(463b) A Model-Based Approach to Identify Optimal Formulations for the Spray Drying of Pharmaceutical Compounds

Spray drying dispersion (SDD) of active pharmaceutical ingredients (APIs) is proven an effective technique for improving solubility, dissolution and bioavailability of poorly soluble drugs, and it is widely used in pharmaceutical manufacturing. However, preparing suitable formulations for SDD remains a challenging task as it requires identifying proper polymer and solvent blends with desired properties to ensure high solubility and stability of the drug [1]. In current practice, experimental protocols and heuristic approaches are extensively employed for pre-screening small sets of commonly used polymers and solvents for the spray drying of APIs. These methods are time consuming leading to longer development timelines. It is therefore important to develop computational methodologies to reduce the number of time-consuming experiments and enhance the quality and attributes of the drug product.

Over the last decade, a few model-based studies have employed thermodynamic tools to predict solubility and phase stability of given APIs and polymers [2-4]. However, they focus on a limited set of pre-defined mixtures with fixed ingredients and do not fully explore the entire design space of different materials and proportions in order to optimise the final formulation. In this work we present a systematic methodology for identifying optimal API-polymer-solvent formulations with desired physical and chemical properties that are used in the spray drying of drug products. In particular, we employ property-prediction models to estimate solubility, miscibility and the glass transition temperature of a wide range of chemical blends. In addition, we use advanced optimisation models [5] to design improved and environmentally friendly formulations that yield high solubility and stability of the drug. The design methodology is applied to the selection of optimal polymers and solvents that maximise the solubility of naproxen. A ranked list of optimal solutions (blends with different chemicals and compositions) is obtained by introducing integer cut inequalities into the model, and proper phase diagrams of the best mixture are constructed. Finally, the proposed model is incorporated in a polymer and solvent selection workflow that is used to identify better-performing designs and guide experimental work.

[1] M. Davisa, G. Walker, 2018. Journal of Controlled Release 269, 110-127.

[2] Y. Tian, J. Booth, E. Meehan, D. S. Jones, S. Li, G.P. Andrews, 2013. Molecular Pharmaceutics 10, 236−248.

[3] K. Bansal, U. S. Baghel, S. Thakral, 2016. AAPS PharmSciTech 17, 318-327.

[4] K. Lehmkemper, S.O. Kyeremateng, O. Heinzerling, M. Degenhardt, G. Sadowski, 2017. Molecular Pharmaceutics 14, 157-171.

[5] S. Jonuzaj, J. Cui, C.S. Adjiman, 2019. Computers & Chemical Engineering 130, 106518.