Building Efficiency Through Continuous Characterization And Connected Bioprocessing

Developing efficient and cost-effective fermentation and cell culture processes is an important part of biomanufacturing, but it can be long and costly. Researchers and process developers need to run tests on a regular basis. Results from these tests help them understand the impact of changing parameters on the system, whether developing a process from scratch or validating changes to existing setups, such as using new cell lines or changing perfusion rates. Similar measurements can be used to monitor a process once it is up and running.

The current process

In the current in-process testing and assessment approach, samples need to be taken from the bioreactors at certain points in the cell culture cycle to answer a number of questions:

• How much of my target molecule do I have?
• What is its quality?
• When can I harvest, based on the balance between quality and quantity?
• Do I harvest when the yields are “most” or “best”?
• How does changing parameters affect the quality and quantity of the target molecule?
  • Cell density, quality, and viability
  • Temperature
  • Media components and nutrient levels
  • Gas levels and mass transfer (oxygen and carbon dioxide)
  • pH and temperature
  • Metabolites

The current approaches to process analytical technologies raise a number of challenges. The initial issue, as with taking samples from any biological system, is the risk of introducing contamination such as airborne bacteria or yeasts. The key issue, however, is the time lag between taking the sample and getting the results.

Samples taken to assess the purity, quality, and yield of biologics need to be purified first, as these not only contain the target biologic but also a mixture of cells, media, nutrients, and off-target proteins and peptides. The separation and purification process takes time, and meanwhile the process in the bioreactor has moved on, which is an issue in both industry and research. If the purification process is complex, the time lag is even greater.

Meeting the challenges

Real-time analytics have potential to reduce the need for offline testing and, therefore, the associated time lag. The aim of development at the National Institute for Bioprocessing Research and Training (NIBRT), in Dublin, Ireland, is to allow testing to be carried out in real time or near real time.

To reduce the time lag, the team at NIBRT is using high-tech approaches such as ultra-high-performance liquid chromatography and high-resolution accurate-mass spectrometry (HRAM). These can monitor a number of different parameters (known as multi-attribute monitoring or MAM) and effectively bring the lab to the bioreactor, rather than requiring samples from the bioreactor to be taken to the lab.

As well as cutting the timelines, NIBRT's approach, based around science and information technology, could increase the targeted information available. This uses the generated Big Data rather than simple offline information to provide both feedback and feed-forward. The increased quality and quantity of data available could be used to support real-time release of production lots.

The system uses Finesse smart vessels and smart controllers. It can take gas and liquid samples via a sterile sampling interface and use noninvasive sensors to measure other parameters. The aim is to have an operational lab-scale version ready for testing in 2018, and data will become available as the project moves forward.

The 24-month project, which is being developed in collaboration with a multinational pharmaceutical company, is funded by the government funding agency Enterprise Ireland and will be completed in mid-2019. Once completed, other systems could be created for other companies as part of collaborative relationships tailored to a company's specific needs.

The system’s benefits

There is potentially a higher capital outlay to set up the NIBRT system, but overall it has potential to reduce costs in research and development by reducing the need for protein purification. By reducing the time lag to near real time and providing data on an hourly basis, the NIBRT system could also speed process development timelines through increasing the understanding of the process chemistry and the impact of changes. This could help a company looking to beat competitors to the market. It could free up the protein purification teams to work on other projects as well.

Additionally, real-time analytics could play a role in the industry to monitor a process once it is developed and thereby reduce the reliance on quality control laboratories to monitor ongoing processes. The increased amount and frequency of data could support rapid decision making, allowing troubleshooting and remediation of an ongoing process. The analytics could also be used to enable increased automation, which in turn could reduce the risk of process contamination.

If feedback is not required as frequently in manufacturing as in research and process development, more than one bioreactor could be linked via a central analytics hub. This includes a redesigned sampling interface that ensures the hub can detect which bioreactor the sample comes from.

Making the change

The biopharma industry can be conservative, reluctant to bring in new and unproven technologies. The key drivers of change for the industry are proven return on investment, especially if it requires additional investment, and impact on speed to market, supported by confidence in its safety and reproducibility. An argument to demonstrate the value of a new process like this one needs to be supported by data from both modeling and the real world.

Data to prove safety and reproducibility in a system could be generated by taking check samples, running them offline, and then comparing the outcomes with results generated as part of the integrated system. Taking check samples on a less-regular basis could also become part of the QA of the system once it is up and running.

The return on investment data, and therefore the process' impact on costs and competitiveness, could be produced initially through financial modeling. The models could look at the costs over an individual process, over a specific period, and over the lifetime of the system. As the system becomes more widely used, these models could be supported by anonymized real-world data.

The future of real-time analytics

Different users need to see different levels of process analytics, from the researchers looking to understand protein chemistry, to the developers who are creating a more efficient biomanufacturing process, to the operators monitoring existing processes in biologics manufacturing.

The goal is to create a user-friendly interface in the style of a smartphone app. This would allow users to see just the level of information they need, while giving them the option to see more detailed data. For example, process developers and researchers would need detailed analytics to see how changing individual parameters affects others. Operators on the factory floor would need a visual interface that highlights trends, shows specifications, and triggers warnings when parameters go outside specified guidelines. And quality assurance teams would need indicators showing the process meets specific compliance needs, which could be modified as regulatory guidelines evolve.

The project was initially developed for batch processing with mammalian cells (Chinese hamster ovary cells) producing antibodies. To meet the changing needs of the industry, it has been designed to be cell- and process-agnostic, and so it has potential be used with other types of cells such as bacteria, algae, and yeast, and with different bioproducts. It could also be adapted to continuous perfusion processing to follow current industry trends, though this would need to deal with higher-density cells and longer processing and growth times.

The bioprocessing industry is moving toward greater integration and automation. This can reduce the risk of contamination and human error and can also allow the use of high-tech approaches without the need for users to learn new skills. The ultimate goal of NIBRT's continuous characterization system is to support the creation of integrated and automated end-to-end upstream and downstream continuous processing. Its analytics system has potential to provide the "blanket of information” required to wrap around and support the technology.