Figure 5. Virus inactivation pH trends. (a) Valve switch to in-line with process during startup. (b) pH trends for sub-run 3. NOR is normal operating range defined by the process description. (c) Virus inactivation residence time trend during sub-run 3.
Residence time through the SEC column was also controlled to a setpoint through feedback control. Rather than simply setting the pump to a fixed flow rate to achieve the required residence time, the last pump in the process changed flow rate to drive the overall mass flow rate of the process to match the flow rate required to achieve the target residence time in the SEC column. This strategy is required when running a truly integrated process as fixed flow rates on any step will inevitably result in vessel over/underflow events due to imperfectly calibrated pumps.

Polishing

A flow-through anion exchange membrane polishing step was performed for this run. A second polishing step was not required to meet product quality specifications. A 0.2µm filter was installed before and after the Pall Mustang Q AEX membrane, which was immediately prior to the final collection bag. The filter train, which consisted of the AEX membrane and two 0.2µm filters (pre and post membrane), was swapped with a fresh filter train once the capacity on either the filter or membrane had been reached. The filters were sized in order to match their capacity with the AEX membrane and were replaced at the same time. The throughput was automatically determined by the PAK system, which alerted the user when a swap was required.
The filter train was swapped once during the process, although ideally, two swaps would have been performed. Unfortunately, the COVID-pandemic impacted global supplies and limited the availability of single-use materials like filters and membranes, thus only one swap was able to be performed. This resulted in an intentional overloading of the second Mustang Q to prevent cutting the run short. The first train was loaded to 14.2 liters while the second was loaded to 30.4 liters. Aside from the consequence of limited membrane area, the polishing step ran as expected.

Product Quality

The continuous purification process performed as expected and demonstrated acceptable product quality results. DNA, HCP, and monomer purity measurements were performed. Samples were taken from the harvest vessel, protein A product vessel, virus inactivation product flow kit, and the final collection bag. The final collection bag contained all the material collected from a single filter train. Product was diverted to a new collection bag when the filter train was swapped. The final AEX product was within acceptable ranges for percent monomer and host cell protein on both days measured. DNA levels were acceptable on day 9 but out of range on day 14. This was attributed to the intentional overloading of the second AEX membrane as described above. Overall, product quality was comparable to the batch downstream process, with the exception of the day 14 AEX DNA datapoint. Results can be seen in Figure 6.