Unifying Continuous Biomanufacturing Operations

Published on: 
BioPharm International, BioPharm International-06-01-2017, Volume 30, Issue 6
Pages: 14–19

The success of a truly integrated continuous processing platform relies on the collaborative efforts of upstream and downstream specialists.

When human cells travel throughout the body, they are constantly being influenced by the environment around them. The most efficient upstream processing systems in biologics manufacturing seek to mimic the body’s treatment of cells, so a constant level of incoming nutrients and outgoing waste products is necessary. Through continuous manufacturing methods upstream, engineers have been mastering the ways to support the growth of cells to desired volumes while preserving cell viability. Downstream continuous purification has seen some success as well, mostly through the integration of various chromatography column sequences. But the continuous production of biologics in a closed, totally integrated stream--from “stem to stern”-still remains somewhat elusive, at least in commercial settings.

Early adopters of this “bioreactor as cell” model incorporating continuous manufacturing upstream may have a competitive advantage over companies that are more reluctant to update their aging facilities. Currently, approximately 20 FDA-approved biologic products are made using perfusion, according to a comment during INTERPHEX 2017 from Parrish Galliher, chief technology officer, upstream, at GE Healthcare Life Sciences. These products include drugs made by Baxter, Genzyme/Sanofi, Biogen, Merck Serono, Bayer, BioMarin, Pfizer, Janssen, Novartis, Shire, and Eli Lilly. Although these drugs only exploit continuous techniques upstream, experience with this type of process upstream may help to eventually inform end-to-end manufacturing solutions. As the concept of integration makes its way from theory to practice, manufacturers will have to carefully weigh the costs and benefits of using continuous methods for biologics. In the future, manufacturers will still have to identify optimal cycle time; weigh the risk of scale-up and explore how to construct continuous models that could eliminate scale-up altogether; test the consequences of continuous lines on product quality and stability; and consider what vendor gaps exist that could preclude a completely end-to-end continuous stream for the manufacture of biologics.

Optimal operation and residence time

Manufacturers may not want to be as resistant to explore truly continuous systems for the production of biologics. Higher productivity per volume means companies can manufacture products at a faster rate. Lawrence Yu, PhD, deputy director in the Office of Pharmaceutical Quality at the Center for Drug Evaluation and Research (CDER), said in a 2016 FDA blog post that processes that take a month with fed-batch could take only a day with the implementation of continuous processing (1). While this estimate sounds intriguing, says Felipe Tapia, researcher at the Max Planck Institute for Dynamics of Complex Technical Systems, he points out that there is no reference nor calculation in Yu’s blog that would support this specific assertion.

Not only is the time to create a batch reduced, but the final product has a shorter residence time within the process. Reductions in protein product residence times are thought to help minimize the effects of adverse chemical or enzymatic modifications.

Researchers and process engineers are still grappling with how to identify the optimal product cycle length. John Bonham-Carter, director of upstream sales at Repligen, says Repligen estimates the median run length for new processes in clinical trials today is approximately four weeks, with approximately four-five days dedicated to start-up time. “There are more and more companies investigating five-to-six-week processes as skills and experience expand, while [the run length time for] commercially launched products in perfusion from the traditional perfusion experts such as Janssen, Genzyme, Bayer, Shire, and others might be several months.” He continues, “The current driver to have shorter runs at only four weeks is a mix of practical operations, company experience, and often, out of concern for the life of single-use bags or other equipment reliability used in the process. Given the lack of experience of many companies starting this journey, it is unsurprising to err on the side of caution until a broader experience is achieved.”

In an analysis, a continuous cell culture ran for 26 days, but could have continued even longer, according to the study authors: the cell culture remained viable after this time period, “and the continuous capture process did not show a decline in performance” (2). A previous experiment by Klutz at el. included the upstream and preliminary downstream steps of a monoclonal antibody (mAb) in 28 days, with the secondary downstream steps taking an additional two-and-a-half days (3).

A 1998 analysis found that perfusion cultures can be in operation for up to six months (4). But that is just the upstream part of the process stream, so how long could end-to-end production processes continue without there being a marked decrease in quality--and when does overpassaging, or keeping cells in culture for too long, start to become a concern?

When one considers the total residence time of the cells in a process stream (not the residence time of the final protein product), perfusion may actually introduce more opportunities for genetic drift. Hughes at al. wrote, “As a result of selective pressures and genetic drift, cell lines, when kept in culture too long, exhibit reduced or altered key functions and often no longer represent reliable models of their original source material” (5). Indeed, as Véronique Chotteau, PhD, researcher at KTH Royal Institute of Technology notes, “Since continuous (perfusion) processes are longer (in time) than fed-batch, the concern [of quality] is higher due to a higher chance to have genetic mutation event.” In other words, genetic instability is more of a concern with cells in a continuous line than in a fed-batch line, even though target protein products are more homogenous in a continuous line and final yields between runs remain similar (6).

Bonham-Carter comments that the proclivity for drift varies depending on the cell line used and on the product molecule being manufactured. “Typically, in antibody and most recombinant protein production, cell lines are very stable. Certain exceptions exist, and certainly, a few companies have published on enzyme-producing cell lines that exhibit instability [and] must be monitored.”

The continuous manufacture of viral vaccines is a bit different, comments Tapia, who specializes in this field. He explains that stable continuous virus production is not possible if the cell and the product (virus) are propagated in the same vessel because viruses are lytic. “For continuous virus production, multi-stage bioreactors are needed, with configurations that can be either cascades of continuous stirred-tank bioreactors (CSTRs) or CSTRs followed by tubular reactors. In all these cases, the residence times, as well as the duration of the production process, have to be clearly identified in order to guarantee product quality stability,” Tapia asserts. However, using perfusion to obtain high cell densities and increase output per bioreactor volume for the production of vaccines is of interest to researchers, too (7).

No need to scale: Eliminating scale-up

A major benefit of integrated continuous processing is that questions about whether a product will retain its properties after scale-up--and if a product can successfully be produced within larger-volume tanks that have different geometric specifications--essentially disappear. Klutz at al. wrote that efforts to achieve equivalency between bench-scale discoveries and commercial applications become null and void, and as a result, the “commercial manufacturing process will run longer in order to meet the manufacturing demands” (3). In other words, in continuous manufacturing, R&D scale is the same as commercial scale, and a researcher can conduct development studies using commercial equipment (8). Process intensification and the use of high-seed cultures, coupled with continuous methods, allows manufacturers to produce the same quantity of protein as pilot-scale resources in half the time. Benchtop systems, then, could be sufficient for the commercial production of some mAbs. Integrated continuous bioprocessing is already regularly being achieved in small research labs, according to Bonham-Carter, but has not yet made its mark in commercial, cGMP environments (8).

Product quality improvements

The known benefits of continuous manufacturing include: reduction in the degradation effects of proteases (9), a smaller footprint, higher cell concentration/higher productivity per volume, lower residence time, reduced buffer consumption, reduction of human error, the elimination of hold steps (which means safer processes), lower risk associated with changing scale, lower capital and operating costs for a facility, and maximization of column capacity/resin use. Some of these benefits could fluctuate based on market trends, however. As Klutz et al. pointed out, if resin prices were to come down, downstream continuous manufacturing might no longer offer an economic benefit (3). Additionally, says Yvonne Genzel, PhD, team leader of upstream processing in the bioprocess engineering group at the Max Planck Institute Magdeburg notes, any reduction in human error has to be carefully balanced with potential new errors related to handling disposable perfusion equipment and the proper control of perfusion.

Karst et al. highlighted a compelling reason to consider adoption of continuous manufacturing: The group showed that in addition to the well-known benefits of continuous processing, including increased flexibility and higher productivity per volume, an integrated biomanufacturing system produced mAbs of higher, more consistent product quality (due to a steady state) than fed-batch systems (2).

This is also likely to be true for the production of non-mAb biologics, says Massimo Morbidelli, a study coauthor, professor of chemical reaction engineering at ETH Zürich, and co-founder of ChromaCon AG, who notes that end-to-end continuous processing would also improve the production quality of fusion proteins. Continuous mode may be especially appealing for products that are toxic to the overall process, says Genzel, because it would reduce the toxic effect of the product remaining in the reactor.


Increased product quality is a common goal of continuous manufacturing, according to Bonham-Carter. But the product quality improvement that the Karst et al. team saw was not just related to product uniformity, but also to product stability. This enhanced stability was directly associated with a shorter product residence time. The researchers wrote, “The enhancement of product quality compared to fed-batch is indeed the most relevant one in a regulatory perspective. The stable operation at steady state and the short residence time in the reactor favors homogenous post-translational modifications of the protein and mitigates the effect of chemical and enzymatic degradation, as well as the formation of aggregates” (2).

The samples in the Karst et al. experiment showed good glycosylation consistency and “minimal variations of the attached glycan species” (2). They added that this so-called “culture and capture” approach is expected to “reduce heterogeneity in other product characteristics such as glycation or oxidation with respect to classical batch operations” (2).

Mechanistic modeling may be key for purification processes to be able to react to the inherent variability of cells upstream that result from perfusion, the Karst team demonstrated. Mechanistic modeling of the capture operation allowed downstream operations to adapt to changes in the perfusion bioreactor, the researchers wrote. They concluded that compared with fed-batch--where charge isoforms, aggregates, fragments, and N-linked glycosylation were variable from batch to batch--“the enhanced control and constant cellular environments in continuous operation allows the modulation and fine-tuning toward desired product characteristics” (2).


Gaps in vendor offerings and processing knowledge

The pilot plant described in the Klutz et al. study was designed to run using primarily single-use, disposable materials. Of these materials, 95% of parts were gamma sterilized. There is difficulty associated with sterilizing chromatography columns; there are “no sterility testing [methods] for chromatography columns,” according to the study authors (3). Also, Klutz and team admitted the single-use equipment they used in the paper was meant for fed-batch, so no analysis of extractables and leachables was conducted; this analysis would typically be a necessary step for the GMP production of biologics. However, there are vendors that are now investigating column sterility, says Bonham-Carter: “Long-term life of the columns is a current topic of conversation in the industry, and Repligen has been examining the effects of gamma irradiation on columns and their feasibility in long-term continuous use.” He adds that there has not yet been a large demand for gamma-sterilized columns, however.

According to Morbidelli, although there are numerous types of sensors on the market, there is still a need for different types of sensors to inform operators about potential disturbances within a process. “If we want to exploit the potential of end-to-end integration, we need to have sensors that tell us what is happening in the various positions along the production train,” he comments. And Morbidelli asserts that while mathematical models for chromatography are well developed already, models and control strategies for bioreactors are not as well supported.

Advancements in sensors informing operators on product quality, glycosylation, viral titer, viral antigen amount, and osmolality, specifically, would be of great use to the industry, noted Genzel, and Tapia added that all sensors used in a continuous train must be tested and validated for their use in processes with long operational times.

Continuous purification campaigns are the most challenging to implement, say many experts. Kartz et al. described the potential set-up for the end-to-end production of a mAb in a 2015 study, in which the pilot plant moved the bottleneck from downstream back to the upstream processes. In the study, it was determined that countercurrent downstream technologies were key to keeping up with bottlenecks: “The cyclic process was able to load a single column beyond its dynamic binding capacity without losing product by loading the unbound protein onto the second step” (2). According to a study by Andersson et al., to optimize continuous downstream operations, engineers have to consider product residence time, total size of the columns, automated methods to control disturbances, and packing variations of columns (10).

In terms of cell retention devices, alternating tangential filtration (ATF) allowed a significant technological step ahead compared with tangential-flow filtration (TFF), comments Morbidelli, although he says the scale-up for both ATF and TFF remains, to a certain extent, an open issue. Bonham-Carter notes that using ATF for cell retention is preferable, and adds, “ATF is used in approximately seven commercial processes today and is the industry standard.” In a six-step model of an end-to-end system originally proposed in 2013 by Peter Tiainen and fellow researchers from Novo Nordisk, an ATF perfusion cultivation system and a modified ÄKTA pure protein purification system (GE Healthcare Life Sciences) were integrated (11).

In fact, Tiainen and Novo Nordisk are listed on a United States patent (filed in March 2017) describing the integrated setup, which they say is the first investigation that attempts to match the flow of upstream processing units with downstream units without the use of a surge vessel or intermediate hold vessel (12). In the patent, they say the use of surge vessels--which would typically be required in integrated continuous streams to manage overflow and overpressure--are not suitable for the production of fragile proteins with stability issues (12). Prior to the Tiainen/Novo Nordisk patent, Bayer HealthCare was granted a patent in Europe in 2016 for the integrated continuous manufacture of biologics. The two proposed platforms in that patent, however, used surge tanks, and in some embodiments of the invention, the use of a convective adsorption/desorption system was only semi-continuous (13). While the Bayer patent only demonstrated the efficacy of an integrated biomanufacturing train for the production of recombinant blood coagulation Factor VIII, the authors of the patent argued “the inventions can clearly be expected to be similarly useful for the production of other proteins and biological molecules, in particular, complex inherently unstable proteins such as Factor VII, Factor IX, Factor X, and others” (13).

Another company, Univercells, also intends to capitalize on improvements in processing knowledge. The company was granted a US patent in December 2016 for a portable, integrated, automated method for the manufacture of antibodies. Specifically, the patent describes the production of Synagis (palivizumab), a respiratory syncytial virus-antibody-based vaccine. Univercells received a $12-million grant from the Bill and Melinda Gates Foundation to develop the continuous biomanufacturing platform (14). According to language in the patent, “The invention allows providing third-world countries with national production systems and also enables pharmaceutical companies without [a] biotech background to produce cell products such as antibodies in bulk” (15). Interestingly, Univercells also stakes some claim to the processing methods further down the line, including processes that could occur after the cell-secreted product is collected and purified: “For the purpose of equivalents the formulations may be lyophilized if desired. Thus, the invention encompasses production of lyophilized forms of the formulations” (15).

Other hurdles to process integration are cell lines and media, says Morbidelli, which are currently optimized for fed-batch operations and are not ideal for the purpose of perfusion. “We observe, many times, a decrease in productivity of the cell line after 30 days.” To combat this problem, Morbidelli says vendors should focus on better clone selection for cell lines destined for perfusion-based processes. “If a cell line is robust, in time, longer perfusion runs would be possible.”

According to Bonham-Carter, “Until this month [April 2017], there has been no off-the-shelf media available for perfusion culture. Therefore, almost everyone utilizes the same methodology of taking an existing basal or fed-batch media and adjusts it according to spent media analysis studies performed at various different steady states.”

Another area that could benefit from industry improvement is downstream, continuous viral inactivation, comments Morbidelli. He also notes that he expects to see more membranes--specifically, multi-membranes--being used downstream (in addition to downstream chromatography).

Products to support and enhance the continuous manufacture of vaccines (i.e., media) are fairly underdeveloped as well, according to Genzel. Adds Tapia, “Continuous production is currently not a popular topic among vendors and stakeholders in the vaccine industry. Most of their efforts are focused on process development using traditional batch operation.” Tapia explains that “previous studies have shown that virus replication in continuous mode is limited by the accumulation of defective interfering particles (DIPs)” and is associated with the risk of unwanted viral antigenic aberrations when viruses undergo too much replication. Tapia, Genzel, and their collaborators are currently investigating new methods to enable the continuous production of cell-culture-based vaccines in perfusion and multi-stage bioreactors (16).

When process specialties collide

The best way to achieve process symbiosis and integrate upstream and downstream operations, according to Bonham-Carter, is for vendors in both sectors to begin collaborating. “Integrating up and downstream R&D often does not mean each unit operation is developed and integrated continuously with others, since equipment supply is mostly from different vendors for each unit operation.” Thus, he notes that “close collaboration between groups and rapid iterative developmental strategies” should be examined. “For example, within one run of a reactor, four different cell concentrations (40, 60, 80, and 100e6/mL) could be held for a week each, and the downstream group may evaluate the product quality and impurity profile for each state,” he adds. “The next run might modify media composition and perfusion rates, and repeat the same four steady states.”

Genzel comments that the aforementioned developmental strategies would be “difficult for academics,” as experiments like these are generally too cost prohibitive for this sector. She adds, “Vendors, as well as companies, are not sponsoring such research enough.”

Indeed, there are product offering gaps to fill, technology improvements to be made, and investments in process collaboration studies that need to occur before end-to-end process integration for continuous biomanufacturing can be adopted on a widespread scale. Concludes Bonham-Carter, there are “new product opportunities in nearly all unit operations, given there are no drugs commercially produced in a continuous downstream process yet.”


1. L. Yu, “Continuous Manufacturing Has a Strong Impact on Drug Quality,” FDA Voice, accessed April 28, 2017.
2. D.J. Karst et al., Biotechnol. Bioeng. 114 (2), 298-307 (February 2017).
3. S. Klutz et al., J. Biotechnol. 213, 120-130 (2015).
4. S.M. Woodside, B.D. Bowen, and J.M. Piret, Cytotechnology 28 (1-3), 163-175 (November 1998).
5. P. Hughes et al., BioTechniques 43, 575-586 (November 2007).
6. D.J. Karst et al., Biotechnol. Prog., doi: 10.1002/btpr.2421 (Dec. 20, 2016).
7. Y. Genzel et al., Vaccine 32 (24), 2770-2781 (May 19, 2014).
8. J. Bonham-Carter, “Inspiring Advances in Bioprocessing,” Repligen presentation (March 14, 2017), accessed April 28, 2017.
9. S.B. Mohan et al., Biotechnol. Bioeng. 42, 974-986 (1993).
10. N. Andersson et el., Biotechnol Prog. (Epub ahead of print), doi: 10.1002/btpr.2434 (Jan. 5, 2017).
11. Tiainen et al., “End-to-End Continuous Production of Complex Recombinant Proteins--Integration of Perfusion Cultivation and Automated Multi-Step Purification,” abstract presented at Integrated Continuous Biomanufacturing I (Barcelona, Spain, Oct. 2013), accessed May 2, 2017.
12. M. Aakesson et al. (Novo Nordisk A/S), “Integrated Continuous Biomanufacturing Process,” US patent application 20170058308, March 2, 2017.
13. J. Vogel et al. (Bayer HealthCare), “Devices and Methods for Integrated Continuous Manufacturing of Biological Molecules,' European patent EP2550971, Sept. 11, 2016.
14. Univercells, “Univercells Receives $12 Million Grant to Develop Breakthrough Vaccine Manufacturing Platform,” accessed May 5, 2017.
15. J. Castillo (Univercells NV), “System, Apparatus, and Method for Biomolecules Production,” US patent 20160355572, accessed May 5, 2017.
16. F. Tapia et al., Appl. Microbiol. Biotechnol. 100, 2121-2132 (Jan. 13, 2016).

Article Details

BioPharm International
Volume 30, Number 6
June 2017
Pages: 14–19


When referring to this article, please cite it as R. Hernandez, " Unifying Continuous Biomanufacturing Operations," BioPharm International 30 (6) 2017.