Recombinant CD19 Glycomutant Accurately and Reproducibly Detects CD19-Directed CAR-T Cells By Flow Cytometry

Background

CD19-directed chimeric antigen receptor (CAR)-modified T cell therapy expansion and persistence in peripheral blood has been shown to correlate with responses against B cell malignances (Hay, Blood, 2018; Ayuk, Blood Adv, 2021). CAR-T cell enumeration is typically achieved through quantitative PCR (qPCR) or flow cytometry (FC). qPCR, however, requires techniques that may not be locally validated and, unlike FC, cannot simultaneously address T cell immunophenotype. Existing FC reagents to detect CD19 CAR-T cells are mostly specific for a given CAR sequence or bind non-specifically to the introduced protein (e.g. Protein A). Utilizing glycosylation data from the structure of CD19(Teplykov, Proteins, 2018), we created a stable, biotinylated CD19 ectodomain glycomutant reagent (gmCD19) that binds to the CD19-directed scFv in CAR constructs and can be detected by FC.

Methods

The gmCD19 ectodomain (amino acids 21-227, N138Q) was expressed in 293 Freestyle™ cells with a C-terminal 6x Histidine-AviTag™ to facilitate expression, purification, FC detection (with fluorophore-bound streptavidin) and enable tetramer formation. We used FC to determine limit of detection (LoD), defined as the percent of detectable CAR-T cells when spiked into peripheral blood mononuclear cells (PBMC) at known concentrations. We used a CD19-directed CAR-T cell that co-expresses truncated EGFR (EGFRt) to benchmark gmCD19 and compare our reagent with commercially available detection reagents. This was quantitated by Pearson correlation coefficient and Bland-Altman measure of bias - defined as the mean of differences between tests on the same sample. We also tested reagent stability by assessing for decrease in mean fluorescence intensity (MFI) and the percent of CAR-T cells detected after multiple freeze-thaw cycles of the gmCD19 reagent. We compared CAR-T cell detection by gmCD19 with qPCR from actual patient samples treated with axicabtagene ciloleucel (axi-cel) and collected into three different anticoagulants.

Results

gmCD19 detected as few as 0.25% CAR-T cells by FC and was highly correlated with EGFRt expression (r=0.9993, p<0.0001). When benchmarking gmCD19 with EGFRt expression and comparing gmCD19 quantitation with other commercially available CD19 scFv detection reagents (Acro and BioSwan), gmCD19 showed the least bias (0.04 vs -1.225 and 61.66, respectively), and was the only method that demonstrated statistically significant agreement with EGFRt. Following each freeze-thaw cycle of gmCD19, there was no statistically significant decrease in percent of CAR-T cells detected with only a slight decrease in MFI (~2% decrease per cycle). Detection of axi-cel from patient samples was highly correlated with CAR copy number determined by qPCR, regardless of the anticoagulant in which the patient sample was collected (r=0.9387, 0.9849 and 0.9373 for sodium heparin, sodium citrate and EDTA, respectively) with equivalent coefficients of variation (11.0% vs 11.4% for gmCD19 and qPCR, respectively).

Conclusion

The availability of multiple CD19-directed CAR-T cell products and the importance of monitoring CAR-T cell expansion and persistence in patients undergoing CAR-T cell therapy creates the necessity for an easily applied, stable, and reliable quantitation FC method. We show that our gmCD19 accurately measures CD19-directed CAR-T cells across a variety of CAR-T cell constructs, including commercially available products.