Automated cytotoxicity assessment of natural killer cells by flow cytometry

Natural killer (NK) cell-based therapies are emerging as highly promising candidates for cancer treatment, but their development and quality control depend on robust assessment of key critical quality attributes, particularly their cytotoxicity. Despite the availability of various approaches for assessing cytotoxicity, existing techniques often suffer from high data variability and show limited reproducibility. We compared commonly used approaches for NK-cell cytotoxicity assessment, including calcein release, lactate dehydrogenase release, and flow cytometry (FCM)-based analysis, using NK-92 effector and GFP-labelled K562 target cells. Our results provide insight into the shortcomings of these methods, as well as problems resulting from unharmonized evaluation criteria and limitations of endpoint measurements, which are commonly applied in literature. We selected FCM as the most suitable platform for standardized evaluation and developed an automated gating workflow for cytotoxicity analysis. The automated workflow was benchmarked against three independent manual evaluations to assess agreement, bias, and performance. The optimized workflow showed agreement with manual analysis while remaining essentially unbiased. In addition, automated analysis reduced evaluator dependence by producing deterministic outputs from identical input data. Further comparison revealed directional bias in manual gating, indicating that a relevant portion of measurement variability arose from manual evaluation, rather than biology alone. We present autogating as a fit-for-purpose, automated FCM-based strategy for NK-cell cytotoxicity evaluation that preserves agreement with manual analysis while improving standardization and reproducibility, thereby providing a practical route toward more harmonized cytotoxicity testing in cell therapy workflows.