Points to Consider for Continuous Downstream Bioprocessing

As therapeutic monoclonal antibodies (mAbs) move through the drug development pipeline—from R&D to clinical trials to market approval by regulatory authorities—chemistry, manufacturing, and control (CMC) plays a critical role. Nearly all commercial mAb products are currently manufactured in batch mode with high cost, long manufacturing time, low process efficiency, and a large footprint of equipment and facility. Biopharmaceutical companies have therefore been seeking alternative strategies and/or novel technologies to address these issues.

One strategy for protein drug manufacturing is the adoption of continuous bioprocessing (CBP), which has been successfully used for decades to produce amino acids, vitamins, and antibiotics. CBP has demonstrated multiple advantages, including high efficiency, high productivity, and high consistency. CBP also employs real-time monitoring, control, and automatic operating with a small footprint for hardware (e.g., column, skid, resin, tank) and facility.

• Implementing CBP has many challenges based on the following:
• Protein complexity (e.g., isoelectric point [pI], size, hydrophobicity, post-translational modification)
• Protein stability (e.g.,sensitive to extreme pH, conductivity, heat, light, protease)
• Biosafety (e.g., virus from cell line, mycoplasma, animal-derived components, bacteria, endotoxin)
• Product purity and quality (e.g., monomer percent, aggregates, fragments, charge profile, residual host cell proteins [rHCPs], and residual DNA [rDNA])
• CBP-dedicated technologies (e.g., inline analytical equipment and method, process-related hardware, software, and facility)
• Regulatory guidance
• Business risks (e.g., investment and return).

In this study, the authors design and propose a continuous mAb manufacturing and purification process —from harvest to final formulation—based on the understanding of these challenges and including the specific upstream process (e.g., high titer and/or large harvest volume). The purified drug substance (DS) from the CBP is expected to meet pre-determined safety, quality, identify, purity, and efficacy (SQIPE) requirements.

Progress in upstream process productivity

The upstream process is conducted by fed-batch or perfusion cell culture for mAb production. Significant progress has been made in both cell line development (e.g., Chinese hamster ovary, PER.C6, or murine myeloma [NS0] cell line selection, clonality, clone stability, and cell banking) and upstream process development (e.g., media selection, feed strategy, viability, viable cell density, titer, product quality, stability, and potency). High titer (e.g., > 5 g/L) from fed-batch cell culture or large harvest volume from perfusion cell culture (e.g., two harvest media volume/bioreactor volume/day, 60 days per batch) has enhanced upstream productivity to satisfy market demand. Either the higher titer or larger harvest volume presents challenges for the downstream process development and manufacturing (1).

Points to consider for harvest

Prior to downstream purification, the mAb is usually harvested by depth filtration (DF), particularly when a smaller fed-batch bioreactor (e.g., ≤ 2000 L) is used to produce product for early clinical trials. Depth filters with different sizes and properties remove viable cells and dead cell debris by size-excusive sieving retention and/or charge adsorption (2). The DF filters are made from cellulose or polymer with filter aids, such as diatomaceous earth, charged resins, glass, silica, or activated carbon. High-grade DF filters with low extractable and/or low endotoxin are available as well. The DF filtrate quality (i.e., rHCP, rDNA, aggregates, fragments) and recovery are closely related to upstream cell line, titer, viable cell density (e.g., > 10x10⁷ cell/mL), cell viability (e.g., > 70%), filter property (e.g., size, or positively charged filter), and feed flux (e.g., < 600 LMH), pressure (e.g., < 30 psi), load ratio (e.g., primary: < 100 L/m², and secondary filter: < 200 L/m²).

The current industrial practice for DF is a manual operation mode. To design a continuous DF process, the authors used a multi-round DF system (3M) as an example. This system provides sufficient filter area for the process need. The system cannot be assembled and performed automatically due to manual assembly with tubing. To upgrade the system for continuous harvest, a filter holder can be designed as illustrated in Figure 1. Each round of filter can be performed sequentially. After finishing the DF from each round, the holder can be disassembled and re-assembled robotically with a new set of filters. It is expected that this type of hardware could be designed and applied for future continuous DF (e.g., automation of operation and filter replacement).

For late-stage and commercial-production scale, larger bioreactors (e.g., ≥ 10,000 L) are often used. It is proposed to use continuous-flow centrifugation followed by DF for harvest. Key centrifugation process parameters include feed flux, bowl speed, and discharge frequency. The load cell density, viability, titer, and viscosity usually impact the parameter establishment. The centrate quality can be monitored by real-time turbidity using a turbidity meter (e.g., Beer Turbidity Meter InPro 8600i, Mettler Toledo) and by particle size distribution using a laser diffraction testing instrument (e.g., Malvern Panalytical Mastersizer 2000, Particle Technology Labs).

Points to consider for downstream CBP

A fully integrated CBP for a mAb manufacturing process and purification process based on the author’s current knowledge, experience, and future expectations is shown in Figure 2. The CBP is designed based on a mAb biological property, process philosophy, operational intensification, process consistency, robustness (e.g., load range, eluate concentration, pH, conductivity, volume), viral clearance, and bioburden control (e.g., retro- and parvo-virus, microbial control). The CBP design considers impurity removal, process monitoring, and control (e.g., upstream additives, rHCP, rDNA, aggregates, fragments, charge variance) based on analytical equipment and method availability. It also addresses equipment, hardware, and software availability (e.g., inline dilution, concentration, pH, and conductivity adjustment).

Continuous mAb capture (e.g., BioSMB from Pall, PCC from Cytiva) has been successfully used from process development labs to manufacturing scale in large biopharma companies to avoid potential process bottlenecks. The continuous capture improves process efficiency (e.g., multiple columns in continuous mode) and product quality (e.g., shorten holding time and minimizing product stability from harvested cell culture fluid) and reduces cost of goods (COGs), such as raw material (e.g., resin, column, buffer). It also reduces hardware footprint (e.g., number of bioreactor tanks and size) and facility size. As this technology and system matures, it is expected to be adopted by more medium-sized biotech companies and contract development and manufacturing organizations (CDMOs). To adopt continuous capture, the starting point is from the Protein A resin screening step. There are a variety of resins available for screening (e.g., MabSelect Sure PCC from Cytiva, Toyopeal AF-rProtein A HC-650F from TOSOH, Poros ProA from ThemoFisher). The screening is typically focused on resin property (e.g., ligand matrix from highly cross-linked agarose- or polymer-based resin), binding capacity (e.g., > 60 g/L), process efficiency (e.g., short residence time, less compression factor), product purity (e.g., rHCP, rDNA, residual Protein A [rProA], aggregates, fragments), resin performance (e.g., resin stability, binding capacity, elution peak broadness, elution volume, yield), bioburden, and endotoxin control.

Once the resin is selected, capture step parameters will be assessed and established based on process design, equipment, and software. Process design based on product quality and recovery is key.The software program should be created and tested based on column size, equilibration volume/time, load titer/density, washing residence time/volume, elution peak cut and eluate volume, column post usage regeneration, sanitization, and reusing. For example, washing step optimization (e.g., pH, conductivity, detergent, or amino acids addition) not only results in impurity and/or virus removal but also facilitates multicolumn operational programming (Figure 2).

Integration of capture and low pH viral inactivation (VI) into a single step is ideal for a CBP strategy. The eluate pool pH could be designed and controlled by elution buffer strength, buffer pH, and eluted volume (3). An inline pH adjustment system can be used for pH fine-tuning if needed (4).

If the mAb is sensitive to low pH treatment, detergent treatment (e.g., 1% v/v Polysorbate 20 + 0.03% tri-n-butyl-phosphate) is an option as it inactivates enveloped virus effectively within five minutes (e.g., > 5 log reduction) (5).Detergent treatment can be conducted in the harvest tank prior to capture load or in one of the washing steps by on-column VI strategy. The residual detergent can be removed by an additional wash step without detergent added in the buffer before the elution step.

Challenges and solutions

MAb aggregation after low pH hold should be closely monitored and controlled. Lot-to-lot pH variation from the elution pool should be well-studied to understand the impact from the variation of elution buffer pH and column load density. Intermediate homogeneity can be achieved by using disposable mixers (e.g., Allegro from Pall, or Mobius Power MIX from Millipore). Aseptic sampling can be conducted manually using aseptic sampling systems (e.g., Sartorius Take One or Millipore Nova Septum). Bioburden and endotoxin monitoring and control should start at raw material release and follow through to process operation.

To reduce microbial contamination, inline 0.2-µm filtration, before load and after elution, is essential.Current 0.2-µm filtration systems are manually assembled for single-set usage. A multiple 0.2-µm filter assembly (similar design principle from Figure 1) should be invented for CBP. Each individual filter sizing should be pre-determined based on load property (e.g., turbidity, concentration, pH, conductivity) and filter property (e.g., polyethersulfone, polyvinylidene difluoride).

In-line, real-time testing methods and technologies for HCP, DNA, capillary gel electrophoresis, glycan, and capillary isoelectric focusing (cIEF) assay are essential, but they are available.

Continuous polishing step

It is highly recommended to use a two-column process for downstream CBP. If the second column load pH, conductivity, and/or concentration needs to be adjusted (4), one may make adjustments using the Cadence pH adjustment (Pall) or an in-line dilutor or in-line concentrator (e.g., from Pall or Merck Millipore).

If the product purity and/or quality does not meet its SQIPE requirement after the second column, other resins, such as anion exchanger, cation exchanger, mixed-mode resin, or hydrophobic interaction resin, may be screened (3). A third column should be avoided in a CBP strategy.

Most recently, the authors have studied the cation exchange chromatography (CEX) of flowthrough mode versus the traditional bind and elute mode. The preliminary data showed that the flowthrough mode had a much higher process efficiency (e.g., four times faster), lower cost (e.g., 90% of resin cost reduction), and higher impurity removal for high-molecular species (> 2x) and HCP (> 1.5x) than the bind and elute mode. The flowthrough mode is effective in removing high molecular species and/or residual host cell proteins, which are difficult to be reduced to acceptable level due to their evenly distribution in the elution peak from the bind and elute mode. The results demonstrate that a continuous process can be possibly conducted in a flowthrough mode from both anion exchange (AEX) and CEX with the same load density, residence time, buffer pH, and conductivity.

Continuous viral filtration

Viral filtration (VF) is a robust viral clearance step recognized by regulatory authorities that is routinely performed in batch mode. Current VF technology and systems pose a bottleneck for CBP. To implement continuous VF processing, the authors recommend designing a novel multi-pack VF system (similar to the principle and design in Figure 1). Each filter pack would be designed for a fixed amount of product based on its capacity, process efficiency, and recovery, which can be controlled based on filtration flux decay. The filtrate from each filter pack would be collected individually. After post-use integrity is tested, the individual filtrate would be pooled for formulation by ultrafiltratioin and diafiltration (UF/DF). This practice can minimize the failure of the entire VF process because post-use integrity testing can pinpoint the possible individual filters where the failure occurred.

Continuous UF/DF

Current UF/DF is conducted in batch-mode for final drug substance concentration and buffer exchange for formulation. To adopt a continuous process technology for UF/DF, single-pass tangential flow filtration (SPTFF) has been used for protein concentration (6).One such system (Cadence, Pall) has achieved a concentration factor range of 3–25 at 10–70 L/m²/h permeate flux within 25–65 psid feed pressure drop (7). This technology helps in-process sample concentration prior to column or viral filter load. Multi-module SPTFF may allow mAbs to reach high concentration (i.e., 100 g/L), but a single pass UF/DF system is not commercially available. One study, however, demonstrated a three-stage single-pass diafiltration that resulted in > 99.75% buffer exchange (8).That is encouraging news for future continuous UF/DF process and manufacturing.

Conclusion

CBP enhances process efficiency, productivity, consistency, and product quality. It can reduce overall manufacturing cost from raw materials, equipment, hardware, and facility footprint. Therefore, it would be advantageous to implement continuous process strategies in downstream bioprocessing to avoid potential bottlenecks caused by upstream high titer cultures and/or harvest volume.

The authors of this study proposed a fully automated CBP applied for downstream process and manufacturing based on a novel process design, continuous pH and conductivity adjustment for intermediates, and real-time process and product quality monitoring and control (9). The authors anticipate that specialized computer software (e.g.,digital manufacturing, artificial intelligence, remote control) and hardware (e.g., artificial intelligence-controlled probes, valves, tanks, skids) will be invented and successfully applied to future downstream CBP.

References

1. D.J. Sewell, et al., Biotechnol Bioeng. 116 (6) 1315–1325 (2019).
2. S. Le Merdy, “Evolving Clarification Strategies to Meet New Challenges,” bioprocessintl.com, Oct. 16, 2014.
3. Y.R. Ding, M.P. Marino, and H. Kumar, BioPharma International 32 (12) 24–29 (2019).
4. M. Schofield, GENS 38 (15) (2018).
5. P.L. Roberts, D. Lloyd, and P.J. Marshall, Biologicals 37, 26–31 (2009).
6. C.A. Teske, B. Lebreton, and R. van Reis, Biotech Progress 26 (4) 1068–1072 (2010).
7. C. Casey, et al., J. of Membrane Science 384 (1–2), 82–88 (2011).
8. J. Rucker-Pezzini,et al., Biotechnol. Bioeng. 115 (8) 1949–1957 (2018).
9. M.M. St. Amand, B.A. Ogunnaike, and A.S.Robinson, Biotech Progress 30 (1) 249–255 (2014).

About the authors

Yanhuai (Richard) Ding^*, PhD, rding@Anaptysbio.com, is director, Downstream Process Development and Manufacturing, and Margaret (Peggy) Marino, PhD, is senior vice-president, Program Management; both at AnaptysBio.

^*To whom correspondence should be addressed.

Article Details

BioPharm International
Vol. 33, No. 8
August 2020
Pages: 22–26

Citation

When referring to this article, please cite it as Y.R. Ding and M.P. Marino, “Points to Consider for Continuous Downstream Bioprocessing,” BioPharm International 33 (8) 22–26 (2020).