Focal delivery and sampling within heterogeneous tissues

Abstract

Heterogeneity exists on multiple scales within the body, at the organ, tissue, and cellular level. Such heterogeneity presents a challenge when attempting to modulate or interrogate different tissues. The brain is one such tissue. Many neuropsychiatric disorders, for example, arise from pathologic signaling from a single neural node. Patients often fail to respond to oral medication therapy due to poor control over spatial and temporal resolution of medication, and off-target effects. Tumors are also inherently heterogeneous. This renders analysis of them difficult in experimental studies investigating tumor sensitivity to therapies. This thesis describes approaches for targeting of heterogeneous brain and tumor tissues with high spatial and temporal resolution. Chronic miniaturized neural drug delivery systems (MiNDS) were fabricated for repeated targeting of brains structures. Modular assembly techniques enabled variation in probe length and modalities (electric, fluidic, etc.).

Positron emission tomography (PET) was used to visualize intracerebral infusions of various volumes through chronic probes. Muscimol, a GABA agonist, was delivered using MiNDS and its effect on modulation of neuronal firing and behavior was characterized. These findings highlight that volume is as important to consider as dose when targeting neural structures. Other technologies for refined neural targeting were also developed. Concentric marking electrodes (CME) allow for simultaneous marking and stimulation. Marks are visible on MRI for real-time modification of experimental approach, and remain visible for up to 10 months. Polished microprobes were developed for single-modality interfacing in a 60 ptm footprint, minimizing chronic gliosis. Polishing microprobes allowed for independent implantation and steering. Varying the probe polish angle, material, and size resulted in distinct insertion trajectories.

A computational framework was also developed based to optimize drug delivery to any given brain structure. Microdevices containing micro-depots of various drugs can be used to assay drug efficacy on patientderived xenograft (PDX) tumor models. However, PDXs have significant tissue heterogeneity in tissue. A machine learning method of classifying tumors into various tissue types was developed, and allowed for greater accuracy in characterizing drug efficacy on individual PDXs.