Angiogenesis Inhibition and Choroidal Neovascularization Suppression by Sustained Delivery of an Integrin Antagonist, EMD478761

Abstract

purpose. To evaluate the angiogenic inhibitory effects of an α_vβ₃/α_vβ₅ integrin antagonist, EMD478761, released from a polymeric implant in a chick chorioallantoic membrane (CAM) assay and laser-induced experimental choroidal neovascularization (CNV) in rats.

methods. Polyvinyl alcohol-based reservoir implants releasing EMD478761 were designed for placement onto a CAM or intravitreally in rats. In vitro release rates of the implants were measured using HPLC. Angiogenesis was induced on 10-day-old chick embryos by basic fibroblast growth factor (bFGF), and areas of neovascularization were measured. Experimental CNV was induced in the Brown-Norway rat with a diode laser. EMD478761 or sham microimplants were placed within the vitreous chamber of Brown-Norway rats. Two weeks later, areas of CNV were determined by FITC-dextran staining of choroidal flatmounts.

results. Sustained delivery of EMD478761 significantly inhibited bFGF-induced angiogenesis in CAM, as determined by a reduction in angiogenesis areas, without drug toxicity to the normal CAM vasculature. In an experimental rat model, intravitreal EMD478761 implants significantly suppressed laser-induced CNV compared with intravitreal sham implants, with the mean area reduced by 63% (P < 0.05).

conclusions. Sustained delivery of EMD478761demonstrates potent antiangiogenic properties in vivo. These results suggest that an EMD478761 implant may be beneficial in the treatment of neovascular ocular diseases.

Age-related macular degeneration (AMD), both dry and wet forms, is the leading cause of legal blindness among aged persons in the developed countries. ¹ Wet (neovascular) AMD, characterized by choroidal neovascularization (CNV) and fluid accumulation, is responsible for 90% of the vision loss, and subjects with dry AMD are at risk for the wet form. ^{2 3} The number of people affected is expected to double by 2020 because of the aging of the population. ⁴ Current treatments available for CNV, including conventional laser photocoagulation, photodynamic therapy (PDT), and repeated intravitreal anti-VEGF therapies, while successful in halting vision deterioration in most patients, are only effective in improving vision in 6% to 33% of patients. ^{5 6 7 8} In addition, patients undergoing these treatments often have recurrences of AMD and require repeated treatments. Given the short half-life of most drugs injected into the vitreous, ^{9 10} repeated intravitreous injections may be required to maintain drug concentration within the therapeutic window. However, multiple intraocular injections, to patients who may need months or years of treatment, have been associated with cataract formation, retinal detachment, and endophthalmitis. ^{11 12 13} To improve the therapy for CNV, alternative therapeutic modalities, including novel antiangiogenic compounds and delivery methods, are being explored. 

CNV is a result of pathologic angiogenesis, which is partially mediated by the alteration of integrin expression. Integrins, a superfamily of transmembrane glycoprotein receptors, participate in cell-to-cell and cell-to-substrate interactions. ¹⁴ Integrins are heterodimers consisting of noncovalently associated α and β subunits. To date, 24 integrins, with different combinations of α and β chains, have been described in mammalian cells. In ocular angiogenesis, integrin α_vβ₃ is selectively expressed in the CNV, whereas both α_vβ₃ and α_vβ₅ are present in diabetic retinopathy tissue. ¹⁵ Anti-α_vβ₃ antibodies and other anti-α_vβ₃ compounds appear to block cytokine- and tumor-mediated angiogenesis in several animal models. ^{16 17 18} These findings suggest that integrins α_vβ₃ and α_vβ₅ might be good targets for ocular angiogenesis inhibition. 

Given the adverse effects of intraocular injections, sustained drug delivery devices offer a promising alternative to systemic injections in that they can deliver predictable therapeutic levels of drug directly to the site of action for an extended period. In addition, continuous delivery of antiangiogenic drugs has been shown to improve the efficacy and potency of therapies in tumor regression and lung metastasis compared with intermittent bolus injections, ^{9 19 20} suggesting that a combination of integrin antagonist and sustained delivery may provide a better treatment modality for ocular angiogenesis. In the present study, with the use of bFGF-induced CAM angiogenesis and laser-induced CNV models, we have demonstrated that sustained release of EMD478761, a nonpeptide diastereomer benzoxazinone molecule with dual antagonism for integrins αvβ3 and αvβ5, appears to have significant antiangiogenic properties. 

Methods

Implant Design

Two types of EMD478761 (EMD) containing implants (A and B) were designed (Fig. 1)according to the following procedure: a 15% (wt/vol) polyvinyl alcohol (PVA) solution was formulated by placement of 3 g PVA (Airvol 125; Air Products and Chemicals, Inc., Allentown, PA) in 20 mL molecular biology-grade water (Eppendorf Scientific, Inc., Westbury, NY) in a closed vial and heated for 3 hours at 100°C. The freshly made PVA solution was poured onto a glass plate, forming a thin film. Compressed pellets, containing either 2 mg EMD478761 (implant A) or 250 μg EMD478761 (implant B), were embedded within this film during its wet phase. After the film was allowed to dry at room temperature, sections of the film, with the drug core centered within the section, were cut with a 2-mm biopsy puncher (implant A) or a razor blade (implant B). Implant A measured 2 mm in diameter, and implant B measured 1 × 1 × 2 mm. Sham implants were made in a similar fashion except that no drug was incorporated. All implants were sterilized (3 × 10⁴ grays, gamma radiation) before use. 

In Vitro Release Rate Determination

In vitro release rates on representative implants of each design were measured daily with HPLC. Five implants from each design were randomly selected and placed in a closed vial with a constant volume of phosphate-buffered saline (PBS; pH 7.4) at 37°C. PBS was replaced every 24 hours. The drug assays were performed with an HPLC system (HP1100; Agilent Technologies, Palo Alto, CA) equipped with a G1329A autosampler, a G1315A diode array detector, a G1312A binary pump, and a computer workstation (Dell, Roud Rock, TX) that controlled the operation of HPLC and analyzed the data. A reverse-phase column (5 μm, 4.6 × 250 mm; Ultrasphere C-18; Beckman Coulter, Inc., Fullerton, CA) was used for separation, and detection was set at 280 nm. The flow rate used was 1 mL/min, with a mobile phase of 20% acetonitrile, 40% water, and 40% methanol by volume. The retention time was 5 minutes, and the detection limit was 10 ng/mL. Release rates were determined by calculating the amount of drug released in a given volume over time and recorded for each implant design in μg/d (±1 SD). The cumulative amount of drug released was calculated based on daily release rates and was recorded separately for each implant design. 

CAM Assay

The CAM assay was performed using 10-day-old embryonated eggs (CBT, Charlestown, MD), as described previously, with slight modification. ²¹ A doughnut glass coverslip was constructed by creation of a 2.5-mm hole in an 8-mm glass coverslip. On embryonal day 3, 3 mL ovalbumin was removed from each egg. After opening windows on embryonal day 10, the angiogenic stimulus (100 ng bFGF), previously dried on the precut doughnut glass coverslips, was applied to the CAM. Implant A was placed centrally in the previously cut circular hole in the coverslip for local delivery of the drug. Eggs were photographed 3 days after the stimulus was added, and the angiogenesis response was examined. The neovascularization area was quantified with modular imaging software (Openlab, version 3.1.5; Improvision Inc., Lexington, MA). 

Choroidal Neovascularization Induction in Rats

All procedures were performed with strict adherence to the guidelines for animal care and experimentation proposed in the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and by the National Eye Institute/National Institutes of Health. Male adult (250 g) Brown Norway rats (Charles River Laboratories, Raleigh, NC) were anesthetized for all procedures with intramuscular ketamine at 70 mg/kg and xylazine at 30 mg/kg, and the fundus was visualized with a slit lamp biomicroscope (Haag-Streit, Mason, OH) using a slide coverslip as a contact lens. A diode red laser (810 nm; Iris Medical Instruments, Inc., Mountain View, CA) was used to induce CNV by rupturing the Bruch membrane. Laser parameters were set to 75-μm spot size, 100-ms exposure, and 150-mW power. A series of four laser lesions was placed at approximately equal distance around the optic discs of both eyes of the animals. Eyes that bled or showed no bubble formation were excluded. 

Intravitreal Injection and Implantation

Intravitreal injections of 10 μL of 0.5 μg/μL EMD478761 or PBS were conducted through a 32-gauge needle (Hamilton Co., Reno, NV) under a Zeiss (Thornwood, NY) operating microscope. Injections were performed on five animals immediately after laser photocoagulation. 

For placement of EMD or sham implants into the vitreous cavity, the conjunctiva was dissected and a small single incision was made with a 20-gauge needle in the superior globe, just posterior and parallel to the ora serrata, allowing the microimplant to be inserted within the vitreous. The posterior segment was evaluated immediately after insertion to confirm proper placement of the microimplant into the vitreous cavity. All eyes were treated with topical antibiotic ointment (Fougera; E. Fourera & Co., Melville, NY). 

CNV Quantification Using FITC-Dextran Perfusion

Rats were killed with 100% CO₂ 2 weeks after laser treatment and were perfused with 5 mL PBS, 5 mL of 4% (wt/vol) paraformaldehyde (PFA), and 5 mL of 10 mg/mL high molecular–weight FITC-dextran (Sigma, St. Louis, MO). Eyes were immediately enucleated and fixed in 4% PFA. Choroidal flatmounts were made, as previously described, by hemisecting the eye and removing retina. ^{22 23} CNV from implant treatment, intravitreal injection, and laser controls was thereafter examined under a fluorescence microscope (BX50; Olympus, Tokyo, Japan). Images of the neovascular lesions were captured with a CCD color video camera (Retiga EX; Retiga, Burnaby, BC, Canada) coupled to a computer (MacIntosh; Apple, Cupertino, CA). Areas of CNV were measured with Image J (version 1.32j; National Institutes of Health, Bethesda, MD). 

Histopathologic Study

Five eyes from each treatment group were selected 2 weeks after laser treatment for histologic studies. After the rats were humanely killed, their eyes were immediately collected, placed in plastic molds filled with embedding medium (Tissue-Tek OCT Compound; Sakura Finetek, Torrance, CA), and flash frozen on dry ice. Frozen eyes were sectioned transversely on a cryostat at −20°C, and 10-μm thick sections were stained with hematoxylin and eosin. 

Statistical Analysis

All data were presented as mean ± SD and were statistically evaluated by unpaired Student’s t-test. 

Results

In Vitro Release Rates of the Implants

Implant A delivered an initial burst of drug on the first day, followed by constant release for the next 10 days, which then gradually decreased over time (Fig. 2A) . The steady state cumulative release amount between days 2 and 11, calculated from release rates, followed a zero-order release kinetics (Fig. 2A) , typical for a diffusion-controlled reservoir implant. ²⁴ The steady state release rate was 119 ± 30 μg/d. Implant B (microimplants) had a release profile similar to that of implant A throughout the assay period (Fig. 2B) . The mean release rate estimated from steady state was 18.5 ± 3.8 μg/d. These results indicated that the implants could release drug for more than 17 days. 

Effect of EMD478761 Implant on Angiogenesis Inhibition in CAM

After placing a bFGF-containing coverslip on the CAM, implant A was placed at the center of the doughnut glass coverslip. Results showed that EMD implant A (Fig. 3C)dramatically inhibited bFGF-induced neovascularization compared with sham implant (Fig. 3B)and controls (Fig. 3A) . Of bFGF CAMs, 83% had positive neovascular response (Fig. 3D) , inducing an average of 6.8 mm² of neovascularization area (Fig. 3E) . The vehicle (sham)-treated CAMs did not show an effect on inhibiting new vessel formation and followed a similar response in which the neovascularization area remained unchanged. On the other hand, only 19% of the CAMs were weakly angiogenic in the presence of EMD implant A, and the angiogenesis area was dramatically reduced to an average of 0.47 mm², which is significantly different from those of bFGF-treated CAMs (P = 0.0008) and sham-treated CAMs (P = 0.0001). In addition, there were no signs of local drug toxicity and no effect on the normal CAM vasculature (Figs. 3B 3C) . 

Effect of Intravitreal Injection and EMD Microimplant on Laser-Induced CNV

To examine whether bolus administration of EMD478761 can lead to significant suppression of the development of laser-induced CNV, a single intravitreal injection of EMD478761 at the dose of 5 μg was performed on five animals immediately after laser injury. Fourteen days after treatment, the area of CNV in eyes treated with EMD478761 was reduced by 12% compared with PBS control (Figs. 4B 4C 4F) . However, the differences between them were not statistically significant (P = 0.68). 

An EMD microimplant or a sham was then placed into the vitreous cavity of five previously laser-treated animals. Fourteen days after implantation, animals were killed and the area of CNV was measured. EMD microimplants inhibited CNV relative to the sham implant and laser controls in a statistically significant fashion (P < 0.05). In the eyes that received EMD microimplant, the mean CNV area of the lesions decreased by 65% and 63%, respectively, compared with those of laser controls and sham implant-treated eyes (Fig. 4) . In addition, there was no significant difference between sham implant and laser controls (Fig. 4F) . 

Histopathologic Study

Two weeks after CNV induction, histopathology of eyes treated with laser or PBS revealed numerous new blood vessels arising from the disrupted RPE and Bruch membrane (Figs. 5A 5B) . Eyes treated with intravitreal bolus injection of EMD478761 demonstrated responses similar to those of eyes treated with PBS (Fig. 5C) . In contrast, eyes with the EMD microimplant treatment had thin proliferative membranes and few new blood vessels under the retina (Fig. 5E) . Sham implant-treated eyes had responses similar to those of laser controls (Fig. 5A) , indicating that there was no inhibitory effect of the sham implant on CNV. These results are consistent with those of CNV areas calculated from choroidal flatmounts. Examination of the slides showed an absence of inflammatory cells in EMD implants and sham implants. 

Discussion

Although numerous antiangiogenic compounds, such as triamcinolone acetonide, TNP-470, interfering RNA, and certain monoclonal antibodies, have been investigated to target choroidal neovascularization, ^{17 20 25 26 27 28 29} results are far from satisfactory. In this study, we have demonstrated the significant inhibitory effect of EMD478761 released from polymeric implants on bFGF-induced angiogenesis and laser-induced experimental CNV. 

bFGF stimulates the expression of integrin α_vβ₃ on the neovasculature in CAM. ^{16 30} Therefore, we first examined the effect of the sustained-release EMD478761 implant on bFGF-induced angiogenesis in CAM. In contrast to previous CAM assays in which the testing compounds were often added exogenously to the angiogenic stimuli, we used a sustained release implant to deliver the drug continuously and locally. Our results showed that sustained delivery of EMD478761 significantly suppressed bFGF-induced angiogenesis without affecting normal CAM vessels. These findings suggest that EMD478761 is likely to interact with the integrin α_vβ₃ present on the sprouting CAM vessels. ^{16 18} Potent inhibition of bFGF-induced angiogenesis by this integrin antagonist demonstrates its antiangiogenic activity and the potential applications to CNV. 

Angiogenesis is partially regulated by integrins. ^{15 18} The α_vβ₃ and α₅β₁ integrins appear to be overexpressed in laser-induced CNV in rats ³¹ and mice ³² but not on normal choroidal vessels. The high expression and upregulation of these integrins suggest that they may play an important role in CNV. Indeed several antagonists to these integrins have been developed, and they potently inhibit cytokine- and tumor-mediated angiogenesis in different animal models. ^{17 32 33 34 35 36 37} Intravitreal injection of cyclic RGD, an α_v-integrin antagonist, effectively inhibited the progression of laser-induced CNV in rats. ²⁷ Systemic administration of JSM6427, an integrin antagonist to α₅β_1, significantly suppressed the development of CNV in mice by 33% to 40%. ³² In this work, we observed that intravitreal bolus injection of EMD478761 failed to attenuate the development of CNV. The low efficacy of this delivery route was not surprising because the dosage used in this study, which was limited by its aqueous solubility (∼500 μg/mL), was 20× lower than that used for cyclic RGD peptides. ²⁷ Other possible reasons could have been the quick turnover of the integrin α_vβ₃ receptor ³⁸ and the rapid elimination of the drug from the vitreous. ^{9 10} Limited drug solubility, small vitreous volume (estimated at approximately 56 μL), ³⁹ and lower levels of drug extraction efficiency from ocular tissues (less than 65%; Kim H, unpublished data, 2007) preclude direct, in vivo pharmacokinetic measurement of EMD478761 in the rat eye. However, all these possibilities could be obviated by the use of a sustained delivery device. In addition, to our best knowledge, sustained delivery of an integrin antagonist with an implant has not been previously studied. For these reasons, we designed a sustained-release delivery system that would provide a therapeutic dose for this compound for an extended period. 

Sustained-release microimplants have been beneficial in ocular delivery of other antiangiogenic compounds because not only can they bypass the main ocular delivery barriers by direct insertion into the vitreous cavity, but they also prolong the half-life of the drug. For example, it has been shown that a sustained-release 2-methoxyestradiol intravitreal implant was safe in normal rabbit and that it suppressed CNV in rats. ⁴⁰ In another study, Ciulla et al. ⁴¹ demonstrated the potent inhibition of laser-induced CNV in rats by sustained release of triamcinolone acetonide from matrix-type microimplants. The present study demonstrates that EMD478761 can be delivered intraocularly using a sustained-release reservoir microimplant. Animals treated with the EMD478761 microimplant exhibited a 63% reduction in CNV area 2 weeks after implantation compared with sham-treated animals. Furthermore, new vessels and thicker neovascularization membranes were observed in untreated and sham-treated eyes compared with EMD478761 implant-treated eyes. The increased potency observed with the implant may be attributed partially to the continuous presence of EMD478761 around the neovascularized tissues. Although previous studies revealed that polyvinyl alcohol tended to accumulate at the CNV lesion site, ⁴² the fact that there was no significant difference between sham implant-treated eyes and laser controls indicated that CNV inhibition was not an artifact of the PVA polymer but the result of EMD478761 released from the microimplant. Histologic studies performed on those eyes treated in the same manner confirmed this finding. 

The mechanism by which EMD478761 exerts its antiangiogenic effect is likely to be similar to that by which other integrin antagonists act. A predominant characteristic of CNV is the excessive growth of new blood vessels from the preexisting vasculature. Endothelial cells play a pivotal role as they form a tube in every blood vessel and are actively involved in vascular remodeling during CNV. Therefore, EMD478761 could exert its inhibitory effect through antiadhesion of endothelial cells to the extracellular matrix, ²⁷ induction of apoptosis of endothelial cells, ⁴³ or changes in the regulation of integrin expression. Additional studies of the drug’s action mode are under way to further assess the antiangiogenic efficacy of EMD478761 through this sustained delivery system. 

In conclusion, sustained-release EMD478761 implants inhibited bFGF-induced angiogenesis in CAM assay and suppressed laser-induced CNV in rats. These results provide new evidence that sustained release of EMD478761 may be useful in the treatment of neovascular ocular diseases. 

 Submitted for publication April 19, 2007; revised July 11, 2007; accepted September 25, 2007.

 Disclosure: Y. Fu, None; M.L . Ponce, None; M. Thill, None; P. Yuan, None; N . S . Wang, None; K . G. Csaky, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Implant designs. Representative images of implants A and B made from 15% PVA. Sham implants contain no drug. The final dimension of implant A was 2 mm in diameter, whereas implant B measured 1 × 1 × 2 mm.

Figure 2.

 

(A) In vitro mean release rate and cumulative release amount of implant A. (B) In vitro mean release rate for implant B.

Figure 3.

 

Effect of EMD implant on angiogenesis inhibition in the chick CAM assay. (A) bFGF-CAM. (B) bFGF/sham-CAM. (C) bFGF/EMD implant A-CAM. (D) Percentage of positive CAMs in the presence of bFGF only, bFGF/sham, and bFGF/EMD implant A. (E) Neovascularization areas from quantitative image analysis after treatment with bFGF, bFGF/sham, and bFGF/EMD implant A. **P = 0.0008 compared with bFGF-treated CAMs. Original magnifications, 1.28× (A–C); 2.56× (insets).

Figure 4.

 

Laser-induced CNV inhibition by EMD478761 implant in a rat model. (A–E) Representative images of CNV in choroidal flatmounts with laser control, intravitreal injection of PBS, intravitreal injection of EMD478761, sham implant, and EMD478761 microimplant treatment, respectively. The vasculature was labeled with FITC-dextran 2 weeks after laser photocoagulation. (F) Quantitative CNV areas measured from lesions 2 weeks after laser photocoagulation. *P < 0.05 compared with laser control. Scale bar, 100 μm.

Figure 5.

 

Hematoxylin-eosin–stained light microscopy images of CNV 2 weeks after laser photocoagulation. Each image shows the center of CNV lesions. (A–E) Histologic sections of the eye with laser control, intravitreal PBS injection, intravitreal EMD 478761 injection, sham implant, and EMD478761 microimplant treatment, respectively. Scale bar, 100 μm.

The authors thank Mathias Wiesner (Merck KGaA, Darmstadt, Germany) for EMD478761 compound synthesis.