extrapolating the results of this study to humans.2 Compared with humans, the rabbit has a lower mean scleral thickness, higher choroidal flow rates, smaller vitreous cavity, and a poorer vascularized retina. The rabbit eye has lower volume and surface area for drug clearance: the volume of the human eye is 3.9 ml compared with the rabbit’s eye of 1.5 ml. Furthermore, albino rabbits are preferred over pigmented animals because ocular pigmentation may protect against the toxic effects of the drug. On an ultrastructural level retinal layers in normal rabbits present reasonable degree of vacuolization which may impair analysis of retina toxicity to drugs. Nevertheless, rabbits are the most commonly applied animal model to study retinal toxicity of drugs for intravitreal injection.12

The cat has served as an important research model for neurophysiology and retinal degenerative disease processes. In addition, felines may also be useful for the study of retinal toxic reactions after intravitreal injections of drugs such as human recombinant tissue-type plasmi­ nogen activator in the management of subretinal hemorrhage.13 Anatomically, some diameters of intraocular structures in cats may be similar to those of humans. Their vitreous volume is quite similar to that of human eyes, whereas the transverse and anteroposterior diameters of the globe are around 22.30 mm. The transverse diameter of the cornea averages 16 mm. The volume of the anterior chamber varies from 0.8 to 1.0 ml, and the depth of the anterior chamber is very large, around 4.5 mm. Some recognized characteristics of the cat eye are a considerably larger lens than humans. Also, the vascular supply in the cat eye is different from that of human eyes: there is no central retinal artery and cats have a vascularized inner retina. The single medial and lateral long posterior ciliary arteries pass around the globe within the sclera to form a large vascularized plexus in the pars plana region with the anterior ciliary arteries. The ora ciliaris retina, corresponding to ora serrata in humans, is approximately 6 mm behind the limbus in the cat eye. The large vessels of the plexus can be seen through the sclera, and in attempts to avoid these vessels, the sclerotomies of pars plana vitrectomy should be done as far posterior as possible.14

The pig eye may also be a good model for toxicology study because the size and anatomy are similar to those of the human eye.15 There has been interest in evaluating the suitability of the pig as a model of ocular diseases, since it is phylogenetically close to the human and is much more available, cheaper, and less ethically barren than the monkey. The porcine retina is even more similar to the human retina than that of other large mammals such as the dog or cow. Some examples of similarities in anatomical and histological characteristics of pigs in comparison to humans include nontapetal fundus with a holangiotic vascular pattern and retinal layers of similar thickness. The pig retina has four main arteriovenous branches that arborize to the retinal periphery. Unlike human eyes, these vessels do not typically form a single central artery and vein before reaching the lamina cribrosa, making the pig eye a poor disease model for central retinal vein occlusion. With the conceptions of the lack of a fovea, the porcine retina shares some significant similarities within the photoreceptor mosaic of humans and other primates. Otherwise, the porcine optic nerve structure is similar to that of a human, containing a prominent lamina cribrosa, as well as the retinal blood vessels and vascular supply. Additionally, tools employed for diagnostics in ophthalmology, such as optical coherence tomography, corneal topography imaging, or multifocal electroretinography can easily be applied to the pig eye.

The macula can only be found in primates and birds, and the animal model for macular degeneration and macular toxicity is therefore mostly limited to monkeys. The size and macroscopic eye measurements of primate eyes depend on the species of primate; some species like common marmoset monkeys (Callithrix jacchus) exhibit very small ocular dimensions which may impair surgical techniques for investigation in local drug therapy. In the retina ultrastructure some differences among primates’ eyes include density and distribution of mitochondria, as well as the complexity of on/off cones. Overall, cynomolgus monkeys (Macaca fascicularis) and rhesus macaque (Macaca mulatta) have been competently used to analyze the safety and efficacy of intravitreal injection of drugs and chemical compounds for various decades.2

MAJOR CLASSES OF DRUGS AND THEIR SAFETY PROFILE AFTER LOCAL OCULAR APPLICATION FOR RETINA THERAPY

CORTicosteroids

Corticosteroids are powerful drugs widely used in ophthalmology, including in the therapy of retinal diseases. In the past two decades much research has been performed to evaluate the indications of various types of steroids for vitreoretinal diseases. However, beyond efficacy, one of the most important issues for the application of corticosteroids is finding the limitation of their side-effects.

Fluocinolone acetonide is a synthetic corticosteroid with low solubility in aqueous solution. Experiments in rabbits provided the preclinical toxicity profile of fluocinolone to the retina. Clinical examination, electroretinography, or histologic examination determined the safety of both 2 and 15 mg synthetic fluocinolone acetonide in one preclinical study. Neither electroretinographic alterations nor histology alterations were observed in this in vivo investigation. However, further clinical experience with implants containing 2 or 6 mg fluocinolone acetonide for therapy of uveitis revealed that a considerable amount of patients do experience complications, which include glaucoma, cataract, and retinal vein occlusion.16

A series of contradicting studies have recently been published regarding retinal toxicity after intravitreal triamcinolone acetonide injection. On the one hand, in various experiments intravitreal injection of 4, 16, 20, or 30 mg triamcinolone acetonide promoted normal histological and electroretinographic retinal findings after 7 months. In contrast, in one investigation the authors injected escalating doses from 0.5 to 20 mg of suspended preservative-free triamcinolone acetonide in rabbits, and found at doses 4 mg or higher prominent retinal damage manifested by destruction of photoreceptor outer segments and retinal pigment epithelium/photoreceptor interdigitation.17 Only a few studies have thus far addressed the subretinal toxicity of triamcinolone acetonide: in one investigation the researchers examined subretinal injection of 3 mg/ml triamcinolone acetonide in primate eyes and described neither ultrastructural nor cellular retinal damage, but Maia et al. in a morphologic study disclosed disturbance to photoreceptor segments after subretinal injection of preservative-free triamcinolone acetonide, although no clinical abnormality on fundoscopy or angiography examination was observed.18 These opposite findings suggest that other factors may contribute to intravitreal retinal toxicity of triamcinolone acetonide (Figure 15.1).

Much controversy exists in animal studies whether triamcinolone acetonide itself or the vehicle plays the most significant role in retinal damage. Some colleagues demonstrated severe damage to photoreceptor only in a group with triamcinolone acetonide with vehicle compared with preservative-free triamcinolone acetonide. In contrast to these results, other authors revealed a safer profile of the vehicle only; they reported normal retinal structure after intravitreal injection of vehicle only in rabbits.17,18 It is also unclear which chemical element from the vehicle solution of triamcinolone acetonide may induce retinal injury, although research so far has focused only on the organic colorless preservative benzyl alcohol. Other than benzyl alcohol, the vehicle of commercially available triamcinolone acetonide contains a second preservative named polysorbate at 0.4 mg/ml. Future studies should elucidate the safety of triamcinolone acetonide vehicle and whether benzyl alcohol, polysorbate, sodium carboxymethycellulose, or other factors such as pH or osmolarity pose the higher risk to the retina.

Clinical experience in recent years unraveled the risks of patients injected with intravitreal or periocular triamcinolone acetonide. The two most frequent complications are cataract and glaucoma, which in most patients can be managed by surgery and topical eye drops. Much less common complications are endophthalmitis and pseudoendophthalmitis, encountered in around 0.5% of patients.

Dexamethasone is a synthetic glucocorticoid class of steroid hormones with potent anti-inflammatory and immunosuppressant activities. In a classic work performed by Kwak and D’Amico, the authors

delivery drug retinal for routes and models Animal • 2 section

Drugs of Application Ocular to Toxicity• 15Ocularchapterand Retina

A E I

B F J

C G K

D H L

Figure 15.1  After subretinal injection of control, preservative-free triamcinolone, and triamcinolone with preservative, different types of retinal cells manifest various signs of damage. The damage induced by the preservative is more severe in neuroretinal posterior cells than that caused by the drug itself.

delivery drug retinal for routes and models Animal • 2 section

Drugs of Application Ocular to Toxicity• 15Ocularchapterand Retina

in vivo animal study with monkeys showed no toxicity at the commonly used concentration of 2.25 mg ceftazidine, but others showed that ceftazidime may not be toxic in vitrectomized rabbit eyes.21 However, it appears that the cephalosporin may cause some degree of toxicity at otherwise nontoxic concentrations in a silicone-filled eye. Based on those preliminary data, surgeons have applied ceftazidime intravitreally in doses up to 2.25 mg for the therapy of endophthalmitis.

Vancomycin remains the antibiotic of choice targeting highly pathogenic Gram-positive microorganisms, usually applied in patients intravitreally in the dose of 1 mg in 0.1 ml. In vivo animal studies have shown no toxic effect of this antibiotic when used in an infusion solution that was given intraocularly after or during vitrectomy in rabbit. However, in a silicone-filled eye, nontoxic concentrations of vancomycin may cause toxicity, so vitreous status should be evaluated when vancomycin is the antibiotic of choice as well. In the clinical practice, vancomycin has been associated with postoperative cystoid macular edema when infused as intracameral injection for prophylaxis during cataract surgery.

The ocular toxicity of another class of antibiotics, the quinolones, has been investigated in the recent past. Intravitreal injection of ciprofloxacin has not been associated with toxicity at therapeutic levels of 100– 500 g in rabbits, and significant retina damage has been present only at 2 mg.22 The fourth-generation quinolones, which include moxifloxa-

cin and gatifloxacin, have had greater attention for clinical use. An in vitro study showed that, at concentrations higher than 160 g/ml,

moxifloxacin induced adverse effects on primary retinal pigment epithelium and neuronal retinal cell proliferation and viability. Further

studies in vivo showed that intravitreal injection of moxifloxacin did not cause retinal toxicity up to 100 g/ml in mice or 150 g in rabbits.23 In

vivo, intravitreal­ injection of the other quinolone, gatifloxacin, at doses varying from 50 to 400 g caused no retinal toxicity assessed clinically and microscopically in rabbits. Clinical experience on retina toxicity studies revealed that the current recommended dose for intracameral

injection of ciprofloxacin is less than 25 g. In humans intravitreal injections of ciprofloxacin 100 g, ofloxacin 50 g/ml, trovafloxacin 25 g or less, moxifloxacin 160 g/0.1 ml or less, and pefloxacin 200 g/0.1 ml

are considered nontoxic to the retina and intraocular structures.24 Fungal infections are difficult-to-treat causes of endophthalmitis.

Amphotericin B has traditionally been used for treatment, either systemically or intravitreally, usually injected at doses varying from 1 to 50 g. In one study with application of escalating doses from 10, 20, 30, and 50 g, the three higher doses of amphotericin B appeared to be associated with stronger degrees of retina toxicity.25 Based on the animal experiments, intravitreal amphotericin B in doses of 5 or 10 g remains an appropriate therapeutic option for patients with severe fungal injection, for instance secondary to Aspergillus.

An alternative to amphotericin is the use of intravitreal fluconazole. An in vitro study showed no toxicity at a 20 g/ml exposure to fluconazole. Further animal data revealed no retinal toxicity resulting from vitrectomy with a 2 mg/ml fluconazole infusion in an experimental model of candidal endophthalmitis. For a single intravitreal injection fluconazole at a concentration of 100 g and above caused harmful retinal changes with disorganization of the photoreceptor outer segments. Clinical experience revealed that intravitreal injection of 10 g/0.1 ml fluconazole may be the safe dosage for intraocular fungi injection.25

For the therapy of cytomegalovirus infections ganciclovir is a commonly used antiviral medication. Ganciclovir dosages of up to 200 g/0.1 ml appear to be safe for serial intravitreal injections in rabbit eyes following vitrectomy and silicone oil insertion. In unvitrectomized eyes ganciclovir in doses above 300 g induced severe morphologic retinal damage, although at lower doses of 200 g ganciclovir promoted only small functional damage characterized by changes on electroretinography b-wave in rabbits. Similar to other types of antibiotics, ganciclovir also induced sporadic cases of macular infarction in patients. A case report of inadvertent intravitreal injection of a high dose of ganciclovir (40 mg/0.1 ml) for cytomegalovirus retinitis in a patient with AIDS led to permanent retinal damage and visual loss.26 Currently ganciclovir has been injected with consecutive intravitreal injections in

doses varying from 400 g to 4 mg for therapy of intraocular viral infections.

MONOCLONAL ANTIBODIEs

AND FRAGMENTs

The two anti-vascular endothelial growth factor (VEGF) monoclonal antibodies (mAbs), bevacizumab (Avastin, Genentech), and ranibizumab (Lucentis, Genentech), were shown to promote clinical control of ocular neovascularization in the last few years. Although endovenous administration of anti-VEGF mAbs has demonstrated acceptable toxicity profile, their intraocular injection has decreased the risk of systemic complications. In addition, intravitreal injection of anti-VEGF agents may increase the amount of drug available to intraocular tissues such as the retina.

In vitro cellular assays exposed to various concentrations of bevacizumab (0.08 g/ml to 1 mg/ml) have shown little toxic effects to neuronal cells like the ganglion cells, neuroretinal cells, as well as retinal pigment epithelial cells. To study the effects of bevacizumab on various types of retinal cells, one investigation exposed cultured adult porcine neurosensory retinas joined to the retinal pigment epithelial/choroid layer to three doses of bevacizumab (0.25, 0.5, and 1.25 mg/ml) for 3 days. Their results showed no toxic effects on ganglion or photoreceptor cells observed at any concentration of bevacizumab. However, they observed significantly enhanced smooth-muscle actin expression in retina blood vessels in the presence of bevacizumab, which may imply a loss of smooth-cell modulation in normal retinal vessels by VEGF.27 In contrast to those data, others found no toxicity to microvascular retinal cells in vitro after their exposure to 0.125, 0.25, 0.50, and 1 mg/ ml of bevacizumab for up to 24 hours.

A large body of animal studies has been released about the biocompatibility and safety of bevacizumab and ranibizumab for ophthalmology. Consecutive experimental investigation in rats, rabbits, and primates revealed that intravitreal bevacizumab at various concentrations up to 3 mg/ml demonstrated no functional or morphologic toxicity to the retina.28 However, a few recent experimental publications demonstrated some signs of retinal damage after intravitreal bevacizumab. In primates, intravitreal bevacizumab application induced choriocapillaris abnormalities manifested by reduced choriocapillaris endothelial cell fenestrations by densely packed thrombocytes and leukocytes within the vascular lumen. Moreover, Manzano et al. reported signs of ocular inflammation after intravitreal injection of high-dose bevacizumab at 5 mg in rabbits’ eyes.28 Also in rabbits, intravitreal 1.25 mg or 3 mg bevacizumab has caused both mitochondrial changes in the inner segments of photoreceptors and intensive apoptotic protein expression of bax and caspase on immunohistochemistry in comparison to control, although on light microscopy and electroretinographic examination no signs of toxicity were detected. The clinical relevance of such apoptotic retinal findings is yet to be clarified. The preclinical safety of ranibizumab has been evaluated in primate eyes, 0.5 mg injection of the monoclonal antibody fragment promoted reduced leakage from choroidal neovascularization whereas no signs of retinal toxicity have been encountered.

Intraocular bevacizumab and ranibizumab injection has promoted few clinically relevant ocular side-effects to date. In contrast to the crystalline steroid drug triamcinolone acetonide, intravitreal bevacizumab has not been shown to induce glaucoma or cataract progression. Further clinical experience with intravitreal bevacizumab revealed few sporadic cases of uveitis, vitreous hemorrhage, retinal pigment epithelium tears, or endophthalmitis.29

Clinically Rosenfeld et al. suggested the maximum tolerated dose of ranibizumab is 0.5 mg, as higher doses above 1 mg promoted clinically relevant intraocular inflammation.30 Additional clinical investigation disclosed that intravitreal ranibizumab has induced few severe complications such as endophthalmitis, uveitis, vitreitis in the fellow eye, while minor reported ocular events were conjunctival bleeding, eye pain, and floaters. In 2008 a head-to-head comparison of ranibizumab versus bevacizumab to treat advanced age-related macular degeneration funded by the National Eye Institute started, and this should clarify