CD64 CAR-T Therapy Targets Venetoclax-Resistant Monocytic Acute Myeloid Leukemia

Introduction: Acute Myeloid Leukemia (AML) continues to have a poor prognosis despite therapeutic advances. Venetoclax-based therapy induces remissions in approximately 70% of patients, but duration of response is limited and relapses are common. We have previously shown a higher propensity for venetoclax resistance in AML with a monocytic (FAB M5) phenotype and correspondingly inferior overall survival after venetoclax/azacitidine (Ven/Aza) treatment relative to patients with non-M5 AML (Pei, et. al. Cancer Discovery, 2020). We further demonstrated that while leukemia stem cells (LSCs) with an undifferentiated phenotype are sensitive to Ven/Aza, a unique monocytic LSC (m-LSC) population, which uniformly expresses CD64, can drive monocytic AML relapses in patients treated with Ven/Aza (Pei, et. al., Cancer Discovery, 2023). As venetoclax is integral in many AML treatment regimens, we developed a novel CD64-directed chimeric antigen receptor (CAR) to target the Ven/Aza-resistant m-LSC population. We have previously shown CD64 CAR T cells are effective against CD64+ AML cell lines in vitro and in vivo. To further optimize this CAR T cell product for clinical translation, we have evaluated the efficacy of AML patient-derived CD64 CAR T cells, characterized their in vivo persistence, tested their synergy with venetoclax against patient-derived AML with mixed undifferentiated and monocytic populations, and validated a new humanized scFv. Collectively, these pre-clinical data support the development of our CD64 CAR T cells for a planned phase I/II clinical trial for patients with post-Ven/Aza relapsed AML. Methods: Second-generation CD64 CAR constructs incorporating different scFv's and costimulatory molecules were transduced into T cells from healthy donors or AML patients. Phenotype, in vitro function, in vivo efficacy and persistence were evaluated for CD64 CAR T cells both alone and in combination with Ven/Aza. Results: We previously demonstrated that CD64 CAR T cells effectively clear monocytic AML cell lines in xenograft models, however the optimal design to maintain efficacy and enhance persistence is unknown. We found the 4-1BB costimulatory domain drove a final CAR T cell product that was predominantly CD4 cells with either a T stem cell memory (TSCM) or T central memory (TCM) phenotype. In contrast, the CD28 costimulatory domain generated a CD64 CAR T cell product that was balanced between CD4 and CD8 cells, which surprisingly were heavily skewed towards a TSCM phenotype. Despite these differences, we found ~10-fold greater persistence of CD64BBz CAR T cells, relative to CD6428z, in vivo after leukemia clearance (p<0.05). Persisting CAR T cells predominantly displayed a T effector memory phenotype and showed limited expression of the inhibitory receptors PD1 and TIM3. In patients the immunogenicity of the CAR molecule can impact persistence, so we next tested a novel humanized CD64 scFv. Using low doses, humanized CD64BBz(hu) CAR T cells were able to completely clear established THP1 leukemia, whereas CD64BBz CAR T cells with the murine scFv (CD64BBz(mu)) only cleared 66.7% of mice and CD64-28z(mu) was only effective in 11.1% of engrafted mice. Given the intention to use CD64 CAR T cells for patients relapsed after Ven/Aza, we tested the ability to generate CAR T cells from AML patient-derived T cells. We successfully manufactured patient-derived CD64 CAR T cells, which were functional against AML cell lines and could eliminate autologous CD64+ AML. Furthermore, we found differential susceptibility to Ven/Aza and our CAR T cells in patient-derived AML samples consisting of a mixture of primitive/undifferentiated and monocytic populations. Treatment with Ven/Aza specifically reduced the CD34+/CD64- (primitive) AML subpopulation, whereas our CAR T cells eliminated the monocytic CD64+ subpopulation. Conclusions: Through these pre-clinical studies, we have identified the optimal scFv and costimulatory domain of our CD64-directed CAR to maximize efficacy and persistence as well as defined the ability of our CD64 CAR T cells to target Ven/Aza resistant, monocytic AML. These data suggest that CD64 CAR T cells can target venetoclax-resistant m-LSCs and support the development of a phase I/II clinical trial to test the safety of this potential salvage regimen for patients who relapse after Ven/Aza therapy.