Development of a decision tree to classify the most accurate tissue-specific tissue to plasma partition coefficient...

Area of Science:

• Pharmacokinetics and Drug Metabolism
• Computational Toxicology
• Pharmacometrics

Background:

• Physiologically based pharmacokinetic (PBPK) modeling is crucial for drug discovery and risk assessment.
• Tissue to plasma partition coefficients (Kp) are key PBPK parameters for predicting drug distribution.
• In silico Kp prediction methods exist but lack standardization and vary in accuracy.

Purpose of the Study:

• To develop a novel, accurate, and tissue-specific Kp prediction method.
• To improve the reliability of PBPK model outputs by enhancing Kp prediction.
• To provide a tool for Kp prediction with limited input parameters.

Main Methods:

• A decision-tree-based classifier was developed using six existing Kp prediction algorithms.
• A dataset of 122 drugs was used to train and validate the classifier.
• Three versions of tissue-specific classifiers were created based on input availability.

Main Results:

• The developed classifier achieved higher prediction accuracy than any single algorithm across all tissues.
• The method successfully identified the most accurate tissue-specific Kp prediction algorithm for a given drug.
• The tool provides reliable Kp predictions even with limited input parameters.

Conclusions:

• The novel decision-tree Kp prediction tool enhances the accuracy of PBPK modeling.
• This method improves confidence in predicting drug tissue distribution.
• The tool offers a valuable solution for PBPK model building with limited data.