GSK v. Teva: Skinny Label Approval is not a Patent Safe Harbor

In an important pharma decision, the Federal Circuit has doubled-down on its prior ruling that Teva’s sales of the drug carvedilol induced infringement of a GSK method-of-treatment patent, even though Teva’s product is approved and labeled only for non-infringing uses (as required by statute).  In the new decision, however, the court makes clear that the inducement liability is due to Teva’s additional marketing activity that encouraged others to infringe.

GlaxoSmithKline v. Teva Pharm (Fed. Cir. 2021) (GSK v. Teva 2021)

The GSK Patent covers the use of carvedilol to “decrease a risk of mortality caused by congestive heart failure.” RE40,000 (original US Pat 5,760,069). The FDA approved Teva’s sales of the same drug, but only under a “skinny label” that carves-out the patented use.  Teva only labeled its version for treatment of hypertension and left ventricular dysfunction following a heart attack (“post-MI LVD”).  Despite the label differences, the drugs are otherwise therapeutic equivalents indicated by the FDA’s “AB rating.” Thus, many people (insurance companies,  pharmacies, doctors, patients, etc) began using the cheaper generic for the patented purposes and Teva profited.

GSK sued for inducing infringement and the jury sided with GSK and awarded $200 million in damages.  The district court rejected the jury verdict and instead found insufficient evidence to show causation — that insufficient evidence Teva’s actions caused the underlying infringement.  Then on appeal the Federal Circuit flipped again and reinstated the jury verdict in its October 2020 decision.

The problem with the Oct 2020 decision is that it included an indication that a Generic could be held liable for inducement by simply selling a generic version of a drug and noting the therapeutic equivalence with the branded drug — without ever mentioning the patented use.  The Federal Circuit has now released a new decision clarifying that such a straight use of the Skinny-Label could not justify inducement liability.  In this case, however, the court found sufficient evidence that Teva went further — labelling its drug in a way that encouraged the patented use.

An important element of this decision: the FDA’s Skinny-Label Carveout approval process does not create a genuine safe-harbor for the generic launch.

The new opinion is listed as per curiam, signed by Chief Judge Moore and Judge Newman.  Judge Prost wrote in dissent.

If a brand drug company (here, GSK) has a patent on one of a drug’s uses, it tells the FDA which use is patented. In fact, it tells the FDA exactly what language from its label is covered by its patents. The FDA will then permit a generic version of that drug to come to market if the manufacturer “carves out” such use from its drug label by omitting the language that the brand drug company identified. That’s what happened here. GSK’s sworn FDA filings identified just one use as patented. So Teva carved out that use and came to market with its “skinny” label. It played by the rules, exactly as Congress intended. It sold its generic for years without controversy.


= = = =

The opinion raises the doctrine of equitable estoppel.  The argument is premised on the idea that GSK’s statements to the FDA led Teva to believe that its skinny-label would not induce infringement.  Since Teva relied upon those statements to its detriment, GSK should now be estopped from suing Teva for inducement.  This issue was not yet decided by the district court and so will be raised on remand.

= = = =

Note on the damages: $200 million was GSK’s lost profits; Teva only profited $13 million from the infringement because of its generic pricing.

= = = =

I expect that we’ll see a congressional fix following this decision.

34 thoughts on “GSK v. Teva: Skinny Label Approval is not a Patent Safe Harbor

  1. 8

    Are generics wisening the brands by providing well made and thorough arguments in the FLB (factual and legal basis) with a PIV certification even before suit is filed. If brand feels that the claims can very well be held invalid in view of cited art(s), it files for a narrowing reissue or a continuation (if prosecution is open) to bolster the claims and then files suit with heightened confidence on patent validity. It is somewhat similar to what happens when ANDA dossier is sent to brand along with FLB to explain how the claims are being circumvented by the generic product. Brand then knows what generic is missing or avoiding and then accordingly ‘instructs’ its team to cover these very aspects either in continuations or reissue patents. probably OCA/Protective orders are mere showpieces and have no deterring effect (prosecution bar or regulatory bar)?

  2. 7

    The more I study this case, the more convinced I am that the majority is correct that the exception to skinny-labeling here is very narrow and fact specific. The outcome here rather depends on the fact that the FDA made Teva amend their label to include a patented use. That occurrence, however, is not very common.

    The mistake here is the FDA’s. I feel a little bit sorry for Teva here (not a common occurrence for me). What can they do when the FDA instructs them to expand their label? The FDA is not usually this clumsy, however, so this precedent is not going to chew much into the usual skinny-label process.

    1. 7.2


      This brings up a few interesting questions – is the brand owner required to comply with the FDA “request” to update the label to include the new indications? What happens if the generic manufacturer refuses? Also, the government basically ordered the generic manufacturer to infringe the patents, so this would seem to create an interesting indemnification issue that might take the Court of Federal Claims to resolve.

    1. 6.1

      I have never quite figured out how the other two members of a three judge panel can have the cheek to sign an opinion “per curiam” when there is a dissent. One sees it often enough, however, that I have ceased to worry about the incongruity.

  3. 5

    What what drove the FDA “in 2011, [to direct] Teva to revise its label to include the CHF indication”?

    1. 5.1

      The district court opinion kind-of-sort-of explains this. GlaxoSmithKline LLC v. Teva Pharma., 313 F. Supp. 3d 582, 587 (D. Del. 2018):

      In April 2011, the FDA sent Teva a letter in response to the de-listing of certain GSK patents from the Orange Book, instructing Teva to “revise [its] labeling to include the information associated with [the de-listed] patent.” (Id. at 461-63; PTX-15) One of the patents that had been de-listed was GSK’s ‘069 patent, which had been reissued in 2008 as the ‘000 patent. (See PTX-15; Lukas Tr. at 352-53) Teva, therefore, amended its label in 2011 to be essentially a copy of GSK’s full label, thereby covering all three indications: hypertension, CHF, and post-MI LVD. (Pastore Tr. at 461-65) The ‘000 patent expired on June 7, 2015, the date the ‘069 patent was originally set to expire.

      It appears, in other words, that someone at the FDA noticed that the only patent associated with Use Code 233 had been delisted, and said to themselves “if there are no patents associated with Use Code 233, then the generic labels should be expanded to include that use as well.” Evidently, this person did not notice that ‘069 had only been delisted because it had been forfeited in favor of RE40,000.

      That is not a very satisfactory explanation, but it is all I can make out from the decision. Also, it makes (kind of) sense as to how it could happen that way.

  4. 4

    Any indication in the decision on what drove the FDA “in 2011, [to direct] Teva to revise its label to include the CHF indication”?

  5. 3

    This case looks well positioned for SCOTUS review. As for a Congressional fix however, I will believe that when I see it. There are lots of statutes in need of Congressional fixes, but Congress does not seem overmuch concerned (they are too busy inveighing against transgender athletes and woke tech companies).

    1. 3.1

      Inveighing against…?

      How about “captured by”?

      That you both advocate FOR corporatacracy and espouse Liberal Left philosophies rather colors your perceptions (with the single exception of the desired Liberal Left give-away of COVID IP (and critically not limited to patents) waiver.

      Go figure, right?

  6. 2

    Bros, I don’t even remember hearing on here that the first “xenobots” (micro I believe), or “living robots” have in fact been reduced to practice! Made em from frog embryos and they don’t really do much yet, but they can self-repair, and maybe carry stuff around.

    link to

    Big news for techies.

      1. 2.1.1

        Prob tough to find if they filed a patent app for it. Not sure if it is real world “useful” yet even. Might not even work as a doorstop just yet.


          Typically a university like Tufts will file prior to publicizing.

          It should be EASY to find.

          The Guardian may be, shall we say, embellishing.


            Don’t think they’re embellishing bruh they published in pnas. Seems to say about the same thing. Again, not sure if it is “new and USEFUL” just yet. Or they may not see any $$$ in 20 years time so maybe not worth filing just yet, might even have some other more fancy things going on in the lab they want to file on a bit later w this just as teaser.

            link to


              Thanks 6 – any suggestions for a public search of the USPTO site to see if there was a filing (perhaps with different terminology)? Any shot an examiner can share (of course public) info?


                I’ll see if I can pull up anything by the inventors n see what all is public if it’s that big a deal to you.

                1. The only one with any hits is Michael Levin and I checked his pubs through 2020 (public). Nothing there that I think is related to this. They may have filed intl app (or a prov to be followed by intl if that chain is possible, I forget) and we’re still waiting on ntl phase apps to be filed and to get publications out (18 mo n all).

  7. 1

    Just out of curiosity, who are the DIRECT infringers of a method-of-treatment patent involving taking a particular pill? Is it only the individual patients? That makes inducing infringement almost essential, but are not the main inducers the insurance companies?

    [The last sentence above may be underestimating the lobbying effectiveness of the pharmaceutical companies?]

    1. 1.1

      Direct infringer is the patient.

      Indirect infringers would include prescribing physicians, and possibly insurance companies through pricing structures that encourage prescription of the generic.


          There is only one claim in this patent, and that single method claim has only one step: “using a compound of Formula I… for decreasing mortality resulting from congestive heart failure… .” It seems to me that the physician could be said to “use” the drug to decrease mortality from CHF, but so could the patient. Both of them count as direct infringers (although the doctor is, of course, statutorily immune from liability).


            Does a Doctor’s “use” (except AS a patient) actually decrease anything?

            Direct infringement is ‘direct’ because of ALL elements of the claim, eh?

            Telling someone ELSE to directly infringe really does not make one a direct infringer, does it? (maybe Pharma is different….)


              Claim is to a method of administering. Doctors administer drugs. Can be easily proven by scripts.


                I think that the claim should recite administering, but strangely it does not. The only active step in the whole claim is “using.”

                Certainly a doctor could inject a patient with the drug, in which case you might well describe the doctor as “using” the drug to reduce CHF mortality in her patient. On the other hand, a patient might well take the drug herself, in which case one might equally well describe the patient as “using” the drug.

                Either the patient or the physician could be a direct infringer, depending on circumstances.


                As noted by Greg, a doctor MAY administer.

                But even such administering is not likely to be “using,” as only one of a doctor administering and the patient that is being administered to is actually USING.

                Greg tries a little too hard with his the Doctor is using in her patient, as the action IN the patient is NOT the action OF the doctor, and you violate the single actor rule for direct infringement.

                The ONLY way that the Doctor is a direct infringer is if the Doctor happens to ALSO be the patient.

    2. 1.2

      There was a R&R from Magistrate Judge Hall in of Delaware this week in the Vascepa litigation that raised the issue of inducement by a health insurer on patented uses not covered by a skinny label. The insurance companies will absolutely be the next targets for requesting substitution for all uses.

      I think Judge Prost has the right answer here. Congress was aware of this issue in 1984 and allowed carve-outs. If the generic company issuing press releases saying it has approval is enough to induce infringement, it’s hard to see how you can really carve uses out. The majority does a lot to try and thread a needle, but their attempts to distinguish the case seem illusory to me. Hatch-Waxman seems to resolve this issue. If you carve out an indication and do not do anything to specifically promote off-label uses, then the kinds of activity here shouldn’t be enough. GSK is upset about the fact that once an AB-rated is on the market, pharmacies can substitute the drug even if the use the doctor is prescribing it for is carved out. But Congress intended this. If they didn’t, Congress should fix it.

      This will get a strong petition for rehearing en banc and cert petition and I would not be surprised to see this overruled by the Supreme Court if the Federal Circuit doesn’t rehear it and reverse if.

      1. 1.2.1

        This was not a Hatch-Waxman case. It was a straightforward case involving a post-approval launch tried to a jury. GSK convinced the jury that Teva was unsuccessful in carving out the protected indication from its label. That alone is enough support a finding of inducement. The sky is not falling. SCOTUS will not get involved.


          It appears (without diving into the case and merely reflecting the write-up and comments here) that “carve-out label” ALONE should be dispositive.


          That’s certainly how the majority framed it. It was a one-off case, based on particular facts, no big deal. And certainly no larger implications for the mine run of skinny label generic launches. That’s what they’d say I think.

          The dissent basically accused them of trying to have their cake (infringement in this specific case) and eat it too (nothing for a typical skinny label user to worry about). Per the dissent, the problem is that nothing really distinguishes this case from what any other skinny label generic would be doing. So, as Prost said, they failed to thread the needle. Moreover, impeding the use of skinny labels in this fashion goes against the manifest intent of Congress.

          I tend to find the dissent more convincing, but YMMV of course.

Comments are closed.