In silico ADME modelling: prediction models for blood-brain barrier permeation using a systematic variable selection...

Area of Science:

• Medicinal Chemistry
• Computational Chemistry
• Pharmacokinetics

Background:

• Predicting blood-brain barrier (BBB) permeation is crucial for drug development.
• Existing computational methods require validation and improvement for accuracy.

Purpose of the Study:

• To develop and validate Quantitative Structure-Property Relationship (QSPR) models for predicting in vivo blood-brain barrier (BBB) permeation (logBB).
• To assess the performance of these models using diverse compound datasets and literature comparisons.
• To evaluate the utility of the developed QSPR models in virtual screening applications.

Main Methods:

• Utilized in vivo blood-brain permeation data (logBB) for 88 diverse compounds.
• Employed 324 molecular descriptors and a systematic variable selection method (VSMP).
• Developed and validated QSPR models, including a three-descriptor and a four-descriptor model, assessing their statistical performance (r, q, F, SE).

Main Results:

• The best three-descriptor model incorporated Atomic type E-state index (SsssN), AlogP98, and Van der Waal's surface area (r=0.8425).
• The best four-descriptor model included Kappa shape index, SsssN, AI topological descriptor (AIssssC), and AlogP98 (r=0.8638).
• Models demonstrated an 82% success rate in virtual screening for both BBB+ and BBB- compounds on a 91-compound test set.

Conclusions:

• The developed QSPR models accurately predict blood-brain barrier permeation using computed molecular descriptors.
• These models serve as effective tools for virtual screening, aiding in the selection and prioritization of potential drug candidates.
• The VSMP method proved efficient for developing robust QSPR models for logBB prediction.