Sterilizing filtration—Principles and practice for successful scale-up to manufacturing

Abstract Filtration process design using micro-filtration membranes includes the proper sizing and scale-up of filtration units and requires an understanding of the effects of membrane fouling on filter capacity. This fouling is usually quantified by some fouling factor K . The filter capacity (related to surface area A ) also impacts the rate of filtration which is usually quantified by the initial normalized flux through the filter J 0 . For ideally scalable systems, the normalized flux J 0 should be the same for filtration across the same membranes with different surface area. Since fittings and device design can significantly affect filter resistance, scale-up factors must be taken into account in designing a scalable filtration step. In this manuscript we describe the principal approach and tools utilized to successfully transfer the sterile membrane filtration process for a pharmaceutical product from laboratory through pilot plant to manufacturing. Generalized models of pore fouling based on the standard blocking or complete blocking mechanisms are derived to simulate filtration of solutions at different filter set-up configurations using filters of different filter areas and construction. The analysis shows that specific resistance of filters resulting from differences in construction, and resistances of other flow details must also be considered in scale-up. Three different scales are compared to illustrate the application of the models to evaluation of filtration data and to the scale-up of filtration for a pharmaceutical solution. Millipore hydrophilic PVDF sterilizing filters were used at all scales. Scale-up criteria based on surface area (filter capacity) and flow rate (filtration duration) were employed at different scales. The case study results confirm that successful scale-up of filtration of pharmaceutical products requires understanding and correct prediction of both, the fouling characteristics (impacted mostly by properties and interactions between solution and membrane) and the flow characteristics (impacted by the filter construction and by the filtration system configuration).