The competitor's product became the prior art

How to read the ledger below. Each numbered point carries the kind of evidence behind it. ESTABLISHED is stated in the authority's own text, with a locator. IMPLEMENTED is the rule as the authority actually applied it. OBSERVED is carried by a source outside this one. ASSUMED is the drafter's inference, flagged as such and not the authority speaking. Quotations are checked against a cached copy of the primary source by script.

In one paragraph

The Federal Circuit reversed a jury verdict in Seagen v. Daiichi Sankyo on 2 December 2025 and held the patent invalid on two independent grounds. The 2004 parent application disclosed an enormous genus of peptide units but gave no reasonably specific supporting disclosure for the 81 member subgenus first claimed in 2019, so the continuation could not claim the earlier priority date and the competitor product publicly disclosed in December 2015 anticipated it. The claims also failed enablement because a broad construction of one term plus a functional limitation covered a scope the specification did not teach how to reach without undue experimentation.

Reader summary

Seagen's 2004 application described antibody-drug conjugates with a peptide unit built from any of 39 amino acids and any length from two to twelve residues. In 2019 it filed a continuation claiming a four-residue peptide made only of glycine and phenylalanine, an 81-member subgenus. Daiichi's Enhertu, publicly disclosed in December 2015, uses exactly such a peptide.

The court is unusually direct about what that sequence looks like, and equally direct that looking that way is not the problem. Filing to encompass a competitor's product is permissible if the application is entitled to a date before the product was disclosed. Everything therefore turned on whether the 2004 application supported the 2019 claim.

It did not, on two independent grounds. First, the disclosed genus was too large relative to what was claimed: over 47 million tetrapeptides were encompassed, and the 81 claimed species were an infinitesimal fraction of them, which is not the reasonably specific supporting disclosure a subgenus claim requires. Second, there were no blaze marks. Seagen's expert testified that a skilled artisan would take a straightforward leap from the disclosed GFLG tetrapeptide to an all-glycine-and-phenylalanine one, and the court treated that as an admission: what you must leap to is not there.

The inventors' own testimony compounded it. Each admitted they had not seen such a peptide until Enhertu or until the 2019 claims themselves.

Losing the priority date was fatal rather than merely damaging. The 2015 disclosure of Enhertu then anticipated the claims, on the principle that what would infringe if later anticipates if earlier. The jury's infringement finding became the mechanism of invalidity.

Enablement failed separately, and for a reason that should worry anyone who has won a broad claim construction. The district court read "D is a drug moiety" to cover any drug moiety, and the claim also carries a functional limitation about intracellular cleavage. Together those meant the patent covered any drug that cleaves in a patient, with no common quality disclosed and an assay required for each candidate. Under Amgen v. Sanofi that is undue experimentation.

Why this matters when you are drafting

The permission in claim 2 is real, and so is its condition. Drafting a continuation to capture a competitor's product is expressly permissible. It is permissible only if the parent supports the claim as of a date before that product was disclosed, and the support question is decided later, by a court, on the parent's four corners.

A subgenus carved out of a combinatorial disclosure is the exposure. Claims 5 through 8. A parent that recites lists of substituents and lengths describes a space, and a space is not a description of any particular region of it. The question at drafting time is whether the parent individually identifies the thing you now want to claim, in the way Driscoll's fourteen moieties were identified.

Expert testimony about what a skilled artisan could derive is dangerous to your own case. Claim 10. Testimony framed as a skilled artisan readily getting there from the disclosure concedes that the disclosure does not contain it. If a continuation claim needs that testimony, the record it creates cuts against possession.

Inventor testimony is discoverable and it decides possession. Claim 11. Before filing a branch that reaches subject matter the inventors first encountered in a competitor's product, consider what those inventors will say under oath about when they conceived it.

Losing priority is not a partial loss. Claim 12. Where the accused product's public disclosure sits between the parent and the continuation, the same evidence that proves infringement supplies anticipation. Priority is the only thing separating the two outcomes.

Watch the interaction between claim construction and enablement. Claims 13 and 14. A broad construction of a component term, combined with a functional limitation, defines a scope the specification may not reach. Winning breadth at construction can cost validity at trial, and the two are usually litigated by different people at different times.

What the authority establishes

[ESTABLISHED, the disposition] "we reverse the district court's denial of judgment as a matter of law that the '039 patent is not invalid for failure to meet the written description and enablement requirements." (P. 19.) Two independent grounds, either sufficient.

[ESTABLISHED, and this is the sentence a practitioner should read first] "One cannot avoid the suspicion that the '039 patent was filed specifically to encompass Enhertu, which of course is permissible if it was entitled to a filing date antedating any public disclosure of Enhertu." (P. 6.) The strategy is lawful. The support is the whole question.

[ESTABLISHED, why the date decided everything] Because the jury found Enhertu met every limitation, "the priority date of the '039 patent became critical". (P. 7.) Validity and infringement collapsed into one question.

[ESTABLISHED, the statutory route] To gain the benefit of an earlier application's priority date, that earlier application must disclose the invention "in the manner provided by section 112(a)". (P. 9.) Section 120 buys nothing that section 112(a) does not support.

[ESTABLISHED, the genus-size finding, and the figures describe THIS specification rather than setting any rule] The 2004 application encompassed over 47 million tetrapeptide units, and although the 81 claimed species fall within them, "they are merely an infinitesimal fraction of those peptide units generally included." (P. 12.) No ratio, count or fraction is stated as a threshold anywhere in the opinion, and none can be derived from it.

[ESTABLISHED, the rule the size finding applies] A disclosure of a broad genus, "without more, is inadequate to satisfy the written description requirement for claims directed to a" particular subgenus or species contained in it. (P. 11.)

[ESTABLISHED, what a subgenus claim needs instead] Claims to a particular species or subgenus require "reasonably specific supporting disclosure" to show the inventor possessed the specific compound. (P. 12.)

[ESTABLISHED, the contrast that marks the boundary] Driscoll found support where "the exact subgenus claimed is clearly discernible in the generalized formula" (P. 13.), which is "a far cry from the 14 moieties individually listed in the specification in Driscoll." (P. 13.) An enumerated list of fourteen is a different artifact from a combinatorial space.

[ESTABLISHED, the second and independent written-description failure] "the claimed subgenus of 81 Gly/Phe-only tetrapeptides is never mentioned in the 2004 application" (P. 13.), and the blaze-mark test asks whether the disclosure marks a trail: "We are looking for blaze marks which single out particular trees." (P. 14.)

[ESTABLISHED, and it is the most quotable line in the opinion] Seagen's own expert called the step a straightforward leap, and the court answered: "That which one must leap to is obviously not there." (P. 15.) Testimony that a skilled artisan could get there is testimony that the disclosure did not.

[ESTABLISHED, inventor testimony as evidence of absence] The conclusion was "further supported by admissions from the '039 patent's named inventors that they had never contemplated an ADC with a Gly/Phe-only tetrapeptide as of the 2004 application's priority date." (P. 13.) What the inventors say they conceived is admissible against the priority claim.

[ESTABLISHED, the consequence of losing the date] "The '039 patent thus cannot claim priority from the 2004 application" (P. 12.), and the 2015 disclosure then anticipated, on the principle that "which would literally infringe if later anticipates if earlier." (P. 12.)

[ESTABLISHED, the enablement half, and the trap inside it] The district court had construed "D is a drug moiety" "to encompass any type of drug moiety." (P. 17.) Combined with the functional cleavage limitation, "Taken together, these two limitations mean that the '039 patent covers an ADC containing any drug moiety with the recited function of cleaving in a patient." (P. 18.)

[ESTABLISHED, the standard from Amgen, as this opinion states it] What Amgen required and this specification lacked was "a quality common to every functional embodiment" (P. 17.) The locator is the Amgen discussion; the same phrase recurs on the next page applied to this patent, and per hazard 12 the shorter span resolves to the first occurrence.

14a. [ESTABLISHED, why that scope was not enabled here] Without it, "it is undisputed that ADC science was so unpredictable that a skilled artisan would be required to use an assay to test whether any given ADC with a given drug moiety meets that functional limitation." (P. 18.) So a skilled artisan needed one assay per candidate, and "where the scale of trial and error is so vast and the science so unpredictable, such experimentation is undue under Amgen" (P. 18.)

[IMPLEMENTED, the standard-of-review posture] This is a reversal of a denial of judgment as a matter of law after a jury verdict, so the court held that no substantial evidence supported the findings. Claim 1. It is a determination about this trial record, not a holding that subgenus claims of this shape always fail.

[ASSUMED, drafter's inference, flagged as such] Claims 2 and 13 together describe a squeeze that recurs in late continuation practice: the claim must be broad enough to read on the target and narrow enough to be supported by a specification written years earlier, and a broad construction won at Markman moves the enablement burden in the direction that hurts. The opinion does not frame it as a strategy tension; it follows from those two claims read against each other.

Quotations, verified against the source

All from the Federal Circuit's opinion. Page locators computed from the docket stamp.

"One cannot avoid the suspicion that the '039 patent was filed specifically to encompass Enhertu, which of course is permissible if it was entitled to a filing date antedating any public disclosure of Enhertu." (P. 6.)
"the priority date of the '039 patent became critical" (P. 7.)
"in the manner provided by section 112(a)" (P. 9.)
"without more, is inadequate to satisfy the written description requirement for claims directed to a" (P. 11.)
"reasonably specific supporting disclosure" (P. 12.)
"they are merely an infinitesimal fraction of those peptide units generally included." (P. 12.)
"The '039 patent thus cannot claim priority from the 2004 application" (P. 12.)
"which would literally infringe if later anticipates if earlier." (P. 12.)
"the exact subgenus claimed is clearly discernible in the generalized formula" (P. 13.)
"a far cry from the 14 moieties individually listed in the specification in Driscoll." (P. 13.)
"the claimed subgenus of 81 Gly/Phe-only tetrapeptides is never mentioned in the 2004 application" (P. 13.)
"further supported by admissions from the '039 patent's named inventors that they had never contemplated an ADC with a Gly/Phe-only tetrapeptide as of the 2004 application's priority date." (P. 13.)
"We are looking for blaze marks which single out particular trees." (P. 14.)
"That which one must leap to is obviously not there." (P. 15.)
"to encompass any type of drug moiety." (P. 17.)
"Taken together, these two limitations mean that the '039 patent covers an ADC containing any drug moiety with the recited function of cleaving in a patient." (P. 18.)
"a quality common to every functional embodiment" (P. 17.)
"it is undisputed that ADC science was so unpredictable that a skilled artisan would be required to use an assay to test whether any given ADC with a given drug moiety meets that functional limitation." (P. 18.)
"where the scale of trial and error is so vast and the science so unpredictable, such experimentation is undue under Amgen" (P. 18.)
"we reverse the district court's denial of judgment as a matter of law that the '039 patent is not invalid for failure to meet the written description and enablement requirements." (P. 19.)

Tensions and open questions

No line is drawn between a supported subgenus and an unsupported one. Claims 5 through 8 give two data points, fourteen enumerated moieties on one side and 47 million combinations on the other, and say nothing about the enormous interval between them. A practitioner cannot compute where a given disclosure falls.

"Never mentioned" and "infinitesimal fraction" are two different theories. Claim 9 says the subgenus is absent from the parent; claim 5 says the parent's genus is too large. If the subgenus had been mentioned once among millions of alternatives, the opinion does not say which theory would control.

The relationship between the two grounds is not explored. Written description and enablement failed independently, and the opinion says the requirements are distinct, but the same feature (a very broad disclosure paired with a narrow claim and a broad construction) produced both. Whether that is coincidence on these facts or a common cause is not addressed.

The parallel post-grant review is left hanging. The Board separately held the claims unpatentable and that appeal was dismissed as moot in a companion decision this record does not cover. Nothing here says how the two proceedings would have interacted had the outcomes diverged.

What this record does and does not cover

The written-description analysis in both of its halves, the enablement analysis, the priority framing and the disposition were read end to end.

What was NOT read closely, and is therefore outside what this record can support: the chemistry, the trial record beyond the passages quoted, the damages and royalty dispute, and the parallel post-grant review. This record says nothing about whether the claims were obvious, whether the damages theory was sound, or what the Board decided and why. Ruschig and Driscoll are known here only through this opinion's use of them.

Currency

This record was verified against its primary source on 2026-08-15. That is a fidelity check: the quotations, locators and attribution are accurate to that source. Whether the authority has since been narrowed, distinguished, abrogated or reversed has not been checked here. Confirm it still stands before relying on it.