Bispecific antibodies targeting EGFR and HER2 for effective killing of solid tumor cells via a safe CD47-SIRPA modulation

INTRODUCTION: The CD47-SIRPalpha axis is a crucial immune checkpoint, allowing tumor cells to evade immune destruction. Initial efforts to target CD47 with monoclonal antibodies and SIRPa-Fc fusion proteins encountered significant hurdles due to ubiquitous and abundant CD47 expression in particular on blood cells (including RBC and platelets). To overcome the limitation of mono-specific CD47 targeting, we generate bispecific antibodies (bsAbs) that block CD47 selectively on tumor cells, upon co-engagement of a tumor-associated antigen (TAA) on the cell surface. Harboring an immunologically active Fc portion (human IgG1 Fc) these TAA/CD47 bsAbs mediate efficient tumor cell killing through antibody-dependent cellular phagocytosis (ADCP) and antibody-directed cellular cytotoxicity (ADCC). Importantly, this strategy has now been clinically validated: NI-1801, our mesothelin/CD47 bsAb, has demonstrated selective and safe CD47 targeting together with encouraging antitumor activity in heavily pretreated ovarian cancer patients (1). METHODS: EGFR/CD47 and HER2/CD47 bispecific antibodies were assembled using the κλ-body technology/platform (2). For that goal, high affinity antibody arms specific to different regions of the extracellular domains of EGFR or HER2 were paired with a low affinity anti-CD47 arm. The resulting hIgG1 bsAbs were tested in vitro for activity in ADCP assays with TAA-positive tumor cells and monocyte-derived macrophages as effector cells, and in ADCC assays using unfractionated PBMCs as effector cells. RESULTS: The EGFR/CD47 and HER2/CD47 bsAbs efficiently induce ADCP and ADCC of tumor cell expressing varying levels of EGFR or HER2, respectively, including low level-expressing cancer cell lines. Interestingly, we could observe that the killing activity depended not only on the affinity of the TAA arm, but also on the region of the ECD bound by the TAA arm, suggesting that the geometry of the antigen-co-engagement on tumor cell surface may affect the efficiency of the immunological synapse. This was particularly apparent when comparing the killing activity mediated by different HER2xCD47 bsAbs. CONCLUSION: The “unbalanced affinities” design of TAA/CD47 bispecific antibodies effectively addresses the safety and pharmacokinetic liabilities observed with monospecific CD47 antibodies. Importantly, our clinical experience with NI-1801 (MSLN×CD47) validates both the safety and the potential clinical benefit of this tumor-restricted CD47-blocking strategy (1). Together, these findings support TAA/CD47 bsAbs as a valuable addition to the anti-cancer armamentarium, with the EGFR/CD47 and HER2/CD47 bsAbs described here representing promising development candidates. (1) De La Motte Rouge T. ESMO 2025 (Abstract #1512O) (2): Fischer N. Nat Commun, 2015 Feb 12:6:6113