Harnessing the Power of CAR.HER2 T Cells: A Novel Approach to Target Osteosarcoma and Ewing’s Sarcoma

Pediatric and adolescent bone sarcomas, particularly osteosarcoma (OS) and Ewing’s sarcoma (EWS), are aggressive and treatment-resistant. Despite multimodal therapy, survival for recurrent or metastatic disease remains below 30%. The transmembrane tyrosine kinase receptor HER2 is overexpressed in a substantial fraction of OS and EWS. Although anti-HER2 monoclonal antibodies such as Trastuzumab deruxtecan (T-DXd) are safe, their efficacy is limited by low antigen density and an immunosuppressive microenvironment. CAR T cells, with antigen-specific cytotoxicity, may offer a potent therapeutic option. We developed and characterized third-generation CAR.HER2 T cells for pediatric bone sarcomas, aiming to enhance efficacy in HER2-low tumors while ensuring safety via a ΔCD19 suicide gene strategy. Two constructs, (4D5)CAR.HER2 and (2C4)CAR.HER2, were derived from trastuzumab and pertuzumab, respectively, incorporating optimized scFvs which design was based on structural bioinformatic analyses, and dual co-stimulatory domains. Flow cytometry of pediatric tumor samples detected HER2 expression in 46.2% of OS and 33.3% of EWS. Both CAR constructs exhibited high membrane expression ((4D5) 84 ± 9%; (2C4) 88.4 ± 6.2%), similar phenotypes, and preserved TCR diversity. In co-culture assays, both CARs eradicated HER2-high OS (SAOS-2, p < 0.05) and HER2-low EWS (A-673, p < 0.0001), even at low effector: target ratios (1:16). Repeated tumor stimulations confirmed durable activity up to the third challenge (p < 0.05). In 3D EWS models, both CARs showed robust cytotoxicity at 20 h (E:T 1:1; p < 0.0001). CAR.HER2 T cells also targeted tumor models resistant to T-DXd. Both constructs rapidly eliminated HER2-low JIMT-1 cells (T-DXd-resistant) and A-673 cells (minimally sensitive to T-DXd). In in vivo orthotopic OS xenograft model in which NGS mice were engrafted with HER2-high U2-OS cells, both CAR T cells eradicated tumor at day+40, with bioluminescence dropping to 2.2×10⁶ ± 2.5×10⁶ photons/s (for 2C4) and 1.1×10⁶ ± 2.4×10⁶ photons/s (for 4D5), while NT controls remained around 3.8×10⁹ ± 4.5×10⁹ photons/s (p < 0.0001). In a more challenging HER2-low EWS mice model (A-673), only (4D5)CAR.HER2 T cells achieved complete tumor eradication by day +24 (1.3×10⁶ ± 5.8×10⁵ photons/s, p = 0.0163 vs control cohort), whereas (2C4)CAR.HER2 T cells merely delayed tumor progression respect NT T cells (5.7×10⁸ ± 6.1×10⁸ and 2.2×10⁹ ± 1.4×10⁹ respectively; p = 0.0163). Marked CAR T cell expansion in three mice triggered xenoreactivity, which was effectively resolved by activating the ΔCD19 safety switch using Blinatumomab. This led to the selective depletion of CAR T cells, mitigated systemic toxicity, normalized IFNγ, TNFα and Granzyme B levels and restoring body weight, reactivity, and coat appearance. (4D5)CAR.HER2 T cells exhibit a significant anti-sarcoma activity, and durable persistence in the in vivo model, enabling rapid tumor regression even in HER2-low pediatric bone sarcoma model. The ΔCD19 safety switch provides controllable mitigation of possible on-target/off-tumor toxicity. Compared to prior CAR constructs, (4D5)CAR.HER2 T cells demonstrate superior functional fitness, balanced immunophenotype, and enhanced therapeutic efficacy, supporting translation into a phase I clinical trial. These results suggest that other antibody-resistant tumor antigens could be effectively targeted using similar CAR T cell strategies.