CEX Background
At the previous viral clearance symposium, insights into the general mechanism of viral clearance by cation exchange chromatography (CEX) were shared by Amgen and Novartis (1). Amgen presented data from multiple monoclonal antibody (mAb) products that consistently revealed ≥4 log10 murine leukemia virus (MuLV) clearance when run at a pH of 5.0. Clearance was substantially reduced when CEX was performed with pH 5.5 or 6.0 and was abolished by pH 6.5. MMV clearance was consistently poor under all pH conditions tested. During the discussion session it was agreed that this was quite surprising, as the pI of MuLV and MMV is 5.8 and 6.2, respectively; from a mechanistic standpoint and assuming that virus–mAb interactions do not occur, if CEX was based exclusively on ionic interaction, then robust (i.e., not reduced) log reduction values (LRVs) would be expected when runs are performed at pH values close to the virus pI.
The Novartis presentation in the 2009 meeting displayed various MuLV clearance data from CEX runs performed at pH 5.0 and above for a number of mAb products. It was noted that the pH 5.0 runs tended to have the highest LRVs, implying that in addition to column partitioning there existed a pH-dependent inactivation component. However, data by Amgen showed otherwise, in that substantial clearance of (1) Pseudorabies virus (PRV, an enveloped virus) was achieved in runs performed at pH 6.0, a level where inactivation is highly unlikely; and (2) Reovirus type 3 (REO-3 a non-enveloped virus) from runs performed at pH 5.0. By the end of the CEX session discussion at the 2009 Symposium, it became evident that additional efforts would be required to better understand the mechanism of viral clearance for CEX.
MuLV Removal by CEX Chromatography (Lisa Connell-Crowley, Amgen)
Additional evidence from an in-house database supported the notion that CEX could effectively and reproducibly clear MuLV (Table I). Building on the original data presented two years prior, Amgen had continued its efforts focusing on the strong CEX exchanger Fractogel SO3− and had determined that the dominant mechanism for MuLV clearance was retention of the virus via adsorption instead of inactivation. Though the key findings were presented at the symposium, a more extensive description was eventually published as a peer-reviewed journal article (2).
Presented was a design space for effective MuLV removal by Fractogel SO3− with respect to operational pH, elution ionic strength, loading, and load/equilibration buffer ionic strength (Figure 1 and Tables II & III). MuLV is able to bind to other CEX resins, such as Fractogel COO− and SP Sepharose Fast Flow (Figure 2), suggesting that this phenomenon is not restricted to one type of CEX resin. Taken together, the data indicate that while ion exchange critical process parameters (pH, conductivity) are likely influential in determining the extent of MuLV clearance, other parameters such as loading or buffer species/strength may not be as critical to ensuring a robust and reproducible removal of MuLV.
An Amgen Study Of CEX Operation At Ph 5.0 (A) And Ph 6.0 (B) Demonstrating Xmulv LRV Of >4.4 And 1.4 Log10, Respectively. Open symbols in (a) and (b) indicate no virus detected (by TCID50). (c–e) Defining sodium concentration (c) and pH (d) and cation concentration (e) for effective xMuLV clearance using Fractogel SO3- in bind and elute mode. Note from (c) and (e) that binding strength increases as pH decreases, and from (e) that xMuLV elutes earlier as EQ/Load cation concentration is increased. A “*” in (e) indicated no virus detected in the respective fraction.
In An Amgen Study, xMuLV Binds To Other CEX Resins (Figure 4 In CC Et Al., 2011). The extent of binding appears to be ligand-dependent. Fractogel SO3- exhibits strongest binding of xMuLV.
Evaluation of CEX as a Capture Step for mAb Purification (G. Miesegaes, CDER/FDA)
Similarly, FDA presented design space results for effective binding, purity, and parvovirus (PPV) partitioning from a model mAb provided by collaborators at Eli Lilly and Company, based on a project that sought to determine the feasibility of utilizing CEX as a mAb capture step. At the previous symposium, the FDA presented results from its viral clearance database (1, 3) that suggested CEX as a capture step did not perform as well as when it is used for polishing (1.8 versus 2.7 log10; P = 0.01). This led to a general inquiry as to whether CEX when run within a defined operating space could adequately serve as an alternative to protein A for product capture from harvested bulk material. To address this, CEX was investigated through a design of experiment (DoE)-based series of experiments; although partially presented at the 2011 Symposium, this work was later published in its entirety as a standalone peer-reviewed journal article (4).
In this project, five commercially available CEX resins were systematically evaluated under capture conditions commonly found in the database and that resulted in acquiring sufficient (i.e., manufacturing relevant) product yield. From these condition-screening studies a design space was then experimentally defined (Table IV). This working design space was next assessed for viral clearance of PPV (infectivity assay) first at a defined center point condition (factor “0” in Table IVa) and later across multiple conditions by use of a resolution III DoE (Table IVc). Assessments at center point resulted in two of the original five resins being dropped from the study. With regard to the DoE study, it was found that for all remaining resins tested, the conditions that utilized the lowest pH values, independent of conductivity, did not achieve favorable PPV clearance (<1.0 log10). Interestingly, this loss of viral clearance also trended with lower levels of product purity (4). Additional key findings included that, while attributes such as feedstock quality may affect capture CEX performance, a range of conditions was eventually identified that achieved LRVs comparable to that for protein A. Afterwards, reassessment of the regulatory database found that the distribution of LRVs for CEX and protein A capture steps were comparable. That is, the majority of records for both operations fell within claims of 2–3 log10 for retrovirus and 1–2 log10 for PPV clearance, respectively (Figure 3). This suggests that under appropriately defined conditions, CEX may be a suitable alternative to protein A based on product capture desirables such as yield and purity, as well as viral clearance.
Conditions Identified in a Collaborative CDER/Eli Lilly Study To Provide an Effective Range of Product Capture Using CEX. In early-on screening studies, buffer pHs below 5.0 or above 5.7, or conductivities above 6.0, did not favor adequate levels of product yield and thus served as outer boundaries for a subsequent resolution III DoE-based study. Viral clearance of PPV was then determined by TCID50 using an arbitrarily-defined centerpoint based on these boundaries (see Values column). Next, a DoE approach was employed to investigate these product yield–favorable parameter ranges to clear PPV (quantitative polymerase chain reaction). As part of optimizing run fractions, two elution pooling criteria were tested, one which consisted of fewer fractions but was slightly more pure (“narrow pool”) and one containing more fractions so as to provide greater yield (“wide pool”) at the cost of purity. This work in its entirety was eventually published (4).
Average CEX Capture Lrvs Versus Protein A From A Database Of CDER Regulatory Records. There exists substantial clearance across each operation, from the standpoint that (i) for retrovirus: 37% database records for both operations range from 2–3 log10 clearance, and (ii) for parvovirus: 52% database records and 42% for CEX and protein A respectively, or the majority of records for each operation, range from 1–2 log10 clearance.
Detergent Inactivation Background
It was generally agreed upon during the 2009 Symposium that when compared to plasma products, solvent/detergent (S/D) methods applied to mAbs are just as robust in terms of inactivating enveloped viruses such as MuLV. Concerns arose over the need to eliminate cytotoxicity effects due to properties of the solvent. Although data was presented on detergent-only methods, it was decided that more work would be needed in order to directly address robustness of inactivation in the presence/absence of solvent. Also left unresolved were issues pertaining to cell culture and buffer composition, and lipid content, with respect to influencing the robustness and reliability of S/D methods overall. As presented below, some of these remaining issues were addressed during the 2011 event.
Virus Inactivation by Detergents (L. Norling; Genentech, a Member of the Roche Group)
Shown in Figure 28, experiments were performed to evaluate the effect of temperature and Triton X-100 concentration on MuLV inactivation using mAb 1- and mAb 2-harvested cell culture fluid (HCCF). For both mAbs, inactivation is slower at 12 °C than 20 °C, demonstrating that lower temperature and shorter time are worst-case conditions. No difference in MuLV inactivation was observed between 0.3% and 0.5% Triton X-100.
The effect of protein concentration and lipid content on virus inactivation by Triton X-100 was also evaluated. mAb 2 had a high percent packed cell volume (% PCV) and low viability. High % PCV and low cell viability likely leads to higher cellular components such as total proteins and lipids in the HCCF, which could be potential worst-case conditions for viral inactivation by detergent. The lipid content and protein concentration in mAb 2 (low viability and high % PCV) and mAb 3 HCCF are shown in Table V. MAb 3 concentrated 10× by tangential-flow filtration had higher total protein and lipid than mAb2.
Ranges of Protein and Lipid Content in Genentech mAb 2 and mAb 3 HCCFs. Total protein was determined by Bradford assay. Total lipids was determined using Phospho-vanillin assay.
MuLV inactivation in 0.3% and 0.5% Triton X-100 was tested in 1× and 10× mAb 3 HCCF at 20 °C. Complete virus inactivation was obtained in 1× and 10× HCCF, indicating that levels of protein and lipid present in Genentech's HCCF does not affect retrovirus inactivation by Triton X-100 (Figure 4)> for this mAb.
A Genentech Study Of The Effect Of Temperature (A), Time (B), HCCF Load (C), And Triton X 100 Concentration On X-Mulv Inactivation. For (c), the 1×, 0.3% Triton X-100 inactivation data is from regular titration. The 10×, 0.3% and 0.5% Triton X-100 inactivation was done using large-volume testing.
Eight mAbs have been tested for MuLV inactivation by 0.3% Triton X-100 at 20 °C for 60 min. These mAbs represent a wide range of percent cell viability and % PCV during cell culture, as well as impurity levels for CHO cell proteins (CHOP), DNA, and aggregates. Effective virus inactivation was obtained for all eight mAbs (Table VI). mAbs tested represent a wide range of levels for HCP, DNA, aggregates, percent cell viability, and % PCV. Effective inactivation across mAbs and understanding of parameter effects support a modular claim for clinical products of 4.6 LRV.
Effective MuLV inactivation by Triton Detergent across Eight Genentech mAbs (IgG1 and IgG4 Isotypes). Manufacturing conditions for Triton X-100 hold in HCCF are as follows: 0.5 ± 0.2% Triton X-100, ≥20 °C, ≥60 min (target: 0.3% Triton, 20 °C, 60 min).
Next, a number of commonly available and proprietary detergents were assessed to obtain a cumulative feel for viral clearance capability across a variety of detergent types (Table VII). Considerations to be made involved the need to meet global disposal requirements, and to have minimal impact on product quality, all while still allowing for adequate viral inactivation. The more novel or less known of the detergents to be tested, as they are not currently being used in a pharmaceutical manufacturing setting, included non-toxic and/or biodegradable seed ester alcohols from the following renewable sources: alcohol ethoxylate (biodegradable), alkyl polyethylene, and alkyl glucoside.
Summary of Genentech Virus Inactivation Data Evaluating Various Detergents. Conditions include 1% Detergent (left column), incubated at 20 °C for 1 h.
In summary, it was determined that Triton X-100 was a robust and effective detergent for MuLV inactivation when used at 0.3% at 20 °C and for 60 min. No impact was observed with protein concentration, lipid content, or impurity levels. This study was based off a total of eight different mAb products and collectively led to a modular claim of 4.6 log10.
- © PDA, Inc. 2014
