Bispecific antibodies for cancer therapy: evolution of structural formats and co-targeting strategies from wet-lab to AI-driven in silico modeling.

Bispecific antibodies (BsAbs), designed to recognize two different antigens or two different epitopes on the same antigen, are generated in laboratories using various techniques, including chemical conjugation (introduced in the 1960s), cell fusion (introduced in the 1980s), and protein engineering (first introduced around the 1990s). The activity of BsAbs is dependent on how the two co-targets are strategically selected; BsAbs can be designed to redirect immune cells or block signaling pathways and to deliver therapeutic activities spatiotemporally to where they are needed. BsAbs are revolutionizing treatment of cancer and other diseases. As of this writing, at the beginning of 2026, 16 BsAbs and 1 BsAb-like fusion protein are approved in the US and/or China, of which 15 are for cancer treatment and 2 are for non-cancer indications, and many more BsAbs are in clinical trials. Next-generation BsAb development is transitioning from traditional wet-lab approaches to artificial intelligence (AI)-powered platforms. AI-assisted tools are poised to facilitate the identification of new targets, predict protein structures, and guide the development of promising new BsAb formats for the treatment of cancer and other diseases.