General Procedure
The laboratory scale experiments were designed to simulate the feeding of a perfusion fermenter with a fermenter feed media solution containing MMV as a contaminant. The intention of each laboratory scale experiment was to investigate the maximum process feed load capacity of an investigated filter when loading it with fermenter feed media spiked with MMV at a ratio of 1:250 (i.e., 0.4% spike load). The MMV-spiked, 0.2 µm filtered fermenter feed media solution (“spiked starting material”) was passed over a filter train consisting of one 0.1 µm prefilter followed by a virus removal filter (each with a stipulated pore size of approximately 20 nm, see Materials and Methods). During each investigation the impact of multiple interruptions during the loading process on virus clearance of a given filter was also investigated. This was done when the filtrate collecting flask was replaced by a fresh one (samples for virus titration were drawn from the filtrate and stored at ≤-60°C until the end of the experiment when all filtrate samples were titrated together) – during this activity filtration was interrupted intentionally for at least one hour. One of these intentional interruptions lasted at least 72 hours to represent the pause between the first filling of the fermenter and the build-up of the inoculum before the actual perfusion fermentation started. Filtrate collection flask replacement – and thus process interruption – took place approximately once a day. The overall number of interruptions per experiment ranged from 1 to 21 depending on filter performance. Filtration was continued until a flow decay of around 90% was reached or the duration of the experiment had exceeded 30 days (duration ranged from one to 34 days). Virus removal filtration was performed at constant transmembrane pressures (TMP, i.e., the pressure difference between feed and filtrate – which was at ambient pressure and, therefore, taken as 0.0 bar for calculations) over the virus removal filter. The TMP was set to 1.0 to 2.6 bar for each experiment (the maximum TMP set-point was determined either by the maximum value suggested by the filter manufacturer or by the maximum pressure allowed by the tubing used) and to 0.0 bar during the intentional interruptions. Filtration was performed under ambient temperature for all experiments. pH was measured for the spiked starting material and for each filtrate fraction.
Controls for aggregates (EMEA, (2010)) were performed for (a) the virus stock and (b) the spiked starting material matrix. For (a) the titers of the thawed virus stock and of the 0.1 µm filtered virus stock (which then was used for spiking) were compared – a titer reduction of less than 1 log10 is not considered to be significant according to regulatory guidance (EMEA, (1996)). For (b) the titer of the spiked starting material was compared to the titer of the pre-filtrate drawn during the first interruption.
In order to evaluate the influence of the freeze/thaw cycle, to which the filtrate fractions were subjected, on the virus titer of these samples, freeze controls were performed (EMEA, (1996)). Two freeze control levels were investigated: (c) “high titer” and (d) “low titer”. For (c) the titer of the spiked starting material (titrated immediately) was compared to the titer of the same matrix which had been stored at ≤-60°C throughout the experiment and titrated together with all other samples which had been stored frozen. For (d) the titer of the spiked starting material diluted with starting material to a theoretical MMV titer of 2.0 log10 [TCID50/mL] (titrated immediately) was compared to the titer of the same matrix which had been stored at ≤-60°C throughout the experiment and titrated together with all other samples which had been stored frozen.
In order to evaluate the influence of ambient conditions during an experiment on the virus titer, process controls were performed. To this purpose an aliquot of the spiked starting material was kept at ambient temperature in the biosafety cabinet where the virus filtration was performed, until the end of the NF procedure. At evenly spaced intervals samples were drawn and stored at ≤-60°C throughout the experiment and titrated together with all other samples which had been stored frozen. The virus titers of these samples were compared to that of the spiked starting material (titrated immediately).
Un-spiked control runs were performed for the virus clearance validation studies. All settings were identical to those for the virus spiked runs with the exception that no intentional pressure interruptions were performed. Samples were drawn for the analysis of glucose concentration and osmolality. Furthermore, a cell proliferation use-test was performed with the pooled filtrate in order to evaluate if the composition of the fermenter feed media was significantly altered during filtration with respect to supporting cell growth. The use test consisted of the analysis of the cell density, cell viability and cell productivity after a 3-day incubation period with the filtered media.