Incorporating a prefilter in virus removal filter validation

In some cases, aggregates removed during adsorptive pre-filtration can re-form over time. When the pooled prefilter filtrate is spiked and challenged to the virus filter this can lead to a reduction in throughput.

At best, this approach models the process-scale filtration less effectively; at worst, it fails to validate the target process volume per filter. You should always conduct a decoupling study to assess how prefilter decoupling affects throughput performance, before starting a virus validation study and selecting a validation method. Figure 2 illustrates the recommended test comparison and Figure 3, typical example data. The decoupled prefilter must be run at a flowrate as close to that of the coupled prefilter / virus filter as possible in order to keep an equivalent residence time. Depending on the product stability, dropping the prefilter filtrate from a large height and splashing into the collection vessel can also significantly affect decoupling; directing flow to the bottom of the collection vessel with appropriate downstream tubing is best practice. Avoid over-challenging the prefilter to prevent overload. Once you have the data, use the decision tree in Figure 1 to assess whether prefilter decoupling will succeed.

Fig 2. Decoupling study to determine if the time between prefiltration and virus filtration influences the filterability.

Fig 3. Graph displaying typical throughput data for coupled and decoupled prefiltration.

Figure 4 illustrates the preferred approach to validate virus filters. First, prefilter the test solution, then spike the pooled, prefiltered solution with virus before challenging only the virus filter. As with the decoupling study, you must match the prefilter flow rate to the flow rate of the coupled prefilter / virus filter combination as closely as possible.

Fig 4. Set-up for validation with a decoupled prefilter.

Figure 5 shows how to evaluate the retention of the virus filter by comparing the performance of the coupled prefilter and virus filter with that of the prefilter alone. Again, you should match the prefilter flow rate as closely as possible to that of the coupled prefilter/virus filter combination. Although this test reduces accuracy due to errors from two separate challenge calculations and the potential prefilter variation, it most accurately reflects the process scale set-up. This method allows you to test virus removal in solutions that show significant decoupling, but it does not work well for enveloped viruses such as murine leukemia virus (MuLV) as the prefilter may remove too much of the virus spike. For minute virus of mice (MVM), Pegasus™ Protect filters have shown virus reduction up to 1.5 logs, although this varies depending on the test solution properties and specific process parameters, such as overall throughput and is typically less than 0.5 logs.

Fig 5. Coupled prefilter vs prefilter only to determine the contribution of the prefilter alone to the total virus reduction.

In circumstances where decoupling the prefilter significantly reduces throughput, and the prefilter removes too much of the virus spike to allow coupled prefilter vs prefilter only testing, then in-line spiking offers the best alternative. This approach addresses both decoupling and excess virus removal. The test solution flows through the prefilter and filter in series. Between the filters, a small percentage of virus spike is injected into the test solution flow, mixed, and then challenged to the virus filter (Fig 6).

Fig 6. In-line mixing set-up.

A study was carried out to demonstrate decoupled and coupled virus validation procedures. A peristaltic pump (Masterflex) driven filterability control system (Pendotech) maintained a constant pressure 2.1 bar (30 psi, 0.21 MPa) to the test filters (Fig 7). By using two pump heads on a single drive it was possible to match the flow rate of the Pegasus™ Protect prefilter only test to the coupled Pegasus™ Protect prefilter and Pegasus™ Prime virus filter (Fig 8). Pump driven control systems offer optimal control over the prefilter-only flow rate to ensure the correct residence time, however contact your selected virus test laboratory to ensure that the primary data capture during viral clearance studies complies with GLP. A humanized IgG-1 mAb was purified by protein A and cation exchange chromatography to represent an example input stream to the virus filter operation.

The chosen mAb test solution (8.1 g/L, 75 mM Tris, pH 7.3, 5.7 mS/cm) shows lower throughput performance when tested with the prefilter decoupled (Fig 9). This would prove problematic if the target process throughput at full scale were 750 L/m² or higher and demonstrates the risks of not addressing decoupling issues before the test takes place. Many feed solutions can also display a more dramatic impact on filter capacity upon decoupling of the prefilter. The chosen mAb also shows a significant challenge to the Pegasus™ Prime virus filter, causing high flux decay. In addition to this, a 25 L/m² post-use buffer flush was carried out after a 1 h pressure interruption. These are all known risk factors for the maintenance of high viral retention.

The data in Table 1 demonstrates how the different approaches to virus filter validation with a prefilter successfully achieve high parvovirus removal with Pegasus™ Protect prefilters and Pegasus™ Prime virus filters.

Fig 7. Example of validation equipment.

Fig 8. Flow rate of coupled Pegasus™ Protect / Pegasus™ Prime filters compared to flow rate of Pegasus™ Protect prefilters only, demonstrating that the prefilters used in parallel experience the same flow rate and flow decline irrespective of the presence (or not) of a subsequent virus filter. This provides additional justification that the virus reduction of the coupled prefilter can be estimated from the prefilter only retention data and therefore the retention performance of the Pegasus™ Prime virus filters can be evaluated.

Fig 9. Flux profiles during the virus filter challenge test comparing coupled and de-coupled tests.

Table 1. Pegasus™ Prime retention data for MVM in an 8.1 g/L mAb solution. 2.1 bar (30 psi / 0.21 MPa). A 1 h pressure interruption was incorporated into each test, followed by a 25 L/m² post-use buffer flush. Various lots of filters were utilized to demonstrate consistency.

* Virus detected below the limit of quantification.
† Coupled prefilter results evaluated by subtracting the LRV of a parallel Pegasus™ Protect filters only test from the Pegasus™ Protect prefilter and Pegasus™ Prime filter combination.