(682c) A Framework for Using a Well Characterized Performance Variability In a Virus Filtration Process towards a Highly Consistent, Economical Unit Operation

Virus filtration is the most common orthogonal step to clear virus in downstream processes for monoclonal antibody production in mammalian cell cultures. Virus clearance is required by regulation to ensure safety of the biological drug. As parvovirus ( 20 nm) needs to be removed yet monoclonal antibody ( 150 kd) needs to be recovered in the filtrate, several types of parvovirus filters are commercially available that achieve this objective. 2 main classes of parvovirus filters exist: low flux filters that are often sized based on flux and high performance filters that are sized based on a capacity and flux component. Flux for the second type of filter is often higher.

As virus filtration needs to be implemented in manufacturing, a robust process needs to be implemented that takes into account various possible sources of variability. QbD approach or other process characterization can be used to define performance differences due to changes within your process window of feed, process and membrane device. This paper will describe the impact of these changes from average performance from a process development perspective. Additionally, this paper will address how a partial pore blockage model in combination with range of input variables can be used to predict the manufacturability of a virus filtration step from a quality as well as from a economic perspective. More advanced models or actual filtration data can be used through a similar frame work.

The learnings of both process development and manufacturing view will finally be translated into some direction towards development of robust virus filtration steps.