Analysis of the impact of material properties on tabletability by principal component analysis and partial least squares regression

Principal component analysis (PCA) and partial least squares regression (PLS) were combined in this study to identify key material descriptors determining tabletability in direct compression and roller compaction. An extensive material library including 119 material descriptors and tablet tensile strengths of 44 powders and roller compacted materials with varying drug loads was generated to systematically elucidate the impact of different material descriptors, raw API and filler properties as well as process route on tabletability. A PCA model was created which highlighted correlations between different powder descriptors and respective characterization methods and, thus, can enable reduction of analyses to save resources to a certain extent. Subsequently, PLS models were established to identify key material attributes for tabletability such as density and particle size but also surface energy, work of cohesion and wall friction, which were for the first time demonstrated by PLS as highly relevant for tabletability in roller compaction and direct compression. Further, PLS based on extensive material characterization enabled the prediction of tabletability of materials unknown to the model. Thus, this study highlighted how PCA and PLS are useful tools to elucidate the correlations between powder and tabletability, which will enable more robust prediction of manufacturability in formulation development.