The terms, per the grant. On April 15, 2025, Hoffmann-La Roche was issued US12275794B2, covering bispecific antigen-binding molecules with an anti-FAP binder. The CPC tags — C07K 16/2878, C07K 16/40, A61K 39/3955, plus the C07K 2317 engineering series — describe both a specific binder (anti-FAP, a tumor-stroma target) and the bispecific construction.

Why format is its own asset: a bispecific molecule's value splits between the binders it carries (target-specific) and the engineered format that holds them together (reusable across targets). A company with a proprietary, well-behaved format can deploy it across many binder pairings — making the format a platform asset that collects value across programs.

“The invention relates to novel bispecific antigen binding molecules, comprising (a) at least one antigen binding domain capable of specific binding to Fibroblast Activation Protein (FAP) comprising FAP clone 212 or variants thereof, and (b) at least one antigen binding domain capable of specific bin…”— U.S. Patent No. 12,275,794 source

The structure point: a bispecific deal can be priced as a binder license plus a format-platform license, each with its own royalty. For a model, the format component is the higher-multiple asset because its royalty base grows with every new construct built on the same scaffold.

What the grant does not promise: a clinical result, an approval, or exclusivity over all bispecific formats. It is an exclusivity claim on a specific construct using a defined binder, within a competitive format landscape.

The takeaway: in a bispecific term teardown, value the engager format separately from the binders, because the format is the reusable, compounding asset. Roche's April 2025 FAP-bispecific grant is a dated example of a construct whose value splits across format and binder.