Photo of Bexis

The Legal Intelligencer has a story today about the preemption loss in Knipe v. SmithKline Beecham. Since that adverse decision has made its way into the legal version of “mainstream media,” that means, under our self-imposed guideline of “first, do no harm” to the defense position in such matters, we’re now free to blog about it.
These restrictions also mean that we’ll probably never be the first to mention quite a few adverse cases. For instance, we saw Knipe on Pharmalot last week. But our objective isn’t to be first (although we like it when we are), but rather to provide our thoughts and hope that our readers might find them useful. If we have to blog with one arm behind our backs, so be it. As defense lawyers, our first duty is to our clients.
But blogging about Knipe is doubly difficult because one of us had some prior involvement in Paxil litigation. That means that we can’t talk about the facts, and in Knipe the facts are both discussed at great length (it’s 55 pages!) and are of considerable importance.
So with this one, we’ve got two arms tied behind our backs. But like Monty Python’s Black Knight (here’s a link for the culturally deprived who don’t get the reference), we’ll just keep at it anyway.
“Tis but a flesh wound.”
Staying away from the facts in Knipe, the first thing that strikes us about the case is how much things have changed in preemption-land since we first started practicing. It used to be that the other side had the distinct advantage in terms of generally applicable principles. They had arguments that, if accepted (and they often were), would dispose of many cases at once. These were arguments like, “tort claims aren’t regulatory and can’t be preempted,” “conflict preemption only reaches ‘positive’ state enactments,” “‘requirements’ (or similar words) in preemption clauses don’t refer to common-law standards imposed by the judiciary,” “there’s a presumption against preemption,” and “Congress couldn’t have meant to preempt tort claims without specific comment.”
We, on the other hand, mostly had to fight preemption battles in the interstices – urging courts to look at particular – often peculiar – sets of facts, and urging them to find that there‘s gotta be preemption here because of the obvious conflict with what the FDA wanted the client to do. In short, we’d be arguing facts and the other side would be arguing law.
Back then we usually lost.
But how things have changed. One by one, the legal propositions that the plaintiffs trotted out for years in preemption cases have been done in by the Supreme Court. Tort claims are no longer “different” in any way that exempts them from preemption generally. With “requirements” language, the court in Riegel has gone the other way and announced that, “requirements” will be interpreted generally to include tort cases unless there’s some good, statute-specific reason not to. Riegel v. Medtronic, Inc., 128 S.Ct. 999, 1008 (2008). Riegel also put some dents in the “can’t assume Congress intended to do that” line with the retort that “operation of a law enacted by Congress need not be seconded by a committee report on pain of judicial nullification.” Id. at 1009.
And the generality of the presumption against preemption? It was ignored altogether in not only Riegel but Geier v. American Honda Motor Co., 529 U.S. 861 (2000). And the Court shot a big hole in it in Buckman Co. v. Plaintiffs’ Legal Committee, 531 U.S. 341, 347-48 (2001), specifically holding that it doesn’t apply to situations (such as refereeing what manufacturers tell the FDA) where the states don’t have a legitimate interest. There are good arguments that no presumption should apply in conflict cases where preemption arises by direct operation of the Supremacy Clause. The Third Circuit acknowledged the strength of these arguments in Colacicco v. Apotex, Inc., 521 F.3d 253, 263 (3d Cir. 2008), and those arguments that are before the Supreme Court in the Wyeth v. Levine case.
So what does all this have to do with Knipe, anyway?
Well, the turnaround from just a few years ago is striking. The main reason the Knipe opinion goes on for 55 pages is that, to avoid the logic of preemption in a post-Colacicco world (for anybody who doesn’t know, Knipe is by a Third Circuit district court that’s bound to follow the Colacicco precedent), the court had to rely upon an excruciatingly detailed discussion of the facts. It’s the first post-2006 drug case rejecting preemption arguments despite: (1) not relying strongly upon a presumption against preemption, slip op. at 23 (“although the outcome of this case turns ultimately on the existence or non-existence of an actual conflict between state and federal law, the general presumption against preemption remains viable”), and (2) giving some (albeit low level) deference to the FDA’s 2006 preemption preamble. Slip op. at 32-33 (Skidmore deference).
With just about all of the other side’s broad preemption legal principles either gone or questionable, Knipe had to do what our side used to have to do when we were starting out – rely on nice factual distinctions to distinguish its way around otherwise controlling legal principles. In short, it’s plaintiffs, not us, who are now engaged in guerilla warfare in the preemption fight.
In Knipe the court essentially atomized the preemption issue – producing a holding that’s extraordinarily case specific. The ruling is specific to a particular (off-label) patient population, slip op. at 29-30 (distinguishing Colacicco on facts), 41-42; specific to a particular moment in time, slip op. at 35-36, 37 n.26, 42-47; and even specific to a particular company’s product. Slip op. at 47-49 (distinguishing Kallas amicus brief because the FDA was discussing a different drug in the same class).
From any perspective, that’s cutting the salami pretty thin. But that’s what plaintiffs have to do to avoid preemption when they don’t have any broad legal propositions left. If our side wins in Levine, then Knipe is preview of what a post-Levine preemption world might look like. To a sympathetic judge, almost any case is distinguishable.
All this emphasis on what the FDA had and hadn’t done at the precise moment of this particular suicide results in some rather unusual and misleading nomenclature. After (quite properly) identifying the defendant’s position as involving solely implied conflict preemption, slip op. at 21, the court goes on to describe its preemption analysis as involving “express preemption.” E.g., slip op. at 26 (heading 1). That’s because the court sets out on a search for “express” FDA administrative actions as of a particular date (something that, of course, will vary from case to case). Slip op. at 29. Guerilla legal warfare – producing opinions so fact bound as to have little precedential value.
The second interesting aspect of Knipe is how the opinion grabs hold of the Fellner case, Fellner v. Tri-Union Seafoods, L.L.C., 2008 WL 3842925 (3d Cir. Aug. 19, 2008), like a drowning man to a life preserver. While Fellner is certainly precedential, at one point this death grip gets just a little too tight. Knipe uses the “informal” FDA opinion rationale in Fellner as a basis for ignoring the FDA’s opinion as expressed in its prior amicus brief in the Kallas case, which involved the same patient population and same time period. Slip op. at 50. As we pointed out in our analysis of Fellner, that case involved food, not drugs, and is therefore not all that on point in drug cases. The “informal” FDA documents at issue in Fellner, we pointed out, were a private letter, a “backgrounder,” and a “consumer advisory.” See Fellner, 2008 WL 3842925, at *8-9. Apparently there are no established FDA procedures for when the agency is supposed to/allowed to take any of these types of actions.
One thing Fellner certainly did not involve was an FDA amicus brief – which is what Knipe ignored in reliance on Fellner. The FDA can’t just willy-nilly decide to write an amicus brief. For one thing, the FDA doesn’t appear in court unrepresented. It has to go to the Department of Justice and have the Department write the brief. We’re almost positive that there are other internal requirements that must be met before the FDA will file an amicus brief, probably in some obscure manual that, because neither of us ever worked for the FDA, we don’t know what it’s called. Yo, all you FDA alums over at FDA Law Blog – help us out here.
Specifically with respect to FDA amicus briefs, the Third Circuit has accorded them deference in at least two recent preemption cases, in Colacicco, 521 F.3d at 275, and in Horn v. Thoratec Corp., 376 F.3d 163, 177-79 (3d Cir. 2004). On this point, at least, Knipe is just plain wrong. Fellner is simply not authority that courts may disregard of agency positions asserted in amicus briefs, while Colacicco and Horn are directly-on-point decisions holding that FDA preemption positions expressed in such briefs are entitled to deference.
Third, and finally, to us Knipe is a poster child for why the FDA needed to change the CBE (“changes being effected”) regulation, 21 C.F.R. §314.70(c)(6)(iii), to tighten the standard of proof. We’re not going to rehash our prior posts on this subject, for more, see here and here. Repeatedly, Knipe hinges its no-preemption decision on a (typical, but the FDA would say, wrong) reading of that regulation as authorizing changes to approved labeling without FDA pre-approval upon a low “reasonable evidence. . .a causal connection need not have been proved” standard. Slip op. at 24, 39, 42, 53.
The FDA is of the opinion that, for new data to warrant the extraordinary step of bypassing normal agency pre-review, it must have at least as much causal connection as needed to justify a normal agency-approved warning. Thus, it has now changed this evidentiary standard specifically to require “sufficient evidence of causal association”:

FDA proposed amending these provisions to affirm that a CBE supplement. . .may be used to add or strengthen a contraindication, warning, precaution, or adverse reaction only if there is sufficient evidence of a causal association with the drug, biologic, or medical device.

73 Fed. Reg. 49603, 49604 (FDA Aug. 22, 2008) (final rule).
That means that the rationale the court adopted for the time period relevant in Knipe is no longer available – and has been affirmatively rejected by the FDA.
Toward the end, Knipe recognizes that the low “causation need not have been proved” standard diverges significantly from the common-law standard in a tort case:

[S]uch litigation will not result in exaggerated risks associated with drugs. Because the standard for adding a warning under a CBE requires only reasonable evidence of association of a serious risk with the drug, and because state failure to warn claims under New Jersey law require the more stringent proof of causation between the drug and the injury, only a drug manufacturer who, in fact, possessed such reasonable evidence, yet failed to add a warning, would be potentially liable under state law.

Slip op. at 53 (citation omitted). We think Knipe drew exactly the wrong conclusion from this fact. If it hadn’t been for the lower standard of causation that Knipe found applicable to CBE submissions, it’s pretty clear that this action would have been preempted, since there’s no finding anywhere that the evidence that supposedly warranted a CBE submission (evidence, we’re not going to discuss) could also meet the higher common-law causation standard. Thus, what the court has allowed to proceed is – in everything but name – an action against the defendant for allegedly violating a purported obligation to file a CBE submission. That cause of action should be preempted. As any number of courts have held, there’s simply no private right of action under the FDCA. Rather, “[t]he FDCA leaves no doubt that it is the Federal Government rather than private litigants who are authorized to file suit for noncompliance with the medical device provisions.” Buckman, 531 U.S. at 349 n.4 (citing 21 U.S.C. §337(a)). Section 337(a) is “clear evidence that Congress intended that the MDA be enforced exclusively by the Federal Government.” 531 U.S. at 352.
The only arguable way to avoid preemption in this circumstance is by asserting a “parallel” violation claim. But the claim in Knipe – that the defendant should have filed a CBE supplement and didn’t – plainly doesn’t “parallel” any state law. It can’t, since Knipe agrees that the regulation imposes a much different, and lesser, standard of causation than the common law. That’s no “parallel” claim. Rather, that’s precisely the sort of claim Buckman held was preempted, because it’s based entirely upon an administrative obligation that’s not equivalent to any pre-existing common-law action:

In the present case, however, the fraud claims exist solely by virtue of the FDCA disclosure requirements. Thus, although [Lohr] can be read to allow certain state-law causes of actions that parallel federal safety requirements, it does not and cannot stand for the proposition that any violation of the FDCA will support a state-law claim. In sum, were plaintiffs to maintain their fraud-on-the-agency claims here, they would not be relying on traditional state tort law which had predated the federal enactments in question. On the contrary, the existence of these federal enactments is a critical element in their case. For the reasons stated above, we think this sort of litigation would exert an extraneous pull on the scheme established by Congress, and it is therefore pre-empted by that scheme.

Buckman, 531 U.S. at 352-53 (emphasis added).
Just like the fraud claims in Buckman, the allegations in Knipe that the defendant is liable for failing to submit a CBE submission exist “solely by virtue of the FDCA [CBE] requirement.” The opinion admits that there’s no common-law duty to warn where “causation need not be proven.” Slip op. at 53. The CBE standard utilized by the court thus is not “traditional state tort law” – which requires proof of causation. Rather the existence of the CBE regulation and its lesser standard of proof “is a critical element in [Knipe’s] case.”
Thus, in our view (no surprise), Knipe was wrongly decided. Even if the opinion’s multiple factual contortions (which we, unfortunately, can’t go into in detail) could be sufficient to escape preemption under Colacicco (same caveat), that only results in preemption of the resultant regulatory-based claim under Buckman.