Cocrystal Prediction of Bexarotene by Graph Convolution Network and Bioavailability Improvement

Area of Science:

• Computational chemistry and materials science.
• Drug discovery and pharmaceutical development.
• Artificial intelligence in drug design.

Background:

• Bexarotene (BEX) is approved for cutaneous T-cell lymphoma (CTCL) but suffers from poor aqueous solubility and low bioavailability.
• Limited clinical application of BEX necessitates strategies to enhance its physicochemical properties.

Purpose of the Study:

• To develop and validate a deep learning model for in-silico screening of bexarotene cocrystals.
• To identify novel cocrystal forms of BEX with improved solubility and bioavailability.
• To integrate computational prediction with experimental validation for efficient cocrystal discovery.

Main Methods:

• Development of a Graph Convolutional Network (GCN)-based deep learning model (CocrystalGCN) for virtual screening of coformer candidates.
• In-silico scoring of 109 coformer candidates using CocrystalGCN.
• Experimental validation of top-ranked cocrystal candidates.
• Characterization of obtained cocrystals using single-crystal X-ray diffraction, powder X-ray diffraction, differential scanning calorimetry, and thermogravimetric analysis.
• Pharmacokinetic studies to evaluate in-vivo performance.

Main Results:

• CocrystalGCN demonstrated high performance in screening coformer candidates.
• Successfully synthesized and characterized novel cocrystals: BEX-pyrazine, BEX-2,5-dimethylpyrazine, BEX-methyl isonicotinate, and BEX-ethyl isonicotinate.
• All synthesized cocrystals exhibited superior solubility and dissolution rates compared to pure BEX.
• Pharmacokinetic studies revealed significantly enhanced plasma exposure (AUC0-8h) for BEX-pyrazine (1.7x) and BEX-2,5-dimethylpyrazine (1.8x) compared to commercial BEX powder.

Conclusions:

• The developed CocrystalGCN model is effective for in-silico cocrystal screening of poorly soluble drugs.
• Novel BEX cocrystals with enhanced solubility, dissolution, and pharmacokinetic profiles were successfully discovered.
• This integrated approach of virtual screening and experimental validation offers a promising strategy for developing new cocrystals of water-insoluble drugs.