The naive question: an antibody is the body's targeting protein, but it grabs one thing. What if a single molecule could grab two things at once? That's a bispecific antibody — and in cancer, the 'two things' are chosen on purpose, then engineered so the molecule can hold both at the same time.

Think of it like a matchmaker with a hand on each party's shoulder. One arm of the bispecific latches onto a marker on a tumor cell. The other latches onto a T-cell, the immune system's effector. Holding the two together forces an introduction — the T-cell ends up parked on the tumor.

“The disclosure relates to means and methods of treating a subject for a CLEC12A positive cancer.”— U.S. Patent No. 11,873,338 source

Merus's grant US11873338B2 is a concrete example: CLEC12AxCD3 bispecific antibodies. The 'x' in the name is the giveaway — it binds CLEC12A (a marker found on certain leukemia cells) and CD3 (a component of the T-cell receptor complex). One molecule, two targets, deliberately bridged. But what's striking when you read this particular patent is that its claims are not really about the molecule's shape at all — they are about how you dose it. That is the engineering problem this estate stakes out.

The reason is a side effect specific to this class. When a bispecific suddenly bridges large numbers of T-cells to target cells, the T-cells can fire all at once and flood the body with signaling molecules — a dangerous over-activation. The Merus claims address that head-on with a "step-up" regimen written directly into the patent: a first, deliberately sub-therapeutic dose, followed by larger doses in subsequent administrations, each higher than the last, before settling at a constant maintenance dose (a dependent claim sets that at least 15 mg, then at least 60 mg, with the first dose 9 mg or less). In other words, the invention is a ramp — start low so the immune system isn't shocked, then climb. The patent frames this as an engineering control on a powerful bridge, and even claims a related idea: a "hemopoietic stem cell sparing" schedule, dosing so that healthy CLEC12A-positive blood cells can recover while malignant ones are cleared. The disease context recited in the claims is acute myeloid leukaemia and related myeloid disorders.

The design space is wider than direct tumor-to-T-cell bridging. Roche's publication US20240002546A1 covers a combination using a PD1-LAG3 bispecific together with a CD20 T-cell bispecific — here the bispecific format is pointed at immune checkpoints (PD1 and LAG3 are brakes on T-cells) rather than purely at direct killing, and the inventive idea is pairing two different bispecifics. Publication US20240398951A1 covers T-cell engagers built against a different tumor target altogether, showing that the same two-handed architecture is being re-aimed at new markers.

Here's the 'so what.' A bispecific can do something no single-target antibody can: physically connect two cells that wouldn't otherwise meet. That capability is the whole point — and it is also the whole danger, because forcing that connection too hard or too fast has consequences. The 2024 records make clear that the engineering work has shifted accordingly. Some of it is still about which two targets to bridge (CLEC12A and CD3; CD20 and a T-cell; checkpoint pairs). But a striking amount of it — as the Merus claims show — is about when and how much: turning a step-up dosing schedule into a patentable, controlled way of switching the bridge on gradually.

The fact that Merus chose to claim a dosing schedule rather than a new molecular shape is itself revealing about where the modality's difficulty has migrated. The two-handed architecture — one arm to a tumor marker, one to CD3 — is by now well established; what remains genuinely hard, and genuinely dangerous, is switching such a powerful bridge on without overwhelming the patient. By writing a low first dose, an ascending ramp, and a fixed maintenance level into the claims, the patent treats the tempo of activation as the invention. The recited numbers — a first dose of 9 mg or less, maintenance of at least 15 mg and then at least 60 mg — are the engineering specification of that ramp.

The "stem-cell-sparing" claim adds another layer of control logic. CLEC12A is found not only on malignant cells but on some healthy blood cells, so an indiscriminate, sustained attack would damage the very system that regenerates blood. The claimed schedule is designed to clear malignant cells while leaving room for healthy CLEC12A-positive cells to repopulate — a timing-and-dose problem, not a targeting one. Set beside the Roche checkpoint-combination publication and the T-cell-engager publication against a different target, the 2024 records together show a class that is being engineered along two axes at once: which two targets to bridge, and how to modulate the bridge over time.

It is also worth noting how the naming convention itself encodes the engineering. A label like CLEC12AxCD3 reads as "this target by that target," and across the 2024 records you can read the design intent straight off the name: a tumor marker paired with CD3 for direct engagement, or a checkpoint pair like PD1-LAG3 for releasing immune brakes. The two-target shorthand is, in effect, a compressed statement of what the molecule is built to bridge.

The short version: a bispecific antibody is a molecule with two hands, and in cancer it usually holds a tumor-associated marker in one and an immune cell in the other. The 2024 patent record is full of variations on which two targets to bridge — but it also shows that controlling the bridge, through step-up dosing and stem-cell-sparing schedules, is now its own field of claimed engineering.