Trial Design and Oversight
This multicenter, double-blind, randomized, placebo-controlled, event-driven, noninferiority trial involved patients with type 2 diabetes and established atherosclerotic cardiovascular disease.12 The protocol (available with the full text of this article at NEJM.org) was approved by the relevant regulatory authorities and ethics committees responsible for each trial site.
In collaboration with a group of academic investigators who comprised the scientific advisory committee, representatives of the sponsors (Merck Sharp & Dohme [a subsidiary of Merck] and Pfizer) designed and oversaw the conduct of the trial. A clinical research organization, Parexel International, selected and monitored the trial sites and managed and stored the data, with oversight from the sponsors. An independent, external data and safety monitoring committee monitored the interim unblinded data. Lists of the trial committee members, investigators, and sites are provided in Section S1 in the Supplementary Appendix, available at NEJM.org; information regarding data handling and quality assurance is provided in Section S2.
Analyses were performed by employees of Parexel International, and the results were independently confirmed by the sponsors with the use of original data. The academic authors ensured the accuracy and completeness of the data and were able to request additional analyses at their discretion. The first and last authors drafted the first version of the manuscript, and all the authors contributed to revisions. The decision to submit the manuscript for publication was made jointly by the authors, who vouch for the completeness and accuracy of the data and for the fidelity of the trial to the protocol.
As reported previously,12 the original protocol was finalized in August 2013 and included a planned sample size of approximately 4000 patients. After the results of the Empagliflozin Cardiovascular Outcome Event Trial in Type 2 Diabetes Mellitus Patients (EMPA-REG OUTCOME) became available,6 the protocol was amended in March 2016 without knowledge of any interim results. The key changes were to double the sample size to approximately 8000 patients and to include efficacy objectives for superiority with respect to cardiovascular and renal outcomes. Patients who had been enrolled in the trial before the March 2016 amendment were designated as cohort 1, and those who were enrolled after the March 2016 amendment were designated as cohort 2.
A full list of the trial eligibility criteria is provided in Section S2. Patients were eligible if they were at least 40 years of age and had type 2 diabetes (with a glycated hemoglobin level of 7.0 to 10.5%) and established atherosclerotic cardiovascular disease involving the coronary, cerebrovascular, or peripheral arterial systems. Key exclusion criteria were a history of type 1 diabetes or ketoacidosis and an estimated glomerular filtration rate below 30 ml per minute per 1.73 m2 of body-surface area. All the patients provided written informed consent.
Eligible patients were randomly assigned in a 1:1:1 ratio to receive 5 mg or 15 mg of ertugliflozin or matching placebo once daily, added to background standard-of-care treatment. Randomization was performed at a central location with the use of an interactive voice-response system and was based on a computer-generated schedule with randomly permuted blocks, stratified according to geographic region. The rationale for the selection of the ertugliflozin dose and a detailed description of the randomization criteria are provided in Section S2.
Doses of background antihyperglycemic medication were held constant for the initial 18 weeks of the trial except in the patients who met the criteria for glycemic rescue (Section S3) and those with clinically significant hypoglycemia. Patients who discontinued ertugliflozin or placebo prematurely were followed for outcomes, except if they withdrew consent or were lost to further follow-up. Extensive efforts were made to collect full outcome data from all the patients.
The primary outcome, assessed in a time-to-event analysis, was a composite of death from cardiovascular causes, nonfatal myocardial infarction, or nonfatal stroke (i.e., a major adverse cardiovascular event). The key secondary outcomes, assessed in time-to-event analyses and in a hierarchical statistical testing sequence, were a composite of death from cardiovascular causes or hospitalization for heart failure; death from cardiovascular causes; and a composite of death from renal causes, renal replacement therapy, or doubling of the serum creatinine level. Additional outcomes and definitions are provided in Section S4. All the primary and secondary outcome events were centrally adjudicated by a cardiovascular adjudication committee in a blinded manner. The trial included three glycemic substudies (results not reported here) (Section S5).
Changes from baseline in glycemic measures, body weight, and blood pressure were also assessed. Safety was assessed on the basis of adverse-event monitoring and, for certain adverse events (e.g., genital mycotic infection, hypovolemia, and amputations), on the basis of a priori definitions (Section S6).
Sample size and power for the analyses of the primary and key secondary outcomes have been described previously12; the original sample-size calculation before the March 2016 protocol revision is outlined in Section S7. For the final sample-size calculation, we estimated that, with a total of 8000 patients, 939 primary major adverse cardiovascular events would be accrued in approximately 6.1 years. With the data from the two ertugliflozin dose groups pooled for analysis, the trial had approximately 96% power to show noninferiority of ertugliflozin to placebo (the primary objective) by ruling out a hazard ratio for major adverse cardiovascular events of 1.3, in accordance with guidelines from the FDA; the power was determined under the assumption of no difference between the trial groups (i.e., hazard ratio for major adverse cardiovascular events of 1.0). If noninferiority was shown for the primary outcome, then tests of superiority for the key secondary outcomes (a composite of death from cardiovascular causes or hospitalization for heart failure; death from cardiovascular causes; and a composite of death from renal causes, renal replacement therapy, or doubling of the serum creatinine level) were to be performed with the use of a sequentially rejective graphical testing procedure.13
The noninferiority analysis of the primary outcome was performed with data from all the patients who had undergone randomization and received at least one dose of ertugliflozin or placebo. For the patients who discontinued the assigned trial regimen prematurely, only major adverse cardiovascular events that occurred up to 365 days after the confirmed last dose were included in the primary analysis, in accordance with guidance from the FDA. Tests of superiority with respect to the secondary outcomes were performed on an intention-to-treat basis in all patients who had undergone randomization, with no limit on the time window for the ascertainment of outcomes. These analyses were also performed according to cohort (cohort 1 vs. cohort 2). Sensitivity analyses were performed with the use of an intention-to-treat approach and an on-treatment approach, in which confirmed events that occurred between the day of the first dose of ertugliflozin or placebo and 14 days after the last dose were included in the analysis (Section S7).
One preplanned interim analysis had been scheduled to evaluate efficacy and futility with the use of a Lan–DeMets alpha-spending function with an O’Brien–Fleming boundary to control the type I error rate14; this analysis took place after 715 major adverse cardiovascular events (73%) had accrued among the patients during treatment and up to 365 days after the last dose and after 351 deaths from cardiovascular causes had accrued among the patients in the intention-to-treat population. The trial was continued on the basis of the results of this interim analysis. The testing boundaries and confidence intervals for the final analyses were adjusted according to the actual alpha spent at the interim analysis (Section S7).
A stratified Cox proportional-hazards model that included the trial group as a covariate and cohort of enrollment as the stratification factor was used to evaluate the primary outcome. After adjustment for the interim analysis, the upper boundary of a two-sided 95.6% confidence interval for the hazard ratio was used for the noninferiority test. The Kaplan–Meier method was used to estimate the cumulative incidence (first occurrence) of an outcome event over time in each trial group. Additional details are provided in Section S7.
The data from the two ertugliflozin dose groups were prespecified to be pooled for the assessment of cardiovascular and renal outcomes. Individual trial-group responses are presented for safety outcomes. Safety analyses included all patients who had undergone randomization and received at least one dose of ertugliflozin or placebo.