(327b) Hybrid Modeling with Phase Contrast Microscopy Images for the Design of Mesenchymal Stem Cell Cultivation Processes

Mesenchymal stem cells (MSCs) are promising cell source for regenerative medicines due to their therapeutic properties. In anticipation of the demand growth of MSCs, quality manufacturing by process design is required to produce clinical-grade cells at industrial scale [1]. In static cell culture, the spatial distribution of seeded cells affects their growth due to contact inhibition. In fact, different initial cell distributions were investigated and the work showed that uneven distributions significantly slowed down cell growth, especially towards the end of cultivation [2]. Recently, our group developed a kinetic model for MSC cultivation processes to incorporate system dynamics and variabilities including contact inhibition [3]. However, the way of predicting the effects of initial spatial distribution of MSCs on their subsequent growth is yet to be presented.
In order to improve our previous model, this work presents an integration of both mechanistic and data-driven modeling approaches for designing MSC cultivation processes considering initial cell distribution using phase contrast microscopy images. The hybrid model was built in three steps: (i) First, three initial distribution cases were experimentally investigated to compare their effects on the resulting growth dynamics. (ii) Second, the spatial distribution was quantified by phase contrast microscopy images to define the standard deviation of the cell number fraction for the 2D cultivation. (iii) Third, a new parameter was integrated with our previous mechanistic model [3] by formulating a novel growth inhibition term for the Monod equation to represent the decrease in growth rate. The developed model showed a good agreement with experimental results with sufficiently small values of the normalized root mean squared error for various variables (i.e., attached cell numbers and metabolite concentrations). The new model was then used in a dynamic simulation of MSC cultivation, which enabled the determination of appropriate time for cell harvesting based on their initial distribution. The developed hybrid model can serve as the basis for quantitative decision-making in the design of MSC cultivation processes.

References
[1] Lipsitz, Y. Y., Timmins, N. E. & Zandstra, P. W. Quality cell therapy manufacturing by design. Nat. Biotechnol. 34, 393–400 (2016). https://doi.org/10.1038/nbt.3525
[2] Zygourakis, K., Markenscoff, P. & Bizios, R. Proliferation of anchorage-dependent contact- inhibited cells. II: experimental results and validation of the theoretical models. Biotechnol. Bioeng. 38, 471–479 (1991). https://doi.org/10.1002/bit.260380505
[3] Hirono, K., A. Udugama, I., Hayashi, Y., Kino-oka, M. & Sugiyama, H. A Dynamic and Probabilistic Design Space Determination Method for Mesenchymal Stem Cell Cultivation Processes. Ind. Eng. Chem. Res. 61, 7009–7019 (2022). https://doi.org/10.1021/acs.iecr.2c00374