In the Wake of the Supreme Court's Amgen v. Sanofi decision: What’s Next for Biotechnology Claims?

By Kevin E. Noonan --

Section 112 of the patent statute, which in earlier years was something of a backwater in patent law, has had a tumultuous quarter century beginning with the Federal Circuit decision in Regents of the University of California v. Eli Lilly & Co., which (in the view of many) heightened the written description requirement for biotechnology inventions.  This was particularly true more generally for inventions directed to isolated nucleic acids (those were the days!) and proteins, based on the principle that the complexity of such molecules required disclosure of more than a recitation of the expected product (in Lilly, human cDNA encoding insulin) and methods for making/obtaining it.  This decision led to a series of cases from that Court, including Enzo Biochem. v. Gen-Probe, University of Rochester v. G.D. Searle, Carnegie Mellon University v. Hoffman LaRoche, and culminating in the en banc Ariad v. Eli Lilly & Co. decision that imposed the interpretation that written description and enablement in 35 U.S.C. § 112(a) (then, first paragraph) were separate requirements.  The enablement requirement in this regime had the steadier interpretation, being a question of law grounded in the factual determinants of the In re Wands factors, so much so that the isolation, discovery, and characterization of a novel antigen was enough for an applicant to be granted a patent for antibodies specific for that antigen without actually disclosing said antibodies, under the Court’s 2004 Noelle v. Ledermann decision.  Many recognized that this decision was contrary to the Ariad decision but that apprehension was grounded in the written description rather than enablement requirement.

That situation began to change during the tenure of Chief Judge Prost, who wrote the first Federal Circuit Amgen v. Sanofi opinion harmonizing the Court's precedent by overruling on written description grounds the Ledermann decision.  But lurking in the Court's consideration of that case was an inkling that the enablement requirement also contained a limitation, based on the principle of "no undue experimentation" from the Wands decision that bore fruit in several other cases, including Wyeth & Cordis Corp. v. Abbott Laboratories, Enzo Life Sciences, Inc. v. Roche Molecular Systems, Inc., Pacific Biosciences of California, Inc. v. Oxford Nanopore Technologies, Inc., and Idenix Pharmaceuticals LLC v. Gilead Sciences Inc. (the latter being a conventional chemical compound/genus case that, it had been presumed, encompassed molecules of sufficient predictability that did not implicate the complexity issues of biological molecules).  This trend reached its apogee (for now) in the Supreme Court's Amgen v. Sanofi decision, where the Court approved the limiting principle that the enablement requirement was satisfied if (but only if) the claims were commensurate in scope with what was disclosed in the specification.  The Federal Circuit has applied the Supreme Court's reasoning (and their own) since the Amgen decision, in Baxalta Inc. v. Genentech, Inc.

In the wake of this decision (once one gets beyond the weeping and gnashing of teeth that accompany most Supreme Court forays into patent law) there have been a number of views asserted about how (if at all) biological molecules (and their chemical counterparts) could be protected with claims of sufficient scope not to be easily designed around.  An intriguing one has been (independently) promulgated by Bob Armitage and Tom Irving, who have suggested that "means-plus-function" (MPF) claims under  35 U.S.C. § 112(f) would provide a way to expand the scope of claims having great sequence or other complexity while avoiding the conundrum created by (or more accurately, affirmed by) the Amgen decision.  As a reminder this provision of the patent statute provides:

An element in a claim for a combination may be expressed as a means or step for performing a specified function without the recital of structure, material, or acts in support thereof, and such claim shall be construed to cover the corresponding structure, material, or acts described in the specification and equivalents thereof.

Under this proposal,* exemplified with regard to Amgen's antibody claims, the stratagem would be implemented as follows:

A pharmaceutical composition of an isolated monoclonal antibody preparation comprising a pharmaceutically acceptable excipient and an antibody means that specifically binds to [a particular antigen/epitope/pathogen/etc.].

The specification would then provide (as did Amgen's) expressly disclosed antibodies as exemplars.

Reservations arise regarding this approach because MPF claims are limited to the species expressly disclosed ("the corresponding structure [or] material") and equivalents thereof.  For antibodies (as in Amgen) this would have been the about two dozen expressly disclosed antibodies plus "equivalents."  But this begs the question because what is an "equivalent" will need to be experimentally determined and thus be subject to the same "undue experimentation" objections that forms the basis for both the Supreme Court's Amgen decision and the several Federal Circuit decisions in recent years limiting their scope even in conventional small molecule/chemical claims.  And while the doctrine of equivalents is a thin reed to rely upon to police, at best, "trivial" modifications (like substituting an isoleucine residue with an valine residue, or vice versa, insofar as the difference in these amino acids is one methylene (-CH₃-) group) the scope of MPF claims is also more limiting in this regard.

A possibly more fruitful course (which has the benefit of also providing an argument for satisfying the written description requirement) would be the alternative ways of providing a sufficient disclosure first enunciated by Judge Lourie in California v. Eli Lilly & Co. and later in Ariad (v. Eli Lilly & Co.):

• Rely on species and subgenus claims as much as possible, while avoiding broad genus claims, to provide a "representative number" of species

• Focus on structure/function relationships and properties of antibodies (CDRs, for example) not antigens

(or a combination of these).  There has been a convergence in the relationship between claim scope and the amount of disclosure supporting claims required for satisfaction of the written description and enablement requirements that can be addressed using Judge Lourie's rubrics.

Another approach would be to align the CDR sequences in expressly disclosed antibodies and provide a sequence alignment that could be used to identify a consensus sequence having three types of alignment.  Specifically, there would be 1) invariant amino acids at some positions (which would indicate that these are important for antigenic specificity or structural stability or both); 2) positions having amino acids of related chemical properties (basic, acidic, aliphatic, etc.) that provide more structural variability/scope; and 3) positions having little consistency, except most likely not having proline or glycine which are known to disrupt protein secondary or tertiary structure.  The number of variants for such consensus sequence CDRs, while large, would then have a more limited and scientifically rational basis resulting from evolution and would not involve the "trial and error" characteristics that raised judicial disparagement at both the Federal Circuit and Supreme Court.

There is no magic wand that can ensure that a claim will not be found wanting by a court in satisfying the disclosure requirements of Section 112.  And of course, sound patent claiming strategies (having claims of varying scope, including genus, subgenus, and species, for example) are useful in reducing the likelihood that a competitor will be able to practice an invention with impunity by clever designing around schemes.  The Amgen claims (as well as the claims in Baxalta) were prosecuted under interpretations of the enablement requirement that have now been abrogated by the courts.  While it is likely that patents having similar claims will be subject to invalidation if asserted the benefit of decisions like Amgen is that the standard, having been established by the Supreme Court is less likely to change in future, as not relying on the Federal Circuit's recent uncertain precedential value.

Comments

[Darden]

Thank you for this post. I think the alternatives discussed are the most likely to work, but am mindful that a number of examiners in Art Unit 1600 have rejected any claim language directed to allowing some even small amount of sequence variation (including for example sequences with very high (99%) percent identity or sequences varying by "less than" a small number of amino acids or nucleotides in a molecule with known conserved regions). Examiners have also rejected claims encompassing the *use* of antibodies that bind to a known antigen even when there are several well established examples in the art, which seems at odds with Capon v Eshhar (although maybe that case has now been made obsolete). So unfortunately I think in practice it may now be nearly impossible to get claims of any useful breadth.

[Tom Irving]

I think Kevin's post is, as usual, brilliant. Also, recently, Sharon Crane of Rothwell Figg gave a brilliant treatment of this written description topic, and Joanna Brougher of the BioPharma Law group, masterfully dealt with MPF, both on AIPLA's September 20 Hot Topics in Chem and Pharma program. For those who missed it, recordings are available through AIPLA. I stand strongly for MPF, having spent years promulgating it. To be sure, what exactly is a literal equivalent will depend on the specific facts and circumstances. In the pharma space, I am aware of only one case, Wockenhardt, where a pharma MPF claim was litigated. The case settled, but what caught my eye was that the defendant accepted an injunction as part of the settlement. Before I retire in 2066, I expect we will have much more light and knowledge about MPF in pharma. I am using it with my pharma clients at The Marbury Law Group and my former law firm has done the same.

[Kevin E Noonan]

Dear Darden:

While I appreciate your bleak outlook, I think the only way to address the issue is to abandon the "picnic blanket" approach to claim scope and adopt a "picket fence" approach (or perhaps a pointillist approach) in identifying as many related sequences that can be verified so that the "millions and millions" meme is not applicable. (I'm sure the sequence comparison software will identify the number of permutations in any consensus sequence.) Coupled with the routine nature of the binding tests to verify function based on structure at the minimum competitors will need to invest the time and effort to design around (which is sufficiently significant that the "first mover" advantage may be enough to make the effort worthwhile).

Thanks for the comment.

[skeptical]

Dr. Noonan, this is outside of my practice area, but a potential overlap with an area that I am familiar with occurs with your comment of, "the sequence comparison software will identify the number of permutations in any consensus sequence."

If - and this has been admitted to by at least one pharma company in public discussions on Artificial Intelligence (AI) - the "identifying" action is by 'software,' AND that is the action that is setting out the patentable distinction, then you will have a circumstance in which an invention is being made by a non-human actor (the AI, in at least a co-inventor role), and no viable patent will be forthcoming -- either by the prong of no patents for non-human inventors OR the prong that anything from an AI may well be considered obvious [if AI's are deemed non-inventive].

Further, you advance a notion of "first mover" advantage, but I think such a notion is still-born given the (same) burgeoning use of AI for speed of development.

Again, pharma is typically outside of my bailiwick, but it does not take operating in that sphere to see some serious oncoming issues for an industry that has so heavily depended on the patent model to eek out payback (and often in the later days of patent coverage).

[Kevin E Noonan]

Dear Skeptical:

Going in inverse order, the "first mover" idea is that the hypothetical patentee will have actually made several antibodies, subjected them to testing and development, and obtained FDA approval to market them. Given the effort all that requires (which will also be required by copycats and design arounds) there is likely to be a significant lag in competing antibodies being marketed, and that is the "first mover" effect I posit might give an advantage.

As for AI, in this instance it is just a tool, because it is not creating anything but the comparison, which is based on the sequence of the antibodies themselves that are real world molecules. Producing the comparison does not rely on any of the properties that make AI intelligent; indeed, such comparisons have been routine for ~ 40 years.

Thanks for the thoughts.

[skeptical]

Thank you Dr. Noonan - I think that you are not addressing the novel characteristics of the use of AI, which brings it to be MORE than merely using a tool.

IF in fact, one wants to depend on "just using a tool,"
THEN this necessitates the view that the tool is NOT providing any 'inventive step' and the results of the use of the tool is de facto obvious -- obvious to a NEW, albeit still non-real person but instead legal fiction) Person Having Ordinary Skill In The Art.

There is a Hobson's choice in the current "mere use of" AI that needs be addressed.

[David C Diamond]

I too have long been an advocate of MPF claims in biology (though not nearly as prominently as Tom) and completely agree with him that exactly what an equivalent will be in the view of the courts is an open question. However, it has to be broader than simply claiming the disclosed sequences. What value that has will likely vary widely from case to case. Kevin worries about the testing that will be necessary to confirm an equivalent is in fact one and suggests pursuing consensus sequences instead. By all means, construct a consensus sequence. No one claim type will be a panacea. But please realize that there will still be testing required to determine if any particular sequence encompassed by the consensus sequence has the associated function expected.

[Kevin E Noonan]

Dear Skeptical: I have no trouble confessing ignorance about AI. I just don’t think it’s relevant because (to make an analogy) what I suggest can be performed with an abacus and doesn’t need a digital computer.

And I am discussing sufficiency of disclosure - the novelty and nonobviousness derive from the biology and its application (I recognize the risk of the section 101 folks making an eligibility argument).

Thanks