Trimming Some Fat From Statin Litigation With Daubert and Preemption
Eric Alexander
One of the greatest spectacles on earth is the annual migration of wildebeest, zebras, and antelope in a loop around the Serengeti. The Blog will have its own migration shortly, which we hope will be a one-time event without dangerous river crossings or, well, loads of crap for either us or our readers. Next week, the Drug and Device Law Blog will be moving to an upgraded platform via LexBlog. All the old posts will be available and going to the old site should redirect you to the new site. That’s the easy part. The comparatively difficult part should take you about 30 seconds, less time than it usually takes to google one of McConnell’s obscure movie references. Click HERE to fill out the new e-mail sign-up form. (Personal information will not be used for holiday cards from Bexis or any other nefarious purpose.) After you complete the form, you will receive a confirmation e-mail and you will have to click to confirm your subscription. Then you will get posts by email once we have completed the not-so-great migration. As an added bonus (?) from the migration, we will not be posting on May 23 or May 24. If your DDL addiction is too strong for that much of a break, then feel free to peruse old posts. Checklists and surveys might do the trick. If you need a stronger fix, then try searching for obscure or inflammatory words used in past posts. Hours of fun.
* * *
In the mythical days of product liability yore, a big drug litigation was spurred by a notable event—like a study describing a new risk appearing in a major journal, a significant labeling change because of new or greater risk, or the withdrawal of the drug—and was built on identifiable and otherwise uncommon injuries in patients who took the drug. Because there was a notable event and there were identifiable injuries, the manufacturer could generally expect when lawsuits about those injuries would need to be filed and that it could figure out if a plaintiff was suing over an injury that fit the larger liability issues. In the purportedly distinct modern era of drug litigation, the only notable events sometimes are the start of plaintiff lawyer conferences and advertising on the drug and the injuries claimed may not be particularly identifiable or otherwise uncommon. The cases brought probably include non-product liability claims for economic harm and product liability claims loosely tied to the risk of the injury actually sustained by the individual plaintiffs. Even if there is no good science backing up the plaintiffs’ claims, it can take quite a bit of time and effort for the manufacturer to get the rulings needed to show the claims and evidence offered in support of them to be bogus.
This seems to be what is going on in litigation involving the drug Lipitor, which has been one of the most widely prescribed drugs over the last twenty years (although generic entry had an effect a few years ago). Along the way, the manufacturer has faced litigation over liver injuries and rhabdomyolysis (like the other statins), antitrust claims, and some other issues. Meanwhile, the drug continued to be recommended as first line therapy for an expanding range of indications related to improving lipids and reducing cardiovascular risk. From what we can tell, without anything terribly dramatic, the drug was also one of several in the class associated with a possible small increase in the risk of developing Type 2 diabetes. If you have been paying attention over the last few decades, then you probably know that the rates of Type 2 diabetes have climbed dramatically, basically hand-in-hand with increases in the rates of obesity, hypertension, unhealthy eating, inactivity, and some other bad habits that call to mind the scooter-bound humans of the future in “WALL-E.” You might also know that the risks of heart attacks and strokes go up when you add a bad lipid profile to the mix and go down when the lipid profile improves. That is why doctors and patients care about LDL, HDL, and triglycerides in trying to prevent first or second heart attacks, for instance, just as they try to monitor and control obesity, hypertension, diabetes, sleep apnea, etc. In this context, it makes sense that doctors might continue to prescribe Lipitor or any of the other statins pretty much as they did before the labels warned of a possible increase in blood sugar levels and the rate of diabetes. That might make you think plaintiffs suing over a risk of diabetes might have hard time establishing either medical causation or proximate cause for failure to warn of diabetes. But the plaintiff lawyers filed enough cases that an MDL was established, so they must have mustered some good theories and evidence to support their claims, right?
Starting about six months ago, we reported on a series of Daubert decisions from the MDL (here, here and here) that generally excluded plaintiffs’ experts on general and specific causation for plaintiffs’ injuries. We recently also saw that the Second Circuit affirmed the dismissal of a separate False Claims Act case about the purported off-label use and promotion of Lipitor based on a nonsensical interpretation of the drug’s label, which happened to have been one of our first posts. We now get two more orders from Lipitor MDL, which we read as being more of the delayed inevitable response to a questionable litigation. The second in time, In re: Lipitor (Atorvastatin Calcium) Mkt’g Sales Practs. & Prods. Lib. Litig., MDL no. 2:14-mn-02502-RMG, -- F. Supp. 3d --, 2016 WL 2851445 (D.S.C. May 11, 2016), is similar to the prior Daubert decisions, particularly the one on the other expert in the same two cases. In this decision, Dr. Handshoe, whose pulmonology background seems ill-suited to weigh in on whether a lipid drug causes diabetes, could not offer a reliable causation opinion for either plaintiff. The court started by saying that the relative risk of 1.25 that Dr. Handshoe thinks applies to Lipitor and diabetes means that “20% of the people who take Lipitor and develop diabetes did so only because they took Lipitor.” We can quibble with this discussion of attributable risk exposed calculations—it presupposes general causation, for one—but it does show the uphill battle of connecting diabetes to the drug for a specific plaintiff. The case-specific facts for the first plaintiff made it even harder, as she had a strong family history of Type 2 diabetes, was overweight and had “pre-diabetes” when she started Lipitor and was obese when she was diagnosed as diabetic six years later. (This case, as one might expect with a litigation like this, was brought more than ten years after her diagnosis and without regard to the massive improvement in her lipid while on the drug whose manufacturer she sued.) She also had hypertension and hyperlipidemia (duh), which he considered risk factors for diabetes, and was a smoker, which he did not think was a risk factor even through some studies and governmental publications say otherwise. He also could not interpret relative risk data and relied heavily on the timing of plaintiff’s progression from pre-diabetes to diabetes occurring while she was on the drug. Without belaboring it too much, this shoddy case-specific causation opinion failed just about every criterion for reliability. The analysis for the second plaintiff was not quite as egregious, but was also excluded. While this plaintiff had a weaker family history of diabetes and no recorded history of abnormal blood sugar readings when she started a lower dose of Lipitor (taken irregularly), she had risk factors from her weight (overweight and continuing to gain weight), age, ethnic background, hyperlipidemia, and hypertension, which Dr. Handshoe could not rule out or reliably place into context. If you consider that these plaintiffs took a dose of Lipitor as to which one of the plaintiffs’ epidemiologists could not offer a general causation opinion, then the inability of an expert to offer a reliable case-specific causation opinion for plaintiffs with multiple risk factors other than the drug should not be surprising.
The second decision, In re: Lipitor (Atorvastatin Calcium) Mkt’g Sales Practs. & Prods. Lib. Litig., -- F. Supp.3d --, MDL No. 2:14-mn-02502-RMG, 2016 WL 2840215 (D.S.C. May 6, 2016), is somewhat more interesting and more controversial to us for a few reasons. First, it spells out that the plaintiffs—it applies to all of them—have asserted warnings and design claims predicated on the lack of efficacy in preventing coronary heart disease in women who have not had a coronary event. We think those claims are highly suspect even if true—although there might be other claims associated with false affirmative representations about efficacy for a subgroup. Second, it evaluated the impact of Bartlett and Mensing on preemption of these “lack of efficacy claims” for a branded prescription drug. Third, it used both Daubert and preemption in excluding most of the challenged expert testimony supporting these claims. Although the results here are pretty good, we think the court could have gone a little farther.
A little more background is needed to understand these issues. Among the various approved indications for Lipitor was one for “primary prevention of cardiovascular disease” for men and women based on the ASCOT study, as to which the label remarked “due to the small number of events, the results for women were inconclusive.” Following the Levine game plan, the plaintiffs claimed that a CBE could have been used to change the label to essentially re-frame the approved primary prevention indication as being for men only. Noting that “any claim that a drug label should be changed based on information previously submitted to the FDA is preempted because the CBE regulation cannot be used to make a label change based on such information,” the court held that plaintiffs could not use a for-litigation reevaluation of the ASCOT study as basis for their claims. The plaintiffs also could not use new risk information as a basis for requiring a labeling change as to efficacy through a CBE, given that the CBE regulations themselves require a link between the new data and the aspect of the label to be changed. The court also recognized that claims other types of labeling provided to healthcare professionals should have been changed are just as preempted as claims that the label itself should have been changed. And it recognized that plaintiffs cannot get around any of this preemption by claiming that the manufacturer could have just stopped selling the drug for primary prevention in women. All of that is good and part of the progression of branded drug preemption that we have urged for some time.
Where we think the court could have gone farther is on some issues that would have required more analysis. The court elected not to analyze if the law of any of 52 jurisdictions actually imposed duties on the manufacturer to take the steps plaintiffs claimed related to lack of efficacy and left open that the defendant could re-raise the arguments that specific requirements for liability based on advertising to the general public were preempted. It also did not dig deeply behind plaintiffs’ allegations that a study called CASHMERE could have been used as the basis for a CBE to change the approved primary prevention indication. While the “Defendant argues that ‘the results of CASHMERE have no bearing on the primary prevention indications’ [and] such arguments concern the sufficiency of evidence to survive summary judgment, not preemption,” there is the issue of whether CASHMERE was ever evaluated by FDA and what FDA thought about it. A temporary labeling change that would have been reversed by FDA because it still thought its approval of the primary prevention indication for both sexes was still correct has little impact on the claims of most plaintiffs. Thus, the conclusion that “[t]o the extent that Plaintiffs claim that a state law duty required Defendant to include different statements on Lipitor’s label regarding Lipitor’s efficacy for prevention in women, based on CASHMERE or other newly acquired information, the claims are not preempted” seems incomplete. We also think—without having tried to re-do the regulatory arguments—that there is a fairly direct conflict between the requirement for advertisements directed to the general public must truthfully summarize the efficacy of the drug and a contention that the advertisements should have said primary prevention indication was wrong in not excluding women. With a little more digging, it should be fairly obvious that all the “lack of efficacy claims” involve second guessing the FDA’s repeated decisions about efficacy, which should lead to broad preemption if any such claims actually exist under state law.
The Daubert portion of the ruling also seemed somewhat inconsistent, although the defendant got almost all of what it wanted on the five experts challenged. First, it did not look like any of the experts opined that CASHMERE—as opposed to just reinterpreting ASCOT—should have led to a revision of the drug’s indication, which undercuts the preemption ruling above. Second, while one expert was allowed to offer his own for-litigation re-analysis of ASCOT (which was also a peer-review publication) because “I think I was the only one to ever check the proportional hazards assumption” underlying the statistical test used in the original analysis to determine whether the results could properly be applied to men and women. We are not sure how this is relevant to any non-preempted claim in the case or the basis of the expert’s belief that nobody else checked the assumptions. We would wager that FDA only approved the primary prevention indication based on the ASCOT study after getting its data and checking it thoroughly, which would raise the issue of how plaintiffs’ claims are predicated on FDA making informed but incorrect decisions. When another expert opined that “the ASCOT data did not establish efficacy of Lipitor in women for primary prevention, and the label was misleading on this point,” his opinions were excluded as irrelevant based on the preemption decision. They were also confusing and misleading under Fed. R. Evid. 402 and 403 given that “such allegations cannot be the basis of Plaintiffs’ claims.” It seems like that same analysis could have been applied to the first expert’s opinions. Likewise, these rules provided the basis for precluding a third expert from offering his opinion that “I don’t think FDA should have approved Lipitor for primary prevention in women.” The court was consistent, however, in excluding opinions from multiple experts—phrased in slightly different ways—that there was no evidence that Lipitor (or other drugs of its class) was effective in women for primary prevention. In doing so, the experts had no reasoned basis for disregarding multiple large published studies and meta-analyses that found efficacy in women. As such, we will not dwell on how the different experts failed in trying to offer their lack of efficacy opinions.
We see this all as related to the problem we identified up front: a drug litigation that is not driven by science suggesting a significant risk of an identifiable and otherwise uncommon injury will struggle to fit into the existing legal framework. Proving that overweight hypertensive patients developed diabetes because of a drug that might increase the risk by 25%--far less than the patient’s weight or blood pressure does—is difficult. It is also hard to prove that a warning about such a risk would have changed any physician’s decision to prescribe a lipid medication to try to avoid heart attacks and strokes. That is why the plaintiffs make up “lack of efficacy claims” that should not be cognizable in most states, walk right into conflict preemption, and set up experts to offer unsustainable opinions. We suspect that it will take several more decisions from this court before the plaintiffs pack it in, though.