January 1, 2017 (Vol. 37, No. 1)
Angelo DePalma Ph.D. Writer GEN
Scaleup Involves More Than Multiplying Ingredient Quantities, Container Sizes, and Cooking Times
If you did any baking during the holidays, you learned (or relearned) that scaleup involves more than multiplying ingredient quantities, container sizes, and cooking times. As with adjustments to baking recipes, so it is with more complicated scaleup projects—such as those encountered in bioprocessing.
Scaleup projects in bioprocessing continue to grapple with cell harvesting challenges, says Parrish Galliher, chief technology officer, upstream, GE Healthcare Life Sciences: “As mammalian cell densities climb and product titers rise, separating the drug from cells is increasingly concerning. There are no elegant ways to do this.”
High cell densities can overburden the usual cell harvesting approach, depth filtration. Although alternatives to traditional depth filtration are being investigated, none are yet ready for prime time, notes Galliher.
GEN has previously reported on two attractive alternatives: Pall’s acoustic wave separation and Repligen’s alternating tangential filtration (ATF). Acoustic wave separation uses sound to coagulate cells, causing them to fall from suspension. Pall has demonstrated a 75% reduction in post-separation filter footprint (as with centrifugation, unfortunately, still necessary) and reduced usage of buffers and water for injection.
Similarly, ATF operated in a concentrated fed-batch mode is an attractive alternative to depth filtration or centrifugation in that it produces cell-free permeate. Continuous operation is also possible but adds complexity and validation burden.
Centrifugation is a traditional process that removes most cells but still requires some filtration as the centrate is not totally cell free. Single-use centrifuges, however, have lower throughput compared with stainless-steel systems.
Some companies have tried extracting product directly, either with or without cell flocculation, “but these methods are expensive and/or difficult to execute,” states Galliher. Flocculant containing cell mass presents a increased logistics and disposal burden as do filtration-enhancing agents. “When it comes to clarification,” informs Galliher, “there is no free lunch.”
Adding to the dissatisfaction with conventional, inelegant harvest methods is the proliferation of single-use equipment and ever-rising cell densities. “Per-cell specific productivity has leveled off somewhat,” Galliher observes. “So to get more product, processors are jacking up cell densities to 70, 80, 100 million cells per milliliter.”
Factors Affecting Quality
Maintaining product quality through scaleup is challenging as several parameters change, including equipment and components used to manufacture the product. Annelies Onraedt, Ph.D., director of marketing for cell culture technologies at Pall Life Sciences, provides the example of bench-scale processes often using peristaltic pumps, where at higher flow rates other pumps are specified to maintain quality under different conditions of shear stress on cells and proteins.
Whereas some downstream operations, such as tangential flow filtration, are nominally scalable, bioreactor scaleup requires deep understanding of the multiple parameters affecting the oxygen uptake rate. “There are general rules, like keeping the impeller tip speed constant,” says Dr. Onraedt, but smooth scaleup demands deep process understanding.
Another example of components affecting scaleup is the availability of single-use sensors that measure pH and dissolved oxygen. During cell culture, these sensors are typically optical devices with limited ability to undergo gamma irradiation. Higher radiation doses employed for scaled-up systems limit the use of these sensors in large single-use bioreactors.
With single-use equipment having become common over the past 15 years (and entrenched at certain scales over the past 5 years), one would think that more robust sensors would now be available at all scales. But a significant gap still exists between need and availability. “I can’t speak to why,” admits Dr. Onraedt. “I know there’s a lot of effort in developing robust sensors.”
On the subject of process changes to accommodate changing requirements at various scales, Dr. Onraedt notes that significant alterations are never welcome, even though they are necessary in many cases.
“You’ve made clinical materials, and only when you begin scaleup do you notice the flaws in your process,” warns Dr. Onraedt. “That’s when you feel the pain of high-cost equipment and operations. Things that seem affordable at small scale, that enable you to get into the clinic, suddenly become impractical.”
Related is the increasing reliance on multiple contract manufacturers at various scales. “You must accept that process changes will occur,” Dr. Onraedt adds. “If you are to accommodate these changes more easily, you need to be prepared. Before it is scaled up, your process should be well characterized at the smaller scale. Understanding the design space of your process will speed up your path to a robust large-scale process.”
Favorable Turndown Ratios
Employing one large-volume single-use bioreactor across several scales is highly desirable but not always achievable. Practical working volumes depend on the system’s turndown ratio—the quotient of total volume and working volume. Thus, a single-use bag with a turndown ratio of 5:1 allows operation of a 100 L single-use biorector bag at just 20 L, providing the culture with “room to grow” without changing bioreactor bags.
Mixing and sparging are the main limitations to high turndown ratios, notes Surendra Balekai, product manager, Thermo Fisher Scientific: “In lower-volume operations, bottom-driven and top-driven mixing both claim advantages, such as the ability to grow seed cultures at small scale and ramp up from there.” Thermo Fisher uses top-driven mixing whereas most other vendors to the bioprocessing industry use bottom-driven mixing.
During an investigation into mixing techniques, Thermo Fisher scientists noted the significance of headspace in the turndown ratio calculation. Simply put, sparging the metabolic gases accumulated from headspace becomes more significant at higher turndown because the gas volume impact on cell culture is much greater at 80 L than at 40 L working volume.
“If metabolic gases are not flushed out effectively,” says Balekai, “the effect on cell culture is significant, creating an acidic blanket.”
Given the cost of acquiring and disposing of single-use bioreactors, the economic drivers for cross-flow sparging can be substantial. “Operating at higher turndown reduces the number of bioreactors you need to go from seed to medium production scales, even up to 2,000 L systems, depending on seeding ratio,” explains Balekai.
Conventional scaleup requires separate bags for several intermediate scales. “If you eliminate just one bioreactor during scaleup, you’re saving significantly on operational expenses,” insists Balekai, “not to mention maintenance capital investment, floor space in clean areas, and maintenance.” Cross-flow sparging, a patented technology, is undergoing alpha testing at several Thermo Fisher customer sites.
Volumetric capacity improvements through process intensification is one goal of efficient scaleup—the ultimate “economy of scale.” According to Ken Clapp, senior manager for applications, technology, and integration at GE Healthcare Life Sciences, the competitive aspects of intensification arise from improved process economics or business models.
“Although process intensification may be interpreted or implemented in a variety of ways, I believe it still falls under four main categories: performance, compatibility/coordination, liquid management, and operability,” Clapp asserts. “Process intensification highlights the requirement for quality by design (QbD) in the early stages of scale-up and associated requirement for appropriate process analytical technologies.”
Clapp offers as an example a bulk perfusion cell culture process with a vessel volume or more per day of fresh media input and waste media output. The scaling up of a properly intensified analogous process, one that eliminates the bulk media dependence in favor of smaller volume of specific media constituents, is facilitated through the selective use of single-use components and systems.
“Perhaps this type of application, especially with removal of the bulk media processing dependency, broadens the process operations that fall within the scope of single-use technology,” Clapp speculates, “spanning everything from media preparation to bulk biologic product and beyond.”
Platform Processes: A Mixed Blessing
Over the years, bioprocessors have adopted template or platform approaches, particularly for monoclonal antibody products, that compress process development times. These approaches, however, have a downside, observes Willem Kools, Ph.D., vice president and head of technology management at MilliporeSigma: “Developers have less data on the development and manufacturability of specific molecules. In these cases, specific molecules that present development issues that differ from the norm might become more challenging, particularly during scaleup.”
That specificity may express itself in plant fit (or lack thereof), or a reduced understanding of process variability. “As therapeutic bioprocess technology becomes more industrialized,” Dr. Kools indicates, “minimizing process variability is needed to design a well-controlled process.”
Moreover, as bioprocessors deploy single-use technology at different scales, from bench to 2,000 L and beyond, they must devise strategies for managing equipment and consumables. Such strategies address the storage and installation of disposable bags and the decontamination of consumables after their (single) use, as well as disposal itself. Both decontamination and disposal are more cumbersome if they follow the processing of highly potent molecules.
Another mixed blessing occurs in process modeling, which increasingly relies on smaller and smaller scaling tools for screening cells, media, feeds, and process conditions. Process modeling techniques may shorten development times and reduce the mass of protein required for process development, but during scaleup they can lead to greater scaleup factors and hence uncertainty.
As inter-plant technology transfer becomes more and more the norm, porting a process to another country or regulatory jurisdiction without adequate process knowledge becomes quite risky. “What you know about your process affects how smoothly technology transfer will go,” comments Dr. Kools. With so many processes going global in terms of patients and markets, having insight into various regulatory landscapes will facilitate efficient process transfer.
On the matter of process changes to support technology transfer, Dr. Kools notes that “with the right quality programs from vendors that take into account global regulations, you will have greater peace of mind when a process changes geography or is produced for a different country.” He adds that MilliporeSigma spends a lot of time assuring that it “has the right product, with the right level of information,” so site changes can be managed in a more straightforward manner regardless of where the product is made.
As quality-by-design is implemented more widely, bioprocessors increasingly work within a design space. Changes occurring within that design space during scaleup or technology transfer proceed much more easily.
In the Right Order
The objective of scheduling during scaleup is to obtain a good estimate of the production rate—the number of batches a facility can reasonably produce within a given time period. While scaleup usually focuses on unit operations such as cell culture, there is much more to consider. Another way to view scheduling is to understand better the true constraints with respect to timing, and thereby predict productivity at larger scale.
Duration of operations is infrequently linear with scale. “Cell culture lengths of time often change during scaleup, sometimes to longer and sometimes to shorter culture times,” says Charles Siletti, Ph.D., a principal at process modeling firm Intelligen.
Steps that seem insignificant at small scale, such as transferring materials or cleaning tanks, take on independent significance at large scale. Additionally, operations that work one way at the bench may not be suitable at production scale. Chromatography is one example. “You can scale by using a larger bed volume and more resin, or you can cycle the column more frequently,” explains Dr. Siletti. “Each scenario affects scaleup scheduling differently.”
Multiproduct facilities do not present any special scheduling concerns. “The suites are normally separated and do not share equipment without shutdown and cleanout between campaigns,” Dr. Siletti elaborates. Theoretically then, one could fit the available time and resources for two or more products sequentially.
Other factors to consider are water for injection, solution, and media preparation resources, which may be constrained by the number or scale of manufacturing campaigns and the facility’s capacity. “You need to calculate how much buffer must be staged for a given batch and have room to hold it,” urges Dr. Siletti. “Floor space calculations for storing single-use bags or capacity for tanks could be useful.”
Similarly, the need for pure water, which is used in large quantities for cleaning, might affect the decision to use stainless-steel or single-use equipment. These devices trigger their own scale-dependent resource requirements.
An often-overlooked scaleup issue involves manpower. “Small-scale and larger facilities utilize human resources differently,” Siletti notes. The former often operate in discrete shifts of one or two per day; the latter operate 24/7. “Scheduling will help with manpower allocation as well,” Dr. Siletti advises.