There is a well-worn saying in electrical engineering — if you want to break a circuit, you can cut the wire or trip the breaker. Plaintiffs in several consolidated spinal-cord stimulator cases recently tried something more creative — they attempted to rewire the entire regulatory grid by suing the FDA itself, hoping to knock out the device manufacturer’s preemption defense before the defense could do its job. It was an audacious gambit that did not work. The court in Keys v. Medtronic, Inc., 2026 WL 2030069 (D. Minn. July 14, 2026), dismissed all but a sliver of the plaintiffs’ claims, and the novel APA strategy fizzled on the launchpad.
Keys is actually four cases decided together. All four plaintiffs were implanted with defendant’s spinal-cord stimulator, a Class III medical device that first obtained PMA in 1984 and has since been the subject of more than 400 PMA supplements. Id. at *2. The plaintiffs’ core theory was that post-2017 PMA supplements had so significantly altered the device’s design that it had become a materially different product from what was originally approved. Id. at *3. On that foundation, the plaintiffs constructed an unusual two-track attack — standard product-liability claims against the device manufacturer, and a simultaneous APA challenge against the FDA itself.
The obvious problem with suing the FDA over a PMA decision is Buckman, which forecloses private FDCA enforcement suits. Plaintiffs tried to sidestep Buckman by styling the claim as an APA “arbitrary and capricious” challenge rather than a direct FDCA enforcement action. Clever — but the court was not impressed.
The plaintiffs needed to explain what injury gave them standing to bring this FDA suit in the first place. Their theory was that the injury was the manufacturer being able to assert a preemption defense in the product-liability cases. That is, the “harm” was not anything the FDA did to plaintiffs physically — plaintiffs expressly disclaimed that theory in briefing — but rather the availability of a legal defense in a different lawsuit. Id. at *4. Standing, however, requires an injury in fact that exists at the outset of the litigation. But by plaintiffs’ own account, the “injury” they suffered — defendant invoking preemption — did not occur until defendant filed its motion to dismiss. That happened after the complaints were filed. So at the moment each plaintiff filed suit, the asserted injury did not yet exist. No injury, no standing. The APA claims were dismissed for lack of jurisdiction. Id.
The court also addressed the merits, finding the APA claims would fail regardless. Plaintiffs leaned heavily on 21 C.F.R. § 814.39(a), claiming it required a new PMA whenever cumulative changes were sufficiently significant. The court read the regulation and found that it does exactly the opposite of what plaintiffs claimed — § 814.39(a) governs when a PMA supplement is required (as opposed to a mere report to the FDA), and says absolutely nothing about when a new PMA, as opposed to a supplement, might be required. Id. at *4-5. Plaintiffs cited no statutory or regulatory provision that identifies any such threshold. Zero. The merits ruling was swift and conclusive.
The court then turned to the products liability claims. The four plaintiffs were from either Kentucky or Tennessee, states with one-year statutes of limitations for personal injury and product-liability claims. Making nearly every claim dead on arrival on timeliness grounds.
The plaintiff from Kentucky was implanted in May 2018, developed severe complications, and was advised to turn off his device by mid-2019. One of the Tennessee plaintiffs was advised to turn off her device by mid-2019; and another experienced shock-like symptoms and urinary incontinence within two weeks of her February 2020 implantation. They all filed suit in 2025.
The court found all of their claims time-barred on their face. Id. at *5-7. The fourth plaintiff brought claims related to a 2019-implanted device, which was time-barred on the same logic; but had a second device implanted in March 2024. His claims relating to the 2024 device survived the statute of limitations and moved on to preemption analysis.
Here, the court applied the familiar Riegel/Buckman framework: a parallel state-law claim must allege conduct that violates the FDCA, but the claim cannot be premised on the FDCA violation itself. The gap is narrow, and the court found it swallowed most of plaintiff’s claims.
Indeed, the lone survivor was manufacturing defect. Plaintiff alleged that his device deviated from defendant’s design specifications and included specific allegations regarding improper assembly. This is the archetypal parallel claim. Further, the Eighth Circuit has counseled leniency in dismissing manufacturing-defect claims given plaintiffs’ limited access to PMA manufacturing data. So, it is hardly surprising that the court found plaintiff had done enough for this claim to survive a motion to dismiss. Id. at *9.
Nothing else did. Plaintiffs alleged that the defendant marketed the device as safe and effective while concealing adverse events and post-market modifications. But they never alleged that defendant failed to disclose anything the PMA required it to disclose. That omission was fatal. On that ground the court dismissed several information-based claims (consumer protection, failure to warn, fraudulent concealment). Id. at *8-9.
Plaintiff’s negligence-per-se theory based on alleged failures to submit adverse-event reports was impliedly preempted under Buckman. To the extent the claim was based on failing to update labeling, it survived Buckman but failed on pleading because plaintiff never identified any specific labeling language required by the FDA that defendant omitted. Finally, to the extent plaintiff was relying on an alleged cGMP violation, it did not support a negligence per se claim, but the court left open the possibility that, if properly developed, a cGMP violation could support the manufacturing-defect claim. Id. at *9-10.
Plaintiff’s safety-and-effectiveness warranty claims were conflict preempted because a jury verdict finding the device not safe and effective would directly contradict the FDA’s PMA determination. The remaining warranty allegations failed because plaintiff never alleged that defendant warranted anything beyond what the PMA required — expressly preempted. The claims also would have failed independently for lack of pre-suit notice under Tennessee’s UCC. Id. at *10.
Finally, plaintiff made two sales representative claims. He alleged that defendant was negligent in using unlicensed field representatives to influence intraoperative decisions without ensuring informed consent. While defendant’s representatives were present during the surgery, the court dismissed the claim because plaintiff did not allege that field personnel failed to disclose anything required by the PMA. Id. at *11. Preempted.
Plaintiff’s most unusual claim alleged that defendant’s representatives committed negligence per se by violating Tennessee’s statute prohibiting the unauthorized practice of medicine. Defendant argued there is no private right of action under that statute. Plaintiff did not respond. Under Tennessee law, the burden falls on the plaintiff to establish a private right of action exists. Silence equals concession. Claim dismissed. Id.
Plaintiffs asked for leave to amend but failed to follow the district’s local rules for doing so and — more fundamentally — identified no additional facts that would save their claims. The court, unsurprisingly, declined to let them take another swing. Id.
While manufacturing-defect claims continue to be the gap in the PMA preemption armor, the novel APA strategy that was supposed to short-circuit preemption entirely got unplugged before it ever got started.
Recent Comments