Integrating Pharmacokinetic Modeling and in vivo Imaging

Advancements in small-animal imaging technology over the past decade enable quantitative assessment of dynamic in vivo distribution of radiolabeled compounds as well as quantitative sub-organ analysis in preclinical studies. Drug distribution and targeting depend on a large number of factors including affinity, immunoreactive fraction, radiolabel, molecular weight, blood clearance, cellular internalization rate, antigen density, tumor vascularity, dose, and specific activity. inviCRO develops and employs biologically-relevant mathematical models with physiologically meaningful parameters as tools for researchers in guiding the design, preclinical development, and translation of therapeutics and imaging biomarkers.

Models of Transport in Tumors

inviCRO has developed several compartmental and distributed PK models of molecular transport in tumors to predict the tumor accumulation of radiolabeled biologics and to estimate tumor properties in vivo (i.e. antigen density and tumor vascularity) from molecular imaging data. These models are proprietary extensions of the published Krogh cylinder model (Baxter and Jain, Br J Cancer, 1996; Jackson et al. Br J Cancer, 1999; Thurber et al. J Nucl Med, 2007) and a published compartmental model (Schmidt and Wittrup, Mol Cancer Ther, 2009). These models can be customized and validated for specific applications through a combination of imaging, autoradiography, and immunohistochemistry analysis.

Pre-clinical Data vs. Model Prediction

Correlation between time activity curve (right) generated from inviCRO's tumor accumulation pharmacokinetic predictive model and in vivo image data (left).

A general compartmental model is available as a tool in inviCRO’s VivoQuant software to simulate tumor accumulation over time for common tumor-targeting scaffolds (e.g., IgGs, minibodies, peptides). The tool allows the input of biological and image study parameters that affect tumor accumulation (affinity, antigen density, injected activity, dose, etc.). The tool calculates specific activity, isotope-to-ligand ratio, and initial blood concentration from the input parameters and displays the model simulation results in various units (%ID, %ID/g, μCi, μCi/mm³, nM, SUV).

Distributed models (i.e. that simulate spatial and temporal distribution) have also been developed and are currently used in several collaborations.

inviCRO’s tumor PK models can be used to:

• Predict the effect of dose, specific activity, affinity, immunoreactivity, blood clearance, radiolabel, internalization rate, antigen expression, vascularity, molecular weight on tumor uptake and tumor-to-background ratios
• Analyze microdistribution/penetration of ligand within the tumor and the effect of tumor, ligand and dosing parameters on microdistribution
• Estimate in vivo tumor parameters (such as antigen expression, tumor vascularity, and ligand internalization) from imaging data using nonlinear optimization
• Estimate relative fractions of free, bound, and internalized ligand over time
• Guide decision-making in study design (i.e. time points, isotope selection, specific activity)
• Correlate biological parameters with efficacy
• Connect in vitro/ex vivo characterization of new compounds with in vivo tumor imaging analysis to support preclinical development and clinical translation of tumor-targeted therapeutics and imaging biomarkers

Human Data vs. Model Prediction

Beer et al. demonstrated a significant correlation between SUV and integrin expression for imaging ¹⁸F-galacto RGD in human tumors (left). The clinical results fall within the expected range of results from distributed PK model simulation (right).

Simulations were performed for both a high vascularity (R = 50 um) and low vascularity (R = 100 um) case (right) and compared to experimental ex vivo immunohistochemistry data (left) for ¹⁸F-galacto RGD. The model accurately predicts the mean standardized uptake value (SUV) of ~0.4 in human tumors with varying integrin levels. The clinical PET data shows an increase in the variability of mean SUV with increasing integrin levels. The model captures this increase in variability when tumor vascularity is varied over typical values for human tumors, suggesting that there is a correlation between antigen expression and SUV.

The results of this comparison indicate that vascularity impacts standardized uptake values (SUV) in imaging with equivalent if not greater impact than antigen expression. These results suggest that a static total-tumor SUV analysis at a single time point is a limited approach if imaging without biopsy is to succeed as a means of patient selection/stratification and/or response to therapy.

Models of Transport in the Brain

In addition to tumor PK models, inviCRO is developing mathematical models of molecular transport in the brain for radiolabeled biologics.

For more information regarding these models and how they could be customized for a particular application, please email This e-mail address is being protected from spambots. You need JavaScript enabled to view it .