Viral filtration operations are evaluated and validated using scaled-down models of production processes. In a scaled-down study, a sample of representative process feed is “spiked” with a known quantity of virus to simulate a viral contamination.
Virus removal and hydraulic performance of the spiked feed across a filter are measured, and this data is used to determine the viral clearance capability of the filter as well as the filter size necessary to reach throughput targets at production scale.
Virus spiking can present a challenge because the virus stocks used often contain impurities derived from the cell cultures used to produce them. In fact, many virus preparations consist mainly of nonviral substances that are not relevant to the drug manufacturing process.
These impurities can, in some cases, interact with the filter or feed material to dramatically diminish filter performance, reducing throughput and producing results that do not represent the production-scale process. Furthermore, lot-to-lot variability of virus stock quality can hinder reproducibility and confidence in results.
Consequently, insufficiently purified and characterized virus preparations can cause validation failures.
Researchers at EMD Millipore have developed new methods that produce “ultrapure” virus stocks that have minimal levels of impurities, are high titer, consistent, and well characterized. These improved virus preparations were designed to facilitate validation of Viresolve® Pro Solution filters by minimizing the risk of spike-induced throughput reductions and maximizing consistency of results.
Furthermore, the ability to spike to higher titers will add value by enabling greater virus removal claims.
Virus Production and Characterization
Existing virus-preparation methods were rebuilt and optimized in order to deliver ultrapure stocks. New methods were developed for four viruses commonly used as part of a virus clearance validation panel (Table) The viruses under study were minute virus of mice (MVM), pseudorabies virus (PRV), reovirus 3 (Reo 3), and xenotropic murine leukemia virus (X-MuLV).
While each virus type needed its own specific approach, all methods followed these general principles:
- Virus was harvested from cell culture under serum-free conditions, which eliminated a major source of impurities.
- Virus is purified using a combination of common operations, proprietary products from EMD Millipore, and optimized buffers utilized in a specific sequence. Aspects of the methods are patent-pending and are designed to be readily transferable to our partner Charles River Laboratories.
- Virus stocks produced are high titer, low protein, and minimally aggregated.
- Virus stocks are consistent, and characterized in accordance with the recommendations of PDA® Tech Report #47: Preparation of Virus Spikes Used for Virus Clearance Studies.
TrueSpike™ Comparison Studies
MVM stocks were prepared by the TrueSpike method and compared to MVM preps made by conventional methods at a contract testing laboratory. The virus preparation methods were as follows:
- Crude: Clarified lysate from infected cells
- Q-membrane: Virus purified by bind/elute chromatography
- Ultracentrifuged: Virus pelleted by ultracentrifugation
- TrueSpike: Serum-free virus purified by ultrafiltration, ultracentrifugation, and flow-through chromatography
The protein content of the different virus-preparation methods is shown in Figure 1. Nonviral proteins present in virus preparations may cause filter fouling. Analysis of the protein content of the various preps shows that TrueSpike virus contains much less protein than conventionally purified stocks.
The ratio of protein to infectious virus (pg protein/PFU) generally correlates well with spiking hydraulic performance, with low values being most favorable.
Figure 2 shows the spiking hydraulic performance of the various preparations methods. A 9 g/L monoclonal antibody feed was spiked with MVM and processed using Viresolve Pro Solution. The conventionally purified virus preparations caused filter fouling and a dramatic loss of throughput at spike levels of 104–105 PFU/mL.
However, TrueSpike MVM had no negative impact at spike levels as high as 106 PFU/mL.
The complete panel of TrueSpike viruses (MVM, X-MuLV, Reo3, and PRV) was tested for spiking impacts using several antibody feeds (Figure 3). SeraCare™ IgG (0.1 g/L) was spiked with viruses from the TrueSpike panel and measured for hydraulic impact in filtration runs using Viresolve Pro Solution.
Standard, less purified X-MuLV was included for comparative purposes. Spike levels (TCID 50/mL) were as follows: TrueSpike MVM, 2x106; TrueSpike X-MuLV, 1x105; TrueSpike Reo3 2x106; TrueSpike PRV 2x106; standard X-MuLV 1x105.
The target was for virus spiking to cause less than a 10% reduction in throughput. This target was achieved for all four viruses in all antibody feeds tested.
The TrueSpike virus preparations described in this article contain mostly virus, minimizing the potential for filters to be fouled by virus stock impurities. Consistency of the virus stock enables reliable filtration performance without the need to prescreen or cherry-pick virus lots.
These highly purified, well-characterized virus preparations were developed to ease validation of filtration by minimizing the risk of spike-induced throughput reductions and maximizing consistency of results.