The hardest problem in CAR-T is moving from blood cancers to solid tumors, which requires a binder for a solid-tumor antigen. US11707487B2 - "EpCAM antibody and CAR-T cells," issued July 25, 2023 to AffyImmune Therapeutics, Inc. - claims a binder for EpCAM and the CAR cells built around it.

In a chimeric antigen receptor, the antigen-binding domain is the module that decides what the engineered T cell attacks. This claim fixes that domain to EpCAM (epithelial cell adhesion molecule), a marker expressed on many carcinomas. The CPC tags - C07K 16/30 (anti-tumor antibody), C07K 14/7051 (CAR component), A61K 35/17 (T-cell therapy) - capture both the antibody and its incorporation into the receptor.

“The present invention provides EpCAM antibodies with different affinities. The present invention also provides chimeric antigen receptors (CARs) specific to EpCAM.”— U.S. Patent No. 11,707,487 source

The claim's two-part structure matters. It covers the EpCAM antibody itself and the CAR-T cells that use it as their binding domain. That dual coverage means a competitor could infringe either by making the antibody or by building CAR cells with a binder falling within the claimed EpCAM-targeting domain - two distinct infringement theories from one grant.

The scope question is the binder's sequence and specificity. EpCAM is expressed on healthy epithelium as well as tumors, so the on-target-off-tumor risk is real, and the value of a particular EpCAM binder lies in its affinity and selectivity profile. A competitor's EpCAM CAR infringes only if its binding domain falls within the claimed antibody's scope - which turns on the CDR sequences, not the antigen name.

For the landscape, solid-tumor CAR-T antigens like EpCAM each anchor their own small estate of binder-and-construct claims. Freedom-to-operate for a solid-tumor cell therapy means clearing the specific antigen-binder claims for its target, separately from the foundational CAR-architecture layer - a reminder that 'CAR-T patent' fractures by antigen as much as by construct.