Downloaded from journal.pda.org on September 18, 2016
Session 2/3: Integrated Viral Clearance Strategy and Case Studies Hannelore Willkommen
PDA J Pharm Sci and Tech 2015, 69 183-194 Access the most recent version at doi:10.5731/pdajpst.2015.01042
Downloaded from journal.pda.org on September 18, 2016
CONFERENCE PROCEEDING
Session 2/3: Integrated Viral Clearance Strategy and Case Studies HANNELORE WILLKOMMEN Regulatory Affairs & Biological Safety (RBS) Consulting ©PDA, Inc. 2015 Background Integrated virus clearance strategy in manufacture of medicinal products is the implementation of unit operations that are effective for virus clearance, well understood in its mechanism of action (MoA), and well to control because the critical process parameters (CPPs) are identified and maintained in the accepted range. If the virus clearance capacity shall be predictive for other products, the investigation of the MoA should clearly show that virus clearance by the selected unit operation is not product-dependent. Unit operations following different MoAs (these are orthogonal steps) should be combined in manufacturing. The data base needed to develop the integrated virus clearance strategy can be built up from the investigation of similar unit operations used for production of several products (platform manufacturing) or from systematic investigations identifying those parameters or parameter ranges that influence virus removal/inactivation by a particular unit operation (design-of-experiment, DoE, studies). There are unit operations that are well understood in their MoA. These are especially anion exchange (AEX) chromatography in flowthrough mode, filtration using virus-retentive filters of small pore size for larger viruses (like retrovirus), and solvent/detergent (S/D) treatment using tri-N-butyl-phosphate and Triton X100 to inactivate enveloped viruses. Other methods are not yet sufficiently investigated to fully understand the MoA or the MoA is so complex that the virus clearance capacity cannot be predicted. The two sessions of the 2013 Virus Clearance Symposium were attributed to three areas: to discuss (1)
e-mail: [email protected] doi: 10.5731/pdajpst.2015.01042
Vol. 69, No. 1, January–February 2015
factors influencing virus clearance and selection of model viruses, (2) problems arising from testing for adventitious agents, and finally (3) testing of re-used resins and its need for Biologic License Application/ Marketing Authorisation Application (BLA/MAA) submissions. The general descriptions and case studies are reported in this order and summarized at the end of this report. DoE-Based AEX Technology Comparison (George Miesegaes, CDER/FDA) [A larger body of this work has been published recently (1)] For many years AEX resin-based chromatography has been used for robust partitioning of the desired protein intermediate from host cell DNA and host cell protein (HCP). In addition, consistent viral clearance, with values reaching upwards of ⱖ4.0 log10, is often possible (2). More recently, there has been an observed increase in the use of AEX chromatographic membranes, also named membrane adsorbers, as an alternative to the traditional column-based approach (3). Key advantages include increases in capacity and flow rate, which afford reduced processing times and overall resource consumption, and other features such as disposability (Table I). While the mechanism of impurity removal is considered the same, a direct, sideby-side comparison of each modality was warranted. In this study, AEX column (n ⫽ 4) and adsorber (n ⫽ 3) brands were systematically assessed to investigate the extent of performance overlap. A single flowthrough mode run condition was first identified by entering queries into a previously reported viral clearance database (2) and was found to be similar to that used by Norling et al. (4). These parameters had later served as the centerpoint condition for a resolution III DoE study, published elsewhere (1). The types and amounts of impurities present in a feedstock may affect viral clearance (e.g., presentation by J. Glynn, Pfizer (5)). To further investigate this, Xenotropic murine leukemia virus (XMuLV) and por183
Downloaded from journal.pda.org on September 18, 2016
TABLE I General Features of AEX Resin and Membrane Technologies Feature
AEX Resin
AEX Membrane
Scalability
Yes
Yes
Reuse
Yes
No; disposable
Flow/particle capture
Diffusion-based
Adsorption-based
⫹⫹⫹⫹
?
Robustness of viral clearance
cine parvovirus (PPV) clearance was assessed under a range of conditions where DNA and protein impurities were spiked into the feed at known concentrations (Figure 1). Log10 reduction values (LRVs) obtained [polymerase chain reaction (PCR) assay] were found robust (ⱖ4 log10) under most of the conditions tested. While two outliers were noted (Figure 1d), their clearance values were approaching 3 log10 and were considered adequate. It was concluded that under the conditions tested, both AEX technologies were capable of achieving equivalent LRV ranges.
Viral Clearance for Vaccine Candidate Using Baculovirus Expression System Focusing On Aex Bind-Elute Chromatography (Adam Kristopeit, Merck, Sharp and Dohme, Inc.) Strategy for Clearance Studies Using the Baculovirus System Data presented is from a virus clearance evaluation for a vaccine candidate produced in an insect cell/Baculovirus expression system. In this expression system, antigens of interest are produced by infecting insect cell culture with a recombinant Baculovirus that contains the genetic information for the antigen. Virus clearance studies are required to demonstrate removal of Baculovirus as well as model viruses. The intended virus clearance regulatory package for this vaccine candidate includes the viruses shown in Table II. The process scheme for this candidate includes detergent inactivation followed by chromatography and virus filtration steps. Because of the presence of Bac-
Figure 1 Viral clearance for XMuLV (a, b) and PPV (c, d) across resin (a, c) and adsorber (b, d) brands and six conditions differing in spike impurity levels. Conditions defined as follows: Cond 1, HCP; Cond 2, Genomic DNA; Cond 3, Fragmented DNA; Cond 4, No Spike Control; Cond 5, HCP ⴙ Fragmented DNA; Cond 6, HCP ⴙ Genomic DNA. 184
PDA Journal of Pharmaceutical Science and Technology
Downloaded from journal.pda.org on September 18, 2016
TABLE II Viruses Intended for Inclusion in Virus Clearance Demonstration for Vaccine Candidate Produced in Insect Cell/Baculovirus System Virus
Virus Type
Baculovirus XMuLV
Enveloped Enveloped
PPV
Non-enveloped
Size
Rationale for Inclusion
250–300 nm 80–130 nm 18–26 nm
9
High titer (⬃10 vp/mL) in cell culture harvest Model retrovirus; reverse transcriptase activity is present in some insect cell lines Non-enveloped virus. PPV was selected over MMV for this vaccine candidate because the process streams presented significant toxicity issues for the indicator cells used for the MMV infectivity assay
ulovirus in process streams, the strategy for demonstrating Baculovirus clearance varies slightly depending on the unit operation. The feeds to the detergent inactivation and capture chromatography steps contain high levels of Baculovirus; therefore, clearance can be measured without spiking Baculovirus to the feed. For more downstream steps including virus filtration, no infectious Baculovirus remains, so infectious Baculovirus stock is spiked to the step in order to measure virus clearance.
cal. Although it was intended to include data for only one parvovirus in the regulatory submission, data was gathered for both MMV and PPV to help understand effects of the parvovirus spike on elution profile, as discussed below. PPV concentration was quantified using an infectivity assay for this study. MMV concentration was measured by quantitative PCR (qPCR), because the process streams presented significant toxicity issues for the indicator cells used for the MMV infectivity assay.
Results from Clearance Studies for AEX Capture Step
The data in Table III demonstrates that the virus clearance capability of bind/elute AEX depends strongly on the NaCl concentration used for elution. Significantly better clearance of Baculovirus and PPV are obtained for purification of Product 1, which is eluted using 150 mM NaCl, than for Product 2, which is eluted using 400 mM NaCl. Only feed and product fractions were analyzed, so full virus mass balances were not obtained. However, based on the data collected, our hypothesis is that when 150 mM NaCl elution is used, most Baculovirus or PPV remains on the AEX column during elution. The 400 mM NaCl required to elute Product 2 results in elution of more Baculovirus or PPV, so the virus clearance capability of the step is compromised. The relationship between
The vaccine candidate contains multiple recombinant proteins. The purification process for two of these proteins, Product 1 and Product 2, uses a bind/elute AEX capture step; the amount of NaCl required to elute these target proteins varies based on the protein characteristics. Virus clearance data was collected for Baculovirus and two non-enveloped viruses—PPV and mouse minute virus (MMV)—for the AEX step for each of these proteins, and is compiled in Table III. Note that XMuLV was not explored in this study; the process contains multiple orthogonal steps capable of removing XMuLV and demonstration of the capability of AEX to remove XMuLV was not considered criti-
TABLE III Virus Clearance Results for Bind/Elute AEX for Two Product Proteins, Which Require Different NaCl Levels To Elute the Product Protein from the AEX Column
Virus
Assay
Logs Virus Clearance, Product 1 Elution with 150 mM NaCl, pH 8
Baculovirus PPV MMV
qPCR Infectivity qPCR
5.5 3.7 1.3
Vol. 69, No. 1, January–February 2015
Logs Virus Clearance, Product 2 Elution with 400 mM NaCl pH 8 1.2 ⬍1 1.7 185
Downloaded from journal.pda.org on September 18, 2016
Figure 2 Elution profiles for PPV runs using Product 1, which is eluted using 150 mM NaCl. A non-spiked run was performed using the same processing scale as the PPV virus clearance runs and is included for reference. The elution profile following spike of 2% PPV stock is significantly different than the non-spiked run.
elution salt concentration and the virus clearance capability of bind/elute AEX has been reported previously (6). The data also shows the differing levels of clearance obtained for PPV and MMV. For MMV, the amount of virus clearance is relatively low even using 150 mM NaCl elution.
product recovered during the elution. In order to achieve a more representative elution peak, runs were performed with lesser amounts of PPV spiked to the feed. As shown in Figure 2, reducing the spike to 0.2% results in an elution profile that more closely resembles the non-spiked AEX run. The data in Table IV shows that the amount of PPV clearance demonstrated by the step is not significantly affected by lowering the amount of PPV spiked.
Effects of Parvovirus Spike on AEX Elution Profile During the virus clearance experiments using bindelute AEX, the elution profile was significantly affected by the model virus spike for the initial PPVspiked run. Because the virus clearance experiment must resemble the production operation as closely as possible, ways to eliminate the effect on the elution profile were explored. Because both the target protein and PPV are known to bind to AEX resins, it was hypothesized that PPV preferentially binds during the column load, displacing the target protein and reducing the amount of 186
The runs are described in terms of percent virus spike; the concentrations of the virus stocks used were re-
TABLE IV Log10 PPV Removal by the AEX Capture Step, for Varying PPV Spike Level PPV Spike
Logs Virus Clearance for AEX (Product 1)
2% 0.2% 0.02%
3.7 3.6 ⬎3.3
PDA Journal of Pharmaceutical Science and Technology
Downloaded from journal.pda.org on September 18, 2016
Figure 3 Elution profiles (A280) for bind-elute AEX (Product 1) using PPV as model virus. A non-spiked run was performed using the same process scale and is shown for reference. The elution step for these runs uses 150 mM NaCl at pH 8. ported by the Contract Research Organization (CRO) as 108 virus particles/mL for PPV (infectivity assay) and 9 ⫻ 108 virus particle/mL for MMV (qPCR) (Figure 3). These values indicate that a 2% MMV spike contains more viral particles than a 2% PPV spike, but the use of the qPCR assay means that the MMV value could contain an unknown percentage of genomes not associated with an infectious viral particle. Discussion MMV and PPV performed differently in these AEX chromatography studies with respect to clearance obtained and effect on the elution profile. The isoelectric point of MMV has been reported as 6.2 (7). Discussion at the symposium indicated that the isoelectric point of PPV has been measured at 5.5. This difference would suggest that PPV may have a stronger affinity for AEX resins than MMV. While the virus partitioning in these virus clearance studies has not been fully characterized, the data from Vol. 69, No. 1, January–February 2015
these studies would also suggest a stronger AEX affinity for PPV than for MMV. A spike of 2% PPV stock to the AEX feed significantly affected the product elution profile, while a spike of 2% MMV did not. It is hypothesized that the effect on the elution profile is due to displacement of the product during the load; with this hypothesis in mind, the data suggests that PPV was able to displace the product more effectively than MMV. Also, the amount of PPV clearance obtained at 150 mM and 400 mM NaCl elution suggest that PPV is strongly bound to the column at 150 mM NaCl. The fact that MMV clearance using 150 mM NaCl elution was not as effective suggests that MMV may not be as strongly bound to the resin as PPV under these conditions. Similar studies with full characterization of MMV and PPV partitioning would have to be run to confirm these hypotheses regarding MMV and PPV performance. It must also be noted that some of the difference observed in the MMV and PPV clearance values may be due to the different assays used (qPCR versus infectivity). An MMV infectivity assay could not be 187
Downloaded from journal.pda.org on September 18, 2016
Figure 4 RVLP quantifiction by qPCR: (a) Absolute and linear assay behavior; (b) sample freeze/thaw stability. used for this study because the process streams present significant toxicity issues for the indicator cells used in the MMV infectivity assay. Quantification of Retrovirus-Like Particle (RVLP) Removal at Manufacturing Scale (Christian H. Bell, Jutta Mayr, Marit Raible, Stefan Hepbildikler, Roche Diagnostics GmbH) RVLPs are the major viral contaminant in the production of biopharmaceuticals through cultivation of genetically engineered Chinese hamster ovary (CHO) cells. Because these particles are non-infectious, they can only be quantified by either transmission electron microscopy (TEM) or quantitative reverse transcriptase PCR (RT-qPCR). We have developed a method to isolate, purify, and quantify retroviral RNA using automated nucleic acid purification and RT-qPCR. The method allows one to study the actual, endogenously expressed viruses 188
within the actual manufacturing environment and is thus reflecting retrovirus removal without any model bias. RVLP counts from a variety of in-process samples including cell culture fluids can be measured. The method is fully validated and run in a quality control (QC) environment. Total nucleic acid extraction is performed using the high-throughput, automated MagNA Pure LC 2.0 (MagNA Pure and LightCycler are trademarks of Roche) system followed by DNase treatment. Quantification is carried out by real-time one-step RTqPCR on a LightCycler® 480 instrument using RVLP specific primers and probe [based on de Wit et al. (9)]. The assay allows absolute quantification of RNA with an external standard curve down to 2.5 copies/L and shows linear behavior over a range of over 6 logs (Figure 4a). Sample storage conditions have been established that enable freeze/thaw stability (Figure 4b). PDA Journal of Pharmaceutical Science and Technology
Downloaded from journal.pda.org on September 18, 2016
TABLE V RVLP Removal by Protein A Affinity Chromatography Protein A NL2 Mab6 Run1 Mab6 Run2 Mab7 Mab8 Mab9 PZ14 1 RVLP and XMuLV
LRV RVLP1
LRV XMuLV1
1.97 1.66 1.59 3.08 2.38 2.18 1.85 were quantified by
2.73 2.21 2.21 2.96 2.05 2.02 2.35 qPCR.
Utilizing this method we studied RVLP removal by affinity chromatography for six independent mAb purification processes at manufacturing scale. To verify our findings we also performed classical validation studies for these products using the model virus XMuLV in a validated small-scale model system. Our data show excellent comparability between RVLP removal at scale and XMuLV removal in small scale (Table V). This is in line with previous findings (8) and further strengthens the validity of claiming RVLP removal from at-scale samples to ensure viral safety. Using this approach, viral clearance data, at least for affinity chromatography, can be conveniently generated in a Biosafety Level 1 (BSL-1) laboratory along with other routine in-process controls. For quality-bydesign strategies, this means that RVLP removal can be studied in the same model system using the same experimental (DoE) set-up used for process characterization and validation (PC/PV) studies (Table V). RVLPs simply represent an additional impurity along with other possible critical quality attributes like HCP or DNA content. Because the possible parameter effect on RVLP and thus virus removal can be evaluated directly from PC/PV studies, this method offers a convenient, fast, and cost-effective way of establishing a design space for affinity chromatography (10). Viral Clearance Strategy—General Considerations (Hannelore Willkommen, RBS-Consulting) Integration of the viral clearance strategy means the implementation of effective orthogonal steps for virus removal in the manufacturing process. For the characterization of process steps for virus clearance, models are used that are associated with limitations which Vol. 69, No. 1, January–February 2015
are greater if data are simply accumulated from the investigation of similar processes used for production of similar products (manufacturing platform) than if operational parameters are systematically investigated in a DoE-like study. It needs to be considered that results of virus clearance studies may not allow predictions or conclusion on the effectiveness of a unit operation in a general way. Limitations may come from ● Type and number of viruses selected for experimental studies: normally a set of three or four viruses are used, other viruses may differ in size, charge, and other properties so that, for example, partitioning in removal steps and resistance to inactivation may differ; possible virus contamination of raw materials is not completely predictive and cannot fully considered in selection of model viruses; test methods in place to control bioreactor harvests for adventitious virus are not able to detect everything. ● Operational parameters which in combination may influence virus clearance or additional process parameters might be critical but are not studied sufficiently or are not identified so far. ● Inherent inaccuracy of the downscaling: the model used may not completely reflect manufacturing conditions; sampling in chromatography steps (flowthrough, wash, pre-elution, elution, post-elution, and strip) may differ to some extent from manufacturing; the different fractions are normally not tested separately so that small changes in virus partitioning are not noticed; virus spike may influence the composition of the intermediate and influence virus partitioning, and their influence cannot be analyzed because virus spike preparation and possibly analytical parameters are not sufficiently provided in the study report. The AEX chromatography (e.g., Q-Sepharose SFF and similar) in the flowthrough mode is well characterized and the proposed design-space parameters are pH 7.0 – 8.5, conductivity ⬍14 mS/cm, effective loading ⬍100 mg/mL resin, and the resin shows not different properties if reused up to 50 times (11). It was understood that HCP and DNA may influence virus binding, so the concentration of both should be used as an additional parameter. The parameters developed preferentially for MuLV and MMV are working well in many applications, but deviations have been observed (12). Despite the fact that AEX chromatography is a well189
Downloaded from journal.pda.org on September 18, 2016
characterized step, it might be wise to perform a control experiment to show that the design-space predictions are valid for the case in question. Chromatography in the binding mode is normally more difficult to characterize. Overlapping of product peak and virus peak is possible so that the correct sampling is important; the effectiveness of the chromatographic step for virus clearance may be only moderately effective and more variable in such case; it might even be possible that the results differ from one study (or laboratory) to the other (laboratory) dramatically. There are, of course, cases where such steps are effective for virus clearance and are reproducible, but the current knowledge does not allow general predictions, so that currently the product-related virus clearance study or at least a short confirmation study is needed. It is surprising that although it is known that virus preparation properties can affect the outcome of virus clearance studies, so little information is provided in the study reports. Virus titers and reduction factors should be reported with the 95% confidence limits, and if virus is quantified by PCR then the virus stocks should be controlled for free virus-specific DNA/ RNA. The ratio between infectivity titer and PCR titer would be useful information. Overall, virus clearance results from the combination of several steps, and it is requested that the overall capacity of the process significantly exceeds the possible level of contamination. This is a compensation of the inaccuracies possibly associated with the individual virus clearance studies. In the analysis of the individual steps and in particular in general conclusions taken from them, a critical scientifically sounded analysis must be applied and any remaining uncertainty should be clarified by appropriate experiments.
vectors, vaccines, and mAbs. In 2001, the FDA Vaccine and Related Biological Products Advisory Committee supported the use of Per.C6 in the production of a live viral vaccine. Human cells also have been used as cell substrates for vaccines, for cell therapies, and more recently stem cells have been proposed for a variety of therapies. The use of human cells has a history of safety even when there is minimal or no purification. With regard to endogenous retroviruses, rodent cells are known to express endogenous retroviruses. There is no evidence of endogenous retrovirus expression in Per.C6 cells and no known reports of endogenous retrovirus expression in other human cell substrates or in human cell therapies. Although human endogenous retrovirus expression was described in an early report (13), it is, in fact, very rare. Evaluations of viral clearance studies for mAbs follow the ICH Q5A guideline. Because there is no evidence of endogenous retrovirus expression or other viruses in Per.C6 cells, the action plan for process assessment would be considered Case A: viral clearance studies using non-specific model viruses because there is no specific virus to use as a model. Also, it seems that a safety factor calculation for retroviruses, as is performed for rodent cell banks that use TEM retrovirallike particle counts as the amount or “load” entering the process, is not warranted for Per.C6 because there is no basis for the amount of virus in the bulk harvest. However, the typical non-specific model virus clearance should meet the target of approximately 6 logs (14).
Opportunities and Challenges of Working with a Human Cell Substrate (Dominick Vacante, Janssen Research & Development, LLC)
In summary, as human cells have been used for production of products with limited or no purification, it seems reasonable that Per.C6 cells may provide an opportunity to produce large-molecule products that might not be amenable to standard low pH or virus filter and/or have limited purification. The benefit or opportunity would be balanced by an appropriate riskbased approach.
Monoclonal antibodies (mAbs) are typically produced using rodent cell substrates (CHO, murine myeloma); however, development of the human cell substrate Per.C6® provides an opportunity to also use these cells as a substrate for mAb production. Per.C6 cells are human cells immortalized using recombinant DNA technology to provide a continuous cell line. The cells were designed to facilitate production of gene therapy
Although production with human cell substrates may have benefits, there are also challenges. One of these challenges is the potential for false positives in TEM testing. Various structures in TEM such as coated vesicles, nuclear pores, and non-descript debris may be interpreted as RVLPs by a conservative test lab. A thorough investigation that includes a risk assessment may be necessary to resolve a false positive from a
190
PDA Journal of Pharmaceutical Science and Technology
Downloaded from journal.pda.org on September 18, 2016
true contamination. Additional testing may also be required. Because TEM is a visual observation of structures and substructure characteristics, external experts can review the same photographic images and provide an independent interpretation. The images themselves can be reprinted to possibly adjust contrast to visualize substructures for more accurate identification. Additional sections from the same TEM sample block can also be analyzed to provide, in principle, many more images of structures for interpretation within the same sample. Lastly, additional tests may be performed for detection of the presence of retroviruses: ● PCR-based reverse transcriptase (RT) testing. ● Cultivation studies with susceptible detector cell lines. ● Analyses of the end of production cells, if available. 䡩 TEM, RT, and co-cultivation studies. To ensure adequate control of the TEM test, an action limit and a pre-determined action plan for resolution of reported structures may be implemented. In conclusion, human cells may provide opportunities for production products with limited purification and/or limited strategies for viral clearance such as mAbs. However, there are challenges that are unique to this cell type and one of those, TEM testing, has been described which included an action limit and action plan to ensure the test is adequately controlled. Risk-Based Justification for Why Master Cell Bank (MCB) and Unprocessed Bulk (UPB) Are Screened by In Vitro Viral (IVV) Testing with Different Assay Durations (Dayue Chen, Eli Lilly) IVV testing is used to detect adventitious viruses in cell substrates such as MCB and UPB. The IVV assays for adventitious viruses screening are designed to detect a broad range of viruses by using multiple readouts such as cytopathogenic effects (CPE), hemagglutination, and hemadsorption. These assays typically employ three or more cell lines and can be carried out with either 14 day or 28 day duration. The 28 day assay involves an additional passage intended to improve the assay sensitivity. MCB and UPB are screened for adventitious viruses by the cell based IVV assay with 28 day and 14 day duration at Eli Lilly Vol. 69, No. 1, January–February 2015
and Company, respectively. This is based on the distinct risk profiles of MCB and UPB as well as their respective positions in the good manufacturing practice (GMP) manufacture chain as outline below. First, MCB is the starting cell substrate for the life cycle of the product, and a contaminated MCB would potentially affect the entire product franchise. In contrast, each individual UPB represents only one single bioreactor harvest batch, hence any contamination would be limited to the given batch involved. Secondly, because MCB is the starting cell substrate, low-level viral contaminant could turn into a fullblown contamination event through adaptation and propagation in subsequent extensive cell expansions. On the other hand, UPB is at the end of the cell culture process, and low-level viral contamination introduced during harvest will not be able to further propagate and will be readily removed by the downstream purification process. Thirdly, MCB production involves multiple manipulations from vial thaw, cell passage, centrifugation, re-suspension, to final vial aliquot, all taking place in an open environment. Viral contaminants could be inadvertently introduced into the MCB during any of these manipulations with similar probability via culture medium, cryopreservation medium, or operators. Conversely, UPB is produced by gradual bioreactor-to-bioreactor scale-up process in a closed system and viral contaminant is most likely introduced into the UPB either at the time of inoculation of the production bioreactor or earlier, should it occur. Assuming a single infectious virus is introduced into UPB during the inoculation of the production bioreactor, the single virus would have sufficient time to adapt and produce enormous amount of progeny viruses if it can initiate the infection, making it readily detectable by 14 day in vitro assay. This notion can be best illustrated by the vesivirus 2117 example experienced by Genzyme. It was found that vesivirus 2117 isolated from the contaminated bioreactor failed to cause noticeable CPE without passage in CHO cells when inoculated at very low level. However, inoculation of CHO cells with the cell culture materials from the vesivirus 2117– contaminated bioreactor consistently resulted in obvious CPE between 5 and 10 days post-inoculation (15, 16). If the virus could not initiate the infection, it does not pose any safety threat as the contaminant will certainly be removed by the downstream purification process. Finally, there is an inherent risk of viral contamination during in vitro virus assay, resulting in a false-positive testing result. The most vulnerable points for introducing viral contami191
Downloaded from journal.pda.org on September 18, 2016
TABLE VI Resin Age and Column Performance
Step
Resin Age
Chromatography Step A
New EOL New EOL New EOL
Chromatography Step B Chromatography Step C
Achieved Viral Reduction (Log10 TCID50) XMuLV MMV PRV Reo-3 Ab1 Ab2 Ab1 Ab2 Ab1 Ab2 Ab1 Ab2 2.94 3.33 ⱖ5.99 ⱖ5.47 ⱖ3.57 3.73
2.68 2.77 ⱖ4.99 4.59 ⱖ4.91 3.84
2.24 2.03 4.35 3.47 NT NT
1.49 1.63 5.22 ⱖ4.55 NT NT
NT NT ⱖ5.99 ⱖ6.09 NT NT
NT NT ⱖ6.48 ⱖ6.74 NT NT
0.94 1.05 5.94 4.97 NT NT
NT NT ⱖ6.83 ⱖ6.30 NT NT
NT⫽ not tested. nants are during the initial inoculation and subsequent medium change or culture passage. The additional passage in 28 day assay inevitablely increases the risk of false-positive testing results. Based on the above assessment, it makes scientific sense to test MCB using the 28 day IVV assay in order to maximize the possibility of detecting low levels of viral contaminants. The increased risk of potential false positives associated with the 28-day assay is worth taking in order to maximize the detection of low-level or slowgrowing viruses. Design of Viral Clearance Study To Support Marketing Authorization (Olga Galperina, GSK)
elution pool, strip, sanitization fractions are sampled and tested. y Virus carry-over run is performed only on the endof-lifetime (EOL) resin. - No assessment of viral distribution is done on EOL resin, but samples are collected and frozen. If new and EOL resin clearance results are significantly different, distribution assessment can be done for the aged resin samples. - Column is cleaned using batch record cleaning procedures and put into storage buffer for the shortest time allowable in production before execution of carry-over runs.
Viral clearance study in support of BLA submission can be a challenging exercise, especially for chromatography steps. The question is, how to design a scientifically sounded, but manageable, study that meets all objectives outlined in ICH Q5A. This presentation summarized GSK Rockville approach to such studies, which allows minimizing the number of runs in the virus clearance study for each chromatography step:
This approach results in three runs per model virus (two with spiked load and one carry-over run) and allows us to achieve all ICH Q5A objectives. In-house data on the performance of new versus EOL resins indicate that there is no significant impact of resin age on the performance of chromatography steps (Table VI).
y Only steps where effective viral clearance is expected are selected for viral validation. For example, cation exchange or hydrophobic interaction chromatography columns can be excluded or only used to assess clearance of some viruses (for example, retroviruses).
Additionally available data suggest that conventional cleaning storage procedures, which include at least 60 min exposure to 0.5 N or 1 N NaOH can effectively prevent virus carry-over between chromatography runs. Two chromatographic columns were investigated with XMuLV, MMV, Reo-3, and pseudorabies virus (PRV) and in no case virus was detected in the post-cleaning pool.
y Virus distribution is assessed only on fresh resin. - Multiple process fractions like flowthrough, wash (multiple washes could be combined or assessed separately depending on the project), 192
One of the main objectives of this presentation was to initiate discussions on the creative alternative approaches to design of BLA viral clearance studies. PDA Journal of Pharmaceutical Science and Technology
Downloaded from journal.pda.org on September 18, 2016
Summary The experimental comparison of AEX column chromatography and AEX membranes resulted in the conclusion that both methods provide equivalent LRV ranges. Even if ”outliers” were observed when the AEX membrane was challenged with PPV, clearance remained at a level of 3 log10. Impurity levels (HCP, DNA) have to be considered in both cases as critical parameters. AEX chromatography was used as a capture step in the production of a recombinant protein in the Baculovirus/insect cell system; Baculovirus is produced at a level of 109 particles per milliliter so that this virus can directly be measured in the processed intermediates. It was demonstrated that virus clearance (Baculovirus as well as PPV and MMV) by this AEX capture step depends on the salt concentration of the elution buffer (150 mM versus 400 mM NaCl), and the lower revealed better clearance. Differences between PPV and MMV clearance were observed, but it was not finally clarified whether this result is caused by a different binding strength of both parvoviruses or is related to the detection methods (infectivity assay versus qPCR). These results will further be investigated. RVLPs are produced by genetically engineered CHO cells partly at very high levels. They can be quantified by RT-qPCR. The method is fully validated and runs in a QC environment. Six independent mAb purification processes at manufacturing scale were compared with down-scaled experiments using XMuLV as model. The data showed complete comparability between RVLP removal at scale and XMuLV removal in small scale. This allows applying this method on inprocess samples that can be analyzed in a BSL-1 laboratory along with other routine in-process controls. Human cells, in contrast to rodent cells, do not produce retrovirus or RVLPs. Human cell substrates, like Per.C6 cells, fall therefore in the Case A category of the ICH Q5A guideline. Although production with human cell substrates may have benefits, beside the absence of endogenous retrovirus or RVLPs, there are also challenges. One of these challenges is the potential for false positives in TEM testing. Various structures in TEM such as cell-related coated vesicles, nuclear pores, and non-descript debris may be interpreted as RVLPs by a conservative test laboratory. The test itself and the interpretation of the results should therefore be adequately controlled. Vol. 69, No. 1, January–February 2015
IVV assays for adventitious viruses screening are designed to detect a broad range of viruses. These assays typically employ three or more cell lines and can be carried out with either 14 day or 28 day duration. The additional passage in the 28 day assay improves the assay sensitivity but may increase the risk of introducing viral contaminants during culture passage. For UPB testing, the 14 day assay may be the choice as it provides adequate virus detection and minimizes the chance of potential false positives. Virus clearance studies in support of BLA/MAA submission are cumbersome for chromatography steps. The strategy for the evaluation of a chromatographic step requires at least three runs per virus (two runs with new resin, one run with EOL resin, the last is combined with a virus carry-over run to achieve all ICH Q5A objectives). In-house data indicate that there is no significant impact of resin age on the performance of the chromatographic step, and the question remains whether this needs to be controlled for each product and resin. Virus clearance data provided for clinical trials can be based on in-house data, but this approach is not accepted so far for BLA/MAA submissions. A discussion of this regulatory approach should be initiated. References 1. Miesegaes, G. R.; Lute, S.; Read, E. K.; Brorson, K. Viral clearance by flow-through mode ion exchane columns and membrane adsorbers. Biotechnol. Prog. 2014, 30 (1), 124 –131. 2. Miesegaes, G. R.; Lute, S.; Brorson, K. Analysis of viral clearance unit operations for monoclonal antibodies. Biotechnol. Bioeng. 2010, 106 (2), 238-246. 3. Brorson, K.; Miesegaes, G. R.; Tounekti, O.; Skene, J.; Blumel, J. Conference summary: gaps, lessons learned, and areas for improvement. PDA J. Pharm. Sci. Technol. 2014, 68 (1), 83– 89. 4. Norling, L.; Lute, S.; Emery, R.; Khuu, W.; Voisard, M.; Xu, Y.; Chen, Q.; Blank, G.; Brorson, K. Impact of multiple re-use of anion-exchange chromatography media on virus removal. J. Chromatogr., A 2005, 1069 (1), 79 – 89. 5. Brorson, K.; Levy, R. Proceedings of the 2011 Viral Clearance Symposium (South San Fran193
Downloaded from journal.pda.org on September 18, 2016
cisco, CA). PDA J. Pharm. Sci. Technol. 2014, 68 (1), 1. 6. Miesegaes, G. R.; Lute, S.; Brorson, K. Analysis of viral clearance operations for monoclonal antibodies. Biotechnol. Bioeng. 2010, 106 (2), 238 –246. 7. Strauss, D. M.; Lute, S. C.; Tebaykina, Z.; Frey, D. D.; Ho, C.; Blank, G. S.; Brorson, K.; Chen, Q.; Yang, B. Understanding the mechanism of virus removal by Q sepharose fast flow chromatography during the purification of CHO-cell derived biotherapeutics. Biotechnol. Bioeng. 2009, 104 (2), 371–380. 8. Zhang, M.; Lute, S.; Norling, L.; Hong, C.; Safta, A.; O’Connor, D.; Bernstein, L. J.; Wang, H.; Blank, G.; Brorson, K.; Chen, Q. A novel, Q-PCR based approach to measuring endogenous retroviral clearance by capture protein A chromatography. Biotechnol. Bioeng. 2009, 102 (5), 1438 –1447. 9. De Wir, C.; Fautz, C.; Yu, Y. Real-time quantitative PCR for retrovirus-like particle quantification in CHO cell culture. Biologicals 2000, 28 (3), 137–148. 10. Zhang, M.; Miesegaes, G. R.; Lee, M.; Coleman, D.; Yang, B.; Trexler-Schmidt, M.; Norling, L.; Lester, P.; Brorson, K. A.; Chen, Q. Quality by design approach for viral clearance by protein A chromatography. Biotechnol. Bioeng. 2014, 111 (1), 95–103.
194
11. Miesegaes, G.; Bailey, M.; Willkommen, H.; Chen, Q.; Roush, D.; Blumel, J.; Brorson, K. Proceedings of the 2009 Viral Clearance Symposium. Devel. Biol. (Basel) 2010, 133, 3–101 12. Roush, D. Viral clearance using traditional, wellunderstood unit operations (Session I): Anion exchange chromatography (AEX). PDA J. Pharm. Sci. Technol. 2014, 68 (1), 23–29 13. Lower, R.; Boller, K.; Hasenmaier, B.; Korbmacher, C.; Muller-Lantzsch, N.; Lower, J.; Kurth, R. Identification of human endogenous retroviruses with complex mRNA expression and particle formation. Proc. Nat. Acad. Sci. USA 1993, 90 (10), 4480 – 4484. 14. ICH Q5A: Quality of biotechnology products: viral safety evaluation of biotechnology products derived from cell lines of human or animal origin. www.ICH.org. 15. Qiu, Y.; Jones, N.; Busch, M.; Pan, P.; Keegan, J.; Zhou, W.; Plavsic, M.; Hayes, M.; McPherson, J. M.; Edmunds, T.; Zhang, K.; Mattaliano, R. J. Identification and quantitation of vesivirus 2117 particles in bioreactor fluids from infected Chinese hamster ovary cell cultures. Biotechnol. Bioeng. 2013, 110 (5), 1342–1353. 16. Plavsic, M.; Qiu, Y.; Jones, N.; Keegan, J.; Woodcock, D.; Morris, J.; Davis, C.; Palermo, A.; Pomponio, R.; Scaria, A. Caliciviridae and vesivirus 2117. BioProcess. J. 2011, 9 (2), 6 –12.
PDA Journal of Pharmaceutical Science and Technology
Downloaded from journal.pda.org on September 18, 2016
An Authorized User of the electronic PDA Journal of Pharmaceutical Science and Technology (the PDA Journal) is a PDA Member in good standing. Authorized Users are permitted to do the following: ·Search and view the content of the PDA Journal ·Download a single article for the individual use of an Authorized User ·Assemble and distribute links that point to the PDA Journal ·Print individual articles from the PDA Journal for the individual use of an Authorized User ·Make a reasonable number of photocopies of a printed article for the individual use of an Authorized User or for the use by or distribution to other Authorized Users Authorized Users are not permitted to do the following: ·Except as mentioned above, allow anyone other than an Authorized User to use or access the PDA Journal · Display or otherwise make any information from the PDA Journal available to anyone other than an Authorized User ·Post articles from the PDA Journal on Web sites, either available on the Internet or an Intranet, or in any form of online publications ·Transmit electronically, via e-mail or any other file transfer protocols, any portion of the PDA Journal ·Create a searchable archive of any portion of the PDA Journal ·Use robots or intelligent agents to access, search and/or systematically download any portion of the PDA Journal ·Sell, re-sell, rent, lease, license, sublicense, assign or otherwise transfer the use of the PDA Journal or its content ·Use or copy the PDA Journal for document delivery, fee-for-service use, or bulk reproduction or distribution of materials in any form, or any substantially similar commercial purpose ·Alter, modify, repackage or adapt any portion of the PDA Journal ·Make any edits or derivative works with respect to any portion of the PDA Journal including any text or graphics ·Delete or remove in any form or format, including on a printed article or photocopy, any copyright information or notice contained in the PDA Journal
Session 2/3: Integrated Viral Clearance Strategy and Case Studies Hannelore Willkommen
PDA J Pharm Sci and Tech 2015, 69 183-194 Access the most recent version at doi:10.5731/pdajpst.2015.01042
Downloaded from journal.pda.org on September 18, 2016
CONFERENCE PROCEEDING
Session 2/3: Integrated Viral Clearance Strategy and Case Studies HANNELORE WILLKOMMEN Regulatory Affairs & Biological Safety (RBS) Consulting ©PDA, Inc. 2015 Background Integrated virus clearance strategy in manufacture of medicinal products is the implementation of unit operations that are effective for virus clearance, well understood in its mechanism of action (MoA), and well to control because the critical process parameters (CPPs) are identified and maintained in the accepted range. If the virus clearance capacity shall be predictive for other products, the investigation of the MoA should clearly show that virus clearance by the selected unit operation is not product-dependent. Unit operations following different MoAs (these are orthogonal steps) should be combined in manufacturing. The data base needed to develop the integrated virus clearance strategy can be built up from the investigation of similar unit operations used for production of several products (platform manufacturing) or from systematic investigations identifying those parameters or parameter ranges that influence virus removal/inactivation by a particular unit operation (design-of-experiment, DoE, studies). There are unit operations that are well understood in their MoA. These are especially anion exchange (AEX) chromatography in flowthrough mode, filtration using virus-retentive filters of small pore size for larger viruses (like retrovirus), and solvent/detergent (S/D) treatment using tri-N-butyl-phosphate and Triton X100 to inactivate enveloped viruses. Other methods are not yet sufficiently investigated to fully understand the MoA or the MoA is so complex that the virus clearance capacity cannot be predicted. The two sessions of the 2013 Virus Clearance Symposium were attributed to three areas: to discuss (1)
e-mail: [email protected] doi: 10.5731/pdajpst.2015.01042
Vol. 69, No. 1, January–February 2015
factors influencing virus clearance and selection of model viruses, (2) problems arising from testing for adventitious agents, and finally (3) testing of re-used resins and its need for Biologic License Application/ Marketing Authorisation Application (BLA/MAA) submissions. The general descriptions and case studies are reported in this order and summarized at the end of this report. DoE-Based AEX Technology Comparison (George Miesegaes, CDER/FDA) [A larger body of this work has been published recently (1)] For many years AEX resin-based chromatography has been used for robust partitioning of the desired protein intermediate from host cell DNA and host cell protein (HCP). In addition, consistent viral clearance, with values reaching upwards of ⱖ4.0 log10, is often possible (2). More recently, there has been an observed increase in the use of AEX chromatographic membranes, also named membrane adsorbers, as an alternative to the traditional column-based approach (3). Key advantages include increases in capacity and flow rate, which afford reduced processing times and overall resource consumption, and other features such as disposability (Table I). While the mechanism of impurity removal is considered the same, a direct, sideby-side comparison of each modality was warranted. In this study, AEX column (n ⫽ 4) and adsorber (n ⫽ 3) brands were systematically assessed to investigate the extent of performance overlap. A single flowthrough mode run condition was first identified by entering queries into a previously reported viral clearance database (2) and was found to be similar to that used by Norling et al. (4). These parameters had later served as the centerpoint condition for a resolution III DoE study, published elsewhere (1). The types and amounts of impurities present in a feedstock may affect viral clearance (e.g., presentation by J. Glynn, Pfizer (5)). To further investigate this, Xenotropic murine leukemia virus (XMuLV) and por183
Downloaded from journal.pda.org on September 18, 2016
TABLE I General Features of AEX Resin and Membrane Technologies Feature
AEX Resin
AEX Membrane
Scalability
Yes
Yes
Reuse
Yes
No; disposable
Flow/particle capture
Diffusion-based
Adsorption-based
⫹⫹⫹⫹
?
Robustness of viral clearance
cine parvovirus (PPV) clearance was assessed under a range of conditions where DNA and protein impurities were spiked into the feed at known concentrations (Figure 1). Log10 reduction values (LRVs) obtained [polymerase chain reaction (PCR) assay] were found robust (ⱖ4 log10) under most of the conditions tested. While two outliers were noted (Figure 1d), their clearance values were approaching 3 log10 and were considered adequate. It was concluded that under the conditions tested, both AEX technologies were capable of achieving equivalent LRV ranges.
Viral Clearance for Vaccine Candidate Using Baculovirus Expression System Focusing On Aex Bind-Elute Chromatography (Adam Kristopeit, Merck, Sharp and Dohme, Inc.) Strategy for Clearance Studies Using the Baculovirus System Data presented is from a virus clearance evaluation for a vaccine candidate produced in an insect cell/Baculovirus expression system. In this expression system, antigens of interest are produced by infecting insect cell culture with a recombinant Baculovirus that contains the genetic information for the antigen. Virus clearance studies are required to demonstrate removal of Baculovirus as well as model viruses. The intended virus clearance regulatory package for this vaccine candidate includes the viruses shown in Table II. The process scheme for this candidate includes detergent inactivation followed by chromatography and virus filtration steps. Because of the presence of Bac-
Figure 1 Viral clearance for XMuLV (a, b) and PPV (c, d) across resin (a, c) and adsorber (b, d) brands and six conditions differing in spike impurity levels. Conditions defined as follows: Cond 1, HCP; Cond 2, Genomic DNA; Cond 3, Fragmented DNA; Cond 4, No Spike Control; Cond 5, HCP ⴙ Fragmented DNA; Cond 6, HCP ⴙ Genomic DNA. 184
PDA Journal of Pharmaceutical Science and Technology
Downloaded from journal.pda.org on September 18, 2016
TABLE II Viruses Intended for Inclusion in Virus Clearance Demonstration for Vaccine Candidate Produced in Insect Cell/Baculovirus System Virus
Virus Type
Baculovirus XMuLV
Enveloped Enveloped
PPV
Non-enveloped
Size
Rationale for Inclusion
250–300 nm 80–130 nm 18–26 nm
9
High titer (⬃10 vp/mL) in cell culture harvest Model retrovirus; reverse transcriptase activity is present in some insect cell lines Non-enveloped virus. PPV was selected over MMV for this vaccine candidate because the process streams presented significant toxicity issues for the indicator cells used for the MMV infectivity assay
ulovirus in process streams, the strategy for demonstrating Baculovirus clearance varies slightly depending on the unit operation. The feeds to the detergent inactivation and capture chromatography steps contain high levels of Baculovirus; therefore, clearance can be measured without spiking Baculovirus to the feed. For more downstream steps including virus filtration, no infectious Baculovirus remains, so infectious Baculovirus stock is spiked to the step in order to measure virus clearance.
cal. Although it was intended to include data for only one parvovirus in the regulatory submission, data was gathered for both MMV and PPV to help understand effects of the parvovirus spike on elution profile, as discussed below. PPV concentration was quantified using an infectivity assay for this study. MMV concentration was measured by quantitative PCR (qPCR), because the process streams presented significant toxicity issues for the indicator cells used for the MMV infectivity assay.
Results from Clearance Studies for AEX Capture Step
The data in Table III demonstrates that the virus clearance capability of bind/elute AEX depends strongly on the NaCl concentration used for elution. Significantly better clearance of Baculovirus and PPV are obtained for purification of Product 1, which is eluted using 150 mM NaCl, than for Product 2, which is eluted using 400 mM NaCl. Only feed and product fractions were analyzed, so full virus mass balances were not obtained. However, based on the data collected, our hypothesis is that when 150 mM NaCl elution is used, most Baculovirus or PPV remains on the AEX column during elution. The 400 mM NaCl required to elute Product 2 results in elution of more Baculovirus or PPV, so the virus clearance capability of the step is compromised. The relationship between
The vaccine candidate contains multiple recombinant proteins. The purification process for two of these proteins, Product 1 and Product 2, uses a bind/elute AEX capture step; the amount of NaCl required to elute these target proteins varies based on the protein characteristics. Virus clearance data was collected for Baculovirus and two non-enveloped viruses—PPV and mouse minute virus (MMV)—for the AEX step for each of these proteins, and is compiled in Table III. Note that XMuLV was not explored in this study; the process contains multiple orthogonal steps capable of removing XMuLV and demonstration of the capability of AEX to remove XMuLV was not considered criti-
TABLE III Virus Clearance Results for Bind/Elute AEX for Two Product Proteins, Which Require Different NaCl Levels To Elute the Product Protein from the AEX Column
Virus
Assay
Logs Virus Clearance, Product 1 Elution with 150 mM NaCl, pH 8
Baculovirus PPV MMV
qPCR Infectivity qPCR
5.5 3.7 1.3
Vol. 69, No. 1, January–February 2015
Logs Virus Clearance, Product 2 Elution with 400 mM NaCl pH 8 1.2 ⬍1 1.7 185
Downloaded from journal.pda.org on September 18, 2016
Figure 2 Elution profiles for PPV runs using Product 1, which is eluted using 150 mM NaCl. A non-spiked run was performed using the same processing scale as the PPV virus clearance runs and is included for reference. The elution profile following spike of 2% PPV stock is significantly different than the non-spiked run.
elution salt concentration and the virus clearance capability of bind/elute AEX has been reported previously (6). The data also shows the differing levels of clearance obtained for PPV and MMV. For MMV, the amount of virus clearance is relatively low even using 150 mM NaCl elution.
product recovered during the elution. In order to achieve a more representative elution peak, runs were performed with lesser amounts of PPV spiked to the feed. As shown in Figure 2, reducing the spike to 0.2% results in an elution profile that more closely resembles the non-spiked AEX run. The data in Table IV shows that the amount of PPV clearance demonstrated by the step is not significantly affected by lowering the amount of PPV spiked.
Effects of Parvovirus Spike on AEX Elution Profile During the virus clearance experiments using bindelute AEX, the elution profile was significantly affected by the model virus spike for the initial PPVspiked run. Because the virus clearance experiment must resemble the production operation as closely as possible, ways to eliminate the effect on the elution profile were explored. Because both the target protein and PPV are known to bind to AEX resins, it was hypothesized that PPV preferentially binds during the column load, displacing the target protein and reducing the amount of 186
The runs are described in terms of percent virus spike; the concentrations of the virus stocks used were re-
TABLE IV Log10 PPV Removal by the AEX Capture Step, for Varying PPV Spike Level PPV Spike
Logs Virus Clearance for AEX (Product 1)
2% 0.2% 0.02%
3.7 3.6 ⬎3.3
PDA Journal of Pharmaceutical Science and Technology
Downloaded from journal.pda.org on September 18, 2016
Figure 3 Elution profiles (A280) for bind-elute AEX (Product 1) using PPV as model virus. A non-spiked run was performed using the same process scale and is shown for reference. The elution step for these runs uses 150 mM NaCl at pH 8. ported by the Contract Research Organization (CRO) as 108 virus particles/mL for PPV (infectivity assay) and 9 ⫻ 108 virus particle/mL for MMV (qPCR) (Figure 3). These values indicate that a 2% MMV spike contains more viral particles than a 2% PPV spike, but the use of the qPCR assay means that the MMV value could contain an unknown percentage of genomes not associated with an infectious viral particle. Discussion MMV and PPV performed differently in these AEX chromatography studies with respect to clearance obtained and effect on the elution profile. The isoelectric point of MMV has been reported as 6.2 (7). Discussion at the symposium indicated that the isoelectric point of PPV has been measured at 5.5. This difference would suggest that PPV may have a stronger affinity for AEX resins than MMV. While the virus partitioning in these virus clearance studies has not been fully characterized, the data from Vol. 69, No. 1, January–February 2015
these studies would also suggest a stronger AEX affinity for PPV than for MMV. A spike of 2% PPV stock to the AEX feed significantly affected the product elution profile, while a spike of 2% MMV did not. It is hypothesized that the effect on the elution profile is due to displacement of the product during the load; with this hypothesis in mind, the data suggests that PPV was able to displace the product more effectively than MMV. Also, the amount of PPV clearance obtained at 150 mM and 400 mM NaCl elution suggest that PPV is strongly bound to the column at 150 mM NaCl. The fact that MMV clearance using 150 mM NaCl elution was not as effective suggests that MMV may not be as strongly bound to the resin as PPV under these conditions. Similar studies with full characterization of MMV and PPV partitioning would have to be run to confirm these hypotheses regarding MMV and PPV performance. It must also be noted that some of the difference observed in the MMV and PPV clearance values may be due to the different assays used (qPCR versus infectivity). An MMV infectivity assay could not be 187
Downloaded from journal.pda.org on September 18, 2016
Figure 4 RVLP quantifiction by qPCR: (a) Absolute and linear assay behavior; (b) sample freeze/thaw stability. used for this study because the process streams present significant toxicity issues for the indicator cells used in the MMV infectivity assay. Quantification of Retrovirus-Like Particle (RVLP) Removal at Manufacturing Scale (Christian H. Bell, Jutta Mayr, Marit Raible, Stefan Hepbildikler, Roche Diagnostics GmbH) RVLPs are the major viral contaminant in the production of biopharmaceuticals through cultivation of genetically engineered Chinese hamster ovary (CHO) cells. Because these particles are non-infectious, they can only be quantified by either transmission electron microscopy (TEM) or quantitative reverse transcriptase PCR (RT-qPCR). We have developed a method to isolate, purify, and quantify retroviral RNA using automated nucleic acid purification and RT-qPCR. The method allows one to study the actual, endogenously expressed viruses 188
within the actual manufacturing environment and is thus reflecting retrovirus removal without any model bias. RVLP counts from a variety of in-process samples including cell culture fluids can be measured. The method is fully validated and run in a quality control (QC) environment. Total nucleic acid extraction is performed using the high-throughput, automated MagNA Pure LC 2.0 (MagNA Pure and LightCycler are trademarks of Roche) system followed by DNase treatment. Quantification is carried out by real-time one-step RTqPCR on a LightCycler® 480 instrument using RVLP specific primers and probe [based on de Wit et al. (9)]. The assay allows absolute quantification of RNA with an external standard curve down to 2.5 copies/L and shows linear behavior over a range of over 6 logs (Figure 4a). Sample storage conditions have been established that enable freeze/thaw stability (Figure 4b). PDA Journal of Pharmaceutical Science and Technology
Downloaded from journal.pda.org on September 18, 2016
TABLE V RVLP Removal by Protein A Affinity Chromatography Protein A NL2 Mab6 Run1 Mab6 Run2 Mab7 Mab8 Mab9 PZ14 1 RVLP and XMuLV
LRV RVLP1
LRV XMuLV1
1.97 1.66 1.59 3.08 2.38 2.18 1.85 were quantified by
2.73 2.21 2.21 2.96 2.05 2.02 2.35 qPCR.
Utilizing this method we studied RVLP removal by affinity chromatography for six independent mAb purification processes at manufacturing scale. To verify our findings we also performed classical validation studies for these products using the model virus XMuLV in a validated small-scale model system. Our data show excellent comparability between RVLP removal at scale and XMuLV removal in small scale (Table V). This is in line with previous findings (8) and further strengthens the validity of claiming RVLP removal from at-scale samples to ensure viral safety. Using this approach, viral clearance data, at least for affinity chromatography, can be conveniently generated in a Biosafety Level 1 (BSL-1) laboratory along with other routine in-process controls. For quality-bydesign strategies, this means that RVLP removal can be studied in the same model system using the same experimental (DoE) set-up used for process characterization and validation (PC/PV) studies (Table V). RVLPs simply represent an additional impurity along with other possible critical quality attributes like HCP or DNA content. Because the possible parameter effect on RVLP and thus virus removal can be evaluated directly from PC/PV studies, this method offers a convenient, fast, and cost-effective way of establishing a design space for affinity chromatography (10). Viral Clearance Strategy—General Considerations (Hannelore Willkommen, RBS-Consulting) Integration of the viral clearance strategy means the implementation of effective orthogonal steps for virus removal in the manufacturing process. For the characterization of process steps for virus clearance, models are used that are associated with limitations which Vol. 69, No. 1, January–February 2015
are greater if data are simply accumulated from the investigation of similar processes used for production of similar products (manufacturing platform) than if operational parameters are systematically investigated in a DoE-like study. It needs to be considered that results of virus clearance studies may not allow predictions or conclusion on the effectiveness of a unit operation in a general way. Limitations may come from ● Type and number of viruses selected for experimental studies: normally a set of three or four viruses are used, other viruses may differ in size, charge, and other properties so that, for example, partitioning in removal steps and resistance to inactivation may differ; possible virus contamination of raw materials is not completely predictive and cannot fully considered in selection of model viruses; test methods in place to control bioreactor harvests for adventitious virus are not able to detect everything. ● Operational parameters which in combination may influence virus clearance or additional process parameters might be critical but are not studied sufficiently or are not identified so far. ● Inherent inaccuracy of the downscaling: the model used may not completely reflect manufacturing conditions; sampling in chromatography steps (flowthrough, wash, pre-elution, elution, post-elution, and strip) may differ to some extent from manufacturing; the different fractions are normally not tested separately so that small changes in virus partitioning are not noticed; virus spike may influence the composition of the intermediate and influence virus partitioning, and their influence cannot be analyzed because virus spike preparation and possibly analytical parameters are not sufficiently provided in the study report. The AEX chromatography (e.g., Q-Sepharose SFF and similar) in the flowthrough mode is well characterized and the proposed design-space parameters are pH 7.0 – 8.5, conductivity ⬍14 mS/cm, effective loading ⬍100 mg/mL resin, and the resin shows not different properties if reused up to 50 times (11). It was understood that HCP and DNA may influence virus binding, so the concentration of both should be used as an additional parameter. The parameters developed preferentially for MuLV and MMV are working well in many applications, but deviations have been observed (12). Despite the fact that AEX chromatography is a well189
Downloaded from journal.pda.org on September 18, 2016
characterized step, it might be wise to perform a control experiment to show that the design-space predictions are valid for the case in question. Chromatography in the binding mode is normally more difficult to characterize. Overlapping of product peak and virus peak is possible so that the correct sampling is important; the effectiveness of the chromatographic step for virus clearance may be only moderately effective and more variable in such case; it might even be possible that the results differ from one study (or laboratory) to the other (laboratory) dramatically. There are, of course, cases where such steps are effective for virus clearance and are reproducible, but the current knowledge does not allow general predictions, so that currently the product-related virus clearance study or at least a short confirmation study is needed. It is surprising that although it is known that virus preparation properties can affect the outcome of virus clearance studies, so little information is provided in the study reports. Virus titers and reduction factors should be reported with the 95% confidence limits, and if virus is quantified by PCR then the virus stocks should be controlled for free virus-specific DNA/ RNA. The ratio between infectivity titer and PCR titer would be useful information. Overall, virus clearance results from the combination of several steps, and it is requested that the overall capacity of the process significantly exceeds the possible level of contamination. This is a compensation of the inaccuracies possibly associated with the individual virus clearance studies. In the analysis of the individual steps and in particular in general conclusions taken from them, a critical scientifically sounded analysis must be applied and any remaining uncertainty should be clarified by appropriate experiments.
vectors, vaccines, and mAbs. In 2001, the FDA Vaccine and Related Biological Products Advisory Committee supported the use of Per.C6 in the production of a live viral vaccine. Human cells also have been used as cell substrates for vaccines, for cell therapies, and more recently stem cells have been proposed for a variety of therapies. The use of human cells has a history of safety even when there is minimal or no purification. With regard to endogenous retroviruses, rodent cells are known to express endogenous retroviruses. There is no evidence of endogenous retrovirus expression in Per.C6 cells and no known reports of endogenous retrovirus expression in other human cell substrates or in human cell therapies. Although human endogenous retrovirus expression was described in an early report (13), it is, in fact, very rare. Evaluations of viral clearance studies for mAbs follow the ICH Q5A guideline. Because there is no evidence of endogenous retrovirus expression or other viruses in Per.C6 cells, the action plan for process assessment would be considered Case A: viral clearance studies using non-specific model viruses because there is no specific virus to use as a model. Also, it seems that a safety factor calculation for retroviruses, as is performed for rodent cell banks that use TEM retrovirallike particle counts as the amount or “load” entering the process, is not warranted for Per.C6 because there is no basis for the amount of virus in the bulk harvest. However, the typical non-specific model virus clearance should meet the target of approximately 6 logs (14).
Opportunities and Challenges of Working with a Human Cell Substrate (Dominick Vacante, Janssen Research & Development, LLC)
In summary, as human cells have been used for production of products with limited or no purification, it seems reasonable that Per.C6 cells may provide an opportunity to produce large-molecule products that might not be amenable to standard low pH or virus filter and/or have limited purification. The benefit or opportunity would be balanced by an appropriate riskbased approach.
Monoclonal antibodies (mAbs) are typically produced using rodent cell substrates (CHO, murine myeloma); however, development of the human cell substrate Per.C6® provides an opportunity to also use these cells as a substrate for mAb production. Per.C6 cells are human cells immortalized using recombinant DNA technology to provide a continuous cell line. The cells were designed to facilitate production of gene therapy
Although production with human cell substrates may have benefits, there are also challenges. One of these challenges is the potential for false positives in TEM testing. Various structures in TEM such as coated vesicles, nuclear pores, and non-descript debris may be interpreted as RVLPs by a conservative test lab. A thorough investigation that includes a risk assessment may be necessary to resolve a false positive from a
190
PDA Journal of Pharmaceutical Science and Technology
Downloaded from journal.pda.org on September 18, 2016
true contamination. Additional testing may also be required. Because TEM is a visual observation of structures and substructure characteristics, external experts can review the same photographic images and provide an independent interpretation. The images themselves can be reprinted to possibly adjust contrast to visualize substructures for more accurate identification. Additional sections from the same TEM sample block can also be analyzed to provide, in principle, many more images of structures for interpretation within the same sample. Lastly, additional tests may be performed for detection of the presence of retroviruses: ● PCR-based reverse transcriptase (RT) testing. ● Cultivation studies with susceptible detector cell lines. ● Analyses of the end of production cells, if available. 䡩 TEM, RT, and co-cultivation studies. To ensure adequate control of the TEM test, an action limit and a pre-determined action plan for resolution of reported structures may be implemented. In conclusion, human cells may provide opportunities for production products with limited purification and/or limited strategies for viral clearance such as mAbs. However, there are challenges that are unique to this cell type and one of those, TEM testing, has been described which included an action limit and action plan to ensure the test is adequately controlled. Risk-Based Justification for Why Master Cell Bank (MCB) and Unprocessed Bulk (UPB) Are Screened by In Vitro Viral (IVV) Testing with Different Assay Durations (Dayue Chen, Eli Lilly) IVV testing is used to detect adventitious viruses in cell substrates such as MCB and UPB. The IVV assays for adventitious viruses screening are designed to detect a broad range of viruses by using multiple readouts such as cytopathogenic effects (CPE), hemagglutination, and hemadsorption. These assays typically employ three or more cell lines and can be carried out with either 14 day or 28 day duration. The 28 day assay involves an additional passage intended to improve the assay sensitivity. MCB and UPB are screened for adventitious viruses by the cell based IVV assay with 28 day and 14 day duration at Eli Lilly Vol. 69, No. 1, January–February 2015
and Company, respectively. This is based on the distinct risk profiles of MCB and UPB as well as their respective positions in the good manufacturing practice (GMP) manufacture chain as outline below. First, MCB is the starting cell substrate for the life cycle of the product, and a contaminated MCB would potentially affect the entire product franchise. In contrast, each individual UPB represents only one single bioreactor harvest batch, hence any contamination would be limited to the given batch involved. Secondly, because MCB is the starting cell substrate, low-level viral contaminant could turn into a fullblown contamination event through adaptation and propagation in subsequent extensive cell expansions. On the other hand, UPB is at the end of the cell culture process, and low-level viral contamination introduced during harvest will not be able to further propagate and will be readily removed by the downstream purification process. Thirdly, MCB production involves multiple manipulations from vial thaw, cell passage, centrifugation, re-suspension, to final vial aliquot, all taking place in an open environment. Viral contaminants could be inadvertently introduced into the MCB during any of these manipulations with similar probability via culture medium, cryopreservation medium, or operators. Conversely, UPB is produced by gradual bioreactor-to-bioreactor scale-up process in a closed system and viral contaminant is most likely introduced into the UPB either at the time of inoculation of the production bioreactor or earlier, should it occur. Assuming a single infectious virus is introduced into UPB during the inoculation of the production bioreactor, the single virus would have sufficient time to adapt and produce enormous amount of progeny viruses if it can initiate the infection, making it readily detectable by 14 day in vitro assay. This notion can be best illustrated by the vesivirus 2117 example experienced by Genzyme. It was found that vesivirus 2117 isolated from the contaminated bioreactor failed to cause noticeable CPE without passage in CHO cells when inoculated at very low level. However, inoculation of CHO cells with the cell culture materials from the vesivirus 2117– contaminated bioreactor consistently resulted in obvious CPE between 5 and 10 days post-inoculation (15, 16). If the virus could not initiate the infection, it does not pose any safety threat as the contaminant will certainly be removed by the downstream purification process. Finally, there is an inherent risk of viral contamination during in vitro virus assay, resulting in a false-positive testing result. The most vulnerable points for introducing viral contami191
Downloaded from journal.pda.org on September 18, 2016
TABLE VI Resin Age and Column Performance
Step
Resin Age
Chromatography Step A
New EOL New EOL New EOL
Chromatography Step B Chromatography Step C
Achieved Viral Reduction (Log10 TCID50) XMuLV MMV PRV Reo-3 Ab1 Ab2 Ab1 Ab2 Ab1 Ab2 Ab1 Ab2 2.94 3.33 ⱖ5.99 ⱖ5.47 ⱖ3.57 3.73
2.68 2.77 ⱖ4.99 4.59 ⱖ4.91 3.84
2.24 2.03 4.35 3.47 NT NT
1.49 1.63 5.22 ⱖ4.55 NT NT
NT NT ⱖ5.99 ⱖ6.09 NT NT
NT NT ⱖ6.48 ⱖ6.74 NT NT
0.94 1.05 5.94 4.97 NT NT
NT NT ⱖ6.83 ⱖ6.30 NT NT
NT⫽ not tested. nants are during the initial inoculation and subsequent medium change or culture passage. The additional passage in 28 day assay inevitablely increases the risk of false-positive testing results. Based on the above assessment, it makes scientific sense to test MCB using the 28 day IVV assay in order to maximize the possibility of detecting low levels of viral contaminants. The increased risk of potential false positives associated with the 28-day assay is worth taking in order to maximize the detection of low-level or slowgrowing viruses. Design of Viral Clearance Study To Support Marketing Authorization (Olga Galperina, GSK)
elution pool, strip, sanitization fractions are sampled and tested. y Virus carry-over run is performed only on the endof-lifetime (EOL) resin. - No assessment of viral distribution is done on EOL resin, but samples are collected and frozen. If new and EOL resin clearance results are significantly different, distribution assessment can be done for the aged resin samples. - Column is cleaned using batch record cleaning procedures and put into storage buffer for the shortest time allowable in production before execution of carry-over runs.
Viral clearance study in support of BLA submission can be a challenging exercise, especially for chromatography steps. The question is, how to design a scientifically sounded, but manageable, study that meets all objectives outlined in ICH Q5A. This presentation summarized GSK Rockville approach to such studies, which allows minimizing the number of runs in the virus clearance study for each chromatography step:
This approach results in three runs per model virus (two with spiked load and one carry-over run) and allows us to achieve all ICH Q5A objectives. In-house data on the performance of new versus EOL resins indicate that there is no significant impact of resin age on the performance of chromatography steps (Table VI).
y Only steps where effective viral clearance is expected are selected for viral validation. For example, cation exchange or hydrophobic interaction chromatography columns can be excluded or only used to assess clearance of some viruses (for example, retroviruses).
Additionally available data suggest that conventional cleaning storage procedures, which include at least 60 min exposure to 0.5 N or 1 N NaOH can effectively prevent virus carry-over between chromatography runs. Two chromatographic columns were investigated with XMuLV, MMV, Reo-3, and pseudorabies virus (PRV) and in no case virus was detected in the post-cleaning pool.
y Virus distribution is assessed only on fresh resin. - Multiple process fractions like flowthrough, wash (multiple washes could be combined or assessed separately depending on the project), 192
One of the main objectives of this presentation was to initiate discussions on the creative alternative approaches to design of BLA viral clearance studies. PDA Journal of Pharmaceutical Science and Technology
Downloaded from journal.pda.org on September 18, 2016
Summary The experimental comparison of AEX column chromatography and AEX membranes resulted in the conclusion that both methods provide equivalent LRV ranges. Even if ”outliers” were observed when the AEX membrane was challenged with PPV, clearance remained at a level of 3 log10. Impurity levels (HCP, DNA) have to be considered in both cases as critical parameters. AEX chromatography was used as a capture step in the production of a recombinant protein in the Baculovirus/insect cell system; Baculovirus is produced at a level of 109 particles per milliliter so that this virus can directly be measured in the processed intermediates. It was demonstrated that virus clearance (Baculovirus as well as PPV and MMV) by this AEX capture step depends on the salt concentration of the elution buffer (150 mM versus 400 mM NaCl), and the lower revealed better clearance. Differences between PPV and MMV clearance were observed, but it was not finally clarified whether this result is caused by a different binding strength of both parvoviruses or is related to the detection methods (infectivity assay versus qPCR). These results will further be investigated. RVLPs are produced by genetically engineered CHO cells partly at very high levels. They can be quantified by RT-qPCR. The method is fully validated and runs in a QC environment. Six independent mAb purification processes at manufacturing scale were compared with down-scaled experiments using XMuLV as model. The data showed complete comparability between RVLP removal at scale and XMuLV removal in small scale. This allows applying this method on inprocess samples that can be analyzed in a BSL-1 laboratory along with other routine in-process controls. Human cells, in contrast to rodent cells, do not produce retrovirus or RVLPs. Human cell substrates, like Per.C6 cells, fall therefore in the Case A category of the ICH Q5A guideline. Although production with human cell substrates may have benefits, beside the absence of endogenous retrovirus or RVLPs, there are also challenges. One of these challenges is the potential for false positives in TEM testing. Various structures in TEM such as cell-related coated vesicles, nuclear pores, and non-descript debris may be interpreted as RVLPs by a conservative test laboratory. The test itself and the interpretation of the results should therefore be adequately controlled. Vol. 69, No. 1, January–February 2015
IVV assays for adventitious viruses screening are designed to detect a broad range of viruses. These assays typically employ three or more cell lines and can be carried out with either 14 day or 28 day duration. The additional passage in the 28 day assay improves the assay sensitivity but may increase the risk of introducing viral contaminants during culture passage. For UPB testing, the 14 day assay may be the choice as it provides adequate virus detection and minimizes the chance of potential false positives. Virus clearance studies in support of BLA/MAA submission are cumbersome for chromatography steps. The strategy for the evaluation of a chromatographic step requires at least three runs per virus (two runs with new resin, one run with EOL resin, the last is combined with a virus carry-over run to achieve all ICH Q5A objectives). In-house data indicate that there is no significant impact of resin age on the performance of the chromatographic step, and the question remains whether this needs to be controlled for each product and resin. Virus clearance data provided for clinical trials can be based on in-house data, but this approach is not accepted so far for BLA/MAA submissions. A discussion of this regulatory approach should be initiated. References 1. Miesegaes, G. R.; Lute, S.; Read, E. K.; Brorson, K. Viral clearance by flow-through mode ion exchane columns and membrane adsorbers. Biotechnol. Prog. 2014, 30 (1), 124 –131. 2. Miesegaes, G. R.; Lute, S.; Brorson, K. Analysis of viral clearance unit operations for monoclonal antibodies. Biotechnol. Bioeng. 2010, 106 (2), 238-246. 3. Brorson, K.; Miesegaes, G. R.; Tounekti, O.; Skene, J.; Blumel, J. Conference summary: gaps, lessons learned, and areas for improvement. PDA J. Pharm. Sci. Technol. 2014, 68 (1), 83– 89. 4. Norling, L.; Lute, S.; Emery, R.; Khuu, W.; Voisard, M.; Xu, Y.; Chen, Q.; Blank, G.; Brorson, K. Impact of multiple re-use of anion-exchange chromatography media on virus removal. J. Chromatogr., A 2005, 1069 (1), 79 – 89. 5. Brorson, K.; Levy, R. Proceedings of the 2011 Viral Clearance Symposium (South San Fran193
Downloaded from journal.pda.org on September 18, 2016
cisco, CA). PDA J. Pharm. Sci. Technol. 2014, 68 (1), 1. 6. Miesegaes, G. R.; Lute, S.; Brorson, K. Analysis of viral clearance operations for monoclonal antibodies. Biotechnol. Bioeng. 2010, 106 (2), 238 –246. 7. Strauss, D. M.; Lute, S. C.; Tebaykina, Z.; Frey, D. D.; Ho, C.; Blank, G. S.; Brorson, K.; Chen, Q.; Yang, B. Understanding the mechanism of virus removal by Q sepharose fast flow chromatography during the purification of CHO-cell derived biotherapeutics. Biotechnol. Bioeng. 2009, 104 (2), 371–380. 8. Zhang, M.; Lute, S.; Norling, L.; Hong, C.; Safta, A.; O’Connor, D.; Bernstein, L. J.; Wang, H.; Blank, G.; Brorson, K.; Chen, Q. A novel, Q-PCR based approach to measuring endogenous retroviral clearance by capture protein A chromatography. Biotechnol. Bioeng. 2009, 102 (5), 1438 –1447. 9. De Wir, C.; Fautz, C.; Yu, Y. Real-time quantitative PCR for retrovirus-like particle quantification in CHO cell culture. Biologicals 2000, 28 (3), 137–148. 10. Zhang, M.; Miesegaes, G. R.; Lee, M.; Coleman, D.; Yang, B.; Trexler-Schmidt, M.; Norling, L.; Lester, P.; Brorson, K. A.; Chen, Q. Quality by design approach for viral clearance by protein A chromatography. Biotechnol. Bioeng. 2014, 111 (1), 95–103.
194
11. Miesegaes, G.; Bailey, M.; Willkommen, H.; Chen, Q.; Roush, D.; Blumel, J.; Brorson, K. Proceedings of the 2009 Viral Clearance Symposium. Devel. Biol. (Basel) 2010, 133, 3–101 12. Roush, D. Viral clearance using traditional, wellunderstood unit operations (Session I): Anion exchange chromatography (AEX). PDA J. Pharm. Sci. Technol. 2014, 68 (1), 23–29 13. Lower, R.; Boller, K.; Hasenmaier, B.; Korbmacher, C.; Muller-Lantzsch, N.; Lower, J.; Kurth, R. Identification of human endogenous retroviruses with complex mRNA expression and particle formation. Proc. Nat. Acad. Sci. USA 1993, 90 (10), 4480 – 4484. 14. ICH Q5A: Quality of biotechnology products: viral safety evaluation of biotechnology products derived from cell lines of human or animal origin. www.ICH.org. 15. Qiu, Y.; Jones, N.; Busch, M.; Pan, P.; Keegan, J.; Zhou, W.; Plavsic, M.; Hayes, M.; McPherson, J. M.; Edmunds, T.; Zhang, K.; Mattaliano, R. J. Identification and quantitation of vesivirus 2117 particles in bioreactor fluids from infected Chinese hamster ovary cell cultures. Biotechnol. Bioeng. 2013, 110 (5), 1342–1353. 16. Plavsic, M.; Qiu, Y.; Jones, N.; Keegan, J.; Woodcock, D.; Morris, J.; Davis, C.; Palermo, A.; Pomponio, R.; Scaria, A. Caliciviridae and vesivirus 2117. BioProcess. J. 2011, 9 (2), 6 –12.
PDA Journal of Pharmaceutical Science and Technology
Downloaded from journal.pda.org on September 18, 2016
An Authorized User of the electronic PDA Journal of Pharmaceutical Science and Technology (the PDA Journal) is a PDA Member in good standing. Authorized Users are permitted to do the following: ·Search and view the content of the PDA Journal ·Download a single article for the individual use of an Authorized User ·Assemble and distribute links that point to the PDA Journal ·Print individual articles from the PDA Journal for the individual use of an Authorized User ·Make a reasonable number of photocopies of a printed article for the individual use of an Authorized User or for the use by or distribution to other Authorized Users Authorized Users are not permitted to do the following: ·Except as mentioned above, allow anyone other than an Authorized User to use or access the PDA Journal · Display or otherwise make any information from the PDA Journal available to anyone other than an Authorized User ·Post articles from the PDA Journal on Web sites, either available on the Internet or an Intranet, or in any form of online publications ·Transmit electronically, via e-mail or any other file transfer protocols, any portion of the PDA Journal ·Create a searchable archive of any portion of the PDA Journal ·Use robots or intelligent agents to access, search and/or systematically download any portion of the PDA Journal ·Sell, re-sell, rent, lease, license, sublicense, assign or otherwise transfer the use of the PDA Journal or its content ·Use or copy the PDA Journal for document delivery, fee-for-service use, or bulk reproduction or distribution of materials in any form, or any substantially similar commercial purpose ·Alter, modify, repackage or adapt any portion of the PDA Journal ·Make any edits or derivative works with respect to any portion of the PDA Journal including any text or graphics ·Delete or remove in any form or format, including on a printed article or photocopy, any copyright information or notice contained in the PDA Journal
