Correlation-based prediction of tissue-to-plasma partition coefficients using readily available input parameters

Area of Science:

• Pharmacokinetics and Drug Metabolism
• Computational Chemistry
• Toxicology

Background:

• Tissue-to-plasma partition coefficients (Kp) are crucial for physiologically based pharmacokinetic (PBPK) models, influencing drug distribution predictions.
• Accurate Kp values are essential for early-stage compound evaluation and PBPK model development.
• Existing methods for Kp determination can be resource-intensive or rely on parameters not available early in drug development.

Purpose of the Study:

• To develop an empirically derived algorithm for predicting tissue-to-plasma partition coefficients (Kp).
• To utilize input parameters that are readily available early in a compound's investigation.
• To enhance the accuracy and applicability of PBPK models in drug development.

Main Methods:

• Developed a Kp prediction algorithm using a dataset of 97 compounds, divided by acidic/basic properties.
• Employed multiple stepwise regression to correlate experimentally derived Kp values with rat volume of distribution at steady state (Vss) and physicochemical parameters.
• Incorporated parameters like lipophilicity, ionization, and protein binding to address inter-organ distribution variability.

Main Results:

• Generated prediction equations for Kp values across 11 different tissues.
• Validated the model with a separate dataset of 20 compounds, achieving a two-fold error deviation for 65% of predicted Kp values.
• Demonstrated superior prediction accuracy compared to existing empirical and mechanistic tissue-composition algorithms.

Conclusions:

• The developed algorithm provides a novel and efficient method for predicting Kp values.
• Readily available input parameters and reasonable prediction accuracy enhance PBPK model usability.
• This approach increases confidence in PBPK model outputs for early drug development decisions.