Photo of Eric Alexander

Not that our readers are dying for a glimpse behind the curtain into the making of the sausage (and mixed metaphors) of the blog, but we do try to first figure out what decisions may be blogworthy before we start writing up the posts each week. We cannot say how blogworthiness relates to spongeworthiness, but we do know that the former involves asking a few questions, one of which is “is this just the same old same old?” In other words, even if the decision comes from a product liability case with a drug or device and addresses something that we think matters, we will not write about it if it adds nothing more than being another decision like we have seen many times before. Maybe it will be added to some cheat sheet or other compilation post, but it will not be worthy of its own post on this illustrious blog. (So illustrious are we that have noticed other blogs trying to rip off our name in the hopes that they will confuse search engines into directing traffic their way.)

When Levine came out with its misreading of the CBE regulation and novel “clear evidence” standard for impossibility preemption, we certainly did not think it would become so commonplace for prescription drug manufacturers—branded, in particular, although the liability of generics was not much of a concern then—to win warnings claims based on preemption. We are not yet there, but we can envisage a day where wins like in Seufert v. Merck Sharp & Dohme Corp., No. 13cv2928, 2016 WL 3369512 (S.D. Cal. May 11, 2016), just end up in our handy-dandy Levine cheat sheet For now, Seufert is still blogworthy.

Here are the basic facts. Plaintiffs sued over allegedly inadequate (absent) warnings on a purported risk of pancreatic cancer with two anti-diabetes drugs—the same compound is in each, with one adding metformin—within a class of drugs called incretin mimetics or incretin-based therapies. There are a number of approved drugs in this class, including one with its active ingredient derived from Gila monster venom. The two at issue were approved in 2009 and 2010 and there has been some degree of attention by FDA to a proposed risk of pancreatic cancer since they came on the market. It is not clear if plaintiffs contended that there was pre-marketing evidence of a pancreatic cancer risk specific to these two drugs, but we assume they did not have much if anything. In 2009, FDA reviewed adverse event data for the class of drugs and concluded “a causal association . . . is indeterminate as this time.” In 2013, FDA issued a “Drug Safety Communication” stating that it was analyzing the issue of pancreatitis and pancreatic cancer with the class of drugs, but that “it had not reached a new conclusion regarding whether incretin mimetics cause pancreatic cancer, and advised health care professionals to continue following the prescribing recommendations in the drug labeling”—which did not mention pancreatic cancer. Based on the plaintiffs’ case numbers, it looks like this Communication started them suing.

In February 2014, FDA and the European Medicines took the fairly unusual step of publishing an article in the New England Journal of Medicine to report on the results of the FDA’s analysis mentioned in the Communication and the EMA’s own analysis.

Both agencies agree that a causal association between pancreatitis and pancreatic cancer as expressed recently in the scientific literature and media are inconsistent with the current data . . . . The FDA and EMA believe that the current knowledge is adequately reflected in the product information and labeling.

In March 2014, FDA denied a citizen petition to remove a particular drug in the class from the market based on the purported risk of pancreatic cancer and other things. (We checked and it was filed by—shocker—Public Citizen. One need not dig deeply to find the ties between that group and the lawyers who bring these cases.) The denial noted there was “no new evidence regarding the risk of pancreatic carcinoma in association with the use of [the drug] that would support any changes to the current approved labeling”—which did not discuss pancreatic cancer. Based on this history, the defendants moved to dismiss the warnings claims based on conflict preemption. The court denied it so that the cases could proceed to discovery before addressing the merits of preemption. Since then, FDA has approved other drugs in the class without any warnings about a risk of pancreatic cancer in their labels. Apparently, none of the manufacturers have asked to add such a warning to the label, so the FDA has never rejected such a request.

For those following along on preemption of prescription drug warnings claims, the battle lines for summary judgment were clearly drawn from these facts. For those really following, you might recall that we have posted previously on two decisions finding preemption with other drugs of this same class, including in a decision by the same judge With this background, the plaintiffs first advanced two arguments to try to avoid a detailed consideration of the regulatory record on the question of whether there was clear evidence that the FDA would have rejected the plaintiffs’ proposed labeling change. They claimed that the FDA had to have rejected a proposed labeling change from the/a manufacturer, but Levine did not say that and cases had found preemption without such a rejection. They also claimed that only FDA’s consideration of a proposed warning counts as evidence, not actions relating to whether there is a risk. There is some irony to this latter argument as plaintiffs tend to be loath to specify the form and content of the warning that would have been adequate, even while arguing about the actual label and whether it should and could have been changed. In any event, the court patiently rejected what was a pretty silly argument given the facts of this case. Again, nothing in Levine—including the regulatory history of the drug and warning at issue—suggested such a limitation. More importantly, these plaintiffs argued that a risk not in the label needed to be added, which the court characterized as plaintiffs seeking a new warning “that incretin mimetics cause or increased the risk of pancreatic cancer.” To add such a warning, in any part of the label, requires an assessment of whether FDA believes such a risk is supported by the data. As set out in our discussion of the facts, “[t]he FDA’s review therefore directly corresponds to the claims at issue in this litigation,” even the review was not occasioned by a proposed labeling change.

We think this is right, but a broader understanding of FDA’s regulation of drugs would show that consideration of risk and labeling implication are commonly connected, which is evidenced by the references to labeling in the Communications, NEJM publication, citizen petition denial, and at least the vast majority of approval letters for NDAs and NDA supplements and even internal memoranda evaluating periodic submissions. It is said that FDA approves labels more than it approves drugs. Anyone who has spent much time reviewing submissions to FDA and documents generated by FDA about those submissions and analyses that FDA does on its own should come to the realization that labeling is the linchpin. The same conclusion can be reached by reviewing the chunk of 21 CFR focused on prescription drugs. Approval letters, for instance, say the drug is approved because of proof that it is “safe and effective for use as recommended in the label” (give or take on the language) and post-marketing submissions relating to risk and the review of those are always tied to whether the current label says what it should about the risk. FDA evaluations of risk information, whether from studies, collections of spontaneous reports, or some more systematic review of spontaneous reports, always involves a question of whether labeling (draft or approved) needs to be changed in some fashion. A statement in an FDA review memorandum that “no action is recommended” or “no changes to labeling are recommended” really means the reviewer, in her official capacity, has concluded that the current label appropriately describes the current evidence of the risk of each adverse event discussed in what was reviewed, including where the label does not mention the event at all. The FDA reviewers are not writing their memoranda with an eye towards how some judge will interpret their preemptive effect, so a greater understanding of the overall scheme will have to suffice.

Returning from our segue, we note the court’s thoroughness in explaining that the evidence showed “that the FDA would have rejected a pancreatic cancer labeling change” and the plaintiffs’ arguments to the contrary were unavailing. “The FDA’s repeated conclusion that scientific data did not support warning of pancreatic cancer risk coupled with the FDA’s statement that product labeling was adequate amounts to clear evidence that the FDA would have rejected a pancreatic cancer labeling change.” With regard to the Levine argument that a CBE was available to make a temporary labeling change, “Defendants did not have a duty to submit a pancreatic cancer warning to the FDA because the risk of pancreatic cancer was never readily apparent from available data.” Not only would FDA have rejected a CBE the same way it would have rejected a prior approval submission, but public policy runs contrary to requiring requested labeling changes without sufficient scientific support:

A rule to the contrary would encourage prophylactic labeling changes by manufacturers, which, in turn, could inundate the FDA with labeling submissions. This could lead to overwarning consumers and deterring potentially beneficial use of prescription drug,

Consistent with the “well-settled maxim of jurisprudence that the law does not require idle acts, without sufficient data to justify a labeling request, “Defendants did not have to submit a pancreatic cancer labeling change to establish conflict preemption.” In short, the evidence was about as clear as one could envision in the absence of a denial of a request to change the label in the same way plaintiffs contend was required.

Plaintiffs raised two additional arguments. First, because FDA did not consider its statements “final” while additional risk evidence was being accumulated, particularly clinical studies, plaintiffs claimed a rejection of a proposed labeling change was not clear. The court saw that risk evidence on marketed drugs—or classes of drugs—so waiting for a truly “final” statement would allow the court to do its job. As with Daubert rulings, the implications of the present science can be tested at an appropriate time after the plaintiffs chose to bring their cases. Second, plaintiff s contended that FDA had been inconsistent with regard to allowing pancreatitis in labeling while saying a causal relationship had not been established. This was a red herring because FDA described the level of proof of association differently for the different conditions.

Plaintiffs also asked to extend discovery to “obtain information about the FDA’s actions and potential new safety information in Defendants’ possession.” We will not belabor it, but no last minute charge from the Knights of the Vale was coming to change the result here. It also seems that, although the court did not rely on any of it in finding conflict preemption, that on-going studies of the particular drugs at issue showed no increased risk of pancreatic cancer. That should provide some comfort that the timing of the decision did not work against plaintiffs. Speaking of timing, we wonder why the publicly available information back in the spring of 2014—when the motion to dismiss was denied—would not have been enough to reach the same result. It does not look like discovery—which we presume was costly for both sides, but more costly for defendants—turned up anything that was critical to the decision. It will probably be a while until we start compiling cases with branded prescription drug warnings claims preempted before defendants are subject to any discovery.