Background
The aim of Session 1.5 was to review the viral clearance properties of steps that have not traditionally been evaluated for their virus clearance capabilities. These include newer unit operations in harvest and downstream processing such as multi-modal and membrane chromatography, ceramic hydroxyapatite chromatography, as well as chemical agents such as sodium caprylate.
Inactivation of Retroviruses by Sodium Caprylate at Neutral pH (Olga Galperina, GSK)
Sodium caprylate solution at low pH is used in the plasma industry for viral inactivation (1). Its ability to inactivate lipid-enveloped viruses has been explored, and it has been suggested that the non-ionizable form of the caprylate acid disrupts the integrity of the lipid bilayer and membrane associated proteins of enveloped viruses (2). Available information suggests that there is an equilibrium of ionized and non-ionized forms in caprylate solutions. Only the non-ionized form is effective for viral inactivation. Lower pH favors the formation of non-ionized caprylate.
Sodium caprylate can be used for product elution from certain affinity columns (for example, Cibarchrome Blue Affinity Resins). Concentrations of caprylate in the range of 75 to 150 mM are typical for elution, which is significantly higher than the concentration of caprylate used in the plasma industry (10–40 mM). Elution is usually performed at neutral pH, 7.2–7.5.
According to published data (3), higher concentrations of caprylate can result in sufficient amounts of the non-ionized caprylate form even at pH values close to or above neutral (>6.5).
The purpose of this work was to investigate the possibility of using sodium caprylate for inactivation of enveloped viruses in the pH range close to neutral.
Sodium Caprylate Solubility
Sodium caprylate at concentrations >50 mM can precipitate from the solution at lower pH. The higher the concentration of caprylate, a higher pH is required to maintain acceptable solubility. To define ranges for the proposed study, turbidity of the sodium caprylate solutions at different concentrations and pH values were measured in the presence and absence of product. Turbidity >50 TBU was considered unacceptable. Table I summarizes the results of the solubility tests.
Solubility of Sodium Caprylate
A caprylate inactivation study was performed at BioReliance with the results summarized in Table II.
Inactivation of XMuLV by Higher Concentrations of Sodium Caprylate
Inactivation Conditions
Time: 1 h
Temperature: 15 ± 1 °C
Caprylate concentrations: 50–150 mM
Protein Concentration: 4–6 mg/mL
Buffer composition: Tris/NaCl
Main Conclusions
Sodium caprylate at concentrations above 100 mM can be used to inactivate retroviruses even at neutral pH.
125 mM and 150 mM solutions of sodium caprylate completely inactivate Xenotropic murine leukemia virus (XMuLV) at pH 6.5–7.0.
Achieving Robust Viral Clearance with Mixed-Mode Anion Exchange Chromatography (Mark Teeters, Janssen Pharmaceutical Research and Development)
Several years ago a new mixed-mode anion exchange resin was implemented in our downstream monoclonal antibody (mAb) platform, replacing an older anion exchange resin as one of two polishing chromatography steps. Clearance of XMuLV and minute virus of mouse (MVM) across Capto™Adhere mixed-mode anion exchange resin was found to be robust across a wide range of operating pH and conductivity when operating in a flowthrough mode and as the third chromatography step in the process.
In characterization of a design space for pH between 4.5 and 6.5 and conductivity between 10 and 30 mS/cm, buffer solutions (sodium acetate with sodium chloride, no mAb) were spiked with XMuLV or MVM and loaded on 10 mL Capto™Adhere columns at a residence time of 4 min. A load of 10 column volumes (CVs) was applied to represent a typical operation for this stage (e.g., 100 g/L column load with a feed concentration of 10 mg/mL). XMuLV and MVM were quantified in the load (10 CV), five load-effluent fractions (2 CV each), an additional-wash fraction (2 CV), and a column strip (3 CV) for each run. The additional wash applied the respective buffer solution without virus. Results for the total virus in each fraction are presented in Figure 1. For XMuLV, no residual virus was detected in any load-effluent or additional-wash fraction under any of the four conditions evaluated. For MVM, small amounts of residual virus were detected in at least one fraction of each condition evaluated. Table III summarizes the log clearance of each virus under the four conditions, with minimum LRVs of >5.7 and 5.3 observed for XMuLV and MVM, respectively. These results suggest robust binding of XMuLV and MVM to Capto™Adhere resin across the pH and conductivity conditions evaluated.
Total XMuLV and MVM quantified in different solution conditions (no mAb) across a Capto™Adhere mixed-mode anion exchange column. FT represents the flowthrough load effluent fractions. An asterisk denotes where no virus was detected in a fraction.
Virus Reduction Observed for XMuLV and MVM in Different Solution Conditions (no mAb) Across a Capto™Adhere Mixed-Mode Anion Exchange Column. LRV was calculated from the total virus loaded and the sum of all effluent load and additional wash fractions
Historical XMuLV and MVM clearance data for multiple mAb processes that applied Capto™Adhere resin in a flowthrough mode are presented in Figure 2. In all cases, the column was the third chromatography step in platform mAb processes, and operating pH and conductivity were within the range characterized above (pH 4.5–6.5, conductivity 10–30 mS/cm). As the third chromatography step, residual process-related impurities in the load were at or near the limit of detection, product-related impurities such as aggregate were typically low (less than 3% in the load and reduced to less than 1% in the eluate), and column loads ranged from 50 to 150 g/L. For XMuLV runs, no virus was detected in any of the eluate pools, and LRVs between >4.5 to >6.1 were observed. For MVM runs, small amounts of residual virus were detected in most eluate pools, and LRVs between 3.6 and 7.0 were observed. Ten of 11 MVM runs had LRVs of 4.5 or greater, and the lowest observed LRV (3.6) was not reproduced in subsequent runs for that process. The historical data presented were generated at multiple contract laboratories over a span of about 5 years.
Historical virus reduction data for several different mAbs that applied Capto™Adhere mixed-mode anion exchange chromatography in a flowthrough mode and as the third chromatography step in each process. All processes had an operating pH between 4.5 and 6.5 and conductivity between 10 and 30 mS/cm. An asterisk denotes where no virus was detected in the eluate pool.
Together, these data suggest that clearance of XMuLV and MVM across Capto™Adhere mixed-mode anion exchange steps are robust when operating in a flowthrough mode, as the third chromatography step in a platform process (low levels of residual impurities), and at an operating pH between 4.5 and 6.5 and conductivity between 10 and 30 mS/cm.
Evolution toward Membrane Chromatography and the Impact upon Virus Reduction (James Berrie, Lonza)
With some clear process and economic drivers indicating membrane chromatography as an attractive manufacturing-scale alternative to packed bed chromatography, consideration was given here to the impact upon virus reduction claims. An evaluation of data from an evolved platform purification process migrating from traditional packed bed chromatography to membrane-based cartridges was presented and is discussed here.
Data was presented on the virus reduction capacity of an anion exchange membrane chromatography technology (Figure 3) used widely at Lonza where the data shows similarity to that obtained for packed bed anion exchange resins operated in flowthrough mode. The use of hydrophobic interaction chromatography (HIC) membranes was also discussed and consideration given to the reuse of anion exchange and HIC membrane cartridges. Multicycling of both anion exchange and HIC membranes is routine for intra-batch operation, and data on the successful reuse of HIC membranes up to 200 cycles was presented and discussed, which raised the question of inter-batch membrane chromatography reuse. Opinion was wide ranging on this concept, but in general the consensus opinion was that data would have to be generated to demonstrate membrane reuse batch to batch.
Virus reduction capacity of flowthrough mode anion exchange membrane chromatography technology at Lonza.
Viral Clearance on Ceramic Hydroxyapatite Chromatography (Matthew Dickson, Ph.D., MedImmune)
Ceramic hydroxyapatite (CHT) chromatography is sometimes used as a polishing step in mAb processes for removing process-related (e.g., host cell proteins) and product-related impurities (e.g., high-molecular-weight protein aggregates). A review of MedImmune viral clearance data was performed for three mAbs that included CHT as a polishing step within their clinical manufacturing processes. Table IV provides a summary of the operating parameters used for the CHT chromatography step for the three mAbs. Two different types of CHT chromatography medium were used for the mAbs, and the processes were performed under load pH conditions ranging from 6.7 to 7.8. Other chromatography operating conditions—including linear flow rates, load challenge, load conductivity, load concentration, and equilibration and elution buffer compositions—were similar for the three mAbs.
Operating Parameter Summary for the Three CHT Chromatography Processes
Four model viruses were included in the viral clearance assessment for the CHT chromatography processes to represent a broad set of virus classes and physiochemical properties. This panel of viruses included XMuLV, pseudorabies virus (PRV), simian virus 40 (SV40), and porcine parvovirus (PPV). Virus titer was measured in samples of the CHT load, hold control, nonbound, and product fractions using infectivity quantitation methods appropriate for each model virus.
Figure 4 provides a summary of the viral clearance results for the three mAbs. CHT chromatography was found to provide consistently good clearance (>4 log10) for XMuLV for the three mAb processes. CHT provided moderate clearance for PRV (∼3 to 4 log10). PPV clearance was found to be mainly negligible (≤1 log10), with the exception of one replicate for mAb 2. Clearance was found to be most variable for SV40 with values ranging from 4 to <1 log10.
Viral clearance results for three mAbs on CHT chromatography.
The partitioning pattern for CHT showed that clearance was generally achieved through virus binding during loading and maintenance of binding during product elution. For example, the partitioning pattern for XMuLV on the CHT column (Figure 5a) demonstrated that XMuLV clearance can be attributed to virus retention during column loading as well as during elution, with the majority of virus likely remaining on the column post-elution. The remaining three viruses (Figures 5b, c, and d) showed less clearance with variable virus retention during loading and more substantial co-elution in the product pool.
XMuLV partitioning.
SV40 partitioning.
PRV partitioning.
PPV partitioning.
Of the three mAbs, mAb 2 showed broadest range of clearance with >3 log10 clearance for three of the four model viruses (Figure 4). mAb 2 consistently shows virus retention for all four viruses during loading with no detectable virus in the nonbound. This mAb also showed variable virus retention during product elution, with PPV showing limited retention (Figure 2d) and XMuLV, SV40, and PRV showing more substantial retention (Figures 5a, b, and c). mAb 3 showed the least broad range of clearance of the three mAbs with negligible SV40 clearance (Figure 4). It is not clear when examining the operating conditions for the three mAb processes (Table IV) why mAb 2 demonstrated better SV40 clearance and why mAb 3 showed negligible SV40 clearance. Two possible reasons might be that virus retention during elution is affected by higher linear flow rate or the different elution condition (slightly higher ionic strength and slightly lower pH). Another possibility is that variability of mAb physiochemical properties may play a role in variable SV40 retention during loading and elution (Figure 5b). Additional experiments would need to be conducted to elucidate the role of these parameters on SV40 retention during elution.
In conclusion, this review has shown that CHT can offer good viral clearance for some model viruses, with better clearance shown for XMuLV and PRV but negligible clearance shown for PPV. The partitioning pattern for CHT showed that clearance was generally achieved through virus binding during loading and maintenance of binding during product elution. SV40 showed the most variable clearance, and it is not yet clear what operating factors contribute to SV40 clearance variability although linear flow rate, elution buffer composition, and mAb properties may play a role.
Assessment of High-Temperature/Short-Time Treatment as Viral Barrier for Various Cell Culture Media (John Mattila, Regeneron)
Regeneron assessed high-temperature/short-time (HTST) treatment of media as a viral barrier to prevent contamination of Chinese hamster ovary (CHO) cell culture with MVM. The evaluation was conducted as a bracketing study using a fully scalable, commercial benchtop HTST module (Armfield FT74X). Two base media and two concentrated nutrient feeds were independently treated in duplicate to assess the robustness of HTST treatment. Media included chemically defined and non-chemically defined proprietary medium. All media were processed at 102 °C for 10 seconds following a 0.1% (v/v) MVM spike.
The study revealed complete inactivation of MVM for all cell culture media with clearance greater than 5.0 log10 reduction factor (LRF) for all tests (Figure 6). The results indicate HTST treatment at 102 °C for 10 seconds provides robust parvovirus inactivation in a variety of cell culture media, providing an effective viral barrier.
HTST treatment at 102° C for 10 seconds provides an effective viral barrier by reducing MVM infectivity by at least 5 log10 in four classes of cell culture media. * denotes virus levels reduced below detection limits of the infectivity assay.
Strategies for Effective and Cost-Efficient Viral Clearance Studies To Support Early-Phase Programs (Michael Clark, Ph.D., AbbVie)
Through years of evaluating downstream antibody purification processes, many biopharmaceutical companies have compiled large amounts of viral clearance and inactivation data. These data sets have allowed for the implementation of platform operating conditions that provide robust and predictable viral clearance claims. This process knowledge now presents opportunities to streamline viral clearance studies for early-phase programs, which can serve the need to more efficiently drive clinical programs while maintaining the top priority of patient safety.
Among these opportunities is the ability to focus only on dedicated and/or robust viral clearance steps, which include low-pH inactivation, virus-retentive filtration, and anion exchange chromatography. The combined capabilities of these three steps are often suitable to meet the viral safety targets that have generally been accepted within the industry: at least two orthogonal clearance steps, ≥6 logs of clearance for parvovirus and ≤1 retrovirus-like particle (RVLP) per 1 million doses (calculated based on particle quantity in cell culture harvest). Thus, steps that are less robust and not dedicated to viral clearance, such as antibody capture by protein A chromatography and fine purification steps including hydrophobic interaction chromatography and cation exchange chromatography, can be eliminated from viral clearance studies for early-phase programs.
In addition to removing non-dedicated process steps from viral clearance testing, the existence of platform data for dedicated steps may allow for the implementation of modular viral clearance claims (Points to Consider in the Manufacture and Testing of Monoclonal Antibody Products for Human Use, USDHHS, FDA, CBER, 1997). The modular approach involves applying a standard log reduction claim based on viral clearance data acquired from previous projects that have used the same process step under comparable operating conditions. Attempts to apply modular claims are best suited for robust process steps that have well-understood mechanisms of action. Low-pH inactivation and large virus filtration using small virus–retentive filters seem the most amenable to this strategy. Modular claims can be maximized by optimizing viral clearance study design. For example, Figure 7 shows that by increasing the amount of virus spiked into the low-pH inactivation starting material and by using large volume testing, significantly greater viral inactivation capability is detected. This strategy is predicted to be applicable to removal of retrovirus by small virus–retentive filtration as well, as breakthrough of retrovirus on a small virus retentive filter has never been reported (Proceedings of the 2009 Viral Clearance Symposium, IABS, 2010).
Low-pH inactivation platform data for nine representative mAbs.
A recent challenge in meeting retrovirus clearance targets has come from improvements in cell culture conditions that allow for high cell densities and resulting titers, but also higher RVLP content in cell culture harvests. RVLP content is used to calculate a theoretical value for endogenous retrovirus per dose of drug substance, in accordance with ICH Regulatory Guidance Q5A. This states that the “measured or estimated concentration of virus in cell culture harvest” be used in the calculation. In scenarios with high harvest RVLP content, even applying three robust viral clearance steps when calculating theoretical RVLPs per dose may not be sufficient to reach the target of ≤1 per one million doses. To address this, we have pursued a strategy to more precisely quantify the true RVLP content entering the downstream process for clinical batches. In this approach, we evaluated RVLP, as measured by transmission electron microscopy (TEM), for unprocessed bulk harvest as well as clarified harvest (clarified by centrifugation and depth filtration). As shown in Table V, this reveals that the true average RVLP content entering the downstream process (clarified harvest) is approximately 2.1 logs lower than would be predicted by measuring unprocessed bulk. The majority of the decrease observed across the harvest step (about 1.7 logs) can be attributed to a lower limit of detection for the TEM assay post–harvest clarification. This is because the resulting clarified harvest sample pellet is smaller than the pellet from the unprocessed bulk, and thus each TEM section represents a larger ratio of the total pellet for clarified harvest. Therefore, this approach not only is more representative of the true RVLP content entering the downstream process, it is also more sensitive for the detection of RVLPs. While there is no known example of this approach being used in a regulatory filing, the concept is scientifically justifiable and seems reasonable for assuring the clinical safety of specific manufacturing batches of drug substance. However, this method should not be considered a way to validate the harvest process for viral clearance or a substitute for testing unprocessed bulk harvest for RVLP content, as testing of bulk harvest remains an important factor for assuring viral safety.
TEM Results for Unprocessed Versus Clarified Harvest Samples
Summary
A number of processing steps were discussed that have not been extensively studied and are not well understood for their ability to clear or inactivate virus. Sodium caprylate has been used to inactivate virus in the plasma industry. It was reported in the session that high concentrations of sodium caprylate (above 100 mM) can be used to inactivate retroviruses even at neutral pH.
The data presented suggest that clearance of XMuLV and MVM on Capto™Adhere mixed-mode anion exchange steps are robust when operating in a flowthrough mode, when positioned as the third chromatography step in a platform process (low levels of residual impurities) at an operating pH between 4.5 and 6.5 and conductivity between 10 and 30 mS/cm. Although this data suggests that the mixed-mode anion exchange resin Capto™Adhere is a robust step for removal of virus, additional data on the influence of contaminants on a robust design space for viral clearance are needed.
Membrane chromatography continues to gain adoption for use in the manufacture of recombinant protein therapeutics. A comparison of anion exchange membrane chromatography to traditional packed bed anion exchanged resins operated in a flowthrough mode was presented. The results indicated that anion exchange membranes provided similar viral clearance properties to traditional anion exchange resins. Additional information on the effective reuse of membranes and the impact to viral clearance properties is needed.
Ceramic hydroxyapatite (CHT) chromatography also has the potential to provide additional viral clearance to the downstream process. The partitioning pattern for CHT indicated that viral clearance is generally achieved through virus binding to the resin and being retained during product elution. The weakest retention was found with PPV, while XMuLV and PRV all showed strong retention to the resin. SV40 showed the most variable clearance, and it is not yet clear what operating factors contribute to SV40 clearance variability although linear flow rate, elution buffer composition, and mAb properties may play a role. Additional studies are required with CHT to better understand the resins' viral clearance attributes.
HTST treatment of cell culture media has been widely adopted by the biopharmaceutical industry to reduce the risk of adventitious virus contamination. Results indicate HTST treatment at 102 °C for 10 seconds provides robust parvovirus inactivation in a variety of cell culture media and provides an effective viral barrier.
The last talk was presented by AbbVie and discussed the how to accurately assess the RVLP burden that enters the downstream process. The RVLP content entering the downstream process (clarified harvest) was found to be approximately 2.1 logs lower than would be predicted by measuring unprocessed bulk, with approximately 1.7 logs of clearance being observed across the harvest step. It was suggested that measuring the TEM value post–harvest clarification is more representative of the true RVLP content entering the downstream process and is more sensitive for the detection of RVLPs. There was no consensus on the appropriateness for the use of this for clinical trial applications.
- © PDA, Inc. 2015
Jump to section
- Article
- Background
- Inactivation of Retroviruses by Sodium Caprylate at Neutral pH (Olga Galperina, GSK)
- Sodium Caprylate Solubility
- Inactivation Conditions
- Main Conclusions
- Achieving Robust Viral Clearance with Mixed-Mode Anion Exchange Chromatography (Mark Teeters, Janssen Pharmaceutical Research and Development)
- Evolution toward Membrane Chromatography and the Impact upon Virus Reduction (James Berrie, Lonza)
- Viral Clearance on Ceramic Hydroxyapatite Chromatography (Matthew Dickson, Ph.D., MedImmune)
- Assessment of High-Temperature/Short-Time Treatment as Viral Barrier for Various Cell Culture Media (John Mattila, Regeneron)
- Strategies for Effective and Cost-Efficient Viral Clearance Studies To Support Early-Phase Programs (Michael Clark, Ph.D., AbbVie)
- Summary
- References
- Figures & Data
- References
- Info & Metrics