Finding Efficacy in a Safety Trial
Empagliflozin and Cardiovascular Death
In late June 2016, the Endocrinologic and Metabolic Drugs Advisory Committee of the US Food and Drug Administration (FDA), on which we serve, met to consider whether empagliflozin, an inhibitor of the sodium-glucose cotransporter 2 currently approved for the treatment of type 2 diabetes mellitus, qualifies for a new indication to reduce the risk of cardiovascular death. The committee’s discussion focused on the results of the EMPA-REG OUTCOME [BI 10773 (Empagliflozin) Cardiovascular Outcome Event Trial in Type 2 Diabetes Mellitus Patients] cardiovascular outcome trial, which was designed primarily as a cardiovascular safety study to meet the FDA’s 2008 guidance for new antihyperglycemic medications. That guidance requires pharmaceutical companies to statistically exclude the possibility that their medications are associated with a ≥30% increase in the risk of cardiovascular events.1 Thus, EMPA-REG OUTCOME was designed with a prespecified primary end point of the composite of cardiovascular death, myocardial infarction, and stroke and randomized 7020 patients to once-daily empagliflozin 10 mg, empagliflozin 25 mg, or placebo in a 1:1:1 ratio. In EMPA-REG OUTCOME, the combined empagliflozin groups had a reduction in the primary composite outcome of 14% (hazard ratio, 0.86; 95.02% confidence interval, 0.74–0.99; P=0.038).2 This is the first large, prospective, randomized trial of an antihyperglycemic drug under review by the FDA to show a cardiovascular benefit.
The reduction in the primary cardiovascular end point in EMPA-REG OUTCOME was driven by a 38% reduction (hazard ratio, 0.62; 95% confidence interval, 0.49–0.78; P<0.0001) in cardiovascular death. Analyses of secondary end points suggested that empagliflozin reduced the risk of heart failure and progression of kidney disease, prompting the FDA to ask the committee to discuss these findings, in addition to the key finding on cardiovascular death. The committee was asked to vote on 2 questions: whether EMPA-REG OUTCOME had demonstrated the cardiovascular safety of empagliflozin and whether EMPA-REG OUTCOME provided “substantial evidence” that empagliflozin “is indicated to reduce the incidence of cardiovascular death,” a new efficacy label requested by the sponsor (Boehringer Ingelheim).3 “Substantial evidence” is the regulatory threshold that drugs are required to meet to claim a given indication, and the FDA typically requires >1 well-conducted study to meet this standard. However, the FDA has approved new indications on the basis of single studies if they are large, multicenter, well-controlled studies that show a consistent effect on a number of prospectively defined end points with statistically persuasive findings (typically P<0.001).3 Furthermore, as the FDA noted, there is no precedent for using this type of diabetes safety trial for an efficacy claim.
The committee voted unanimously that EMPA-REG OUTCOME results met the required standard for cardiovascular safety. The committee was intrigued by the effects of empagliflozin on heart failure and kidney end points, but many on the committee concluded that those end points were largely exploratory. The trial was not primarily designed to test those end points, and as a result, important baseline covariable data that would allow the findings to be placed in context were not adequately captured. In terms of the second voting question, there was a clear lack of consensus on whether EMPA-REG OUTCOME met the standard of substantial evidence for the proposed indication for cardiovascular death reduction, with 12 members voting yes and 11 voting no.
The sponsor used a prespecified hierarchical, stepwise statistical strategy to test first for noninferiority of the primary outcome (the composite of cardiovascular death, myocardial infarction, or stroke) and, if satisfied, then the key secondary outcome (the primary outcome plus unstable angina). Once noninferiority was demonstrated (the primary question of safety), then statistical testing continued for superiority on those 2 outcomes. Empagliflozin was unambiguously noninferior to placebo for both the primary and secondary outcomes and, as noted above, was superior to placebo for the prevention of the primary outcome but not for the key secondary outcome that included unstable angina. The committee felt that the observed reduction in the primary composite outcome, with a P value that just met the threshold for superiority of empagliflozin over placebo (P=0.038) in a single trial, was not particularly robust. Any dilution in the cardiovascular death component would have made the primary end point nonsignificant.
The proposed indication focused just on the component of the primary composite outcome for which empagaliflozin had demonstrated a substantial effect: cardiovascular death. Many on the committee raised concerns about changes to the EMPA-REG OUTCOME protocol, end point definitions, and analysis plan during the course of the trial, as well as the statistical weakness of the observed reduction in the primary end point. Others raised concerns about the possibility of a chance finding for cardiovascular death, given that the mechanism of the effect of empagliflozin on cardiovascular mortality is unknown, that the finding did not have a precedent, that a large proportion of the deaths included in the cardiovascular death end point were of unknown cause but were presumed to be cardiovascular, and that EMPA-REG OUTCOME represents only 1 trial. Also concerning to many committee members was the focus on only 1 component of the primary end point, an approach that emphasizes only favorable findings and ignores neutral or unfavorable data such as a nonsignificant increase in the risk of stroke with empagliflozin.
However, others noted the clinical importance of cardiovascular death as an end point and that adjudication of death is far more resistant to errors of misclassification or differential ascertainment than other clinical events, including hard clinical end points such as myocardial infarction and stroke. Although EMPA-REG OUTCOME is only 1 trial, empagliflozin therapy led to a reduction in the absolute risk of cardiovascular mortality of 2.2% (from 5.9% to 3.7%), a statistically robust (P<0.0001) finding based on a large number of events (309 total). Each of the individual doses of empagliflozin also offered a similar relative reduction in the risk of cardiovascular death compared with placebo (hazard ratio, 0.65; 95% confidence interval, 0.50–0.85; P=0.0016 for 10 mg; and hazard ratio, 0.59; 95% confidence interval, 0.45–0.77; P=0.0001 for 25 mg).2,4 Several committee members noted that the consistency across both dose arms and across different subgroups favorably influenced their assessment of the validity of the finding. In addition, the reduction in cardiovascular mortality was still significant after unknown deaths were reclassified as noncardiovascular (P<0.0001).
In our opinion, the FDA asked the committee to address 2 separate but related questions. First, was the committee willing to accept data from a safety trial as evidence for efficacy, establishing a new precedent in the process? We fully support the concept that trials such as EMPA-REG OUTCOME, which was mandated by the FDA’s 2008 guidance,1 should allow analysis plans to support a claim of efficacy. Second, if the committee decided to accept that precedent, did the data from EMPA-REG OUTCOME provide substantial evidence to establish that empagliflozin reduces cardiovascular death? Here, the committee’s opinion was divided. We anticipate that ongoing studies with other sodium-glucose cotransporter 2 inhibitors will shed further light on their effects on cardiovascular death and other important cardiovascular outcomes.
Circulation is available at http://circ.ahajournals.org.
- © 2016 American Heart Association, Inc.
- 1.↵US Department of Health and Human Services Food and Drug Administration Center for Drug Evaluation and Research (CDER). Diabetes mellitus—evaluating cardiovascular risk in new antidiabetic therapies to treat type 2 diabetes. 2008. http://www.fda.gov/ucm/groups/fdagov-public/@fdagov-drugs-gen/documents/document/ucm071627.pdf. Accessed July 21, 2016.
- Zinman B,
- Wanner C,
- Lachin JM,
- Fitchett D,
- Bluhmki E,
- Hantel S,
- Mattheus M,
- Devins T,
- Johansen OE,
- Woerle HJ,
- Broedl UC,
- Inzucchi SE
- 3.↵US Food and Drug Administration. FDA presentations for the June 28, 2016 meeting of the Endocrinologic and Metabolic Drugs Advisory Committee. http://www.fda.gov/downloads/AdvisoryCommittees/CommitteesMeetingMaterials/Drugs/EndocrinologicandMetabolicDrugsAdvisoryCommittee/UCM510017.pdf. Accessed July 9, 2016.
- Woerle H-J