OR WAIT 15 SECS
Single-use equipment, analytics, and automation may add complexity, but they are reducing scale-up costs and timelines.
Bioprocess scale-up has become more complex, but asking the most difficult scalability questions at the very start of process development is leading to significant progress. In the end, the goal is standardized processes and increased flexibility.
Converging trends are moving biopharma companies in this direction. One of the most important has been single-use equipment’s ascendance in all phases of R&D, pilot, current good manufacturing practice (CGMP), and even commercial manufacturing. Innovator companies and their contract development partners are relying more on single-use equipment throughout the value chain and taking a more data-driven approach to scale-up, says Joe Makowiecki, director of business development for Cytiva’s FlexFactory and KuBio products. The results have been reductions in timelines and costs, as well as the bottlenecks that have traditionally separated upstream and downstream operations, he says.
“While the principles of scale-up are universal, each product presents its own unique set of challenges,” notes Byron Rees, senior manager of process development services at Pall Biotech. “This is particularly true for therapies with less established platforms (e.g., the viral vectors used for gene therapies, or viral vaccines),” he says.
“Long cell-doubling times for mammalian cell systems slow down upstream process development and the ability to generate material for downstream process development to optimize for yield,” he explains. Another potential problem is analytical cell-based assays with long turnaround times, he says. “In this case, low throughput and product complexity may also present challenges, so efforts must be made to identify rapid surrogate assays that can be used for process development,” Rees explains.
Many suppliers provide small-scale process development equipment that allow for high-throughput and design of experiments, Rees says. One example is Pall’s Mustang Q ion-exchange chromatography, which is available in 96-well format for early-stage development. Parameters determined on the early-stage product can then be translated into full-scale versions using a larger form of the same product, he says.
Partnering with equipment suppliers can often accelerate process development significantly, says Rees. Recently, Pall’s process development services division in Portsmouth, UK, helped scale up the AstraZeneca/Oxford University COVID-19 vaccine. “They were able to scale up the adenovirus-based vaccine, design a full single-use manufacturing facility, and install and commission it for the first commercial run in a record eight weeks,” Rees says.
Over the past five years, some of the biggest changes in bioprocess scale-up have been seen with single-use technology. Once reserved for the earliest pre-clinical developmental stages, single-use equipment is now being used throughout the pharma value chain. Cytiva’s AKTA XL and FLUX XL systems, for example, represent extensions of its original “Ready” platforms, allowing the technologies to be used on much higher volumes of product, says Makowiecki.
Meanwhile, process intensification is transforming upstream and downstream processes, he says. Downstream, process intensification can be achieved through newer chromatography resins that are more rigid, provide increased flow rates and higher binding capacities, and can accommodate more stringent cleaning regimens. New to the market are single-use chromatography nanofibers and membrane-based units that are designed for rapid cycling and single batch production, he says. Cytiva’s new Protein A Fibro device, for instance, can be rapidly cycled, resulting in similar or higher throughput in less time, and in a compact, disposable format.
The switch from stainless-steel to single-use bioprocess equipment, if welcome from a cost and operational perspective, can still pose some technical challenges, and in some cases, has necessitated cell-line optimization efforts. “We have had to develop new scale-up paradigms to deal with single use bioreactors,” says Colin Jaques, technical director, R&D, at Lonza Pharma & Biotech.“With stainless-steel, developers control the process by considering the geometry of the vessel and its materials of construction, which allows scale-up to be based on firm assumptions. With single-use systems, however, the design and geometry for each piece of equipment differs from vendor to vendor,” he says.
Thus, in the stainless-steel days, when geometric similarity and materials of construction were assured, scale-up could be based on physicochemical considerations such as pressure/volume (P/V), mass-transfer coefficient (KLa), superficial gas velocity, mixing times, and feed strategies, Jaques explains. “Without this assurance, cell culture characterization has become much more important. At Lonza, we have moved to using multivariate data analysis methods, such as principle component analysis (PCA) and projection to latent structures (PLS), to compare culture performance across scales and between vessel types,” he says.
Scale-up is becoming increasingly data-driven. As analytical instrumentation collects more data, and better models are developed for scale-down and other efforts, developers can apply enhanced process knowledge to improve results.“Historically, we would have only applied engineering principles when scaling processes up to bioreactors larger than 250 L, but now we add insight from all our data and experience in the large-scale and small-scale relationships,” says Timothy Morris, group leader of manufacturing process development at Catalent Biologics.“This approach adds robustness to our small-scale early-stage development work, and helps us maintain a high level of success in our large-scale processes,” he says. His team trends data sets to ensure that all known potential failure modes are removed prior to scaling, Morris explains, a process that requires seasoned scientists and experience working on scale-up programs.
Catalent use multiple tools for scale-up, Morris says. In initial development stages, one of the most common is a scale-down model approach. “Based on our experience with robust CGMP systems, we use multiple factors to detect known industry failure modes, such as pH variations in large bioreactors,” he says.
In an ideal bioreactor, everything would be homogenous, Morris explains, but empirically, a 2000-L bioreactor is going to have some stratification. He recalls one case where the team saw lower bioreactor productivity at large scale and concluded that cell productivity was pH sensitive. Although the online systems read similar values, the scaling differences were not detected. “In a small-scale model, we were able to detect the same failure mode at the edge of the pH values, and incorporated that learning into our process scale-up. Modeling alone might not have detected the productivity loss,” Morris says. In the end, the manufacturing process was scaled up to achieve higher yields.
The most important scale-up approach in the future will rely on the “multiplexing of analytics,” says Morris. In the area of antibodies, complexity has already created a plethora of advanced analytics, he notes. As use of biotherapeutics grows, he expects the industry to move to even more complex molecules such as multi-specific antibodies, novel proteins, and other biomimetic modalities. These therapies will require more advanced analytics than the techniques being used today, if developers are to better understand the relationship between process and product variation. “We will see high-throughput and more sensitive analytics that provide a cost-effective way to screen variations in order to create in-process specifications,” he says,
Lonza is also using more data-driven approaches and more historical data to leverage process understanding to improve scale-up. As Jaques notes, the company has started using multivariate analysis to aid in comparison of cultures performed at different scales and in different reactor formats.
In one case, Jaques recalls that his team was characterizing a new single-use bioreactor. Data from the new reactor fit well with data generated by a comparable stainless-steel vessel, he says. However, when the bioreactor was running only half full, the behavior of the cell culture changed completely.
Using principal component analysis, the team found that data from runs performed at half volume did not fit in with those from full volume runs. Even worse, data from the cultures performed at half volume at different scales did not fit expected patterns, Jaques says. Analysis of the Q residuals for the PCA models indicated that operation at half volume had fundamentally changed the behavior of the cultures. In this case, changes to a single geometric parameter, the depth of the culture in the vessel, had a significant impact on performance.
In the future, manufacturers and contract development and manufacturing organizations (CDMOs) will have to be better prepared for scale-up, says Atul Mohindra, senior director for biomanufacturing, R&D at Lonza. “First, we have to take into account the diversity of technologies being used for current and future clinical manufacturing processes, and how these will differ when scaled up for commercial production.”
He points to the challenges of scaling up from early-stage single-use equipment to larger stainless-steel bioreactors, and notes that historical experience doing these transfers has been crucial to achieving and being able to replicate good results. Mohindra also believes that the industry needs to expand its use of data modelling and the concept of digital twins, and to get better at transferring process information, both within and between companies.
As high-density cell cultures become more important in the industry, so will the need to achieve representative scale-down models to generate high density inocula for use in high-throughput models, says Jaques. “Cell line characteristics that are required to maximize process performance in high-density cultures are different from those required to maximize performance for low-density cultures. As a result, selection decisions made early in the cell-line construction may be less than optimal if they are based on existing scale-down models,” he says.
Operating small-scale perfused N-1 cultures for high-throughput cell line selection is not practical, he adds. “Crude methods of mimicking a high-density inoculum can be developed, but these are often labor intensive and compromise the starting conditions in the production model,” says Jaques. Makowiecki sees adoption of Biopharma 4.0 concepts potentially halving the time required for standard programs (i.e., from the pretoxicology phase on) to six months.
A number of platforms permit rapid scale-up, once the process has been fully characterized and the team is satisfied with the performance of the upstream production process, recovery yield, and product quality at the process development scale. Rees cites upstream suspension processes as an example. In this case, he says, process development scientists need to evaluate the bioreactor agitation and gassing strategy. Pall’s Allegro STR allows users to base scale-up on power input and superficial gas velocity. “The key for a streamlined scale up is to keep development runs to a minimum, so choosing a well-characterized platform is vital,” Rees says.Downstream, a linear scale-up strategy can be adopted for many of the unit operations. “Here, we need not only to make sure that the filters and chromatography stationary phases are sized correctly for the process and that they are scalable, but that all the equipment and fluid-transfer manifolds are capable of operating at the required flow rates and pressures needed to maintain the process conditions from the laboratory scale,” says Rees.
As technology improves, biopharmaceutical process scale-up remains a work in progress. Given increased outsourcing, optimizing approaches will be crucial to reducing costs. In collaborative research, Amgen Corp. and MilliporeSigma have described a unified scaling approach that would tie upstream and downstream scaling approaches. “Upstream and downstream process scale-up is most commonly approached in parallel, but separately. We felt that a unified approach that applies to all bioprocesses would be helpful to the bioprocess engineers concerned with process scale-up. We also noted that the scaling techniques employed by industry are time-tested and have proven reliable. Often, scale-up failures are due to facility fit and system design constraints,” comments Joshua Arias, technical lead, BioReliance End-to-End Delivery, global pharma processing, MilliporeSigma. Ongoing automation improvements and better data analytics promise to help sort out the real signals from the noise in pharma’s scale-up projects.
Vol. 33, No. 11
When referring to this article, please cite it as A. Shanley, “Bioprocess Scale-Up: Getting Smarter Sooner,” BioPharm International 33 (11) 2020.