Current methods to assess T cell function use labels that prevent non-destructive quality control of T cell infusions. Here, we use autofluorescence imaging of NAD(P)H and FAD, co-enzymes of metabolism, in quiescent and activated T cells for label-free, non-destructive determination of T cell activation state and subtype. Logistic regression models achieved 97-99% accuracy for classification of T cell activation, and random forest models of achieved >97% accuracy for four-group classification of quiescent and activated CD3+CD8+ and CD3+CD4+ T cells. These results indicate that NAD(P)H and FAD imaging is a powerful method for label-free, non-destructive quality control of T cell infusions.
|