The most direct means of delivering drug to the vitreous humor and retina is by intravitreal injection. While this method of administration has been associated with serious side effects, such as endophthalmitis, cataract, hemorrhage, and retinal detachment (198), aggravated by the conventional need for serial injections further increasing the risk, intravitreal injection continues to be the mode of choice for treatment of acute intraocular therapy and has become the standard of care for providing treatment of several chronic ocular diseases, such as age-related macular degeneration (AMD) and associated retinal edema. For example, Lucentis (ranibizumab), injected monthly, can be highly effective in decreasing the thickness of the edematous retina in AMD patients.
In response to well-known risks of recurring intravitreal injections, intravitreal inserts have been, and are continuing to be, developed to deliver drug for protracted periods following implantation. The first was the Vitrasert® (Bausch & Lomb), delivering ganciclovir for treatment of cytomegalovirus (CMV) retinitis (240,241). Later, a similar device, the RETISERT™ (Fig. 15), was developed to deliver fluocinolone acetonide for
Figure 15 Drawing showing the optimum placement of an intravitreal sustained-release device in the eye. Source: From Ref. 240.
the treatment of uveitis (242). Surmodics' I-Vation™ device is a corkscrew-like thumbtack and has been used to incorporate triamcinolone acetonide for treatment of diabetic macular edema (DME) (243). Neurotech encapsulated cell technology is being investigated for delivery of ciliary neurotrophic factor (CNTF) for neurodegenerative retinal diseases (244). Because these devices must be surgically implanted and subsequently removed, biodegradable alternatives also are being evaluated. Two such devices under development are Allergan's Posurdex®, delivering dexamethasone (245), and Alimera Sciences Medidur™, delivering fluocinolone acetonide, both drugs used for the treatment of DME.
Owing to the nonstirred nature of the vitreal space, the kinetic behavior of intravitreally delivered drugs is governed by mechanisms of diffusion, hydrostatic and osmotic pressure, convective flow, and active transport (246). Diffusion is the predominant mechanism for transvitreal movement for small- to moderately sized molecules, such as fluorescein or dextran (18,246), with kinetics similar to that observed in water or saline. Although low-level convective flow has been observed within the vitreous, this flow has only a negligible effect on transvitreal movement in comparison with diffusion for drug molecules of low molecular weight (247,248). However, the change in viscoelasticity with age might increase the contribution of convective flow.
Two primary mechanisms of vitreal drug distribution and elimination are (i) diffusion from the lens region toward the retina with elimination via the retina-choroid-sclera and (li) anterior diffusion with elimination via the hyaloid membrane and posterior chamber (18). Distribution to the retina from an intravitreal injection site is relatively slow, considering juxtaposition of the vitreous and retina, with the time for maximum drug concentration (/max) in retina typically achieved at 4 to 12 hours, and reflects the inefficiency of diffusion over the distances encountered within the vitreous body. For example, ranibizumab, an ocular specific monoclonal VEGF antibody, distributes to the retina with imax of 6 to 24 hours. While relatively rapid therapeutically, this is slow compared with the rate of redistribution in stirred compartments. (249).
The dominant path of distribution and elimination in the vitreous depends on a molecule's physicochemical properties and its substrate affinity. Lipophilic compounds, such as fluorescein (250) or dexamethasone (251), and compounds subject to active transport mechanisms, tend to be eliminated via the retina (Fig. 16). On the other hand, hydrophilic substances, such as fluorescein glucuronide, and compounds with poor retinal permeability, such as fluorescein dextran, tend to exit the vitreous anteriorly through the hyaloid membrane into the posterior chamber and subsequently into the anterior chamber, where they are subject to elimination pathways for aqueous humor (250). In general, shorter vitreal half-lives are associated with elimination through the retina, with its high surface area, whereas longer half-lives are indicative of elimination through the hyaloid membrane.
Volume and location of an intravitreal injection can impact the patterns of ocular distribution and elimination. Friedrich et al. demonstrated a substantial effect of both on vitreal distribution and elimination of fluorescein and fluorescein glucuronide (252), with evaluation of four extreme positions and two injection volumes, 15 or 100 |iL. The mean drug concentration remaining in the vitreous 24 hours postdose varied up to 3.8-fold with injection position, and increasing injection volume reduced this effect.
Aphakia and retinal inflammation are common pathophysiological conditions that can alter vitreal retention and kinetics of elimination. Elimination is generally faster in aphakic eyes, especially for drugs with low retinal permeability and injected in the near vicinity of the lens capsule. For example, Wingard et al. showed that intravitreally injected amphotericin B progressively accumulated in the sclera-choroid-retina in control
Figure 16 Contours of fluorescent intensity in frozen sections of the rabbit eye following 15 |xL injection of marker solutions in the central vitreous cavity; injection was conducted through the superior rectus muscle. (A) Fifteen hours following injection of 0.2% sodium fluorescein. (B) Fourteen days following injection of 0.1% FITC-dextran, molecular weight 66,000. Source: From Ref. 250.
phakic eyes, a phenomenon not observed in aphakic eyes (253). Half-life in the whole phakic eye was 6.9 to 15.1 days, whereas that in the aphakic eye was only 1.8 days. Retinal inflammation is known to increase blood-retinal barrier permeability. Drug diffusivity and retinal permeability are important factors that determine elimination from the vitreous, particularly when the blood-retinal barrier is compromised (254).
In the case of infected eyes, Ben-Nun et al. demonstrated that the elimination rate of intravitreally injected gentamicin was greater in infected than normal eyes, attributed to alteration in blood-retinal barrier (255). The vitreal half-lives of ceftizoxime, ceftriaxone, ceftazidime, and cefepime in rabbits ranged from 5.7 to 20 hours in rabbits with uninflamed eyes and from 9.4 to 21.5 hours in rabbits with infected eyes (256).
As noted above, some compounds are actively and rapidly transported out of the vitreous through the blood-retinal barrier. The so-called efflux transporters of the blood-retinal barrier include P-glycoprotein (P-gp), multidrug resistance protein (MRP), and breast cancer-related protein (BRCP) (257). Evidence for active transport in the blood-retinal barrier has been provided by Mochizuki, who investigated the transport of indomethacin in the anterior uvea of the albino rabbit, both in vitro and in vivo, the latter following intravitreal injection (258). An energy-dependent carrier-mediated transport mechanism with low affinity was observed in the anterior uvea of the rabbit that could have accounted for the drug's rapid clearance (30%/hr) from the eye. In another example of active transport in the retina, Yoshida et al. characterized the active transport mechanism of the blood-retinal barrier by estimating its inward and outward permeability in monkey eyes using vitreous fluorophotometry following intravitreally injected fluorescein and fluorescein glucuronide (259). Outward permeability (Pout) was 7.7 and 1.7 x 1CT* cm/min, respectively, and Pout/An was 160 for fluorescein and 26 for fluorescein glucuronide. Moreover, intraperitoneal injection of probenecid caused a significant decrease in Pout for fluorescein but had no effect on fluorescein glucuronide Poat. Concomitant intraperitoneal injection of probenecid has been shown to prolong the vitreal half-life of the cephalosporins indicating a secretory mechanism. Barza et al. studied the ocular pharmacokinetics of carbenicillin, cefazolin, and gentamicin following intravitreal administration to rhesus monkeys (260).
The simplest model for exploring the kinetics of elimination assumes a well-stirred vitreous body; this applies almost exclusively to studies employing larger injection volumes, 100 |iL or more, where the injection process itself can "stir" and alter the vitreal space. More relevant and sophisticated modeling explores diffusion through a relatively stagnant vitreous humor. For example, Ohtori and Tojo applied Fick's second law of diffusion to understand the process of elimination of dexamethasone sodium m-sulfobenzoate (DMSB) following injection in the rabbit vitreous body. They demonstrated that the rate of elimination was greater in vivo than in vitro (261). The model assumed a cylindrical vitreous body with three major elimination pathways: posterior aqueous chamber, retinal-choroidal-scleral membrane, and lens. The results showed that the retina, with its large surface area, is a major pathway for elimination. In a separate study, Tojo and Ohtori used the cylindrical model approach to demonstrate three potential pathways of elimination, including the annular gap, the lens, and the retina-choroid-sclera (262). The drug's site of injection or initial distribution profile affected retinal levels. Drug injected into the anterior segment of the vitreous was shown to exit rapidly through the annular gap into the posterior chamber. It follows that that drugs injected into the posterior vitreous can prolong therapeutic levels in the retina.
More precise modeling of vitreal pharmacokinetics employs the engineering technique of finite element analysis, which has now been conducted by a number of investigators (247,248,252,254,263-268). This approach accounts for the geometry and boundary conditions of the vitreous and predicts with relatively high resolution the concentration gradients. The power of these methods for accurately predicting the disposition of drug in the eye is illustrated in Figure 17, where finite element analysis was used to simulate the experiments of Ref. 250. When the retinal permeability is high, the contours resemble the experimental result for fluorescein, with the highest concentration immediately behind the lens. When the only pathway allowed is through the hyloid membrane, the contours resemble those obtained for the dextran polymer, with the highest concentration at the rear of the vitreous cavity. Other physiological details included in simulations have been the hydrodynamics of the aqueous humor (264), the influence of the directionality of release from an intravitreal device (247,265), dynamic partitioning of drug between various ocular tissues (266) (also simulated in Fig. 17), and IOP (247). Figure 18 demonstrates that IOP does not appreciably alter the concentration profile for small drug molecules. The very marked concentration gradients depicted have been confirmed by experimental measurements using nuclear magnetic resonance (NMR) imaging of a paramagnetic drug surrogate (269).
While intravitreal injection can be quite effective and has become the route of choice by retinal specialists, in spite of known risks, periocular drug administration, using
Figure 17 Finite element modeling of concentration profiles established after central intravitreal bolus injection, using FIDAP, 3D geometric model for the posterior rabbit eye, similar to Ref. 247, comprising the vitreous region, the retina, choroid, and sclera. At time zero, a spherical bolus of drug having unit concentration was placed in the center, behind the lens. (A) Contours of drug concentration simulated six hours after injection, assuming a reasonable value (250) for the diffusion coefficient (6 x 10-6 cm2/sec) and efficient clearance by the choroid. Permeability across the lens and hyloid membrane is assumed to be negligible. (B) Simulated concentration profile two days after injection of FITC-dextran (diffusion coefficient 6 x 10-7 cm2/sec), assuming zero retinal permeability and hyloid permeability 8 x 10-5 cm/sec. Drug partition coefficients in the vitreous, retina, choroid, and sclera were assumed to be roughly 1, 4, 4, and 2, respectively.
Was this article helpful?