Stability of Jurkat cells during short-term liquid storage analyzed by flow imaging microscopy

The viability of cell-based medicinal products (CBMPs) is a critical quality attribute and must be assessed throughout the product lifecycle to contribute to a safe and potent drug product. In this study, we investigated the impact of short-term liquid storage conditions, encountered during manufacturing of CBMPs, such as holding times outside cell culture conditions and medium composition, on cell viability. As a model for T cells Jurkat cells were used and stored in different storage media for up to 24 h outside a freezer and outside of cell culture conditions. The effect of storage in different storage media, i.e., cell culture medium or phosphate buffered saline (PBS), dimethyl sulfoxide (DMSO) at different storage temperatures as well as the impact of pH on the cell viability was assessed. The viability of the cells was assessed by (i) flow cytometry with an Annexin V and CalceinAM staining or (ii) machine learning tools, leveraging the morphological information of flow imaging microscopy images. Cell images obtained by flow imaging microscopy were analyzed with both the ParticleSentry^AI imaging software and a convolutional neural network (CNN) for fast and semi-automated viability assessment. Throughout storage conditions similar to those during processing of CBMPs, a decrease in cell viability was observed over time for all conditions based on Annexin V and CalceinAM staining. Additionally, we observed a damaging effect of DMSO over time, whereas this effect was more pronounced at room temperature compared to refrigerated temperatures. The ParticleSentry^AI software was useful to detect qualitative differences in the cell viability as a shift towards non-viable cells could be observed throughout storage based on flow imaging microscopy images without prior sample preparation. Viability determination based on the CNN underestimated cell viability when compared to the flow cytometry assays, however the same trends were determined. In summary, non-frozen storage of CBMPs should be kept to a minimum. However, standing times throughout the manufacturing of CBMPs as well as during in-use cannot be completely avoided and therefore, the optimal storage conditions have to be carefully evaluated. Additionally, further analytical development is needed to implement machine learning tools suitable for reliable viability quantification without additional sample preparation such as staining.