Allegro™ Connect virus filtration system: Automated leak and filter integrity testing, and process control

All flow kits were installed and operated on the system in accordance with the Allegro™ Connect virus filtration system Operating Instructions.

System and flow kits

All flow kit sets selected incorporate four prefilters and two virus filters. This configuration would not normally be used with filters smaller than 508 mm (20 in.), with more compact options available. However, we used a higher number of small filters to test a more complex setup, which represented a greater challenge for the tests. This allowed us to demonstrate the integrity of the most complex flow kit and show the ability to separately integrity test two virus filters in series.

Fig 1. Allegro™ Connect virus filtration system configuration with four virus prefilter capsules and two virus filter capsules.

Some of the main equipment used in this study were:

• Allegro™ Connect virus filtration system: PLC 208 VAC, software automation
• Pegasus™ Protect membrane in 254 mm (10 in.) Kleenpak™ Nova capsule (in-line style)
• Pegasus™ Prime membrane in 254 mm (10 in.) Kleenpak™ Nova capsule (in-line style)

Leak test

We installed flow kits according to the instructions in the standard installation phase on the system. Once installation was complete, the recipe continued to a flow kit leak test phase. Prior to water or product commitment, the Allegro™ Connect virus filtration system activated the leak test function on the Palltronic Flowstar IV filter integrity test instrument at 1.2 bar (17.4 psi, 0.12 MPa) flow kit pressure for 10 min. The test is capable of measuring flow rates between 0.1 and 1000 mL/min using volume-dosed flow measurement technology.

Forward flow integrity test

Before pre-use filter integrity testing, we carried out automated standard preparation phases to prime and flush the system and filters. All phases complied with the specific slow filling of water and flushing at the required backpressure, as per the instructions for use for the Pegasus™ Prime virus removal filter.

After full filter processing, the standard buffer chase phase carried out a water flush to prepare the system for post-use integrity testing.

We carried out both integrity tests in series using the same standard recipe phase. The Allegro™ Connect virus filtration system activated the forward flow integrity test function of the Palltronic Flowstar IV filter integrity test instrument at 4.15 bar (60.2 psi, 0.42 MPa) and 10 min for each filter in series, to generate individual integrity test results for each virus filter installed.

Processing control

To demonstrate the processing control of the system, we filtered 400 L of protein solution through Pegasus™ Protect virus prefilter capsules and Pegasus™ Prime virus removal filter capsules. We used a human immunoglobulin G (IgG) solution in phosphate buffered saline (PBS) to foul the virus filters while the pump control was set to either constant differential pressure control or fixed flow control. We used the standard phases for product filtration and buffer flush to transition smoothly by first opening the buffer valve and then, after a user defined time (6 s), closing the product valve. Throughout the valve transition, the pump control remained active and unchanged to allow effective transition without significant deviation from the set-point.

Proportional–integral–derivative (PID) control parameters were set individually for each control strategy, with a deadband of 0.05 bar (0.73 psi, 0.005 MPa) and 0.05 L/min either side of the target values of 2.10 bar (30.5 psi, 0.21 MPa) and 2.50 L/min, where control corrections were not applied.

Product recovery

Further to the IgG filtration by differential pressure control experiment, we collected the buffer chase material in 7.5 L aliquots, equivalent to the system and filter hold-up volume. We measured the concentrations of the IgG in the 400 L initial feed and the aliquots to demonstrate the capability to rinse out and recover product. IgG concentrations were determined using ultraviolet-visible (UV-vis) spectroscopy at a wavelength of 280 nm.

Flow kit differential pressure distribution

We recirculated a dilute salt solution through the flow kit to demonstrate even flow distribution between the two virus filter positions. Breaking the flow path at any point to measure relative flow through, the two virus filter positions would not offer a true representation of the fluid dynamics during processing. Instead, we replaced the 254 mm (10 in.) capsule filters with restricted tubing with pressure sensors at either end as detailed in Figure 2. Equal measured pressure equates to equal flow. Each restricted tubing set gave a resistance equivalent to a 762 mm (30 in.) Kleenpak™ Nova capsule with Pegasus™ SV4 virus removal filter membrane and we carried out the testing at a typical processing flow rate for two such filters of 7 L/min at 2.1 bar (30.5 psi, 0.21 MPa) differential pressure.

To ensure that the results were not influenced by variation in the equipment, we tested each combination of the two restricted tubing sets and the two pressure sensor sets on both virus filter position 1 and position 2. Average differential pressures were calculated to demonstrate the flow distribution between the two virus filter positions independent of any variation from the pressure sensors or the restricted tubing.

Fig 2. Schematic of the experimental setup during equal flow kit differential pressure testing.

The built-in Palltronic Flowstar IV instrument allows flow kit leak testing on the system as part of a fully automated process, without any need for a separate integrity test instrument skid or any manual connections post-installation.

Following installation, the flow kit leak test was carried out dry to make sure there is a leak-free process before any product is committed to the process, and before the flow path is wetted with water. The status and results of the leak test are displayed on the human machine interface (HMI) screen as the test progresses. An example data set is displayed in Figure 3, demonstrating the highly accurate pressure control and typical decrease in flow rate before stabilization towards the end of the test. The pass result confirms the leak-free status of the installed flow kit.

Fig 3. Example of a flow kit leak test performed post-installation by the integrated Palltronic Flowstar IV instrument.

We performed a filter integrity test to ensure the integrity of the virus filters and the virus retention performance at the point of use. This was done after priming and flushing the filters (pre-use), and after the processing, buffer flush, and water flush (post-use). The progress is again monitored on the HMI screen during the test. Each of the two virus filter positions can be integrity tested in parallel or in series, one after another.

Figure 4 shows a summary of the filter integrity test results generated in series, both pre-use and post-use. The test was done with 254 mm (10 in.) Kleenpak™ Nova capsules with Pegasus™ Prime virus removal filter membranes. The pass results not only confirm the virus retention integrity of the filters tested, but also the consistent success of the fully automated, slow-fill priming and filter flushing (with back pressure) required to effectively wet the virus filters.

Fig 4. Filter integrity test result summary for 254 mm (10 in.) Kleenpak™ Nova capsules with Pegasus™ Prime virus removal filter membrane in both filter positions.

The importance of differential pressure and flow as critical parameters to maintain virus filtration performance is well known in the industry. Effective virus filter processing must allow control of the differential pressure across the virus filter and flow through the virus filter throughout the product filtration and buffer chase. A critical stage for this is the transition from product filtration phase to buffer chase phase, which can cause deviations if not properly controlled. Typical transitions during product processing are detailed in Figure 5 (differential pressure control) and Figure 6 (fixed flow control). The Allegro™ Connect virus filtration system approach to virus filter transitioning allows virtually seamless switching between product protein solution and buffer.

Fig 5. Differential pressure control focusing on the final 12 min of a longer human IgG filtration demonstrating a smooth transition to the buffer flush phase, avoiding deviations in the critical pressure differential processing parameter.

Fig 6. Fixed flow control focusing on the final 9 min of a longer human IgG filtration demonstrating a smooth transition to the buffer flush phase, avoiding deviations in the critical flow rate processing parameter.

During buffer chase under differential pressure control the largest flow variation occurs before the transition has taken place. The differential pressure drifts down during the buffer chase as the more viscous protein solution is flushed out of the filters but remains within the PID deadband control limits. This drift could be avoided by tightening the deadband if required. The setting decision is a balance between adherence to the exact pressure differential and both greater flow variation and more frequent pump adjustments.

Within the fixed flow control buffer chase phase there are two small fluctuations in flow and pressure. These are not related to the timing of the valve changes during transition but are related to manual adjustments of the product line during collection of the filter permeate which would not typically occur during processing. This demonstrates the sensitivity of the flow and pressure to slight manipulations of the outlet tubing and reinforces the tight control of these parameters during the rest of the process run.

After processing 400 L of IgG during differential pressure control processing, we collected the buffer flush in 7.5 L aliquots equivalent to one hold-up volume of the flow kits and filters used during the test. Table 1 summarizes the concentrations measured in each of five hold-up volumes of buffer flush. The data demonstrate that more than 100% of the IgG present in one hold-up volume of feed is recovered during just two hold-up flushes, confirming that the system does not hinder product recovery. This demonstrates that there is also rapid recovery of protein from material held up inside the filter membrane and capsule during processing.

The concentration of IgG in the buffer flush drops quickly with each subsequent buffer flush as expected for virus filtration processing and the Allegro™ Connect virus filtration system displays its ability to allow effective product recovery. Choice of the buffer flush volume is a product-specific and user-preference choice, but we recommend typically either two or three hold-up volumes of buffer flush. The fourth and fifth hold-up volumes combined only added an additional recovery of < 0.05% of the original 400 L feed of igG quantity.

Table 1. IgG Concentration during buffer flush

The measurement of flow distribution between the two virus filter positions cannot be reliably measured by breaking the flow paths to collect liquid. This disrupts the overall fluid dynamics in the flow kit and is influenced by multiple factors other than the true flow distribution. Based on the direct relationship between flow and pressure predicted by the Darcy-Weisbach equation and other modifications of this relationship, equal pressure must indicate equal flow. Therefore, we carried out testing measuring the individual differential pressure across a tubing restriction in each virus filter position created to simulate two identical virus filters.

Figure 7 visualizes the distribution in average differential pressure and therefore the distribution in flow between the two virus filter positions. This is independent of any variation caused by the two different restrictions, and two different pressure sensor sets used during testing, since results were averaged from all different combinations of this equipment on the two virus filter positions. The differential pressure variation between the two pressure sensors sets was < 0.1% (Fig 7C) and between the two restricted tubing sets was < 0.3% (Fig 7B), demonstrating a consistent measurement and the potential to demonstrate equivalence. Chart A in Figure 7, shows that there was a 1.1% higher differential pressure across virus filter position 2 compared to position 1. Based on empirical measurements of the flow and pressure of the tubing restriction, this equates to a 1.1% higher flow and therefore a negligible 0.5% higher throughput compared to the total throughput measured for both virus filters.

Based on the equal differential pressure distribution described, we can reliably infer that there is equal flow distribution across the two virus filter positions.

Fig 7. Average differential pressure (bar) measurements: (A) across a flow restriction in the two virus filter positions during 7 L/min flow; (B) across the different tubing restrictions; and (C) across the pressure sensor sets.