Modelling convection-enhanced delivery in normal and oedematous brain.

Convection-enhanced delivery (CED) could have clinical applications in the delivery of neuroprotective agents in brain injury states, such as ischaemic stroke. For CED to be safe and effective, a physician must have accurate knowledge of how concentration distributions will be affected by catheter location, flow rate and other similar parameters. In most clinical applications of CED, brain microstructures will be altered by pathological injury processes. Ischaemic stroke and other acute brain injury states are complicated by formation of cytotoxic oedema, in which cellular swelling decreases the fractional volume of the extracellular space (ECS). Such changes would be expected to significantly alter the distribution of neuroprotective agents delivered by CED. Quantitative characterization of these changes will help confirm this prediction and assist in efforts to model the distribution of therapeutic agents. Three-dimensional computational models based on a Nodal Point Integration (NPI) scheme were developed to model infusions in normal brain and brain with cytotoxic oedema. These models were compared to experimental data in which CED was studied in normal brain and in a middle cerebral artery (MCA) occlusion model of cytotoxic oedema. The computational models predicted concentration distributions with reasonable accuracy.

Quantitative three-dimensional analysis and diffusion modeling of oligonucleotide concentrations after direct intraparenchymal brain infusion.

We compared quantitative experimental results on the diffusion of 35S-labeled phosphorothioate oligonucleotide (PS-ODN) after intraparenchymal infusion in rat brain, with the distributions predicted by Fick's second law of diffusion. Fischer 344 rats underwent identical intracerebral infusions of 36S-PS-ODN. After 0, 5, 11, 23, and 47 h, groups of animals were sacrificed and sequential brain cryosections subjected to autoradiography. The resulting experimental data were compared to the predicted distributions, for estimation of the apparent free diffusion coefficient, D*. Volumes of distribution and total content of 36 S-PS-ODN in the parenchyma were also computed, to monitor loss of total material. The values for D* were within the expected range for the 21-mer PS-ODN used, but a progressive decrease in D* over time was noted. The model requires D* to remain constant and, thus, does not adequately explain the spread of 35S-PS-ODN following infusion. The progressive slowing of spread over time suggests that at later time points, 35S-PS-ODN may be fixed by tissue binding or cellular uptake in the brain. Loss of material via vascular and CSF clearance may also contribute to the lack of fit. Our results highlight issues to be addressed in the modeling and experimental design of the intraparenchymal infusion process.