(695d) Developing an Automated Bioreactor Feedback Control Toolbox for Cell Culture Processes | AIChE

(695d) Developing an Automated Bioreactor Feedback Control Toolbox for Cell Culture Processes

Authors 

Lowe, C. - Presenter, University of Massachusetts-Amherst
Tuozzolo, G., Takeda Pharmaceuticals Inc
Bayer, B., Takeda Pharmaceuticals
Karmiy, L., Takeda Pharmaceuticals
Houle, Z., Takeda Pharmaceuticals
Process analytic technology (PAT) has long held promise to deepen understanding of cell culture processes. Great advancements have been made in extracting additional information through online probes and at-line analytical equipment but in current deployments, that data is often under-utilized. Cell culture processes have long utilized in-process criteria to trigger process actions but still rely on operators to interpret and manually input process changes such as feed delivery, temperature shifts, gassing adjustments, and others. To fully leverage the increased process information provided by PAT technologies, feedback control should be established between at line analyzers or online probes and the bioreactor control systems.

This presentation will describe automated tools developed by Takeda’s Biotherapeutics Process Development team, both with existing biologics infrastructure and emerging PAT tools, and discuss their potential for deployment across development labs, pilot plants, and GMP manufacturing spaces. Closed loop bioreactor feedback control enabled by integration of bioreactor control software (DeltaV), auto sampling infrastructure (MAST or Nova Flex2 OLS), and cell culture analyzers (Nova Flex2) already present in process development laboratories has reduced hands on operator time by >90% for routine sampling and feeding activities. These tools are implementable with existing instruments and infrastructure and have been successfully deployed across 38 bench scale bioreactors in the process development labs and to large scale bioreactors (50L, 200L, and 500L) in the pilot plant. In addition to automation of daily sampling and feeding operations, manual process intervention for temperature shifts, feed initiation, or perfusion rate modifications has been automated using online capacitance probes. A soft-sensor model was developed to translate the online capacitance signal to offline cell counter measurements on the Flex2. This model was implemented into the DeltaV bioreactor control software (1) to directly control perfusion rate in an n-1 perfusion bioreactor and (2) to automatically initiate temperature shift and feed delivery in a fed batch bioreactor process. These advances allow for increased process consistency, by initiating process changes automatically when the culture dictates, regardless of operational constraints, and allows development labs to approximate 24-hour coverage without the need for 24-hour staff. Additionally, the combined implementation of these tools allows for end-to-end hands-off operation of a fed-batch bioreactor where automated features maintain the bioreactor entirely from inoculation to harvest. Collectively this set of automated tools not only allows for reduced operational burden but also greatly reduces the possibility of human error during bioreactor sampling or in interpretation of in process data and action criteria.

Continued development of these automated tools is enabling flexible, fit for purpose deployment across development and manufacturing spaces. While throughput gains and reduced operational resources are needed in the small-scale process development labs, reliability, process consistency, and reduced operational risk are critical for manufacturing spaces. These automation initiatives not only provide immediate benefit to the end users, but also better prepares these spaces for future adoption of advanced process tools as they continue to evolve. Overall, this suite of automated tools allows for flexible implementation to meet the needs of different facilities and provides a foundational backbone for advanced processes that will be enabled by PAT in the future.