Caramel: Dual-targeting (CD123/CD33) CAR-T cell as a bridge to allogeneic transplantation for Relapsed/Refractory Acute Myeloid Leukemia. preclinical Results.

Abstract

Abstract Introduction Relapsed or refractory acute myeloid leukemia (R/R AML) remains a major therapeutic challenge. The application of chimeric antigen receptor T-cell (CAR-T) therapy in AML has been limited by clonal heterogeneity and the lack of leukemia-specific antigens. Currently, clinical evidence on the field is scarce and mainly based on single-target CAR-T cells directed against CD33 or CD123, showing limited efficacy. To date, no clinical experience has been reported using a dual-targeting strategy against both CD33 and CD123. In this study, we developed a dual CD33/CD123-targeted CAR-T cell therapy, directed against the two most frequently expressed antigens in AML, as a bridge to allogeneic hematopoietic stem cell transplantation in patients with R/R AML. Both antigens are targets of already approved drugs, supporting their feasibility and safety, as well as their translational potential to the clinical setting. Methods In-house single-chain variable fragments (scFvs) targeting CD123 and CD33 were generated, followed by the production of second-generation single-target CAR-T cells via lentiviral transduction of T lymphocytes derived from healthy donors. The single constructs that demonstrated optimal in vitro performance were subsequently evaluated in vivo using both a CDX (cell line-derived xenograft) model and a PDX (patient-derived xenograft) model of AML, both luciferase-positive (allowing bioluminescent tracking). Next, dual CAR constructs were developed using four different strategies: dual transduction, bicistronic CAR, tandem CAR, and loop CAR. Comparison of the different strategies were performed by cytotoxicity assays using MOLM-13 cells. Results Among the four single CAR constructs generated, αCD123-VL-VH-CD28-CD3ζ and αCD33-VL-VH-CD28-CD3ζ demonstrated the most favorable in vitro profiles, with high transduction efficiency, robust proliferation, and potent in vitro cytotoxic activity.These results were confirmed in vivo using CDX models. Phenotypic analysis showed a predominance of differentiated T cells (CCR7−/CD45RA−) in CAR-T cultures, with no major differences compared to mock-transduced or non-transduced controls. Then, we developed eight dual CAR constructs using four distinct strategies: co-transduction, bicistronic (x4), tandem (x2), and loop (x2) formats. All dual CAR-T products exhibited effective cytotoxic activity in vitro. Notably, the co-transduction strategy yielded the most potent product, achieving superior cytotoxicity at lower effector-to-target ratios and enhanced proliferative capacity, while preserving antigen specificity. In comparison to single CAR-T cells, dual CAR-Ts efficiently eliminated both CD123+ and/or CD33+ AML targets in vitro, potentially overcoming antigen escape and improving therapeutic efficacy. Phenotypic characterization of dual CAR-Ts revealed an enrichment in CD4+ T cells, a higher proportion of central memory cells (CCR7+/CD45RA−), and increased representation of differentiated subsets (CCR7−/CD45RA−). Moreover, dual CAR-Ts displayed elevated expression of LAG-3 and PD-1 relative to controls, though no significant differences were noted among the various dual formats. Cytokine profiling showed that co-transduced CAR-Ts secreted higher levels of IFN-γ, TNF-α, IL-2, Granzyme A, and Granzyme B, consistent with a more activated effector state. In vivo evaluation of the co-transduction dual CAR-T construct is currently underway. Conclusions The dual αCD33/CD123 CAR-T construct generated via co-transduction exhibited the most favorable profile in terms of cytotoxicity and proliferation. To date, this represents the first CD33/CD123 dual CAR-T developed in Europe for the treatment of AML. Based on these preclinical findings, next steps include the evaluation and potential design of a clinical strategy following the academic development model established at our institution.