Multivariate data analysis to assess dry powder inhalers performance from powder properties

The study aimed at investigating the correlations among the physical and bulk properties of carrier based dry powder inhaler formulations and the performance of the powder inhaler device estimated by in-vitro tests for a specific active pharmaceutical ingredient (API), and at obtaining predictive models for the in-vitro performance. Samples from scale-up process batches having different formulations, process settings and bulk size, were characterized by rheological, density and particle size tests. In vitro performance was evaluated by several parameters obtained by a dosage unit sampling apparatus (DUSA) and a next generation impactor (NGI). Correlations between powder properties and performance propertieswere established using partial least square (PLS) regression analysis. Variable importance in projection (VIP)was used in order to assess the most influential powder characterization variables to estimate the analytical ones. Particle size, density and rate of flowability are significant for modeling the Delivered Dose of the API and the total quantity of powder related to each dose. Powder characterization variables, describing the degree of cohesiveness and the flow properties of powder, are related to the total amount of the active ingredient for different formulations. DUSA test variables were satisfactory predicted on the basis of powder characterization variables, while NGI performance variables were predicted with higher error.