Meeting Report: PDA Virus and TSE Safety Forum

2.1. Virus Contamination and Strategies of Risk Mitigation

This first session dealt with recent virus contamination incidents in biotech and vaccine industry, the history of virus safety guidelines, and individual strategies of risk mitigation.

Jim Robertson (National Institute for Biological Standards and Control, NIBSC) reflected on the evolution of virus safety guidelines that had been triggered by virus transmissions through plasma-derived products and vaccines. The safety approach that has been adopted for biotechnology products by the International Conference on Harmonisation (ICH) Q5A (1) guideline is based on careful sourcing and testing of starting materials, testing of production intermediates, and the evaluation of virus inactivation/removal steps of the downstream process. Other guidelines dealing with virus safety of, for example, starting materials such as bovine serum or investigational medicinal products (16), apply similar principles. Although no virus transmission caused by a biotechnology product has been reported, bioreactor runs have been affected, in the majority of cases by the use of contaminated, animal-derived supplements. In this context, the recent contamination of a human vaccine by porcine circovirus (PCV) was addressed. As a consequence of this incident, European Union (EU) guidance on the use of porcine trypsin, the potential contamination source, will be developed. The World Health Organization responded with a draft guideline on assessing the risk when an adventitious agent is found.

New methods on virus detection and identification—such as microarrays, polymerase chain reaction (PCR) mass spectrometry, and MPS—are neither validated nor standardised; therefore, the speaker reminded the conference that the implementation of blood/plasma testing by hepatitis C virus (HCV)-specific PCR took time and required proficiency studies to assess and improve the method performance and collaboration studies to establish adequate standards. Such studies would also be required for these new methods. Viral contaminations may have serious consequences not only for the patient, either by a virus transmission or by deprivation of medicine due to product withdrawal, but also for the pharmaceutical company that may be affected by drastic economic implications.

Mark Plavsic (Genzyme, a Sanofi-Aventis Company) discussed the concept of integrated viral risk mitigation. Although viral contaminations in biotechnology industry are rare events, proactive prevention initiatives should be implemented in order to prevent a virus entering the manufacturing process. Containment and rapid response, in case of a contamination incident, should also be developed and implemented. A viral mitigation approach usually starts with an appraisal of the potential contamination risk that is defined, on one hand, by the physico-chemical characteristics of the virus and its replication and, on the other hand, by the facility design and production process. The potential sources of a viral contamination need to be identified and understood, as existing viral detection methods could also affect the level of risk. The limitations of existing cell culture monitoring tools and traditional virus testing methods, such as in vitro tests, should be understood and based on a process risk evaluation, while additional monitoring tools and tests for specific viruses should be considered. Once a viral risk is understood, an integrated, holistic viral risk mitigation strategy could be implemented and should span across the complete supply chain. This includes the replacement, reduction, and refinement of animal-derived raw materials, treatment of animal raw materials for virus inactivation, the utilisation of single-use manufacturing equipment, effective viral clearance (inactivation and removal) steps, manufacturing facility design or modifications for segregation and containment, as well as adequate sanitisation procedures of facilities and equipment. Addressing the existing virus testing limitations should also be part of the strategy. New technologies, such a microarrays, MPS/deep sequencing, and proteomic analysis, are looked upon as valuable contributions to viral risk mitigation because they complement traditional tests.

Sally Baylis (Paul-Ehrlich-Institute) summarized the results of several studies dealing with the recent contamination of the human rotavirus vaccine Rotarix by porcine circovirus 1 (PCV1). Circoviruses are very small (∼17 nm), non-enveloped viruses with a single-stranded DNA genome. They are very resistant to standard virus inactivation procedures such as gamma-irradiation, low pH, or heat treatment. PCV1 usually infects pigs, but can persist in cell lines of different origin without any visible effect. In March 2010 the European Medicines Agency (EMA) was informed about the contamination by the manufacturer of Rotarix. As PCV1 DNA has been detected in the Vero production cell substrate, virus seed stocks, and bulk and final product of the vaccine, porcine trypsin was suggested as the most likely source of contamination. It is used during the cultivation of the cell substrate and cleavage of the rotavirus outer coat protein to improve infection of the cell substrate. The contamination of the rotavirus vaccine was detected by an MPS and pan-microbial array approach (2). Further analysis of vaccine lots has shown loads of 6–7 log₁₀ genome copies per dose. Because the viral genome present in the vaccine is resistant to nuclease treatment, it is most likely packaged in capsid/capsid-like structures. In vitro experiments at the Robert-Koch-Institute, Germany, using porcine cells, showed the transcription of PCV genes and protein expression after incubation with control virus preparations, but not with the final doses of the vaccine. However, based on studies performed by the manufacturer using vaccine bulks (equivalent to >1500 doses) neutralised with anti-rotavirus antibodies, it was estimated that the titre of infectious PCV1 is low, in the order of 3–100 TCID₅₀ per dose. In other investigations, PCV1 DNA replication was observed in porcine cells after rotavirus inactivation. In the human cell lines MRC-5, U937, and Hep2, no viral protein expression was observed. In studies performed by the manufacturer, infants immunized with Rotarix did not develop antibodies against PCV1 after vaccination; however, stool samples from some vaccinees were positive for PCV1 when tested by PCR.

The EMA concluded that this contamination incident did not raise a public health concern because PCV1 does not cause a disease in humans. There was no withdrawal of the vaccine from the market due to its effectiveness in preventing rotavirus infections.

Gay Gauvin (Amgen Inc.) outlined two cases of viral contamination events and the adopted strategy for risk mitigation. The first contamination case came to the fore by the decrease of cell viability accompanied by a reduction in aeration rate of a large-scale bioreactor. Testing for bacterial contamination and heavy metals was negative, but the analysis of samples derived from different scales of the upstream process by the in vitro adventitious virus test showed a cytopathogenic effect on cell lines derived from different species. The contaminating virus was identified as Cache Valley virus, a bunyavirus that is widely distributed throughout the cattle population of North America, and nonirradiated foetal calf serum of U.S. origin used in this process was assumed to be the source of the contamination. As short-term corrective action, gamma-irradiated serum was used and the development of a serum-free manufacturing process was initiated.

The second case study dealt with a mice minute virus (MMV) contamination. MMV is a member of the Parvovirus family and is characterised by a considerable resistance to physico-chemical inactivation procedures. The contamination was indicated by a pH shift of the culture and a decrease in cell viability. The culprit was finally identified by quantitative PCR to be an immunosuppressive strain of MMV (MMV_i). Although a comprehensive investigation was performed, the portal of viral entry was never identified. Decontamination of the facility included, in addition to the routine clean-in-place (CIP) and steam-in-place (SIP) procedures, the use of bleach to inactivate any residual infectivity and the replacement of equipment parts made of elastomer. Additional activities prior to the start-up of manufacture included the tightening of gowning requirements and restriction of the personnel and materials flow. Concomitantly, the awareness of the personnel was increased by special training, for example, on contamination prevention and pest control, as well as by revision of standard operating procedure and training documents. The strategy to mitigate the recurrence of a contamination event aims at the early detection of any viral contaminant by applying rapid and sensitive methods and also includes an improved containment policy.

Marie Murphy (Eli Lilly and Company) reported the evaluation of the high-temperature, short-time (HTST) technology using MMV as a model virus due to its resistance against temperature-mediated inactivation. This technology had been implemented by Genentech in the 1990s to treat cell culture solutions. The evaluation started with a small-scale model that reflected suitable process conditions, such as temperature ramp-up and cool-down rates. The effectiveness of treatment was measured as the probability of achieving complete inactivation in low-level contaminated solutions. Cell culture media were spiked with low levels of MMV, and HTST was performed using gradually higher temperatures (85–102 °C) and a constant residence time. The treated media was split into several flasks of cells that were considered as individual bioreactors. They were observed for cytopathogenic effects, and the supernatant of those that stayed negative was inoculated onto fresh indicator cells. The negative flasks were analyzed in a titration experiment, and the probability of complete MMV inactivation was calculated. The experiments showed that by using this model HTST treatment is only effective in inactivating a low-level MMV contamination (3), at higher temperatures (>102 °C).

Glenda Silvester (EMA) emphasized the importance of epidemiological data in the risk assessment of plasma products, referring to the recently revised EU guideline on epidemiological data on blood-transmissible infections (4). The history of virus transmissions by these products indicates a greater potential risk than for biotech products; this is aimed to be mitigated by extensive donor screening, testing of individual donations for antiviral antibodies, or viral proteins and pool testing for HCV by nucleic acid amplification techniques (NAT). In the plasma master file (PMF), epidemiological data on human immunodeficiency virus (HIV)-, hepatitis B virus (HBV)-, and hepatitis C virus (HCV)-positive donors in a specific donor population are collected to get information about the infection risk of these viruses, to support the selection of adequate donors and to indicate the effectiveness of donor screening. Prevalence and incidence are complementary, as they provide information on past and current infection risk in a donor population. The EU guideline recommends the evaluation of the prevalence in first-time tested donors and of the incidence in repeat tested donors in a given period of time. PMF holders are expected to stipulate acceptance ranges for the data collected and to perform a trending analysis. If the ranges are not met by the collection centres, a root cause analysis should be performed and corrective and preventive action (CAPA) measures initiated. However, further details, for example, how acceptable ranges for the data can be set or how significant trends should be defined, remain to be discussed.

Houman Dehghani (Amgen Inc.) described an industry consortium under the auspices of the Massachusetts Institute of Technology (MIT) Center for Biomedical Innovation on “Biotech Virus Contaminations” with a view to collecting data on contamination incidents, performing a risk analysis, and developing mitigation strategies. As detailed information about contamination events in the biotech industry is rarely available, this project is expected to identify contamination risks and effective barriers. Pre-drafted questionnaires asking for information about the viral contaminants identified, detection/identification methods, source and frequency of contamination, raw materials and in-process controls, CAPA measures, and so on are employed for data compilation. Each company providing its data becomes a member of the consortium and is able to guide the project. The confidentiality of source data is maintained by individual nondisclosure agreements between consortium members and MIT. The findings will finally be reported, anonymized in the form of a white paper, and may be used for assessment of best industry practices for prevention and control of virus contamination.

2.2. New Technologies for Virus Detection

In session two, four speakers presented recent advances in technologies that ultimately should enable detection of any viral contaminant in biotechnology processes. During the introductory statement the audience was reminded by Thomas R. Kreil (Baxter BioScience) that it had taken just a decade from the Nobel Prize–worthy invention of PCR testing in 1983 to the initiation of large-scale application of the technology, initially to prevent contamination of plasma for fractionation by HIV, HBV, and HCV. Given the even more rapid pace now of putting basic science to use in an industrial setting, the broad utilization of cutting-edge virus detection methods such as microarrays or deep/massively parallel sequencing needs to be considered imminent. However, the interpretation of results generated with these technologies still leaves a number of questions unanswered, ranging from aspects of validation and path to licensure to the ultimate interpretation of findings with respect to biological safety of the products tested. It is important to remember that the first product contamination detected by metagenomics and a panmicrobial microarray (2) did not result in regulatory restrictions of product use, as the risk-benefit assessment was still considered positive.

Opening the series of presentations, David Onions (BioReliance) introduced the different analytical concepts, with a proposed hierarchy of use depending on the target application, going from low cost and only hours of analysis time, to high cost with a few days required for analysis. Known pathogens would best be tested for by (panels of) PCR tests, emerging pathogens by microarrays, and then advanced/totally unknown pathogens by next-generation sequencing methods.

Going into more depth on the use of massively parallel sequencing as a fully unbiased method to potentially detect any virus contaminant, the good manufacturing practice (GMP) qualification for such an approach was considered entirely possible, with potentially the one remaining issue being the bioinformatics processing of the large volume of data generated. As a basis to support such an approach, a database with >300,000 curated sequence entries has already been established. Given the significant effort associated with the technology, proposed applications would primarily include cell bank testing for qualification, rather than, for example, lot-by-lot release testing.

Finally, case studies were presented that included the detection of porcine circovirus, a new bovine parvovirus (5), bovine Norwalk virus, and bovine kobuvirus by massively parallel sequencing, viruses that could have been missed by conventional tests.

Rangarajan Sampath (Ibis Biosciences, an Abbott Company) gave a presentation on the application of their PLEX-ID PCR panel system in a number of case studies with Amgen, Baxter, and BioReliance. In these, the technology was capable of identifying several virus strains spiked into samples provided in a blinded fashion. In another collaboration with Genentech, an unknown viral contaminant could finally be identified as Bluetongue virus (6).

Houman Dehghani (Amgen Inc.) then presented an overview of measures designed to assure the safety of biological products from mammalian cell culture, as well as a perspective on the future of in-process testing utilizing molecular methods to complement traditional cell culture–based methods. Specifically, a universal biosensor approach that could be utilized to detect bacterial, mycoplasmal, as well as viral targets was discussed, as well as approaches towards the demonstration of such a technology's utility (7, 8).

Finally, Thomas Briese (Columbia University) offered a broad discussion of technology options for the discovery and surveillance of pathogens, to employ a staged strategy for pathogen detection quite similar to the one discussed earlier by David Onions.

Regarding the individual technologies, the presentation also included a discussion of several recent successes, specifically, the discovery of a novel human rhinovirus by mass-tag PCR (9), diagnosis of malaria in the context of a filovirus outbreak by microarray (10), and the discovery of a new hemorrhagic fever–associated arenavirus (11) or of a novel astrovirus associated with an unexplained encephalopathy by high-throughput sequencing (12).

A lively panel discussion focused on paradigmatic issues, such as pan-virus testing versus pan-virus removal by, for example, nanofiltration, similar to current sterile filtration practices, as well as implementation issues, such as the requirements for validation and GMP compliance for the new molecular methods to attain regulatory approval.

While there was wide acceptance that ultimately these new methods have the potential to supplement or eventually even replace currently established virus detection methods, it was equally evident that considerable more work will be required before these brilliant research tools will become part of accepted routine practices in the biotechnology arena.

2.3. Virus Reduction

In this session, there were seven presentations covering various aspects of virus reduction, ranging from history and evolution of regulatory requirements, to specific unit operations, to biological and biophysical characterizations of model viruses commonly used for virus clearance studies.

Hannelore Willkommen (Regulatory Affairs & Biological Safety Consulting) gave an overview of the history and evolution of virus safety regulations. Many current requirements such as statistical analysis of virus titration methods, inactivation kinetics, and demonstration of virus partitioning across columns were born of studies following the occurrence of virus transmission cases via plasma products in the 1980s and early 1990s. In addition, several examples were provided to illustrate the importance of understanding the mechanism of action and process parameters in order to assure safe and effective control. The potential impact of model virus strains as well as virus preparation methods was discussed. For example, significant variation of hepatitis A virus (HAV) inactivation by pasteurization has been reported previously. A recent study has shown that different HAV variants vary significantly in thermostability, resulting in >2 log₁₀ of differences in inactivation under the same heat treatment conditions (13). Finally, several key elements for providing safety assurance were emphasized, such as appropriate design/execution of virus clearance studies considering the principles of good virological practice (GVP), the use of well-characterized indicator cells and model viruses, the science based data interpretation, and a clear understanding of CPPs.

Kurt Brorson (Center for Drug Evaluation and Research/U.S. Food and Drug Administration, CDER/FDA) summarized the overall experience and lessons on the robustness and reliability of virus reduction by various unit operations used for production of mAbs. The summary was based on a large body of information and data collected by CDER from regulatory submissions over the years as well as information shared during the 2009 Viral Clearance Symposium in Indianapolis (14). Reliability is a measure of virus clearance consistency, predictability, scalability, independence from product effects, and operational feasibility. Robustness is a measure of independence of clearance with respect to small defined and controlled changes in CPPs. As per these definitions, unit operations involved in virus reduction were placed into three categories based on the above described data sources: (1) those demonstrated to be robust and reliable, (2) those close to robust and reliable, and (3) those where improvement is needed or gaps exist. Furthermore, specific control strategies were proposed for individual unit operations that had been extensively studied and well understood with regard to effective virus reduction. Low pH treatment for inactivation of retrovirus and virus filtration of large viruses are classed as robust and reliable methods where ≥5 log₁₀ of virus can be inactivated or removed. Detergent inactivation and anion exchange (AEX) flow-through chromatography are considered close to robust and reliable, as low log reduction values have been observed in some cases dependent on the detergent or resin type used. Protein A and cation exchange (CEX) chromatography and virus filtration with small-virus filters are methods that require further evaluation before an effective control strategy based upon their use can be defined.

George Miesegaes (CDER/FDA) described a systematic evaluation of CEX chromatography for potential use as an alternative mAb capturing and virus removal step. Five CEX resins (Poros HS, Unosphere S, Hyper D, Capto S, and GigaCap) were assessed using a DoE approach. Based on predetermined criteria for recovery, yield, and purity, Hyper D, Capto S, and GigaCap were selected for further investigation using porcine parvovirus (PPV) as a model virus. High pH and low conductivity within the range evaluated improved PPV clearance as well as the purity profile for all three resins. Together, the data indicate that it is possible to use CEX chromatography as an alternative primary capturing step for mAb production and achieve comparable yield, impurity removal, and parvovirus clearance. However, prior development studies are needed in order to identify optimal process conditions in a case-by-case manner.

Albrecht Groener (CSL Behring) presented an extensive review on virus removal by chromatography used for production of plasma-derived products. Various chromatography columns (ion exchange, size exclusion, affinity, and hydrophobic interaction) used in production of plasma-derived products have been evaluated for virus reduction. Several learning points concerning the effectiveness of virus removal by chromatography were summarized based on the large body of data accumulated over the years. First, chromatography as a whole is capable of removing a wide range of viruses. Second, different chromatography columns often vary in reduction effectiveness for a given virus. Third, it is possible for a given chromatography column to provide robust and consistent reduction for a given virus within the narrow range of process specifications. Fourth, chromatography contributes to overall virus reduction capacity of the manufacturing process for plasma-derived products. Finally, appropriate design and execution of laboratory-scale virus clearance is essential to assure product safety. Because process conditions often vary for different types of chromatography based on partitioning mechanism and product, it is impractical to apply a one-size-fits-all platform approach for chromatography.

Dayue Chen (Eli Lilly and Company) described a study demonstrating that a small-virus filter (Planova 20N) can consistently and reliably retain retroviruses. Laboratory-scale virus clearance studies were carried out by either spiking parvovirus and retrovirus together (co-spiking) or separately (singly spiking). Parvovirus breakthrough was observed in both singly spiked and co-spiked runs. However, the data from co-spiked experiments provided direct and compelling evidence indicating that Planova 20N can consistently and reliably achieve complete retrovirus clearance even when parvovirus breakthrough was taking place. Based on the results from co-spiking studies using five different molecules, a modular retrovirus clearance of ≥6.0 log₁₀ was proposed in support of clinical trials for future molecules as long as the Planova 20N filter is used in the production.

Hans Rogl (F. Hoffmann–La Roche Ltd.) presented a study comparing two small-virus filters (Millipore's Viresolve Pro and Asahi's Planova BioEx) using a DoE approach. Effects of several parameters such as product concentration and feedstock pH on parvovirus reduction were assessed and compared between the two filters in a total of 51 individual experiments. Robust parvovirus reduction was demonstrated for both small-virus filters. Based on the results from the study, a reduced virus validation program was proposed for clinical development of future mAbs provided that the unit operation is operated within the defined ranges.

Horst Ruppach (Charles River Laboratories) reported a study aimed at an improved understanding of model viruses used for virus clearance studies. Standard protocols were developed and used for phenotypic characterization of virus stocks. Specifically, growth kinetics, cell tropism, pH susceptibility, thermostability, freeze/thaw sensitivity, and aggregation propensity were evaluated. The specified phenotypes were used to characterize the model viruses and also to define significant differences even between closely related viruses. As an example, xenotropic murine leukemia virus (XMuLV), MMV, and pseudorabies virus (PRV) exhibited very different cell tropism and one-step growth kinetics, making it possible to identify and differentiate these viruses based on these characteristics. In a head-to-head comparison and as an example, XMuLV and PRV showed different and distinct thermo-stability and pH susceptibility properties, indicating that not all enveloped viruses behave the same. Matrix components such as the protein content and salt concentration can influence thermostability, and this effect can be different for different model viruses as was demonstrated for XMuLV and PRV. Different parvoviruses showed very similar phenotypes but had apparent different inactivation profiles when subjected to alkaline pH treatment (pH = 12) within a given time range. The phenotypes of two different strains of feline caliciviruses were almost identical except for a slight difference in the thermostability and a significant difference in inactivation by isopropanol treatment. In summary, the relative stability/resistance of the different model viruses is a function of the type of treatment and is adequately reflected by the specified phenotypes.

2.4. Virus Safety by QbD/Platform Technology

Increasingly, platform cell culture and purification processes are being used for the production of cell-derived medicinal products, and this session concerned the implementation of QbD and platform technology to assure virus safety. Biopharmaceutical manufacturers are building virus clearance databases as they develop multiple products. Spurred by the QbD initiative, they are acquiring process understanding by testing the impact of process parameters and investigating the mechanism of action of virus clearance unit operations. Speakers presented regulatory and industry perspectives on the application of platform data and scientific knowledge to assure virus safety with an emphasis on products in clinical development, but also to support Marketing Authorization Application or Biological License Application filings.

Johannes Blümel (Paul-Ehrlich-Institut, PEI) presented a perspective on the overall EU regulatory approach described in the Committee for Proprietary Medicinal Products (CPMP) guideline, “Note for guidance on virus validation studies: The design, contribution and interpretation of studies validating inactivation and removal of viruses.” Robustness of virus clearance by particular unit operations needs to be demonstrated, and studying critical parameters is the key to this understanding. A recent retrospective, cross-industry data analysis from the Plasma Protein Therapeutics Association (PPTA) (15) has shown that detergent choice is a critical process parameter for solvent/detergent-based enveloped virus inactivation. However, the effects of all potential factors (e.g., lipid content, temperature, aggregates) and possible interactions are at present not completely understood.

For biopharmaceutical products, platform approaches for virus clearance are increasingly being advocated. At the present time, low pH and detergent-based inactivation appear to be the best candidates for platform approaches, although unsettling low log₁₀ reduction values (LRV) “outliers” are found when PEI performs retrospective analysis of archived data. Certain unit operations, like filtration of parvoviruses, need additional work to understand mechanisms and failure modes before any platform claims can be made. Other unit operations, like flow-through AEX, appear to be an effective reduction step for some nonenveloped viruses (MMV, PPV, simian virus 40 (SV40)) but not others (poliovirus). A major unaddressed question is whether and to what extent existing unit operations can clear circoviruses, a virus that only recently emerged as a major concern for bioprocessing. PEI allows reduced testing for clinical trial products, with a general expectation that the safety factor for retrovirus is sufficient for a quantitative risk assessment as per ICH Q5A (1), and for parvovirus that there is at least one effective step or sufficient overall reduction.

Kurt Brorson (CDER/FDA) presented expectations/experiences on modular validation for antibody products and current thinking on viral clearance data supporting QbD filings. The presentation started with a historical overview of the streamlining concepts proposed by FDA in the 1990s that ultimately fed into the QbD initiative. These were expressed in two key FDA documents, “Content and Format of Investigational New Drug Applications (INDs) for Phase 1 Studies of Drugs, Including Well-Characterized, Therapeutic, Biotechnology-derived Products (1995)” and “Points to Consider (PTC) in the Manufacture and Testing of Monoclonal Antibody Products for Human Use (1997).” The first document stated that “present regulations allow a great deal of flexibility in the amount and depth of various data to be submitted in an IND (investigational new drug application) depending in large part on the phase of investigation.” The second document introduced the streamlining concepts of modular validation (application of validation data from one product to a second, similar product) and bracketing of CPPs during a validation study. The bracketing concept led directly to the DoE studies that are an integral part of the QbD approach to product development. The PTC document also introduced the concept of robust virus clearance as “steps … that work well under a variety of conditions … with a variety of mAb.” Robustness is a key concept supporting QbD's reliance on mechanistic understanding and risk assessments for DoE-based design spaces and process control strategies. The presentation ended with a discussion of QbD case studies available in the public literature. While these case studies should not be viewed as guidance, gold standards or a template for all products, they have great value as educational tools (see description of David Paolella's presentation, below). Despite minor variations, they all possessed a common flow: identification of quality therapeutic product profile → risk assessments → design space identified via DoE studies → control strategy.

Mahmood Farshid (Center for Biologics Evaluation and Research, CBER/FDA) described the CBER's expectations regarding virus clearance studies for plasma-derived products. Unlike recombinant protein products, the starting material for this product class (source plasma and recovered plasma) is taken directly from humans. Historically, the transmission of human blood–borne viruses HBV, HCV, HIV, HAV, and B19 has been a particular concern because of the pathogenic and in some cases deadly nature of these viruses and, thus, a number of precautionary steps have been taken to prevent the risk of infection. These include (1) screening donors for heath status and for risk factors for transmissible disease, (2) implementation of virus clearance steps in the production process, (3) adherence to GMPs in the plasma fractionation facility, and (4) the presence of neutralizing antibodies in some products (e.g., immunoglobulins). Surveillance for new viruses (e.g., West Nile virus (WNV)) and implementation of NAT as a sensitive detection tool are growing parts of the plasma-derived product safety program. CBER takes a conservative approach towards plasma product safety, often obtaining advice and concurrence from an independent expert body, the Blood Product Advisory Committee. While differing in some details (e.g., choice of model viruses), the virus clearance overall strategy for plasma products possesses commonalities with that of biotech products.

David Paolella (GlaxoSmithKline) presented an overview of the published model mAb (A) QbD case study, with an emphasis on the A-mAb downstream process and virus clearance strategies (low pH inactivation, AEX chromatography, small-virus retentive filtration). A-mAb is a fictitious, but realistic, monoclonal antibody drug substance and drug product, and the highly comprehensive case study is publicly and freely available as a teaching tool for industry and agencies (see www.casss.org or www.ispe.org/pqli). The study covers protein structure, quality attributes, upstream and downstream processing, drug product fill and finish, control strategy, and regulatory aspects. The intent is to illustrate the benefits of a QbD development approach with sections that are realistic and represent selected QbD principles. However, it is not a mock dossier, a template for all applications, or a gold standard. All of the major steps and strategies of QbD are outlined, including identification of critical quality attributes (CQAs); identification of CPPs; linkage of CPPs to CQAs; a rational approach to defining a control strategy that reflects product and process understanding; and finally a risk-based, lifecycle approach to managing continual improvement. The presentation illustrated how all of the above could be applied to virus clearance.

Daniel Strauss (Eli Lilly and Company) presented a strategy to apply QbD principles to virus clearance purification process steps and demonstrate process robustness. While testing all combinations of variations of process parameters is not feasible, the QbD risk-based approach provides tools such as risk assessment and DoEs to choose parameters and efficiently test them. During the risk assessment phase for each unit operation, all process parameters, feedstock attributes, and key raw materials are considered across the intended operating ranges. A risk rating is given for each parameter based on its potential to affect virus clearance, supported by in-house and literature data and an understanding of the mechanism of action. Experiments then are designed taking into consideration the model virus, unit operation parameters, and information desired from the study such as main effects vs interactions. Protein A chromatography in a mAb purification process was given as an example for the risk assessment and DoE design. Removal of PPV was studied using a fractional factorial design evaluating bed height, load pH, load conductivity, load density, loading flow rate, and elution flow rate. Parameter impact was evaluated using statistical analysis, and was considered significant only if P ≤ 0.05 and the scaled estimate (LRV change) ≥0.5. It was shown that virus clearance by Protein A chromatography is not significantly affected by process parameter variations over the ranges tested.

Rachel Specht (Genentech) presented an approach to leverage an in-house virus clearance database for platform purification process design, modular virus clearance validation, and QbD. Due to the many commercial and development mAbs derived from Chinese hamster ovary cells, the company was able to accumulate an extensive virus clearance database for mAbs from validation and development studies. This database included over 70 studies performed between 2005 and 2010 for AEX and virus-retentive filtration. The majority of AEX processes uses QSFF (Q-Sepharose Fast Flow) operated in flow-through mode, resulting in effective and reproducible virus removal. XMuLV removal is effective for all mAb processes using either large- or small-virus retentive filters. The in-house experience and knowledge is used to guide platform purification process design; where well understood, dedicated virus inactivation and removal steps such as low pH, detergent, and virus retentive filtration are incorporated. These steps, as well as AEX chromatography, are operated under the most robust conditions to achieve planned and predictable virus clearance. For commercial products, the in-house database is a critical component for QbD study design where a thorough understanding of parameters that affect viral clearance supports risk ranking, DoE study design, and parameter-acceptable ranges. Multivariate studies used to support QbD may be applied to future QbD studies to streamline study design.

Herbert Dichtelmüller (Biotest AG) summarized a data collection from the PPTA member companies on contribution to safety of immunoglobulin and albumin from virus partitioning by cold ethanol fractionation. PPTA is the primary advocate for the world's leading source plasma collectors and producers of plasma-based therapeutics. The cold ethanol fractionation is a method developed by Cohn and Oncley in the 1940s and 1950s to separate fractions of human plasma, and it is an essential process to make human plasma–derived immunoglobulins and albumin medicinal products. By addition of ethanol and manipulating pH at low temperature, the protein of interest is separated from impurities as they are precipitated or remain in the supernatant and are collected by centrifugation or filtration. During this process, viruses can be precipitated along with protein impurities thus removed from the intended products. Six PPTA member companies (Baxter, Biotest, CSL Behring, Octapharma, Kedrion, and Talecris) participated in a data review of over 600 virus clearance studies by cold ethanol precipitation. It was found that cold ethanol fractionation contributes significantly to removal of certain viruses, often in the range of >4 to >5 log₁₀. However, bovine viral diarrhea virus (BVDV) and PPV are more challenging to remove. Furthermore, different model parvoviruses such as CPV, PPV, and MMV show different fractionation behavior in this scheme. When process parameter effects were evaluated for Fraction III, it was found that separation of precipitates by filtration was more effective for virus removal than centrifugation. For immunoglobulin preparations (IgG) from Fraction III, and albumin from Fractions IV-1/IV-4, better case parameters could be defined in this study for BVDV and PPV removal. Some parameters, such as precipitation time, do not affect virus removal significantly, while others, such as a suitable pH, can favor virus removal.

A panel discussion was held at the end of the Virus Safety by QbD/Platform Technology Session to further facilitate the dialogue among participants on topics of shared interest. The in-depth discussion focused on topics related to regulatory compliance and QbD as summarized below:

• On whether evaluation of a single step for parvovirus clearance is sufficient to support clinical development if robust (>4 log₁₀) clearance is demonstrated per EMA guideline (16), it was agreed that specific consideration should be given to the mechanism of action to ensure that not only the step achieves LRVs >4 of parvovirus clearance but is also capable of clearing a broad range of viruses.

• A QbD approach requires a significant number of virus clearance studies to increase the probability of encountering outlier result. The potential risk of such outlier results is a major concern when firms consider applying QbD concepts in virus clearance. An effective approach is to repeat the experiment a few times to determine whether the initial result is real or a true outlier so that appropriate decision can be made based on data and scientific reasoning. Whenever possible, a thorough investigation should be conducted to determine the root cause. It was concluded that the potential risk of outlier results should not deter the industry to embrace QbD.

• The DoE method is an essential component of QbD and a powerful tool for assessing the clearance robustness. However, DoE significantly increases the number of clearance studies needed to support regulatory filing. The question as whether DoE data are “nice to have” or “must have” for demonstration of clearance robustness was raised and discussed. While implementation of QbD in virus clearance is considered voluntary at this time, demonstration of clearance robustness is required for marketing authorization applications. Understanding of the critical parameters and their potential impact on clearance is the key to assure virus safety. In this regard, QbD methodology (e.g., risk assessment and DoE) makes it possible to assess the effects of critical parameters in a multivariate manner, providing basis for setting appropriate design space and effective control.

• The question of whether knowledge obtained from a DoE study using a single virus for a given step can be generalized was vigorously discussed. Several examples were given to emphasize the fact that effectiveness of virus clearance for a given unit could vary significantly for different model viruses. Therefore it is inappropriate to assume that the information from one model virus can be extrapolated to other ones. However, when the mechanism of action is well established and sufficient data are provided, it is scientifically justifiable to apply conclusions from the worst-case model virus to other viruses. For example, retrovirus clearance by virus filtration step can be assumed based on LRV achieved using parvovirus.

3.1. The Current Situation with BSE and vCJD

Session 1 reviewed the current knowledge on the agents responsible for TSEs and the implications for the safety of medicinal products. This was followed by an update on sporadic and vCJD and on progress to develop assays for vCJD agents.

Linda Detwiler (University of Maryland) gave a review on animal TSE and, particularly, on the evolution of BSE cases, country classification, emerging TSE diseases, and other potential risks. The change from a passive to an active surveillance system has had an important impact in the evolution of BSE cases in many countries. The revised BSE country classification from World Organisation for Animal Health (OIE) was presented (17). In general, a decline in classical BSE is observed together with a significant decline in TSE research funding and a relaxation of controls. The greatest risks remaining, however, are apathy and complacency. Atypical disease and potential risk of spread to other species need to be considered. The origin of atypical cases is still unknown, but potential reasons were discussed (low level of feed contamination cannot be excluded; perhaps sporadic cases that may occur around the world). The pathogenesis is not fully defined and current standards do not cover atypical scrapie. Recent findings regarding chronic wasting disease (CWD) were also presented. Remaining threats to the pharmaceutical industry include sourcing from countries not yet classified, lack of compliance, premature relaxation of regulations, emerging TSEs, CWD, TSE spreading to other species, or presence of infectivity in the absence of the Proteinase K (PK)-stable (PrP^res) form of the pathogenic form of PrP (PrP^TSE).

Robert Will (University of Edinburgh) gave a review of the number of cases of vCJD in relationship to the BSE epidemic in the UK and in Europe. The number of vCJD deaths per annum in the UK is in clear decline after a peak in 2000, and no cases in persons born after 1999 have been identified. The findings of the TMER (Transfusion Medicine Epidemiology Review) study were presented (18). In the UK a total of 175 cases of vCJD have been identified, and blood components from 18 donors who developed later vCJD were transfused to 67 recipients. Four instances of probable transfusion-transmitted infection have been identified. The time from donation till development of clinical symptoms and the survival of recipients of contaminated components has been traced. It clearly shows that infectivity is present in blood before symptoms appear. Ten vCJD cases were recipients of blood transfusion. The interval from transfusion to onset of illness varies from 4 to more than 16 years. The findings strongly suggest that blood transfusion is a transmission route. Possible further outbreaks are possible but are considered limited. Plasma-derived medicinal products have also been considered as a potential vehicle for transmission of vCJD agents. Recently the detection of PrP^res in the spleen of an adult hemophilia patient (without any clinical symptoms) has been published (19). The risk of transmission has been assessed for 787 UK patients with an inherited bleeding disorder prospectively followed for 10–20 years. The absence of clinical vCJD cases in this cohort to date suggests that either plasma fraction infectivity can be estimated as extremely low, or the incubation period is longer for this cohort than for implicated cellular blood product recipients. Further follow-up of this cohort is needed. The transmission of sporadic CJD by blood transfusion has been recently reviewed (20). The information available remains limited; however, no significant risk has been found in the UK. There may be bias and interpretation in the studies conducted, so this issue needs to be kept under evaluation.

Jillian Cooper (NIBSC) described the work performed at NIBSC in collaboration with the UK CJD Steering Committee (including experts from UK and Europe) to address the issues surrounding the evaluation of tests for potential blood donor screening (21). Four transfusion-related transmissions have been described, therefore, a blood screening test would be very valuable to identify carriers and possibly to allow intervention before clinical symptoms develop. There are a number of challenges for a screening test evaluation, for example, availability of test material (vCJD plasma), volume of test material available, and the need for a confirmatory assay. Different approaches have been tried to overcome these limitations, such as the use of vCJD tissue homogenates spiked into human plasma, or blood or blood components from animals affected by TSE as initial test material. The process followed for a particular candidate test was presented. Results obtained using sheep plasma showed problems with the specificity of the assay as the scoring of positive and negative samples was not satisfactory. The results from plasma samples spiked with primate BSE brain tissue (10% homogenates) from animals at the preclinical and clinical stage of the disease were in line with the level of detection previously observed at NIBSC. However, the test failed to score primate plasma samples positive. Specificity studies were also carried out, including a large number of negative samples. Fresh and frozen plasma samples spiked with vCJD brain were included. The test failed to score two plasma samples from clinical cases of vCJD. In conclusion, a higher level of sensitivity is likely to be needed. Alternatively, there could be a signal that may be detected at earlier time points in the course of the disease. Pros and cons of other test alternatives under development were discussed.

3.2. Regulatory Update

The European Commission published a revised TSE guideline in March 2011 (22). Several changes related to the classification of countries and regions in BSE risk categories, for example, the replacement of the European system by the OIE classification according to Chapter 11.4 of the Terrestrial Code, as well as changes related to specific bovine-derived materials were introduced. Sol Ruiz (EMA/Spanish Medicines Agency) explained the principles of the risk assessment that is required to demonstrate compliance with these requirements. The introduced OIE classification is different in some respects from the previous European classification, as the former four risk categories are replaced by the three OIE categories (23) while resolution No. 17 adopted at the 79^th general session, May 2011, provided the classification of further countries so that 15 countries or regions are now acknowledged as having a negligible BSE risk, whereas 32 countries were recognized of having a controlled BSE risk including the USA and most European countries (17). Considerations of specific, bovine-derived products are provided in the guideline also; chapters related to collagen, gelatine, and bovine-blood derivatives were amended; and an additional chapter relating to peptones was introduced. The European Directorate for the Quality of Medicines & HealthCare (EDQM) has successfully used its certification process to assess compliance of raw and starting materials with the TSE guideline. Related to the BSE risk in Europe, the epidemiological situation has changed dramatically since the stringent EU prevention measures (TSE Roadmap 2) were introduced in 2001. From that time the recorded cases of BSE dropped from 2167 in 15 EU member states to 67 cases in 27 EU member states in 2009; there was also a significant fall in the number of birth cohort BSE cases, from 1870 in 1995 to 23 in 2001.

Alberto Ganan Jimenez (EMA) provided the view of the EMA on the risk of CJD transmission by human plasma–derived medicinal products. It is confirmed in the position statement released in 2011 by the Commission for Medicinal Products for Human Use (CHMP) (23) that there is no epidemiological evidence for the transmission of classical CJD by blood, blood components, or plasma products. But four instances of secondary transmission from healthy donors who developed vCJD 17–40 month following blood transfusions were observed, and there is one presumed case observed in an elderly hemophilic patient where the most likely route of infection was the receipt of a plasma-derived medicinal product. Estimated vCJD prevalence in different UK studies range from 237 cases per million (95% confidence limit 49–692) (24), to 0–113 per million (reflecting the 95% confidence limit calculated from 1961–1995 birth cohorts), and finally to 109 infections per million (95% confidence limit 3–608) (25). vCJD cases occurred also in other countries. Therefore donor selection criteria defined in Annex III of Directive 2004/33 remain in place, and manufacturers of plasma-derived medicinal products are furthermore requested to assess the capacity of the manufacturing process to remove the TSE agent. The recommendation of batch recall where a donor to a plasma pool subsequently develops vCJD is also maintained. The detection of a low level of infectivity (3.8 ID/mL) in the urine of TSE-infected animals has been reported (26). This resulted in a requirement to estimate the potential reduction capacity of the manufacturing process of urine-derived medicinal products and encouraged record keeping to trace back when possible to the donor. The CJD risk of Advanced Therapy Medicinal Products (ATMPs) relates to cell-based medicinal products where the risk assessment should consider the potential risk related to human products used as raw materials or excipients.

Dorothy Scott (CBER/FDA) explained current U.S. FDA considerations for the TSE safety of medicinal products derived from human plasma. Although occurrence of BSE cases has declined worldwide, CWD in the U.S., atypical scrapie and atypical BSE cases are of concern, and the exposure of humans to TSE agents may still occur. Donor deferrals for vCJD risk may decrease the risk of vCJD contamination of human plasma but cannot exclude every infected individual. Performing TSE clearance studies that demonstrate the removal of TSE agents by manufacturing processes is optional in the U.S., but a labeling claim can be obtained. To get a labeling claim accepted, study conditions must replicate manufacturing conditions, reduction factors should be reproducible, and the bioassay should be used to determine infectivity; it should be shown that mechanistically similar steps are or are not additive, and studies should be designed to include possible conditioning of the spiked TSE agent by prior manufacturing steps (such as detergent treatment) that may affect physical state and clearance properties of the TSE agent. Logistical and scientific challenges relate to the TSE spike and the assays used. Currently exogenous, brain-derived spiking preparations are used that may contain multiple forms of infectivity (membrane-bound or soluble); the spectrum goes from clarified brain homogenate to highly solubilized infectivity (detergent-treated, sonicated, ultracentrifuged supernatant). Endogenous blood-born infectivity or spleen-derived infectivity studies are challenging because of the low titer in these tissues (2–100 i.c. LD₅₀/mL blood). Recently published, more sensitive methods for detection of TSE infectivity or PrP^TSE (27⇓–29) could have potential for use in endogenous infectivity studies. Bioassay in hamsters or in mice remains the gold standard for infectivity assays because of high sensitivity and lack of reliance on resistance to proteinase K digestion, which is the basis for the detection of PrP^TSE by Western blot (WB) and most other in vitro methods. Alternative cell culture bioassays are not yet well characterized for this purpose, and their relevance and application for clearance studies is currently uncertain. A continuing area of concern is the sanitization of reused manufacturing resins. NaOH sanitization removed TSE infectivity under specific conditions, but a range of resins and conditions have not been systematically studied (30). Critical parameters for sanitization are likely to be NaOH concentration and dwell time. However, further research is needed.

Marc Martin (Agence Francaise de Sécurité Sanitaire des Produits de Santé, AFSSAPS) discussed a review of TSE removal investigational studies from the European perspective. The CHMP position statement (23) requires manufacturers of plasma products to “estimate the potential of their specific manufacturing processes to reduce infectivity” and to consider the recommendations provided in the guideline (EMEA/BWP/5136/03, 21 October 2004) (31). The physicochemical nature of TSE agents in blood is unknown and brain-derived spiking agents (brain homogenates, microsomal fraction, caveola-like domain) are used. Most published studies were performed with brain homogenate (with and without detergent and sonication), although increasing numbers of studies were performed with the microsomal fraction. Twenty-three of 32 published studies used the 263 K strain (hamster adapted scrapie), and the remainder with other TSE strains (human, bovine, mice-adapted scrapie). In 250 experiments (1 step/1 strain/1 inoculum/1 assay), 195 used WB and 55 the bioassay for detection of TSE agents, while 26 experiments used both. No discrepancy was found in 24 of the 26 experiments; in the two experiments where the bioassay revealed lower reduction than the WB, the difference was less than 1 log₁₀. The outcome of the studies demonstrating removal by partitioning does not appear to be influenced by the nature of the TSE agent used for spiking. Filtration studies, however, demonstrated that small virus–retentive filters cannot remove all infectivity with detergent (TnBP/Tween 80, 0.1% sarkosyl, or lysolecithin) and sonication influencing the size of the particles (32). The effectiveness of virus-retentive filters can therefore not be clearly concluded from the experimental studies performed so far. Chromatographic steps (41 experiments) and precipitation steps (113 experiments) demonstrated variable effectiveness for removal of TSE agents. TSE agents can be grown in several neuroblastoma cells and in epithelial cells. Before the use of in vitro assays can be considered as an appropriate assay to replace current bioassays, analytical validation showing stable susceptibility of cells with passage, reproducible titre, and high sensitivity is required.

3.3. vCJD Threat for Safety of Biopharmaceuticals—Current Challenges

Two presentations addressed the prion reduction capacity of the manufacturing process of plasma-derived products.

Albrecht Gröner (CSL Behring) briefly described the risk of TSE transmission by animal-derived material and its mitigation, and then focused on the evaluation of the capacity of manufacturing steps of plasma-derived products to remove prions, the causative agent of TSEs including vCJD. Parameters as outlined in the “Guideline on the investigation of manufacturing processes for plasma-derived medicinal products with regard to vCJD risk” (31), such as the choice of spiking agent and its preparation, the choice of assay for prion quantification, and summation of reduction factors of individual manufacturing processes were discussed in depth based on industry data. It was concluded that the use of different prion preparations from brain homogenate of infected laboratory animals enhances the confidence in the overall prion reduction capacity of the manufacturing process, that summing up reduction factors of individual manufacturing steps should be experimentally assessed, that similar prion reduction factors using bioassay and biochemical/serological assays for a range of plasma-derived products could be demonstrated, and that sanitisation of equipment and material avoids batch-to-batch-contamination. An appropriate safety assessment of plasma-derived proteins regarding vCJD transmission can then be provided.

Mikihiro Yunoki (together with K. Hagiwara and K. Ikuta, Benesis Corporation) discussed TSE removal by filtration when prion spike material of different sizes due to detergent treatment and/or sonication are used. Sonication resulted in residual PrP^TSE, demonstrated by WB, in the filtrate compared to non-sonicated spike preparations. When a 15 nm filter is used to remove extensively sonicated PrP^TSE, no signal in the filtrate could be detected by WB, in contrast to a bioassay where infectivity could be detected in the pellet and the supernatant of the filtrate centrifuged at 150,000 g for 1 h. In another set of experiments, filtration removed prion material from different species (263 K derived from hamster brain and vCJD derived from mouse brain (FVB/n strain) comparably. The removal of prion spike material could also be achieved by depth filters and endotoxin removal filters. However, the prion removal capacity by filtration depends on the composition of the spiked material.

The panel discussion highlighted current discussion about TSE spike preparations and detection systems that are used in investigational studies and current concerns related to the epidemiological situation with TSEs and exposure to vCJD worldwide.

• The nature of the vCJD agent in human blood is not known, and the most appropriate properties of a TSE preparation used in spiking experiments cannot be defined. It was recommended to use two preparations with distinct aggregation states and to precondition spikes (for example if an Solvent/Detergent treatment precedes the TSE clearance step). The most dispersed TSE material is the worst case for a filtration step, but the TSE agent can also re-aggregate during processing. Proteinase K (PK)-stable (PrP^res) and PK-sensitive (PrP^sen) forms of PrP^TSE should have different conformations and surface charge, and may partition differently under the same manufacturing conditions. Therefore the interpretation of data needs to be undertaken very carefully. The question as to how much reduction should be achieved to assure the TSE safety of plasma-derived products could not be answered, but it was affirmed that current data confirm that certain production processes are likely to be effective for PrP^TSE removal.

• WB is mostly used and accepted for quantification of TSE agents, but the bioassay—that is, testing of infectivity in an animal model—is still considered the gold standard. Correlation of the WB with the bioassay must be demonstrated. As TSE infectivity can be associated with PrP^sen, the detection system should be independent of PK digestion. The conformation-dependent immunoassay (33) is therefore a powerful tool but is not used widely, most probably because of patent issues. Other test systems like cell culture assays are not yet ready for use as “bioassays.”

• BSE is declining worldwide; atypical cases of BSE and scrapie remain concerning, and in the former case may be more difficult to detect. CWD is in the environment in North America and provides exposure to grazing animals. A feed ban (ban on feeding meat and bone meal to animals) was very successful in limiting transmission of BSE; the feed ban is currently in place for ruminants but not for other species (pigs, fish, etc.). These factors suggest that humans and animals will continue to be exposed to TSE agents for the foreseeable future.