Lyophilization scale-up to industrial manufacturing: A modeling framework including probabilistic success prediction

Scaling-up a lyophilization cycle to the manufacturing scale successfully at the first pass is of crucial importance to the bio-pharmaceutical community in order to save expensive and sparsely available drug products for clinical and industrial manufacturing. The constraints on time, cost and energy consumption during process development are prohibitive to performing multiple experimental studies at the manufacturing scale. Most process analytical techniques to obtain cycle data are often not available at the manufacturing scale due to sterility concerns with intrusive tools and experimental uncertainties at industrial manufacturing scales. Modeling techniques offer an attractive alternative solution under these circumstances to gain knowledge about the manufacturing scale equipment limitations and predict the probability of success prior to performing experimental trials in order to minimize the risk associated with scale-up. In this paper, we present a detailed characterization of equipment capability curves for lyophilizers across laboratory, pilot and manufacturing scales using Computational Fluid Dynamics (CFD) modeling and highlight the flow features in manufacturing-scale equipment with different geometric attributes. We present the equipment, process and product parameters which determine the outcome of the cycle and develop guidelines for robust scale-up practices from the laboratory and pilot scales to the manufacturing scale using vial heat and mass transfer modeling. We present examples of cycles which would seem to scale-up to the manufacturing-scale successfully using a deterministic model but would have a high probability of failure when process excursions, deviations, and input uncertainties are accounted for by applying a Monte-Carlo based probabilistic model. We demonstrate the method to reduce the failure probability and de-risk the scale-up of such processes. We also present the significant reduction in primary drying time that can be achieved by implementing a lyophilization recipe with varying setpoints of chamber pressure and shelf temperature for the primary drying stage in a manufacturing-scale lyophilizer. All the observations from our modeling analyses and example studies indicate that CFD simulations in combination with deterministic and Monte-Carlo based probabilistic vial heat and mass transfer modeling would significantly improve the success of scale-up to industrial lyophilized drug manufacturing.