Abstract
Appropriate performance of virus validation studies and testing of unprocessed bulk harvests for retrovirus particle count are procedures in the demonstration of an acceptable level of viral safety for cell-derived biotechnology products. Product-specific validation studies on virus reduction with two model viruses [usually murine leukemia virus (MuLV) and a parvovirus] performed in duplicate runs are standard for clinical trial applications. For the retroviral safety margin, a 6 log reduction is normally expected. Retroviral particle counts are measured traditionally by transmission electron microscopy (TEM) and are commonly performed at contract laboratories. These procedures are quite time-consuming and can be associated with significant costs. In particular, the time factor is a hurdle for companies that want to quickly bring their new products to the clinic. In this session, several strategies on how to lower time, cost, and workload in the evaluation of viral safety for early clinical trial applications, while still ensuring sufficient level of viral safety of the product, were presented. In addition, virus reduction strategies for molecules that do not have the standard antibody structure are presented. Also presented in this session is the feasibility of the use of retrovirus-like particle (RVLP) in the prevalidation of virus removal and the use of quantitative polymerase chain reaction (qPCR) as an alternative to infectivity assays in virus validation studies as well as its use as an alternative to quantitative TEM analysis for determining RVLP count in the bulk harvest of a perfusion bioreactor.
LAY ABSTRACT: In this session, several strategies on how to lower time, cost, and workload in the evaluation of viral safety for early clinical trial applications of cell-derived biotechnology products, while still ensuring sufficient level of viral safety of the product, were presented. In addition, virus reduction strategies for molecules that do not have the standard antibody structure are presented. Also presented in this session is the feasibility of the use of retrovirus-like particle (RVLP) in the prevalidation of virus removal and the use of quantitative polymerase chain reaction (qPCR) as an alternative to infectivity assays in virus validation studies as well as its use as an alternative to quantitative TEM analysis for determining RVLP count in the bulk harvest of a perfusion bioreactor.
- Viral clearance
- Project acceleration
- Novel format molecules
- Retrovirus-like particle
- Quantitative polymerase chain reaction
- Safety margin
Background and Session Overview
Appropriate performance of virus validation studies and testing of unprocessed bulk harvests for retrovirus particle count are procedures in the demonstration of an acceptable level of viral safety for cell-derived biotechnology products. Product-specific validation studies on virus reduction with two model viruses (usually murine leukemia virus [MuLV] and a parvovirus) performed in duplicate runs are standard for clinical trial applications. For the retroviral safety margin, a 6 log reduction is normally expected. Retroviral particle counts are measured traditionally by transmission electron microscopy (TEM) and are commonly performed at contract laboratories. These procedures are quite time-consuming and can be associated with significant costs. In particular, the time factor is a hurdle for companies that want to quickly bring their new products to the clinic. In this session, several strategies on how to lower time, cost, and workload in the evaluation of viral safety for early clinical trial applications, while still ensuring sufficient level of viral safety of the product, were presented. In addition, virus reduction strategies for molecules that do not have the standard antibody structure are presented. Also presented in this session is the feasibility of the use of retrovirus-like particle (RVLP) in the prevalidation of virus removal and the use of quantitative polymerase chain reaction (qPCR) as an alternative to infectivity assays in virus validation studies as well as its use as an alternative to quantitative TEM analysis for determining RVLP count in the bulk harvest of a perfusion bioreactor.
Retrovirus Clearance Risk Evaluation and Clearance Target for Chinese Hamster Ovary Cells
Bin Yang, Genentech
To ensure virus safety, assessing the capacity of production processes to clear infectious viruses is one of three principal and complementary approaches required by ICH Q5A (1). Chinese hamster ovary (CHO) cells express endogenous C-type RVLPs. However, CHO RVLP infectivity has never been observed for >30 years at Roche or across the biotechnology industry. ICH Q5A also acknowledges that CHO RVLPs are extensively characterized and CHO cell has no reported safety concerns (1).
Molecular characterization of CHO RVLP genomic and cDNA sequences also supports the noninfectious nature of RVLPs. A full-length CHO provirus sequence was identified and multiple nucleotide substitutions, insertions, and deletions were observed in both the gag-pol and env open-reading frames (ORFs). This provirus genome was predicted to not be able to produce Type C particles from its locus (2). More interestingly, a sequence of cDNA clones isolated from purified CHO RVLP revealed that there is no functional endonuclease owing to multiple interruptions of ORFs (3). A more recent study carried out in CHO by RNA sequencing detected only minimal env or pol transcripts, while gag transcripts were more abundant (4).
Currently, the same virus clearance target (e.g., ≤1 virus particle per 106 doses), as shown in Table I, applies to CHO and NS0 cell line–derived products and plasma-derived products with distinctly different virus risk profiles. Owing to the noninfectious nature of CHO RVLPs and the safety record of CHO cell-derived products, it is suggested to re-consider the retrovirus clearance target for CHO-derived biotechnological products. A reduced retrovirus clearance target was proposed for CHO-derived products by validation of two orthogonal steps with retrovirus removal and/or inactivation capacity, one of which is an effective step such as small virus retentive filtration.
Virus Safety Margins for Cell-Derived and Plasma-Derived Medicinal Products
A Proposal for Future BLA Submissions on Virus Filtration: Parvovirus Removal Studies Only
Rachel Specht and Qi Chen, Genentech
Small virus retentive filtration is a key unit operation in ensuring virus safety for biologic molecules. Robust and effective large virus removal by both large and small virus retentive filtration has been shown with no reports of significant large virus passage. Yet, companies continue to perform small virus retentive filtration virus clearance validation studies using both small and large model viruses for license applications. Because the virus filter is designed to retain small viruses, Genentech proposes to study parvovirus removal and extend the reduction factor to larger model viruses such as xenotropic murine leukemia virus (X-MuLV) and simian virus type 40 (SV40). An example is given in Table II of how experimental constraints owing to the assay type and prefiltration loss can lead to significant difference in log reduction values (LRVs), causing the reported large virus LRV to under-estimate the removal capacity of the virus filter while not giving meaningful insight into the virus removal capacity of the filter. While the percent virus spike is held constant across three model viruses, the prefiltration loss varies greatly between the viruses and results in large virus LRVs below 4, while the minimal amount of prefilter loss for minute virus of mice (MVM) results in complete and highly effective removal of >6 logs. Because MVM is the smallest virus retained by small virus retentive filtration and is tested under worst-case conditions of maximum volumetric throughput and high and low flux including a 1-h pause (as shown in Figure 1), the MVM LRV gives a more accurate total virus removal capacity and robustness for the virus filter than the larger model viruses and thus can be applied to larger viruses.
Example of Log Reduction Values Divergence Based on Model Virus and Experimental Conditions
Flux profile during virus filtration of MVM.
qPCR as an Alternative to TEM and Infectivity Assays
William Rayfield, Merck
RVLPs are budding (Type C) viral particles that are related to several known murine retroviruses as determined by sequence similarity (2–3, 5). They have been shown to be noninfective and originate from provirus sequences endogenous to the CHO genome as opposed to adventitious virus contamination. ICH Q5A states that, “the risk of viral contamination is a feature common to all biotechnology products derived from cell lines. Such contamination could have serious clinical consequences and can arise from the contamination of the source cell lines themselves (cell substrates) or from adventitious introduction of virus during production” (1). Endogenous retrovirus has been identified as a potential contaminant of mammalian cell lines that bear the risk of becoming infectious during the production process. Regulatory guidance requires that representative unprocessed bulk material is tested for RVLPs on a limited number of lots to assess process variability. Detection methods include TEM for particles that exhibit similar size/morphology of retroviruses and qPCR assay that can detect retrovirus genome sequences.
TEM quantitation of RVLPs consists of manual scoring by morphology of viral particles in concentrated and micro-sectioned cell culture supernatant. This method requires high sample volumes, has a high limit of detection, and requires capital equipment that is considerably difficult to access. As such, samples have historically been sent to contract laboratories for TEM analysis, where other barriers arise such as high sample cost and long turnaround time (>1 month). Nonetheless, TEM is currently a standard for RVLP quantitation. Alternatively, quantitative real-time (RT)-PCR methods have been described for RVLPs [and other viruses; (5–6)]. This is an orthogonal method of detecting RVLPs that present several advantages over TEM, namely, higher throughput, higher sensitivity, and lower cost, and the required equipment is relatively easy to acquire and operate at a manufacturing facility. One caveat of this technique is the assumption that the number of RVLP genomic copies is a surrogate measure for the number of intact RVLP. However, in this meeting session, data were provided by Merck suggesting that TEM and qRT-PCR measurements have reasonable concordance [Figure 2a–c; (6)].
Comparison of RVLP by TEM vs qRT-PCR. Units are in particles per milliliter. Limit of detection is 1 × 106 for TEM and 3.08 × 104 particles/mL via qPCR. (a) qPCR values versus TEM values; (b) Log difference between qPCR and TEM values; and (c) kinetic of qPCR and TEM values in perfusion bioreactor and permeate; open symbols indicate RVLPs not detected by TEM assay.
Merck evaluated the feasibility of qRT-PCR as an alternative assay for process development, which can potentially be used as an in-process test or an in-process control (IPC) for future-state manufacturing. The comparison of TEM and qRT-PCR measurements is shown in Figure 2. Figure 2a suggests a general concordance between the two measurement modalities, with differences measured from the same sample calculated at approximately one order of magnitude or less. Similarly, Figure 2b compares the log difference between the two measurements of the bioreactor material, which shows that the qPCR measurement is generally higher than the TEM measurement, but is within approximately 1 log. The assay variability among replicates as well as difference in detection methodology can account for these differences. Furthermore, based on these data (Figure 2c), there appears to be an accumulation of RVLP over time in the bioreactor, and significantly lower levels of RVLP were detected in the permeate samples. This was observed in both TEM and qRT-PCR data.
The experiments described here suggest that qRT-PCR can be used as an alternative to TEM. Although TEM and qRT-PCR measure different aspects of the RVLP (visualizing intact particles vs quantitating genome copy number), the trends are preserved between the two methods. While the bioreactor values from the qRT-PCR assay are typically 1 log higher than the TEM values, qRT-PCR has the advantage of being more sensitive and enables higher throughput, which allows for a near real-time analysis of bioreactor and perfusion samples without the significant turnaround time associated with TEM.
Feedback from the 2017 Viral Clearance Symposium indicates that qRT-PCR would be an acceptable method of quantitating RVLPs. Furthermore, feedback from the meeting also indicates that using permeate RVLP levels from a perfusion culture to calculate RVLP burden could be acceptable for a clinical-phase regulatory filing, particularly if the qPCR testing was performed as a potential IPC. The accessibility and the high-throughput nature of the qPCR test method enable this type of strategy to be implemented for a manufacturing process.
Viral clearance studies have used tissue culture infectious dose 50 (TCID50) or plaque-based infectivity assays for neutral pH streams owing to their specificity. However, such assays require extensive turnaround times and significant virology/biology lab support for culturing virus-specific indicator cells. The use of qPCR as an orthogonal method of detecting virus would present several advantages—namely, higher throughput and lower cost. To compare the two methods, fraction and pool samples from a viral filtration MVM viral clearance study from both the Planova 20N and Viresolve® Pro filter runs were analyzed by both qPCR and infectivity assays. Here, qPCR samples were treated with benzonase prior to virus DNA extraction, and then the samples were assayed for DNA detection and quantitation by qPCR using TaqMan technology primers and probe specific to MVM. An MVM DNA standard was used for quantitation. Data from the runs show log reduction of virus that is comparable between the two assay methods (Figure 3). For each run set, the qPCR result was within 1 log of the infectivity value (7), which suggests that the qPCR assay could be used as a surrogate for infectivity data.
Differential MVM LRV in viral filtration pool using both infectivity and qPCR assays.
The Use of RVLP in Research and Development Viral Clearance Studies
Moritz Bennecke, Roche Pharma
The growing pipeline of modern therapeutic antibodies consists of many different antibody formats. The purification processes of new-format molecules might require more attention with respect to removal of process-related impurities compared with standard, monoclonal antibodies (mAb). As a consequence, adaptations of the downstream process (DSP) can include the implementation of new resins or the development of less well-known process step conditions (Figure 4).
Design of a “new-format” validation study.
To allow more efficient validation approaches for new-format antibody purification processes, Roche developed a protocol to use RVLPs in spiking studies for prevalidation experiments. RVLP were successfully purified with an easy-to-follow protocol for research and development (R&D) purposes and were shown to have a quality comparable with commercially available retrovirus stocks. With a specific qPCR assay, results can be obtained by conducting chromatography experiments in an S1 lab setting.
A number of experiments were carried out in several purification processes, including new-format molecules. The results (LRV reduction), as shown in Table III, fit well to the data obtained from validation studies using X-MuLV spikes. A good overall comparability between RVLP R&D and virus validation study results was observed.
Comparison of Test Results of R&D Studies with Validation Studies
As a conclusion, the use of RVLP spikes in the purification process of new-format molecules supports early identification of challenges and enables efficient process development for viral removal capacity of individual process steps.
Getting Medicines to Patients Fast
Sherrie Curtis, Genetech
Demonstration of virus safety for cell line–derived products is a Health Authority requirement for entering clinical trials and for marketing authorization. Virus clearance validation is one of the three pillars to ensure product virus safety, and it is sometimes on the critical path to getting medicines to patients. There is significant advantage to a company sponsoring a clinical trial to enable early patient enrollment. For a product that is entering the commercial realm, there is a health authority expectation to have a robust manufacturing process for virus clearance. In the life cycle of a product, both speed and robust virus clearance are needed.
The following four strategies are implemented to remove virus clearance from the critical path and enable speed for clinical trial initiation: dedicated virus clearance steps, modular validation, qPCR for RVLP estimation, and earlier initiation of validation activities. The following three dedicated virus clearance steps are included in the purification process: detergent, low pH, and small virus retentive filtration. The impact of process parameters and intermediate composition on virus clearance has been evaluated, and reproducible LRV between products and processes has been shown. For each of the dedicated steps, a space has been experimentally defined, in which virus clearance is robust and these parameters are controlled in manufacture as shown in Table IV. The use of prior in-house knowledge enables application of modular validation per health authority guidance (8–9). The use of qPCR assay to estimate the number of endogenous CHO RVLP also significantly shortens the standard lead-time compared with the time-consuming TEM assay. Additionally, initiation of virus clearance validation can begin prior to GMP manufacture by using material from earlier production (e.g., material used to supply toxicology trials) upon demonstration that the material is representative of GMP-manufactured material. Implementing these four strategies have achieved significant time saving.
Controlled Parameters in Manufacturing for Dedicated Virus Reduction Steps
For a product entering into commercial production, a robust process for virus clearance is needed. Because virus clearance studies are expensive and time-consuming, an efficient study design with few experiments to support process parameter ranges is desirable. For the robust virus clearance steps, key parameters that affect virus clearance have been investigated, and the directionality of these parameters has been identified. By evaluating parameters at the worse-case direction (e.g., high pH, low temperature, short time for low pH virus inactivation), or by conducting a small design of experiments, the manufacturing parameter ranges can be supported with virus clearance data. For small virus retentive filtration, as the mechanism of virus removal is based on size, only the smaller virus is assessed for removal, and the LRV obtained for the smaller virus is applied to larger viruses. Prior in-house data for affinity chromatography have shown that no parameter affects virus removal, and as such, the virus clearance validation at manufacturing set points can support the manufacturing ranges. The stability of virus clearance after repeated use of chromatography resins has been shown to have no impact on affinity and ion exchange resins across many molecules, processes, and cycles.
Strategies for virus clearance vary depending on the stage of development. At clinical trial stages, a “speed” approach is adopted by predefining robust conditions for virus clearance and implementing the robust conditions in manufacture, using prior in-house knowledge enabling modular validation, using of qPCR assay for RVLP estimation, and initiating validation prior to GMP manufacture. At the commercial stage, reduced experimentation to support the manufacturing ranges is made possible by understanding the mechanism of virus clearance and by understanding the directionality of important parameters that can impact virus clearance. At Genentech-Roche, virus clearance strategies have been implemented keeping virus clearance off the critical path while assuring patient safety at both clinical and commercial stages.
Streamlining Virus Clearance (VC) Studies to Reduce Costs and Time to Clinic
Lisa Connell-Crowley, Just Biotherapeutics
The mission of Just Biotherapeutics is to dramatically expand global access to biotherapeutics. To achieve this, Just Biotherapeutics works toward lowering costs and reducing development time to clinic. Viral clearance studies performed at contract labs can be costly and require significant time to prepare, execute, and receive the final testing results. Typically, the viral clearance study is conducted toward the end of development with material from a large-scale run using a locked manufacturing process, which means that study completion and receipt of the TEM data for RVLP burden can be on the critical path for a regulatory filing.
MSD Sharp & Dhome and Just Biotherapeutics have evaluated ways to streamline a first-in-human (FIH) viral clearance study to reduce cost and time. One approach is to reduce the number of runs tested for the viral clearance study. Figure 5 shows a platform mAb purification process and a comparison of viral clearance testing for a full FIH study compared with a streamlined study. Both study designs use the following two model viruses: X-MuLV and MVM. The full FIH study includes evaluation of each chromatography step and the viral filter using both viruses with duplicate runs and X-MuLV tested in duplicate for the low pH viral inactivation step. The streamlined study evaluates only those steps that provide effective clearance: low pH viral inactivation, viral filtration, and one effective chromatography step. A second chromatography step could be added if more clearance is needed. Protein A chromatography, which can exhibit low clearance depending on the molecule, would not be tested unless needed. Additionally, the viral filter is tested with MVM only, as it is the smaller of the two viruses and thus represents worst case for sized-based removal by the filter (10).
Proposal for reduced FIH viral clearance testing.
For platform mAb processes, testing could be further reduced by using viral clearance data sets to support viral clearance claims on effective steps. For example, the American Society for Testing and Materials (ASTM) standard E2888-12 for low pH inactivation of X-MuLV supports a clearance worth 5 logs if the operating parameters specified in the standard are used (11). Platform data sets can be used to support modular claims for viral filtration, as well as chromatography steps such as anion exchange chromatography (AEX) or cation exchange chromatography (CEX), in which the mechanism of clearance is understood and the data package represents a range of operating parameters. Such data sets can then be used to reduce duplicate testing to single confirmatory runs.
Additional streamlining could be achieved by using qPCR assays in place of TEM for RVLP burden and infectivity for virus removal by chromatography and filtration (Table V). Turnaround times for TEM testing at contract labs are typically 8–12 weeks, while those for qPCR testing are significantly faster. Because qPCR requires a smaller equipment footprint and setup than cell-based infectivity assays and TEM, it could be feasible to bring virus-spiking studies and qPCR testing for infectivity and RVLP burden in-house, which would reduce costs, provide flexibility, and facilitate faster turnaround times. qPCR testing of RVLP levels would not require a special virus lab for sample handing, as the particles are noninfectious. Virus-spiking studies and sample analysis would require a dedicated biosafety level 2 virus lab, which would add additional cost and complexity.
Using qPCR in Place of Infectivity and TEM Assays
Viral Clearance Strategy for Early-Phase Program
Mi Jin, Teva Pharmaceuticals
Robust viral clearance strategy is an integral part of the overall control strategy to ensure patient safety. Fast-to-clinic demands reliable viral clearance with minimal process development time, and therefore, dedicated platform unit operations based on orthogonal virus inactivation or removal mechanisms are desirable. However, cell line–specific properties may increase the perceived viral contamination risk, and molecule-specific properties may challenge the applicability of certain platform unit operations, requiring additional measures to ensure viral safety. In the 2017 Viral Clearance Symposium, Teva Pharmaceuticals presented proposed viral clearance platform unit operations for early-phase program, as well as a risk-based approach to improve viral clearance performance or using an additional unit operation to expand the viral clearance platform toolbox.
The proposed viral clearance process consists of the following three unit operations: low pH viral inactivation (VI), AEX (membrane) chromatography, and virus reduction filtration (VRF). Two model viruses, X-MuLV and MVM, are used for early-stage viral clearance studies. Process parameters either operate at worst-case conditions based on viral clearance mechanisms or bracket the operating range for manufacturing, with specific considerations for manufacturing procedure. For example, for VRF step, protein load-to-membrane surface area ratio is controlled at a high end of operating range as worst case, while differential pressure is controlled bracketing the high and low end of the operating range defined for manufacturing, and a 30-min pause is incorporated between load and chase steps to reflect the manufacturing procedure worst-case scenario. Other considerations for viral clearance study include virus-spiking levels and virus stock purity. Large-volume titration is used to improve the virus detection limit.
The above three orthogonal viral clearance steps are robust and provide an overall virus reduction of 16–18 LRV under typical operating conditions. However, cell line–specific and molecule-specific properties can pose higher risk for viral safety. Table VI lists some examples of risk and mitigation strategy.
Examples of Risk for Viral Clearance and Mitigation Strategy
For the VRF step, protein molecular properties can significantly affect the fouling behavior of the virus filter. A case study was presented for a fusion protein that severely fouled platform filter owing to its intrinsic molecular properties. Further screening of alternative virus filters and condition optimization was necessary to achieve the required filtration throughput. Data showed that membrane filter type, solution conditions, and protein concentration all exhibited a significant impact on the filtration performance.
Freeze–thaw processes can also generate fouling species for VRF membranes that causes a significant problem during small-scale VC studies. Data were presented for the evaluation of microfiltration prefilters. Filters with different material of construction displayed drastically different capabilities in restoring virus filter performance, suggesting an adsorptive-based mechanism in foulant removal by the effective prefilter.
To enhance the viral clearance capability for programs with higher viral-clearance risk, Protein A affinity chromatography step was optimized. Data were presented on selected Protein A chromatography wash conditions that enhanced viral clearance capability by >2 log reduction, thus making the Protein A step another robust VC step.
ASTM Update on Standards for Low pH Inactivation and Virus Retentive Filtration
John Schreffler, Janssen J&J
ASTM is a not-for-profit, international organization that develops consensus standards with direct stakeholder involvement. Following the 2009 Viral Clearance Symposium, an initiative was started to develop a series of standards within ASTM to help ensure viral safety with specified practices based on data compiled over the past 30 years.
This collaboration has produced two viral inactivation standards for retrovirus inactivation to date, namely, Standard Practice for Process for Inactivation of Retrovirus by pH and Standard Practice for Process Step to Inactivate Rodent Retrovirus with Triton X-100 TM Treatment. To ensure utility and broaden applicability, less restrictive low pH inactivation specifications and alternative detergents are being considered as future iterations of these standards.
Pharmaceutical company and virus filter vendor representatives are currently developing new standard practices for filtration of large (MuLV) and small (MVM) viruses. Both proposed standards present difficulties to approval not previously seen. The main challenges stem from difficulties in defining what a virus filter is and what tests assure that classification. Initial draft standards using bacteriophage retention definitions specified in PDA Technical Report 41 for small and large virus filters have been created (12). Standards using this approach would likely have prolonged approval cycles, requiring approval of at least six additional standards specifying bacteriophage challenge assay preparation, performance, and quantitation. These procedures are not standardized across filter vendors, and qualification and standardization would require resources currently unavailable. To avoid these extended approval times, additional draft standards are being developed in parallel that attempt to leverage available vendor- or company-specific bacteriophage or virus challenge data.
Specifications for process parameters and material attributes across filter types present additional challenges for both standards. Development of appropriate ranges for all small or all large virus filters is difficult owing to the number of filters currently available. For small virus filter standards, creation of ranges is even more complicated when parameters such as depressurization time are assessed. One possible solution could be operation within filter-specific ranges validated using standardized bacteriophage challenges.
Current efforts are focused on the creation of a large virus or MuLV filtration standard, hoping to leverage the robust, well-documented validation history of large virus filtration. Careful considerations and continued collaborations across pharmaceutical companies, filter vendors, and regulatory authorities are needed to ensure viral safety is maintained for this and all future standards.
Summary
Several strategies were proposed by Genentech and Just Biotherapeutics to enable speed to clinic: validation of dedicated steps that are believed to provide efficient virus reduction, use of in-house data to make modular claims for virus reduction, use of low pH ASTM standard, use of qPCR for RVLP estimation in unprocessed harvest, early initiation of validation studies, and implementation of in-house testing. With the implementation of all or most of these measurements, a three-month time saving was calculated; however, this approach has not been tested yet by a real clinical trial submission. It was suggested that an early initiation of validation studies can be accomplished when using non-GMP material (e.g., material for toxicology studies), which requires demonstration of comparability with the GMP material. Regarding the RVLP count, only a half-log difference was found between GMP and non-GMP material according to Genentech's experiences. Using prior knowledge reduces the extent of virus validation and thus saves time and costs. However, regulatory agencies will ask for a substantial amount of in-house/platform data in support of a modular claim. Modular claims can be possible for virus retentive filtration and for well-understood AEX or CEX, but in each case, single confirmatory runs are needed. For virus retentive filtration, one product-specific run with a parvovirus is the minimum requirement for a modular virus filtration claim. Another proposed way to complement virus validation studies is the use of standards developed by ASTM International. As presented by Janssen J&J, so far, the following two standards have been approved: one for low pH inactivation and one for detergent inactivation using Triton X-100. Efforts are now being made to develop standards for virus retentive filtration; however, their establishment faces several problems requiring the development of more than one standard. At this moment, it will not be foreseeable if and when such a standard may be ready for use. In addition, the use of ASTM standards in application for clinical trials bears the risk for nonacceptance by regulatory agencies. It has yet to be revealed and discussed between companies and regulatory agencies if those standards can be accepted, also in Europe, maybe in support with in-house data.
The parvovirus' only claim in the context for marketing authorization application (MAA) was presented for discussion by Genentech. Using qPCR for determining virus titers and performing prefiltration in the validation studies can lead to a significant reduction of LRV for larger viruses (here MuLV and SV40), thus underestimating the true retentive capacity of the virus filter. In contrast, highly effective removal without prefiltration loss was observed with the small virus MVM. Prefiltration, however, seems to be a necessary step in validation studies to avoid filter clogging if the loading material was frozen before application to the virus filter (as presented by Teva). Parvovirus-only validation is already accepted in clinical trial applications and was also already accepted in single cases for established virus retentive filters in MAA. Validation of MVM removal by virus retentive filtration was performed by Teva and Genentech under worst-case conditions using a pressure release and according to Teva also by bracketing the differential pressure at the high and low end of the operating range defined for manufacturing. It was stated that low pressure (or low flow) should be studied as a worst-case parameter. It was further noted that it might be possible to leverage in-house experience/data for a certain filter brand to support low pressure (low flow) conditions. Based on the study results, lower limits for pressure (or flux) ranges should be defined at manufacture for MAA.
qPCR appears to be an attractive alternative to infectivity assays in validation studies for virus reduction, in particular for flow-through chromatography steps and viral filtration. Because contract laboratories may charge the same costs for qPCR and infectivity assays, performing qPCR in-house in addition to spiking studies—according to Just Biotherapeutics—could also lower time and costs. One drawback of qPCR is to show that only infectious viral particles and not free viral genomes are measured. For example, a side-by-side comparison performed by Merck on virus filtration using qPCR and infectivity assay showed mostly a slightly lower LRV for the qPCR assay. In addition, in downscale studies on virus filtration performed by Genentech, a virus filter breakthrough was observed for X-MuLV when measured by qPCR, which is normally not observed in infectivity assays. These observations suggest that free viral genomes may have passed the filter and thus pretend lower LRV or virus breakthrough. Therefore, it is very important to ensure proper nuclease treatment of the samples to exclude false-positive signals and to get reliable virus reduction data. In addition, matrix interference needs to be determined and excluded. qPCR was further proposed by several companies (presentations from Merck, Genentech, and Just Biotherapeutics) as a suitable tool to determine the number of RVLP in the unprocessed harvest as a replacement method for TEM, especially in light of the 8- to 12-week performance when TEM is done by contract laboratories. A method-to-method (qPCR vs TEM) comparison was shown by Merck in the context of analyzing RVLP count in a perfusion bioreactor. Here, similar results were found with both methods for RVLP counts in the bioreactor and permeate, even though the higher sensitivity of qPCR enabled also the tracking of RVLP in the permeate, which was not possible by TEM where all measurements were at the limit of detection. With both methods, an increase of RVLP was observed in the bioreactor over time, whereas a decrease was observed in the permeate. Further data are needed for confirmation and a better understanding of the mechanisms leading to the RVLP distribution in perfusion bioreactors. It was discussed that it might be possible to determine RVLP titer in the permeate for RVLP per dose calculation for clinical trial application if it is measured for each lot as IPC and if it is shown to be consistent among lots.
Roche investigated the feasibility of using CHO RVLP for analyzing the virus removal capacity of several chromatography steps. While using RVLP in virus validation for Protein A chromatography has already been published by Zhang et al. (13), Roche investigated their suitability as spiking agent for other chromatography resins (CEX and AEX and hydrophobic interaction chromatography resin) and compared the results with data from validation studies with MuLV. In 38 runs, similar removal or only slightly lower LRV were measured for RVLP compared with the validation study results. However, the RVLP data are considered still preliminary because LRVs for both spiking agents were determined in independent experiments and the titer was determined by different assays (qPCR for RVLP and infectivity assay for MuLV). It was noted by the audience that in another instance RVLPs measured by qPCR was not able to detect one of the two CHO strains tested. Therefore, sequences from a more conserved genome region might be needed. In addition, RVLP cannot substitute for MuLV in virus inactivation studies.
Recombinant engineering enables the design of new molecule formats, resulting in molecules with physicochemical characteristics different from classical antibodies. Those molecules may require modification of the standard virus purification process, for example, by inclusion of new chromatography resins. Roche examined the suitability of MM AEX and hydrophobic interaction chromatography resins for the purification of new-format molecules using RVLP. Here, good RVLP removal was found with MM chromatography resins (however, no specific resin types were specified) and RVLP results were similar to validation study results using MuLV. Teva presented a case study of a fusion protein where molecule properties and process conditions significantly affected the reduction capacity of a virus filtration step. Hydrophobicity of the product obviously has led to severe fouling of the platform filter, which made it necessary to screen for alternative virus filters. In this case, conditions such as solutions and protein concentration needed to be optimized for good filtration results.
Genentech provided considerations about the retrovirus safety margin for CHO cells. CHO RVLPs are defective particles, and no single case of CHO RVLP infectivity has been observed across the biotech industry. Likewise, ICH Q5A states that no virus transmission has been associated with biotechnology products derived from cell lines (1). Based on these arguments, Genentech questioned the expected safety margin of 6 log for CHO-derived biotechnology products in light of the same safety expectation for biotechnology products derived from cell lines that are known to produce higher numbers of RVLPs (e.g., NS0 cells). For plasma-derived medicinal products that have the risk of containing human infectious viruses, a high safety margin is also expected (e.g., HIV), but it is not always possible for all viruses (e.g., parvovirus B19) and the guideline on plasma-derived medicinal products does not define a specific limit because the virus reduction factor is subject to various qualitative aspects of interpretation (14). According to the arguments stated above, a <6 log safety margin is desired by the companies for CHO RVLP. It was noted by the audience that in U.S., a lower safety margin of 4 is acceptable. A lower safety margin is also accepted in Europe for clinical trial applications; however, MAA still requires the more strict 6 log LRV. To better understand the molecular basis of the noninfectivity of CHO RVLP, it was agreed that recent CHO genome sequences should be further analyzed. It was furthermore speculated, if rather a safety margin for contaminants such as MVM would be needed because of the real infectivity (for CHO cells) of this virus and the occasional observation of bioreactor contamination.
Conflict of Interest Declaration
The authors declare that they have no competing interests.
- © PDA, Inc. 2018