Matthew A. Nix ^1,2* Caleb A. Lareau ³ Jeffrey Verboon ¹ David G. Kugler ¹

• ¹ Cartography Biosciences, South San Francisco, United States
• ² Department of Laboratory Medicine, University of California, San Francisco, San Francisco, United States
• ³ Computational & Systems Biology Program, Sloan Kettering Insitute, Memorial Sloan Kettering Cancer Center, New York, United States

Targeted antibody-based therapy for oncology represents a highly efficacious approach that has demonstrated robust responses against single tumor-associated antigen (TAA) targets. However, tumor heterogeneity presents a major obstacle for targeting most solid tumors due to a lack of single targets that possess the right on-tumor/off-tumor expression profile required for adequate therapeutic index.Multi-targeting antibodies that engage two TAAs simultaneously may address this challenge through Boolean logic-gating function by improving both therapeutic specificity and efficacy. In addition to the complex engineering of multi-targeting antibodies for ideal logic-gate function, selecting optimal TAA combinations ab initio is the critical step to initiate pre-clinical development but remains largely unexplored with modern data-generation platforms. Here, we propose that single-cell atlases of both primary tumor and normal tissues are uniquely positioned to unveil optimal target combinations for multi-targeting antibody therapeutics. We review the most recent progress in multi-targeting antibody clinical development, as well as the designs of current TAA combinations currently exploited.Ultimately, we describe how multi-targeting antibodies tuned to target pairs nominated through a datadriven process are poised to revolutionize therapeutic safety and efficacy, particularly for difficult-totreat solid tumors.