DETERMINANTS OF RESISTANCE TO ENGINEERED T‐CELL THERAPIES TARGETING CD19 IN LYMPHOMA

Abstract

Introduction: Anti-CD19 chimeric antigen receptor (CAR19) T-cells have activity in patients with relapsed/refractory large B-cell lymphoma (rrLBCL), but over half of patients ultimately relapse. We applied cell-free DNA (cfDNA) analysis to patients receiving Axicabtagene Ciloleucel (axi-cel) to identify determinants of resistance and characterize molecular thresholds predictive of treatment failure. Methods: We developed a novel hybrid-capture approach allowing evaluation of both circulating tumor-derived DNA (ctDNA) and CAR19-derived cfDNA (Fig 1A). We applied this to 381 plasma, tumor, and germline DNA samples from 64 rrLBCL patients, including prior to and during treatment and at relapse. We evaluated all samples for somatic alterations across 246 genes, as well as quantitative levels of ctDNA and CAR19-cfDNA. Results: The median follow-up for the cohort was 12.5 months, and 55% (35/64) progressed after axi-cel. We identified 100.5 mutations/case, which was similar to a cohort of 136 untreated DLBCL patients (P>0.8). Notable differences included more alterations in TP53 (P = 0.02), EP300 (P = 0.02), SETD1B (P = 0.04), ARID5B (P = 0.04), BTK (P = 0.04), CD79A (P = 0.03), CXCR4 (P = 0.03) and RHOA (P = 0.03) in rrDLBCL, and fewer alterations in CD79B (P = 0.04) and PIM1 (P = 0.03). When considering ctDNA quantity, both pre- and on-treatment levels were prognostic for PFS, with higher levels correlating with adverse outcome (Pre-LD [HR = 1.5, CI = 1.1-1.9], Day 0 [HR = 1.6, CI = 1.2-2.3], Day +7 [HR = 1.5, CI = 1.1 - 2.0], Day +28 [HR = 1.7, CI = 1.4- 2.2]; Fig 1C). In contrast, higher CAR19-cfDNA levels at Day +7 were associated with favorable outcome (HR 0.52, CI 0.32-0.87). CAR19-cfDNA was also correlated with CAR19 T-cell levels by flow cytometry (Pearson r = 0.7, P < 0.001; Fig 1B). When we assessed the effect of specific mutations on outcomes, we observed recurrent emergence and clonal selection of variants at relapse, including mutations in CD19, PAX5 and TP53 (Fig 1D). Finally, mutations in multiple genes were identified as being prognostic for outcome, including CD58, PAX5 and IRF8 (Fig 1E). Notably these include both immune-mediated and target-mediated putative mechanisms of resistance. For example, CD58 encodes a costimulatory T-cell molecule which has been implicated in CAR19 resistance (Majzner et al., Blood 2020), and PAX5 and IRF8 encode transcriptional regulatory elements that are central to B-cell differentiation and phenotype. Conclusions: Baseline and interim ctDNA and CAR19-cfDNA measurements have prognostic significance in LBCL patients being treated with CAR19 T-cells. Additionally, genomic alterations in several genes, including CD19, CD58, PAX5 and IRF8 are associated with inferior outcomes, and thus represent candidate resistance mechanisms that warrant further study with the goal of improving future generations of CAR T-cell therapy.