Robust on-line sampling and analysis during long-term perfusion cultivation of mammalian cells

• A novel platform has been developed to address the unmet need for robust automated on-line sampling and analysis of mammalian cell bioreactors.
• This includes sterile sampling, sample preparation and transport to online analyzers, and communication with process automation systems.
• Robust operation of the automated sampling and analysis platform has been demonstrated in a 102 day perfusion process.
• This higher resolution data set generated by this system can help accelerate process development by intensifying decision making.
• In a commercial facility, this advance can enable successful realization of PAT both as a diagnostic tool and for routine process monitoring.

In an attempt to support robust automated sampling and analysis of mammalian cell bioreactors, an integrated platform, BaychroMAT®, was developed which includes an innovative sterile sampling device, automated sample transport, a sample preparation module, online analyzers, and communication interfaces to process automation systems. The robustness of this platform was verified by applying it to a laboratory-scale perfusion bioreactor that was operated for over 100 days. Both manual and automated samples were collected over the course of the run and a comparison was made for cell density, viability, glucose, and lactate concentrations. The highest variability (14.4%) was seen for cell density estimates while those for viability, glucose, and lactate were 0.7, 12.9, and 8.2%, respectively. In addition, cell density set-point changes were made towards the end of the perfusion culture and the high frequency automated samples provided a higher resolution description of the dynamics of cell density change compared to less frequent manual sampling. Overall, our results indicate stable and robust operation of the BaychroMAT® platform in a long-term perfusion culture. This success should readily translate to shorter duration fed-batch cultures thereby enabling feed-back control based on real-time nutrient measurements.