CD19 CAR T Cells Maintain Efficacy in the Allogeneic Environment but Mediate Acute Graft-Versus-Host-Disease Only in the Presence CD19+ Acute Lymphoblastic Leukemia

A significant portion of patients with relapsed acute lymphoblastic leukemia (ALL) who are eligible for treatment with chimeric-antigen-receptor (CAR) T cells have undergone a previous hematopoietic stem cell transplantation. We and others have previously demonstrated that the allogeneic environment, even in the presence mild graft-versus-host disease (GVHD), severely impairs tumor-directed T-cell immunity. To study the behavior of CAR cells in the allogeneic setting we chose a murine model of minor mismatched allogeneic T-cell depleted bone marrow transplantation (BMT) followed by CAR T (CD19-28-z) cell infusion in recipients bearing a murine B-precursor ALL model (positive for CD19). Importantly, the leukemia persists following myeloablative radiation thus mimicking post-transplant minimal residual disease. Donor-derived, post-transplant injection of CD19 CAR T cells eliminated residual leukemia in the syngeneic as well as the allogeneic settings. CD19 CAR T cells harvested from allogeneic recipients maintained in-vitro ability to produce IFN-gamma and degranulate in the presence of ALL ex vivo, at levels comparable to syngeneic CD19 CAR T cells. However, CAR T cells administered in the allogeneic environment had the potential to mediate severe acute GVHD with early mortality of recipients not typically seen in this minor mismatch model. This occurred across multiple T cell doses capable of clearing leukemia (0.3e⁶-5e⁶), in transduced CD19 CAR T cells generated from donors tolerized to allogeneic antigens, and when CD19 CAR T cell infusion was delayed following BMT. In-vivo tracking of transferred cells showed comparable expansion and persistence of the CD19 CAR T cells in the allogeneic and syngeneic environments. However, syngeneic CAR T cells tended to develop later into central memory T cells (CD62L⁺CD44⁺), whereas the profile of allogeneic cells was significantly skewed toward effector T cells (CD62L^-CD44⁺). Remarkably, the process of CAR-driven acute GVHD in this minor mismatch model was only seen in the presence of leukemia, indicating a CAR-target interaction influences the induction of GVHD. Indeed, pro-inflammatory cytokines IFN-gamma and IL-6 were elevated only in the presence of both ALL and CAR T cells, whereas TNF alpha levels were undetectable in all instances. We are currently testing whether neutralization of cytokines can prevent GVHD in these models. Altogether, these data demonstrate efficacy of CD19 CAR to clear residual leukemia in an immunocompetent mouse model, and maintain initial cytotoxicity despite the potentially suppressive allogeneic environment. However, we demonstrated potential risks of allogeneic CAR T cells aggravating GVHD in the presence of residual leukemia, likely via cytokine-mediated manner. Clinically, as IL-6 was shown to be significant in cytokine release syndrome, this may represent a murine model to study potential interventions.