Multi-parameter flow cytometric analysis of therapeutic EBV-specific T cells: visual assessment of product development and differentiation using dimensionality reduction with t-SNE.

Abstract

Background & Aim Adoptive transfer of Epstein-Barr virus (EBV)-specific T cells (VST) has demonstrated proven efficacy for therapy of EBV-related malignancies such as Post-Transplant Lymphoproliferative Disease (PTLD). Clinical manufacture of virus-specific T cells requires extensive quality testing and characterisation, and flow cytometric analysis is fundamental to the assessment of cell product quality and potential efficacy. A broad range of surface and intracellular markers can be used to identify activation, differentiation, memory and functional status of therapeutic virus-specific T cells (VST), but flow cytometry can also identify bystander and contaminating cells. The principal issue with a large panel of markers comes in the challenge of analyzing such high dimensional data effectively, and making meaningful qualitative assessments. Methods, Results & Conclusion We have developed two flow cytometric panels of 16 parameters in total for assessment of our recently-established GMP-compliant EBV-VST bank, which are used to characterize source leukocytes, VST throughout culture, and final product for release. The release criteria (RC) panel covers lymphocyte lineages and T cell memory subpopulations. The in-process control panel assesses both surface markers and intracellular cytokines (IFNγ, TNFα and IL-2) to identify activation and differentiation status of different populations in response to EBV peptide stimulation. We have applied t-Stochastic Neighbour Embedding (t-SNE) as a dimensionality reduction tool to assess all parameters together. By first clustering cellular populations into CD8/CD4 and memory types based on CD45Ro/CD45Ra/CD62L expression, a single image can then be generated incorporating the distinct cytokine profile of each subpopulation present. In this way, t-SNE provides a simple and visual mechanism for assessing the quality of the final product. As an in-process control during manufacture, t-SNE analysis can also identify lines that deviate from accepted parameters, providing a go/no go point in a manufacturing run that would otherwise fail to meet final product release criteria. The visualisation of cytokine response within VST cell subpopulations provided by t-SNE clustering is a powerful tool for dissection of the T cell compartment using a single image. This visual representation allows clear and powerful comparison of cell population heterogeneity which makes it a practical qualitative evaluation which could be applied to other cell therapy products.