October 15, 2015 (Vol. 35, No. 18)
Angelo DePalma Ph.D. Writer GEN
Many of the Scale-Up Mismatches Can Be Avoided With Careful Planning
A volume change in bioprocessing changes everything else, which is why scale-up can be so unpredictable. It isn’t easy anticipating which upstream or downstream processes will require special attention. It is even hard to say whether upstream processes or downstream processes such as purification will scale more easily.
“The upstream/downstream question depends on the process design,” says Mats Lundgren, Ph.D., customer applications director, GE Healthcare Life Sciences. “For example, if you’re producing monoclonal antibodies from suspended CHO cells, upstream is probably easier to scale because you can increase the size of the bioreactor. In that situation, downstream might present more challenges because you’re producing a lot of material that you must purify.”
In other words, the dreaded upstream/downstream capacity mismatch could become reality unless purification was developed specifically for scalability. Which is why Dr. Lundgren advises process developers to select technologies and unit operations that are inherently scalable.
“You never know how a project will develop,” Dr. Lundgren cautions. “If demand for your product is higher than anticipated, you may have to address serious scale-up issues.”
Yet the upstream/downstream dichotomy does not always apply, even for cell culture. With anchorage-dependent cell lines (common with vaccine production), developers use T-flasks, roller bottles, or products such as Thermo Fisher Scientific’s Cell Factory™ systems. But here scaling up actually involves scaling out—adding more bottles or flasks.
“Having thousands and thousands of flasks is cumbersome, introduces logistical problems, and adds to the difficulty of finding contaminated vessels,” Dr. Lundgren points out. In these situations, Dr. Lundgren recommends using microcarriers, which perform well in bioreactors. “Suddenly you’re in more scalable-friendly territory for upstream operations.”
Another perspective is offered by Kenneth Ludwig, business director, bioprocess and cell therapy, Corning Life Sciences. According to Ludwig, scaling attachment-dependent cell cultures is not as straightforward as for suspension cells: “Adherent cell culture can be difficult to scale, not only because of the biological, chemical, and vessel differences that must be considered, but also because of the space required for growing an exponentially greater number of adherent cells.”
In some cases, larger bioprocessors must make capital investments, which affect overall process economics. However, there are technology solutions. For example, Corning’s HYPERStack® (High-Yield PERformance) technology maximizes the culture surface area per cubic meter and reduces the headspace required for gas transfer within the vessel by the use of a gas-permeable culture surface. “HYPERStack vessels,” Ludwig asserts, “allow manufacturers to increase cell production without expanding the footprint of their cell culture operations.”
GEN has been keeping an eye on upstream/downstream harmonization. It used to hold “next big thing” status, but it has come to seem less urgent—or at least less of a challenge. The consensus today is that biomanufacturers can overcome the more serious ramifications of mismatches by judicious planning and scheduling.
Still, the best-laid plans of bioprocess developers and biomanufacturers often go awry. For example, upstream/downstream mismatches can arise during development as processes become more efficient and productive. As Dr. Lundgren notes, “When you have a lot of material coming out of a relatively small bioreactor, your purification process has to be ready.”
Such scenarios typically pose clarification issues. To see how, consider a scenario in which improved or more-efficient upstream processing is combined with very high cell densities. Initially, centrifugation might look like a workable option. But the only way to scale that unit operation upward would be to purchase more centrifuges, hire more personnel to operate them, and devote more floor space to purification. Scale-up is not, in general, centrifugation-friendly.
On a similar note, the harvesting of highly dense mammalian cell cultures may also be limited by the capacity of depth filters, the principal alternative to centrifuges.
Thanks to the exponential relationship between diameter and column volume, chromatography is more scalable than centrifugation in terms of space and personnel, but chromatography columns are filled with expensive resins, and larger columns require more buffer for elution and regeneration. The additional expenses may be partially offset if modern chromatography resins are used. These resins have very high capacity and remain usable for many cycles—attributes that may positively affect process economy.
Thanks to continuous chromatography, achieving more separation per unit of resin is close to reality. GE Healthcare Life Sciences is deeply committed to continuous chromatography, which allows running the equivalent of several “columns” in one system. “Continuous chromatography is of great interest,” promises Dr. Lundgren, “as an approach which will maximize the productivity of chromatography operations and as such reduce the overall cost of manufacturing biotherapeutics.”
Detailed, Rational Processes
To succeed, both upstream and downstream scale-up require “a detailed and rational process,” says Aurore Lahille, head of process development at EMD Millipore. “They both involve the transfer of a process scheme, process parameters, and critical process parameters to reach similar process performances between the two scales.” But upstream and downstream operations encounter significantly different hurdles.
Downstream scale-up difficulties arise from the process scheme definition, according to Lahille. The size of each column, cassette, and filter as well as the number of cycles per step depend on the bioreactor titer, which itself is linked to successful upstream scale-up. These factors become even more relevant for continuous processes where there are no hold steps.
“For clinical process scale-up,” Lahille points out, “the best compromise between the size of the column, the number of cycles, and the cost of the resin should also be defined.”
Upstream scale-up difficulties are related to what Lahille describes as “the biological side of the process.” Each clone has its own sensitivity to oxygen bubbles, shear stress, and process parameter regulation. Cells also retain physiologic memories, so seemingly insignificant events that occurred long before the production bioreactor can affect the final bioreactor titer and molecule quality.
“However, we cannot say that upstream operation scale-up is more difficult than downstream,” Lahille adds. “There are just more parameters to consider and a higher probability of obtaining differences between the small and the large scales.”
Scale-up success therefore depends on in-depth knowledge of equipment at both scales and implementation of a rational, clear scale-up plan developed by upstream and downstream groups from early in process development. “As usual,” Lahille states, “defined deliverables, planning, and communication flow path are key.”
Considerations at Very Large Volumes
Compared with productivity-enhancing developments upstream, conventional downstream operations still scale-up much as they did 25 years ago, says Christine Gebski, vice president of product management and field applications at Repligen. “The principles and equipment have remained relatively constant,” she maintains. “Engineering concepts become more critical as you scale, but the process has become an activity-centered, time-based exercise rather than something that is technically challenging.”
Upstream scale-up is different, however. According to Repligen’s John Bonham-Carter, director of upstream sales, only a handful of companies globally possess the expertise required to scale to 20,000 L. “They’re successful at this scale because they know what to look for,” he insists. “Nonetheless, [scaling up] is not straightforward. Scaling from 2 L to 20,000 L sometimes presents unexpected challenges.”
Bioprocessors would prefer to avoid scaling to 20,000 L because cell culture at larger volume is inherently more risky, as is the combination of hold steps and large columns required to process all that protein. The move toward smaller bioreactors—driven by rising protein titers for fed-batch cultures and higher potencies—supports risk avoidance and associated scale-up issues. “It’s a lot easier to go from 2 L to 2,000 L than from 2 L to 20,000 L,” Bonham-Carter notes. “With some of today’s technologies (such as our ATF™ cell-retention device), manufacturers can achieve dramatically improved cell concentrations. They might produce from a 2,000 L bioreactor the same volume of protein as they previously did in a 20,000 L tank. New upstream technologies are helping to eliminate process steps and are allowing smaller bioreactors to do the same job.”
Higher awareness of scale-up operations is part of a more comprehensive focus on manufacturing efficiency. “Bioreactor efficiency, purification throughput, and facility operation are getting much more attention than before,” observes Bonham-Carter. “Fifteen years ago, bioprocessors readily scaled up and solved those problems at considerable cost. But those drugs served very large markets and populations. Money wasn’t the problem. Now cost and money are problems, so people focus on doing things better in ways they didn’t have to before.”
One cannot discuss scale-up without mentioning the potential for capacity mismatches between upstream and downstream operations. Commenters and consultants ran with this idea 5 to 10 years ago, but the excitement seems to have wound down. The current calm, says Gebski, is due to improvements in downstream processing (such as high-capacity resins) that allow bioprocessors to achieve the right balance between productivity, process design, and facility fit/utilization.
Another factor that comes into play is continuous processing, which provides a constant upstream product flow and maximum utilization of downstream resources. Gebski notes, however, that “once you get into continuous processing, the sequencing of downstream operations becomes more challenging, as does the elimination of hold steps for quality control.”
“Companies running 20,000 L and up were looking at chromatography columns with 2-meter diameters,” adds Bonham-Carter. “Today the industry is evaluating how to approach continuous downstream processing, using multiple smaller columns. This reduces the risk of packing and qualifying large columns, drives better resin utilization, and ultimately enables the use of prepacked chromatography columns, eliminating the art of column packing from the workflow.”
So Which Is It?
The relative difficulty of upstream versus downstream scale-up comes down to one’s perspective.
“In my view, upstream, and specifically cell culture, is the greatest scale-up challenge,” says Nick Hutchinson, market development manager at Parker Hannifin Manufacturing. “There are clear models from chromatography that work well. These include normal and tangential flow filtration. But a stirred cell culture vessel is more difficult. Some parameters, such as tip speed, average shear, and kLa, are hard to maintain as you go up in scale.
“That being said, I think most people nowadays maintain constant kLa and find it works reasonably well. If it doesn’t, however, you can have all kinds of problems with downstream being oversized or undersized. Both have happened to me. Neither one is good, and so predictability is the key.”
An alternative view is offered by Steve Garger, director of isolation and purification in Bayer’s global biological development group. Unsurprisingly, Garger believes that downstream scale-up presents more compelling challenges: “There is more variability in the feed streams that we receive, and by that I mean product titer which can range quite a bit. There can also be impurities such as host cell proteins and DNA, depending on cell viability at harvest.”
Garger provides a specific example: harvest filtration/clarification. Bayer typically uses filtration because they operate mostly at 1–2 kLa, where filtration is disposable, relatively easy to use, and avoids the worry of cleaning centrifuges. “But if you go much above 2,000 L,” warns Garger, “filtration becomes more unwieldy because of the number of filter units required.”
The other challenge during purification is the decision either to increase the number of cycles for chromatography steps or to employ a sublot strategy. The latter makes sense if a downstream process has equipment capacity to handle less than a full batch. “You purify a sublot, which might take three or four days,” Garger tells GEN, “then purify the remaining clarified harvest.”
Regardless of the adopted strategy, downstream processors could run into limits on buffer capacity—basically on making and storing buffers.
Finally, Garger mentions dealing with clarified harvest from larger processes: “A 2,000 L process is more challenging than a 1,000 L process because of the required tank capacity and mixing capacity. Even though single-use bioreactors can handle 2,000 L, these devices are not very portable. They can encounter facility restrictions, which become more problematic the larger the batch size.”
UVC Disinfection for Biopharmaceutical Development