On EPIC
So how confident are you that in EPIC, we won't see any imbalance on any of these events?
Harvey J. Berger
I don't know if we'll see no imbalance but -- because I can't predict that. But I believe we'll see [indiscernible] that are totally fine and are of no concern whatsoever.
Terence C. Flynn - Goldman Sachs Group Inc., Research Division
Okay, and do you guys get -- do you get blinded safety data? Or has only the DMB have blinded safety data [indiscernible]?
Harvey J. Berger
We had -- we get reports from the DSMB and we get blinded series adverse event reporting. So we have a handle on what's really important and the highly detailed analysis comes recurrently from the DSMB, which I see and Frank Haluska, our Chief Medical Officer, sees, so the rest of the company does it, but -- we do.
Terence C. Flynn - Goldman Sachs Group Inc., Research Division
And what -- I mean, in terms of -- just remind us what a win is from EPIC, in terms of the primary endpoint. What do you guys need to show? And then, how important is that study to driving frontline use ahead of entry of a generic Gleevec?
Harvey J. Berger
How important is an understatement of the year? So how important is we have to win? So what we have to show is driven statistically. There's no absolute magic number of what percent is important. The trial was designed to show a 15% difference between the 2 arms, but what the number is, is determined by the statistics. So it's a capital [indiscernible] analysis of the event rate curves. So we need to show a statistically significant difference in 12-month major molecular response between the Gleevec and the Iclusig arms. The trial was designed with 500 patients and a 90% power to show a 15% difference between the 2 arms, assuming Gleevec did the best that it has ever done in any randomized trial, which is the upper boundary of the 95% confidence interval of the best point estimate for Gleevec ever, which would have had Gleevec then at 34%. Bear in mind, it usually does in these sort of trials, 22%, 24%. So we assume Gleevec did 34%, add 15% on top of that gives you 49%. So it assumes, for trial design purposes, that Iclusig does 50% or better. Having said that, not that that's the requirement if Iclusig does 50% or better, it's a smashing home run, that would be viewed by everybody as, "Oh my goodness, that's extraordinary." It does not have to do that to win unequivocally, but nilotinib and dasatinib and their trials came out in the mid-40s. So just from a comparative trial point of view, if Iclusig ends up in the 50s, it would be great from the point of view of cross-trial comparison, it would be great from the absolute number being, wow, that's really extraordinary, and you'd be sure to win. So that's sort of the way I look at the trial. I mean, if Gleevec did 18%, you could win with way less. If Gleevec did 28%, you probably would need to break 50%, although it's never done that. But certainly, if it were 50% or greater, you could feel confident that it's an unequivocal, great outcome. And yes, we do have to win the trial.
Terence C. Flynn - Goldman Sachs Group Inc., Research Division