Accelerating perfusion process optimization by scanning non-steady-state responses

Abstract

Perfusion processes provide consistent culture conditions, high productivity and low product residence times. However, process development can be slow due to the 1 week or more required to reach each steady state. The objective of this work was to accelerate process development in perfusion cultures by scanning non-steady-state transient responses to qualitatively predict steady-state performance. The method was tested using a shift in temperature every 3 days, scanned down by steps of 2 degrees C from 37 degrees C to 31 degrees C, then scanned up to 37 degrees C. Higher t-PA concentrations were predicted at lower temperatures, confirmed by subsequent pseudo-steady-state results. In most cases, transient values on the 3rd day were in close concordance with pseudo-steady-state values. To further accelerate process development, transient scanning was applied to small-scale, non-instrumented cultures. Similar results were obtained, although quantitative t-PA values were 15-30 times lower than in high cell density perfusion cultures. The method was further explored by investigating 1 day transient shifts in temperature where more variability was observed, suggesting that the cells were still adapting to the new environment. Nonetheless, the overall response again qualitatively predicted the pseudo-steady-state temperature response. Use of transient scanning in conjunction with pseudo-steady-state verification and refinement of optimal results could reduce process development time to a third or less of comparable steady-state-based optimization.