Introduction

Many practitioners read Amgen Inc. v. Sanofi, 598 U.S. 594 (2023), as the end of functional genus claiming in the life sciences. Three years later, however, in Teva Pharmaceuticals International GmbH v. Eli Lilly & Co., 172 F.4th 1367 (Fed. Cir. 2026), the Federal Circuit held that disclosure of a single humanized antibody species could support a genus claim, and in doing so reversed a judgment of invalidity and restored a $176.5 million verdict.

Teva does not repudiate Amgen. Instead, it revises a doctrine from two decisions of the Court of Customs and Patent Appeals that the panel cites as its authority: In re Fuetterer, 319 F.2d 259 (CCPA 1963), and In re Herschler, 591 F.2d 693 (CCPA 1979). Together they are the source of what might be called the auxiliary-genus doctrine — the principle that a functional term describing a component of an invention is not the same thing as a functional term describing the invention. Under that doctrine, the question is not how many species a specification discloses, but whether the genus is the invention. Where it is not — where the claim recites a known class, the invention merely uses — the specification need only lead the skilled artisan to that class. Also, while both Fuetterer and Herschler focused on written description, Teva appears to have extended the principle to enablement for the first time.

Whether the genus is the inventive contribution or merely auxiliary is a fact about the invention, not a drafting choice. What drafting controls is how that auxiliary genus is recited. A claim may name a class the art already recognizes, leaving the skilled reader with a bounded set whose members can be identified from the literature and made by routine means. Or, it may define the same component by the result it achieves, in which case the set has no boundary at all: it contains whatever anyone later builds that produces the result, and membership can be determined only by testing. The first recitation invites the auxiliary-genus doctrine; the second forecloses it, because there is no class to which the specification could lead the artisan. The consequence is that § 112 outcomes turn on that choice as much as on the predictability of the field.
 

How Amgen Was Over-Read

Amgen held non-enabled a genus of antibodies defined by function. The received wisdom — that functional antibody claiming was finished — drew support from a written-description overlay pointing the same way: Regents of the University of California v. Eli Lilly & Co., 119 F.3d 1559 (Fed. Cir. 1997) (function is not description); Ariad Pharmaceuticals, Inc. v. Eli Lilly & Co., 598 F.3d 1336 (Fed. Cir. 2010) (en banc) (representative species or common structural features); AbbVie Deutschland GmbH & Co. v. Janssen Biotech, Inc., 759 F.3d 1285 (Fed. Cir. 2014) (roughly three hundred antibodies insufficient because all shared a common heavy chain — representativeness is diversity, not count).

None of these cases asked whether the functional genus was the invention or merely a component of it. In each, the two were the same thing. The doctrine developed on facts where the questions collapsed into one, and the profession stopped asking the first.
 

The Dormant Line of Authority

However, the question had been asked and answered before. Two CCPA decisions had squarely addressed the case where a functional genus is a component rather than the contribution, and they remained good law throughout — cited, when they were cited at all, only to be distinguished.

Fuetterer involved a tire-tread composition claim reciting "an inorganic salt that is capable of holding a mixture … in colloidal suspension in water," supported by four disclosed salts. Judge Rich, for a plurality, stated that the invention "is the combination claimed and not the discovery that certain inorganic salts have colloid suspending properties," and that the claims "should not be so restricted that they can be avoided merely by using some inorganic salt not named." That the claim might reach salts the applicant never identified was not a defect, because the salts themselves already existed: inorganic salts were a catalogued class, and whether a given one held a colloid in suspension could be established by routine test. What remained to be discovered was the property of a fixed set, not new members of an open one.
Notably, this was a composition claim — which undercuts the shorthand that method claims enjoy special latitude — and, as a plurality decision, its authority was uncertain for sixteen years.

Herschler resolved that, reversing the rejection of a method of enhancing skin penetration of a "physiologically active steroidal agent" with DMSO on one disclosed corticosteroid. It adopted Fuetterer as "the line first clearly drawn" and stated the governing principle: use of known compounds "in a manner auxiliary to the invention must have a corresponding written description only so specific as to lead one having ordinary skill in the art to that class of compounds." The reservation matters equally — "were this application drawn to novel 'steroidal agents,' a different question would be posed." This was never a license for functional claiming, but a rule about the relationship between the functional term and the inventive contribution. Its "auxiliary" language survives in MPEP § 2163, which keeps the argument available in prosecution. The line went quiet because the cases stopped fitting it. See University of Rochester v. G.D. Searle & Co., 358 F.3d 916 (Fed. Cir. 2004) (distinguishing Herschler rather than repudiating it).
 

Teva v. Lilly

Teva's headache patents — U.S. Patent Nos. 8,586,045, 9,884,907, and 9,884,908 — claim methods of treating headache by administering a humanized anti-CGRP antagonist antibody and cover Ajovy (fremanezumab). Teva asserted them against Lilly's Emgality (galcanezumab) on inducement. The jury awarded $176.5 million; the district court granted JMOL of invalidity; and the Federal Circuit reversed and reinstated the verdict (Prost, J., joined by Cunningham, J., and Andrews, D.J.).

The disclosure described several murine anti-CGRP antibodies, one humanized species (G1), and eighty-four G1 variants, which the district court had discounted.

The reasoning runs in three steps. The invention was the therapeutic method — the discovery that CGRP antagonism treats headaches, not the antibodies. The genus was well known at the priority date, and humanization was routine. Every member performs the same function within the method, so the scope is coextensive with the discovery.

The panel relied on Herschler, Fuetterer, and Ajinomoto Co. v. International Trade Commission, 932 F.3d 1342 (Fed. Cir. 2019), to hold that background knowledge can supply the genus component of a claim whose contribution lies elsewhere. Critically, Teva appears to extend the auxiliary principle from written description to enablement — the move Lilly's rehearing petition, filed June 17, 2026, principally attacks.

Rochester is distinguished on two grounds: there the genus was genuinely unknown, and the claim was circular.
 

Wyeth: The Limit

Teva is not a general relaxation. Twelve weeks later, in Wyeth LLC v. AstraZeneca Pharmaceuticals LP, No. 2024-2325 (Fed. Cir. July 9, 2026), the court affirmed JMOL of non-enablement on claims requiring a daily "unit dosage" of an irreversible EGFR inhibitor, holding that skilled-artisan knowledge "may not [be used to] provide the only means to enable these specific claim limitations." The limitation had been added in prosecution to overcome a reference — and was the one limitation the specification could not support.

Both cases ask where the claim locates its technical uncertainty. In Teva it sat in solved, cataloged territory, and the art filled the gap. In Wyeth it sat exactly where the specification was silent.
 

Juno, Re-read

The obvious question is whether Teva rehabilitates the claims that produced the largest § 112 loss of the last decade. The more useful question is narrower: how much of that loss was caused by the technology, and how much by seven words of claim language.

In Juno Therapeutics, Inc. v. Kite Pharma, Inc., 10 F.4th 1330 (Fed. Cir. 2021), claims to nucleic acids encoding a second-generation chimeric antigen receptor — a CD3ζ signaling domain, a CD28 costimulatory region, and "a binding element that specifically interacts with a selected target" — were invalidated for lack of written description, reversing a verdict of roughly $1.2 billion. Two scFvs were disclosed for the binding element, without amino acid sequences and without any correlation between scFv structure and binding function.

On its face the fit with Teva looks close: the inventive contribution was the second-generation CAR architecture — the CD3ζ signaling domain paired with a CD28 costimulatory region — and the binding element was arguably just an auxiliary component that pointed that architecture at a tumor. What defeats the argument is not the framing but the words the drafter added on top of it. Had the claim stopped at a binding element, it would have named a component. Instead, it specified one "that specifically interacts with a selected target" — and those seven words changed the character of the limitation. They define the element by what it accomplishes rather than what it is, so membership in the class can be determined only by testing: build a candidate, aim it at a target, see whether it binds. The added language does not identify a class — it identifies an outcome, and an outcome cannot be cataloged in advance.

Moreover, the seven words appear unnecessary. Prior-art rejections based in part on the inventors' own publication were overcome by narrowing the costimulatory region to specific sequences. The functional term earned nothing under § 102 or § 103 and cost everything under § 112.
 

Lessons

For the drafter, the practical instruction is to identify the inventive contribution before drafting claims and to keep every functional genus term outside it, because a term that coincides with the point of novelty invites an Amgen analysis that no framing survives. Where a genus term is needed, a structural class the art already catalogs will usually do the work that outcome-defined language was reaching for, at a cost in scope that is more theoretical than real — and the specification should carry the evidence that the class was known and its members obtainable, rather than reserving the point for expert testimony a decade later. In prosecution, the discipline is narrower but no less demanding. Every amendment deserves an audit asking not whether the invention is supported but whether that limitation is supported across the class the claim still recites, since the limitation added to distinguish a reference is precisely the one nobody has stress-tested. That was Wyeth. And any functional term that neither distinguishes over the art nor draws support from the disclosure should be struck, because Juno's binding element earned nothing under § 102 and cost everything under § 112.
 

Conclusion

The auxiliary-genus doctrine is not a concession to patentees but a statement about the relationship between a claim and a discovery: where a functional term names a component the invention uses rather than something it supplies, the specification need only lead the artisan to that component's class. Fuetterer said so of inorganic salts in a tire tread, Herschler of steroids delivered through skin, Teva of antibodies administered to treat headache.

Its limits are internal. It protects a component, so it fails when the functional term is the invention, as in Amgen. It requires a class the art can bound, and so it fails when the term names an outcome, as in Juno. And it lets background knowledge fill a gap only where the art holds that knowledge — which is why Wyeth's dosing claims failed despite being method claims of the Teva form, and why the doctrine can never substitute for asking what the field actually made predictable at the priority date.

Those two variables, architecture and predictability, are not equally within the drafter's control. Predictability is a fact about the art, fixed when the application is written. Claim architecture is a choice. Juno's binding element and Teva's antibody genus were components of comparable importance to their inventions; one was recited as a class, and the other as a result, and seven words separated them. That is the durable lesson of these cases: the doctrine rewards a drafter who can say precisely what the invention is, and it is at drafting — not in litigation — that the answer is settled.