Data-Driven Design of Novel Polymer Excipients for Pharmaceutical Amorphous Solid Dispersions

Area of Science:

• Pharmaceutical Sciences
• Polymer Chemistry
• Materials Science

Background:

• Many active pharmaceutical ingredients (APIs) exhibit poor aqueous solubility and low oral bioavailability, hindering drug delivery.
• Amorphous solid dispersions (ASDs) are a key strategy to enhance solubility and bioavailability by preventing API crystallization within polymer matrices.
• Understanding the structure-function relationships of polymers in ASDs is crucial for developing effective formulations, with high glass transition temperature (Tg) being a critical performance indicator.

Purpose of the Study:

• To investigate how polymer structural features influence the glass transition temperature (Tg) of amorphous solid dispersions (ASDs).
• To design novel copolymers with improved Tg for enhanced ASD stability and performance.
• To develop a predictive machine learning (ML) model for identifying high-Tg ASD formulations.

Main Methods:

• Synthesis of a library of copolymers using automated photoinduced electron/energy transfer-reversible addition-fragmentation chain-transfer (PET-RAFT) polymerization.
• Preparation and characterization of 50 amorphous solid dispersions (ASDs) using the model drug probucol.
• Application of a machine learning (ML) algorithm, specifically a Random Forest Regressor, trained on Tg data to identify key structure-Tg relationships.

Main Results:

• The ML model accurately predicted Tg for both polymers and probucol-loaded ASDs, achieving an average R² > 0.83 across 10-fold cross-validation.
• Key polymer features influencing Tg were identified, including backbone methylation and nonlinear side chain geometry.
• The study demonstrated the potential of ML in guiding the design of ASDs with desirable solid-state properties.

Conclusions:

• Machine learning effectively captures structure-Tg relationships in ASDs, enabling the prediction of novel copolymers.
• The developed ML model serves as a powerful tool for designing ASDs with high Tg, crucial for improving shelf stability and preventing drug crystallization.
• This approach facilitates the rational design of advanced oral drug delivery systems for poorly soluble APIs.