Abstract
Antibody-drug conjugates (ADCs) is a growing field requiring specific handling due to the highly potent drug or highly active pharmaceutical ingredient (HAPI). The generic ADC process is composed of chemical conjugation steps followed by purification (removal of small molecules) and buffer exchange steps through Tangential Flow Filtration (TFF). In this article, we investigate the use of the ProStream® and HyStream® TFF cassettes from Repligen in comparison to reference cassettes in 1) a side-by-side comparison of a standard process and 2) in a challenging process (high load and high concentration factor). We obtained comparable process performance results (processing time and yield) and product quality attributes (including but not limited to Drug-to-Antibody Ratio (DAR), High Molecular Weight (HMW) and Low Molecular Weight (LMW) levels, residual organic solvent and free drug content) as in the reference processes. Thus, this evaluation provides encouraging results to support ProStream and HyStream cassette use in ADC processes.
Introduction
Antibody-drug conjugates (ADCs) is a growing therapeutic modality achieving maturity as demonstrated by the FDA approval of 12 ADCs since 2000 with 5 in the last 3 years. [1] ADC is the translation of Paul Ehrlich’s magic bullet concept [2]: to a targeting antibody is attached a highly potent small molecule drug through a stable linker. The antibody delivers selectively the small molecule drug to a specific cell type, triggering its death in the case of an oncology drug, which is the main current field. [3]. By design, ADCs are complex entities. Indeed, only 0.01% of the injected dose reaches intracellular target, [4] which requires the attached small-molecule drug to be very potent to trigger its mechanism of action. This highly potent drug or highly active pharmaceutical ingredient (HAPI) requires specific handling. [5] The generic ADC process is composed of at least four steps: chemical reactions in aqueous buffer (including some organic solvent), purification, buffer exchange through Tangential Flow Filtration (TFF) and formulation. [6] The purification (removal of small molecules) can be also performed during the TFF. [7]
In this article, we investigate the use of the ProStream® and HyStream® TFF cassettes from Repligen as alternatives to reference cassettes in 1) a side-by-side comparison of a standard process and 2) in a challenging process (high load and high concentration factor). TFF performance evaluation was based on permeate flux values during the concentration and diafiltration steps. For the impact on the ADC process and quality attributes, we assessed the yield, protein concentration, Drug-to-Antibody Ratio (DAR), High Molecular Weight (HMW) and Low Molecular Weight (LMW) levels, as well as residual solvent and remaining free drug content after the TFF step. For the challenging TFF process, the comparison of the quality attributes was performed after the TFF step directly, and on the formulated drug substance with complementary analysis of the mass variant purity, the charge variant purity, the drug distribution profile and the free drug content.
Materials and protocol
Materials
The proprietary ADC (IgG1 antibody structure) used in the studies is obtained from 2 different productions (representative batches), and the material attributes are reported in Table 1.
The first test (side-by-side comparison with a standard process) was performed with a 30 kDa ProStream TangenX “PRO” flat sheet cassette (modified PES, 0.01 m2, H-screen, reusable, product code: PP030WP1H) by Repligen Corporation (Waltham, US). The selection criteria were based on comparable membrane chemistry (modified polyethersulfone or mPES), molecular weight cut-off (30 kDa), screen type (tight screen) and a similar surface area (~0.01 m2).
The second test (high load and high concentration factor, i.e., challenging process) was performed with a 30 kDa HyStream TangenX “SIUS” flat sheet cassette (modified PES, 2x 0.01 m2, E-screen, single-use, product code: XP030MP1E) by Repligen Corporation (Waltham, US). The selection criteria were based on the 30 kDa cassettes (modified cellulose, ~0.02 m2, medium screen) used in the reference process.
TFF Protocol
For the first test (side-by-side comparison), the cassette was flushed with purified water to remove the storage solution, and the reference normalized water permeability (NWP) was determined before use according to the supplier’s recommendations.
The same feed flow rate (5 L/min/m2) and comparable loading ratio (183 g/m2) was used for the ProStream cassette as for the reference one to support the side-by-side comparison.
Optimal transmembrane pressure (TMP) was determined according to flux-TMP testing and applied for the subsequent experiments. The ADC solution (0.80 L) was concentrated from 2.3 to 22 g/L, followed by diafiltration with 10 diavolumes (DVs) of 10 mM sodium acetate, pH 5.5. The product was recovered after TFF by flushing the system with 20 mL (equivalent of 0.3 DVs) of the diafiltration buffer to obtain a final ADC concentration of 19.2 g/L.
The cassette was cleaned after use by recirculating 0.1M NaOH for 30 minutes, followed by the testing of NWP for comparison with the reference value.
The second test (challenging process) was performed with the same filter preparation procedure as in the first test, but with the omission of NWP testing because of the single-use nature of the membrane.
The reference cassettes were operated at 300 g/m2 loading ratio and a constant 5 L/min/m2 feed flow rate. The method was performed in three parts: intermediate concentration up to 35 g/L, followed by diafiltration for 10 DVs, followed by the final concentration up to 120 g/L. Upon meeting the final concentration target, the product was recovered by flushing the system with diafiltration buffer, whose volume had to be limited to 0.125 DVs to maintain the product concentration above 100 g/L.
The HyStream cassette evaluation was performed by increasing the final concentration target to 200 g/L so that higher flushing volume (0.5 DVs) could be applied at the end to maximize product recovery. The loading ratio was also increased to 500 g/m2.
The optimal feed flow rate of 18 L/min/m2, and the optimal feed-channel pressure drop (ΔP) of 0.7 bar was determined experimentally by generating permeate flux–TMP curves at various feed flow rate values (6, 9, 12, 15, and 18 L/min/m2). The ADC solution (1.56 L) was concentrated from 6.4 to 35 g/L, followed by diafiltration with 10 DVs of 10 mM sodium acetate, pH 5.5, followed by an additional concentration step to reach 200 g/L. The product was recovered after TFF by flushing the system with 25 mL (0.5 DVs) of the diafiltration buffer to obtain a final ADC concentration of 127.4 g/L.
Analytical assays
DAR was measured by ultraviolet spectroscopy (UV – SoloVPE, Repligen); monomeric purity, HMW and LMW levels were determined by Size Exclusion Chromatography (SEC-UPLC – Acquity, Waters – Column: Acquity UPLC Protein BEH SEC 200A 1.7μm 4.6x300mm), organic residual solvent and free drug content were determined by 2D-High Performance Liquid Chromatography (2D-HPLC), mass variants were determined by Non-Reducing Capillary Gel Electrophoresis (NR cGE – Maurice, Bio-Techne), charge variants were detected by image capillary Isoelectric Focusing (icIEF – Maurice, Bio-Techne).
Results
First test (side-by-side comparison)
In the studied range of TMP (0.4-1.1 bar), the optimal value was determined between 0.8 and 1 bar. The resulting ∆P was 0.7-0.8 bar.
During the actual run the permeate flux and the processing time obtained with the ProStream cassette were comparable to the reference process during both concentration and diafiltration steps (Table 2).
The slightly lower yield could be attributed to the lower flushing volume (0.3 DVs instead of 0.8 DVs), but minor differences in non-specific binding to the membranes may have had an influence as well.
The ProStream membrane showed equivalent hydraulic performance in terms of permeate flux (Figure 2A) compared with the reference cassette. The ProStream membrane showed also a lower ΔP than the reference cassette.
This could be attributed to the differences in cassette design (Figure 2B).
Because of the higher permeate flux, the reduction of conductivity in the permeate line during diafiltration was also slightly quicker with the ProStream cassette, but the same level of buffer exchange efficiency and clearance profile could be obtained with both cassette types (Figure 3).
After the TFF step, all analyzed product quality attributes such as protein concentration, DAR, HMW and LMW levels, residual organic solvent and free drug content were within the targeted acceptance criteria: the final protein concentration was above 10 g/L, the DAR, HMW and LMW ratios were comparable to the reference process, while the organic solvent and free drug content were under the quantitation limit of the analytical methods (Table 3).
Utilizing the same cleaning procedure for the ProStream cassette as for the reference one, the NWP value after its first use remained above 90%, which was a significant improvement over the currently used cassettes, whose NWP values dropped to 60-70% after their first use.
Despite of the promising first results, the cleanability and reuse potential of the ProStream cassette will need to be evaluated further by performing additional runs and cleaning cycles.
Second test (challenging process with high protein load and high concentration factor)
As the 6-12 L/min/m2 feed flow rates (supplier’s recommendation) did not provide the desired permeate flux, the investigation was extended with 15 and 18 L/min/m2 feed flow rates. The optimal conditions appeared to be 18 L/min/m2 feed flow rate, and TMP between 0.7 and 0.9 bar, which provided a permeate flux of 77- 80 L/hr/m2 and a ΔP of 0.7 bar (Figure 4A).
As expected, the permeate flux dropped to around 30 L/hr/m2 with the concentrated material at 35 g/L (Figure 4B) with the same range of TMP.
Both permeate flux values were deemed reasonable and sufficiently high to proceed with the actual test.
No impact on product quality was detected based on the preliminary assessment of the concentrated samples in terms of monomeric purity (determined by SEC-UPLC).The obtained permeate flux values and processing times fulfilled the expectations: the permeate flux was consistently higher than in the previous iteration of the process, which could shorten the total processing time by 30%, despite of the 1.5x higher loading ratio (Table 3).
The permeate flux decreased by 76% by the end of the intermediate concentration step but could be recovered to 75% of the starting value during diafiltration with the HyStream membrane (Figure 5A). The ΔP regulation at 0.7 bar provided a more balanced performance compared to the traditional constant feed flow rate operation. The latter resulted in a step increase of ΔP from 0.2±0.1 bar to 1.5 bar during the final concentration step in the reference process (Figure 5B).
As in the first test, the higher permeate flux provided a quicker transition from intermediate concentration to diafiltration, but the same level of buffer exchange efficiency and clearance profile could be obtained with both cassette types (Figure 6).
Based on the analyzed quality attributes of the concentrated materials after 0.5/0.2 μm filtration both cassette types were able to provide the same high-quality outcome: the protein concentration, DAR, monomer purity, HMW and LMW levels were comparable, the residual organic solvent content was below the detection limit. There was only a negligible clearance of fragment levels detected by NR cGE with HyStream, but a minor improvement was detected in the free drug ratio compared to the reference process, which may have been caused by method variance (Table 4).
Since the concentrated samples post-TFF had high turbidity, which was known to be interfering with the UV method, the protein concentration was measured after the 0.5/0.2 μm filtration only. The yield post-TFF and before the 0.5/0.2 μm filtration could only be estimated by calculation, which indicated comparable performance by both cassette types.
The final drug substance (FDS) produced by both cassette types and methods was analyzed for the key quality attributes (Table 5), and the results of both samples were comparable in terms of appearance, protein content, DAR, monomeric purity, HMW and LMW content at the precision of the analytical method.
Discussion
The standard process and parameters either with ProStream (modified PES, H-screen or “fine” screen) or reference cassette (comparable characteristic) showed similar performance and product quality. It could be noted that beside equivalent hydraulic performance, the ProStream membrane showed also a lower pressure drop (ΔP) leading to a faster buffer exchange step (monitored through the conductivity trend).
The second test was performed to assess a more challenging process: higher concentration factor and higher loading, to gain in process productivity. Additionally, improved pressure control was implemented for the final concentration by ΔP regulation combined with TMP-control.
For this second test, the ProStream was replaced by the HyStream membrane, which is also made of modified polyethersulfone (mPES) but has a hydrophilicity comparable to regenerated cellulose (supplier information). The E-screen (or “coarse” screen) was selected to mitigate the expected pressure drop with the high concentration to be assessed.
The HyStream cassette had to be operated with very high feed flow rate (18 L/min/m2) to achieve the desired permeate flux. This setting is difficult to scale-up due to the high pump capacity requirement at manufacturing scale. As a potential follow up it would be beneficial to investigate whether the same membrane material equipped with the tighter L-screen (or “medium” screen) would provide comparable permeate flux at a lower feed flow rate, and how this change would influence the pressure profiles.
Based on a survey of major pharmaceutical companies, reusing TFF membranes is not common for ADCs compared to other modalities such as mAbs (monoclonal Antibodies) (7). Repligen offers traditional reusable cassettes (“PRO”) and purpose-built single-use cassettes (“SIUS”) with the same ProStream and HyStream membrane materials to facilitate the transitions in both directions. Calculations have shown that the adoption of single-use cassettes can provide significant cost reduction compared to the traditional reusable cassettes due to the reduction in filter cost, and the savings in buffer consumption, utility cost and manpower (due to not having to perform cleaning after and/or before use). The only caveat is the absence of H-screen (or “fine” screen) in single-use format, therefore the transition in the standard process would require additional validation work with cassettes equipped with L-screen (or “medium” screen), which is available in both reusable and single-use formats.
Repligen offers the single-use cassettes (“SIUS”) incorporated into closed and sterile flow-path assemblies called “SIUS Gamma”, as well as built into the recently launched, closed and gamma irradiated self-containing format called “SIUS SC”. In the “SIUS Gamma” devices the cassettes are adhered onto polypropylene filter plate inserts (FPI) to avoid product contact with the stainless-steel holder, but the latter is still necessary to keep the cassettes under compression. The “SIUS SC” devices on the other hand are ready-to-use on their own, and the omission of holder and torquing not only simplifies and shortens the preparation procedure, but mitigates the risk of product loss caused by insufficient compression, which is one of the most common causes of product loss with flat sheet cassettes. Both types of devices are equipped with genderless AseptiQuik® G connectors (CPC, Arden Hills, MN) on the inlets and outlets. The “SIUS Gamma” devices are available with both L-screen and E-screen, while the “SIUS SC” devices are available with L-screen only at this time. The introduction of E-screen into the “SIUS SC” format is expected later in 2024. The use of self- contained equipment is very important for operator safety regarding the high potency of the product and free drug impurities (5).
Conclusions
The ProStream and HyStream cassettes of Repligen have been tested side-by-side with reference cassettes in a standard ADC process, and in a challenging ADC process with high load and high concentration factor. The obtained results demonstrated comparable process performance (processing time and yield) and product quality attributes (including but not limited to DAR, HMW and LMW levels, residual organic solvent and free drug content) as in the reference processes.
Potential next steps would be to pursue benchmark work on additional programs and assess the self-containing “SIUS SC” device at pilot scale.
References
[1] Dumontet C, Reichert JM, Senter PD, Lambert JM; Beck A. Antibody–drug conjugates come of age in oncology. Nat Rev Drug Discov 2023 (22): 641–61. Available from https://www.nature.com/articles/s41573-023-00709-2
[2] Riccardi F, Dal Bo M, Macor P, Toffoli G. A comprehensive overview on antibody-drug conjugates: from the conceptualization to cancer therapy. Front Pharmacol 2023 (14):1274088. Available from https://doi.org/10.3389/fphar.2023.1274088
[3] Wang Z, Li H, Gou L, Li W, Wang Y. Antibody-drug conjugates: Recent advances in payloads. Acta Pharmaceutica Sinica B 2023 13(10): 4025-59. Available from https://doi.org/10.1016/j.apsb.2023.06.015
[4] Teicher BA, Chari RVJ. Antibody Conjugate Therapeutics: Challenges and Potential. Clin Canc Res 2011 17(20): 6389-97. Available from https://doi.org/10.1158/1078-0432.CCR-11-1417
[5] Denk R. The Challenge of cGMP in the Manufacturing of Antibody-drug Conjugates. ADC review 2016 April. Available from https://www.adcreview.com/articles/the-challenge-of-cgmp-in-the- manufacturing-of-antibody-drug-conjugates/
[6] Bulger PG, Conlon DA, Cink RD, Fernandez-Cerezo L, Zhang Q, Thirumalairajan S, Raglione T, Liang R, Zhou J, Chalger i A. Drug-Linkers in Antibody–Drug Conjugates: Perspective on Current Industry Practices. Org Process Res Dev 2023 27(7): 1248–57. Available from https://pubs.acs.org/doi/10.1021/acs.oprd.3c00136
[7] Fernandez-CerezoL, Holstein M, Thirumalairajan S, Romanini D, Kos P, Raglione T, Bulger PG. Strategies for UF/DF-Based Impurity Removal in the Post-conjugation Purification of Antibody–Drug Conjugates. Org Process Res Dev 2023 27(7): 1258–1268. Available from https://pubs.acs.org/doi/full/10.1021/acs.oprd.3c00137
Authors: Benjamin Ha*1, Zoltán Gulyás*2
1 Sanofi, ADC Conjugation Development; 13 quai Jules Guesde, 94400 Vitry-Sur-Seine, France; 2 Repligen, Sr. Field Applications Scientist, DSP | EMEA West/South,
Corresponding Author: Benjamin Ha and Zoltán Gulyás.
Key terms: ADC, antibody-drug conjugate, Tangential Flow Filtration, TFF, free drug content, permeate flux values, protein concentration, Drug-to-Antibody Ratio, DAR, High Molecular Weight, HMW, Low Molecular Weight, LMW.
Published In: ADC Review| Journal of Antibody-drug Conjugates
DOI: https://doi.org/10.14229/jadc.2024.11.08.001
How to cite:
Benjamin Ha and Zoltan Gulyas
Methodology for assessing Repligen ProStream® and HyStream® Tangential Flow Filtration Cassettes in an Antibody-Drug-Conjugate Process – J. ADC. November 8, 2024. DOI: 10.14229/jadc.2024.11.08.001.
Last Editorial Review: October 28, 2024
Article History:
- Original Manuscript Received October 3, 2024
- Review results received October 25, 2024
- Manuscript accepted for publication October 31, 2023
- Article published online: November 8, 2024
Featured Image: Researcher in Lab. Courtesy: © Fotolia/Adobe 2010 – 2024 . Used with permission.
Acknowledgement
The authors thank their colleagues from Sanofi* and Repligen$ for their technical support/discussions/suggestions on this work: Benjamin Fabre*, Kevin Brower*, and Eddy de Bock.
Author Contribution
BH and ZG designed the experiments, reviewed the data and redacted the manuscript. BH performed the experiments. All authors have seen and approved the final version of the manuscript being Submitted.
Conflict of Interest
BH declares no financial nor personal conflict of interest. ZG is a full-time employee of Repligen as a Field Applications Scientist in Europe, and the ProStream and HyStream cassettes selected for testing were provided by Repligen free-of-charge.
Funding Declaration
Experiments were funded by Sanofi. The cassettes used in the studies were provided by Repligen.
Figure Legends
Figure 1: TMP optimization for the first test with ProStream
Figure 2A: comparison of permeate flux (ProStream vs. reference cassette) Figure 2 B: comparison of ΔP (ProStream vs. reference cassette)
Figure 3: Comparison of permeate conductivity (ProStream vs. reference cassette)
Figures 4A: TMP-flux graphs at 6.4 g/L
Figures 4B: TMP-flux graphs at 35 g/L
Figure 5A: Comparison of permeate flux (HyStream vs. reference cassette) Figure 5 B: Comparison of ΔP (HyStream vs. reference cassette)
Figure 6: Comparison of permeate conductivity (HyStream vs. reference cassette)
DOI: 10.14229/jadc.2024.11.08.001.
