Data in Excel, Codebook (variable list) in Excel, Plot in Excel.
The Enhanced Suppression of the Platelet IIb/IIIa Receptor with Integrilin Therapy (ESPRIT) trial was a double-blind, multi-center, randomized, placebo-controlled trial of the treatment of patients with Integrilin (a GP2b3a inhibitor) in patients scheduled to undergo percutaneous coronary intervention (PCI) with stent implantation. A total of approximately 2,400 patients were planned to be enrolled at centers in both the U.S. and Canada. Informed consent was obtained from patients meeting eligibility criteria before enrollment in the study. The study drug was initiated immediately before percutaneous coronary intervention (PCI). The primary endpoint was death, MI, urgent revascularization, or crossover to the other treatment (bailout) at 48 hours. The DSMB was also concerned about the standard endpoint of death or MI at 30 days.
The drug, Integrilin, had been shown to be moderately effective in previous studies such as the PURSUIT Trial. In addition, other GP2b3a inhibitors had been shown to be extremely effective (the EPISTENT Trial). Thus the FDA was reluctant to even allow this placebo-controlled trial to be run. The logic for the ESPRIT Trial was that these were lower risk patients who were normally not given a GP2b3a inhibitor. Finally, after finding an ethicist who supported the trial (in a letter to the FDA), the trial w as allowed to commence. The Data Safety and Monitoring Board) DSMB reviewed the results as they came in. The DSMB was aware that the drug could be highly effective, but because the trial was supposed to be a fast enrolling trial, there was no provision made for interim testing to possibly stop the trial. If an efficacy analysis related to safety concerns was required, then an alpha level spending function of 0.0001 per look was specified in the protocol.
Enrollment turned out to be much slower than expected (when isn't it?). After the second DSMB meeting, it was clear that the drug was quite effective and the DSMB was considering stopping the trial (for the safety of the placebo patients). There was a great deal of discussion as to the alpha level to be used because one was not specified. The DSMB decided that the criteria for stopping the trial would be 0.005 for the primary endpoint if all other endpoints were consistent.
At the next DSMB meeting, the results showed that the criteria were met and a recommendation to terminate the trial for efficacy was made. The trial was stopped that night, and a public announcement was made by the sponsor before 8:00 the following morning. That day, the company's stock value rose from $17 to $51 per share.
The sample size for this trial was based on the key secondary efficacy parameter, and will require approximately 1200 patients per treatment group. A sample size of 1200 per treatment group (2400 patients total) has 86% power to detect a 33% relative reduction (3.6% absolute difference) in the percentage of patients with the composite endpoint of death, MI, and urgent target vessel revascularization within 30 days at the alpha=0.05 significance level, assuming a placebo event rate of 11.0%. This sample size will also have 90% power to detect a 33% relative reduction (4.0% absolute difference) in the percentage of patients with the primary composite endpoint of death, MI, urgent target vessel revascularization, and 'bail-out' for thrombotic reasons within 48 hours at the alpha=0.05 significance level, assuming a placebo event rate of 12.0%. This sample size estimate was determined via the Chi-square algorithm in the statistical software package "Power and Precision". The 11% placebo 30 day event rate is consistent with control group rates seen in the EPISTENT trial.
References