Manufacturing chimeric antigen receptor (CAR) T cells involves several key challenges, including high manufacturing costs, unsustainable failure rates, long vein-to-vein processes, and lengthy quality-control (QC) and product-release activities. With support from Sartorius, scientists in the laboratory of Professor Qasim Rafiq (University College London) sought to mitigate these challenges and improve potency and cell yield with intensified production. Intensified production requires scalable, automated bioreactor systems that enable seamless transitions of small-scale models that can be applied for process development in robust manufacturing platforms in large-scale models. The team used the Ambr 250 High Throughput Perfusion system to investigate which process parameters impact CAR-T cell growth kinetics, quality, and functionality.

The Webinar

Experimental Set-Up: To test CAR-T cell expansion in fed-batch stirred-tank bioreactors, primary T cells were isolated, activated with Dynabeads technology, transduced with lentiviruses, and pre-expanded for seven days in T-flasks. Then the CAR-T cells were transferred to an Ambr 250 system with unbaffled vessels and a single impeller to allow gentle agitation. The CAR-T cells were expanded for seven days at five different steering speeds (100–500 rpm), with static CAR-T cells as a control. Each group reached 90–100% viability, demonstrating that stirring had no impact on viability. The final CAR-T cell yield reached in the Ambr 250 system was higher than those in static conditions.

The team then explored possibilities for shortening process times. The seven-day pre-expansion phase, including two activations, was shortened to a three-day pre-expansion phase with a single activation. CAR-T cells that were pre-expanded for 3 days had increased growth and higher-fold expansion during the seven-day culture in T-flasks compared with cells that underwent a seven-day pre-expansion. The shorter pre-expansion process time also resulted in higher frequencies of naive and central memory CAR-T cells and reduced expression levels of activation and exhaustion markers. CAR-T cells obtained from both processes were able to specifically kill target cells.

Process Intensification: A design of experiments (DoE) approach based on a range of process parameters was applied to further reduce process times and increase final cell yields. The aim was to investigate the impact of critical process parameters on CAR-T cell expansion, phenotype, and function while optimizing the perfusion process. An experimental space was set up based on 32 conditions in the Ambr 250 system to identify and optimize critical process parameters for efficient CAR-T cell growth. RPMI medium supplemented with serum and serum-free, xeno-free 4Cell Nutri-T T-cell medium were compared along with the impact of perfusion parameters. To test donor-to-donor variability, T cells from three healthy donors were applied.

The analysis revealed some heterogeneity among the three donors and between the two media, meaning that the best operating window for the process differed in each test group. Thus, the optimized perfusion parameter settings identified by the software were determined systematically for each of the respective media. Compared to fed-batch, perfusion improved CAR-T cell expansion, especially in the Nutri-T medium by >4-fold. Viability reached >90% in all conditions. CAR-T cell phenotypes were analyzed with the iQue 3 flow cytometer. Perfusion did not impact the expression of differentiation or activation and exhaustion markers.

Continued seven-day culture from inoculation to harvest resulted in gradual downregulation of early activation and exhaustion markers. It also reduced the amount of media and other reagents required to produce one dose, thus decreasing the cost of goods.

By optimizing the perfusion parameters based on the DoE approach, process intensification in the Ambr 250 bioreactor resulted in a

4.5-fold increase of CAR-T cell yields during the seven-day cultures. These generated CAR-T cells have a preferable phenotype, being naive and central memory T cells and neither activated nor exhausted.

Questions and Answers

What was the highest cell density reached with the perfusion process? The highest density reached in these experiments was 21 × 106/mL in a 250-mL bioreactor.

How can the current process be improved? Automation can be used to decrease the number of manual handling steps and operator interactions, therefore minimizing the risk of errors and contamination.

Acknowledgments

We would like to thank Professor Qasim Rafiq and his team at University College London for carrying out the experiments.

Find the full webinar online at www.bioprocessintl.com/category/webinars.