Evolution of Approaches to Viral Safety Issues for Biological Products

Early Viral Vaccines

During the first half of the twentieth century, cell culture technology steadily progressed and eventually lead to the landmark efforts of Enders, Robbins, and Weller in 1949 to prepare monkey kidney cell cultures from monodispersed suspensions of minced money kidneys (1). This work led the way to effective methods for (1) propagation of many different cell types from humans and many species of multicellular animals, (2) research studies for isolation and characterization of many types of human and animal viruses, and (3) large-scale preparation of many human viruses for vaccine products. Monolayer cell culture provided a powerful new way to perform virus studies; it ushered in a golden age of animal virology that resulted in the discovery of different families of animal viruses, each with many members, as well as the elucidation of their biology and eventually their molecular biology. Prior to that point, virus isolation and characterization studies relied on use of intact animals or embryonated eggs to support virus replication and to exhibit subsequent pathological effects. This was also true for efforts to produce viral vaccines for human and animal use, which relied on intact animals (e.g., smallpox virus in cattle) or eggs (e.g., yellow fever virus in embryonated hens' eggs) for virus propagation.

Production processes for early viral vaccine products were based on four steps: (1) isolation of the viral agent in eggs, animals, or eventually cell culture; (2) controlled subcultivation or passage of the agent in eggs, animals, or cells for attenuation or selection; (3) attenuation and/or propagation in eggs, animals, or cells; and finally (4) formulation, fill, and finish. Control measures were limited to assessment of vaccine seed purity, purity of the propagated virus, and confirmation of virus inactivation or attenuation (all based on use of the same in vitro and in vivo propagation methods used to isolate and grow the viral agent), as well as visual observation of uninoculated control cultures or sentinel animals, and confirmation of fill-finish process integrity by sterility testing.

Availability of these viral vaccines helped accomplish significant advancement of public health (such as the global eradication of smallpox and effective elimination of indigenous transmission of non-imported measles in the USA) but were not without issues, including viral safety. The observation of primary cells isolated directly from intact animals (and used for vaccine preparation) led to the discovery of dozens of new viruses indigenous to their animal host. The simian virus (SV) series of serially isolated agents from monkeys used for vaccine preparation numbered about 75 isolates discovered over about 25 years (2). Several of these indigenous animal viruses reached humans through viral products, including SV40 in inactivated poliovirus vaccines given to tens of millions (3) and avian leukosis virus (a type C retrovirus) in attenuated yellow fever vaccines given to hundreds of millions of recipients (4). Reagents derived from biological sources used to prepare cell cultures (e.g., enzymes, serum) also introduced viral contaminants. Many of the early childhood attenuated vaccine preparations were contaminated with bacteriophage originating in the bovine serum collected in abbatoirs (5). Fortunately, none of these three contamination events have been associated with disease in human recipients of these vaccine products (3–5).

Other reports of viral infections due to animal-sourced materials used for vaccines did not have benign outcomes. Primates chronically infected with herpesvirus type B proved to be capable of transmitting the infection to animal handlers or tissue culture workers, producing severe encephalitis with numerous fatal outcomes (6). One incident involving monkeys collected for vaccine production led to transmission of the filovirus, Marburg agent, to humans at several European sites (7). Fortunately, these agents were not transmitted to product recipients. But the explosive nature of some of these episodes, and the grave diseases which resulted, left a permanent residual note of caution on the use of biologically derived raw materials.

Inactivated vaccines have been quite successful in some applications, including initial success in achieving reduction of poliovirus incidences in the 1950s, and in efforts for post-exposure prophylaxis for rabies. But both of these successes have also been associated with unfortunate accidents in which virus preparations that were not completely inactivated were administered to humans who became infected and suffered viral disease. In the three incidents involving incompletely inactivated poliovirus vaccines about 25,000 vaccine recipients were exposed, of whom about 260 became paralyzed and 10 died (8). In the rabies incident, all 18 recipients of the incompletely inactivated vaccine perished (9). These incidents dramatically emphasized the usefulness of tests for residual infectious virus (which had not yet been employed for lot release), and in the case of the poliovirus vaccines, established a key principle in validation of viral clearance—that one cannot claim more than one can measure (i.e., extrapolation is not prudent).

The early attenuated viral vaccines also gave rise to products in which the vaccine virus seed proved to be less than completely attenuated. This has always been a difficult question in the development and deployment of such products—in the face of endemic disease in the general population with measurable deleterious effect, how much attenuation is acceptable, or conversely, how much virulence is unacceptable? Possibly the best known of these cases is the ability of attenuated Sabin poliovirus type 3 to cause paralytic disease in approximately one per 1,500,000 vaccine recipients due to genetic reversion to virulence (10). The strain of attenuated mumps vaccine seed (Urabe), used widely in Europe, was discontinued after a number of years due to an increase in the reported rate of neurological complications (11). Several experimental vaccines for Venezuelan equine encephalitis virus (12) and dengue virus type 2 (13), developed by the U.S. Army, were believed to be adequately attenuated for human use based on biological characterization in vivo and in vitro, but produced disease in human volunteers resembling the wild-type infection. The lesson here is that there are special issues of retained virulence and/or reversion to virulence when dealing with live attenuated vaccines, and these issues call for extensive biological characterization and continued vigilance of exposed populations receiving such products. The gradual improvement in knowledge and technology has brought an increasing ability to understand these events and their mechanisms for inducing disease, for example, the issue with Sabin type 3 virus was due to genetic reversion in the viral genome, and that most single-stranded RNA viruses have error-prone replicase enzymes, which make reversion more likely upon extended viral propagation (14). But the improvement in technologies in recent times has not always been able to prevent unexpected events, such as integration by vector virus resulting in the occurrence of several cases of malignant thymoma in children with severe combined immunodeficiency disease that were given an investigational gene therapy product (15).

Formulation of early vaccines typically used human plasma derivatives such as human albumin. Prior to the eventual optimization of the Cohn fractionation process in the 1950s and associated pasteurization treatment, these plasma derivative preparations were often contaminated by various human viruses, including hepatitis B virus (HBV) and measles (16). In the decades prior to the availability of vaccines to prevent these diseases, there were typically tens of thousands of cases of HBV and over 100,000 cases of measles reported annually in the USA, so contamination of plasma donations and plasma pools was probably commonplace. The largest known outbreak occurred following the stabilization of yellow fever vaccine given to Allied military personnel during World War II, in which 20,000 became ill and 26 died among 2.5 million vaccine recipients (17). Early viral vaccines were also filled using primitive aseptic processing technology, without modern controls on air handling, gowning, air quality monitoring, barrier technology, and other features currently accepted as standards for this type of manufacturing process. Thus, there were many reported cases of bacterial and fungal contaminations of early vaccine products, as summarized by Wilson (16). Not reported is whether lack of such critical controls also led to viral contamination from environmental or personnel sources, but clearly the means to prevent them did not exist.

Unquestionably, the early viral vaccines of the 1930s to the 1960s were major advances in public health, saved the lives of millions, and were usually adequately safe for their intended use. Among the important learnings from these efforts were that raw materials and source materials could be a significant source of viral contamination, sometimes with terrifying results. Another important lesson was that testing for viruses with both in vivo and in vitro methods possesses a degree of sensitivity, which could be useful as a measure of product and process safety, but it was also limited by practical considerations of sample volume per test and amounts of material that could be tested, by the host specificity of the contaminating virus(es), and by the fact that there could be species differences in susceptibility to viral infection between humans and laboratory animals, even primates. It became very clear that virus inactivation studies were useful only within the context of the actual data observed and the test systems used, and that extrapolation beyond the actual data was not a prudent practice. Finally, early accidents and incidents with aseptic processing technology used to manufacture the final product container used by health care providers showed a clear path for future improvements, both in terms of manufacturing technology as well as in the development of what is known today as good manufacturing practices, or GMP regulations.

Early Plasma Derivatives

A related family of learnings came from the early use of plasma derivatives. These were frequently observed to be contaminated with undetected human viruses, which were likely to be associated with the collection of source material from the human donors. Donation screening by laboratory testing became feasible in the 1950s with the development of credible test methods and laboratory standards. About the same time, Cohn fractionation based on acidification and use of ethanol provided a modicum of viral clearance for the later fractions, and a reasonably standardized method of manufacturing. Introduction of pasteurization (treatment at 60 °C for 10 h) was possible for normal serum albumin and some other later fractions, which provided additional viral clearance. Once these testing and manufacturing methods were introduced, especially with the introduction of HBV serology in the 1970s, significant safety from viral transmission risks of later Cohn fraction-derived products such as albumin and immunoglobulins was achieved (18, 19). The use of such materials as vaccine excipients or as therapeutic biologicals contributed to the significant disappearance of the types of viral contamination events described above. Since the 1960s, albumin has not been reported to be associated with transmission of any viral disease, despite being administered in large quantities (up to 100 grams per patient) to millions of patients annually. Similarly, immunoglobulin preparations have not been associated with transmission of viral diseases since the 1930s—with the exception of hepatitis C virus (HCV) transmission associated with use of anti-HCV antibody screening to remove antibody-positive units from plasma pools, a prudent precaution with unintended consequences (20).

However, viral transmission risks still existed for early Cohn fraction-derived products that did not receive much viral clearance from the processing, and which were thermolabile and therefore could not be pasteurized. Products like the anti-hemophiic factors (Factor VIII and Factor IX) were derived from cryoprecipitate (Cohn fraction zero) and still contained live viral contaminants (21). As a result, the life expectancy of hemophilia patients in the 1950s to 1970s was relatively short, as they experienced multiple viral infections associated with the biological products used for their therapy. But as testing capabilities improved during the 1950s to 1970s, and more and more viruses and other agents could be excluded from the donor pools by screening, the life expectancy of hemophila patients continued to improve. A major advance was the availability of HBV tests in the 1970s, which greatly reduced the number of exposures to HBV (19). In the 1990s, significant advances came from the use of new polymerase chain reaction (PCR) technology for nucleic acid–based tests for infectious agents that were difficult to propagate, such as HCV, a major cause of disease associated with use of blood and plasma derivatives (22), and parvovirus B-19, associated with fifth disease (23).

Between 1980 and 1984, approximately 19,000 US hemophilia patients were infected with newly emerging human immuodeficiency virus type 1 (HIV-1), and by 1995, when effective antiretroviral therapies became widely available, most had died. An entire generation of hemophilia patients was infected by an infectious agent for which there was no screening tool and no effective therapy, showing the vulnerability of products that were without much viral clearance capability, to transmit unidentified/emerging viruses. The rapidity of HIV spread throughout the population globally was also profoundly shocking. While the scope of this problem in hemophilia patients was greatest in the USA due to widespread treatment with pooled cryoprecipitate products, it had important impacts elsewhere as well. In France in the early 1990s, several Ministry of Health officials were accused of manslaughter for delays in approval of effective HIV detection kits for blood and plasma donations, which resulted in hundreds of French citizens becoming infected with HIV (24). In Japan, the heads of the plasma derivative companies pled guilty to charges of professional negligence and were forced to publically apologize to the Emperor of Japan for approximately 400 deaths, which occurred among hemophilia patients there (25).

Lest anyone believes that HIV-1 was a unique and unparalleled historical event that could not happen again on that scale, we should be reminded that dozens of emerging viruses and other infectious agents have been discovered or recognized since about the time of the discovery of HIV-1. Among these have been human T-lymphotropic virus types I and II (HTLV I and II), West Nile virus, Chagas disease (caused by a parasite), and new variant Creutzfeld-Jakob disease (nvCJD, caused by a transmissible spongiform encephalopathy agent). The explosive spread of HIV-1 in the US was more recently matched by that of West Nile virus, first diagnosed in New York City in the summer of 1999 (26) and by 2003 had reached all the continental US states except Oregon and Washington (27). The virus spread rapidly epizootically, and it infected humans only accidentally, being detected in blood and organ donations (28). PCR technology was used to detect the presence of viral genome in clinical samples (26), and a test for blood and plasma screening was quickly developed, validated, and deployed.

Plasma derivatives proved to be a source of life-saving therapeutic biological products and important reagents such as for blood and tissue typing, but like cells and tissues, the source materials could be contaminated by viral agents that may be transmitted to the product recipient. Screening donors for potential viral contaminants provides substantial control and prevention, but these measures are limited by the sensitivity of the methods employed, the natural history of the particular infection and host immune response, the emergence of novel viral agents or changes in the epidemiology of existing agents. Analytical methods can ultimately only be expected to detect what they are designed (and validated) to detect, and are never foolproof or 100% effective. Pooling of large numbers of donations was also seen as a major risk factor, and today pool size is limited to a maximum of 40,000 donations in an effort to limit exposure risks to rare agents like nvCJD.

Modern Biopharmaceuticals

Development of the first clinical products from modern biotechnology came along in the late 1970s and early 1980s with murine monoclonal antibodies and the first recombinant DNA (rDNA) products. These efforts benefited enormously from the lessons learned from the development of early vaccines and plasma derivatives, and they were occurring as the HIV story was unfolding. In addition, use of these technologies required the utilization of continuous cell lines and, therefore, required the U.S. Food and Drug Administration (FDA) and counterpart health authorities outside the USA to revisit their ban on the use of materials of tumor origin or tumorigenic nature for the preparation of biological products. Continuous cell lines are capable of immortal growth and tumor formation in immunosuppressed animals, typically have aneuploid karyotypes, and frequently express endogenous retroviral genes and/or retrovirus-like particles.

By late 1982, the FDA became aware that rDNA-derived therapeutic proteins were expressed in continuous mammalian cells that were tumorigenic in immunosuppressed rodents, and that the cells expressed retrovirus-like particles (29). Because the cells themselves could not be made safe, the approach to ensure product safety was to ensure that neither the cells nor anything from the cells associated with endogenous retrovirus-like particles or cellular nucleic acids ended up in the product (30, 31). Gene cloning technology and modern protein purification methods were extremely important in helping to provide this assurance. It was also clear that cell bank characterization, clearance of cell protein and cellular DNA, and clearance of endogenous retroviruses were important elements in providing this assurance for product safety. Finally, the role of GMP manufacturing controls (both procedural and engineering) and adventitious virus testing was also acknowledged as an important part of the product safety strategy. When the data became widely available a few years later from leading rDNA manufacturing firms, and was reassuringly consistent from firm to firm, the health authorities universally accepted the adequacy of this approach to assurance of product safety from continuous cell line risks and endogenous retroviral risks, and approval of these new products occurred worldwide.

As part of this effort, scientists at Genentech evaluated the capability of Chinese hamster ovary (CHO) cells to support replication of a variety of human viruses. They found that most (except parainfluenza viruses and reoviruses) did not replicate in CHO. They also reported that even the viruses that did replicate in CHO cells were cleared by the same purification process steps that were used to effectively clear endogenous retroviruses. Additionally, in vitro virus screening assays could also be employed to detect the viral agents that were capable of replication in CHO cells, and would serve as a further step to ensure that viral agents capable of replication and pathogenesis in humans would not be transmitted to product recipients (32). This set of strategies was very effective in preventing human viruses from contaminating the production systems and the products, but it overlooked the possibility of exogenously introduced rodent viruses that may not be found in the cell bank but could replicate in CHO cells upon infection. This is discussed further below.

Besides reliance on traditional virus detection technologies like in vitro infectivity assays and viral clearance technology, viral safety for early rDNA products also relied heavily on microfiltration technology. Early applications of rDNA technology in mammalian cells typically relied on bovine serum for cell growth, and thus the cells were potentially exposed to bovine infectious agents. In the late 1970s, the development of reliable microfilters capable of 0.1 micron nominal particle removal made it possible to remove large bovine viruses such as bovine rhinotracheitis (a herpesvirus) and parainfluenza type 3. Contamination of large-scale cell culture facilities with these two agents was not reported after the development and introduction of 0.1 micron filters, whereas there was widespread occurrence of these agents in bovine serum collections not processed with such filters.

An extension of this concept was practically achieved for smaller viruses in the early 1990s and applied initially to plasma derivatives. The discovery of parvovirus B 19 in human plasma (33), and occasional contamination of cryoprecipitate products with picornaviruses like hepatitis A virus (HAV) (34) led to the desire for filters capable of removing small viruses from plasma derivatives. The introduction of nanofilters with nominal porosity in the range of 20–40 nanometers showed considerable ability to remove the smallest known infectious viruses, and such nanofilters were rapidly deployed in the plasma derivative industries. Health authorities urged their adoption for rDNA products as well, in part due to non-specific concerns for contamination by small viruses, and in part due to the emergence of rodent parvoviruses as described below.

In the early and mid-1990s, scientists at Genentech described several incidents in which minute virus of mice (MVM, a parvovirus) contaminated 12,000 L CHO cell cultures. The contamination was discovered by the in vitro cell culture adventitious agent assay put in place to detect human viral contaminants putatively present in preharvest samples. In response to the first of these MVM contamination events, Genentech also put in place a PCR-based screening assay for preharvest samples to avoid the need to decontaminate purification facilities (35). A second incident occurred a few years later, and the PCR-based screen functioned as designed to limit the MVM contamination to a single, large-scale cell culture vessel (35). This example is analogous to the use of PCR technology to screen for HCV and later West Nile virus in blood and plasma, thereby preventing transmission of the viral contaminant to the ultimate product recipient.

A few years later, the emergence of mad cow disease in cattle (and later, its associated human form nvCJD) led to renewed concerns from health authorities regarding use of animal-derived raw materials to prepare biopharmaceuticals and vaccines, particularly the use of serum to propagate mammalian cells (36). This led to efforts to either remove animal-derived raw materials where possible, or to limit their use—for example, from sources from geographical regions with low/no incidence of transmissible spongiform encephalopathy (TSE)-based diseases—in cases where it was not possible or practical to use animal-free raw materials. With rDNA-derived products from mammalian cells like CHO cells, this kind of removal was generally possible (37). Many continuous cell lines can be propagated consistently in media containing recombinant growth factors, or can be rendered independent of exogenous sources of growth factors. For vaccines, removal was not always possible, so limitation of exposure and geographic controls were more common. Even where replacement of serum with rDNA-derived growth factors was possible for some vaccines and cell therapy applications, serum-derived attachment factors were typically still required for cell attachment to surfaces for adherent cell populations. In addition to serum for cell growth, analogous considerations also apply to other animal-derived raw materials, including enzymes for cell dissociation or detachment, lipid sources for nutrition, and other applications.

The use of screening technologies based on infectivity and/or nucleic acid methods was inadequate to assure the absence of viral contamination in raw materials, and inadequate to prevent contamination of cell cultures that used these materials. The degree of assurance provided was too nominal to be useful (35). Contamination of a number of GMP manufacturing sites were reported, including epizootic hemorrhagic disease of deer virus (EHDV, an orbivirus, 38), Cache Valley virus (CVV, a bunyavirus, 39), and more recently Vesivirus 2119 (a mammalian coronavirus, 40). These are possibly associated with the use of bovine serum. Other types of agents associated with other raw materials include Spiroplasma (a member of Mollicutes, 41) from plant hydrolysates and porcine parvovirus from trypsin (42). Current thinking on best practice if such raw materials must be used is that the material should be screened for the presence of relevant adventitious agents, and in addition, the raw material or the in-process material containing the raw material should be treated in a robust manner (e.g., gamma irradiation) to significantly reduce the concentration of the suspected adventitious agent.

A further refinement of this best practice concept is seen in events following the second MVM contamination at Genentech. Although the PCR-based detection technology prevented the MVM contamination from reaching beyond the cell culture process, something else was clearly needed to prevent MVM cell culture contamination from occurring every few thousand bioreactors inoculations. Given the ubiquity of mice in the environment, the transmission of the virus in nature through feces and urine, and the large amount of raw materials needed for 12,000 L cell cultures, screening raw materials was dismissed as inadequately insensitive and impractical. Instead, an engineering solution was deployed called high-temperature short-time (HTST) treatment of cell culture media prior to cell culture. Effectively, this introduced a heat barrier that inactivated substantial amounts of MVM (43). While it is possible to estimate how effective this procedure is at inactivation of known amounts of MVM, it is still difficult to estimate the putative MVM load in potentially contaminated media, or how often the MVM contamination occurs (35). Nonetheless, Genentech reports that it has not experienced a subsequent MVM contamination event after the first two occurrences in the early and mid-1990s (R. Kiss, personal communication). Some other firms have applied this same technology for similar applications. However, HTST technology is not compatible with serum-containing media. Irradiation of complete serum-containing media with UV-C has been reported to be effective by Amgen scientists (44).

Viral safety of rDNA-derived products from cell culture has been accomplished by an integrated approach of multiple systems and controls, derived from lessons learned from the vaccine and plasma derivative industries, more recent experiences, as well as applications of newer technologies. These approaches include (1) the use of cryopreserved banked cells rather than primary cells or intact organisms that can be tested prior to the release of the cell banks for manufacturing use; (2) testing programs for preharvest samples by infectivity in vitro for general viral contamination, and newer PCR-based methods that can be very effective for specific agents; (3) fractionation and purification to achieve viral clearance from active ingredients by chromatographic separation, inactivation, and filtration (micro and nano); (4) in some cases, prevention of viral contamination by exposure of raw materials to inactivation procedures (irradiation, heat); and (5) integration with prudent quality systems and risk assessment principles, knowledge of the putative infectious agents, knowledge of the physical and chemical properties of the active ingredients, and the practicalities of production.

Summary

The lessons of history amply demonstrate that viruses are ubiquitous in nature and may contaminate biological products unless proper precautions are taken. Without such precautions, contaminations can occur, sometimes with horrific consequences. In the beginning of the vaccine and plasma derivative era, there were few tools available for detection and limited understanding of where viruses came from or how they might be transmitted. For drug products based on thermostable small molecules, terminal sterilization is feasible, and provides a simple, reliable, one-step procedure to prevent viral contamination of such products. But the thermolability of biological products makes terminal sterilization impossible; thus, viral safety assurance for large-molecule products is built upon a patchwork of less robust technologies. Detection of contaminants alone is inadequately sensitive for the task, and is therefore ineffective by itself. No single precaution is foolproof, but an integrated combination of them can be very effective. These integrated approaches have steadily evolved over the past 80 years, beginning with the development of early viral vaccines and plasma derivative products, learning from each unfortunate accident and each new technology. No rDNA product derived from mammalian cell culture has been associated with transmission of a viral safety problem to product recipients in the past 24+ years of marketing of such products. Occasional viral contamination events have been detected in manufacturing processes for these products (approximately every 1–2 years). Most of these rare events seem to be preventable or partially mitigable through integrated approaches currently available.

Where do we go from here? The industry as a whole has implemented new process and analytical technologies, over time, that meet established needs and provide an improving level of viral safety for the process as well as the product. But in the face of new and emerging technologies, it is important to ask a series of complex questions. What standards are appropriate

• To determine when a new technology is ready for “prime time?”

• To determine when a need is established, and whether all organizations/products have the same needs?

• To introduce new technologies into investigational products?

• To introduce new technologies into licensed/approved products?

Additionally, what is the appropriate expectation for GMP compliance for new technologies? Given that substantial time is required to develop some of these technologies or to establish their relevance (and understand their limitations)—or for process technologies, to understand their effect on product stability—what is an appropriate time frame in which to expect adoption of a proven new technology to occur? We will hopefully begin to find some answers to these questions.