Laundry Lists of Components are Insufficient Written Description for a Particular Combination

by Dennis Crouch

The pending Supreme Court case of Amgen v. Sanofi focuses on a broadly claimed genus and asks whether the disclosure is sufficiently detailed.  The Federal Circuit’s recent UMN v. Gilead is the flip-side.  Regents of the U. of Minnesota v. Gilead Scis., Inc., 2021-2168, — F.4th — (Fed. Cir. Mar. 6, 2023).  UNM’s original disclosure is extremely broad with “laundry lists” of components that might be included in various combinations;  while the later-added claim is directed to a more particular sub-genus combination.  The court looked in vain for some indication in the specification as to why the claimed combination is important or particularly suggested.  Here, they disclosed a large forest, but failed to disclose the tree that turned out to be valuable.

Going forward, a patentee might avoid this particular problem with some patent attorney tricks.  Rather than listing all potential components for each functional group and generally suggesting their combination, an early-stage patentee might have its AI assistant provide a hypothetical example of each possible combination, one at a time.  That situation would provide clear ipsis verbis disclosure and thus more likely survive a written description challenge.

= = =

Most U.S. patents assert priority to at least one prior patent application filing.  The early filing date helps avoid would-be intervening and invalidating prior art.  But, the right to priority is not automatic. In addition to the formal paperwork, the original application must sufficiently disclose the invention as claimed in the later patent.  Sufficiency of disclosure is generally judged under 35 U.S.C. §112(a) with the doctrines of enablement and written description.

A patent that improperly claims priority is not automatically invalid.  Rather, the priority filing date is rendered void and then we consider whether the  invention was still novel and nonobvious as-of the later filing date.  This setup also allows these disclosure issues to be considered during Inter Partes Review  (IPR) proceedings that are statutorily limited to anticipation and obviousness grounds.


In Regents of the U. of Minnesota v. Gilead Scis., Inc., 2021-2168, — F.4th — (Fed. Cir. Mar. 6, 2023), the Federal Circuit faced the issue of whether the  written description found in UMN’s priority applications supported the claims in the resulting US8815830 patent.  Here, the ‘830 patent covers the drug sofosbuvir that has FDA approval for Hep-C treatment.  Claim 1 does not simply cover sofosbuvir, but instead an entire genus of molecules that fit the general structure shown above and where functional groups R1 – R7 are further defined within the claim.  For instance, “R3 is hydroxy … R5 is an amino acid …”

UMN’s initial patent application was a provisional filed back in 2004, followed by a PCT application in 2005, followed by a non-provisional application in 2007 and finally by another non-provisional application in 2013 that led to the ‘830 patent. In 2010, a Gilead patent application published (“Sofia”) that discloses the claimed invention.  In this setup, UMN’s can disregard Sofia so long as it properly claimed priority to at least one of those earlier applications.  But, if priority fails then the ‘830 patent is anticipated.

During the IPR, the Board sided with Gilead — concluding that the priority filings lacked sufficient written description and thus that the claims were invalid as anticipated. On appeal, the Federal Circuit has affirmed in an opinion by Judge Lourie and joined by Judges Dyk and Stoll.

The patent act requires a “written description of the invention” be submitted in the form of a patent specification.  “Judicial gloss” explains this requirement has proving that the inventor “possessed the invention as claimed” at the time of filing.  Written description issues most often arise in situations like this one where the patented claims were not found in the original filing documents; but rather were added later during prosecution.

The courts have been particularly hard on patentees seeking to claim coverage for a large genus of compounds.  In that situation, it is typically a difficult endeavor for the patentee to literally enumerate all operable compounds covered by the claims.  That level of disclosure is not required, but the patentee must describe “the outer limits of the genus” as well as “either a representative number of members of the genus or structural features common to the members of the genus, in either case with enough precision that a relevant artisan can visualize or recognize the members of the genus.”  The court will also consider “blaze marks within the disclosure that guide attention to the claimed species or subgenus.” Slip Op., citing In re Ruschig, 379 F.2d 990 (CCPA 1967).

Ipsis verbis: I mentioned above that written description challenges are most often successful in situations where the claim scope has changed during prosecution.  Claims that are found word-for-word in the original application (ipsis verbis) can still fail the written description requirement, but that result is much less likely.  Here, UMN argued that the original provisional filing disclosed the eventual claim in ipsis verbis form — pointing to a combination of elements of provisional claims 1, 2, 13, 21, 22, 45, and 47 that collectively forming a substantial word-for-word recitation of claim 1 in the ‘830 patent.  On appeal though the Federal Circuit found that combination of this disparate set of claims required a bit too much cleverness.

Following this maze-like path, each step providing multiple alternative paths, is not a written description of what might have been described if each of the optional steps had been set forth as the only option. This argument calls to mind what Yogi Berra, the Yankee catcher, was reported to have said: “when one comes to a fork in the road, take it.” That comment was notable because of its indeterminacy, its lack of direction. Similarly, here, all those optional choices do not define the intended result that is claim 1 of the ’830 patent.

Moreover, Minnesota’s argument is akin to that rejected in Fujikawa, where the applicant “persist[ed] in arguing that its proposed count [wa]s disclosed ipsis verbis in Wattanasin’s application.” Fujikawa v. Wattanasin, 93 F.3d 1559 (Fed. Cir. 1996). As the court explained in Fujikawa: “just because a moiety is listed as one possible choice for one position does not mean there is ipsis verbis support for every species or sub-genus that chooses that moiety. Were this the case, a ‘laundry list’ disclosure of every possible moiety for every possible position would constitute a written description of every species in the genus. This cannot be because such a disclosure would not ‘reasonably lead’ those skilled in the art to any particular species.”

Slip Op.  I would argue that the Federal Circuit erred on this point. The UMN provisional application’s Claim 1 is directed to the same general molecule, but with much broader definitions for each of the functional group. Then, the subsequent provisional dependent claims (2, 13, 21, 22, 45, and 47) refine the scope of those functional groups in a manner that is identical to that eventually claimed in the ‘380 patent.  Here, each of these dependent claims are defined as being dependent upon any of the prior claims.  Thus, provisional dependent claim 47  seemingly includes the identical limitations found in the ‘830 patent claim 1. Yes, it is a little bit complicated, but so is chemistry.

It is not clear that my quibble would change the outcome of the case since ipsis verbis support does not necessarily result in sufficient written description.

= = =

The absence of ipsis verbis support is not fatal, so long as the original filings sufficiently show possession of the full scope of the claimed invention.  In cases involving very broad disclosure (a forest), the courts look carefully for indications that for disclosure of the particular trees actually claimed.  But, long lists of components are not sufficient. Rather, the suggestions – the “blaze marks” — must be clear.  Here, the PTAB found the patentee had failed to provide that roadmap and, on appeal, the Federal Circuit affirmed on substantial evidence.

[T]he structures [disclosed in the provisional] are so extensive and varied that [they] encompasses a significantly larger genus than that claimed in the ’830 patent, are not sufficiently common to that of claim 1 of the ’830 patent to provide written description support.

Slip Op.

UMN made other arguments that were all rejected:

  1. The PTAB should have made a credibility determination or other fact finding regarding expert testimony.  On appeal though the Federal Circuit concluded that the PTAB is not required to address each and every  side contention within its decision.
  2. The PTAB previously issued a decision that is facially inconsistent with this case, and thus is acting in an arbitrary and capricious manner.  On appeal, the court found no problem since the Board is not bound by prior panels.
  3. As a state institution, UMN is immune from IPR challenges.  UMN recognized that this argument was previously decided by the Federal Circuit.  Further, UMN has already lost a case on this argument and so is not just bound by precedent but also collateral estoppel.

Cancellation affirmed.

53 thoughts on “Laundry Lists of Components are Insufficient Written Description for a Particular Combination

  1. 6

    Some clever applicants have used an equation that, in theory, gives each species compound a number depending on a selection of 3 to 5 group variations each with its own stated group structure, e.g., a1,a2, a3,….b1, b2, b3, …. c1, c2, c3, and so on.

    Is naming every tree in the forest written description? It is a clever defensive position to take to make certain another does not obtain a select sub-genus claim.

  2. 5

    >an early-stage patentee might have its AI assistant provide a hypothetical example of each possible combination, one at a time.

    Such a rule would be a triumph of form over substance. Worse, one that favors big/rich filers who can invest in said AI tool.

    So, obviously, the Fed Cir is going to make it happen.


          This was a reply to 5.2.1. Not my fault as I checked it thrice. Something is afoot.


            Meh, missing a reply to the correct sub-thread is not that big a deal.

            But I do wonder if your question is legitimate. If you really are having difficulty understanding how turning the patent system into a shell of its perhaps most important aspect of protecting nascent disruptive technology against established entities that CAN compete on non-innovation grounds, but still want a Sport of Kings system and still want to lure small (and more innovative) entities into attempting to employ the Quid Pro Quo, surrendering their Quo withOUT obtaining their Quid.

            These are just not difficult concepts.

                1. You are doing that projection thing again, my friend with the shifting historical pseudonyms.
                  Assertions with name-calling (I chuckle every time you call me snowflake, as that is a double payout to me for YOUR choice of post) and implications that I have not attended a US law school are hilarious when YOU also never seem able to address the actual point provided.

                  Here, I gave you a direct and explicit answer to your question at 5.1.1, and instead of a proper response of, “Thank you,” are even an intelligent follow-up question, all you can do is confirm my supposition that your question was NOT legitimate.

                  Shall we again explore the notion of cyber-stalking?

                2. So you try the Argument from Incredibility fallacy again? So tired. Nobody is fooled. If they went to a US university. And law school.

                3. Care to explain what you mean by this so-called fallacy that you are accusing another of?

                  I am aware of a named fallacy called “Argument from Incredulity” fallacy (“appeal to common sense, or the divine fallacy, is a fallacy in informal logic. It asserts that a proposition must be false because it contradicts one’s personal expectations or beliefs, or is difficult to imagine.”)

                  Of course, your accusation (if you meant the known item I provide) is false. And your choosing to double down with posting after a known falsehood while evading the point I presented is (yet another) double payout for me from your choice.

                4. Res ipsa loquitur.

                  You’d know what that means without looking it up on your Wikipedia if you had actually gone to law school. Any law school.

                5. Your Latin quip does not avail you of the point that you may think that you are attempting to make.

                  The ‘speaking’ here of the plain black and white is most definitely in my favor.

                6. Thank$ for the game$

                  You’ve descended into your (rather b0r1ng) tactics — let me know if you want to advance a conversation.

                7. This is a known context – but alas, this 0bsess1ve post of yours is only a double pay out.

                  Still and all, an 11x on your choices over three repeat threads in one night is a pretty good side hustle.

                8. The phrase “in black and white” is simply immune to your simple minded twists.

                  You cannot make up in volume your lack of substance.

                9. Ooooh! You fooled everybody with your Red Herring fallacy, Snowflake. Where did you say you went to school again?

                10. As you well know, my enterprising is based solely on your choices of posts.

                  Haven’t followed foreign exchange rates for awhile.

                11. Certainly a guess on my part, you may be choosing from your short script on an entirely random basis.

    1. 5.2

      Likewise, Google and Microsoft are fully capable of defensively publishing all obvious variations of inventions in their areas of operation on a daily basis, thereby narrowing all future disputes to anticipation under Section 102. But they like disputes, as that is an easy pass to copying commercially valuable technology invented by others.

      1. 5.2.1

        You are not off the mark as to why a shell of the patent system is sought to be left in place.

  3. 4

    Correct result, but the problem here is lack of enablement/utility. These decisions too often apply metaphysical standards unmoored from any actual thought about pharmaceutical innovation. It’s a long-standing issue that stems from the Federal Circuit’s disregard for the utility standard set in Brenner.

  4. 3

    The irony of a battle between “your spec is too thin” and “your spec is too thick”

    Be aware of (and beware of) an answer of, “just file a boatload of more individualized applications” as that only feeds the Sport of Kings style of patent dominion.

  5. 2

    “an early-stage patentee might have its AI assistant provide a hypothetical example of each possible combination, one at a time.”

    I haven’t read the decision yet, but the genus shown has 7 substituents that are further defined in the remainder of the claim. If each substituent can have only 2 identities, that’s 2 to the seventh, viz. 128 possible combinations. But if each has 3 identities, that’s already 2187 possible combinations. If each has 4 identities, it’s 16384 (i.e. 128 squared). Practically, it’s not uncommon for there to be hundreds of thousands of possible combinations. While a machine can be programmed to generate a name or structure for each possible molecule represented by each possible combination, that’s an awful lot of pages of specification, which means an awful lot of extra page fees in some jurisdictions (including the PCT), not to mention translation fees if the compounds are listed as words rather than shown as structures.

    It’s not clear to me why ipsis verbis disclosure would fail to meet the written description requirement, provided that enough discussion of general synthetic methods is provided to show that the inventors know how to make every one of the molecules disclosed. For the believers in the bifurcation of written description and enablement, whose view currently holds sway at the CAFC, written description simply means that the applicant needs to show he was aware of what he invented, so ipsis verbis (in haec verba) would seem to do the trick in that regard. It would be more likely to fail under enablement, if specific molecules from within the many shown could be demonstrated to require synthetic methods not mentioned in the patent application.

    Which brings me to the real issue here, and that’s the higher level of disclosure required for the patentee to meet the written description/enablement requirements compared what’s required for a reference to be deemed to make something obvious. Per the CAFC, a prior art reference doesn’t have to explicitly spell out every possible combination of substituents to be deemed to render an applicant’s claimed compound (or genus) obvious, even though the level of description in that reference would itself fail to meet the written description and/or enablement requirement were that same compound (or genus) to have been claimed in the prior art reference. Some consistency here would be appreciated, but it’s been this way for a long time so I’m not expecting any relief on this front.

    Dennis’ suggestion also brings up something I thought about years ago: if someone really wanted to mess up the patenting of drugs, he could publish structures of millions of compounds that hypothetically could be made but have not been synthesized, thus posing a potential novelty challenge. If such a disclosure were to be done in the context of a patent application – as Dennis proposes to do using AI to generate the list of compounds – this could potentially kill patent protection for whole swaths of molecules. Under the current system of protection for new molecules, which is keyed to patent protection, that could create disincentives to invest in the development of many pharmaceutically promising molecules. Maybe it’s time to adopt the suggestion that protection for new molecules be given for 12 years from FDA approval, regardless of patent status.

    1. 2.1

      To address one of your points – using AI to create and then publish large swaths of potential molecules – it is already happening. There are many ways to freely publish information in a format that Examiner’s search. The website comes to mind first. Several years ago I encountered exactly this situation on that website – a huge swatch of “prophetic examples” that were anonymously published that directly read on a known research area for a specific company. Based on the writing style and other information, we were able to track the publication back to a competitor who was working on similar structures and targets. The only difference between this and your suggestion is that I don’t think AI was involved in creating the prophetic examples.


          “ New twist on the notion of ‘poison pill’….?”

          Nothing new about using computers to poison wells in the chem/bio space. It’s been going on for decades and the issues have been discussed expressly here (by myself and others) going back at LEAST ten years.


            “Using computers”…

            Is that really what creates the novelty bar destroying attempts to patent “objective physical structures”…?


            Are you that p0ssessed by your emotive rants?

  6. 1

    “ an early-stage patentee might have its AI assistant provide a hypothetical example of each possible combination, one at a time”

    Dennis Crouch, folks, promoting … something. Probably not progress.

    1. 1.1

      Adding: the “problematic” relationship between the patent system and the existence of machines that can automate information processing (including word processing) is already a half century old (at least). Rather than compound the problem, Dennis, why not just expressly recognize the problem of “computers pumping out massive reams of potentially useful information” versus “granting 20 year (lol!) monopolies based on what a few human beings who need to eat feel is right.”

      It’s an old, old issue for this blog. Attorneys have been using computers for decades to create vast universes of prior art (hello Moderna!) that is subsequently “mined” based on predictable circumstances (there is a new infectious respiratory disease? Vaccines will prevent the deaths of millions? Wow, that is so shocking!).

      Do you think your “neighbors” in Missouri (your fellow citizens) will be better off if we expand this “paradigm”? If so, tell everyone why. For a change.

      1. 1.1.1

        Looking in from Europe, I have the feeling that there is a problem in the USA but not in Europe. Chem/pharma is not my field so in what follows I am open to correction.

        Let’s start from the proposition that an AI can be programmed to output the generic formula of every new molecule XYZ, together with an exhaustive list of all members of the group consisting of molecule XYZ. How does that interfere with the opportunities to patent new and useful chemical compounds? Not at all, I should think. Something more is needed, either in the claim or in the description.

        If in the claim, then the AI’s output has yet to “plant the flag”. If in the description, then the AI’s output hasn’t announced the technical problem solved by the new moleecule and, in Europe, the only subject matter that is patentable is that which has plausibly solved an objective problem in a way that was not obvious.

        For the USA, doesn’t the solution to this WD conundrum lie in further exploration of the concept of “utility”?



          I am a bit surprised that you appear to overlook the straight up loss of novelty (as opposed to any of the versions of obviousness).

          A publication with a blanket “physical structure” (I chuckle at using Malcolm’s ‘rallying cry’), needs neither enablement nor utility to BE novelty destroying.


            I accept your point, that even if new molecule X (as such) is obvious, it might arguably be novel up to the point that somebody (or some AI thing) announces structure X.

            I suggest we wait for somebody (I hope Greg) to comment on what does or does not prejudice patentable novelty of a claim directed purely and simply to “Molecule X”. But does it matter? Does anybody still file claims directed purely and simply to “Molecule X” unaccompanied by a specification teaching that X plausibly solves some problem or other? Is there any point in doing so, with or without the new red herring issue of an AI mindlessly churning out lists of molecular structures?

            The interesting issue will occur when you give an AI structure X that works against some disease or other, and ask it: If this works, what other structures can we plausibly put forward as having the same usefulness, and why? Greg will tell us, I think, that this is already happening.


              What anyone files FOR “X” doing anything cannot overcome the bar of novelty for a publication OF “X” itself.

              That’s the point that I am surprised that you are missing.

              Certainly, patents for methods of using “X” may be (or may be not — see the ‘just following natural law’ anti-patent viewpoint) possible, but — and again chuckling at the Malcolm rallying cry — “the objective physical structure” IS a novelty bar for the structure without need for enablement or utility considerations.

              You need not wait for any pronouncement from Greg, as this is the rule of law for ALL art fields.


              … and for shade that you might be aiming for, Dvan already indicates that this is happening.

              The difference of course is that letting loss a ChatGPT to instantly fill VOLUMES (and perhaps the entire space) is all of sudden a very real poison pill possibility.

              Hence, Malcolm’s angst being very different here, as this is HIS backyard basically being taken out of the patent game.


                “Malcolm’s angst being very different here, as this is HIS backyard basically being taken out of the patent game.”

                [eye roll]

                In fact, I could care less. My “angst” is over the fact that (1) the issue of using computers to “poison the well” and to bloat a special with WD support is well-known and old, and using so-called “AI” is just more of the same; and (2) to the extent that computer-created prior art can potentially negate a lot of patent claims, this SHOULD be seen as a big problem by those folks around here who feel that patents are THE TOOL for “promoting progress.” In other words, more computer-created information —> less space for patents —> less progress (I’m assuming the mindset of the folks mentioned above).

                1. Should he seen as a big problem…?

                  How exactly?

                  What screams out in your reply, in your attempted denial, is only confirmation that my post is 100% accurate.

                  “Big problem” indeed.


              I accept your point, that even if new molecule X (as such) is obvious, it might arguably be novel

              Exactly NOT my point.

              Obtuse – is it deliberate?


              MD: “ I suggest we wait for somebody (I hope Greg)”

              Greg traditionally stops commenting here during Lent so you may have to wait until Easter.


                Had not spotted that about Greg, Phophet. But no matter. I don’t need him on this issue any more. Your succinct “bloat” point covers it nicely.

                1. Sadly, his “bloat” is as errant as your feces thrown against the wall.

                  Even more sadly, this is the exact same gamesmanship from both of you that stretches back more than a decade (for Malcolm, more than 17 years — notwithstanding his approximate 15 month hiatus before being ‘resurrected’ as “The Prophet.”


            I think there would still need to be enablement of how to produce a molecule of that structure for the publication for it to be novelty destroying


              Looks like my pal Shifty is engaging in nym-theft (not the first time he has reached for the best historical name 😉 ) and trying to cause some confusion.

              But no, enablement is NOT required for a published “objective physical structure” to be novelty-destroying.

    2. 1.2


      This is not your typical angst.

      Perhaps there is a bit of NIMBYism in the possibility of “Dennis Crouch, folks, promoting … something

Comments are closed.