Perfusion bioprocesses have gained appeal from biomanufacturers due to their ability to generate high-density seed trains, produce high protein titers, and sustain high viable cell densities. Perfusion bioprocessing allows for new media to be continuously supplied to the bioreactor while used spent media is harvested. The cells remain in the bioreactor and not in the harvested media, thereby allowing the process to stay in a highly productive steady-state. By continually cycling in new media while pulling out the old spent media, the culture is replenished with nutrients while toxic byproducts like ammonia are removed. Additionally, the produced therapeutic protein can be purified from the harvest in near-real-time through downstream processes connected in tandem. Perfusion processes allow for enhanced flexibility with the type of facility that is used for manufacturing. This flexibility can be realized through the volumetric productivities, scheduling, and the stability of the therapeutic being manufactured. However, perfusion processes may have a higher risk of failure. Therefore, they require tight process controls and have stringent cell media demands compared to conventional batch and fed-batch processes.

A team of researchers from the University of Natural Resources and Life Sciences (Austria) and GE Healthcare published a study introducing a design-of-experiment (DOE) workflow to quickly select the best basal and perfusion media blends to yield high cell-specific perfusion rates (CSPR) and process space-time-yields (STY). The team used a CHO-K1 cell line producing an IgG1 antibody. The DOE study was run in parallel with two types of basal media (CDM4NS0 or ActiPro) and eight different feed supplements (HyClone Cell Boost) mixed to varying ratios in shake tubes. Process analytics (e.g., cell concentration, viability, cell size, glucose, lactate, glutamine, glutamate, ammonia, osmolality, and protein concentration) were taken throughout the study to provide data for regression analysis.

In an initial screening experiment, a batch culture was conducted with feed supplements spiked in. Certain feed supplement combinations corresponded to increases in peak titers by 170% and viable cell density of over 85%. Conversely, the addition of other feed supplements resulted in poor productivities and cell growth, indicating a negative impact of specific components within those feed supplements. Carrying on, the team tested the feed supplements on a semi-continuous small-scale perfusion model over 23 runs for the CDM4NS0 base media and 17 runs for the ActiPro base media. Across both cell lines, the enriched blend cultures maintained both high cell viabilities (>80%), and high CSPRs in the stationary-phase of the cultures (days 5 to 11). Also, mean steady-state titers were increased by 30-40% compared to the basal media-only cultures. Regression analysis with the DOE study data was used to help formulate the final recommended feed supplement levels to be added to the basal media.

The team verified the final performance of the optimized processes in both small scale (10 mL) semi-continuous perfusion models and 500 mL perfusion reactor systems. A 56% and 22% increase in titer was observed in the optimized supplemented small scale CDM4NS0 and ActiPro media systems, respectively. Additionally, there were improvements in the specific productivities and viable cell densities of the cultures too. Along with the increases in productivities, there were hardly any differences between time-correlated analysis of charge variants (acid, main, and basic) and size distributions (aggregates, main, and fragments) from days 1 and 11 of harvested media. There were small differences in the N-glycosylation of the supplemented processes between the days and conditions compared to the control systems where only glucose was supplemented to the basal media. These observations were shared in the larger 500 mL perfusion process models. Additionally, in the STYs of the CDM4NS0 and ActiPro supplemented systems approached 1.2 g/L/d and 0.4 g/L/d, respectively. Both optimized systems approached viable cell densities above 90%, increased titers, and low CSPRs in the 10-30 pL/c/d with no significant changes to the critical quality attributes of the IgG1. Collectively, both optimized systems were developed over less than two months of DOE studies.

As manufacturers continue to look at ways to reduce facility costs and development schedules to run more efficiently, perfusion bioprocesses will continue to gain attention. Using a DOE workflow and regression analysis, this study provided a case study to that other groups may follow towards achieving accelerated upstream process development and media optimization.


By, Glenn A. Harris