Computational Screening of Drug Solvates.

Area of Science:

• Pharmaceutical Science
• Computational Chemistry
• Crystallography

Background:

• Solvate formation is a common challenge in pharmaceutical development, potentially leading to undesired by-products and inaccurate solubility measurements.
• Predicting the likelihood of solvate formation is crucial for efficient drug discovery and development processes.

Purpose of the Study:

• To develop a simple computational approach for identifying drug-solvent pairs prone to forming crystalline solid phases.
• To provide a predictive tool to assist in the early stages of pharmaceutical development.

Main Methods:

• Utilized computational liquid phase thermodynamics (COSMO-RS theory) to calculate mixing enthalpy as a measure of drug-solvent complementarity.
• Incorporated additional descriptors related to drug and solvent shape and topology.
• Developed a statistically robust model using an extensive dataset of drug-solvent pairs.

Main Results:

• Mixing enthalpy of the subcooled liquid mixture is a key factor controlling solvate formation, similar to cocrystal screening.
• Solvent shape is another significant factor, especially for solvates forming inclusion compounds.
• The model shows limitations in predicting solvates with channel-like voids.

Conclusions:

• The developed computational approach can predict solvate formation with approximately 81% accuracy (AUC=0.81).
• This method offers a significant advantage over trial-and-error approaches for identifying suitable solvent candidates.
• Early identification of potential solvate-forming solvents can streamline the drug development pipeline.