Antiangiogenic therapies are frequently used with concomitantly administered cancer chemotherapy to improve outcomes, but the mechanism for the benefit of the combination is uncertain. We describe a mechanism by which a specific, cytotoxic antivascular agent causes vascular remodeling and improved chemotherapy results. By selectively killing tumor neovasculature using short-ranged α-particles targeted to vascular endothelial (VE)-cadherin on vascular endothelial cells (by use of 225Ac-labeled E4G10 antibody) we were able both to reduce tumor growth and to increase the efficacy of chemotherapy, an effect seen only when the chemotherapy was administered several days after the vascular targeting agent, but not if the order of administration was reversed. Immunohistochemical and immunofluorescence studies showed that the vasculature of 225Ac-E4G10–treated tumors was substantially depleted; the remaining vessels appeared more mature morphologically and displayed increased pericyte density and coverage. Tumor uptake and microdistribution studies with radioactive and fluorescent small molecule drugs showed better accumulation and more homogenous distribution of the drugs within 225Ac-E4G10–treated tumors. These results show that 225Ac-E4G10 treatment leads to ablation and improvement of the tumor vascular architecture, and also show that the resulting vascular remodeling can increase tumor delivery of small molecules, thus providing a process for the improved outcomes observed after combining antivascular therapy and chemotherapy. This study directly shows evidence for what has long been a speculated mechanism for antiangiogenic therapies. Moreover, targeting the vessel for killing provides an alternative mode of improving chemotherapy delivery and efficacy, potentially avoiding some of the drawbacks of targeting a highly redundant angiogenic pathway. Cancer Res; 70(22); 9277–86. ©2010 AACR.
Therapies focusing on the tumor vasculature play an important part in the management of late-stage malignancies. Several drugs that target the vascular endothelial growth factor (VEGF) pathway have been approved and are in clinical use, whereas others are in different stages of clinical development. Bevacizumab, an anti-VEGF antibody, is routinely used in combination with chemotherapy in the treatment for metastatic colon, non–small cell lung, and breast cancers. The multi–tyrosine kinase inhibitors sorafenib and sunitinib, which target, among others, the VEGF-receptor 2 (VEGFR2), are approved as single agents in the treatment of renal cell carcinomas, gastrointestinal stromal tumors, and hepatocellular carcinomas. As tumor angiogenesis is a process necessary for tumor growth, interference with VEGF signaling could produce strategies that are broadly effective. Moreover, because VEGF affects the host endothelium and not tumor cells, development of resistance against these drugs should be less likely. Unfortunately, in preclinical models and in several clinical trials with potent VEGF inhibitors, many tumors failed to respond. Furthermore, even tumors initially responsive to anti-VEGF/VEGFR2 therapy finally acquire resistance to the treatment, and relapse occurs in virtually all patients (1–3). Thus, although antiangiogenic therapy has proven to be a valuable tool in cancer treatment, there is a need for alternative strategies to target the tumor vasculature; in addition, the mechanisms of action for antiangiogenic therapy in combination with chemotherapy are poorly understood.
Vascular endothelial (VE) cadherin is an endothelial cell–specific protein that is expressed constitutively throughout the vasculature and participates in the formation of adherens junctions between adjacent endothelial cells. The monoclonal antibody (mAb) E4G10 specifically binds to a cryptic epitope exposed only on the monomeric terminal domain of the unengaged form of VE-cadherin, found in the neovasculature, and not on the dimeric form, which is present in mature endothelium (4). We looked to exploit this specificity to investigate the effects of neovascular ablation in tumors by conjugating E4G10 to Actinium-225, an α-particle–emitting isotope generator. α-Particles exhibit high linear energy transfer and deposit large amounts energy (5–8 meV) over a pathlength of cellular dimensions (∼60–80 μm; ref. 5). These properties confer α-particles with the potential to kill individual cells with just one nuclear hit, thus reducing the chance of developing resistance, while specifically destroying targeted cells and limiting bystander cell toxicity (6).
We had previously reported the design and use of a neovascular VE-cadherin–targeted α-particle immunotherapeutic agent that was effective with paclitaxel chemotherapy in a prostate cancer model when used in combination, similar to the reported effects of antiangiogenic drugs (7, 8). Unlike other antiangiogenic or antivascular strategies, 225Ac-E4G10 does not target a tumor-specific pathway that can be circumvented by other angiogenic pathways; rather, it directly kills the new endothelial cells (7, 9). These properties make 225Ac-E4G10–based therapies an interesting alternative to other antiangiogenic or antivascular strategies. In addition, the highly selective nature of the direct killing of the vessels, due to the short range of the α-particle, allows us to isolate the effects of the therapy on the tumor endothelial cells alone without significant tumor cell killing.
The mechanism by which antiangiogenic therapies make chemotherapy more effective remains controversial. One theory explaining the additive effects of antiangiogenic therapy and chemotherapy is “vessel normalization”: morphologic improvements and maturational changes are induced by the therapy in residual tumor vessels, which are initially tortuous, dilated, leaky, and associated with erratic blood flow, hypoxia, and increased interstitial pressure. Normalization of the tumor vessels, with reduced interstitial pressure and hypoxia, would enhance delivery and efficacy of subsequent cytotoxic agents, such as chemotherapeutic drugs, explaining improved outcomes observed in preclinical and clinical trials (10, 11). However, there is little experimental evidence that antiangiogenic therapy improves drug delivery, and alternative models have been suggested. For example, tumor cell repopulation, which would occur after chemotherapy ceases, might be mitigated by angiogenesis inhibition during the chemotherapy-free period (12, 13). Another mechanism, which may be consistent with vessel normalization, suggests that antivascular or VEGF-inhibition therapies may prevent endothelial progenitor cells from infiltrating the tumor and providing the vascular network necessary for further tumor growth (9). Although the preclinical and clinical observations strongly support treating cancers with antiangiogenic or antivascular therapy in conjunction with chemotherapy regimens, optimization of their efficacy will require a better understanding of why these combinations are additive (14).
Here, we report that treatment with 225Ac-E4G10 in a human colon cancer xenograft model results in reduced tumor burden, tumor vasculature remodeling, and improved tumor uptake of three different small molecule agents, as well as enhanced therapeutic response to chemotherapy. These data provide a rational pharmacokinetic mechanism to explain the benefits of combining antivascular therapy with chemotherapy in a schedule-dependent manner, which is something that has long been speculated as the mechanism of antiangiogenic agents but has yet to be shown directly.
Materials and Methods
Female athymic nude mice (NCr nu/nu), 4 to 12 weeks of age, were obtained from Taconic. Animal studies were conducted according to the NIH Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee at Memorial Sloan-Kettering Cancer Center (MSKCC).
Preparation, quality control, and administration of radioimmunoconjugates
225Actinium (Oak Ridge National Laboratory or Curative, Inc.) was conjugated to E4G10 using a two-step labeling method as described (8). Routine quality control of the labeled antibody was performed using instant thin-layer chromatography to estimate radiopurity, and cell binding assays were done to determine immunoreactivity. Mice were anesthetized with isofluorane and injected i.v. (in retro-orbital venous plexus) with the radioimmunoconjugate. The injected volume was 100 μL and the antibody dose was 0.3 to 0.5 μg per 60 nCi injection. Typical radiochemical purity was 96% to 99%.
Tumor implantation in mice
LS174T cells were obtained from MSKCC stocks. Cells were grown in DMEM supplemented with glucose, nonessential amino acids, L-glutamine, 10% fetal bovine serum in an atmosphere of 5% CO2 and air at 37°C. Cells were harvested and approximately 3 million cells were injected in 100 μL of matrigel (BD Biosciences) into the right flank of each animal.
Tumor therapy studies
Randomized control therapy studies with 225Ac-E4G10.
Cohorts of 8 to 10 tumor-bearing animals were treated with either saline, unlabeled E4G10 mAb (10 μg), or 225Ac-E4G10 mAb (60 nCi, ∼0.4 μg). Treatments were administered i.v. on days 5, 7, and 9 postimplantation with 3 million LS174T cells; however, unlabeled E4G10 (10 μg) was given as a single bolus dose administered at day 5 post–tumor implantation. In other studies, 225Ac-E4G10–treated animals were compared with 225Ac-IgG2a isotype control. Tumor size was measured with calipers, and tumor volume was calculated by the formula 0.52 × d12 × d2, where d1 is the smaller diameter and d2 is the larger diameter. Day 5 postimplantation of tumor was selected as the first day of treatment because it is the day during which new vessels first clearly appear, as determined in previous histologic studies. Mice were sacrificed when tumor size was >2,500 mm3.
Controlled, drug sequence trials.
Mice were implanted with 3 million LS174T cells on day 0. Each cohort of 8 to 10 mice then received two different treatments in a specified sequence. Saline, chemotherapy [1.8 mg leucovorin (Teva Parenteral Medicines) + 0.6 mg 5-fluorouracil (FU; Abraxis Pharmaceutical Products), i.p.], or 225Ac-labeled E4G10 (60 nCi, on ∼0.5 μg antibody, i.v.) was administered at days 5, 7, and 9 (first round of therapy), and at days 15, 17, and 19 (second round of therapy) post–tumor implantation as indicated in the groups. Because this model system was designed specifically to look at the effects of the drugs on the neovasculature, not necessarily to achieve the maximum tolerated effects on tumor growth, doses were selected to cause ∼50% reduction in tumor growth rate relative to saline treatment as determined in pilot experiments to allow differentiation between effects of the first round of treatments from the second. Three distinct treatment groups were randomized to: (a) saline followed by saline, (b) leucovorin/FU followed by 225Ac-E4G10, and (c) 225Ac-E4G10 followed by leucovorin/FU. Tumor dimensions were measured and tumor volume was calculated. Animals were followed for up to 111 days post–tumor implantation for survival (statistics determined with log-rank/Mantel-Cox test).
A fraction of the mice (n = 3 per group) from the aforementioned treatment groups were sacrificed at day 15 postimplantation of tumor, and tumors were either cryofixed (OCT Tissue-Tek, SAKURA Finetek U.S.A. Inc) or fixed with 4% paraformaldehyde and paraffin embedded for histopathologic study. Vascular density was determined in tissue sections stained for Pan ECA (pan endothelial cell antigen, Ab-Clone Meca 32, Dako); immunofluorescent staining for collagen IV, a major component of vascular basement membrane, assessed the basement membrane of vessels and helped differentiate between vascular ablation and inhibition of angiogenesis; staining for both α-smooth muscle actin (α-SMA) and neuron-glia antigen 2 (NG2), both markers of pericytes, was performed to quantitate pericyte density and coverage of tumor vasculature. All quantification was done after scanning whole stained tumor sections, three per sample. Epifluorescence or bright field scanning was performed on a Zeiss Mirax Scan system using a 40× lens (Carl Zeiss Microimaging). Positive stained areas in the obtained high-resolution images were quantified using ImageJ (http://rsbweb.nih.gov/ij/). Quantification results were analyzed with Prism software (Version 5.0, GraphPad software) for statistical significance using t-test (unpaired, two-tailed). In addition, two-dimensional and three-dimensional images were acquired by laser confocal microscopy (Leica TCS SP2, Leica Microsystems) or bright field microscopy (Zeiss Axiovert 200M) of selected fields from the same sections to illustrate quantification results.
Preparation of 111In-DTPA
Vascular penetration and drug diffusion into tumors was assessed using the radiolabeled construct 111In-DTPA, which was prepared by adding a mixture of 1.01 mCi of acidic 111In chloride (MDS Nordion) and 3M ammonium acetate (Aldrich) in 0.020 mL to 0.080 mL of a 50 mmol/L solution of diethylenetriaminepentaacetic acid (DTPA) solution at 20°C for 60 minutes, and was tested for purity by instant thin-layer chromatography (Gelman Science Inc.); it was mixed with 1% human serum albumin (Swiss Red Cross) and 0.9% NaCl (Abbott Laboratories) to achieve 0.097 Ci/L the day of injection.
Biodistribution, permeability, and accumulation of small molecules into tumors
Tumor-bearing animals (n = 8 per group) were anesthetized with isofluorane and injected with 111In-DTPA (5 μCi) retro-orbitally and sacrificed after 5 minutes or 60 minutes to evaluate 111In-DTPA vascular space and tissue space, respectively. Animals were also injected with Hoechst 33342 dye for high-resolution fluorescence perfusion studies 20 minutes prior to sacrifice. The percent injected dose per gram of indium-111 in tissues was determined after sacrifice and analyzed with Prism software (Version 5.0, GraphPad software) for statistical significance using t-test (unpaired, two-tailed).
Tumor samples from mice injected (n = 6 per group) with 111In-DTPA 60 minutes earlier were collected, cryofixed (OCT Tissue-Tek, SAKURA Finetek U.S.A. Inc.), and cut with a cryostatic microtome (HM 500 M, Microm GmbH) to 30-μm sections. After air drying, the slide-mounted sections were covered with a sheet of 5-μm-thick mylar and pressed against a phosphor screen (type BI) in a light-tight cassette and the screen was exposed for 3 days. The screen was then scanned in a Molecular Imager System (model GS-363, Bio-Rad Laboratories) using a laser scanner at a spatial resolution of 100 μm. For image analysis, ImageJ was used to convert grayscale images into 8-bit binary images and the number of positive pixels per mm2 was determined.
Graphs were constructed using Prism (Graphpad Software Inc.). Statistical comparisons between the experimental groups were by Student's t-test (unpaired, two-tailed), or log-rank/Mantel-Cox test depending on the analysis.
Pretreatment with 225Ac-E4G10 enhances subsequent chemotherapy effects
The LS174T human colon adenocarcinoma xenograft model was selected because it has been extensively used to investigate therapeutic effects on the tumor vasculature (15–18). Based on earlier pilot studies, the treatment dose of 225Ac-E4G10 was chosen to cause ∼50% reduction in tumor size (from saline treatment) and allow us to determine the effects of subsequent treatments. LS174T-engrafted mice treated with 225Ac-E4G10 (60 nCi, ∼0.5 μg/mice/dose on days 5, 7, and 9) showed significant reduction of tumor growth compared with saline, unlabeled E4G10 (10 μg), and, in a separate experiment, 225Ac-IgG2a (isotype control mAb; Fig. 1A and B). This indicated that the antitumor effect was caused by directed α-particle damage to VE-cadherin–positive endothelium and not by binding of the unlabeled antibody alone to VE-cadherin or by a nonspecific effect of the α-radiation within the vessel lumen or in extravascular tissues (19). The standard management of colorectal cancers includes regimens based on FU, a thymidylate synthase inhibitor, and coadministration of leucovorin. As with 225Ac-E4G10, pilot experiments were performed to determine leucovorin/FU dosage, which suppresses tumor growth by 50% to ensure the effect of a second agent would be discernable (Supplementary Fig. S1A).
Next, we studied the therapeutic efficacy of combining 225Ac-E4G10 with the standard chemotherapy and specifically asked whether the sequence of administration of the two drugs was important in outcome. First, LS174T-engrafted animals received saline, 225Ac-E4G10, or leucovorin/FU in on days 5, 7, and 9 post–tumor implantation, after tumors were fully established and vascularized. Mice that received active treatments (225Ac-E4G10 or leucovorin/FU) showed similar reduction of tumor growth when compared with saline-injected controls for the first 15 days following tumor implantation, as expected based on the dose selection of both drugs (Fig. 1C). On days 15, 17, and 19 post–tumor implantation, the animals received a second round of treatment of the other drug. Animals treated with saline in the first round exhibited rapid tumor growth, and 4 of 9 animals reached the survival end point (tumor volume >2,500 mm3) by day 15 (Fig. 1C). The remaining animals in this group were followed to determine median survival, which was 22 days (Supplementary Fig. S1B). The rapidity with which animals in the saline-only pretreatment group reached survival end points precluded the study of the effect of an active agent only in the second round for these animals.
Animals treated with chemotherapy in the first treatment round followed by 225Ac-E4G10 exhibited continued tumor growth and a median survival of about 26 days, which was not significantly different from the saline control group (P = 0.24). In contrast, animals first treated with 225Ac-E4G10 and then treated with chemotherapy showed little additional tumor growth during this time period compared with the inverse treatment (P = 0.03; Fig. 1C and D). Animals in the group that received 225Ac-E4G10 followed by leucovorin/FU regimens had a median survival of 41.5 days which was significantly higher than leucovorin/FU followed by 225Ac-E4G10 (P = 0.019, log-rank/Mantel-Cox test). These data confirmed that the drug combination alone did not enhance survival; rather, it was the sequence of administration that was important for treatment efficacy. These results led to the hypothesis that our antivascular therapy was specifically affecting the efficacy of subsequent chemotherapy.
225Ac-E4G10–treated tumors show normalized, less abundant vessels and increased pericyte number and density
To further understand the mechanism for the effects of 225Ac-E4G10 pretreatment on the efficacy of subsequent leucovorin/FU, we studied the morphologic changes in the tumor vasculature caused by 225Ac-E4G10 on day 15 post–tumor implantation, the day the second treatment round was started. 225Ac-E4G10 showed significantly lower endothelial cell counts, as visualized by staining for Pan ECA (pan endothelial cell antigen, Ab-Clone Meca 32), when compared with control and leucovorin/FU-treated tumors (Fig. 2A–C). In addition, tumors treated with saline or chemotherapy showed collagen IV staining associated with endothelial cells; however, 225Ac-E4G10–treated tumors showed tube-shaped collagen IV structures similar to those observed in the controls, which were not associated with endothelial cells (Fig. 2D–F). These collagen IV ghost structures often were still filled with blood cells and were indicative of a selective endothelial killing by 225Ac-E4G10 (20). Quantification of tumor sections showed significantly more collagen sleeves with 225Ac-E4G10 treatment (Fig. 2G). Vessels of tumors treated with 225Ac-E4G10 were not only less abundant, but also smaller, consistent with previous data (Fig. 2H). Additionally, tumor tissue staining with E4G10 mAb to probe for the unengaged VE-cadherin on neovascular cells showed significantly less immunoreactivity in 225Ac-E4G10–treated tumors when compared with control (P = 0.01) and chemotherapy-treated tumors (P < 0.01).
Quantification of pericytes by staining for α-SMA was performed on tumor samples. Marked pericyte staining was observed after tumors were treated with 225Ac-E4G10, whereas controls showed very low pericyte abundance (Fig. 3A and B). As endothelial cells were depleted (Fig. 3C) and pericyte counts were increased, 225Ac-E4G10 treatment clearly results in significantly more pericytes per vessel area (α-SMA versus Meca32; Fig. 3D). Thus, pericytes may have infiltrated or proliferated within tumors in the 225Ac-E4G10 group, and remaining blood vessels were more often supported by pericytes when compared with controls (Fig. 3E). These experiments were repeated with the pericyte marker NG2, which is selective for a broader subset of pericytes than α-SMA. Results with NG2 confirmed that 225Ac-E4G10 treatment not only resulted in a selective depletion of immature blood vessels, which lacked pericyte coverage, leaving only mature pericyte-covered vessels, but also promoted the recruitment of pericytes (Fig. 3F).
Apoptosis, as measured by cleaved caspase 3 staining, was also significantly higher in 225Ac-E4G10–treated tumors (Supplementary Fig. S2A). Hypoxia evaluated by staining for carbonic anhydrase IX (CAIX), which is upregulated by hypoxia inducible factor-1 α (HIF-1-α), was not significantly affected by our treatment (ref. 21; Supplementary Fig. S2B). Of note, LS174T tumors exhibited a much higher level of CAIX staining that other tumor sections we had previously encountered (data not shown), and these results may be a consequence of a high baseline hypoxia in control-treated animals such that more minor perturbations in hypoxia with 225Ac-E4G10, if any, would be masked.
Pretreatment with 225Ac-E4G10 enhances tumor accumulation of 111In-DTPA, Hoechst 33342, and [18F]-3′-fluoro-3′-deoxy-L-thymidine
Histologic evaluation clearly indicated more mature tumor vasculature after treatment with 225Ac-E4G10. However, the mechanism by which this effect leads to increased chemotherapy efficacy was unknown; therefore, we decided to test if the vessel remodeling also resulted in an improved delivery of small molecule drugs into the tumor. Biodistribution studies were performed using DTPA-chelated 111Indium, a freely diffusible small molecule tracer (22). Sixty minutes after administration of 111In-DTPA, there was a significant (P = 0.01) increase in total 111In accumulation (percent injected dose per gram) in tumors treated with 225Ac-E4G10 (Fig. 4A). This effect was more apparent when tumor signal was normalized to blood signal (Fig. 4B). Biodistribution studies performed 5 minutes after administration of 111In-DTPA showed no difference in tumor accumulation between 225Ac-E4G10 and 225Ac-IgG2a isotype control (Supplementary Fig. S3A). 111In activity in other organs was also assessed at both time points and no consistent treatment-dependent changes in uptake were observed.
Autoradiograms of the mice examined 60 minutes postinjection showed more tumors with high activity in the 225Ac-E4G10 treatment group when compared with 225Ac-IgG2a controls (Fig. 4C). To measure homogeneity of signal distribution, the percent positive areas per tumor were evaluated and 5 of 6 (83%) of 225Ac-E4G10–treated tumors had >50% positive areas, whereas only 2 of 6 (33%) of animals treated with 225Ac-IgG2a showed >50% positive area. We corroborated these results further by evaluating the distribution of another small molecule, Hoechst 33342, in 225Ac-E4G10–treated and untreated tumors. Hoechst 33342, which is fluorescent and can be detected after intercalation into DNA, was injected i.v. into tumor-bearing animals 20 minutes prior to euthanasia and tissue sections were scanned. Evaluation of the fluorescence signal by epifluorescence or laser-assisted confocal microscopy yields the in-tissue distribution and serves as a marker of perfusion. The area percentage positive for Hoechst 33342 significantly (P < 0.05) increased after 225Ac-E4G10 treatment, and tumors also showed a more homogenous perfusion (Fig. 4D).
We also evaluated tumor accumulation and penetration in experimental groups using [18F]-3′-fluoro-3′-deoxy-L-thymidine (18FLT), a positron emission tomography radiotracer analog of FU for biodistribution studies. Ten days after pretreatment with 225Ac-E4G10 started (day 15 post–tumor implantation), the accumulation of 18FLT in the tumor tissue was significantly enhanced when compared with 225Ac-IgG2a isotype treatment (P < 0.003), suggesting a similar effect for FU and explaining the increased efficacy of the treatment (Supplementary Fig. S3B). Other tissues showed no consistent differences in 18FLT uptake across treatments.
We show that directed, vascular killing of tumor neovascular endothelium by use of short-range (∼60–80 μm) cytotoxic α-particles is an effective treatment modality that enhances subsequent chemotherapeutic efficacy. We describe a mechanism whereby 225Ac-E4G10 induces vessel remodeling and leads to improved penetration of subsequently administered small molecules. A number of studies have shown the additive effects of antiangiogenic compounds and chemotherapy in preclinical and clinical trials; however, the mechanisms to explain the results were unclear and controversial (10, 14). The data here suggest that the order of administration of the two types of agents, antivasculature and antitumor, is critical to successful outcomes and may have an important impact on the design of combination therapies for clinical trials.
Antiangiogenic agents, including anti-VEGF or anti-VEGFR mAbs, have been studied extensively preclinically, and bevacizumab has been successful when combined with chemotherapy. One mechanism that explains the additive effects observed with bevacizumab and chemotherapy is vascular normalization with remodeling caused by antiangiogenic therapy, leading to improved drug accumulation. In mice, treatment with DC101, an anti-VEGF receptor-2 mAb, increased pericyte coverage, reduced microvessel density, reduced interstitial pressure, and improved oxygenation, but these studies stopped short of proving enhancement in drug accumulation (10). Improved accumulation of TRITC-labeled albumin has also been shown after bevacizumab, suggesting improved perfusion, and increased accumulation of subsequent topotecan has been reported by Dickson and colleagues in neuroblastoma xenografts in mice (10, 14). However, the use of bevacizumab in mouse models is problematic as the antibody does not bind or neutralize murine VEGF-A (23). Although bevacizumab can effectively block VEGF derived from the implanted human tumor cells, it has been previously shown that host-derived VEGF significantly contributes in a model-dependent manner to the angiogenic process (24). Thus, although improved drug accumulation or uptake in tumors has been speculated as a mechanism for antiangiogenic therapies, little direct pharmacokinetic evidence for this phenomenon exists.
In contrast to antiangiogenic agents, 225Ac-E4G10 is a cytotoxic, vascular-targeting agent that selectively binds to the neovasculature of tumors, depleting the endothelial cells within tumors, as well as their progenitors in the blood and bone marrow, but not the tumor cells themselves (9). Endothelial progenitor cells have been shown to play a role in vasculogenesis within tumors, and vascular depletion, as well as the antitumor effects observed in our study, could be a result of specific killing of both endothelial cells within tumors and killing of endothelial progenitor cells. The relative contribution of the endothelial progenitor cell population is an interesting question for future studies.
The specificity of E4G10 to the neovasculature allowed us to isolate the effects of antivessel therapy from other antitumor effects in our study, but may not, consequently, generalize to all antiangiogenic therapies. However, like anti-VEGF therapies, 225Ac-E4G10 treatment also results in phenotypic vascular normalization without targeting a highly redundant angiogenic pathway. Thus, 225Ac-E4G10 treatment has the potential to avoid some of the pitfalls observed with classical anti-VEGF pathway therapeutic drugs. However, further experiments are necessary to determine whether 225Ac-E4G10 treatment might be a second-line alternative for anti-VEGF–resistant tumors.
Unlabeled E4G10 at >4,000 times the dose of 225Ac-E4G10 used here has some antitumor effect, but it did not at 10 times the dose we used (shown in control groups; ref. 19). In addition, α-emitters on nonspecific mAb were ineffective. These data show that 225Ac-E4G10 was not operating by binding to, or blocking, VE-cadherin, but by delivering the α-emitter selectively to kill the targeted endothelial cells.
The importance of sequence of administration was evident when animals treated first with 225Ac-E4G10 and then with leucovorin/FU chemotherapy exhibited both decreased tumor burden and improved median survival as compared with animals receiving the inverse drug schedule. Histologic examination of 225Ac-E4G10–treated tumors showed smaller, less densely dispersed vessels. Additionally, vascular basement membrane (collagen IV) structures were present in the absence of endothelial cells with 225Ac-E4G10 treatment, whereas control and chemotherapy-treated tumors showed endothelial cells adjacent to the collagen scaffold, confirming that our treatment ablates endothelial cells (20). These collagen IV structures have been implicated in the tumor revascularization that occurs once antiangiogenic therapy is ceased by serving as a scaffold for the neoendothelium (25). Apoptosis was also higher in 225Ac-E4G10–treated tumors, which could be a combination of endothelial cell death, tumor cell death as a consequence of endothelial cell death, or direct tumor cell killing, which might be expected given that the α-particle path length is several cell diameters (∼60-80 μm); the gross differences in tumor volume may be a consequence tumor cell death resulting from any of these phenomena.
Pericytes are known to contribute to the integrity of mature vessels, and when absent result in abnormal function and leakiness (26). Morphologically, 225Ac-E4G10–treated tumors showed significantly more pericytes surrounding smaller, less dilated vessels and higher pericyte coverage (pericytes per vessel) when compared with saline and leucovorin/FU treatment. Interestingly, 225Ac-E4G10–treated tumors showed a higher total number of pericytes adjacent to vessels, suggesting proliferation of intratumoral pericytes or recruitment of pericytes, rather than simply a product of immature vessel ablation. These histologic results are consistent with what occurs during vascular normalization by anti-VEGF therapy.
Previous studies showed that antiangiogenic therapies have resulted in decreases in intratumor pressure, improved vessel function with decreased vascular leakiness, yet few have determined if there is improved penetration of the chemotherapy into the tumor upon vessel normalization (10, 27, 28). Our findings show a novel and specific method for achieving vessel improvements in tumors. Furthermore, accumulation and homogeneity of tumor distribution of 111In-DTPA by ex vivo tissue harvest and autoradiographic analysis, and of Hoechst 33342 dye by immunofluoresence, were all improved in 225Ac-E4G10–treated animals.
18FLT also entered tumors more effectively as measured by positron imaging and ex vivo tissue harvest in 225Ac-E4G10–treated animals than control. As 18FLT is metabolized by intracellular thymidine kinase, the total signal in tumors after treatment could be an amalgam of increased accumulation driven by 225Ac-E4G10–induced remodeling of tumor vasculature and increased metabolic activity in the tumors. If there is increased metabolic activity, this further supports the selectivity of 225Ac-E4G10 for the vessel and not the tumor.
Because our drug results in vascular remodeling, improved perfusion, and improved small molecule drug accumulation within tumors, one might expect 225Ac-E4G10 to cause a decrease in hypoxia as well. However, hypoxia levels of tumor sections by CAIX staining were not significantly altered between treatments. Typically, we found LS174T tumors to exhibit a much higher level of CAIX staining (at least four times more) than tumor sections derived from other cell lines that we had previously encountered (data not shown), and CAIX staining results may be a consequence of a high hypoxia baseline in these tumors such that more minor perturbations in hypoxia with 225Ac-E4G10 would be masked if present. In addition, although antiangiogenic treatment with inhibitors of the VEGF pathway is hypothesized to cause vascular normalization and at least in one previously discussed setting improved drug uptake (14), it is also frequently reported to actually increase hypoxia both in patients and in model systems (29–32). This observation is often cited as a reason for the failure of these therapies and development of resistance and a more aggressive phenotype (33, 34). Our findings combined with results in the literature suggest that the effects of vascular changes on oxygen and drug transport can sometimes be decoupled. Given these differing data, it seems that our antivascular approach mimics certain effects on the vasculature previously observed with antiangiogenic therapies, i.e., normalized vasculature and improved perfusion, while exhibiting no detectable effects on hypoxia.
In conclusion, 225Ac-E4G10 treatment potently killed neovascular endothelial cells, remodeled tumor vasculature without directly targeting an angiogenic signaling pathway, and improved the pharmacokinetics of subsequently administered small molecules in tumors, thus providing a mechanism for the improved therapeutic efficacy of combinations with chemotherapy. These observations also highlight the importance of therapeutic scheduling when employing agents targeting tumor vasculature.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
We thank ImClone Systems for providing E4G10 mAb, and MSKCC small animal imaging core.
Grant Support: NIH R01, CA55349, CA 33049, DOE DE-SC0002456, NIH MSTP GM07739, Lymphoma Foundation, Glades and Tudor Foundations, and the Experimental Therapeutics Center.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
- Received June 4, 2010.
- Revision received August 12, 2010.
- Accepted September 8, 2010.
- ©2010 American Association for Cancer Research.