What’s Beyond the Bioprocess Automation Starting Line?

Automation drives the production of higher quality and more consistent biologic drugs and regenerative therapies at reduced costs of goods (CoGS) – and with higher flexibility and faster time to market (1, 2). And when it comes to automation of bioprocesses, process analytical technology (PAT) and advanced data analytics are crucial enablers because of the need to measure critical process parameters at all stages.
 
The main advantages of bioprocess automation can be summarized as:

• Consistency in product quality and quantity; variations of critical process parameters (CPPs) are reduced and process robustness is increased (see Figure 1).
• Fast and predictive up- and down-scaling; a well characterized and monitored process alongside scalable hardware significantly reduces the cost and effort of scaling, as variations can be accounted for in an automated and predictive fashion.
• Reduced risk of lost batches and increased process safety; operator errors and contamination through manual sampling are reduced. The timely identification and correction of process irregularities reduces the risk of lost batches.
• Operators are free to work on tasks that cannot be (easily) automated.
• Cell variation arising from different sources – as could be expected from different patients in personalized medicine – can be managed through a process that is flexible and able to dynamically adjust to wide variations – through PAT – to assure high process consistency irrespective of the starting material.

There are a number of key technologies that will help the industry to achieve more automated processes. For example, we believe that spectroscopic techniques will become more abundant in both upstream and downstream bioprocessing because they can be used to perform label-free, online measurements of several analytes, cell properties and product quality attributes – and replace offline measurements during the bioprocess. We envision the use of a combination of different spectroscopic techniques, including NIR, Raman and UV-Vis, to achieve online measurements. That said, there will also be a continuing need to use and further develop other technologies, such as sensors for bio-capacitance and tools for the measurement of nutrients/metabolites; here, spectroscopy cannot provide solutions. To propagate the use of spectroscopy in GMP, biomanufacturers will need a combination of measurement technologies that can act as cross checks.
 
Advanced data analytics must come hand in hand with the application of sophisticated PAT tools. Together, they can have a high impact on commercial processing because measurements can be moved forward in the process to the point of controllability. By using process fingerprints, the state of the process can be assessed at any time. Furthermore, through real-time univariate and multivariate process monitoring, data can be used to simulate and model process design and control – and ultimately lead to prescriptive analytics of product quality.
 
We also believe that flexible, automated skids are an important technological development, particularly for downstream processes. Flexible, automated skids, capable of handling different types of unit operations, all based on S88 compliant recipes (AINSI/ISA-88 is a standard addressing batch process control) would make it possible to run standardized and automated processes in facilities using the “ballroom” concept.

Solutions but also challenges

When discussing the automation of bioprocessing, we also need to evaluate technical feasibility and consider a cost–benefit analysis. And there are also regulatory, logistics and safety issues that need to be solved before automation can really be adopted widely in the biopharma industry. Though there are few applications in the process of biopharmaceutical drug production that would not benefit from automation, we do not envision a high degree of automation added into existing pipelines; for example, well-established fed-batch processes. Unless, of course, the automation adds a significant improvement to the process – as we have seen for automated temperature shifts at a certain viable cell density. Other processes may not benefit sufficiently from automation to warrant a change – perhaps areas where manual interference is already limited; for example, dead-end filtration. Finally, there are also processes that will be challenging to automate – product quantification of a target protein with a background of many other host cell proteins could be one such example.
 
There may also be compatibility and infrastructure challenges. The seamless integration of process equipment and process skids into an automated process, especially when considering flexible manufacturing facilities, is an issue. Communication between competitor solutions is not always guaranteed – an issue of insufficient established standards. Another challenge is presented when aligning the process automation concept of a supplier to the facility automation concept in terms of environmental monitoring, building monitoring and the required level of integration into resource planning systems.
 
And let’s not forget regulatory challenges. Some concepts of modern automation technologies and sensor technologies are not yet covered by regulatory guidelines. The latter issue is especially true for multivariate data analysis, which takes all available data and integrates them into a fingerprint. The adoption of such batch fingerprinting concepts needs to be considered by regulatory bodies. The same questions arise for multi-analyte sensors that are based on computational AI models – as is the case with spectroscopy, for example. How do we validate a model for the use of GMP? What are the characteristics of a “good and robust” model? We should also consider the definition of “a batch” for continuous processing. Regulations that were initially established for a two-week batch process have to be adjusted to processes that can potentially run for months without interruption. Regulatory agencies are well aware of the challenges that come with modernizing the industry, but are open and cooperative to new concepts coming from technical advances in the field of automation, PAT and advanced analytics, as evidenced by the creation of the FDA’s Emerging Technology Program (3, 4).
 
When considering greater use of data, companies must also consider concerns around IT and data integrity. A comprehensive automation strategy for an entire bioprocess, and potentially an entire production site, requires connectivity of all components and a centralized control unit. However, this requires data sharing and access that implies safety risks. We experience reluctance among our customers to adopt new technologies, such as cloud computing and wireless communication of PAT components. And we are convinced that the task of meeting the requirements of next generation manufacturing in terms of hardware, software, data analytics and infrastructure are too demanding and complex to be addressed by just one supplier. To overcome these challenges and to guide developments, we need the collaboration of several industries and a frank and open dialogue with customers.

Change is coming...

In the near future, we expect to see wider adoption of analytics in GMP, such as spectroscopy for metabolite control and bio-capacitance for viable biomass. We also foresee that multivariate data analysis (MVDA) and design of experiments (DOE) will be adopted by more users. Greater standardization will allow a real “plug and produce” scenario in a (multi-product) facility setup. And the field of hybrid modeling, where statistical and deterministic modeling principles are combined, will advance within the biopharmaceutical industry, further improving process understanding and simulation. Within systems biology, for example, these approaches are starting to be applied to enhance the production of cell lines commonly used in biopharmaceutical processing in a pragmatic way (5,6).

Five years from now?

We expect that modern facilities will apply intensified and continuous processing with advanced automation. They will be using state-of-the-art automated process batch management and S88 compliant batch recipe control functionalities, as well as plantwide visualization and electronic batch records. Furthermore, sophisticated analysis tools, such as HPLC and mass spectrometry, will be automated and integrated into the bioprocess. Together with an increased use of data science, quality-by-design approaches can be applied, allowing real-time release testing of product quality based on batch fingerprinting. Robotics will take over tasks, which cannot be automated otherwise, such transporting materials to and from the production location.

Ten years from now?

The far-future vision is highly influenced by the Industry 4.0 approach and related concepts, such as machine learning and the Internet of Things. We will see a fully automated, continuous bioprocessing pipelines that require no operator interventions. Processes can be monitored and controlled remotely. Every process will have a digital twin that can be used for process simulation and prediction. More and different data will be gathered and will reside in the cloud, where data analytics can be applied easily to improve processes, regardless of manufacturing location.