A multivariate methodology for material sparing characterization and blend design in drug product development.

This study developed a material and time saving method for powder characterization. Building on an earlier developed raw material property database for use towards development of pharmaceutical dry powder processes, blends were selected in an efficient way to include maximal variability of the underlying raw material dataset. For both raw materials and blends, powder characterization methods were kept to a minimum by selecting the testing methods that described the highest amount of variability in physical powder properties based on principal component analysis (PCA). This method selection was made by identifying the overarching properties described by the principal components of the PCA model. Ring shear testing, powder bed compressibility, bulk/tapped density, helium pycnometry, loss on drying and aeration were identified as the most discriminating characterization techniques from this dataset to detect differences in physical powder properties. This ensured a workload reduction while most of the powder variability that could be detected was still included. The methodology proposed in this paper could be used as a material-saving alternative to the current "Design of Experiment" approach, which will be investigated further for applicability to speed up the development of formulations and processes for new drug products and building an end-to-end predictive platform.