Validating Pegasus™ Prime virus membrane filters

What virus spike should I use?

The purpose of virus filter validation is to evaluate the removal of viruses during an effective scaled-down simulation of the process, to demonstrate the removal of virus under the conditions seen at process scale.

Selection of a virus spike level must allow for quantification of virus removal, but not at the expense of adverse impact on the scaled down process. At the point of virus filtration, many unit operations have taken place which will significantly reduce any additional contamination that could potentially be caused by upstream viral contamination. Over-spiking during virus validation risks introducing contaminants that would normally not be present at that process position and will impact the filterability performance and reduce the accuracy of the scale-down model. The basis for successful spike selection is to:

• Use the purest spike available (ultracentrifugation essential).
• Use the most sensitive assay technique (large volume assays).
• Spike only what is needed to measure your target log reduction value (LRV).

Selecting a spike

The preferred decision tree for spike selection can be seen in Figure 1. It is based on information that defines the minimum required spike and checks to ensure that the filterability of the scale down mimic is not impacted by this spike. Take care here to ensure a reasonable safety margin and make the most of expertise from your chosen virus validation laboratory by asking them to carry out calculations and assure your target LRV can be demonstrated.

Fig 1. Decision tree for virus challenge spike selection.

Example calculations for Figure 1 flow diagram

Table 1. Example: Murine leukemia virus (MuLV) testing of a product with good prior knowledge

Minimum target LRV	4.5	Chosen by the customer
LRV safety margin	0.5	Low risk due to existing data with this product
Dilution factor (logs)	0	No cytoxicity / interference
Limit of quantification (logs)	-1.2	Based on large volume assay of 50 mL and a 95% confidence interval (1)
Target spike (logs)	3.8	Sum of above LRV values
Target spike (virus titer/mL)	6.0 × 103
Minimum stock titer	3.2 × 106
Spike %	0.2	Target spike / minimum stock titer
Target throughput (L/m2)	1000
Expected stock titer (virus titer/mL)	1 × 107	For calculating likely fouling level
Total virus load (virus titer/m2)	1.9 × 1010	Target throughput × expected stock titer × spike % × 1000 mL/L

The example demonstrates viral clearance test design to high throughput with a more challenging virus such as MuLV. Figure 6 indicates that this level of load is relatively safe for filterability risk. A spiked filterability test would still be recommended, but in this scenario we are presuming prior knowledge with the product / virus combination and testing could go directly to validation.

Table 2. Example: Minute virus of mice (MVM) evaluation of cytotoxic product with limited testing experience

Minimum target LRV	5	Chosen by the customer
LRV safety margin	1	Based on risk assessment and high due to the limited experience of this product
Dilution factor (logs)	0.5	1 in 3 dilution required to remove cytotoxic effect
Limit of quantification (logs)	-1.2	Based on large volume assay of 50 mL (diluted from original filtrate sample as per above) and a 95% confidence interval (1)
Target spike (logs)	5.3	Sum of above LRV values
Target spike (virus titer/mL)	1.8 × 105
Minimum stock titer	3.2 × 107
Spike %	0.6	Target spike / minimum stock titer
Target throughput (L/m2)	750
Expected stock titer (virus titer/mL)	1 × 108	For calculating likely fouling level
Total virus load (virus titer/m2)	4.3 × 1011	Target throughput × expected stock titer × spike % × 1000 mL/L

In this scenario, an increase in flux decay is likely because a high spike level is required. The high spike level is due to the high safety margin, based on unknown risks of the virus stability and also the mild cytotoxic effects on the virus host cells requiring pre-dilution. Looking at the total virus load (Fig 3), this could potentially cause moderate issues with higher and non-representative flux decay in the scale down process simulation. Before running the test we would recommend a spiked filterability test to determine the effect of the spike on product filterability. It is also worthwhile revisiting the volume of sample assayed, options to purify the spiked test sample, and even the target LRV or safety margin.

An alternative approach to spike selection is to use existing data and experience to select a reasonable spike level known not to cause fouling in typical testing, as detailed in Figure 2. This approach includes more risk to throughput but is a faster and easier approach.

Fig 2. Decision tree for selecting a virus challenge spike based on prior data.

For a general guide as to what spike levels will have minimal impact on flow decay during a validation run, Figures 3 to 6 demonstrate the impact of 4 viruses commonly used in validation testing on a low fouling monoclonal antibody (mAb) solution:

• Figure 3: Minute virus of mice (MVM)
• Figure 4: Reovirus Type 3 (REO-3)
• Figure 5: Pseudorabies virus (PRV)
• Figure 6: Murine leukemia virus (MuLV)

Figure 3 to 6 show, at a throughput of 1000 L/m², the percentage flow decay that a ‘total virus load’ causes (data dots), compared to the ‘standard’ flow decay of just the clean mAb ‘unspiked control’ depicted as a dashed horizontal line in each graph. Different total virus loads at 1000 L/m² throughput were achieved by varying the added virus spike standard stock quantity. The general observation is that the lowest total virus loads we tested (between 10¹⁰ and 10¹¹) give filterability comparable to that of the unspiked control. Increasing total virus load, brings significant flow decay additional to that of the unspiked control, varying between the four different virus types and preparations.

All the spikes are standard preparations which have been ultracentrifuged (not gradient ultracentrifugation).
This is a guide only and is specific for the given test mAb, solution conditions and virus testing laboratory spike preparation. Variation in these results due to different mAbs, buffers and spike preparations is expected and the level of fouling could be both lower and higher. Spikes of 1% and higher can potentially be used, as demonstrated, but are often undesirable due to increased fouling of the test virus filter. Some virus testing laboratories offer ultra-high purification spikes; these can be even lower fouling and offer more flexibility in spike levels. When incorporating Pegasus™ Protect virus prefilters into your process, this can significantly reduce spike related fouling for some viruses (see Fig 3).

Fig 3. Effect of different spike levels of ultracentrifuged MVM on the filterability performance of a low fouling mAb solution (1.5 g/L mAb, 75 mM Tris, pH 7.5, 6 mS/cm) at various spike concentrations. Data points are 0.1%, 0.3%, 3% spikes from a stock titer of 7.7 log₁₀ virus titer/mL, tested to 1000 L/m².

Fig 4. Effect of different spike levels of ultracentrifuged REO-3 on the filterability performance of a low fouling mAb solution (1.5 g/L mAb, 75 mM Tris, pH 7.5, 6 mS/cm). Data points are 0.05%, 0.3%, and 3% spikes from a stock titer of 8.0 log₁₀ virus titer/mL, tested to 1000 L/m².

Fig 5. Effect of different spike levels of ultracentrifuged PRV on the filterability performance of a low fouling mAb solution (1.5 g/L mAb, 75 mM Tris, pH 7.5, 6 mS/cm). Data points are 0.05%, 0.3%, and 3% spikes from a stock titer of 8.1 log₁₀ virus titer/mL, tested to 1000 L/m².

Fig 6. Effect of different spike levels of ultracentrifuged MuLV on the filterability performance of a low fouling mAb solution (1.5 g/L mAb, 75mM Tris, pH 7.5, 6 mS/cm). Data points are 0.1%, 0.5%, and 2% spikes from a stock titer of 7.1 log₁₀ virus titer/mL, tested to 1000 L/m².

If you require further assistance or information to help with virus clearance validation of Pegasus virus filters.

https://www.cytivalifesciences.com/support/contact-us

Reference

1. ICH Q5A (R2), Viral safety evaluation of biotechnology products derived from cell lines of human or animal origin, Appendix 3.

CY49586