Figure 1

Schematic of Adaptive Arm-dropping Strategy. This figure illustrates the adaptive arm-dropping strategy planned for the trial, based on the use of logistic dose-response models to integrate information from the placebo and all K-134 containing arms. Initially, the trial begins with equal randomization to the placebo, cilostazol, and the three K-134-containing arms. At each of two planned interim analyses, occurring when 28-day data are available from either 50 or 100 patients, logistic dose-response models are fit for each of the three safety and tolerability endpoints (resting tachycardia [denoted HR > 120], signs of ischemia on ECG, and medication discontinuation), using data from the placebo and K-134-contining arms. If the lower limit of the 80% confidence interval around the logistic model (the upper and lower limits are illustrated by dotted lines) exceeds the maximum tolerable limit of the safety or tolerability endpoint at a particular dose (excluding placebo), then that arm of the trial is to be discontinued. In the hypothetical scenario shown in the figure, the ischemia safety endpoint limit is exceeded (arrow) when N = 50 for the highest (100 mg) dose of K-134 so that arm is discontinued and the four remaining arms are continued until N = 100. At the second interim analysis, however, the maximum tolerable rate for resting tachycardia is exceeded by the lower limit of the 80% confidence interval at both the 50-mg and 100-mg doses (arrows). Since the 100-mg dose was already discontinued in this hypothetical example at the 50-patient review, the 50-mg dose arm would be discontinued at this point and new research subjects would be randomized in a balanced manner to one of the three remaining arms for the remaining duration of the trial.