OR WAIT null SECS
Bioreactor technology advances can offer seamless manufacturing scale-up and can reduce the timeline and cost of biologics production.
In the past decade, upstream scale-up of mammalian cell culture has changed and so have the challenges associated with it. However, the cost of delaying process transfer remains stubbornly high. As manufacturing scales increase, these soon mount up to hundreds of thousands of dollars, with Matzmorr (1) estimating that for a biologic with $1 billion sales annually, the cost of every month lost during technology transfer is an eye-watering $80 million. With figures like this, it is little wonder that getting scale-up right is what keeps bioprocess scientists awake at night. Costs incurred are generally from the media, reagents, and staff resources needed to rerun a process, which can be in the hundreds of thousands of dollars at the 2000-L scale. Also lost are the opportunity costs of a product not being available for clinical trials or for patient use, and, in pandemic situations, this can result in preventable loss of life. Therefore, being able to scale-up a process in a timely manner is critical.
To conduct scale-up in 2010, microplates and shake flasks were used to select four to six clones to take forward for process development in glass bench-top bioreactors (2,3,4). Because these methods did not allow for process control, scientists would cross their fingers and hope that they had selected the best clones to take forward into bioreactors. But they remained uncertain that the clones they selected were the best as a full design-of-experiments (DoE) approach of testing multiple parameters with different clones to generate hundreds of data points would have been virtually impossible at that time unless an army of dedicated scientists was available to manually run and sample flasks constantly. The lucky clones selected, usually after around six-months of development work, were then cultured in benchtop bioreactors. After analysis of results and more finger crossing, scientists would take their star performer through culture in a series of up to six increasingly larger vessels. This would usually begin at N-5 with 125-mL–250-mL flasks being used to seed an N-4 benchtop bioreactor (5 L). That culture would be used to inoculate a series of stirred tank stainless-steel bioreactors starting at the N-3 bioreactor (10 L–50 L), and the process would transfer to N-2 (100 L–200 L), then to N-1 at 1000 L–2000 L and would culminate in an N-manufacturing bioreactor of around 20,000 L. The volumes were large because the slightly haphazard method of clone section and process development didn’t produce significant titers. In fact, the titers were on average between 1.0 g/L–1.5 g/L (5,6), which meant their clones couldn’t produce enough of most biologicals at 2000-L scale.
Fast forward 10 years and upstream cell culture looks different. With the advent of high throughout automated micro bioreactors (10 mL–15 mL) in 2010 (7) and mini bioreactors (100 mL–250 mL) in 2013 (8), suddenly scientists could culture up to 96 clones in a controlled environment using a full DoE design in weeks instead of months. They could select the best clone and really optimize everything about their process from pH and dissolved oxygen (DO) right through to the best media and feeds. In fact, studies by Hsu at Genentech (9) comparing the growth of Chinese hamster ovary (CHO) clones cultured in shake flasks, 2-L benchtop bioreactors, and mini bioreactors, showed that the performance in shake flasks in terms of growth, viability, metabolism, and titer were not comparable to either 2-L bioreactors or mini bioreactors, indicating that shake flasks are quite a poor mimic of stirred tank bioreactors.
Use of mini bioreactors was such a game changer that by 2020, plates and shake flasks have virtually disappeared in process development at biopharmaceutical companies and contract development and manufacturing organizations (CDMOs). Being able to select clones and optimize a process using mini bioreactors has helped significantly reduce process transfer timelines at the beginning of scale-up, and Lange at Cobra Biologics, a CDMO, reported that one of their process development projects was cut from many months to just six-weeks using automated mini bioreactors in place of shake flasks and benchtop bioreactors (10).
Being able to select the best clones and nail down processes has, in the past 10 years, helped double the average product titer to 3 g/L (6), enabling the use of smaller bioreactors for commercial manufacturing. This reduction in final manufacturing scale has made it possible to use single-use (SU) rocking motion bioreactors and stirred-tank bioreactors as part of the scale-up. In many pilot and manufacturing plants today, the scale-up equipment is increasingly fully SU bioreactors, especially in the United States and Europe. In Asia, the scale-up equipment is either fully SU or a hybrid of SU and stainless-steel bioreactors, particularly where companies are still using legacy equipment. A typical scale-up is now transferred through a series of up to five vessels. This usually begins at N-4, with 125-mL–250-mL flasks being used to seed an N-3 benchtop or rocking motion bioreactor (5 L). The culture is used to inoculate a series of SU stirred-tank stainless-steel bioreactors starting at the N-2 bioreactor (50 L), and the process is transferred to N-1 (500 L) and is finished in an N manufacturing bioreactor at 1000 L–2000 L (Figure 1). If a biologic is required at high capacity, then instead of using a 20,000-L stainless-steel bioreactor, facilities that don’t have the ceiling height, floorspace, sufficient capital, or the time to install such a big piece of equipment, will often instead use what Wypych describes as a “six pack” of six 2-000 L SU bioreactors (11).
With all this progress, scale-up should be easier, but it remains tricky. Most delays nowadays tend to occur during transfer from process development scale (10 mL–5 L) to the larger bioreactors because that part of the process is often not scaled in the same facility or even the same country. For example, a process can be developed in an academic laboratory or CDMO and then transferred to another CDMO or biopharma company for manufacturing. This means that a process developed in a shake flask and benchtop bioreactor or mini bioreactor could be transferred to a rocking motion bioreactor for the next scale or to a series of stirred-tank bioreactors that have dissimilar geometry and designs. Studies by Löffelholz show that using different impeller types, stirring methods, and sparging strategies can all make cell culture performance go awry during scale-up (12).
Most bioprocess scientists will say that when they transfer a perfectly fine process from one scale to another, they will often see changes. Work by Brunner describes that these changes are generally in key process indicators (KPIs), such as viable cell concentration (VCC), cell viability, cell diameter, and product titer (13). Brunner also indicates that changes to a biologic’s critical quality attributes (CQAs), such as its glycan profile (13), are equally likely to occur during process transfer. Therefore, at each scale, scientists may have to go for up to three engineering runs and one lock-down run to get the best process performance, which can take several months. Issues with VCC and viability will show up within the first three to four days of a run and are usually the easiest to fix, whereas problems with product quality are more difficult to find and solve and need to be ironed out at the pilot stage, if possible, because, as each bioreactor increases in scale, conducting a number of engineering runs becomes increasingly costly in terms of media and reagents. In addition, at manufacturing scales, scientists won’t be allowed many opportunities to meet their process goals.
The most forward-looking scientists use a quality-by-design approach during cell culture scale-up andwill typically use common scaling strategies, such as volumetric gas flow rate (vvm), mass transfer coefficient (kLa), and power per unit volume (P/V) to determine scale-dependent parameters, including gas transfer and agitation rates, in larger bioreactors (14) and scale-down models (15, 16).
At the start of scale-up, scientists will generally keep a single parameter such as vvm constant throughout the complete range of scales. However, this can cause issues when developing a process across a broad range of scales from mini bioreactors (starting at 15 mL) through to production vessels (2000 L). For example, a specific power input of 30 W/m provides a good working stirrer speed at 2000 L but translates to low stirring speeds in mini bioreactors, which can lead to poor culture performance.
An alternative approach is to use a scaling tool to scale multiple parameters in a bioreactor-size-independent way. This allows scientists to find the “sweet spot” within a parameter set of P/V, impeller tip speed, and Reynolds number (RE) to set the right agitation range for consistent mixing with the least sheer force to minimize damage to cells across scales. Currently, some scientists are trying to make their own scale conversion tools using Excel spreadsheets to build a database of in-house process data, but this task requires a detailed knowledge of bioreactors being used, which may not always be available from suppliers. Commercial versions of scaling tools are being developed by some bioreactor manufacturers using a full characterization of their vessels and information about critical influences on cell lines being cultured. Using one such commercial scaling tool in studies, Ruhl (17) generated a set of scaling parameters that were used in a proof-of-concept cell culture process from 15 mL to 2000 L for a CHO cell line expressing a commercial monoclonal antibody (mAb). This set of parameters was applied to mini bioreactors (15 mL and 250 mL), transferred from single-use pilot (50 L and 200 L) through to commercial manufacturing bioreactor volumes up to 2000 L, and resulted in VCCs and critical process parameters with similar trends across all scales and with comparable performance to historic golden batch data.
Using an in-house scaling tool, where scientists put down all the information they have on their process and vessels and then crunching as many numbers as possible, scientists can scale any bioreactor to any other type of vessel. However, it can take time to perform that amount of manual analysis and the results may not be accurate if information on bioreactors is incomplete, or if scientists are using a range of different types of bioreactors for process transfer. However, scaling becomes easier if scientists use a scaling tool in combination with mini bioreactors—which have similar geometrical height-to-diameter ratios to larger vessels (Figure 2)—for process development and then transfer that process across a single bioreactor range (17).
Finally, analytical tools that are also scalable need to be used alongside the bioreactors to ensure that the process can be monitored and measured in the same way throughout scale-up, so that like-for-like data are being compared. This allows development of a robust process as on/in-line measurements with feedback loops that are not dependent on manual measurements offline. The process can then be rapidly transferred through different scales often with as few as three scale-up steps (17), for more cost-effective process transfer.
Delays in upstream scale-up can run into hundreds of thousands of dollars a day and generally occur because not enough is known about the process before it is scaled from benchtop to pilot and manufacturing culture volumes. In 2020, technology advances, including mini bioreactors for process development, commercial scaling software tools, and bioreactor ranges that have similar geometries and associated analytics, all offer the potential for more seamless scale-up. When combined, these could shave months off scale-up timelines, thus speeding up manufacturing and helping deliver affordable biologics to countries when and where they are most needed.
1. W. Matzmorr, BioPharm International 29 (4) 46–48 (2016).
2. R. Legmann, et al., Biotechnol Bioeng. 104 (6) 1107–1120 (2009).
3. A. Amanullah, et al., Biotechnol Bioeng. 106 (1) 57–67 (2010).
4. B. Kondragunta, et al., Biotechnol Progress. 26 (4) 1095–1103 (2010).
5. S. Pearson and M. Walker, BPI 8 (9) 30–32 (2010).
6. R.A. Rader and E.S Langer, BPI 13 (2) 10–14 (2015).
7. G. Lewis, et al., Bioprocess J. 9 (1) 22–25 (2010).
8. R. Bareither, et al., Biotechnol. Bioeng. 110 (12) 3126–3138 (2013).
9. W. T. Hsu, et al., Cytotechnology 64 (6) 667–678 (2012).
10. I. Lange, S. Chhatre, and B. Zoro, BPI 12 (10) 34–37 (2014).
11. J. Wypych, “Strategies in Large-Scale Manufacturing: Rapid Flexible Capacity Expansion with Novel Utilization of a Unique Single-Use Facility Design,” presentation at Clinical/Commercial Manufacturing: BPI Theater at BIO 2015 (online presentation, Aug. 11, 2015).
12. C. Löffelholz, et al., Chem Ing Tech. 85 (1–2) 40–56 (2013).
13. M. Brunner, et al., Bioprocess Biosyst. Eng. 40 (2) 251–263 (2017).
14. S. Xu, et al., Biotechnol Prog. 33 (4) 1146–1159 (2017).
15. S. Khattak and V. Pferdeort, “Development and Qualification of a Cell Culture Scale-Down Model,” in Cell Culture Engineering: Recombinant Protein Production, G.M. Lee, et al., Eds (Wiley-VCH, Berlin, Germany, 2019), pp 391-405.
16. M. Manahan, et al., Biotechnol Prog. 35 (6) e2870 (2019).
17. S. Ruhl, et al., BPI 18 (5) 44–51 (2020).
Melisa Carpio*, email@example.com, is a global technology consultant for Cell Culture Technologies at Sartorius North America.
*To whom all correspondence should be addressed.
Vol. 33, No. 10
When referring to this article, please cite it as M. Carpio, “Current Challenges with Cell Culture Scale-up for Biologics Production,” BioPharm International 33 (10) 23–27 (2020).