A corneal-PAMPA-based in silico model for predicting corneal permeability

Area of Science:

• Ophthalmic Drug Development
• Computational Chemistry
• Pharmacokinetics

Background:

• Predicting corneal permeability is crucial for ophthalmic drug development.
• Previous models lacked compound diversity or had poor performance.
• Quantitative Structure-Property Relationship (QSPR) models offer a promising approach.

Purpose of the Study:

• To develop robust QSPR models for predicting corneal permeability and membrane retention.
• To assess the correlation between corneal permeability and other drug transport parameters.
• To provide accurate predictive tools for ophthalmic drug candidates.

Main Methods:

• In vitro corneal permeability measurements for 189 diverse compounds.
• Development of two QSPR models using partial least squares regression.
• Inclusion of molecular descriptors and Extended Connectivity Fingerprints (ECFP).
• Rigorous validation including internal and external testing.

Main Results:

• No significant correlation found between corneal-PAMPA permeability and Caco-2, jejunal permeability, or logBB.
• Developed QSPR models achieved R² > 0.90 for overall fit.
• Validation runs demonstrated R²-values not lower than 0.85.
• Models provide quick and accurate predictions for corneal permeability and membrane retention.

Conclusions:

• The developed QSPR models are effective for predicting corneal permeability and membrane retention.
• These models enhance the efficiency of ophthalmic drug development by enabling early-stage predictions.
• The study highlights the utility of QSPR in overcoming limitations of previous predictive approaches.