6.4  Flow Processes in the Vitreous

The movement of drugs around the eye following intra-ocular administration must occur as a function of several processes. These are grouped into three driving effects: hydrostatic pressure, diffusional drive and convective flow (Chastain 2003; Moseley 1981). The relative importance of each in affecting clearance after administration will be reflected in the drug and formulation, the mode of delivery, the physical state of the vitreous, the size and shape of the eye, the relationship of the depot to the intra-ocular structures and time. In addition, active transport mechanisms considered later in this chapter, are important.

6.4.1  Flow Patterns

The turnover of fluid in the rabbit vitreous body was described by Duke-Elder (1930). He proposed that the supply of liquid to the vitreous came from the ciliary body and pars planar region, flowing posteriorly through the vitreous to exit near to the optic nerve head. The observations by Duke-Elder led Fowlkes (1963) to investigate the vitreous flow patterns in the rabbit, using injections of Indian ink and the highly protein-bound blue dye nitro blue tetrazolium chloride, which forms an insoluble formazan-labelled protein in situ. The doses were administered starting in the region of the pars planar moving radially outwards, entering the eye near to the superior rectus and from the temporal side. The eyes were harvested and sectioned whilst frozen. Blue formazan stained the retina immediately posterior to the injection. It was observed that when the marker was injected at a shallow depth into the vitreous humour within 2 mm of the retina, it was swept posteriorly at a rate faster than diffusion. Fowlkes termed this movement meridonal flow and only occurred in live eyes. The behaviour was observed to be similar as the injection site was made radially away from the pars planar.

In the perfused Miyake-Apple preparation, increased movement of particles can be noted in surface zones although thermal effects may contribute to movement. This suggests that very short needle injections into the vitreous might access the posterior pole successfully. In mathematical simulations of flows in the eye, flow velocities were calculated to reach a maximum around the edge of the vitreous boundary (Missel 2002). If the viscosity is lower in the peripheral zone between the retina and the edge of the vitreous, it is likely that the particles near to the surface can spread underneath the retina. This is illustrated in Fig. 6.7.

Injection into the body of denser mid-vitreous within an ovine eye, as shown in Fig. 6.4, demonstrates a more tortuous path for the particles as shown in Fig. 6.6, probably following the cisternal margins described in previous literature (Jongebloed and Worst 1987). The appearance of the needle track will change with the structural and rheological properties of the vitreous humour, with pressure differences contributing to the initial disposition.

Fig. 6.4  Needle track of an injection of 10 mm fluorescent microparticles suspension into ovine vitreous humour. The suspension was well retained within the central vitreous gel phase after injection. Note the tortuous path (adapted from Laude et al. 2010, with permission)

Fig. 6.5  Injection of 10 mm fluorescent microparticles suspension into ovine vitreous humour contained within a cuvette. (a) During injection, (b) immediately after injection, (c) 3 h after injection

If the vitreous humour is decanted into a cuvette, the preservation of cisternal structure is noted, with settling of the particles injected into the humour at the top of the cuvette occurring under the gravitational forces to outline internal boundaries as shown in Fig. 6.5.

Fig. 6.6  The appearance of an intravitreal depot formed in ovine vitreous, viewed in an ovine eye in vitro

6.4.2  Injection and Hydrostatic Effects

The injection of even small volumes of liquid into an enclosed volume under pressure will cause a transient increase in hydrostatic pressure, sufficient to pose a risk of reduced retinal blood flow. The movement of suspensions in the eye can be seen using the Miyake-Apple technique, in which a cover slip is glued to the eye after removing a small circle of sclera creating a window on the vitreous. Illumination of the preparation through the lens using a high-intensity blue light emitting diode, allows the movement of a 10 mm suspension to be followed using a camera. As can be seen, the introduction of the needle creates a channel and a temporary, lowresistance pathway along the track of the injection path, leading to some reflux of the material as shown in Fig. 6.6. This may reflect the reflux scenario seen clinically (Benz et al. 2006; Boon et al. 2008). Morlet and Young reported that the mean IOP immediately following injection of six eyes in four patients with 0.1 mL of formulation was 44.5 mmHg, a mean rise of 38 mmHg which was significantly reduced by previous ocular decompression (Mortlet and Young 1993). Application of pressure to the injection site upon withdrawing the needle has been demonstrated to minimise the reflux of triamcinolone acetonide through the injection hole. Maurice has shown that injection of fluorescein made via the sclera through extra-ocular muscle reduces the regurgitation of large injection volume of 100 mL to 12% in rabbits; however, in this paper the loss in one animal was reported to be 32% (Maurice 1997).

In the study by Boon et al., a significant loss of fluid after injection was reported which was associated with a restoration of IOP to below 24 mmHg. On a further

investigation of 13 patients, the authors mixed fluorescein (1% w/v) with the dose of bevacizumab. Ten patients showed reflux of clear liquid, not stained with fluorescein, whereas one patient had reflux of largely fluorescein-stained liquid (Boon et al. 2008). This suggests that the non-homogeneity of vitreous humour and the induced pressure rise will influence the amount and nature of refluxate according to technique and operator.

Maurice described the consequences of multiple injections on the integrity of the vitreous. He dosed both eyes with fluorescein such that the manipulations on one eye could be compared across both eyes, using one as a control. He found that an injection of the vitreous with a 25G needle, even without introduction of fluid, caused a temporary change in integrity leading to increased loss of a fluorescein marker previously injected. Multiple injections at different sites around the rabbit eye led to even greater losses, which resolved at 48 h (Maurice 1987).

6.4.3  Diffusion

In earlier literature, flow within the vitreous humour was thought to be determined by diffusion alone (Maurice 1957; Moseley et al. 1984). Diffusion can be defined as a “random molecular motion that leads to complete mixing” and can be characterised using Fick’s law (Cussler 2009). Passive diffusion is the most fundamental transport mechanism for small molecules in liquid. It requires a differential gradient to provide motive force (such as osmotic pressure and concentration) towards creation of an equilibrium state at which point, no net diffusion occurs.

The vitreous may not be a simple, uniform gel as the structure of collagen– hyaluronan network varies depending on the local abundance and concentration of the macromolecules. The movement of tritiated water is slower in intact rabbit vitreous than water suggesting that structural elements constituted by the vitreous components impose a diffusional barrier to the transport of even very small molecules (Foulds et al. 1985). Similarly, the diffusion of dexamethasone is 4–5 times slower in the vitreous gel as compared to water (Gisladottir et al. 2009). Unlike small molecules, which can diffuse freely across the vitreous network, the diffusion activity of larger molecules appears to be limited by the fibrillar structure of the vitreous meshwork.

6.4.4  Convective Flow

Convection describes bulk movement in a fluid initiated through an applied force, for example, due to pressure gradients or temperature differences. In the vitreous humour, convection is thought to arise through a pressure drop between the anterior and posterior eye from a steady permeating flow, possibly generated

by pressure and temperature differences between the anterior chamber and the surface of the retina.

The impact of both diffusion and convection on flow within the vitreous has been described by several authors based on experimental data and mathematical modelling (Fatt 1975, 1977; Xu et al. 2000; Maurice 1987). Fatt used data from a study of Na+ flux to create a mathematical model of tracer movement within the vitreous. Using this model, he was able to demonstrate that both diffusion and convection played a role in the movement of the marker. Although the presence of convection was noted, convective processes appeared to have less of an impact in determining tracer distribution, with diffusion having an eightfold greater control over tracer movement. Using a specially constructed diffusion cell, Fatt concluded that hydraulic flow conductivity was greater in the bovine vitreous, when compared to the rabbit vitreous (Fatt 1977). The data was used to calculate the effective channel size through which water flows: the figure for both bovine and rabbit vitreous was approximately 0.4 mm. More recently, the diffusion coefficient of acid orange 8 in bovine vitreous and water were shown to differ (3.4 × 10−6 and 6.5 × 10−6, respectively). Using the data obtained, the hydraulic conductivity of bovine vitreous was determined to be 8.4 ± 4.5 × 10−7 cm2/Pa, suggesting the convection would play a role in the movement of acid orange 8 in bovine vitreous (Xu et al. 2000).

Dr Paul Missel has created a number of interesting finite element models to create a 3D representation of hydraulic flow within the eye (Missel 2002), based on data derived following intravitreal injection of different molecular weight dextrans. When the model was set up to disregard hydraulic flow, the elimination rate of the high molecular weight dextran (157 kDa) was reduced to below the elimination rate expected for a dextran of this size. No notable effect was seen for the lower molecular weight dextrans, leading to the conclusion that convection only appeared to be important for larger molecular weight molecules. Stay and colleagues reached similar conclusions using model compounds with diffusion coefficients of 5 × 10−6 and 1 × 10−7, respectively (Stay et al. 2003). Park’s group used a high diffusivity (1 × 10−5 cm2/s) in their simulation to represent compounds with an approximate MW of less than 100 Da (Park et al. 2005). When convective flow was altered by increasing vitreous outflow, little accumulation at the retina was predicted. An increase in accumulation of only 10% for a highly diffusible small molecule was noted, suggesting convection would have little influence on the pharmacokinetic movement of the drug. On the other hand, when using a low diffusivity to represent larger macromolecules with a molecular weight of greater than 40 kDa, the rate of diffusion was slow and convection appeared to have a more obvious role, with increased vitreous outflow causing a fourfold increase in accumulation at the retina after 50 h.

MRI data has been used in a similar manner to investigate the effect of reducing convective flow on the pharmacokinetic movement of the low molecular weight drug surrogate Gd-DPTA (Kim et al. 2005). Using the model, it was found that predicted changes in Gd-DPTA concentrations (MW 590 Da) were insignificant on switching convective flow on and off.