FIGURE 20.1 Drawing illustrating the principle of convection-enhanced delivery. After exiting the cannula tip, the infusate flows by bulk flow through channels in the extracellular space which are widened in the process.

through the tip and directly into the parenchymal tissues, the resulting pressure gradient then generates bulk flow of the fluid through the interstitial space. Drug solution permeates the targeted region at a final tissue concentration and volume of distribution governed by the infusion parameters, the flow resistance or hydraulic conductivity of the tissue, and the duration of the treatment. In this way, CED can avoid the screening effects of the blood-brain barrier and deliver anti-tumoral compounds and agents to specific locations, at desired concentrations, and over time-periods consistent with pharmacologic tolerance, pharmacokinetic stability of the agent, and central nervous system metabolic activity. Subsequent spread of the infusate within wider volumes can occur due to diffusion of the agent along the concentration gradient and to bulk flow along the pathways of tissue anisotropy, but metabolic uptake, vascular clearance, and sumping within the ventricular system ultimately provide pharmacodynamic and physiological sinks.

From a physical perspective, the CED process is governed in part by Darcy's Law for flow through a porous medium during the early (advective) component of the flow, and by Fick's Laws of diffusion during the latter part of the flow. The mathematical expressions of these laws are typically the starting points for analyses of transport within the intersti-tium. The biophysical basis of the diffusion process is discussed by Nicholson and Phillips and Nicholson, while models of the infusion process have been developed by Basser and Morrison et al. [1-5]. All these are analytical approaches to the transport problem (from which numerical estimates have been made), but computational simulations using finite-element methods have been developed as well, with the latter taking into account both the diffusion and infusion components of the flow [6,7]. Still another model, based on the poroelastic characteristics of tissue, has been developed by Chen et al. [8]. Its predictions regarding the volumes of distribution and pressure profiles of an infusion compare well against experimental data obtained during the delivery of dye into an agarose brain phantom gel that has been shown to be a useful surrogate of living mammalian brain tissue [9].

There is one newly emerging model of the flow patterns that occur during CED in which the predictive capability of the technique has advanced to the point where accurate treatment planning has become possible (see Fig. 20.2). This is the work of Raghavan et al., wherein the model is designed to be broadly applicable to most clinical situations [10]. It incorporates various provisions for sources and sinks of the infusate, e.g., backflow, leakage into the subarachnoid space, drainage into cavities, fluctuations in clearance rates resulting from variations in capillary permeability, etc. Their model uses anatomical and diffusion tensor magnetic resonance images as input, and it allows for anisotropic variations in hydraulic conductivity and diffusivity. The movement of the infusate is described by a stochastic differential equation with advective and diffusive terms. In their approach, the stochastic process is implemented as a simulation of the motion of individual particles, in such a way that local particle density is proportional to local concentration. Drug concentrations are predicted at a particular location and time, and the simulation steps are repeated for many particles. The approach allows re-sampling of particles for any intermediate time point and can be optimized for increased resolution or accuracy. They have shown in human trials that a drug injected at a plausible point near a tumor might flow in such a way as to be clinically useless, but such flow patterns can be predicted within enough time intraoperatively for the injection site to be corrected.

In general, CED bears resemblance to standard direct-injection delivery, but with important differences. For instance, the flow rates through the cannula or catheter are very low, typically between 0.5 and 5.0 ml min"1 (a presently ongoing phase III human clinical trial for recurrent glioblastoma multiforme infuses TransMID™ at 0.2 ml h"1 = 3.33 ml min"1). The goal for CED is one of introducing the infusate

FIGURE 20.2 Illustration of the use of a finite-element approach to model convection-enhanced delivery in an experimental setting. Equations of convection and diffusion in a poroelastic media have been solved over a rat brain hexahedral grid. Several concentration isosurfaces are shown following an infusion of 50 min and diffusion for 20 min.

FIGURE 20.2 Illustration of the use of a finite-element approach to model convection-enhanced delivery in an experimental setting. Equations of convection and diffusion in a poroelastic media have been solved over a rat brain hexahedral grid. Several concentration isosurfaces are shown following an infusion of 50 min and diffusion for 20 min.

in a slow, controlled manner so that the agent gently swells the extracellular matrix and flows inter-stitially as opposed to creating rupture paths in the tissues along which essentially uncontrolled bulk flow might instead occur. Concomitant with this is maintaining an interstitial fluid pressure at the outlet port(s) of the delivery device that is no more than two to three times the background intracranial pressure. This is enough of a differential to cause flow along the resulting pressure gradient, but not so high that hydrodynamically driven damage to the tissue would occur.

The presence of ultra-structural fenestrations in the pia mater might logically suggest that an intra-thecal means of drug delivery across the pia might be another way of delivering drugs into the paren-chymal tissues. However, in their review of modalities for the delivery of neurotrophic factors into the brain, Thorne and Frey reiterate the well-known fact that delivery via the cerebrospinal fluid is indeed able to achieve a wide area of coverage, but is limited in efficacy by cellular barriers to tissue penetration [11]. For instance, in the cerebellum, the Purkinje cell layer in particular appears to block diffusive traversal of large molecules through the pia, and this general class of findings argues strongly that any natural fenestra-tion that might exist in the pial covering of the cortical surfaces, as opposed to that surrounding the spinal cord, is insufficient to mediate a direct transpial transport process of anything but very small molecules [12].

As the ventricles are also one of the principal fluid-carrying compartments within the central nervous system, intraventricular delivery with the goal of obtaining eventual intraparenchymal distributions would also seem to be a potential avenue of approach. In fact, the volumes and flows of the cerebrospinal fluid and interstitial fluid pressure are interrelated, with the coupling between them arising at least in part due to the perivascular spaces of the brain's surface vessels. However, the blood-cerebrospinal fluid barrier at the ventricular walls plays a substantial role in regulating all aspects of exchange between the various compartments, and this in turn places limits on transport into the parenchymal tissues that are just as stringent as those dictated by the blood-brain barrier of the endothelial layer.

For all these reasons, direct intraparenchymal delivery via CED plays a unique role in achieving therapeutic concentrations of anti-tumoral agents on a regional basis within the brain. The technique is also being studied for cell delivery for neuro-degenerative diseases and will likely find several other clinical applications as well [13].

0 0

Post a comment