October 15, 2016 (Vol. 36, No. 18)

A Change in Scale Can Change Everything

The adoption of modern analytics, design of experiment approaches, robust scaledown models, and data analysis software has aided bioprocessors in addressing or solving most of the major engineering complexities related to bioprocess scaleup.
 
As a key focus remains on preserving product quality, GEN spoke to a number of scaleup experts 
to learn how recent technological advances have improved companies’ abilities to optimize their 
biomanufacturing operations.


GEN spoke to a number of scaleup experts.

GEN: How have bioprocess scaleup challenges changed over the past decade or two? What remain as major impediments to a successful biomanufacturing scaleup strategy?

Mr. Keijzer: The challenges we have seen over the past decades were in scaling down and not so much in scaling up. Where a 3 L bioreactor was a standard laboratory volume system 15 years ago, there is a new demand in the range below 500 mL. The challenge in going to smaller-scale bioreactors was generating reliable, reproducible, and scalable results to expedite the scaleup process. These issues have been resolved in our small-scale systems by using computational fluid dynamics and extensive mixing and mass transfer studies. This has resulted in the design of well-mixed small-scale systems down to 5 mL working level.

Dr. Sha: The bioprocess field borrowed most of the scaleup methods from the more mature chemical engineering field and adapted these strategies for biological processes. More recently, we saw the field establishing a scaleup strategy that keeps the impeller power consumption per liquid volume (P/V) constant, as the gold standard for both cell culture and fermentation scaleup.

To calculate P/V, the scientists have to measure the impeller torque for each vessel. The challenge remains that each torque adaptor must be made specific to each vessel by the customers. Constant P/V-based studies provided by the manufacturers can be used as a guide and as a means of increasing the user’s confidence.

Mr. Galliher: In upstream mammalian monoclonal antibody processes, higher expressing cell lines and fed-batch media optimization have boosted cell densities and product concentrations 10–20-fold. Scaleup of these processes can create oxygen transfer limitations in larger bioreactors that are not equipped with high O2 mass transfer capability.

Scaleup of cell clarification operations for much higher cell masses (>30M cells/mL) has overwhelmed depth filtration systems, forcing bioprocessors to employ filter aids or flocculants, or to add centrifugation or acoustic separators as a preclarifying steps to filtration.

Scaleup of higher productivity monoclonal antibody processes has been managed in the downstream with higher capacity protein A resins, larger column chromatography systems, and multiple column cycle operations. However, these larger batch sizes require very large volumes of buffers and large tank farms. This is forcing bioprocessors to limit the number of different buffer chemistries for purification and to also consider buffer concentrates coupled with in-line dilution/conditioning systems that can reduce buffer and tank volumes by 10–20-fold.

Lastly, some companies are avoiding upstream scaleup challenges by scaling down into smaller continuous bioprocessing systems whose volumetric productivity can be 10–20 times that of batch or fed-batch systems. In addition, continuous bioreactor systems that employ filtration for cell recycle perfusion produce a cell-free product stream that can be captured directly on the first capture column, thus avoiding the need for problematic cell clarification steps. Downstream continuous operations offer increased resin utilization and productivity and less resin volume, with the added benefit of reduced buffer volumes.

Increasing capacity of continuous processes is achieved by running longer cycles or scaling out by adding additional production lines of the same scale.

Dr. Hebel: In the past, it was often enough for scaleup to “work.” Partly optimized biomanufacturing processes for patent-protected products did the job and yielded sufficient titers in order to be economically viable. As the patents expire, the cost pressure increases and the processes need to become more efficient.

Therefore, many of these processes are taken back to the laboratory in order to gain further process understanding and to maximize the yield. Precise knowledge of the transfer between the differently scaled bioreactors and a good simulation of the biomanufacturing process in small-scale bioreactors is key for obtaining the optimum results.

Dr. Kools: Process development times have shrunk significantly. People rely on existing information, generating minimal new data on new molecules. This means less data is available to determine process-specific variability.

It is expected that many new molecules will be transferred to multiple sites during their life cycle. For such a change to occur, existing facilities will require different facility fit considerations.

MilliporeSigma and other technology providers have created smaller process development tools. These tools reduce the cost of materials and increase scaleup factors.

Another significant change has been the rise of single-use, which has introduced unique challenges including warehousing, waste disposal, leachables, and different decontamination processes.

Dr. Onraedt: Getting consistent results out of a scaled-up process has been—and remains—a significant challenge. As single-use bioreactors have become prevalent, scientists have also witnessed a paradigm shift in scaleup techniques. The move from customized stainless-steel reactors to fully single-use manufacturing systems enables development teams to easily employ characterization data generated by manufacturers, as well as others who have adopted the technology.

However, even though the engineering burden is now shared, cell physiology remains an impediment. The increased demand for productivity and cell growth over the last two decades has also created new challenges for supporting robust processes at larger scales.

Dr. Fenge: In the past, the paradigm was to scale a process in 10-fold steps to prevent unpleasant surprises, whether concerning growth performance and productivity or product quality. Particularly, product quality has always been a major concern, and step-wise scaleup enabled understanding of differences in bioreactor performance. This has changed dramatically.

Today, companies are much more confident about scaling a process in 100-fold or larger steps.

However, especially in the age of single-use, it remains very important to understand differences and how they affect product quality, as bioreactor designs are not necessarily fully aligned with classical stirred-tank principles and plastic materials pose the risk of toxic leachables.

Mr. Shaw: Initially, scaleup was primarily an engineering problem. Early areas of focus were bioreactor volume and scaleup data, and then downstream capacity. Once the engineering work accomplished 10–20,000 L production as routine, biology became an increasing focus area driving cellular productivity to increase from <1g/L to routinely 3–5 g/L at commercial scale.

As capital and timelines for expansion are tight, facility intensification and changeover are now being scrutinized, and we recognaize that single-use technologies are essential to addressing these challenges. Finally, with more powerful and efficient analytical technologies, we have tools to understand how to control raw materials for consistent production and product quality.

GEN: Which technological developments have taken place in recent years to help companies improve their scaleup efforts?

Mr. Keijzer: The more recent challenge is getting more information out of these smaller-scale systems by using more and smaller sensors and by combining process modelling tools to compute more and derived parameters online for advanced process control. Smaller bioreactors and more sensors seem to be a contradiction, but by using novel sensor technologies and advanced processing software, we can calculate many more parameters and obtain a better understanding of the bioprocess on a small scale.

Nowadays, standard computers are powerful enough to run these calculations online during the process, so a lot more information is available to characterize processes on a small scale. This information can be used to analyze and optimize the performance of the same processes during scaleup to pilot or production scale.

Dr. Sha: The maturation of computational fluid dynamics for power number (Np) predictions enabled P/V scaleup to be optimized based on theoretical modeling. This allowed companies to reduce cost, increase speed, and improve efficiency on the development of scaleup strategies. Certain manufacturers also started to include torque and Np calculations into their bioreactor control software.

However, it is crucial to validate computational fluid dynamics results or software predictions with mechanical torque-based methods. In addition, single-use bioreactor technologies also helped companies to reduce capital equipment investment cost when developing scaleup strategies.

Mr. Galliher: Scaledown simulation modeling and mapping of the process design space using quality by design (QbD) and design of experiment (DOE) techniques are more frequently employed to reveal potential scaleup weaknesses in the process. This approach is used for both upstream and downstream processes, and it requires small-scale benchtop systems that are carefully designed and operated to accurately reflect the process design space and the performance of large-scale systems. Using these techniques and methods, bioprocessors can design out or avoid nonscalable systems or operations.

Some technology providers have developed improved small-scale systems that more accurately reflect the process design space and performance of large-scale systems. In addition, operating these laboratory systems with raw materials sourced from large-scale suppliers can similarly reveal problematic or variable-quality raw materials that can be designed out of the process.

For large-scale reactor systems, computerized fluid dynamic modeling is frequently employed to reveal poorly mixed areas, mass transfer limitations, or excessively high shear zones in the vessel that could create scaleup problems or limitations. These weaknesses can then be designed out of system design, or for existing systems, the process control parameters can be adjusted to overcome reactor design limitations.

Dr. Hebel: Two stages exist for small-scale characterization and optimization of existing or novel biomanufacturing processes. During screening, researchers are identifying key parameters relevant to the process, such as strains or media components. The systems for this first phase should be simple and highly parallel, such as a full-size incubation shaker with microtiter plates (Multitron, 3 mm stroke).

The subsequent characterization of the hits should take place in full-featured bioreactors (such as Multifors or Labfors) equipped with precise feed pumps and similar instrumentation as the large-scale bioreactors used for production. This allows for the precise control of scale-independent parameters such as µ or soft sensors for qs, effectively enabling scaleup. A powerful bioprocess platform software such as eve® assists the researcher in collecting and interpreting data throughout all stages.

Dr. Kools: A wide range of technology developments have taken place, including the introduction of smaller process development tools, more emphasis on data analysis, and more focus on “processability” of a specific molecule. Several vendors, including MilliporeSigma, have also built their considerable knowledge on scaleup of their products from bench- to large-scale operations into best practices that can be shared with users of the technology. Additionally, regulators are requesting more insight into drug manufacturing processes. Analytical services to characterize or validate filters, resins, and other materials, as well as the product itself, are becoming more and more critical.

Dr. Onraedt: Improvements in process analytics and sensor technologies have given process developers new tools to apply to scaling up. For example, improvements in sensor accuracy and response times make it possible to provide better process data for scaling models. Also, new types of sensor technologies including near-infrared and Raman spectroscopy expand parameters for measuring effectiveness of scaling. Biomass monitors and in-line/at-line metabolite analytics also allow rapid evaluation of scaling experiments.

Of course, with these improvements and new capabilities comes the burden of vastly expanded raw data generation, creating additional complexity in the analyses required to understand and manage scaling issues. At Pall Life Sciences, our team of bioreactor experts are driven to support customers thru the challenges of scaling with the right technologies and process expertise.

Dr. Fenge: Quite significant progress has been made over the last 20 years, simply by gathering more practical experience and understanding which parameters affect bioprocess performance at different scales. Also, much more robust cell lines and media systems have been developed. Regulatory filings today require understanding of critical process parameter ranges and how they affect the process and product quality to ensure that a process is robust. Many companies are adopting QbD approaches and are using chemometric tools to support process understanding.

For this purpose, it is essential to understand the extent to which the scaledown model used during process characterization represents the large process. Also, single-use bioreactors closely based on the well-understood classic stirred-tank principles have been developed in recent years, and cover a range from 15 mL to 2,000 L.

Dr. Shaw: Technological advancements in equipment and analytics have helped drive efficiencies in scaleup. In upstream development, the design of small-scale automated bioreactor systems and the adoption of DOE permits us to screen large numbers of cell clones and test large numbers of media combinations, in relatively short periods of time.

Performance information that encompasses metabolomics data provides a powerful tool for understanding and dissecting protein pathways and can lead to better productivity. The other area I would highlight is the adoption of single-use technologies in the area of process development and pilot laboratories.

Single-use technology allows for easy changeover from one program to another, and for parallel development of multiple programs where stainless-steel or glass systems would otherwise be limiting.

Previous articleDolphin Proteins May Hold the Secret to Protection against Hypoxia
Next articleTapping the Human Gut Microbiome, Part 1