Detection of CAR-T Cell Persistence with Digital Droplet PCR: Correlation of In Vivo Expansion with Clinical Outcomes in a Cohort of B-Cell Lymphoma Patients

Introduction: In vivo CAR-T cell expansion is of central importance to clinical response and toxicity in patients with hematologic malignancies. In clinical trials, the area under the curve (AUC) within the first 28 days of infusion was considered a valid tool to describe the concentration-time profile of CAR-T cells in peripheral blood (PB). However, several longitudinal assessments performed by either flow cytometry or quantitative real-time polymerase chain reaction (qPCR) were required. This study evaluates the clinical utility of assessing CAR-T concentration by digital droplet PCR (ddPCR) in patients with non-Hodgkin lymphoma (B-NHL).

Methods: Patients with B-NHL ddPCR test collected on day 28 ± 7 post-commercial CD19 CAR T infusion, and clinical outcome data was extracted retrospectively. In this ddPCR assay, the PCR primers target the mammalian expression vector present in the CAR constructs as well as the ATP2B4 reference gene. The CAR-T and reference gene signals are segregated by differential fluorescence signals of tagged probes. The number of CAR constructs and reference gene signals per microliter (uL) are calculated from sample droplets and data fitting to a Poisson distribution. The assay sensitivity is ~0.003% of CAR construct per reference gene.

Results: Between October 2020 and February 2022, 66 patients (62% male) with diffuse large B cell lymphoma (DLBCL, 73%), Burkitt lymphoma (1%), mantle cell lymphoma (MCL, 24%) and follicular lymphoma (FL, 12%) received axicabtagene ciloleucel (83%), brexucabtagene autoleucel (12%), or tisagenlecleucel (5%), with fludarabine/cyclophosphamide (49%) or bendamustine (51%) lymphodepletion and had testing within the timeframe. Bridging therapy was administered in 82%, and elevated serum LDH prior to CAR T was present in 31%. With a median follow-up of 4.5 months (IQR 3.1-6.2 mo), the day 28 overall response rate was 80%, complete remission 57%, and progressive disease 18%. Median CAR copies/microL was 2.3 (range 0-474); of those with a detectable value, the median was 3.57 copies/microL. Excluding the 7 patients who had progression at the same time the sample was drawn, day 28 landmarked 3- and 6-month progression-free survival (PFS) for patients with non-detectable (22%) vs. detectable (78%) ddPCR construct was 31% vs 81% and 31% vs 73%, respectively (p=0.012, Figure 1B). Three- and 6-month PFS for non-detectable (22%), detectable <= 3.57 (41%), and detectable >3.57 (37%) was 31%, 77%, 86% and 31%, 62%, and 86%, respectively (p=0.039, Figure 1A). Univariable analysis of factors predicting detection of CAR T construct by ddPCR at Day 28, including age, sex, gender, elevated LDH, lines of therapy, bridging, CAR T construct, grade 2+ cytokine release syndrome (CRS), and grade 2+ ICANS was significant only for grade 2+ CRS. Patients with grade 0-1 CRS (44%) vs grade 2+ CRS (56%) had a median of 0 (range 0 - 474) vs 4 (0 - 307) CAR copies/microL (p<0.001).

Conclusion: Assessment of PB CAR-T concentration performed by ddPCR at a single timepoint reliably predicted CD19 CAR-T cell efficacy in patients with B-NHL, with detectable CAR construct at day 28 predicting better 6-month PFS. Further research is needed to define the best timepoint for testing and the impact of using strategies to mitigate CRS and/or ICANS.