Comparing a Statistical Model and Bayesian Approach to Establish the Design Space for the Coating of Ciprofloxacin HCl Beads at Different Scales of Production

The primary objective of this study was to compare two methods for establishing a design space for critical process parameters that affect ethylcellulose film coating of multiparticulate beads and assess this design space validity across manufacturing scales. While there are many factors that can affect film coating, this study will focus on the effects processing conditions have on the quality and extent of film formation, as evaluated by their impact coating yield and drug release. Ciprofloxacin HCl layered beads were utilized as an active substrate core, ethylcellulose aqueous dispersion as a controlled release polymer, and triethyl citrate as a plasticizer. Thirty experiments were conducted using a central composite design to optimize the coating process and map the response surface to build a design space using either statistical least squares or a Bayesian approach. The response surface was fitted using a linear two-factor interaction model with spraying temperature, curing temperature, and curing time as significant model terms. The design spaces established by the two approaches were in close agreement with the statistical least squares approach being more conservative than the Bayesian approach. The design space established for the critical process parameters using small-scale batches was tested using scale-up batches and found to be scale-independent. The robustness of the design space was confirmed across scales and was successfully utilized to establish process signature for the coating process.