In vitro CAR-T cell killing: validation of the potency assay

Abstract

For advanced therapy medicinal products, the development and validation of potency assays are required, in accordance with international guidelines, to characterise the product and obtain reliable and consistent data. Our purpose was to validate the killing assay for the evaluation of autologous anti-CD19 chimeric antigen receptor (CAR) T potency. We used CD4 + and CD8 + lymphocytes or anti-CD19 CAR-T cells as effector cells and REH (CD19 +) or MOLM-13 (CD19 -) cell lines as target cells. After co-culturing target and effector cells (1:1 ratio) for 24 h, samples were labelled with 7-AAD, anti-CD3 and anti-CD19 antibodies and the frequency of CD19 + dead cells was evaluated by flow cytometry. In order to verify the CAR-T specificity for the CD19 + target, the co-culture between CAR-T and REH or MOLM-13 at different effector-to-target ratios was scheduled. Moreover, not transduced CD4 + and CD8 + lymphocytes were tested in comparison with CAR-T from the same donor to demonstrate the assay specificity. Linearity and accuracy were evaluated, and established acceptance criteria were compiled for both parameters (r² ≥ 0.97 for linearity and average relative error ≤ 10% for accuracy). Furthermore, the method was considered robust when performed between 23 and 25 h of co-culture, and the intra-assay, inter-assay and inter-day precision was obtained. Finally, in order to verify the inter-analyst precision, the test was executed by three different operators and the intra-class correlation coefficient was > 0.4 in both cases. In conclusion, we consider this CAR-T potency assay as validated and usable in all steps of product development and quality control.

Keywords: CAR-T; Killing; Potency; Validation.

Fig. 1

Experimental design and gating strategy. a Gating strategy to quantify and characterise CAR-T cells. b Seeding conditions for the killing test. c Representative flow cytometric dot plots containing the gating strategy used to identify dead target cells. REH cells were gated based on their CD3 − /CD19 + phenotype, while effector cells were CD3 + /CD19 − . Dead target cells were assessed by selecting CD3 − /CD19 + /7-AAD + cells. d The histogram shows the overlay of background and killing conditions. All the flow cytometric analyses were performed excluding doublets (not shown)

Fig. 2

Dead target cells at different incubation times and E:T ratios. Mean values obtained after co-culture of REH cells with three batches of CAR-T (continuous lines) or CD4 + and CD8 + cells (dashed lines) were reported. Error bars represent the SD

Fig. 3

Assay validation. a Potency induced by different ratios of three batches of CAR-T and target cells. Continuous lines represent the effect on REH cells; the dashed lines represent the effect on MOLM-13 cells. b Histograms represent the mean and SD of target cell death (%) determined by three batches of CD4 + and CD8 + cells and CAR-T cells. c Number of dead target cells obtained by different dilutions of the killing condition. The x-axis shows the dilution (%) with the live REH cells of the CAR-T + REH well. Dashed lines represent three CAR-T batches, and the continuous line represents the calculated mean value. Acceptance criterion: r² ≥ 0.97. d Relation between expected and average measured values expressed as number of dead target cells induced by three different batches of CAR-T. The expected death value was calculated as the multiple of the proportion of sample after dilution and the average measured value obtained from the undiluted sample, plus the multiple of the proportion of added REH and the average measured value obtained from the REH cells only. Acceptance criterion: average accuracy ≤ 10%. e Comparison between the calculated expected value of the target cell death and the measured value. f Dead events measured with flow cytometry (black continuous line) and trypan blue staining (grey dashed line). To allow comparison between flow cytometric analysis and light microscopic counting, we considered all 7-AAD + events. Error bars represent the SD