Novel multispecific T cell engagers exploit avidity for highly selective targeting of B-myeloid mixed-phenotype acute leukemia

Introduction. B-myeloid mixed-phenotype acute leukemia (B-MPAL) is a high-risk leukemia subtype co-expressing lymphoid and myeloid surface antigens. Cure rates remain low, especially in adults (OS 20% to 50%). Current therapies are heterogeneous, borrowing from either ALL or AML regimens. MPAL cells uniquely co-express both lymphoid markers (e.g. CD19) and myeloid markers (e.g. CD33), a dual expression profile not shared by normal hematopoietic cells. We hypothesize that a Multispecific T cell Engager (MTE) engineered to preferentially bind cells expressing CD19 AND CD33 will achieve high selectivity compared to single-antigen targeting. We thus designed MTEs that bind both CD19 and CD33 on MPAL cells and CD3 on T cells. We fine-tuned the affinity of the CD19 and CD33 binders to leverage avidity; this ensures poor binding to single-positive (SP) normal cells but high-avidity binding to double-positive (DP) leukemia cells. This strategy aims to induce deep remissions and improve safety over single-target agents like Blinatumomab. Methods. We successfully designed and produced 30 tri-specific CD19/CD33/CD3 MTEs, utilizing components derived from de-risked or FDA-approved therapeutics. Our MTE framework is an asymmetric "knob into hole" human IgG1 scaffold incorporating effector silencing mutations. In vitro assays tested binding and T cell cytotoxicity against DP, SP, or negative cell lines. Top candidates were tested in vivo using a luciferase-labeled JIH-5 xenograft (CD19+ CD33+) model in NSG-SGM3 mice. Engrafted mice received repeated weekly doses of MTEs and human T cells, with disease progression tracked via IVIS imaging. Results. In vitro data confirm successful dual-antigen targeting and dramatic specificity improvements using lower-affinity binders. For multiple candidates, we observed much lower cytotoxicity for SP cells, with up to a 7-log difference in IC50​ between DP and SP cells, demonstrating an AND logic-gated treatment within a single molecule. Initial preclinical in vivo data show successful control of disease progression in xenografted mice at well-tolerated doses. Crucially, we conducted safety studies in huNOG-EXL humanized mice (engrafted with CD34+ cells) to test toxicity on normal human hematopoietic cells. Dose titration confirmed the selective sparing of normal human B cells and CD33+ myeloid cells, a major advancement over single-targeting therapies. Conclusions. By taking advantage of the unique DP features of B-MPAL cells, we developed a novel immunotherapy that offers improved half-life, potency, selectivity, and safety. This work establishes proof of concept that the combinatorial use of binders with optimized affinity generates highly specific immunotherapies with minimal normal cell cytotoxicity via an avidity-driven mechanism, paving the way for safer and more effective treatments.