Virus Removal by Filtration: Points to Consider

ABSTRACT

Virus safety of biotech- and plasma-derived therapeutics is ensured through complementary manufacturing and quality control measures that include the control and monitoring of raw materials, the validation and implementation of effective virus clearance technology, and the monitoring of final filled product for the presence of virus. Virus filtration, which is considered a robust and effective virus clearance technology, is a common unit operation in the manufacture of biologicals. In this article, we review the points that need to be considered when selecting a virus retentive filter. The areas covered include regulatory considerations; selecting, optimizing, and validating a virus filtration step; and process scale implementation-areas that are of critical importance to users of virus filters.

Some biological therapeutic products are produced using mammalian cell lines or human plasma. The risk of contamination with either known or unknown viruses in these products has been demonstrated; hence, regulatory agencies have mandated that manufacturers evaluate the risks of virus contamination and take necessary measures to mitigate these risks. In addition to ensuring the purity of source materials, manufacturers are encouraged to institute steps in the purification process that will clear endogenous and adventitious viruses. Filtration has been successfully used in numerous processes as a robust step for virus clearance.

To implement virus retentive filtration successfully within a process, several points should be considered. These can be broadly categorized as follows:

• Regulatory considerations

• Process considerations

• Virus filter-related considerations

This article reviews each of these categories and suggests factors to consider when selecting and implementing a virus filter.

REGULATORY CONSIDERATIONS

To ensure the virological safety of biological therapeutics, regulatory guidance advocates virus control at various stages of the drug manufacturing process. Specifically, manufacturers should (a) select and test source materials for the absence of viruses; (b) test the capacity of the production process to remove or inactivate viruses; and (c) test the product at appropriate stages of production for freedom from detectable viruses.¹

Regulators require that an overall safety margin, such as <1 virus particle per 10⁶ doses, be used to demonstrate the virus safety of the manufacturing process. The drug manufacturer is required to quantify the virus "load" in the process. For biotech products derived from murine cell lines such as Chinese hamster ovaries (CHO) and nonsense oligonucleotide (NSO), this typically translates to ~12–18 log₁₀ clearance for endogenous retroviruses and ~6 log₁₀ removal for adventitious viruses. The following is a composite summary of regulatory guidance on virus clearance, with an emphasis on how the guidance relates to virus filtration. ^2–5

• The US Food and Drug Administration (FDA) requires that manufacturers of biotech products that use murine cell lines are to demonstrate the clearance capability of their manufacturing processes with one relevant retrovirus (murine retrovirus) before starting Phase I studies.

• Regulatory agencies in Germany and France require that manufacturing processes be evaluated to clear non-enveloped parvoviruses in addition to retroviruses.

• Before marketing authorization, manufacturers are required to assess clearance of multiple model and relevant viruses in their manufacturing processes.

Depending on regulatory requirements, one needs to consider if the process requires retrovirus clearance, or retrovirus and parvovirus removal. Virus clearance filters are broadly classified into two categories:

• Filters that provide >4 or >6 log₁₀ removal of large viruses, typically 80–100 nm endogenous retroviruses.

• Filters that provide >4 log₁₀ removal of small and large viruses (larger than 18–24 nm parvoviruses).

Commercially available virus filters are listed in Table 1.

Table 1. Commercially available virus filtration products

Table 1. Commercially available virus filtration products

PROCESS CONSIDERATIONS

After regulatory requirements have been addressed, it is time to look at the protein purification process, not only at the current scale but also at the potential manufacturing scale, to ensure that virus filtration is implemented at the best location in the process.

Parvoviruses have a diameter of ~18–26 nm, but a typical monoclonal IgG antibody has a hydrodynamic diameter of ~8–12 nm. To achieve >4 log₁₀ retention of the viruses and >99% recovery of the protein, parvovirus filters are required to have a very narrow pore size distribution. They are, therefore, generally sensitive to the presence of impurities in the feed solution. Thus, optimizing a virus filtration process involves evaluating the effect of a variety of process parameters to arrive at conditions that will ensure a robust, consistent, economical, and scalable operation.^6–9
 

Impact of Location in the Downstream Process Train

Typically, a normal flow virus filtration step can be implemented at any one of several points in a given downstream process. As shown in Figure 1, for a typical monoclonal antibody process, the virus filtration step can conceivably be implemented at three locations in the downstream purification process: following the low-pH inactivation step, following the intermediate chromatographic operation, or following the final chromatography step. Because protein concentration, impurity concentration, and process volumes can vary throughout the downstream process, the actual filtration requirements-including the required filter area-are highly dependent on where in the process the virus filtration step is located.

Figure 1. Virus filters can be located at various points in a typical protein purification process. Options for locating a virus filtration step are 1) following the low pH inactivation step 2) Following an intermediate column chromatography step and 3) After the final column chromatography step.

Figure 1. Virus filters can be located at various points in a typical protein purification process. Options for locating a virus filtration step are 1) following the low pH inactivation step 2) Following an intermediate column chromatography step and 3) After the final column chromatography step.

Figure 1. Virus filters can be located at various points in a typical protein purification process. Options for locating a virus filtration step are 1) following the low pH inactivation step 2) Following an intermediate column chromatography step and 3) After the final column chromatography step.

Impact of Feed Concentration

Feed solution concentration can affect the virus filtration process by reducing product throughput. The level of the impact will depend on the interaction between the filter and the components in the solution being filtered. In general, higher protein concentrations reduce the average process flux through the virus filter. The dependence of average process flux on changes in protein concentration is both protein specific and virus-filter specific. The effect of increasing filter capacity and flow at lower product concentrations is offset by an increase in process volume as the product is diluted. The interplay of these two competing effects can often result in an optimum feed concentration that minimizes the required filtration area. Figure 2 shows the impact of the feed concentration on the filtration area needed to process the same quantity of protein in a fixed time.

Figure 2. The effect of protein concentration on filter area needed can be significant. In this case, the optimum concentration is between 8 and 10 g/L. The optimum concentration can vary, depending upon the protein purity and the buffer conditions.

Figure 2. The effect of protein concentration on filter area needed can be significant. In this case, the optimum concentration is between 8 and 10 g/L. The optimum concentration can vary, depending upon the protein purity and the buffer conditions.

Figure 2. The effect of protein concentration on filter area needed can be significant. In this case, the optimum concentration is between 8 and 10 g/L. The optimum concentration can vary, depending upon the protein purity and the buffer conditions.

Impact of Prefiltration

Prefiltration of the feed solution can have a dramatic impact on filter performance. Prefiltration is targeted to remove various impurities or contaminants, such as protein aggregates, DNA, and other trace materials. Although larger impurities can be removed by prefiltering using 0.2 μm or 0.1 μm microfilters, smaller impurities, such as protein aggregates that may be only marginally larger than the protein product, are not easily removed using size-based removal methods. Prefiltration through adsorptive depth filtration has been observed to provide significant protection for certain virus removal filters.⁷ The impact of prefiltration can be dramatic, with up to a 10-fold reduction in required filter area.

Impact of Hold Times and Freeze-Thaw Cycles

Some proteins exhibit time-dependent aggregate formation or will form low concentrations of aggregates when subjected to a freeze–thaw cycle. If a hold step or a freeze-thaw cycle is expected to be a part of the process, it is important to evaluate the effects of these during filter optimization. Furthermore, although the actual purification process may not have a freeze–thaw step, feed samples required for virus retention testing are often conveniently submitted in a frozen form due to material stability considerations.

Process Time

Parvovirus filters can be broadly classified into two groups: those with high protein flux but low-to-moderate volumetric capacities, and those with high protein capacities but low-to-moderate protein fluxes. Both of these filters have advantages and disadvantages. For example, for a process that must be completed in a short amount of time (2–4 h), high-flux filters require less filtration area than high-capacity filters do (Figure 3). On the other hand, when processing times are extended to 6 hours or more, high-capacity filters may be more economical. (This example assumes that all filters provide necessary virus clearance.)

Figure 3. The graph shows the filtration area in m2 needed to filter 1,000 L of protein as a function of process time through different parvovirus filters. For short processing times (2-4 h), high flux filters generally require less filter area than high capacity filters do.

Figure 3. The graph shows the filtration area in m2 needed to filter 1,000 L of protein as a function of process time through different parvovirus filters. For short processing times (2-4 h), high flux filters generally require less filter area than high capacity filters do.

Designing Virus Retention Qualification Studies

Spiking studies should be designed to reflect the virus clearance capability of the process-scale unit operation.² Therefore, the level of purification of the scaled-down version should represent the production process as closely as possible, by reproducing the key operating parameters that have an effect on purification and on virus clearance. The critical operating parameters will vary depending on clearance technologies (due to the different mechanisms of action). For filtration, scale-down will focus on such parameters as:

• Identity between process and retention study filter media

• Flow rate or pressure

• Ratio of volume processed to filter-surface area (V/A) or the extent of flow decay

• Equivalence of process and retention study feedstocks

• Product yield and quality.

Typically, the study sponsor may rely on the filter manufacturer to supply data required to support the claim of scale-down validity.

Although not specified in regulatory guidance, some manufacturers choose to use worst-case process-variable settings in the design of the spiking study.

Appropriate guidance on how to plan and carry out virus retention qualification studies is available from regulatory agencies.³ Documents such as the Parenteral Drug Association Technical Report 41 provide guidance on how to determine the quantity of virus spike needed to achieve necessary log reduction values (LRV).¹⁰

Many parvovirus filters exhibit decay in virus LRV with filter plugging.^11,12 Furthermore, impurities present in virus preparations can result in premature plugging of some virus retentive filters.¹³ These points should be considered when designing virus retention studies in order to obtain the desired throughput and the desired virus reduction. When the desired throughput or LRV cannot be obtained because of issues with the protein solution or virus spike quality, it may be useful to discuss alternative spiking methods with the appropriate regulatory agencies. A sample decision tree for designing a virus retention qualification study is shown in Figure 4.

Figure 4. Decision tree for designing a virus retention qualification study

Figure 4. Decision tree for designing a virus retention qualification study

VIRUS FILTER–RELATED CONSIDERATIONS

Implementation

After the virus clearance step has been optimized and virus retention studies completed, an implementation strategy is required for robust process operation. After determining the filter capacity (L/m²) required for a process during process simulation, process scale-up, and virus retention studies, the filter area required for processing a given batch volume can be calculated. Various filter configurations are made available by manufacturers to facilitate large-scale implementation. When multiple filter modules are required to process a given batch volume, the modules may be installed in parallel within a multi-round housing.

Operating Sequence

A typical sequence of operations in a virus filtration process includes the following steps. (Some of these will be discussed later in more detail.)

• Filter installation, flushing, and water permeability measurement

• Sterilization and sanitization

• Pre-use integrity testing

• Buffer pre-conditioning and permeability measurement

• Processing and product recovery

• Post-production integrity testing

Sanitization and Sterilization

In a typical downstream purification process, virus clearance filters are used downstream of a chromatography column and upstream of an ultrafiltration/diafiltration step, neither of which is considered an aseptic operation. However, there appears to be an industry trend to sanitize or sterilize the virus filter to reduce the bioburden. Some virus filters are available pre-sterilized and, therefore, will eliminate the sanitization step. It is important to ensure that the filters are compatible with a sanitization or sterilization method that is likely to be implemented at manufacturing scale. Furthermore, it is important to ensure that process steps used during large-scale processing are also carried out during the scaled-down virus retention studies.

Integrity Testing

To ensure that virus clearance is consistent with results obtained during virus retention studies, it is recommended that filter integrity be checked both pre- and post-use.^14,15 To facilitate this, filter manufacturers have developed a variety of destructive and non-destructive physical integrity tests that are related to virus retention. Ultimately, the objectives of properly designed physical integrity testing are threefold:

• To confirm that the virus removal filter is properly installed

• To confirm that the filter is free from gross defects and damage

• To confirm that the filter removes viruses consistent with both manufacturer's specifications and end-user virus retention studies.

To satisfy these requirements, a series of tests is needed to confirm filter integrity. Some of these tests, typically carried out by the end-user, are better suited for confirming proper installation and for confirming the absence of gross defects. Other tests, generally performed by the filter manufacturer, may be better suited for detecting subtle changes in filter pore size distribution. A more detailed summary of the various tests can be found in Parenteral Drug Association Technical Report 41.

Processing

It is important to ensure that conditions used during processing are within the parameters used during the scaled-down virus retention assessment studies. Because most parvovirus retentive filters exhibit a decline in virus retention ability with flow decay, it is important to ensure that flow decay during processing is comparable to that observed during retention studies. Other parameters to be monitored during processing include solution conditions, volumetric throughput, and post-production buffer flush.

CONCLUSION

Virus filtration is a critical component in the manufacture of biological therapeutics. Implementation of a virus retentive filter is one of many steps a manufacturer will take to ensure product safety. The choice of a virus filter is driven mainly by robust virus retention. Nevertheless, robust retention should be achieved as economically as possible. This brief overview of regulatory-, process-, and filter-related considerations should aid filter users in selecting the right virus filter and in initiating filter optimization studies.

Gerd Kern is technology manager, virus management solutions, for Millipore SAS, 39, route industrielle de la Hardt, 67120 Molsheim,France, +33.(0).3.90.46.93.00, fax: +33.(0).3.90.46.91.93, gerd_kern@millipore.com

Mani Krishnan is program manager, virus and biomolecular clearance, for Millipore Corporation, 80 Ashby Road, Bedford, MA 01730, 800.645.5476, fax: 800.645.5439, mani_krishnan@millipore.com

REFERENCES

1. Ensuring compliance: regulatory guidance for virus clearance validation. Bedford (MA): Millipore Corporation; 2002. Application note AN1650EN00.

2. European Agency for the Evaluation of Medicinal Products, Committee for Proprietary Medicinal Products, Biotechnology Products Working Party. Notes for guidance on virus validation studies: the design, contribution, and interpretation of studies validating the inactivation and removal of viruses. EMEA/CPMP/BWP 268/95;1996 Feb.

3. Q5A viral safety evaluation of biotechnology products derived from cell lines of human or animal origin. Federal Register 1998 Sept 24;63(185):51075-51081.

4. European Agency for the Evaluation of Medicinal Products, Committee for Proprietary Medicinal Products, Biotechnology Products Working Party. Note for guidance on plasma-derived medicinal products. EMEA/CPMP/BWP 269/95; rev 3;2001 Jan.

5. Points to consider in the manufacture and testing of monoclonal antibody products for human use. Rockville (MD): US Food and Drug Administration; 1997.

6. Carter J, Lutz H. An overview of viral filtration in biopharmaceutical manufacturing. Euro J of Parenteral Sci 2002;7(3):72-8.

7. Ireland T, Bolton G, Noguchi M. Optimizing virus filter performance with prefiltration. BioProcess Int 2005 Nov;3(Suppl 10):44-7.

8. Varadarajan U. Experimental evaluation of viral contaminant removal from bio-products using a microfiltration approach. Process Validation of Biologicals Conference; 2001 Jan; San Diego (CA).

9. Graf EG. Virus filtration, an effective strategy for reducing viral contaminants. Process Validation of Biologicals Conference; 2001 Oct; London, England.

10. Parenteral Drug Association, Virus Filtration Committee. Virus filtration. Technical Report 41; 59 Suppl (S-2). Baltimore (MD): Parenteral Drug Association; 2005 March. PDA TR41

11. Bolton G, Cabatingan M, Rubino M, Lute S, Brorson K, Bailey M. Normal-flow virus filtration: detection and assessment of the endpoint in bioprocessing. Biotechnol Appl Biochem 2005 Oct;42:133-42.

12. Hirasaki T, Noda T, Nakano H, Ishizaki T, Manabe S, Yamamoto N. Mechanisms of removing Japanese encephalitis virus (JEV) and gold particles using cuprammonium regenerated cellulose hollow fiber (i-BMM or BMM) from aqueous solution containing protein. Polym J 1994;26(11):1244-56.

13. Cabatingan M. Impact of virus stock quality on virus filter validation. BioProcess Int 2005 Nov;3(Suppl 10):39-43.

14. Bolton G, Cormier J, Krishnan M, Lewnard J, Lutz H. Integrity testing of normal flow parvovirus filters using air-liquid based tests. Bioprocessing J 2006 Spring;5(1):50-5.