Computational models to predict blood-brain barrier permeation and CNS activity

Area of Science:

• Medicinal Chemistry
• Computational Chemistry
• Pharmacokinetics

Background:

• Predicting blood-brain barrier (BBB) permeation is crucial for developing drugs targeting the central nervous system (CNS).
• Quantitative structure-activity relationship (QSAR) models offer a computational approach to estimate drug permeability across the BBB.

Purpose of the Study:

• To develop and validate QSAR models for predicting blood-brain barrier permeation.
• To identify key molecular descriptors influencing blood-brain partitioning (logBB).
• To assess the utility of these models in drug discovery for CNS-active compounds.

Main Methods:

• Modeling blood-brain permeation for 281 diverse compounds using linear regression and genetic partial least squares (G/PLS).
• Utilizing quantitative structure-activity relationship (QSAR) models incorporating logP, polar surface area, and electrotopological indices.
• Validating models with training sets (58 compounds), external validation sets (34 molecules), and a prediction set of 181 drugs.

Main Results:

• High correlations (r > 0.9) achieved for training sets and ensemble model comparisons with experimental logBB values.
• Good agreement (r ≈ 0.7) observed for external validation sets.
• Successful qualitative prediction (>70%) for CNS-permeable drugs, with approximately 60% for CNS-impermeable drugs.

Conclusions:

• Developed QSAR models effectively predict blood-brain barrier penetration.
• Key molecular descriptors like logP and polar surface area are vital for accurate logBB predictions.
• Models demonstrate utility in drug discovery for identifying CNS-penetrant compounds and estimating penetration for large compound libraries.