How do you get a composite MACE endpoint of 4.0X10-7 at the third interim?
I think it's some complex of cardiac events....
If you have an absolutely enormous trial but still want to make sure you see a meaningful clinical benefit (as opposed simply to a stat sig one), then you could create interim parameters that translate to tiny p values required to stop the trial.
To use an example familiar to some of us, imagine that the START trial for tecemotide had had 10 times as many enrollees, but still required a clinical benefit of, say, 6 months survival advantage at the 2nd interim to stop the trial. Can you imagine how tiny that p-value would have been.