When a biotech company talks up a drug platform, the molecules get the headlines. The manufacturing process rarely does. But a patent that issued to Amgen on June 16, 2026 is a reminder that, for antibody drugs, how you make the thing can be as carefully fenced off as the thing itself. The grant, US12655177B2, covers what its own text calls "an improved manufacturing method for the production of bispecific antibody products, which comprise at least two binding domains."

Bispecific antibodies are engineered proteins designed to grab two different targets at once. Amgen is one of the most visible players in a specific flavor of them: T-cell engagers, which tether an immune cell to a tumor cell. The category is commercially material for the company, which makes the question of how the proteins are produced at scale more than an engineering footnote. The newly issued patent describes a step in that production: refolding the construct so its disulfide bonds reform into the protein's "native form" using a buffer that combines a redox or oxidizing agent with a chaotropic agent, controlled within a defined pH window relative to the molecule's isoelectric point.

The present invention provides an improved manufacturing method for the production of bispecific antibody products, which comprise at least two binding domains.— Downstream processing of bispecific antibody constructs, US12655177B2

One patent on a refolding buffer would be a narrow data point. What makes it worth reporting is the company it keeps. In the weeks around this grant, Amgen's other newly issued U.S. patents land in the same neighborhood: the production and purification of antibody drugs. On June 9, 2026, the company received US12649907B2, covering "alternate detergents for viral inactivation" described as eco-friendly options for biomanufacturing operations. Earlier in June, on the 2nd, it received US12643938B2, covering variant Fc-containing molecules "having glycosylation characteristics favorable to large-scale production of therapeutic molecules." Taken together, these read less like scattered filings and more like a process map: how to refold a bispecific, how to inactivate viruses during purification, how to control the sugar profile that comes off the bioreactor.

What the manufacturing cluster suggests

The pattern points to a company treating its production line as patentable territory. For a general business reader, the significance is structural. Antibody drugs are biologics, not small molecules, and they cannot be exactly copied the way a generic pill can; a follow-on competitor makes a biosimilar, a close-but-not-identical version, and must develop its own manufacturing process to do it. Patents that cover specific process steps add a second layer of consideration around that process, separate from the patents on the antibody sequence itself. The filings indicate Amgen is investing in that layer.

Amgen's broader patent record is consistent with a company that has long filed across both the molecule and the method. Its estate includes bispecific antibody constructs such as the EGFRvIII and CD3 engager in US10519241B2, and target-binding antibody programs such as the TREM2-binding proteins in US12649788B2, an antigen-binding protein aimed at conditions associated with TREM2 loss of function, including Alzheimer's disease. Those are the molecule-side filings. The grants issued this month sit on the method side, and that is the distinction worth drawing: the recent activity is weighted toward how the drugs are produced.

It is worth being precise about what a process patent does and does not do. It describes a specific way of carrying out a manufacturing step, under specific conditions. The refolding patent recites particular reagent concentrations and a pH band tied to the construct's isoelectric point. The viral-inactivation patent names a class of detergents. Coverage of that kind attaches to the described method, not to the abstract idea of making a bispecific antibody. A reader should not infer from the grant alone how broadly the claims reach or how a competitor's own process would compare; those are claim-construction questions that the patent text frames but does not, by itself, resolve.

Why process coverage matters for the platform story

For the business reader following Amgen as a name with a large biologics pipeline, the manufacturing cluster connects to a familiar theme: the durability of a biologics franchise often rests as much on the difficulty of reproducing the product as on the original molecule patent. When a company files and is granted patents on refolding buffers, detergent-based viral inactivation, and glycoform control, those filings signal investment in the part of the value chain that is hardest for a follow-on to navigate. The June grants do not tell us how a competition resolves or who wins a market. They tell us where Amgen has been spending its patent-prosecution effort, and the answer this month is the production process behind its antibody drugs.

There is also a quieter signal in the inventor and subject mix. The refolding patent (US12655177B2) is squarely about bispecific constructs with "at least two binding domains," the architecture of T-cell engagers. The Fc glycoform patent (US12643938B2) speaks to "large-scale production," the language of commercial-stage rather than discovery-stage work. The viral-inactivation patent (US12649907B2) is generic biomanufacturing infrastructure that applies across products. Read as a set, the grants point to a company hardening the manufacturing base under a platform it is producing at scale, rather than fencing a single new drug. For anyone tracking the economics of antibody biologics, that is the part of the moat that tends to outlast any one molecule.