Amgen Disclosed 26 Antibodies and Claimed Millions — the Supreme Court Said No in 2023

Amgen and Sanofi each built a cholesterol drug on the same idea: block a protein called PCSK9 so the body clears more LDL from the blood. Amgen's drug is Repatha, Sanofi's is Praluent, and each company held a 2011 patent on the exact antibody inside its own product.⁶ That could have been the end of it - two rivals, two molecules, two patents. Instead, in 2014, Amgen went back and obtained two broader patents that tried to claim not just its antibody, but every antibody that binds and blocks PCSK9 the same way.⁸ Then it sued to enforce them. Nine years later, a unanimous Supreme Court told Amgen it had claimed a thing it never taught anyone to make.

The story most people carry away is that Amgen's cholesterol patent got struck down. Almost none of that is right. The patent on Repatha was never touched. What died was a far more ambitious claim - the attempt to fence off an entire class of molecules, most of which Amgen had never seen, described, or built.

The 26 examples that were supposed to cover millions: Amgen described a handful of antibodies and claimed the right to every one that did the same job

Here is the mechanism, and it is the whole case. A patent is a bargain: you teach the public how to make and use your invention, and in exchange you get a temporary monopoly on it. That teaching requirement is called enablement, and it is written into the Patent Act at 35 U.S.C. Section 112.¹ Amgen's 2014 patents did not claim 26 specific antibodies. They claimed the function - any antibody that binds to certain amino acid residues on PCSK9 and blocks it from binding LDL receptors - and then disclosed the sequences of 26 examples plus a generic 'roadmap' and a 'conservative substitution' method for finding more.⁸ The problem is arithmetic. The functionally-defined class Amgen claimed potentially runs into the millions of distinct antibodies.¹ Twenty-six worked examples and a suggestion to keep experimenting is not teaching someone to make the class - it is handing them a treasure map and calling it the treasure.

The Court reached for an image to make the gap concrete. It likened Amgen's disclosure to handing over 26 working combinations of a lock that has millions of possible combinations - most of which don't open the door - and claiming you'd taught someone to open every lock.³ You haven't. You've told them where 26 of the answers are and left them to guess at the rest. Under settled patent law, that is not enablement; it is a claim to territory you never surveyed.

Every court that heard the question said the same thing: this was not an upset at the top, but a consistent defeat repeated up the entire ladder

The clean way to see this decision is that it wasn't a decision - it was a confirmation. A Delaware federal district court held Amgen's 2014 claims invalid for lack of enablement under Section 112(a) before the case ever reached the top.³ The Federal Circuit affirmed that the function- and epitope-defined claims lacked enablement in its 2021 ruling.⁴ Then the Supreme Court, hearing the substantive question after argument in March 2023, agreed - unanimously - in May.⁷ Four levels of review, one answer. When a company litigates the same claim through the district court, the appeals court, and the Supreme Court and loses at every stop, the loss at the top is not the surprise. The decision to keep going was.

Didn't the Court just make biotech patents impossible?: the ruling that read like a new rule was the Court applying old law to an unusually greedy claim

The strongest objection is that this decision hands defendants a brutal new weapon: if a broad functional claim can be voided for disclosing 'only' 26 antibodies, won't every ambitious biotech patent now fall? It is a fair worry, and the answer is that the Court went out of its way to head it off. It was explicit that its reasoning did not alter the existing enablement standard.⁴ Rather than announce a new doctrine, Justice Gorsuch's opinion leaned on a trio of 19th- and early-20th-century precedents - settled law about the old bargain between teaching and monopoly.⁵ The reason the case looked like a landmark is that the claim was extreme, not that the rule was new. Amgen didn't lose because the law tightened around it. It lost because it asked for more than it had ever taught, and the law has always drawn that line.

Sanofi called the ruling 'an unequivocal win for America's innovation economy' - the self-interested phrasing of the party that got to keep selling its own independently developed drug.² Strip the spin and the real lesson is quieter and harder. Amgen had a perfectly good patent on the molecule it invented, and a rival had a perfectly good patent on a molecule it invented separately. That could have been a stable equilibrium. Amgen chose instead to try to own the whole idea, and pursued it through four courts until a unanimous Supreme Court reminded it of the oldest term in the patent bargain: you get to keep only what you actually taught the world to make.