Abstract 3502: Short-term cytolytic assays are not predictive of chimeric antigen receptor anti-tumor activity mediated by some T cell subsets

Abstract

Abstract Clinical trials of T cells genetically engineered to express chimeric antigen receptors (CAR) are currently underway for multiple malignancies with encouraging results. However, much work remains to be done to optimize CAR T cell therapy with a pressing need to define those elements that result in durable antitumor effects. We have been studying a CAR consisting of a scFv directed against CD19, TCRzeta, and the CD28 co-stimulatory domain, against pre-B acute lymphocytic leukemia (ALL). Whereas CD8+ CD19-CAR T cells readily induce cytotoxicity against CD19+ leukemias in vitro and in xenograft models, other CD19-CAR expressing subsets show antitumor effects that are not predicted by short term cytotoxicity assays. Unselected PBMC, as well as immunomagnetically enriched CD8+ and CD4+ CD19-CAR T cells were generated by retroviral transduction of anti-CD3/CD28 bead activated T cells in the presence of IL-2. In a 4-hour 51Cr release assay, both unselected PBMC and enriched CD8+ CAR T-cells demonstrated similar potent cytotoxicity, while CD4+ T cells mediated no significant cytotoxicity in this assay. CD19-CAR expressing enriched PBMC, CD8+ and CD4+ T cells were then administered to NOG mice 3 days after inoculation with NALM6-GL (human ALL cell line modified to express GFP and firefly luciferase). As demonstrated using bioluminescent imaging, mice receiving the same numbers of CD8+ CAR T cells, CD8:CD4 50:50 mix, and CD4+ CAR T cells eliminated NALM6-GL within 2, 3, and 4 days, respectively, demonstrating unexpected clearance of ALL by CD4+ CAR T cells alone. This result demonstrates that CD4+ CAR T cells mediate significant leukemic killing in vivo but that this occurs with slower kinetics than for CD8+ CAR T cells and is not captured by short term in vitro assays. We postulate that CD4+ CD19-CAR mediated lysis involves non-granule mediated killing, and we are currently investigating these mechanisms. Similarly, CD19 CAR T cells expressing CD45RA and CD62L have equivalent cytotoxicity to CD19-CAR T cells bearing a typical effector cell phenotype in a 4hr 51Cr release assay against multiple ALL cell lines. However, in prolonged co-cultures with ALL cell lines up to 4 weeks CD45RA+CD62L+ CD19-CAR T cells failed to eliminate CD19+ leukemic cells even at high E:T ratios, while effector cell phenotype CAR T cells effectively cleared targets in this long term assay. We conclude that a more complete picture of CAR based antitumor effects can be gleaned in preclinical models by utilization of both short and long-term killing assays involving in vitro and in vivo models and that investigators should not solely rely on short term 51Cr release assays to predict clinical activity of these exciting new therapies. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 3502. doi:1538-7445.AM2012-3502