It is now mandatory to demonstrate that new antihyperglycemic drugs do not elevate cardiovascular risk. The current approach to demonstrate CV safety is through an event-driven trial with adequate power for the upper confidence bound of the 95% confidence interval of the hazard ratio to be below 1.3. It would be desirable, however, to prolong a trial that has satisfied the 1.3 threshold in the expectation of concluding, eventually, that the hazard ratio is below 1. We shall show how this can be achieved in a statistically valid manner, with minimal risk of unnecessary trial prolongation, through an adaptive "promising zone" design. As no antihyperglycemic agent has so far shown actual risk reduction, the first agent to do so would enjoy tremendous competitive advantage besides benefiting the large population with type-2 diabetes. We will illustrate our approach with the help of the recently completed EXAMINE trial (NEJM, September 2014) a group sequential trial of patients with acute coronary syndrome treated with alogliptin.