When a 'Skinny Label' Is Not Skinny Enough: GSK v. Teva and the Limits of the Section viii Carve-Out
The Federal Circuit's reinstated $235 million verdict in GSK v. Teva tests whether a generic's carve-out label can shield it from induced infringement of a method-of-treatment patent.
In GlaxoSmithKline LLC v. Teva Pharmaceuticals USA, Inc., No. 2018-1976 (Fed. Cir. Aug. 5, 2021), the U.S. Court of Appeals for the Federal Circuit confronted one of the most consequential questions in Hatch-Waxman practice: whether a generic manufacturer that uses a “skinny label” (a label that carves out an indication covered by an innovator’s patent) can nonetheless be liable for inducing physicians to infringe a method-of-treatment claim. In a precedential 2-1 rehearing decision, issued as a per curiam opinion for the court by Chief Judge Moore and Judge Newman with Circuit Judge Prost dissenting, the panel reinstated a jury verdict of approximately $235 million against Teva for induced infringement of GSK’s reissue patent on the use of carvedilol to treat congestive heart failure. The Supreme Court denied certiorari on May 15, 2023, in No. 22-37, leaving the decision as the governing authority on the outer limits of the statutory carve-out.
The case has been read, alternately, as a narrow substantial-evidence ruling tethered to unusual facts and as a structural threat to the entire skinny-label system Congress built into the Hatch-Waxman Act. Both readings have purchase, and the tension between them is precisely why the decision still drives ANDA strategy years later.
At a glance
- Caption: GlaxoSmithKline LLC v. Teva Pharmaceuticals USA, Inc., No. 2018-1976 (Fed. Cir. Aug. 5, 2021); cert. denied, No. 22-37 (U.S. May 15, 2023).
- Patent: U.S. Reissue Patent No. RE40,000 (reissue of U.S. Patent No. 5,760,069), claiming a method of decreasing mortality caused by congestive heart failure by administering carvedilol in conjunction with one or more other therapeutic agents selected from the group consisting of an ACE inhibitor, a diuretic, and digoxin.
- Branded product: Coreg (carvedilol), approved for hypertension, congestive heart failure (CHF), and post-myocardial-infarction left ventricular dysfunction.
- Mechanism at issue: A Section viii “carve-out” or “skinny label” under 21 U.S.C. § 355(j)(2)(A)(viii), which lets a generic omit a patented method of use from its labeling.
- Holding: Substantial evidence supported the jury’s finding that Teva induced infringement during both the “skinny label” period and the later “full label” period; judgment as a matter of law was improperly granted.
- Vote: 2-1. Per curiam opinion for the court (Chief Judge Moore and Judge Newman); dissent by Circuit Judge Prost.
- Disposition: District court’s JMOL vacated; jury verdict and damages award (~$235 million: $234.1 million in lost profits plus $1.4 million in reasonable royalties) reinstated.
The Hatch-Waxman bargain and the skinny-label carve-out
To understand why GSK v. Teva mattered so much, start with the statutory architecture. Under the Hatch-Waxman Act, a generic seeking approval files an Abbreviated New Drug Application (ANDA) that piggybacks on the innovator’s clinical data and ordinarily must copy the brand’s approved labeling. But Congress recognized that a single drug is often approved for multiple indications, some patented and some not. Section viii of 21 U.S.C. § 355(j)(2)(A) allows a generic to “carve out” patented methods of use from its label, certifying that it does not seek approval for the patented indication. The resulting “skinny label” is supposed to let a generic enter the market for the unpatented uses while leaving the patented use to the innovator until its method-of-use patent expires.
That bargain is central to generic competition. Carvedilol had three approved indications; GSK’s patent claimed only the CHF method of use. Teva accordingly filed its ANDA with a label that omitted CHF, retaining only hypertension and post-MI LVD. On its face, that is exactly the conduct Section viii was designed to permit. The pivotal legal question was whether doing the carve-out, by itself, immunizes the generic from inducement liability under 35 U.S.C. § 271(b). The answer, the majority held, is that it does not, because inducement turns on the totality of the inducing conduct, not on the label alone.
Induced infringement: the “partial label” and “full label” periods
The Federal Circuit’s analysis divided the relevant timeline into two periods, and the distinction is essential to reading the opinion correctly. During the early “partial label” or “skinny label” period (2007-2011), Teva’s label omitted the CHF indication. During the later “full label” period (after 2011), the FDA directed Teva to amend its label to match GSK’s, restoring language that the jury could find described the patented CHF method.
The majority concluded that substantial evidence supported inducement across both periods. The critical point for litigators is what the court relied on: not the skinny label in isolation, but Teva’s affirmative marketing. Teva’s Spring 2008 and Spring 2009 product catalogs described its carvedilol as an AB-rated therapeutic equivalent to Coreg, and press releases from 2004 and 2007 touted the product as indicated for treatment of heart failure and hypertension and as a generic version of Coreg. Combined with expert testimony that the partial label met each claim limitation and represented to doctors that the treatment decreased mortality caused by CHF, the jury could find that Teva encouraged the patented use. The court stressed that this was a “narrow, case-specific review of substantial evidence” that “does not upset” the Hatch-Waxman balance. On the safe-harbor point, the court quoted an amicus brief filed by Novartis, which said generics “could not be held liable for merely marketing and selling under a ‘skinny’ label omitting all patented indications,” and then agreed that Novartis had accurately stated the law, adding that Teva’s conduct fell outside that description because Teva did not omit all patented indications.
That qualification is the load-bearing sentence of the opinion. The majority insisted it was not holding that a clean carve-out triggers liability; it was holding that on this record, the carve-out was incomplete and the marketing filled the gap.
Judge Prost’s dissent: the systemic risk
Judge Prost’s dissent, which she maintained through both the 2020 panel opinion (issued while she was Chief Judge) and the 2021 rehearing, framed the decision as a betrayal of the Hatch-Waxman bargain. Her concern was practical: if a generic that carves out a patented indication can still be hauled before a jury on a fact-intensive inducement theory built around routine AB-rating statements and standard generic marketing, then the carve-out provides no reliable safe harbor at all. Generics, she warned, cannot control how physicians prescribe, and holding them liable for the foreseeable consequences of lawful substitution erodes the predictability Section viii was meant to provide. The dissent’s through-line is that Congress made a policy choice favoring generic entry for unpatented uses, and the court should not let inducement doctrine swallow that choice.
The disagreement is not really about the law of inducement, which both sides accept requires specific intent and affirmative encouragement. It is about whether the evidentiary threshold the majority tolerated is so low that it converts ordinary generic conduct into a jury question every time.
Open questions
- How clean must a carve-out be? The majority disclaimed liability for generics that omit “all patented indications,” but the carvedilol label retained residual language the jury tied to the patented method. The opinion gives no bright line for when label text crosses from permissibly generic to inducing.
- Does AB-rated marketing inherently carry inducement risk? Press releases touting therapeutic equivalence were central to the verdict. Whether such standard statements can ever, standing alone, support inducement remains contested.
- Who bears the carve-out drafting risk? When the FDA compels a label amendment that reintroduces patented language, is the resulting exposure the generic’s, the agency’s, or the innovator’s problem? The decision leaves the allocation unresolved.
- Will a future panel cabin or extend GSK v. Teva? Subsequent skinny-label disputes (including litigation over different drugs) continue to test whether the case is truly sui generis.
Implications
- For generics: A Section viii carve-out is necessary but no longer sufficient. Counsel must scrub not only the label but press releases, monographs, sales materials, and equivalence statements for anything a jury could read as promoting the patented use.
- For innovators: Method-of-use patents gained meaningful enforcement teeth. A well-papered record of a generic’s marketing can defeat a carve-out defense and support substantial damages.
- For litigators: The case is a substantial-evidence decision. Outcomes will turn on the trial record (expert testimony, marketing documents, and physician-prescribing evidence) far more than on abstract carve-out doctrine.
- For FDA practice: Agency-compelled label amendments can create infringement exposure a generic did not choose, sharpening the need to coordinate regulatory and patent strategy.
- For deal and risk teams: Skinny-label launches now warrant heightened diligence; the predictability that once made carve-out launches relatively low-risk has been materially reduced.
Frequently asked questions
Did GSK v. Teva eliminate skinny labels? No. The majority expressly preserved the carve-out and said generics are not liable for merely selling under a skinny label that omits all patented indications. The holding is that Teva’s particular conduct (an incomplete carve-out plus affirmative marketing) supported inducement on the record before the jury.
Why was the verdict reinstated rather than affirmed outright? The district court had granted Teva judgment as a matter of law, setting aside the jury’s verdict. The Federal Circuit held that JMOL was improper because substantial evidence supported the jury, so it vacated the JMOL and reinstated the ~$235 million verdict and damages award.
What did the Supreme Court do? It denied certiorari on May 15, 2023 (No. 22-37). The order noted that Justice Kavanaugh would have granted review. The denial left the Federal Circuit’s decision in place without a merits ruling.
Authorities and sources
- GSK v. Teva: Federal Circuit Opinion After Rehearing, Cooley LLP: https://www.cooley.com/news/insight/2021/2021-08-17-gsk-v-teva-federal-circuit-opinion-rehearing-induced-infringement-liability-skinny-label
- GlaxoSmithKline LLC v. Teva Pharmaceuticals USA (Fed. Cir. 2021), Patent Docs: https://www.patentdocs.org/2021/08/glaxosmithkline-llc-v-teva-pharmaceuticals-usa-fed-cir-2021.html
- Federal Circuit Vacates Judgment, Reinstates Jury’s Verdict of Induced Infringement, Jones Day: https://www.jonesday.com/en/insights/2021/09/federal-circuit-vacates-judgment-reinstates-jurys-verdict-of-induced-infringement
- SCOTUS denies cert in skinny label appeal, A&O Shearman: https://www.aoshearman.com/en/insights/ao-shearman-on-life-sciences/scotus-denies-cert-in-skinny-label-appeal-from-the-federal-circuit
- The Fate of the Skinny Label, Foley Hoag: https://foleyhoag.com/news-and-insights/publications/alerts-and-updates/2023/may/the-fate-of-the-skinny-label-teva-pharmaceuticals-usa-inc-v-glaxosmithkline-llc/
- Opinion after rehearing, GlaxoSmithKline LLC v. Teva Pharmaceuticals USA, Inc., Nos. 2018-1976, 2018-2023 (Fed. Cir. Aug. 5, 2021), U.S. Court of Appeals for the Federal Circuit: https://www.cafc.uscourts.gov/opinions-orders/18-1976.opinion.8-5-2021_1814836.pdf
- Order denying rehearing en banc (Feb. 11, 2022), U.S. Court of Appeals for the Federal Circuit: https://www.cafc.uscourts.gov/opinions-orders/18-1976.ORDER.2-11-2022_1907121.pdf
- Supreme Court docket, Teva Pharmaceuticals USA, Inc. v. GlaxoSmithKline LLC, No. 22-37: https://www.supremecourt.gov/docket/docketfiles/html/public/22-37.html