In Vivo Analysis of Neuroblastoma Targeting Potential of aGD2-SIRPα Fusion Antibodies for Local CD47 Blockade

Anti-GD2 monoclonal antibodies (aGD2 mAbs) are the standard immunotherapy for patients with high-risk neuroblastoma. This treatment has improved 5-year overall survival; however, long-term efficacy still requires improvement. The "don't eat me" signal CD47 is upregulated on neuroblastoma tumor cells and inhibits aGD2 mAb effector mechanisms. Although GD2 is restrictively expressed on neuroblastoma tumor cells, CD47 expression is ubiquitous, resulting in an antigen sink and on-target off-tumor-related cytotoxicities. Recently, we developed two aGD2-SIRPα fusion mAbs for the murine and human settings. In vitro, these aGD2-SIRPα fusion mAbs restrict CD47 blockade toward GD2+ neuroblastoma tumor cells. In this study, we explored the tumor-targeting ability of aGD2-SIRPα fusion mAbs in a syngeneic 9464D-Luc-GFP and SK-N-AS xenograft neuroblastoma tumor model. Conventional aCD47 and aSIRPα mAbs encountered a dominant antigen sink in the 9464D-Luc-GFP model. Surprisingly, although aGD2 mAbs preferentially targeted tumors, murine aGD2-mSIRPα fusion mAbs accumulated in similar organs as aCD47 mAbs. Binding analysis of murine aGD2-mSIRPα and human aGD2-hSIRPα fusion mAbs to red blood cells (RBC) revealed strong binding of murine aGD2-mSIRPα to RBCs, whereas their human counterparts showed negligible binding. These data indicate species-specific CD47-SIRPα binding patterns. Utilizing a SK-N-AS xenograft model, we show effective tumor targeting of the human aGD2-hSIRPα fusion mAbs with hSIRPα in a C-terminal configuration. These data provide the first proof of principle for neuroblastoma tumor-targeted blockade of CD47 by aGD2-hSIRPα fusion mAbs in vivo and support the further development of aGD2-hSIRPα mAbs as attractive therapeutics to improve aGD2-based neuroblastoma immunotherapy.