The companion diagnostic paradox of B7-H3/CD276-targeted therapy: why antigen expression is insufficient to predict ADC and cell therapy response

B7-H3, also known as CD276, is a transmembrane member of the B7 immune-regulatory family and has emerged as a therapeutically attractive pan-cancer target. Its broad expression across solid tumors, relative restriction in many normal tissues, expression on tumor cells and tumor-associated vasculature, and accessibility to antibody-based and cell-based platforms have accelerated development of monoclonal antibodies, antibody-drug conjugates (ADCs), chimeric antigen receptor (CAR) T cells, T-cell engagers, natural killer-cell engagers and radioimmunotherapeutic approaches. The central translational problem is that B7-H3 expression is required for target engagement but is not sufficient to predict therapeutic response. Clinical and translational studies show that baseline B7-H3 membrane expression may fail to correlate with objective response to B7-H3 ADCs, whereas patient-derived models indicate that payload-specific tumor vulnerabilities such as replication stress, RB1 deficiency, DNA-damage response state and drug-tolerance programs can dominate response. For cellular therapies, antigen density is important, but tumor infiltration, immune synapse formation, stromal barriers, persistence, exhaustion and route of administration are equally decisive. We define the B7-H3 companion diagnostic paradox as a translational biomarker–mechanism mismatch: the same nominal biomarker—B7-H3 positivity, usually determined by immunohistochemistry—is being applied across therapeutic platforms with fundamentally different mechanisms of action and distinct determinants of response. We further argue that B7-H3 companion diagnostics should be platform-specific rather than universal, and that the next phase of B7-H3 translation should move beyond binary positivity toward spatially resolved, functionally validated and cancer-context-aware response prediction.