Fig. 5.2  Structures of drugs delivered to back of eye

inflammatory conditions may enhance drug penetration to the posterior segment via the uvea-scleral route. Thus, topically applied drug which has penetrated into the aqueous humor can transit to the posterior segment via the uvea-scleral route, and potentially access choroid and retina. While it is presently unclear whether all molecules which progress down the uvea-scleral route do so via the suprachoroidal space, it is clear that this path exists. Molecules progressing along the uvea-scleral route are subject to vascular absorption and clearance into the systemic circulation. This clearance route will most likely affect small molecules more than proteins (Stjernschantz et al. 1999; Alm and Nilsson 2009) (Fig. 5.2).

5.3  Eye Drops for Posterior Segment Diseases in the Clinic

The following discussion presents published data for drugs formulated as ophthalmic drops in clinical development, and for which there is data indicating that drug is reaching the posterior site of action via one or more of the transit routes described above. It is presently not possible to sample drug levels in most human ocular compartments without removal of the eye; however, one can sample aqueous and vitreous fluids from patients undergoing elective pars plana vitrectomy or other surgical procedures, and thus obtain a limited understanding of drug distribution patterns in ocular tissues. Emerging technological developments in confocal and laser microscopy utilizing dual photon excitation may make it possible to noninvasively measure drug pharmacodynamics and PK in anterior and posterior segments of the human eye in the future (Wang et al. 2010). Presently, our understanding of drug distribution

in the eye is mainly limited to studies in preclinical animals. In the examples described below, there is a detailed understanding of preclinical ocular PK and tissue distribution, which has yielded insight into the mechanisms of drug transit to posterior tissues. Though it is often challenging to delineate a transit path for a particular asset as being trans-vitreous trans-corneal, uvea-scleral trans-corneal, or periocular transscleral, the data is often suggestive of one route being a major contributor.

TG100801 is a dual inhibitor of VEGFR and Src kinases put into clinical development for the treatment of choroidal neovascularization (CNV) due to AMD (http:\\www.clinicaltrials.gov NCT00509548). It possesses a strong preclinical package of data which supports the ability of the drug to reach the back of the eye.

In a series of publications, TargaGen (Doukas et al. 2008; Palanki et al. 2008; Scheppke et al. 2008) has demonstrated that TG100801 is active in animal models of retinal disease when administered as eye drops. TG100801 is an inactive pro-drug of TG100572, which is rapidly hydrolyzed by esterases in ocular tissues. When a single eye drop (10 mL 0.7% w/v) was applied to a mouse eye, both compounds yielded measurable levels of TG100572 in the sclera, choroid, and retina, but the pro-drug delivered tenfold higher levels of drug to the retina (35 h mg/mL of TG100572 from TG100801 vs. 3.2 h mg/mL TG100572) and sustained drug levels in all tissues for longer periods. In addition, similarly high levels of pro-drug were measured in the retina, and suggests that the higher retinal levels of drug may be due to increased penetration of the pro-drug through the retinal pigment epithelium. This trend is supported by the higher lipophilicity of the pro-drug. Plasma levels were undetectable (<1 ng/mL) at all timepoints. Qualitatively similar results were obtained in rats, though the conversion of TG100801 to TG100572 appeared to be slower.

The PK of TG100572 and TG100801 were also evaluated in rabbits, a larger species whose eye size and geometry more closely mimics human eyes. As in rodents, significant concentrations are delivered to the sclera/choroid/retina tissues. Relative to these drug levels, very little drug was measured in aqueous humor and lens, with intermediate levels measured in the vitreous. This suggests that the trans-corneal routes are not the major paths for drug transit. A radiotracer study with 14C-TG100801 confirms the local nature of the delivery to the posterior eye and absence of significant systemic exposure or distribution to the fellow eye (Struble et al. 2007).

In a mouse laser CNV model in which CNV was induced by laser irradiation in both eyes, treatment of one eye with TG100801 reduced the size of the lesion relative to a vehicle-treated eye. The untreated fellow eye showed no effect, and strongly suggests that the treated eye effects were via local, rather than systemic delivery of drug. Whereas the laser CNV model is a measure of the drug’s activity at the choroid, a VEGF-induced retinal leak model is used to assess the ability of drug to access the retina. TG100801 (1.22% w/v, q.d.) completely abolished the retinal leak. The totality of data for TG100801 (efficacy via local delivery, low aqueous humor drug levels) suggests that it reaches the retina via the periocular trans-scleral route.

ATG-3 is a topical eye drop formulation of Mecamylamine (broad spectrum mAChR antagonist) under development by Comentis for wet AMD. In a 16-week Ph I/II trial for diabetic macular edema (Campochiaro et al. 2010), results suggested that 8/21 patients showed convincing improvements in best corrected visual acuity

and/or foveal thickness. Mecamylamine, as a 0.1 or 1% eye drop solution, was shown to be efficacious in a mouse laser-induced CNV model. (Kiuchi et al. 2008) Though the effect of drug on diseased fellow eye was not reported, significant levels of drug were detected in retina/choroid tissues of the treated eye with no detectable levels in plasma. In data published in a patent application (Zhang et al. 2007), topical administration of a 3% solution to the rabbit eye yielded high ratios of retina/choroid to plasma concentrations (plasma levels <50 ng/mL). Compared to an i.v. dose of 15 mg/kg (>30-fold higher total dose compared to eye drop dose), retina/choroid drug levels were higher via eye drop. Collectively, these data suggest that transit to the site of action is via local route(s). Tissue levels of Mecamylamine are measured to be sclera > retina/choroid > vitreous, but aqueous humor levels are also high. The concentration gradient suggests that Mecamylamine most likely transits by the periocular trans-scleral route, but given the high levels in the anterior segment, the uvea-scleral trans-corneal route may also contribute to posterior drug levels.

Memantine is in Ph III clinical development by Allergan as a topical eye drop for neuroprotection in glaucoma (Hughes et al. 2005; Koeberle et al. 2006). Memantine topically dosed to rabbits (0.1% BID for 7 days) resulted in a retinal drug level of 107 ng/mL, similar to that measured with an efficious oral dose of 2 mg/kg. In addition, relatively low levels of drug were measured in the contra-lateral eye, suggesting that drug is reaching posterior tissues via local routes. Memantine binds to melanin at the in vitro level, and drug accumulates to higher levels in pigmented animals. It was reported that autoradiography using 14C-Memantine indicated passage of drug to the retina via the periocular trans-scleral route (data not reported).

Until recently, there was little information around the ability of proteins to transit to the back of the eye via topical eye drops. Molecules as large as dextran (70 kDa) have been demonstrated to penetrate the sclera, the permeability of which has been shown to be inversely correlated with the radius of the molecule (Ambati et al. 2000; Geroski and Edelhauser 2001). Overall, there is little barrier to the diffusion of small and large molecules across the scleral meshwork from extra-scleral periocular fluid. An engineered 28 kDa single chain variable-region fragment (scFv) was shown to yield, via eye drop administration (50 mL; 0.2 mg/mL; application every 20 min for 12 h), ~3 mg/mL of antibody in the aqueous humor of rabbit eyes (Thiel et al. 2002). In contrast, a full-length 146 kDa IgG antibody, was not detected in this compartment. Levels in posterior tissues were not reported. However, in a subsequent report by the same investigator (Williams et al. 2005), it was shown that antibody fragments can be delivered to the back of the eye. Topical dosing of an eye drop formulation of the 28 kDa scFv (50 mL; 0.2 mg/mL; application every 20 min for 12 h) yielded vitreous drug levels of 50–150 ng/mL at 12 h post dosing. Under the same protocol, the full-length IgG was not detected in vitreous, and indicates that higher molecular weight proteins may not penetrate to posterior tissues. In Dutch-Belted rabbits, antibody was not detected in the serum, suggesting a local path for drug transit to back of eye.

More recently, a single chain anti-TNFa scFv antibody fragment (ESBA105) was reported to yield good penetration to posterior ocular compartments when dosed as a topical eye drop (Furrer et al. 2009). Hourly eye drop application to rabbit eyes of a 10 mg/mL solution of scFv over 10 h resulted in >100 ng/mL concentrations

Table 5.3  10 mg/mL ESBA105 application to rabbit eyes (hourly for 10 h)

of antibody in vitreous, neuroretina and RPE-choroid (Table 5.3). Significantly lower levels in serum were measured.

Much lower levels of antibody were measured in the fellow untreated eye. The ocular tissue distribution patterns of treated and fellow eye are similar, and are significantly different from that observed from an i.v. study. Systemic drug levels are 80–1,000 times lower than that measured in individual ocular compartments. These data suggest that delivery to the posterior compartments is via a local route. Levels of antibody in the aqueous humor are low relative to posterior tissues, and suggest a periocular transscleral path is taken towards the back of the eye. In vitro permeation studies using enucleated eyes are also supportive of this (Ottiger et al. 2009).

ESBA105 has subsequently been reported to show activity in a monkey laser CNV model via eye drops (50 mL; 10 mg/mL; 10 drops per day, 36 days) (Lichtlen et al. 2010). In May 2009, recruitment of patients for an anterior uveitis study with ESBA105 was underway (http:\\www.clinicaltrials.gov NCT00823173).

There are two later stage clinical eye drop assets also worth highlighting for the completion of this discussion. OT-551, a drug with an antioxidant mechanism of action, recently completed a Ph II trial in geographic atrophy in which there was limited or no benefit to patients (Wong et al. 2010). In addition, Alcon is reported to be in the midst of a Ph III study with AL-8309B (tandospirone; 5-HT 1a receptor antagonist), also for geographic atrophy (http:\\www.clinicaltrials.gov NCT00890097). For both examples, there is no preclinical data published which sheds light on how drug reaches the posterior tissues.

5.4  Summary

A number of examples of clinical topical eye drop medications for back of the eye diseases have been reported in recent years. For several of these drugs, the mechanisms of transit from the ocular surface to the back of eye have been assessed in detail by ocular tissue distribution studies and/or efficacy models.

Present understanding of these mechanisms suggests three potential paths for local drug delivery: trans-vitreous, uvea-scleral, and periocular. The first two are characterized by penetration of drug into the anterior chamber, followed by distribution into the vitreous and uvea-scleral tissues, respectively. Access to the anterior chamber is mainly via corneal permeation. Periocular delivery is effected by initial conjunctival penetration, transit of drug around the exterior of the eye globe, followed by diffusion through the sclera and interior tissues. The initial tissue penetration event (cornea or conjunctiva) is relatively inefficient (typically <10%) and will be dependent on the physiochemical properties of the molecule, but the ability to