Global Analysis of Models for Predicting Human Absorption: QSAR, In Vitro, and Preclinical Models

Area of Science:

• Pharmacokinetics and Drug Metabolism
• Computational Chemistry
• Drug Discovery

Background:

• Accurate prediction of intestinal absorption is crucial for drug development.
• Existing models face challenges with complex compounds like "beyond rule of 5" (bRo5) and poorly absorbed drugs.
• Rigorous evaluation of current models' applicability is needed.

Purpose of the Study:

• To statistically evaluate the predictive performance of in silico, in vitro, and in vivo models for intestinal absorption.
• To assess model utility for diverse chemical structures, including bRo5 and low absorption compounds.
• To provide a resource for drug metabolism and pharmacokinetic (DMPK) scientists and chemists.

Main Methods:

• Assembled a diverse dataset of approximately 1000 compounds with known human intestinal absorption data.
• Statistically evaluated quantitative structure-activity relationship (QSAR) models.
• Validated in vitro (Caco-2) and in vivo (rat) models against the human data.

Main Results:

• Quantitative structure-activity relationship (QSAR) models showed variable predictive performance.
• In vitro (Caco-2) models demonstrated moderate predictive accuracy.
• In vivo (rat) models showed the highest correlation with human intestinal absorption data, but with limitations.

Conclusions:

• The study provides a comprehensive evaluation of various models for predicting intestinal absorption.
• Results highlight the strengths and weaknesses of different modeling approaches for diverse compound types.
• This evaluation serves as a valuable resource for optimizing drug development strategies and model selection.