Options
Evaluation modellgestützter Design of Experiments-Methoden zur Auslegung biopharmazeutischer Prozesse
Citation Link: https://doi.org/10.15480/882.4655
Publikationstyp
Doctoral Thesis
Publikationsdatum
2022
Sprache
German
Advisor
Referee
Title Granting Institution
Technische Universität Hamburg
Place of Title Granting Institution
Hamburg
Examination Date
2022-06-10
Citation
Technische Universität Hamburg (2022)
The growing demand for biopharmaceuticals and the need to reduce manufacturing costs are increasing the pressure to develop productive and efficient bioprocesses. Conventional Design of Experiments (DoE) methods are mostly used in process development and optimization. Being an exploratory approach, DoE requires extensive expert knowledge of the investigated factors and their limits, and often results in several time-consuming and costly series of experiments. In contrast to conventional DoE, in model-assisted DoE (mDoE) the recommended experiments are simulated using mathematical process models to evaluate the experimental space. This enables a well-defined experimental space, since changes can be made without additional experimental effort. Recommended experiments are thus implemented in a significantly reduced number. In addition to the significant reduction in the number of experiments, the use of mathematical process models is now seen as a sustainable part of a knowledge-driven bioprocess development strategy.
However, mDoE methods have so far only been applied in the biopharmaceutical field for process design in the presence of broad available data. These processes are classified to late-stage process development, although it is currently unclear to what extent mDoE concepts can be applied to the early-stage of process development. Nevertheless, especially in early-stage processes there is a need for suitable methods for knowledge-driven process design. Therefore, the basic hypothesis of this dissertation is that mDoE methods are applicable to these very processes. For this purpose, early-stage processes for different cell lines were investigated and evaluatetd.
Based on a broad data set for CHO DP12 cells, the components for the successful implementation of an mDoE were defined and integrated into an mDoE workflow. For this purpose, experimental designs were programmed and implemented in an mDoE toolbox. Using adherent-growing cells (mesenchymal stem cells, hMSC-TERT and mouse fibroblasts, L929), the utility of the mDoE workflow was investigated using different developmental stages. The cell behavior of adherent-growing cells was model-assisted as well as uncertainty-based analyzed, and a mathematical process model was established. Thereby, the cultivation conditions of L929 cells and hMSC-TERT were successfully adapted for cultivation in shake flasks. Thus, cultivations as well as bead-to-bead transfer in shake flasks were successfully implemented for the first time. For L929 cells, an 8-fold increase in cell yield was achieved by reducing the initial microcarrier concentration. When adjusting the bead-to-bead transfer of hMSC-TERTs, a 6.3-fold increase in cell concentration was achieved. Finally, correlations were detected at different scales, allowing scale transfer. Overall, the mDoE workflow was successfully applied to the expansion of various cell lines, analyzing cell behavior, increasing cell yields, and optimizing boundary conditions for mDoE-methods. The knowledge-based application of mDoE methods thus offers also in early-stage processes the possibility to design expansion processes and produce clinically relevant cell numbers.
However, mDoE methods have so far only been applied in the biopharmaceutical field for process design in the presence of broad available data. These processes are classified to late-stage process development, although it is currently unclear to what extent mDoE concepts can be applied to the early-stage of process development. Nevertheless, especially in early-stage processes there is a need for suitable methods for knowledge-driven process design. Therefore, the basic hypothesis of this dissertation is that mDoE methods are applicable to these very processes. For this purpose, early-stage processes for different cell lines were investigated and evaluatetd.
Based on a broad data set for CHO DP12 cells, the components for the successful implementation of an mDoE were defined and integrated into an mDoE workflow. For this purpose, experimental designs were programmed and implemented in an mDoE toolbox. Using adherent-growing cells (mesenchymal stem cells, hMSC-TERT and mouse fibroblasts, L929), the utility of the mDoE workflow was investigated using different developmental stages. The cell behavior of adherent-growing cells was model-assisted as well as uncertainty-based analyzed, and a mathematical process model was established. Thereby, the cultivation conditions of L929 cells and hMSC-TERT were successfully adapted for cultivation in shake flasks. Thus, cultivations as well as bead-to-bead transfer in shake flasks were successfully implemented for the first time. For L929 cells, an 8-fold increase in cell yield was achieved by reducing the initial microcarrier concentration. When adjusting the bead-to-bead transfer of hMSC-TERTs, a 6.3-fold increase in cell concentration was achieved. Finally, correlations were detected at different scales, allowing scale transfer. Overall, the mDoE workflow was successfully applied to the expansion of various cell lines, analyzing cell behavior, increasing cell yields, and optimizing boundary conditions for mDoE-methods. The knowledge-based application of mDoE methods thus offers also in early-stage processes the possibility to design expansion processes and produce clinically relevant cell numbers.
DDC Class
570: Biowissenschaften, Biologie
Loading...
Name
Dissertation-KimBeatriceKuchemüller.pdf
Size
33.07 MB
Format
Adobe PDF