Photo of Eric Alexander

The week after Weeks II, we are not in position to tout our prognosticating on that decision.  We had certainly hoped that a rehearing would have signaled a smarter approach to considering innovator liability given that the vast majority of jurisdictions have rejected Conte.  We had not predicted a specific result; however, either through reading entrails, the flight patterns of birds, or tea leaves.  In a recent post on a Daubert challenge in the Zoloft MDL, we did make the prediction that we would be seeing “at least one other opinion on mechanism experts.”  This was an easy prediction because the court had stated there was a Daubert hearing on plaintiffs’ experts and the first opinion only addressed the epidemiologist.  We also made a grim prediction for the plaintiffs’ ability to prove causation:  “Dr. Berard’s causation opinions are gone and, it would seem, so is the plaintiffs’ ability to prove general causation, something the court understood requires competent epidemiological expert evidence.  What is left is a decision that should be useful for other Daubert challenges on general cause opinions.”

Last week, the court in In re: Zoloft (Sertraline Hydrochloride) Prods. Liab. Litig., MDL No. 2342, __ F. Supp. 2d. __, 2014 WL 3943916 (E.D. Pa. Aug. 12, 2014), issued an opinion on the Daubert challenge to three mechanism experts.  As expected, the experts were precluded from offering general causation opinions.  They were allowed, however, to offer some of their other opinions “if they are otherwise admissible,” which we do not think they will be because biological plausibility does not matter absent predicate epidemiological evidence that plaintiffs do not have (according to the first Daubert opinion).  It appears that the court had reserved for another day whether plaintiffs can prove general causation for the birth defects at issue in their cases. They certainly cannot by relying only on the four experts addressed in the two opinions so far.

The three experts, none of whom is a medical doctor, relied on studies in a variety of animal models and in vitro models to offer fairly similar opinions that it is biologically plausible that Zoloft and other selective serotonin reuptake inhibitors can cause a range of birth defect by affecting serotonin signaling in human embryos. This sounds like a pretty broad mechanism and the opinions leaves some of our questions unanswered.  How is signaling affected Which pathways are affected by altered signaling?  How does could the alteration lead to such a wide range of birth defects?  Why would the alteration only lead to defects in tiny portion of affected embryos?  Like the details of the proposed mechanism, the analysis of why each expert got to offer his opinion is sparse.  For the first, Dr. Cabrera, it did not matter that he had never studied any SSRIs or tested his hypothesis because he did a reanalysis of the data from a rat studies from the drug’s NDA.  Id. at **2-3. Even though he considered it “questionable whether any reasoned interpretation could be derived from these studies,” and apparently had nothing else on the drug or class of drugs, his methodology was considered reliable.  Id.  Similarly, even though he relied only on in vitro studies on serotonin (as opposed to the drug or class of drug), the court left it to cross-examination to bring out the limitations of the methodology of the second expert, Dr. Sadler. The third expert, Dr. Levin, actually used SSRIs in his own teratology studies and claimed to have produced certain birth defects in a frog embryo model.  Id. at *4.  That put him ahead of the other experts, but his model reported producing different defects than the plaintiff claims and did not report producing the defects the plaintiffs claimed.  Id. at *5.  Again, the court left it to cross to bring out the lack of relevance of Dr. Levin’s opinion.  The court did exclude Dr. Levin’s opinion that, in addition to serotonin signaling, alterations of ion and calcium channels were plausible biological mechanisms, which the court characterized as a mere “untested hypothesis.”  Id. at *5.  The distinction seemed to be that the studies on which Dr. Levin based his ion and calcium channel opinions involved brain, cancer, and kidney  cells rather than embryonic cells.  Id.  We agree that this is a major methodological shortcoming, but the serotonin signaling mechanisms were not much better.

We suspect that the relatively free pass on mechanism was driven, at least in part, by the fact that biological plausibility was just one piece of evaluating the general causation opinion each expert hoped to give.  The parties agreed that the inquiry on general causation should be guided by the well-known Bradford-Hill criteria and the more specific Wilson’s principles of teratology.  Our view, derived from the speech that Dr. Bradford-Hill gave almost fifty years ago that popularized his criteria, is that there must be an association demonstrated through epidemiological studies before you go the next step of evaluating whether the association is consistent with a causal relationship.  The defendant advocated the same position, and, while the court noted that “[s]everal courts have held that positive human epidemiological studies are required to reach reliable conclusions as to whether an agent is teratogenic in humans,” the court allowed the plaintiffs’ experts to try to meet an easier standard:  “when epidemiological studies are equivocal or inconsistent with a causation opinions, experts asserting causation opinions must thoroughly analyze the strengths and weaknesses of the epidemiological research and explain why that body of research does not contradict or undermine their opinions.”  Id. at *8.  The experts still failed that standard, with two punting on epidemiology and one just saying epidemiological studies may underestimate associations because of spontaneous and therapeutic abortions.  Id. at **8-9.  We do not think this was the right standard, or that requiring that any other causation expert “address” the epidemiological research on Zoloft because there is “a great deal” available helps much.  Id. at *9.  We do think that “failure to reconcile inconsistent epidemiological research with their opinions regarding human causation is a significant methodological flaw,” but go further in that a plaintiff expert without some reliable epidemiology supporting causation should not be allowed to offer a causation opinion no matter how much reconciling she attempts. Id.

The rest of the analysis continued similarly, with the experts failing to meet a somewhat more lenient standard that we think Daubert requires.  The experts relied on animal and in vitro studies, but none of them described birth defects at doses similar to even the maximum clinical dose (in real people).  Id. at *10.  Again, rather than simply determine that studies do not support a reliable causation opinion if they do not involve a relevant dose, the court allowed the experts to try to “reconcile” the different doses, but they could not.  Id. at *11.  The experts tried to rely on studies in species that lay eggs, which mammals (other than monotremes—people always forget about platypuses and echidnas) do not.  Rather than limiting studies to those done in mammals, the court looked to see if the experts could tie in the studies they wanted to rely on—but they could not.  Id.  The experts wanted to rely on studies on serotonin pathways in animals without proof that humans have the same pathways, which the court rightly characterized as speculation.  Id.  Lastly, the experts lacked information about serotonin levels in pregnant women, with or without depression, or in their embryos, which made opinions about the impact of altering those levels mere guesswork.  Id. at *12.

This all added up to a statement that fairly summarized the experts’ failings:

The Court cannot allow unscientific speculation to be offered, even by genuinely talented scientists.  The Court holds that the evidence upon which the experts rely in their reports is not sufficient to support a non-speculative opinion that Zoloft can cause birth defects in humans when used at conventionally prescribed doses.  The Court notes that in vitro and in vivo animal research is useful for generating hypotheses about human causation, but each hypothesis must be tested and scientifically verified before it can form the basis for a conclusion about causation. These experts have not demonstrated that such testing and verification has been accomplished to date.  In addition, the Court notes that these scientists have never published their litigation opinions about human causation for their peers, and neither their opinions about human causation nor their methods for reaching those opinions, which bypass crucial hypothesis testing and analysis, are generally accepted by scientists in their fields.  Finally, many essential questions which would connect the animal data to human embryonic development are unanswered at this time.


Id. at * 13.  As a result, their causation opinions were excluded in their entirety.  As to their mechanism opinions, most survived Daubert because the defendant’s “argument conflates the sufficiency of the evidence with the admissibility of the testimony.”  Id.  Sufficiency is apparently for another day, but we can say that expert opinions on biological plausibility without more should not get the job done for plaintiffs.