Federal Circuit Draws a Hard Line Against “Obvious to Try” Analysis

by Dennis Crouch

Teva Pharmaceuticals, LLC  v. Corcept Therapeutics, Inc., Dockt No. 21-1360 (Fed. Cir. 2021)

This is an interesting pro-pharma obviousness decision coming out of the PTAB regarding obviousness of particular drug dosages.  Here, the particular drug was known to work well, but there were concerns about drug safety.  And, there were a couple of particular tests that clearly would have been obvious to try.  In fact, the patentee was legally required by the FDA to conduct the tests in order to ensure drug safety.  Still, the outcome of the tests were not predictable. Nobody knew whether the particular dosage being was going to turn out to be “safe.” (Here “safe” is interpreted as having “an acceptable risk-benefit profile.”)  The patentee ran the tests, found the dosage to be safe, and then patented a method of administering that safe-dosage to patients.  Teva challenged the patent, but the PTAB and the Federal Circuit both sided with the patentee. They held that the unpredictable outcome of the test meant that the outcome was not obvious since there was no “reasonable expectation of success.”

I expect that the Supreme Court would reject the hard line drawn in this case in the same way that it rejected the TSM requirement in KSR v. Teleflex.  I.e., obviousness is a flexible, open analysis asking “is there an invention here?”  Former Kennedy clerk J.C. Rozendaal handled the appeal for Teva, and so I won’t be surprised to see a petition for writ of certiorari in 2022.

The rest of the story: Corcept’s patent covers a method of treatment of some Cushing’s syndrome  symptoms with the drugs mifepristone and ketoconazole (a “strong CYP3A4 inhibitor”).  U.S. Patent No. 10,195,214.  The basic problem is that these two drugs can interact in problematic ways, and the patent calls for a reduced dose of mifepristone when being taken alongside ketoconazole.  When taken alone, mifepristone is prescribed in dosages of up to 1,200 mg per day; but when taken with ketoconazole, the patent calls for reduction of down to 600 mg per day.

The prior art all comes from earlier interaction between Corcept and the FDA.  Initially, Corcept filed a New Drug Application (NDA) seeking to market mifepristone in dosages of 300, 600, 900, and 1,200 mg per day.  The FDA gave its approval, but suspected interaction with ketoconazole.  That suspicion led to the two key prior art references:

  1. The Korlym Label: The FDA-approved label for Corcept’s original drug product Korlym allowed for dosage of mifepristone at 300, 600, 900, or 1,200 mg increments and also offered a warning against co-administration with a strong CYP3A.  In particular it limited the “mifepristone dose to 300 mg per day when used with strong CYP3A inhibitors.”  Thus, an easy reading of the label seems to be a strong suggestion that dosages of 600, 900, or 1,200 might be problematic.  This label also appears to totally anticipate a dose of 300 mg when taken along with a strong CYP3A inhibitor.
  2. The FDA Suggestion (“Lee”): The FDA also suggested the possibility of interaction between CYP3A and mifepristone and required Corcepts to study that interaction.

When I think about these two references, I see a strong suggestion of an interaction between the drugs, and also a strong motivation to look into that interaction. In fact, the first two research questions implicitly demanded by the prior art is (1) whether a 300 mg dose interacts with CYP3A and (2) whether a 600 mg dose interacts with CYP3A.  Yet, while the two references expressly call for research, the references do not suggest the outcome of the 600 mg test.  If anything, the original label suggested that 600 mg dose would be a problematic — since the FDA limited dosage to only 300 mg if also taking the other drug.

The PTAB followed these lines of analysis and eventually concluded that the references failed to prove obviousness.  In particular, the Board hung its hat on “reasonable expectation of success” — finding that the references did not suggest that 600 mg would be a safe dosage.

On appeal, the Federal Circuit has affirmed — holding that invalidating the claim would require proof of “a reasonable expectation of success in achieving the specific invention claimed, a 600 mg dosage.”

Because there was no expectation of success for any dosage over 300 mg per day, there was no expectation of success for the specific 600 mg per day dosage.

Slip Op.  Affirmed.

Note – Mifepristone is new name for the drug RU-486.

43 thoughts on “Federal Circuit Draws a Hard Line Against “Obvious to Try” Analysis

  1. 5

    I don’t find anything wrong with the holding. People often make too much of motivations (although granted, you usually see it in the opposite direction). KSR took every opportunity to run away from the concept of motivation. Rather the focus is on 1) CAN a change be made? and 2) Are the results of making the change EXPECTED? Motivation is a vestige of the improper textual motivation for combination that KSR swept away.

    The concern here is that the motivation is essentially mandated. But that doesn’t change the fact that the outcome was surprising. It serves the purpose of the patent act to reward for disclosing. In the same way you can cite A&P Tea Co to essentially void a motivation attack, it is not surprising you can bank on surprise results or nonenablement to counter an extreme motivation.

    1. 5.1

      Motivation is a vestige of the improper textual motivation for combination that KSR swept away.

      Not withstanding that ‘motivation’ is an absurdly easy bar to reach, I take it that you have no clue as to what you are talking about.

      As usual.

  2. 4

    I am not sure why Prof. Crouch flags this as a particularly interesting case. The PTAB holding and Federal Circuit affirmation thereof is consistent with numerous prior decisions in the medicinal arts. See below for a few examples.

    Case law has clearly established that “obvious to try” does not equate with an “invitation to experiment.”

    KSR discussed the “obvious to try” rationale. KSR involved a “combination of familiar elements.” The mechanical device made by combining known components to produce a combination having the properties of the known components. Key here is that the resulting combination of known mechanical components produced a “machine” with predictable/known properties. The completed machine was not greater than the sum of its parts.

    In contrast, an invitation to experiment or as Prof. Crouch puts it “a strong suggestion of an interaction between the drugs, and also a strong motivation to look into that interaction” is different. An invitation to experiment, one investigates, via scientific method, variables. Of course one has a hypothesis (an outcome) to test against, but such outcome is not predictable (or known). Furthermore, even though your experiments conducted in your scientific method may be routine and “obvious to try/conduct,” it does not follow that the outcome of those experiments are obvious.

    Where an invitation to experiment rationale has been successful in supporting an obviousness rejection is when the facts establish that the the outcome of the experiments would have been predictable/known.

    See e.g., Sanofi-Synthelabs v. Apotex, Inc.,550 F.3d 1075, 1090 (Fed. Cir.2008, 2009), holding the claims patentable because the result of what was obvious to try was unpredictable.

    Proctor & Gamble Co. v. Teva Pharmaceuticals USA, Inc., 566 F.3d 989, 996 (Fed. Cir. 2009), holding the claims patentable and citing O’Farrell for its discussion of when an obvious to try approach may nevertheless lead to an unobvious invention (Id. at 996-997)

    1. 4.1

      The completed machine was not greater than the sum of its parts.

      Second time in short order that suggestions of synergy ARE required (and the converse, that amalgamations are not allowed).

      Where is Random for his reprise on the MPEP section that says otherwise?

    2. 4.2

      I agree that this case is in line with other cases holding that “obvious to try” does not equate with obviousness. Here, though it seems to me that the facts are more compelling in this case since they were legally required to do the testing.

      1. 4.2.1

        Even if changed from “obvious to try” to “compelled to try” (or as you put it, “required to try”), the legal distinction remains that ‘to try’ and ‘to succeed’ are not the same.

        I think a better thrust would have been to emphasize the “limited number” aspect in that in this case not only was there a limited number, the putative inventor was led to that number by the hand, thereby even raising a question as to who is the actual inventor (in the sense that being told by the government to do something (regardless of predictability of the act so directed) makes the doer into a mere follower of orders from ‘the inventor’).


          I think a better thrust would have been to emphasize the “limited number” aspect in that in this case not only was there a limited number, the putative inventor was led to that number by the hand, thereby even raising a question as to who is the actual inventor (in the sense that being told by the government to do something (regardless of predictability of the act so directed) makes the doer into a mere follower of orders from ‘the inventor’).

          Ugh I agree with anon.

          The question is not whether this is non-obvious, it clearly is. The only real interesting question here is who is the inventor of this invention.


            One thing to note is that it matters that this is an AIA patent. Under pre-AIA §102(f), the fact that it might really have been invented by the FDA staffers could make this information prior art (even for §103 purposes) relative to these applicants. Of course, pre-AIA, Teva could not have filed an IPR.


            For all those worrying about FDA contributing to the invention – What typically happens in these situations is that the FDA looked at the data in the original NDA submission and concluded that there were too many adverse events for their comfort at the >300 mg of mifepristone when co-administered with a CYP3A inhibitor. The FDA would have told Corecept that without further studies, were are going to force you to have this “warning/limitation” on your product label. So instead of a limited label, Corecept goes back and conducts a study to “expand” the label.

            Here’s the timeline. Korlym approved 17 Feb. 2017 with the label limitation. Corecept begins formulating a new clinical study to investigate the drug/drug interaction. They file two provisionals in March 2017 (priority cases to the patent at issue 10,195,214) prior to posting their clinical trial with CT.gov (required by law to be public). They convert the provisionals to a NP in June 2017. The examples in the specification appear to be the treatment arms of the clinical trial. The clinical trial is initiated in Aug. 2017 which is a study in 48 patients with 300 mg mifepreistone drug interaction study. See link to clinicaltrials.gov
            Results obtained in Feb 2018. Patent grants Feb 2019. Corecept updates the label via supplemental NDA in 5 Nov. 2019.


              So in other words, they either filed before they had possession of the ultimates claims or where directed to achieve the possession of the ultimate claims or both.

              Your post only confirms the arguments that I have presented.


                “filed before they had possession of the ultimate claims” We have been over this before and I don’t 100% disagree with you. But reality is that the powers that be deemed that the public should be made aware of all drug clinical trials. How do I file a patent on my method of treatment when I don’t have the results, but I am forced by law to publish the study? What Corecept did was standard practice. You have any better idea?

                “where directed to achieve the possession of the ultimate claims” I would suggest that a mere invitation to experiment is insufficient to have possession of the invention such that the FDA scientists wouldn’t be inventors. I view the FDA as simply “Corecept, you need to do further studies in doses over 300 mg when use din combination with CYP3A inhibitors.” The FDA didn’t create the clinical protocol. That was all Corecept.

                1. But reality is that the powers that be deemed that the public should be made aware of all drug clinical trials

                  Such simply does not create a carve-out for Pharma in the patent law arena.

                  Standard practice or otherwise (and this need not be ‘corrected’ by me).

                2. The FDA didn’t create the clinical protocol. That was all Corecept.

                  Fair enough on that point.

                3. [F]iled before they had possession of the ultimate claims…

                  Wait a minute. In what sense did Corcept not have possession as of their filing date? One has possession as soon as one can describe how to make and use the claimed invention with enough detail for a skilled reader to practice the claimed invention. Actual reduction to practice is not necessary, so the fact that they had not completed actual reduction to practice as of their filing date is an irrelevant observation. The fact that they filed an enabling disclosure is proof that they had possession as of the filing date.


              Here’s the timeline. Korlym approved 17 Feb. 2017 with the label limitation… They file two provisionals in March 2017…

              Right, so the timeline is entirely consistent with the idea that FDA personnel contributed to the conception of this claimed invention. That is not decisive either way, but the door remains open for the idea that this patent’s inventorship is not accurate and complete.

              This decision is only an appeal from a PGR. Presumably there is a parallel Hatch-Waxman litigation running. No doubt, if there is any there there regarding the inventorship issue, we will hear more about it in the Hatch-Waxman verdict.


                we will hear more about it in the Hatch-Waxman verdict

                Only if the ‘direction’ from the FDA rises to the level of inventorship.

                xtian makes a fair point on that subject.

      2. 4.2.2

        If the legal requirement to do testing was a factor in the obvious to try analysis, I think it would consume the obvious to try analysis.

        For example, I could argue all experiments conducted to gain FDA drug approval are obvious to try because the experiments are required by law, e.g., toxicology studies to find the maximum tolerated dose or phase III studies in a patient population. These required-by-law tests would render obvious any claim to a dosage amount to treat a disease and any medical use claim. I personally think this should NOT be a factor. Like the courts, the analysis shouldn’t be on how hard it was to get to your invention, but whether the invention itself was obvious.

        To tie-in Max’s question on expectation of success, all experiments have an expectation of success. Some experiments are easy to conduct (cheap materials, short experimental time, conducted on a bench top, etc). Others are not (a Phase III clinical study in Alzheimer’s patients lasting 7-10 years and costing millions of dollars across multiple clinical sites). Whether the experiments are hard or easy to conduct should have no bearing on whether the result of the experiment is obvious.

        A silly analogy is a COVID test kit. If I want to determine whether I have COVID, it would be obvious to try a rapid COVID test. Does the fact that it is obvious to try a COVID test render obvious that I have COVID? No. Absent other facts, a negative test is just as likely. Similarly, does the act of taking a COVID test create a reasonable likelihood of me successfully having COVID? No.


          I agree that this is tricky here and I’m not calling for a rule that turns every suggestion into a patentability rejection.

          Although a different context, I see the situation as somewhat similar to a researcher who hires a lab-tech to run several tests based upon a hunch about the outcome. The researcher provides the lab-tech with the details and instructions of how to conduct the experiment and what to look for in terms of outcomes. When the experiment is successful, it will be the researcher who is the inventor, not the lab-tech.


          Since we seem to be having a worthwhile exchange here, on expectation of success and obvious to try, I want to bring in the idea of scope of claiming based on a small number of Examples in the patent application as filed. Depending on the scope of the claim, there is or is not a “reasonable” expectation that the utility delivered by the Examples will also be delivered by other molecules within the scope of the claim. Any such expectation will determine whether the claim is over-broad or covetous, or nothing more than a fair scope of protection in return for the contribution to the art.

          Take as a simple illustration the recent case litigated all over Europe, where the only example was the di-sodium salt, with the competitor using the di-potassium salt. Likewise, in any “obvious to try” obviousness enquiry, I say, the legal answer is decisively dependent on the unique matrix of fact that is present.

          Are we agreed on that, I wonder.


            If we are talking US law, it seems you are conflating ‘obvious to try’ with ‘enablement’ and ‘scope of equivalents?’

            I am not familiar with the EDTA (di-sodium salt) case. If you provide me with more details, I will review and provide my thoughts.


              Xtian, yes, I am conflating because i) that’s what is going on in Europe ii) the issue of “plausibility”, now a hot topic in Europe, and and iii) I don’t see how one can avoid spill-over of thoughts on obviousness into thoughts on the DoE. See Link below and reference to the “Formstein Defence”.

              link to taylorwessing.com

              The case on the equivalence of sodium and potassium is the Lilly ALIMTA pemetrexed case. Link for you to come, in a follow-up post.


              Xtian, two replies of mine both blocked by the filter. Perhaps because they include a Link. The sodium case is the Lilly litigation on ALIMTA ie pemetrexed.


              Yes, I am conflating. That’s because I don’t see how to avoid it, whenever the party accused of infringement by equivalent runs the defence that their act was an obvious one. Thus, the UK has its “Gillette” Defence and Germany (a jurisdiction that bifurcates infringement and validity trials) its “Formstein Defence”.


              XIAN, the case I have in mind is the Lilly ALIMTA pemetrexed litigation and the UK supreme court decision within it.

      3. 4.2.3

        Dennis, I’m with Xtian on this point. I do NOT find such facts any more compelling. Were the experts at the FDA of a mind that it was obvious what the further test results would be? Surely not.


          You might want to inform yourself of the terrain. Start with the Act of 1952 (or better yet, pay attention to my posts).

    3. 4.3

      xtian, you make it sound as if there is a binary difference in patentability, as between the predictable arts and the unpredictable arts. That which is “obvious to try” is not patentable in mech/EE. However, within the unpredictable arts, including in your specification the result (which could not have been predicted) gives you a ticket to patentability, regardless what result you got, because that result is progress, and it is the discovery of the result which is your patentable contribution to the art.

      I’m still wondering when an expectation of success (in the unpredictable arts) reaches the “reasonable” threshold that neuters patentability.

    4. 4.4

      The “why this?” aspect of Prof. Crouch’s post is explained by a popular adage, viz.: Publish Or Perish. One day he was shaving some whiskers while fretting about how to destroy some new whitespace. And then — drum roll — Inspiration Reared Its Ugly Head, spewing the usual patter about how a truly gumption-laden soul can always sharpen his/her pencil and “git ‘er done” by way of perspicaciously converting those molehills into mountains.

  3. 3

    I’m not in the pharma space but I am wondering what people understand by “reasonable” in the context of “obvious to try”. I mean, if the proposition is “unreasonable” to try, how do you persuade management to fund trials?

    Putting it another way, does not the supply of funding the proposition itself prove that it was not unreasonable to try the proposition?

    Or does pharma routinely invest in prospects for which it is unreasonable to expect success?

    1. 3.1

      I’d say it depends on “reasonable expectation of success”…often, tens, hundreds, or thousands of compounds are tried, and while there is an expectation that one or more of them will have a desired outcome, it remains impossible to predict a priori which one that is – so any individual compound would not itself have a reasonable expectation of success, but the experiment as a whole would be expected to find at least one compound that is useful.

  4. 2

    There is a serious problem with this decision. The safety of the combination of the reduced dosage of mifepristone and the CYP3A inhibitor is not claimed. That safety profile may be relevant to FDA approval, but, if it isn’t claimed, it is not relevant to patentability. The Fed. Cir. missed the boat on this one. The preamble, assuming it is limiting, only requires treating the different medical indications. “Treating” in the specification is defined extremely broadly such that alleviation of any single symptom would be considered “treating”. Safety is one possible way to define the treatment, but it is not required based on the way the definition is written. A much more focused definition could eliminate that problem, but I think such a broad definition should have lead to a different result in this case.

    1. 2.2

      The safety of the combination of the reduced dosage of mifepristone and the CYP3A inhibitor is not claimed.

      This is not a relevant consideration. “From the standpoint of patent law, a compound and all of its properties are… one and the same thing,” (In re Papesch, 315 F.2d 381, 391 (C.C.P.A. 1963)) which is why “no law requir[es] that unexpected results relied on for patentability be recited in the claims…” (In re Merchant, 575 F.2d 865, 869 (C.C.P.A. 1978)). The patent in question here claimed the drugs, the dosage, and the dosing schedule. The effects of those parameters are inherent properties of the drugs, and thus implicit in the claims even if not explicitly recited.

      1. 2.2.1

        I think that his point was different – that the FDA “item” was not ‘actively’ present in the claim.

        I happen to be ‘with you’ on this – but this (by and large) works AGAINST the Big Pharma practices (e.g., filing prior to having actual possession).

Comments are closed.