Practicing Safe Cell Culture: Applied Process Designs for Minimizing Virus Contamination Risk

Genentech's Virus Contamination Events

Genentech experienced two industrial-scale contaminations of a Chinese hamster ovary (CHO) cell process with a virus subsequently identified as the mouse minute virus, or MMV (formerly known as minute virus of mice, or MVM). MMV is an extremely small (about 20 nm diameter) parvovirus that is highly resistant to inactivation by physical (e.g., temperature) or chemical means, and it is commonly harbored by mice that shed particles in their urine and feces. In 1993, multiple 12kL production bioreactors were determined to be infected with a virus by a routine in vitro virus screen conducted on each production batch. However, as that virus screen testing is cell-based, it takes approximately 2 weeks for the test to conclude. Because of the time delay, contaminated cell culture fluid had been carried forward through the purification process, and the entire drug substance manufacturing facility was likely exposed to significant numbers of virus particles. Accordingly, the facility was shut down and required extensive cleaning and decontamination (including para formaldehyde fumigation) prior to the restart of operations. Given the nature of the contamination and the extent of exposure throughout the facility, significant amounts of raw materials and biologics product were discarded. An extensive root cause investigation ensued, which included screening of operations staff for evidence of significant rodent exposure outside of the workplace, evaluation of the genealogy of the affected bioreactors, significant testing of cell culture processing equipment and facilities for the presence of MMV, as well as significant testing of the cell culture raw materials used in the process (cell banks, media components, etc.) for the presence of MMV. Although no raw materials were observed to be contaminated with MMV, the investigation concluded that the pattern of contamination was consistent with the probable source being a cell culture raw material (4). A number of corrective actions to improve processing practices were implemented prior to the restart of the facility. These included enhanced controls on incoming raw materials via improved cleaning of container external surfaces, improved gowning practices and segregation of staff conducting media preparation and cell culture operations, enhancement of the compressed air (for sparging cell cultures) generation and distribution system to increase virus inactivation and removal efficacy, enhancement of the pre-harvest virus testing program to provide additional analytical barriers to culture and facility contamination, and implementation of a standard procedure to be executed by operations staff in the event of a bioreactor virus contamination. Specifically, the compressed air generation system was modified to ensure that the air exiting the compressor would be exposed to not less than 150 °C for not less than 90 s, and the distribution system was enhanced to ensure that all process gases were filtered through sterilizing-grade filters rated (when dry) to remove particles as small as 10 nm. The pre-harvest virus screening program was enhanced in two ways. First, an additional mammalian cell line (324K human kidney cell line) more sensitive to parvovirus was implemented into routine in vitro testing. Second, a polymerase chain reaction (PCR) assay for murine parvovirus was developed and implemented to allow for a rapid test to be conducted on pre-harvest cell culture fluid as an in-process hold step, thereby preventing a contaminated culture from being harvested and spreading the contamination throughout the facility. In addition, technical development work was initiated to evaluate the feasibility of heat-treating cell culture medium as an additional virus barrier that could provide a comprehensive point-of-use protection applied to raw materials entering a bioreactor, though it was known that such media heat treatment barriers would not likely be possible to implement into good manufacturing practice (GMP) manufacturing prior to the need to conduct the next product supply campaign.

With this set of enhanced practices and analytical barriers in place, another production campaign of the same biologics product was initiated about 1 year later. During this campaign, a single 12kL production bioreactor was determined to be contaminated with MMV via the pre-harvest PCR test for parvovirus. Because the contamination was detected prior to harvest, the culture and bioreactor equipment were decontaminated prior to breaking containment, the contamination was limited to the single bioreactor, and there was no impact to the downstream processing area of the facility. Accordingly, the facility was able to continue operating while an investigation ensued. Interestingly, the MMV strain isolated was different than the strain isolated in the 1993 contamination, and the genealogy of the affected bioreactor demonstrated that the source of the contamination was not likely the continuously passaged cells used as the inoculation source for multiple production bioreactors, as only one of such multiple bioreactors developed a contamination. The investigation concluded, as in the 1993 contamination event, that the most likely source of contamination was a heterogeneously distributed virus presence in a cell culture raw material. Based on this conclusion, and given both internal and regulatory agency concerns with the likelihood of continued virus contaminations, a decision was made to implement at manufacturing-scale the virus barriers based on cell culture medium heat treatment that had been under development as soon as possible.

Assessing the Risk of Virus Contamination

Genentech experienced two separate contaminations of “Process B” separated in time by more than 1 year, and in a facility that had successfully produced “Process A” for multiple years as well as in between the two observed Process B MMV contaminations. If one considers the 1993 contaminations as one precipitating event that spread to multiple bioreactors, then the observed contamination frequency was one contamination per 38 Process B runs (two events over 75 total Process B runs), clearly far too high for sustainable manufacturing. The production facility had manufactured nearly 1000 total 12kL production runs in its history, nearly 700 of those runs with Process A. If one considers the overall facility runs, the observed contamination rate was about 1 in 500 runs, still considered too high! Both Processes A and B utilized cells derived from the DUX B11 host CHO cell line, and both also utilized highly similar raw materials, including bovine serum, bovine insulin, and bovine apo-transferrin. Given the significant difference in observed contamination rate between the two processes, the question was asked as to whether there was a difference in susceptibility of the two CHO production cell lines to murine parvovirus infection. A study was conducted to answer this question, with four different production cell lines inoculated with three different strains of MMV. The study results are summarized in Table I, where it is clearly seen that the cell lines do not all show the same sensitivity to infection. In fact, it is seen that the Process A cell line was not sensitive to any of the MMV strains, whereas the Process B cell line was highly sensitive to all three MMV strains. Interestingly, the other two production cell lines showed significantly higher sensitivity to the MMV strain that was isolated from the 1994 contamination when compared to two MMV strains obtained from an academic laboratory (5). Even more interesting, as shown in Table I, is the apparent correlation between the sialic acid content (specifically Neu5Ac) of the CHO cell membrane proteins and the sensitivity to infection. The Process A cell line contained significantly lower (>10 fold) Neu5Ac content, and was observed to be insensitive to infection by all three MMV strains. These results potentially explain why there had never been a detected parvovirus contamination across nearly 700 runs of Process A. All cell lines are not created equal!

Figure 1 illustrates the MMV infection kinetics as observed in the 12kL production bioreactor contamination event of 1994, as measured by PCR. It is seen that the contamination was not detected prior to a 96-h sampling point, with the PCR limit of detection validated to be 0.1 TCID₅₀/mL, or 2000 copies per milliliter (one copy per particle), at that time. The infection kinetics are observed to be extremely rapid, with amplification from less than detectable to 10⁵ copies per milliliter in about 24 h, and to at least 10⁷ copies per milliliter in about 48 h. The minimum required multiplicity of infection (MOI) to initiate an infection has been shown to be as low as 10⁻⁹ by two separate studies (6, 7). Given this information, it is a sobering result to derive that even less than one MMV particle per liter of cell culture medium is sufficient to generate an infection under typical CHO culture process conditions, and that the infection could easily originate from the particles being introduced at the final production culture stage (the largest volume of medium). Based on this assessment, and the investigations concluding that the likely source of infection was through a raw material, support is indicated for virus barrier approaches targeting the large quantities of raw materials used in cell culture processes.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

MMV infection kinetics as measured by PCR testing from an infected 12kL culture during the 1994 campaign.

When considering raw material usage in cell culture, there are four main classes of materials used—the cell bank vial, the process gases (especially air), the water used to prepare culture media, and the media components (amino acids, sugars, salts, proteins, protein hydrolysates, etc). The cell banks used in the biologics industry for GMP purposes are exhaustively tested for the presence of adventitious virus per regulatory guidelines and are, thus, considered to be a relatively low risk of being contaminated. As mentioned previously, process gases (e.g., air, oxygen, carbon dioxide) can be readily filtered to reduce the likelihood of virus particle presence. Compressed air is typically generated on-site using conditions likely to inactivate viruses, or can be modified to provide inactivating conditions. Therefore, the risk is likely low for these gaseous raw materials to introduce a virus contaminant. However, the cleanliness of the air intake supply should always be considered given the potential for rodents to shed viral particles. Do you know where your intake air supply that goes to your bioreactors comes from? Cell culture media for GMP applications is typically generated in highly purified water obtained via a number of purifying steps that are likely to include a high temperature treatment step (e.g., distillation) and subsequent storage at elevated temperatures in excess of 80 °C. Thus, water for media is considered a low risk. The highest risk type of raw material is believed to be that of the chemical constituents of the media and other solutions used to promote and sustain cell growth and recombinant protein productivity. This class of raw materials will typically include dozens of different components obtained from multiple sources, suppliers, and supply chains. Given that rodent-derived virus particles, such as MMV, are essentially environmental contaminants, any raw material could become contaminated based on its potential exposure history from the original source raw materials, through its manufacturing history, and through its supply chain history. Many elements of the exposure history are very difficult for the end user, the biologics manufacturer, to identify or control.

Table II illustrates a semi-quantitative ranking of the risks of adventitious agent contamination of media raw materials. This approach ranks from low to high when considering the raw material original source, raw material manufacturing process, the raw material supply chain, and the amount of raw material used in the biologics manufacturing process. The highest inherent risk attributed to original material source is that of a raw material derived directly from an animal source (e.g., serum, animal tissue hydrolysate), followed by other materials of biologic origin (e.g., plant tissue hydrolysate, sugars derived from plant matter). From a raw material manufacturing process perspective, materials generated in a process that includes appropriate heating steps would clearly be lower risk. The raw material supply chain is an extremely important consideration given that many steps may be involved in the supply chain, with most of these out of the view and control of the end user. For example, if proper cleaning and segregation of crude versus processed source materials does not occur, risks are increased. If the raw material represents a rodent food source, the risks inherent in the supply chain are also likely increased. Finally, given the likelihood that virus particle contamination of raw materials is a heterogeneous representation within a given solid-based material, the risks associated with using a given material are proportional to the amount used in the cell culture process. Considerations such as those illustrated in Table II further convinced Genentech that the highest impact risk mitigation strategy was to provide an efficacious virus barrier at the point of use in the manufacturing facility.

Upstream Virus Barrier Technology Options and Assessment

Various technologies may be considered when evaluating potential virus barriers for cell culture raw materials. Table III reviews four technologies—high temperature short time (HTST) heat exposure, UV light exposure, gamma irradiation, and filtration—that were considered by Genentech as part of the virus barriers development effort. The technologies are characterized in terms of their virus clearance efficacy and their suitability for large-scale processing. It is seen from Table III that HTST heat exposure is a broad-spectrum and highly effective inactivation method that is also straightforward to implement for large-scale, point-of-use applications. The other technologies are generally considered currently less desirable compared to heat treatment for large-scale processing. However, it is important to recognize that not all media raw materials may be compatible with heat treatment at temperatures required for broad-spectrum virus inactivation, and it may be necessary to utilize one or more of the other technologies to provide the best overall solution for a given cell culture process definition. Additionally, for small-volume applications, some of the non-heat technologies may be operationally simpler to implement.

Development and Implementation of Additional Cell Culture Process Virus Barriers

Based on the barrier technology assessments, Genentech undertook an evaluation of the Process B barrier needs. This evaluation included testing of the virus clearance efficacy of both heat inactivation and virus-retentive filtration when using Process B cell culture media and solutions. Additionally, the compatibility of cell culture media and solutions with heat treatment and virus-retentive filtration was tested extensively. The compatibility of the bovine insulin and apo-transferrin proteins and protein solutions with gamma irradiation and virus-retentive filtration was also tested based on an initial concern that these proteins may not survive heat treatment. Irradiation of serum as a raw material component had previously been demonstrated. But, given the fact that inactivation efficacy was known to be virus-dependent (with the smaller viruses more resistant), serum compatibility was evaluated using HTST treatment as part of a completed medium solution.

Given that parvovirus was known to be one of the most physico-chemically resistant viruses, and that it is one of the most likely contaminants of a CHO-based process, it was selected as the model virus for evaluation of the efficacy of HTST treatment. Figure 2A provides data from HTST inactivation studies with MMV. Serum-containing (10% v/v) cell culture media samples (with and without virus spikes) were rapidly heated to the target temperature by flowing through a heating tube within a hot oil bath, held at that temperature by flowing through an insulated retention coil for up to 10 s, and rapidly cooled by a refrigerated water bath. It is seen that, at 97 °C, complete inactivation (0.7 log₁₀ TCID₅₀ limit of detection) is obtained under all time conditions. The zero time point means that the act of heating to 97 °C and immediately cooling was sufficient to provide complete inactivation. At both 87 °C and 92 °C, surviving virus was detected under some conditions. Similar inactivation results were obtained with 2% serum-containing medium and serum-free medium at 102 °C over the range of hold times. These limited kinetic data were used to construct an approximate kinetic model of MMV heat inactivation using Arrhenius modeling. With such an Arrhenius relation, it was possible to predict virus inactivation over a range of time and temperature exposures for a HTST system, as shown in Figure 2B. An extremely conservative inactivation design target of 15 log₁₀ reduction value was selected based on it allowing a single virus particle to survive HTST treatment only once every 100,000 10kL batches of medium. From Figure 2B, it can be seen that a treatment temperature of 102 °C meets the design inactivation target at less than 1 s of exposure. On the basis of this modeling work, a design target of 102 °C and 10 s exposure was selected for evaluation of media compatibility studies. This target condition, if compatibility were demonstrated, would provide an extremely robust inactivation process flow that could be easily implemented in a large-scale setting.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

(A) MMV heat inactivation kinetics (assay detection limit = 0.7 log TCID₅₀/mL). Data courtesy of C. Fautz (Genentech unpublished results); (B) HTST retention time design modeling. The predictions at 102 °C were based on a kinetic death constant extrapolated from an Arrhenius equation based on data from 87 to 97 °C.

Because bioreactor pH control is accomplished through the regulated addition of fairly significant quantities of a pH control reagent directly to the bioreactor, it was necessary to assess the provision of a barrier for the titrant. Given the inherently high pH (pH 11–12) of the 1N sodium carbonate titrant, it was postulated that the titrant might serve as its own barrier. This was confirmed via the testing shown in Figure 3, where complete inactivation of MMV was observed after about 30 min of exposure to the titrant. Given that titrant is prepared well in advance of process use, no further barrier needed to be considered for this significant process addition.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

1N sodium carbonate inactivation of MMV. Data courtesy of S. Liu (Genentech unpublished results).

Finally, Genentech sought to address the question as to whether a contaminated bioreactor could lead to facility contamination by the release of virus particles through the reactor off-gas (vent) line. To answer this question, the two types (suppliers) of Teflon (polytetrafluoroethylene, PTFE) vent filters used by Genentech were evaluated for their ability to retain MMV particles carried out of a sparged bioreactor as aerosols. To test this, 2 L bioreactors were used to generate high-titer, MMV-infected CHO cultures. The off-gas was collected on an aerosol particle collection impinger both in the presence, and in the absence, of the PTFE vent filters. Figure 4 shows the testing results, where PCR technology was used to show that both vent filters removed at least 4 log₁₀ of MMV particles from the off-gas stream.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Vent filter MMV retention study. Data courtesy of J. Trask and S. Liu (Genentech unpublished results).

For a small-volume, alcohol-based addition, which was obviously incompatible with heat treatment, Genentech collaborated with Millipore Corporation to develop a custom “high-flux” version of the Viresolve-70 ultrafiltration membrane in a normal flow configuration. This filter was shown to deliver greater than 3 log₁₀ reduction value for MMV in the organic solution (data not shown). For small-volume reactor additions that were heat stable and could be heat-sterilized (i.e., 121 °C for 30 min), no assessment of virus inactivation was required. This applied to antifoam, concentrated glucose, and a trace metals addition.

For the study of media and solutions compatibility with the cell culture virus barriers, the main focus was on evaluation of HTST treatment of any media or feed solutions delivered to a bioreactor, as that offered the best point-of-use approach. Genentech's media formulations are similar to the common DMEM/Ham's F12 medium formulations, and were (back then) typically supplemented with protein hydrolysate. For the target HTST conditions of 102 °C and 10 s hold time, treatment of cell culture media and feed solutions showed no significant loss of amino acids or vitamins, no significant loss of methotrexate, a 10–20% loss of insulin (transferrin could not be measured analytically in media), and typically no significant impact on media or feed filtration performance. For the trace element solution that was historically added to the bioreactor separately from the culture medium, an autoclave sterilization cycle treatment did not significantly affect the metals concentrations. For bolus additions of concentrated glucose that would be added up to a few times to the production bioreactor during the course of the run, compatibility was evaluated for heat sterilization, which showed no appreciable loss of glucose concentration.

The ultimate test of barrier compatibility was performed via functional testing, that is, performance of cell cultures executed with barrier-treated media and solutions and evaluation of resultant product quality (9). Studies for Process B were conducted in 2 L stirred tank bioreactors operated under conditions representative of manufacturing conditions. Cell growth and viability were measured via microscopic examination using trypan blue exclusion. Media containing insulin and transferrin and feeds were treated via HTST exposure. All other additions were subjected to the barriers described in the previous two paragraphs. Figure 5 shows N-1 culture growth when subjected to the upstream virus barriers over a range of HTST treatment temperatures from 97 to 115 °C as part of the evaluation of media robustness to heat treatment conditions. The N-1 medium contained 2% serum. There was no significant impact on specific growth rate for the conditions shown, or for a study which evaluated 102 °C treatment for up to 60 s of hold time exposure (data not shown). Similar results were obtained when studying the effect of the virus barriers for up to two growth passages containing 2% serum (data not shown). Figures 6 and 7 show production culture cell growth and viability were unaffected for multiple replicates under target barrier conditions. Table IV provides productivity (titer) and product quality data for the production culture conducted at target barrier conditions. There was no significant impact observed to titer or the numerous product quality attributes.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

N-1 culture growth and viability in 2 L bioreactor system using medium treated with virus barriers (9). HTST processing at 102 °C for 10 s.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6

Production culture growth in 2 L model system with media treatment barriers (9). HTST processing at 102 °C for 10 s.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7

Production culture viability in 2 L model system with media treatment barriers (9). HTST processing at 102 °C for 10 s.

On the basis of the lab-scale (2 L) process characterization studies, the decision was made to design, fabricate, and implement the upstream virus barriers for Process B. This included HTST treatment of media and feeds along with the other small-volume barrier treatments (heat sterilization, virus-retentive filtration) previously described. Given the need to implement as quickly as possible, the decision was made to provide the HTST treatment for the two largest scales of operation—the 2kL N-1 bioreactor and the 12kL production bioreactor and feeds. These two scales of operation represent more than 95% of the total media used in the execution of Process B. As such, this represented a significant reduction in virus contamination risk. It is noted that all hydrolysate used in the process were historically heat-sterilized prior to use, regardless of location in the process. This practice was maintained.

The large-scale, additional virus barriers were successfully installed, started up, and validated for Process B. Figure 8 shows the upstream virus barriers applied to large-scale processing at that time, all of which are in practice today (with one exception—current day processes no longer require the use of the alcohol-solubilized organics). Two HTST skids were used for the manufacturing facility, which contained six production bioreactors, with the larger of the two skids processing medium at 100 LPM for 12kL bioreactor batching operations.

• Download figure
• Open in new tab
• Download powerpoint

Figure 8

Process flow for cell culture virus barriers implementation.

Process B was licensed for use of these additional virus barriers by all worldwide regulatory authorities (where licenses were held) as a post-approval change. A subsequent effort developed a serum-free, transferrin-free, and rhInsulin-utilizing Process B, which was revalidated with the additional virus barriers and successfully licensed. Two subsequent CHO cell culture processes implemented the additional virus barriers post-Phase III material production on the basis of formal lab-scale characterization studies, and they received licensure as part of the initial product approval. Since those regulatory approvals, all late stage development processes implement these additional virus barriers upon their first entry into the commercial facilities, which is typically for PhIII material generation. Earlier phase clinical programs will utilize these additional barriers if they are executed in facilities where the barriers are available for use. No additional lab-scale studies are required for barrier implementation at the late-stage clinical phases, as Genentech considers the platform media and feeds to be validated for use with any process utilizing the platform. If changes to the platform media and feeds are made, an assessment is performed to determine if re-validation of the modified media and feeds with the additional virus barriers is required.

Today, similar upstream virus barrier technologies are implemented at nine total commercial manufacturing sites throughout the world, representing both Genentech/Roche sites and several CMO/partner sites. These implementations represent six commercial processes, and tens of clinical processes across three media/feed platforms, both peptone-containing and peptone-free. Genentech believes that these additional virus barriers are also extremely valuable to processes based on chemically-defined media and feeds, based on the reality of the raw materials supply chain risks described earlier. In some cases, the barriers have been extended as far back as the N-3 bioreactor. Consideration is being given to extending the barriers even further back into the processing in the future, to include all inoculum bioreactors and seed train. It is also noted that, since the implementation of the additional virus barriers after the 1994 contamination, there have been no subsequent detected rodent parvovirus contaminations in more than 200 runs of Process B in that same facility. It is acknowledged that the bovine serum and other animal-derived protein supplements were eliminated from Process B about halfway through those 200 runs. This, of course, cannot be taken as proof that the reason for no further contaminations was the implementation of the barriers. But, the result is certainly consistent with the demonstrated efficacy of the barriers, and it is consistent with the expected outcome.

Future directions in upstream barrier technologies will likely include implementation of UV-C inactivation approaches for components, such as serum or other protein supplements, which are incompatible with HTST heat treatment barriers. Technical work is still needed in developing large-scale UV-C systems for cell culture media streams, specifically in terms of improving the control and validation-readiness of the large-scale flow fields. For small to medium volume applications, virus filtration could be a future technology of choice if the cost-effectiveness can be improved.

Downstream Process Virus Barriers (Clearance)

In the unlikely event that an adventitious virus contaminant is undetected in a harvested cell culture, the downstream purification process is validated to provide significant virus clearance (using model viruses which span a range of virus types—enveloped, non-enveloped, RNA viruses, single and double-stranded DNA viruses). Figure 9 illustrates the typical downstream purification process for a monoclonal antibody, along with the description of the virus clearance modes of action and the typical virus clearance results for MuLV (model retrovirus) and MMV (model adventitious virus). It is seen that multiple, orthogonal processing steps can deliver very effective virus reduction. Figure 10 shows the robustness of virus clearance based on detergent inactivation, low pH hold, anion exchange chromatography, and virus-retentive filtration across seven or more different processes. It is seen that there is very consistent, effective (>4 logs) virus clearance of the indicated processing steps against MuLV, Simian virus 40 (SV40), and MMV.

• Download figure
• Open in new tab
• Download powerpoint

Figure 9

Process flow diagram for typical downstream process including orthogonal clearance steps, showing virus clearance modes of action and removal capabilities. LRV is the log removal value, representing log₁₀ of the ratio of initial virus contained in a sample divided by the remaining virus following the treatment step (LRV = log₁₀[virus_in/virus_out]). Courtesy of Qi Chen and the Genentech Process Virology group.

• Download figure
• Open in new tab
• Download powerpoint

Figure 10

Virus clearance robustness of downstream purification process. LRV is the log removal value, representing log₁₀ of the ratio of initial virus contained in a sample divided by the remaining virus following the treatment step (LRV = log₁₀[virus_in/virus_out]). Courtesy of Qi Chen and the Genentech Process Virology group.

Future directions in downstream virus clearance technologies may include increased use of moderate heat for retrovirus inactivation and UV light treatment for small, non-enveloped virus inactivation. Further improvement in virus-retentive filtration is expected, particularly in terms of improved protein throughput capacity and cost-effectiveness. Mixed-mode chromatography may also be employed as a robust virus removal step capable of operating over wide ranges of pH and conductivity.

Overall Virus Barrier Program in Biologics Manufacturing

Figure 11 shows the overall virus barrier approach as applied to Genentech's biomanufacturing processes. The cell culture process begins with generation of the cell banks. Extensive in vitro and in vivo virus characterization testing of the cell banks provides a significant barrier preventing entry of viruses into the manufacturing process from a compromised bank. As described earlier, virus barrier treatment (e.g., HTST, autoclaving, virus-retentive filtration) provides protection of the large-scale bioreactor processes. The in-process hold step PCR test for rodent parvoviruses prevents parvovirus-contaminated cultures from being harvested and contaminating the rest of the manufacturing facility downstream. The adventitious virus testing by cell-based methods performed on pre-harvest cell culture fluid provides a significant barrier to release of virus-contaminated product to patients. Because these tests take several weeks to complete, they do not protect the downstream facility from a contaminated production bioreactor, but a positive result here would enable shutdown of the affected manufacturing train(s), if appropriate, thereby providing some limitation of the scope of the contamination.

• Download figure
• Open in new tab
• Download powerpoint

Figure 11

Genentech's overall virus barrier approach in the biomanufacturing process.

In parallel to the in vitro virus screening testing, the purification process, by virtue of its validated virus clearance, also provides a significant barrier to release of virus-contaminated product to patients. This overall system of virus barriers is an evolving program. As appropriate, surveillance of emerging viruses with a high likelihood of infecting CHO cells may lead to rationale for adding new virus-specific tests to the control system. Likewise, extension of processing barriers across the entire scope of cell culture processing can provide additional layers of risk mitigation. Development of new technologies may also provide ability to implement more manufacturing-friendly barriers as well as extend the ability to validate higher levels of virus clearance protection in the downstream process.

Lessons Learned and Conclusions

With a project of this magnitude, there are many lessons learned that were derived from the collective efforts of many parts of the organization. Table V provides a summary of various key lessons derived from the investigation of the MMV contaminations and implementation of additional virus barriers. A major focus of the lessons learned is with respect to assessment of raw material risks. At the end of the day, providing comprehensive virus barriers for mammalian cell culture processes is a form of business insurance. Doing so will ensure a manufacturing facility's ability to operate while ensuring that product supplies are available to patients when they need them. Just do it!