CD70-Specific CAR T Cells Have Potent Activity Against Acute Myeloid Leukemia (AML) without HSC Toxicity

The outcome for the majority of patients with AML remains dismal. Adoptive transfer of chimeric antigen receptor (CAR) modified T-cells (CARTs) holds the promise to improve outcomes since they do not rely on the cytotoxic mechanism of conventional therapies. CD70 is a promising CAR T-cell therapy target for AML since it is expressed on AML bulk cells and leukemic stem cells (LSC) whereas expression is absent on normal hematopoietic stem cells (HSCs) in contrast to other antigens, like CD33 and CD123, which are currently being explored as immunotherapy targets for AML. The goal of this project was to construct a panel of CD70-CARs and determine their effector function in vitro and in preclinical AML xenograft models.

We generated a panel of CD70-CARs using a CD70-specific single chain variable fragment (scFv) as the antigen recognition domain, different spacers (intermediate [IMS] or long flexible spacer [LFS]) and signaling domains that contained CD3z combined with one of three co-stimulatory domains (CD27, CD28, or 41BB). An additional CAR comprising the human ligand of CD70, CD27 (including its costimulatory endodomain) fused to the CD3z chain (CD27z) was also included in the comparison. CD70-CARs were efficiently expressed from retroviral vectors as detected by FACS analysis. T-cells expressing IMS-CARs had lower viability and a more differentiated phenotype than T-cells expressing the LFS- or the CD27z-CARs. In sequential co-culture assays with live tumor cells, T-cells expressing the LFS-CAR with a 41BB co-stimulatory domain (LFS-41BBz) or CD27z-CARs had greater proliferative potential and an increased ability to repeatedly eliminate tumor cells than NT- and other CD70-CAR T-cell populations. In addition, CD27z-CAR T-cells secreted the highest levels of type I cytokines (TNF-a and Interferon-g) after tumor cell stimulation and expressed lower levels of markers (LAG3, TIM-3 and PD-1) that are associated with T-cell exhaustion. In vivo, only LFS-28z- or CD27z-CAR T-cells had significant anti-tumor efficacy in the murine Molm-13 xenograft model in comparison to NTs and all other CD70 CARTs resulting in a significant (p<0.001) overall survival advantage. The potent anti-tumor activity of LFS-28z- and CD27z-CARTs was confirmed in a second murine AML (THP-1) xenograft model. While LFS-28z- and CD27z-CARTs had similar anti-AML activity, adoptive transfer of fire-fly luciferase-expressing CAR T-cells into AML-bearing mice revealed that CD27z-CARTs had the greatest proliferative potential in vivo. Crucially, LFS-28z- or CD27z-CARTs did not recognize normal CD34+ HSCs in colony forming unit (CFU) assays, confirming that HSCs do not express CD70 and suggesting that treatment with CD70 CAR T cells is safe.

In conclusion, CD70 is a promising target antigen for CAR T-cell therapy of AML. We demonstrate here that hinge and costimulatory domain greatly influences CD70-CAR T-cell function. LFS-28z- and CD27z-CAR T-cells had potent anti-AML activity in xenograft models with a favorable safety profile, warranting future early phase clinical testing of CD70-CAR T-cells in patients with CD70-positive AML.