De-risking and Simplifying the Bioreactor Scaling Process

Bioreactor scaling is a complicated procedure that must take into account how various bioprocess parameters interact. Relying on a spreadsheet to manage a few single factors using a trial and error approach isn’t the optimal way to create a robust, scalable process.

^{This article is posted on our Science Snippets Blog}

In addition to ensuring consistent quality, biopharma companies need to know that that the production process of their drug or biological product can scale effectively and efficiently from initial development to manufacturing.  Scaling a bioprocess up or down for clinical trials or mass production relies on being able to effectively transfer an entire process from one bioreactor environment to another while maintaining critical quality attributes (CQAs), product quality and quantity at the new scale.

Numerous interacting variables can affect the process, and it is not easy for lab scientists to understand how the various factors may interact, or how to maintain a well-documented process that delivers the critical quality attributes.

Using tools that help ensure optimal scaling can make the process more effective. This includes elements such as bioreactors designed with geometric similarity and predictive software to support scaling without needing deep knowledge of scaling algorithms or modeling. Data analytics tools combined with well-designed modular elements, such as vessels with consistent bag-height-to-bag-diameter ratios, and consistent mixing and oxygen transfer rates between volumes can make scaling more efficient and reliable.

Scaling Affects Critical Process Parameters

The goal of any scaling effort is to transfer an existing process from one scale to another and create reproducible results in a target scale. For example, this can mean creating a scaled down model in milliliter volume multiparallel bioreactors, or scaling up from process development to commercial production.

One of the most challenging aspects of bioreactor scaling  is determining which operating parameters are more important for scaling, understanding the interactions, and determining the design space. De-risking the scaling process means gaining an understanding of the interactions between the bioprocess factors and being able to control them in an effective way.

A quality by design (QbD) approach to bioprocess development relies on using a statistically validated methodology for process design. The same principals apply when it comes to creating a process that will scale well. The key is to create a design space large enough to allow for scale-up or down, while ensuring that Critical Process Parameters (CPP) remain controlled and match what was expected at the starting scale.

Moving from pilot to production scale with minimal deviations to the Critical Process Parameters (CPPs) and Critical Quality Attributes (CQAs) require significant process understanding and expertise. Bioprocess scientists want to simplify scaling, make it more accessible to the broader team, and achieve it quickly and with confidence.

^{Scaling from one size bioreactor to another requires the ability to replicate the biological, physio-chemical and mechanical|structural properties of the cellular micro-environment when transferring from one volume to another.}

Factors Influenced by Scaling

Whether you need to scale up or down, being able to replicate the biological and physiochemical properties of the cell culture environment between bioreactors is the key to any scaling effort. The challenge becomes identifying the criteria that are most important to that process, and prioritizing those which have the most impact on the cell culture.

The complex nature of the bioreactor environment means you have a large number of factors to consider that could have an influence on the quality and output. The way in which you control the bioreactor will influence some of these factors, while others may be more impacted by scale and volume.

Some of the factors that have to be considered when scaling up or down include:

• Vessel geometry (impeller diameter, tank aspect ratio)
• Gas velocities  (superficial gas velocity, gas exit velocity and shear stress)
• Volumetric gas flow rate (unit volume of liquid (QG/V)
• Power per (unit) volume  (energy transferred from the stirrer to the liquid)
• Mixing time (time to achieve homogeneity in the liquid)
• Tip speed  (how fast the impeller tip is moving through the liquid)
• Gas transfer (by the gassing rate, sparger type, and pressure)
• kLa (transfer of oxygen from the gas to the liquid phase)

Design Space Considerations

The scaling process for each bioreactor is affected by the design space, which is impacted by a number of physical and biological parameters. As shown in the diagram below, the boundaries are made up of factors such as the shear limit, heterogeneity and foam bubble formation,  which are influenced by the power input, gassing, oxygen limitation and CO₂ accumulation.

How do you determine which are most important, and how can you expand your design space to ensure a more robust process during scale up?

When changing scale, your priority must be to avoid creating significant impacts or changes to the key performance indicators (KPIs) and the CQAs, which affect product safety and efficacy. To be successful, you must replicate the cell environment as closely as possible between the different bioreactor scales. In addition, you want to minimize the number of runs to be carried out at a large scale to help keep costs down.

Effective scaling of bioprocess parameters is critical to ensure optimal process performance and productivity. BioPAT^® Insights software has been designed to facilitate scaling using multiple scaling parameters simultaneously for supported Sartorius bioreactors.

SinYee Yau-Rose, Product Manager for Ambr^® Software Applications, Sartorius

Conventional Scaling Has Limitations

A conventional approach to scaling typically uses one main common scaling parameter for scale conversion, alongside one or two secondary parameters, which may differ at each scale. For example, one main parameter – such as specific power input (PPV), volumetric mass-transfer coefficient (kLa), or tip speed – is taken in isolation and matched across all scales. This is a simple estimation that doesn’t consider all potential variability.

Some of the limitations of this approach are:

• The bioreactor parameters, such as volume, can vary significantly during the process, which will impact scaling factors such as PPV
• Fixing one factor can neglect the impact of other factors
• Multiple scaling factors are intrinsically linked
• The sensitivities are different between scales

Alternatively, scaling is carried out by a few experts on the team, or consultants, using advanced, often custom-built tools, such as spreadsheets. These tools contain models based on numerous in-house physical characterization experiments or calculated based on bioreactor data supplied by manufacturers. And they may not be visible across all teams, or upstream environments, which limits process knowledge and scaling consistency.
Another, more modern approach, is to use data analytics, along with equipment and technologies designed to support scale-up to improve your success.

Bioreactor Scaling Is Complex

Scaling bioreactors can be highly complex since you are looking at numerous factors across multiple scales with a wide range of settings. It often requires a fine balance between factors selected and output, particularly with large changes in scale.

Scale dependent parameters like agitation speed and gassing rate are difficult to compare across scales without first transforming them into meaningful scale independent parameters, such as specific power input or kLa.

Without having a way to manage multiple variables, you may have to compromise between the factors to achieve efficient scaling, and you may not have a way to consider the entire design space.

Choosing the optimal ranges at the small scale is important because it will affect the risks associated with scale-up to pilot or manufacturing scale. Getting scaling wrong can mean undertaking more engineering runs at each scale to optimize the process, which is costly and time consuming.

Plus, if you want to adapt a process, or simulate how certain scenarios will play out, you need characterized data to scale, compare and understand the processs. You may not always have time to perform the experiments needed to obtain the data. Nor have the expertise in house to turn the data in valid models at different scales. Characterizing bioreactors takes time, money, and effort. How can you reduce the risks and create better scale-up models?

Deriving Scale Models from Characterized Data

A more modern approach to bioreactor scaling uses models derived from empirically characterized bioreactor data to identify the risks and critical process parameters for each specific bioprocess. A data driven software solution for predictive bioreactor scaling calculations allows users to easily visualize the risks and quality outcomes predicted for various design space and process parameter settings.

Some advantages of using a data-driven, risk-based approach:

• Scaling is simplified
• Scaling is accessible and approachable for non-expert users
• Scaling is more consistent across multiple teams within an upstream environment
• Scaling is improved with risk-predictions and visualization

By establishing the transfer of set points for both scale down and scale up, scientists can ensure than the results obtained in the development scale will be achievable in the subsequent stages of the development workflow – all the way to manufacturing.

The novel solution from Sartorius that makes this possible is called BioPAT^® Process Insights. It combines scaling functionality with extensively characterized bioreactor data to deliver simplified, consistent scale up between Sartorius Ambr^®, Univessel^® Glass, Univessel^® SU, and Biostat STR^® bioreactors.

BioPAT^® Process Insights Reduces Scaling Risk

BioPAT^® Process Insights is a new type of PAT designed specifically for bioreactor scaling. The software application eliminates the need for complex formula and statistics to scale production. Instead, the software tool can leverage small amounts of information into successful scaling. It can be combined with techniques like Design of Experiments (DoE) and Multivariate Data Analysis (MVDA) to generate large datasets and improve overall scaling outcomes.

“BioPAT^® Process Insights software simplifies bioreactor scaling to reduce risk and development timelines.”

SinYee Yau-Rose, Product Manager for Ambr^® Software Applications, Sartorius

The simultaneous multi-parameter and multi-scale approach de-risks bioprocess scaling early on in the upstream bioprocess scaling workflow. Early risk identification helps minimize failure rates due to scalability issues.   

BioPAT^® Process Insights provides a platform for users to simulate process outcomes and adapt to different scaling scenarios without performing multiple experiments at a significant time and financial cost. These advances promote speed, agility, and quality, helping facilities to remain competitive in the dynamic biopharmaceutical market.