Retinal vein occlusions (RVO) are common retinal vascular disorders with potentially blinding complications. Loss of vision may result from persistent macular oedema, macular ischaemia or neovascular complications including neovascular glaucoma. Vascular endothelial growth factor (VEGF) is a stimulus for persistent capillary leakage and neovascularisation.1 2 Increased VEGF concentrations have been found in intraocular fluid samples from patients with branch retinal vein occlusion (BRVO)1 and central retinal vein occlusion (CRVO), with the highest level seen in those with neovascularisation.2 In a primate model, blockage of VEGF prevented iris neovascularisation in a laser-induced model of CRVO.3 Since retinal vein occlusions are associated with increased levels of VEGF in the vitreous,1 2 and VEGF has been shown to play a major role in the development of macular oedema,1 the inhibition of VEGF could be of therapeutic value.

Bevacizumab (Avastin) is a recombinant humanised monoclonal antibody that binds all isoforms of VEGF and inhibits its interaction with receptors found on endothelial cells. Recently, several reports have described the off-label use of intravitreally injected bevacizumab for age-related macular degeneration, CRVO and BRVO, showing promising functional and anatomical results.4^–13 Short-term follow-up in a small cohort of patients with CRVO showed visual acuity improvements with reduction in the macular oedema following bevacizumab injections.6 In another recent study on early initiation of intravitreal bevacizumab treatment for CRVO, the patients experienced a dramatic improvement in the visual acuity and clinical fundus appearance, without collateral vessel formation.10 These findings are difficult to explain with current theories of the pathophysiological features of CRVO. The aim of our study was to investigate the penetration and time-related distribution of bevacizumab in the posterior segment, particularly in the macula, the optic nerve and the retinal veins after intravitreal injection into the primate eye. We speculated that the penetration of bevacizumab was the same in healthy eyes as in affected eyes.

MATERIALS AND METHODS

Animals

Four Cynomolgus monkeys (age 8–9 years, supplied by Nafovanny, Vietnam) for the experiments and one Cynomolgus monkey as a control were raised under standard conditions with veterinarian attendance in the Covance Laboratories (Münster, Germany). All experiments were performed according to the ARVO guidelines for the use of animals in research.

Intravitreal injection of bevacizumab

Bevacizumab (Avastin, Roche, 25 mg/ml) for the intravenous use was aliquoted by our pharmacy into tuberculin syringes under sterile conditions. The aliquots of 50 μl contained 1.25 mg bevacizumab and were stored continuously at 4–8°C until use. The animals were anaesthetised with ketamine (10 mg/kg) plus xylazine (2 mg/kg). Before injection, the eyes were examined for any signs of inflammation. Pupils were dilated (tropicamide, phenylephrin), and the corneas were anaesthetised (oxybuprocainhydrochloride). The conjunctival and corneal surfaces were disinfected (povidone iodine 10%). After sterile coating and insertion of a lid speculum, 1.25 mg of bevacizumab was injected into the vitreous cavity using a 27-gauge cannula. When removing the cannula, the injection site was compressed using forceps to prevent reflux. A topical antibiotic (gentamycin) was administered from this time point four times a day for 4 days. Animals were monitored for signs of inflammation until sacrificed. There was no intravitreal injection of bevacizumab in any eye of the control in order to prevent bias related to systemic uptake of bevacizumab.

Immunohistochemistry

Eyes were enucleated on days 1, 4, 7 and 14 after intravitreal injection, fixed in formalin, embedded in paraffin wax, sectioned and deparaffinised. Donkey anti-human Cy3-labelled IgG (709-166-149, dilution 1:500; Jackson ImmunoResearch, West Grove, PA) was used for bevacizumab detection. This polyclonal antibody binds to epitopes of both Fc and Fab portions of human IgG. Untreated control eyes were stained in the same way. The sections were embedded (FluorSave; Calbiochem, La Jolla, CA) and examined using a fluorescence microscope (Zeiss Axioplan2 imaging). All photographs (Hamamatsu ORCA-ER digital camera) were taken with the same camera settings (brightness, contrast, etc) to allow a comparison of staining intensity between different time points and the control.

RESULTS

Branches of the central retinal artery and vein

There was no significant difference in the staining of arteries versus veins. An abundant immunoreactivity for bevacizumab was found in the walls of the blood vessels supplying the inner portion of the retina from days 1 to 4 after injection; the staining was also present but was weaker in the lumen of some of these blood vessels (figs 1B, 2B). From day 7, only the lumen of the vessels in the nerve fibre layer showed a strong immunoreactivity for bevacizumab (figs 3C, 4C). No immunoreactivity was detected in the control (fig 5B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Immunohistochemical localisation of bevacizumab at the posterior pole 1 day after intravitreal injection in the primate eye. (A) Overview of the posterior pole by using transmitted light. The tissue separation in the lamina cribrosa is due to a fixation artefact. Three selected areas are shown enlarged (b–d). (B) Fluorescence image showing branches of the central retinal artery and vein in the nerve fibre layer of the retina. Positive immunoreactivity for bevacizumab is found intensively on blood vessel walls and weakly in the lumen. (C) Fluorescence image of the optic nerve showing homogeneous staining of the axon bundles. (D) Fluorescence micrograph of the region of the macula. The staining is concentrated at the nerve fibre layer (NFL) and the ganglion cell layer (GCL). It is weaker in the inner plexiform layer (IPL) and interestingly it is also strongly present in the outer nuclear layer (ONL) of the fovea centralis. The inner limiting membrane (ILM) (arrow), Bruch membrane (arrowhead), endothelium of the choriocapillaris and deeper choroidal vessels (asterisk) are also labelled. The retinal pigment epithelium (RPE) is non-specifically labelled due to lipofuscin autofluorescence. Magnification: 2.5× for (A); 20× for (B); 10× for (C) and 20× for (D).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2 Immunohistochemical localisation of bevacizumab at the posterior pole 4 days after intravitreal injection in the primate eye. (A) Overview of the posterior pole by using transmitted light. Two selected areas are shown enlarged (b, c). (B) Fluorescence micrograph showing a strongly positive staining for bevacizumab in the walls and in the lumen of blood vessels in the nerve fibre layer. (C) Sheaths of axon bundles in the optic nerve showing immunoreactivity as well. Magnification: 5× for (A); 40× for (B) and 20× for (C).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3 Immunohistochemical localisation of bevacizumab at the posterior pole 7 days after intravitreal injection in the primate eye. (A) Overview of the posterior pole by using transmitted light. Two selected areas are shown enlarged (b, c). (B) Fluorescence micrograph showing the absence of immunoreactivity in the optic nerve. An intense staining is present in the sclera. (C) Lumen of retinal vessels showing a staining for bevacizumab. Magnification: 2.5× for (A); 10× for (B); 20× for (C).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4 Immunohistochemical localisation of bevacizumab at the posterior pole 14 days after intravitreal injection in the primate eye. (A) Overview of the posterior pole by using transmitted light. Two selected areas are shown enlarged (b, c). (B) No specific immunoreactivity found in the optic nerve. (C) Lumen of a vessel within the nerve fibre layer showing an immunoreactivity for bevacizumab. Magnification: 2.5× for (A); 20× for (B) and 20× for (C).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5 Immunohistochemical localisation of bevacizumab at the posterior pole in the control primate eyes. (A) Overview of the posterior pole using transmitted light. One selected area is shown enlarged (b). (B) No immunoreactivity for bevacizumab found in the branches of the central retinal artery and vein and in the optic nerve of the control. Magnification: 2.5× for (A); 10× for (B).

Optic nerve

The most intense immunohistochemical staining for bevacizumab was present in the optic nerve on day 1 after intravitreal injection (fig 1C). The staining was homogeneous and located within the axon bundles (fig 1C). On day 4 after intravitreal injection, the immunoreactivity declined and was just found in the sheaths of the axon bundles (fig 2C). On days 7 and 14 after injection, no positive staining for bevacizumab was found in the optic nerve (figs 3B, 4B). The negative control showed no positive immunoreactivity (fig 5B).

Macula

This anatomical region was only found in the sections of the eye 1 day after the intravitreal injection of bevacizumab (fig 1A). Immunoreactivity for bevacizumab was detected in the fovea centralis. It was intense in the nerve fibre and in the ganglion cell layers, was also strongly prominent in the outer nuclear layer and was weaker in the inner plexiform layer. The inner limiting membrane, the Bruch membrane, the endothelium of the choriocapillaris and deeper choroidal vessels were also stained (fig 1D).

DISCUSSION

Currently, no treatments have proven to be consistently effective for patients with macular oedema resulting from RVO. The Central Vein Occlusion Study demonstrated no statistically significant visual acuity benefit from grid laser treatment.14 More recently, intravitreal triamcinolone acetonide has demonstrated beneficial effects in decreasing macular oedema and improving visual acuity in several small case series.15 16 However, the effects of triamcinolone are inconsistent, transient and associated with complications. Recently, clinical case reports with small patient numbers or retrospective studies have been published on bevacizumab treatment of RVO, and they described a significant reduction in central retinal thickness and improved visual acuity after bevacizumab injection.10^–13 In a previous work, we showed that bevacizumab can penetrate the retina and was also transported into the photoreceptors, the retinal pigment epithelium, the choroid and the blood,17 but so far no experimental work has been published on the effects or time-related distribution of bevacizumab on the vascularised tissues of the posterior pole, such as the fovea, the optic nerve and the branches of the central retinal artery and vein. Our study demonstrates that intravitreally injected bevacizumab penetrates well and quickly into the macula, the optic nerve and the retinal veins. From days 1 to 4 after injection, the staining for bevacizumab was intense in the walls of the blood vessels supplying the inner portion of the retina (figs 1B, 2B), which is the site of secretion and binding of VEGF to its receptor on the retinal endothelial cells.18 In addition, these blood vessels showed an immunoreactivity for bevacizumab in their lumen (figs 3C, 4C), and this intravascular localisation increased from day 1 to 14 after injection. Assuming a regular circulation, a rapid distribution and subsequent elimination of intravascular bevacizumab from the lumen would be expected. The selective accumulation of the antibody in certain vessels may be a sign of a disturbed circulation in some of the blood vessels. If the perfusion is equal in all vessels, we would expect homogeneous distribution of bevacizumab. A higher concentration of bevacizumab may be indicative of slow perfusion. Circulation disturbances after intravitreal bevacizumab have been previously demonstrated for choriocapillaris vessels,19 and thromboembolic side effects were described after systemic use in humans.20 Although it cannot be excluded that the humanised antibody bevacizumab interferes with the monkey’s immunology in the present study, the fact that the number of involved vessels increased with the number of days after injection would rather argue against such assumption.

The very fast penetration of bevacizumab into the macula, the optic nerve and the retinal veins is consistent with the clinically described benefits seen 1 day after the intravitreal injection of bevacizumab in patients suffering from RVO.11 This finding is also consistent with the observation that in patients with CRVO, the abnormalities of retinal veins, optic nerves and maculae seemed to respond in a rapid fashion to intravitreal bevacizumab.10 Our study shows that the intravitreal application mode results in an accumulation of bevacizumab throughout the vascularised tissues of the posterior pole. One day after injection, the vessel walls and the photoreceptors localised in the macula were the preferential and specific site of bevacizumab accumulation. This finding consolidates the use of intravitreally administered bevacizumab for the treatment of macular oedema secondary to RVO. However, given the persistent intraluminal location of bevacizumab in certain vessels, perfusion disturbances cannot be excluded.