(56f) Prediction of Four Classes of Pharmacokinetic Distribution in Tumors Using Multichannel Fluorescence Microscopy

Delivery of drugs and imaging agents to tumors is inefficient and correlated with poor response in the clinic. Uptake and distribution is a complex interplay of the physiochemical properties of the drug and the tumor microenvironment. In this work, we use a combination of partial differential equation modeling to predict tumor tissue distribution and 5 channel fluorescence imaging to quantify the distribution of four drugs in A-431 tumor xenografts. The four agents, including a monoclonal antibody, peptidomimetic, small molecule drug, and fluorescent dye, were chosen based on predicted differences in the rate-limiting step in uptake and heterogeneous versus homogeneous distribution. The fluorescence results were in excellent agreement with predictions, validating the model for use in simulating small molecule to macromolecular delivery to tumors.

The large number of factors that determine tumoral delivery leads to an increased amount of in vivo testing to measure distribution. An alternative method, which is less expensive and time consuming, is to conduct detailed mathematical simulations to study the delivery of agents and to develop a model to predict tumor uptake. Theoretical studies have shown that these agents can be categorized into four different classes based on the rate-limiting step in delivery to its target. The current model calculates three dimensionless groups for each agent and categorizes the molecule into one of four classes: (1) Blood flow limited (2) Extravasation limited (3) Diffusion limited, or (4) Local binding or metabolism limited. The model consists of a system of coupled non-linear partial differential equations describing axial and radial gradients around vessels with time-dependent mixed boundary conditions. These equations are solved using finite differences in MATLAB. To facilitate predictions in other labs, a Graphical User Interface (GUI) was created to make it easier to run simulations without needing any prior knowledge of coding. In order to validate the model, four agents predicted to represent each class were selected, and their distribution was studied using multichannel fluorescence microscopy. Importantly, to isolate the effect of drug properties from the influence of the local tumor microenvironment, all four agents were injected into the same tumor-bearing mouse. The agents have distinct, non-overlapping fluorescence excitation and emission spectra in addition to representing the four classes of pharmacokinetic behavior and could therefore be imaged within the same tissue. The simulations consistently predicted the distribution of all four agents, demonstrating that with the help of in vitro experiments and mathematical models, accurate simulations of tumor distribution are possible. Using this model, the amount of time and resources spent on determining drug delivery can be dramatically reduced by aiding experimental design and data interpretation. Current work involves in vivo imaging of the distribution of fluorescent PARP inhibitors to generate correlations for the transient uptake of small molecule imaging agents. These results will allow better in silico predictions of tumor distribution based solely on the physiochemical properties of the drug.