通过发现肿瘤选择性AND门和OR门双特异性抗体克服癌症中有限的单特异性靶点格局

The pool of membrane proteins suitable for monospecific antibody-drug conjugates (ADCs) is inherently limited, restricting tumor selectivity and leaving many cancers without actionable antigens. Bispecific ADCs unlock a much larger combinatorial target universe by pairing antigens, enabling AND-gate designs that enforce dual expression for engagement and OR-gate designs that extend coverage where individual markers are insufficient. However, rational construction of such bispecifics depends on identifying antigen pairs that are co-expressed in tumors but absent - or expressed only individually - in normal tissues. To meet this need, we developed a discovery framework that combines quantitative proteomic profiling across more than 14 tumor types, matched normal tissues, and cell lines with computational modeling of cancer-normal specificity. This approach systematically evaluates both AND- and OR-logic potential across all bispecific ADC pairings currently under clinical or preclinical investigation, while also uncovering previously unexplored combinations with favorable selectivity profiles. Our analysis confirms that the universe of high-selectivity monospecific targets is extremely small, whereas the application of AND-logic markedly expands the number of viable tumor-restricted pairings by exploiting cancer-specific co-expression signatures. OR-logic designs provide complementary value by maintaining or improving selectivity while accommodating tumor expression heterogeneity. Among existing clinical bispecific ADC programs, only a limited subset demonstrates true AND-gate behavior, with the EGFR/MET pairing showing the most robust tumor-restricted co-expression profile. Beyond the current landscape, the platform identifies novel AND-gate opportunities with predicted improvements in therapeutic index. Functional evaluation of one such candidate validated the model's prediction of avidity-driven dual engagement and tumor-selective cytotoxicity. Collectively, these findings demonstrate that a proteomics-driven AND/OR bispecific discovery platform can substantially expand the therapeutic target space beyond what is achievable with monospecific ADCs, enabling the development of safer and more selective next-generation cancer therapeutics.