Looking inside the ‘black box’: freezing engineering for assuring quality of freeze-dried biopharmaceuticals

The pharmaceutical industry is at a turning point of its history, shifting from a mass production of patented drugs to a more competitive, innovative and highly regulated market. In fact, the last few years has seen the rapid emergence of the industry of off-patented drugs and the increase of the requirements of quality, safety and efficiency set by regulatory authorities. In addition, in the next years, the production of personalized, ‘on-demand’, drugs will replace completely the old, batch, technologies with more flexible systems and continuous processes (Byrn et al. 2015). As the future of the pharmaceutical industry is continuous manufacturing, also freeze-drying of pharmaceuticals needs to be converted to continuous and integrated into the production chain. In this perspective, the design of freezing step plays a central role to reach the more stringent requirements of quality, homogeneity and standardization of freeze-dried products (Barresi et al. 2009; Pisano et al. 2016). Both trial-and-error and black box approaches, that are widely used to design freeze-drying cycles, become not sufficient anymore. A new mindset toward research and development is required to open and look inside the ‘black box’, in order to design freezing protocols that produce dried products with the desired final attributes, to predict its variability within a production and the losses due to undesired off-target products. Of course, the product structure is a key factor in the production of stable and valuable dried products, because it affects the stability of APIs, determines the drying and desorption rate and, finally, the appearance of the product, the residual moisture and the rehydration time. The objective of this work is to develop an innovative way for designing the freezing step in order to have a production of freeze-dried products with the desired attributes, and to have a good estimation of inter- and intra-vial heterogeneity. Our procedure consisted of coupling the distribution of nucleation temperature obtained from simple experimental tests with a detailed mathematical model of freezing. As case study, we compared the performances of two freezing protocols.