Abstract
CONFERENCE PROCEEDING Proceedings of the PDA/FDA Adventitious Viruses in Biologics: Detection and Mitigation Strategies Workshop in Bethesda, MD, USA; December 1–3, 2010
Guest Editors: Arifa Khan (Bethesda, MD), Patricia Hughes (Bethesda, MD) and Michael Wiebe (San Francisco, CA)
For decades conventional tests in cell cultures and in laboratory animals have served as standard methods for broad-spectrum screening for adventitious viruses. New virus detection methods based on molecular biology have broadened and improved our knowledge about potential contaminating viruses and about the suitability of the conventional test methods. This paper summarizes and discusses practical aspects of conventional test schemes, such as detectability of various viruses, questionable or false-positive results, animal numbers needed, time and cost of testing, and applicability for rapidly changing starting materials. Strategies to improve the virus safety of biological medicinal products are proposed. The strategies should be based upon a flexible application of existing and new methods along with a scientifically based risk assessment. However, testing alone does not guarantee the absence of adventitious agents and must be accompanied by virus removing or virus inactivating process steps for critical starting materials, raw materials, and for the drug product.
- Adventitious viruses
- Cell culture tests
- Animal tests
- Virus detectability
- Alternative methods
Introduction
Conventional adventitious virus testing in cell cultures and in vivo have been applied as standard methods to exclude the presence of extraneous viruses in biological raw materials, cell culture substrates, viral seeds, and virus harvests used for biological medicinal products and vaccine manufacturing. More or less identical standard schemes have been adopted and are often perceived as “broad-spectrum” methods that cover most if not all needs. The short compendial descriptions of those methods, in particular of the in vivo elements, suggest fairly straightforward test regimes and do not at all point out the practical limitations, such as (non-)detectability of various agents, time requirements, cost, and, last but not least, ethical aspects of in vivo tests.
Much hope is placed on new broad-spectrum molecular biological viral screening methods to replace conventional testing. However, due to their extremely broad range, validation of those methods for all potential applications is extremely demanding. If applied to new samples and matrices, nonspecific reactions are to be expected and require long and complex investigations. Thus, those methods may not be applicable for all purposes.
The introduction of specific (polymerase chain reaction) (PCR)-based methods to replace conventional methods is hampered by different factors. Firstly, it may be easier and less costly to stay with the old methods because the new methods need to be developed and validated. Secondly, a medicinal product license may be obtained faster and more easily if commonly accepted test schemes are applied, since for the new methods, compendial tests or standards are not yet available to judge the quality of those new methods, and individual decisions are needed. Thirdly, the new methods have a high and defined specificity and, despite their ability to detect any genetic variant of the specified range, even a very encompassing and feasible range of PCR methods or multiple PCRs may not be considered sufficient to replace the conventional tests with their undefined broad range. Thus, it may often be considered easier to continue using the old methods rather than to exploit the benefits of new methods.
Summarizing and discussing different aspects of conventional and new methods, this paper proposes a scientifically based, flexible approach to apply a combination of old and new methods. However, due to the inherent limitations of any virus testing, this paper also emphasizes the need to apply other virus elimination strategies to arrive at virus-safe medicinal products.
Conventional Test Schemes for Extraneous Viruses
Conventional test schemes in vitro and in vivo are described, for example, in the European Pharmacopoeia (1). Very similar recommendations are given in USA guidance documents (2), which, however, are partly more flexible and practice-oriented. The test schemes include in vitro tests in cell cultures or embryonated chicken eggs; in vivo tests in adult mice, suckling mice, and guinea pigs; and antibody production tests. If the test sample contains a known virus, this must be neutralized prior to testing using specific antibodies.
This article mainly concentrates on the standard cell culture and in vivo tests, which are briefly described below. To focus this article on the most critical aspects, testing for mycoplasma, mycobacteria, bacteria, fungi, and specific tests for avian and other extraneous agents, such as retroviruses, will not be dealt with, although these methods are usually applied for the same kind of test items.
Neutralization of Samples Containing a Vaccine Virus
If the sample to be tested (e.g., a virus seed lot or a virus harvest) contains a replicating vaccine virus, this virus must be neutralized using specific antibodies of non-human and non-simian origin (for avian tissues non-avian origin). The antiserum must be prepared from an immunizing antigen produced in a cell culture species that is different from that used for the production of the vaccine.
In Vitro Cell Culture Tests
Samples equivalent to 500 human doses or 50 mL, whichever is greater, or cell lysates equivalent to 107 cells are normally tested by inoculation into three different cell cultures, including a simian kidney cell and a human cell culture, which is usually a human diploid cell culture. The third cell is a cell of the same species and tissue type as that used for production. The cells are normally incubated at 36 ± 1 °C and observed for a period of 14 days. The U.S. Center for Biologics Evaluation and Research (CBER) guidance recommends a subculturing onto fresh cells and observation for at least another 2 weeks. The test is valid if 80% of the cultures remain viable and “pass” if none of the cultures shows evidence for the presence of any extraneous agent not attributable to accidental contamination. Evidence for the presence of extraneous agents is usually verified by microscopic evaluation for the absence of cytopathic degeneration throughout the observation period and at the end of the observation period by tests for haemadsorbing or haemagglutinating viruses.
Tests in Adult Mice
The test sample is inoculated intraperitoneally (i.p., 0.5 mL per mouse) and intracerebrally (i.c., 0.03 mL per mouse) into at least 10 (or 20 for the USA) adult mice and the animals are observed for at least 21 days. Mice that die after the first 24 h or show signs of illness are examined for evidence of viral infections by autopsy and macroscopic observation and by i.p. and i.c. subinoculations of appropriate tissue suspensions into at least 5 additional mice that are observed for 21 days. The test is passed if no mouse shows evidence of infection. The test is not valid unless at least 80% of the original inoculated mice survive the observation period.
Test in Suckling Mice
The test sample is inoculated i.p. (0.1 mL per mouse) and i.c. (0.01 mL per mouse) into at least 20 mice that are each less than 24 h old. The animals are observed for at least 14 days. Mice that die after the first 24 h or show signs of illness are examined for evidence of viral infections by direct microscopical observation and by i.p. and i.c. subinoculations of appropriate tissue suspensions into at least 5 additional suckling mice that are observed for 14 days. The test is passed if no mouse shows evidence of infection. The test is not valid unless at least 80% of the original inoculated mice survive the observation period. In The USA guidance documents require that emulsified tissue pools of all surviving mice are subinoculated into 5 additional suckling mice.
Test in Guinea Pigs
Five milliliters of the test sample is inoculated i.p. into each of at least 5 guinea pigs, which are observed for at least 42 days. Animals that die after the first 24 h are autopsied and examined microscopically (by histopathology) and culturally (by cell culture inoculation of specific samples) for evidence of infection. Animals that survive the observation period are “examined in a similar manner” (1). This sentence in the PhEur is usually understood to indicate that at least histopathology is required for all test animals. The USA guidance document (2) requires also an i.c. inoculation of 0.1 mL but only a necropsy at the end of the observation period.
Antibody Production Tests
Antibody production tests are mostly applied to detect rodent viruses. Specific-pathogen-free (SPF) hamsters, rats, or mice are inoculated with the test material and are monitored for antibodies against specific viruses.
Detectability of Viruses by Conventional Tests
Conventional tests for extraneous agents have never been (and will never be) fully validated in terms of specificity (the range of viruses detected) and sensitivity (applicable detection limits for detectable viruses). Instead, the compendial methods are normally qualified by a few standard laboratory strains of common viruses to demonstrate detectability of those laboratory strains. The detection limits are normally defined by the tests themselves, for example in terms of cell culture infectious doses 50% (TCID50) or plaque forming units (PFU). The minimum dose that can be detected is then defined as 1 TCID/PFU or infectious unit. Thus, cell culture tests' detection limits are not defined by an independent and commonly accepted or standardized measure. For in vivo systems, the corresponding cell culture infectious doses are occasionally provided. However, it should be mentioned that those cell culture infectious doses can vary massively and even by log steps if ideal and suboptimal virus harvests are compared using different cell systems for virus titration.
In the absence of robust virus specificity data that are based on relevant field virus isolates and due to self-referencing detection limits, it is extremely difficult to rate the value of these methods to detect different viruses. At least in part, the data and examples given below illustrate strengths and weaknesses of the conventional methods.
Detectability of viruses in laboratory animals depends on effective growth of the virus to induce visible symptoms of disease or visible organ damage. High variability of pathogenicity is to be expected between different animal species but also between inbred strains of the same species due to innate resistance, natural killer cell responses, or acquired immunity (3).
Pseudorabies virus (PRV), a suid herpesvirus, grows rapidly to high titers and with strong cytocidal effects in various cell culture types and is highly pathogenic for mice. This virus may be taken as a best-case model to demonstrate an ideal situation for the detection of an unknown virus in conventional cell culture tests. Kraemer et al. (4) compared the detection of PRV by inoculation into Vero cells or in NMRI mice or by PCR. Using dilutions series of virus stocks, they observed positive reactions in all three methods down to dilutions of 10−4. At higher dilutions the number of reacting mice declined, and the lowest detectable reaction (<20% positives) was seen at a dilution 10−6. Cell culture and PCR were positive at a dilution of 10−5 and negative at 10−6. A second cell culture passage was added and showed positive reactions down to dilutions of 10−7 of the original stock virus. These results show that in an ideal situation PCR, in vivo animal models, and in vitro cell culture systems tend to have very similar detection limits, but they also show that effective growth in cell culture may be applied to amplify low titers of virus to increase the detection limit, provided that a permissive cell culture is used. Attempts to amplify an unknown virus in a non-permissive cell culture will most likely lead to a reduction of the virus titers and thus lower the chances of detection after that amplification step.
High chance of detection by in vivo mouse model systems has also been demonstrated for other viruses that are highly pathogenic in mice. For a flavivirus chimera (tick-borne encephalitis/dengue) the 50% infectious dose (ID50) after i.c. inoculation into 3-day-old mice was between 1 and 10 PFU/cell culture infectious units (5). For La Crosse virus inoculated i.p. into mice, a minimum infectious dose 50% of 40 PFU has been reported (6). Less pathogenic viruses may require higher doses to induce visible signs in mice. In a mouse model established to induce myocarditis by infection with Coxsackie virus B3, 103.36 TCID50 culture infectious units were applied (7).
Whereas the dengue virus chimera mentioned above showed an exceptionally high pathogenicity, it is very difficult to detect normal dengue viruses in vivo, even if suckling mice are infected intracranially. Many efforts were made to develop mouse models for dengue infection, but most of the laboratory strains and clinical isolates did not replicate efficiently in mice. Only mouse-adapted strains displayed higher pathogenicity (8). It appears that only selected dengue strains inoculated into selected inbred mouse strains at high doses of virus and with long observation periods result in detectable disease (9).
A similar situation was observed during attempts to develop a mouse model for a monkeypox virus. When 38 inbred mouse strains were tested for susceptibility to the virus, only 3 highly susceptible wild-derived inbred mouse strains were identified. Using one of these three mouse strains (CAST/EiJ) the lethal dose 50% was determined at 680 PFU. In standard BALB/c mice, a 10,000-fold higher dose induced no morbidity (3).
A preliminary conclusion from those in vivo examples is that even in ideal situations, in vivo systems seem not to be more sensitive than permissive cell culture tests or PCRs. The titers of active virus needed to give positive signals may not always be present in the test samples.
As in animal models, detectability of unknown viruses by conventional tests in cell cultures is dependent on the induction of general, visible signs. In cell cultures these are either cytopathic effects or, for certain viruses, haemagglutination or haemadsorption. Haemagglutination reactions and similarly microscopically visible cytopathic cell degeneration require quite high virus titers, around 106 cell culture infectious units/mL, for example for Semliki Forest virus, (10) or, for less active viruses, such as polyomaviruses or certain herpesviruses, 107 infectious units/mL (11). These observations are consistent with the author's experience with various viruses that do not induce cytocidal effects unless the measured virus titers clearly exceed 105 infectious units/mL (unpublished results). Thus, detectability of unknown viruses by those conventional methods is dependent on effective virus growth in the chosen cell systems and under the specified conditions.
Primary chicken cells and embryonated eggs are still a very common cell substrate for several vaccine viruses. Thus avian reovirus may serve as an example of a potential virus contaminant. This virus shows highly variable results in cell cultures. Its isolation in primary chicken embryo cells was partly not successful; primary chicken embryo liver cells seem to be a more preferred cell type. Growth in Vero cells has also been reported but seems to be restricted to adapted strains from chicken embryo liver cell passages (12–14). JC virus, a polyomavirus that causes latent infections in humans, may serve as another relevant example. JC virus has a very restricted host range, with human glial cells being the only tissue in which it can replicate at reasonable efficiency (15). Other viruses may be cultivated or isolated in different laboratories, but growth of these viruses seems to be poor, is restricted to specific cultures, or requires specific conditions. Examples of such viruses are rotaviruses, metapneumoviruses, and parainfluenzavirus 4b. Attempts to grow these viruses using recommended cell lines and conditions from the donating laboratories or institutions were unsuccessful (own unpublished results).
Various viruses that have been detected by molecular biological methods in more recent times will probably not grow in any conventional cell culture assays, as many efforts to propagate these viruses in cell cultures have failed. Examples of such viruses are bocaviruses, noroviruses, and sapoviruses. For these viruses, neither cell culture systems nor animal models have been established to date (16, 17).
Finally, and mainly to demonstrate the value of the mouse antibody tests, Moore (18) has summarized results from tests for extraneous agents on more than 500 cell lines of mouse, hamster, rat, human, monkey, pig, and insect origin. The majority were monoclonal antibody producing (rodent) cells. Viral contaminants were found in less than 1% of these cells. Apart from expected retroviral particle findings, contaminating viruses were either detected by mouse antibody production tests or by specifically designed cell culture tests for bovine and porcine viruses (18). Interestingly, the author also mentioned that “an over-reliance on rapid tests only has presented problems to manufacturers and the loss of Master Cell Banks.” It remains unknown whether this indicates that PCR methods were used and detected unknown agents that could not be further identified.
Taken together, we have to conclude that only agents that effectively grow in the inoculated cell cultures or animals will be detectable, while many known (and unknown) agents will not be detected by using standard compendial test regimes. This statement will remain true even if more and specific cell culture and animal systems are added to the current standard test regime.
Unspecific Side Effects and Questionable Results
Cell culture and animal tests are “pass” if none of the cell cultures or animals show evidence of the presence of infection by an extraneous agent. What sounds clear and simple in theory is much more complicated in practice. The test sample for those tests is not a chemically defined molecule but a complex composition consisting of lysed cells, cell culture harvests with medium supplements of biological origin, or virus seeds containing high titers of replicating vaccine virus. In the latter case, high amounts of heterologous antibodies must also be added to neutralize the specific virus in the seed lot. Inoculating those components into animals and cell cultures can often induce various effects, which are not due to an infectious agent. For many test samples, unsuspicious matching control preparations that can be tested in parallel are not available to rule out nonspecific effects. Proving beyond any doubt that nonspecific positive results are not due to an extraneous agent can be a real challenge.
Unfortunately, nonspecific reactions and questionable results from tests on pharmaceutical starting materials are not generally reported publicly. This author's experience from three viral working seed lot tests, which were done to fulfill PhEur requirements, may serve as an example to illustrate that nonspecific side effects and questionable results are to be expected and appear unavoidable. The specific virus in those seeds was neutralized by a potent neutralizing antiserum. Mixtures of virus seed plus serum were inoculated into test cell cultures, mice, and guinea pigs. At the end of the cell culture incubation period, haemagglutination reactions were found at 5 °C (but not at ambient temperature) with guinea pig red blood cells (RBCs), but not with RBCs from other species. These reactions did not occur with cultures inoculated with control samples. Haemadsorption tests with the same test cultures were negative. The consequence of such findings would either be to assume an extraneous agent and to discard the seed lot or to try to identify the suspected agent or root cause. We decided for the latter approach. Additional culture passages were performed and a wide range of PCR tests was conducted to identify the causing agent. However, by those tests we were unable to identify any virus or microbial agent. When studying the haemagglutination reaction in more detail and using all sample materials separately, it was finally found that the neutralizing serum itself caused haemagglutination. Probably influenced by the cell culture incubation, the reactivity was variable and was more or less restricted to guinea pig RBCs and cold temperatures. Furthermore, at those concentrations needed for neutralization, the serum alone did normally not induce visible haemagglutination. This example shows that even in the presence of control samples, nonspecific reaction may occur, which either leads to destruction of the material tested or to expensive and extended efforts to identify the root cause.
In this context it may be noteworthy that nonspecific haemagglutination was also observed with the fetal calf serum used as a medium supplement for the test cultures. In our case, those reactivities were low enough that they were no longer present at the concentrations used.
Testing the same seed lots as described above, guinea pig injections also showed unexpected findings that caused concerns and led to additional investigations. Although all test animals remained healthy throughout the observation period, there were findings of a few, small focal discolorations of the liver surfaces of one or two animals per group. Cell culture tests were done with emulsified preparations from the affected organs but revealed no evidence for any infectious agent. Histopathology examinations also revealed no evidence for infectious events. However, the histopathology report noted that such reactions are not uncommon and seem to be “spontaneous in nature,” that is, of unknown origin. The fact that the same reactions partly also occurred in control groups inoculated with the neutralizing serum only indicates that the reactions were due to that serum and may probably be caused by immunopathological reactions.
The most common and disturbing side effects of in vivo tests are spontaneous deaths, particularly after i.c. inoculations, and cannibalism of suckling mice. Test laboratories normally compensate those deaths by starting with higher-than-needed animal numbers and by starting suckling mice tests only with extra pregnant dams so that tests can be repeated if too many mice are lost. Test method descriptions allow for a 20% loss rate and do not consider deaths occurring within the first 24 h. Dead or bitten, injured animals that are found after more than 24 h after inoculation have to be examined for any likely infectious agent.
Animal Numbers Needed
Summing up the numbers of animals that are required and mentioned in the guidelines and regulations for each test item leads to the conclusion that 35–50 animals will be needed for each test (1, 2). In practice, and for viral working seed lots, the real number of animals used is 5- to 10-fold that number. If one accounts for those animals that are needed to raise neutralizing serum, to pretest adequate serum concentrations that neutralize the specific virus of the seed lot, the animals of control groups and mice for subpassages, the total number of animals sacrificed for a single seed lot sample rises to 250–320. That number of animals presupposes that the neutralizing capacity of a serum is first done in a cell culture test and that only a narrow range (assume three doses of one serum or three different sera) that is expected to neutralize sufficiently is then used to qualify the in vivo tests. Omitting that step is extremely risky and may lead to infectious reactions by the seed lot virus itself. It may then require many more animals to rule out that this was due to the specific virus and not an extraneous agent.
Time and Cost of Conventional Testing
For in vitro cell culture testing with subculturing and observation times of 2 × 14 days, it takes about 5 weeks to complete the whole test regime if the test sample can be used without neutralizing antibodies or if a neutralizing antibody is available and can be used under known/pretested conditions. It takes about 5 months to complete the more time-consuming in vivo tests. If a virus-containing sample must be neutralized and the required antibodies are not available, a year or more will usually be needed until test results are available.
Table I summarizes the activities to be completed and their estimated duration. These time estimates are based on guinea pig test requirements. It assumes that other tests can be done in parallel and within the same time period. Five months should be planned to perform testing that includes in vivo studies in guinea pigs, but preparation of adequate test conditions that require prior neutralization of a virus that is contained in viral seeds and harvests may need substantially more time. It is not uncommon that months are spent to make adequate antibody preparations, which may eventually not qualify so that an entirely new approach must be chosen.
The cost of testing a sample by in vitro cell culture tests, and in vivo in adult and suckling mice and in guinea pigs, amounts to about €100,000 per test item. This sum does not include the cost of preparing a neutralizing antiserum, nor does it consider any other compendial tests that may be required for the same test item.
Neutralization of the Virus Contained in Seed Lots and Virus Harvests
Whenever the test sample contains a replicating vaccine virus, this must be neutralized to avoid interference with the detection of extraneous agents. The immunizing antigen for generating the neutralizing serum should be free from extraneous agents and should be made from a cell substrate originating from a different species and not of human or primate origin. Likewise, the virus strain used should preferably come from a different source. For obvious reasons the neutralizing antiserum can only be considered adequate if it does not contain high antibody levels for relevant extraneous agents. Therefore, the use of monoclonal antibodies is preferred (2).
In the best case, another independent and licensed vaccine may be used to generate neutralizing antibodies. For most purposes, however, such an easy solution is not available, so that an experimental laboratory-scale vaccine must be made. High doses of antigen, potent adjuvants, and repeated immunizations will certainly be needed to induce antibody levels that qualify for the intended purpose. Attempts to generate monoclonal antibodies are certainly warranted, but success is not at all guaranteed.
Purification and concentration of the antibodies may be required to avoid excessive volumes that can not reasonably be applied. However, if the concentration is increased, the load of heterologous protein applied to test animals might become more harmful. In addition, attempts to purify and concentrate the antibodies are often not associated with a rise in activity that is proportional to the concentration factor.
Substantial volumes of serum may be required to neutralize the virus that is present in the test samples. Whereas the sample volumes prescribed for animal tests are fairly low, much higher sample volumes (50 mL or an equivalent to 500 human doses, whichever is the greater) (1) are required for cell culture tests. The CBER Guidance for Industry (2) acknowledges that it may not always be possible to effectively neutralize a viral seed. In such cases one may choose alternative strategies, including testing smaller quantities of the seed.
The model calculation in Table II demonstrates that the volumes of antibodies needed may well amount to several liters. The assumed neutralizing serum titers used to estimate the required volumes in Table II are taken from studies attempting to produce high neutralizing antibody titers by hyperimmunization schemes using complete Freund's adjuvant (19). The author's own experience with antiviral hyperimmune sera also confirms that a 50% neutralization of 100 TCID50 at serum dilutions above 1:10,000 is more the exception than the rule. Neutralization tests normally define a 50% neutralization titer. Achieving neutralization by 100% therefore assumes at least a 2-fold higher amount of serum needed.
Applicability of Conventional Tests for Rapidly Changing Viral Seeds
In vitro cell culture tests and in vivo assays are mainly applied to master cell banks and virus seed lots. Depending on the process-specific risks, in vitro testing (but rarely in vivo animal tests) may also be applied for control cell preparations from each manufacturing run and for viral vaccine harvests. Furthermore, raw materials of biological origin may also require in vitro cell culture testing. As explained above, at least 5 weeks must be planned to complete in vitro testing whereas in vivo tests need 22 weeks. These time periods have to be extended by about 40 weeks if no neutralizing antibody preparation is available to neutralize the vaccine virus that is contained in virus seeds. As a consequence, the required materials must be already made during product development so that the full conventional test scheme can be applied for starting materials, such as virus seeds, which are made during development. The same reagents can then also be used later when a new viral working seed must be made. For normal vaccines this may be required at intervals of 5–10 years or longer.
The situation is entirely different for influenza vaccines for which rapidly changing virus seeds must be tested and released. The virus strains to be incorporated into the annual vaccine can change each year, and different strains might even be needed for the northern and southern hemispheres. As shown in Figure 1, selection of vaccine candidate strains by the World Health Organization's (WHO) Influenza Surveillance Network and formal recommendations for the vaccine strains to be used for the northern hemisphere by regulatory authorities normally occurs in February or March. Viral seeds are prepared immediately thereafter (partly even “at risk” before official recommendations are given). This leaves a very narrow time window to manufacture and license the vaccine that is to be distributed in September, at the latest. In Europe, clinical trials are required for licensing so that the earliest use in humans of the new vaccine already occurs in June or July. Furthermore, newly recommended virus strains are selected based on serologic differences from previous strains, so that almost inevitably new neutralizing antibodies must be made for each new vaccine strain. It is obvious that under those conditions conventional in vitro or in vivo testing cannot be accomplished within the available time period. Consequently, cell culture or animal testing has never been considered for inactivated, conventional influenza vaccines produced in embryonated eggs.
Recognizing these time constraints, it has been proposed that standard testing according to PhEur 2.6.16 (1) should be conducted for new cell culture–derived influenza vaccines, even though the “results of such testing may not be completely available before further processing” (20). However, the requirement to conduct conventional testing has been withdrawn, since it has been recognized that no manufacturer would take the risk of making a whole season's vaccine before essential release test results were known, and that even then final results of seed virus testing might only be available after all batches of the vaccine have already been administered to humans. Instead, other and new methods can also be applied to guarantee the viral safety of the vaccine.
Conclusions on Conventional Test Regimes
In vitro cell culture methods, and even more so the in vivo methods, take a long time complete. Their applicability is restricted mainly to starting materials of biological processes that remain constant. Current recommendations must make many compromises when rapid results must be obtained, for example, for biological raw materials, virus harvests, and for certain virus seeds.
Increasing knowledge about the sensitivity and specificity of in vivo tests in comparison to other methods raises doubts about their value to exclude the presence of unwanted viruses. Thus, and also for ethical reasons, animal testing should not be done simply because it has been standard practice for years but rather should be carried out only when justified scientifically. Whereas this might be difficult for standard in vivo tests in mice and guinea pigs, available data indicate that mouse antibody production (MAP) testing for new rodent cell lines might be justified more easily.
In the past there have been only a few alternatives to cell culture and animal tests to exclude the presence of adventitious agents. Reverse transcriptase–based detection of retroviruses is one of those alternatives that has been routinely applied in the past. For the foreseeable future and for specific samples and applications it will most likely remain an inherent part of any broad-spectrum virus testing.
Nucleic acid–based virus detection methods have identified many virus strains, types, and even whole new families. Those newly identified viruses existed long before, but they could not be detected in the conventional assays because they do not replicate in those systems or do not exhibit the properties for the assay read-out. One should probably add that even many field strains of conventional viruses that are assumed to be detected may not be detected by the conventional methods. With those limitations, we cannot rely on conventional methods alone because they are no longer state-of-the-art.
Alternative Methods and Strategies
Alternative virus detection methods have been developed, which are either highly specific for selected viruses or virus families (e.g., PCRs) or for a selected range of virus families (e.g., multiplex PCRs), or which are very broad-spectrum methods that are in the best case able to detect virtually any existing virus (21–23). Such broad-spectrum methods are based on a sequence-independent nucleic acid amplification with subsequent identification of virus encoding sequencing, for example, by massively parallel sequencing, hybridization to DNA arrays, or mass spectrometry identification of virus-specific nucleic acid base compositions.
Whereas conventional methods have never been validated, the introduction of new molecular biological methods requires extensive validation of the new methods according to today's standards. The wider the claimed detection range of the new method, the more demanding the validation will be. Virus specificity can be determined via the method's sequence specificity. Nevertheless, and in the same way as for the old methods, it will be impossible to determine detection limits for any one of the viruses that can be detected. To keep validations feasible, selected examples must be accepted. Under laboratory conditions the new methods may partly be less sensitive than the old methods if a comparison based on infectious units is made and if ideal virus preparations, which contain a minimum amount of inactive virus particles, are used as a comparator. This does not reflect the usual circumstance, in which high numbers of “dead” virus particles of the contaminating virus are more likely. These dead particles will also be detected by the new method and thus lower the detection limit under practical conditions. Furthermore, and considering the virus-reducing capacity of the entire process, the given detection limits may be fully sufficient for the intended purpose.
In practice, the application of the new methods raises more difficulties due to the detection of viral nucleic acids that are not associated with active viruses and that are due to degraded virions or integrated, cellular proviral sequences. A typical example of integrated virus-like sequences are the retroviral sequences found in many cell types. The fact that the new methods do not differentiate between infectious viruses and noninfectious viral or virus-like sequences necessitates that in some cases conventional “infectivity” tests remain relevant and may be used as an addition to modern broad-spectrum tests.
New broad-spectrum methods and, where needed and justified, also conventional methods may be used for products in development, for example, to test new cell lines or viral seeds of standard vaccines. They may also be applied to gain generic information about potential contaminants of certain critical raw materials. In those cases there is sufficient time to follow up on nonspecific or false-positive results that are likely to occur at high frequency. It is tempting to hope for a universal applicability of those new broad-spectrum methods that should be able to detect any virus. However, based on what we currently know about those methods, they are not a suitable option whenever rapid test results are needed, for example, for raw materials, virus harvests, and rapidly changing virus seeds. For those cases a specific selection of PCRs should be applied. The high specificity of PCRs may partly be overcome by designing multiplex PCRs that detect a wider range of virus families. Nevertheless it will be required to justify the range of viruses covered (or not covered) by the PCR methods by risk assessments.
In any case, one should not rely on testing alone. Any test method has a certain detection limit and cannot exclude the presence of viruses beyond doubt. Other methods are similarly important to avoid contamination by adventitious viruses. Such complementary strategies include
Systematic risk assessments to identify and prioritize risks
Growth studies to identify high-risk contaminants
Avoidance of risks by selection and testing of starting materials and process raw materials
Raw material pretreatment for virus inactivation
Virus inactivation by the manufacturing process
Only the adequate combination of the methods and exercises listed above guarantees a virus-safe product. While much attention is already given to raw material selection and testing, the development of methods to inactivate viruses in those materials may be even more effective to avoid virus contaminants. This may be particularly relevant for vaccine processes that are designed to concentrate viruses and to preserve their antigenicity and thus provide fewer options to inactivate viruses by the manufacturing process.
Conflict of Interest Statement
The author of this article is an employee of Novartis Vaccines and Diagnostics GmbH.
Acknowledgments
The author gratefully acknowledges the contribution of independent time estimates for in vivo tests by Hannelore Willkommen, Regulatory Affairs and Biological Safety Consulting, Erzhausen, Germany. These were combined with personal experience to assemble Table II. Specific experience with unexpected side reactions and numbers of animals needed were gathered when in vitro and in vivo studies were done under a contract with Newlab BioQuality AG (now Charles River Biopharmaceutical Services), Erkrath, Germany.
- © PDA, Inc. 2011
References
- 1.
- 2.
- 3.
- 4.
- 5.
- 6.
- 7.
- 8.
- 9.
- 10.
- 11.
- 12.
- 13.
- 14.
- 15.
- 16.
- 17.
- 18.
- 19.
- 20.
- 21.
- 22.
- 23.