Quality-by-Design (QbD): An integrated multivariate approach for the component quantification in powder blends

The objective of this study was to develop an integrated multivariate approach to quantify the constituent concentrations of both drug and excipients of powder blends. A mixture design was created to include 26 powder formulations consisting of ibuprofen as the model drug and three excipients (HPMC, MCC, and Eudragit L100-55). The mixer was stopped at various time points to enable near-infrared (NIR) scan of the powder mixture and sampling for UV assay. Partial least square (PLS), principal component regression (PCR), and multiple linear regression (MLR) models were established to link the formulation concentrations with the Savitzky-Golay 1st derivative NIR spectral data at various characteristic wavelengths of each component. PLS models based on the NIR data and UV data were calibrated and validated. They predicted the main components' concentrations well in the powder blends, although prediction errors were larger for minor components. As expected from the complete random-mixture (CRM) model, the measurement uncertainties were higher for minor components in the powder formulations. The prediction performance differences between the NIR model and UV model were explained in the context of scale of scrutiny and model applicability. The importance of understanding excipient variability in powder blending and its implication for blending homogeneity assessment is highlighted. Published by Elsevier B.V.