Particle size distribution (PSD) variability in excipients may cause unacceptable prolongation of mixing time needed to achieve blend homogeneity. Therefore, it is vital to modulate mixing through real-time monitoring of PSD variability. Notwithstanding the criticality of PSD variability, real-time measurement of PSD during mixing is relatively unexplored; and this is the focus of the present study. The model excipient was commercial grade lactose with modified PSD that conformed to the manufacturer's specifications. It was mixed with microcrystalline cellulose and chlorpheniramine in a double-cone blender. High and low dose blends were prepared and near infrared spectroscopy (NIRS) was used to collect spectral data, during mixing, for chemometric modelling of PSD. Four modelling approaches based on partial least squares regression (PLSR) were applied. The models were highly interpretable and rapidly measured PSD near the beginning of mixing (5th to 6th rotation), with accuracy (relative standard error of prediction <5.0%, r2 ≈ 1.00, slope ≈ 1.00). Therefore, NIR chemometric modelling is a viable strategy to detect variability in PSD of excipients during blending and could enable real-time control of mixing. Most significantly, this strategy is potentially transferable to the monitoring and controlling of batch and continuous processes, where PSD is either a source of process variability or a critical quality attribute.
Keywords: Blending; Chemometrics; Excipient variability; Near infrared spectroscopy; Particle size distribution; Process analytical technology.
Copyright © 2019 Elsevier B.V. All rights reserved.