The saxagliptin medications are valuable treatments for type 2 diabetes mellitus (T2DM). The SAVOR (Saxagliptin Assessment of Vascular Outcomes Recorded in Patients with Diabetes Mellitus) study was a randomized controlled trial, undertaken by manufacturers at the request of the FDA.[1] As a large (over sixteen thousand patients randomized) double-blinded cardiovascular outcomes trial, SAVOR collected data on many different end points in patients with T2DM, at high risk of cardiovascular disease, over a median of 2.1 years. The primary end point was a composite end point of cardiac death, non-fatal myocardial infarction, and non-fatal stroke. Secondary end points included each constituent of the composite, as well as hospitalizations for heart failure, coronary revascularization, or unstable angina, as well as other safety outcomes.
The SAVOR trial found no association between saxagliptin use and the primary end point, or any of the constituents of the primary end point. The trial did, however, find a modest association between saxagliptin and one of the several secondary end points, hospitalization for heart failure (hazard ratio, 1.27; 95% C.I., 1.07 to 1.51; p = 0.007). The SAVOR authors urged caution in interpreting their unexpected finding for heart failure hospitalizations, given the multiple end points considered.[2] Notwithstanding the multiplicity, in 2016, the FDA, which does not require a showing of causation for adding warnings to a drug’s labeling, added warnings about the “risk” of hospitalization for heart failure from the use of saxagliptin medications.
And the litigation came.
The litigation evidentiary display grew to include, in addition to SAVOR, observational studies, meta-analyses, and randomized controlled trials of other DPP-4 inhibitor medications that are in the same class as saxagliptin. The SAVOR finding for heart failure was not supported by any of the other relevant human study evidence. The lawsuit industry, however, armed with an FDA warning, pressed its cases. A multi-district litigation (MDL 2809) was established. Rule 702 motions were filed by both plaintiffs’ and defendants’ counsel.
When the dust settled in this saxagliptin litigation, the court found that the defendants’ expert witnesses satisfied the relevance and reliability requirements of Rule 702, whereas the proferred opinions of plaintiff’s expert witness, Dr. Parag Goyal, a cardiologist at Cornell-Weill Hospital in New York, did not satisfy Rule 702.[3] The court’s task was certainly made easier by the lack of any other expert witness or published opinion that saxagliptin actually causes heart failure serious enough to result in hospitalization.
The saxagliptin litigation presented an interesting array of facts for a Rule 702 show down. First, there was an RCT that reported a nominally statistically significant association between medication use and a harm, hospitalization for heart failure. The SAVOR finding, however, was in a secondary end point, and its statistical significance was unimpressive when considered in the light of the multiple testing that took place in the context of a cardiovascular outcomes trial.
Second, the heart failure increase was not seen in the original registration trials. Third, there was an effort to find corroboration in observational studies and meta-analyses, without success. Fourth, there was no apparent mechanism for the putative effect. Fifth, there was no support from trials or observational studies of other medications in the class of DPP-4 inhibitors.
Dr. Goyal testified that the heart failure finding in SAVOR “should be interpreted as cause and effect unless there is compelling evidence to prove otherwise.” On this record, the MDL court excluded Dr. Goyal’s causation opinions. Dr. Goyal purported to conduct a Bradford Hill analysis, but the MDL court appeared troubled by his glib dismissal of the threat to validity in SAVOR from multiple testing, and his ignoring the consistency prong of the Hill factors. SAVOR was the only heart failure finding in humans, with the remaining observational studies, meta-analyses, and other trials of DPP-4 inhibitors failing to provide supporting evidence.
The challenged defense expert witnesses defended the validity of their opinions, and ultimately the MDL court had little concern in permitting them through the judicial gate. The plaintiffs’ challenges to Suneil Koliwad, a physician with a doctorate in molecular physiology, Eric Adler, a cardiologist, and Todd Lee, a pharmaco-epidemiologist, were all denied. The plaintiffs challenged, among other things, whether Dr. Adler was qualified to apply a Bonferroni correction to the SAVOR results, and whether Dr. Lee was obligated to obtain and statistically analyze the data from the trials and studies ab initio. The MDL court quickly dispatched these frivolous challenges.
The saxagliptin MDL decision is an important reminder that litigants should remain vigilant about inaccurate assertions of “statistical significance,” even in premier, peer-reviewed journals. Not all journals are as careful as the New England Journal of Medicine in requiring qualification of claims of statistical significance in the face of multiple testing.
One legal hiccup in the court’s decision was its improvident citation to Daubert, for the proposition that the gatekeeping inquiry must focus “solely on principles and methodology, not on the conclusions they generate.”[4] That piece of obiter dictum did not survive past the Supreme Court’s 1997 decision in Joiner,[5] and it was clearly superseded by statute in 2000. Surely it is time to stop citing Daubert for this dictum.
[1] Benjamin M. Scirica, Deepak L. Bhatt, Eugene Braunwald, Gabriel Steg, Jaime Davidson, et al., for the SAVOR-TIMI 53 Steering Committee and Investigators, “Saxagliptin and Cardiovascular Outcomes in Patients with Type 2 Diabetes Mellitus,” 369 New Engl. J. Med. 1317 (2013).
[2] Id. at 1324.
[3] In re Onglyza & Kombiglyze XR Prods. Liab. Litig., MDL 2809, 2022 WL 43244 (E.D. Ken. Jan. 5, 2022).
[4] Daubert v. Merrell Dow Pharms., Inc., 509 U.S. 579, 595 (1993).
[5] General Electric Co. v. Joiner, 522 U.S. 136 (1997).
