that Cilengitide, at this dosage in this mouse model, inhibited the accumulation of platelets in tumor vasculature in response to radiation and, thus, decreased the paracrine interaction with tumor cells. In addition, the α_vβ₃ integrin receptor has been shown to participate in the full activation of the VEGFR-2 (84). Inhibition by Cilengitide could interfere with downstream effects of the VEGFR-2 on endothelial cells, such as decreased growth factor release to adjacent tumor cells (85), which would lead to an overall inhibition of tumor cell proliferation.

Although Cilengitide altered HBT 3477 tumor growth only in combination with RIT, it has been effective at inhibiting tumor growth in medulloblastoma and glioblastoma tumors in mice (100 μg/day). Specifically, the peptide is effective in tumors implanted orthotopically in the brain but not in tumors implanted s.c. in the same mice (75). RGD peptide (250 μg/5x weekly) has also shown efficacy in inhibiting the growth of melanoma tumors when implanted s.c. in mice via direct inhibition of tumor cells (25). These data indicate that some tumors may be more sensitive to inhibition by this peptide, possibly because of increased sensitivity to ligand expression and blockade (75), differences in tumor cell expression of α_vβ₃ receptors (25), and differences in the activation level of receptors (23). The lack of responsiveness of HBT 3477 s.c. tumor growth to Cilengitide inhibition (single modality) likely indicates that, in this model, tumor growth uses existing blood vessels or that sufficient blood vessel growth occurs even in the presence of Cilengitide. After RIT, at a time when blood vessel density has decreased substantially, Cilengitide inhibition may lead to decreased tumor cell and endothelial cell recovery from radiation damage, resulting in increased tumor cell death in waves of apoptosis, related to increased therapeutic efficacy.

In conclusion, our results are consistent with the increased efficacy of RIT of breast cancer tumors resulting from combined treatment with the anti-α_vβ₃ peptide, Cilengitide. Although these results were obtained in a single model of breast cancer, the molecular biology of this tumor has been documented to be relevant to the sources of therapy failures in patients with metastatic breast cancer (36, 86). The therapeutic synergy is likely caused by the combined effects of several mechanisms, leading to increased apoptosis and decreased proliferation. The higher level of endothelial apoptosis observed with CMRIT treatment would, however, be consistent with endothelial loss impacting tumor cell loss, and contributing to the observed increase in cures. The enhanced therapeutic response achieved by the addition of the antiangiogenic agent, Cilengitide, to RIT is remarkable because the therapeutic synergy was achieved without additional toxicity, which indicates immense potential for Cilengitide specifically, and antiangiogenic agents in general, for combination therapy with RIT.

The authors thank Dr. A. Joncyzk, Merck KGaA for the gift of Cilengitide, Dr. Simon L. Goodman, Merck KGaA Darmstadt for his thoughtful review and suggestions, and David L. Kukis, University of California at Davis Sacramento Medical Center, Sacramento, CA for his expert support in providing all radiolabeled material.

Prof. Matzku passed away on May 27, 2002.