iwfal, I think I would agree with all of that, except...
There was talk on YMB last fall about the interims being based on O'Brien Fleming around a .001 threshold, and I don't know what that translates to. Clearly there are a range of interim practices and no single convention on what "O'Brien Fleming" means.
It is also possible, given the oddities of this trial (viz. the halt), that the DMC was more conservative in its recommendation, especially since full enrollment hadn't been reached at the 1st interim. FDA guidelines about interim stoppages talk about this kind of stuff. (I'm not saying this DID happen, of course.)
Also, most of the estimates I have seen are based on constant hazard rates (monthly) that would not be appropriate if Stimuvax were working the way that the trial sponsors think it will -- that is to say, that since it's immunotherapy, it takes awhile to "kick in". That would result in the KM curves separating later than you would see in many other trials.
Given that the 1st interim event trigger (presumably) was hit in late summer or early fall of 2010, the weighted average duration of enrollees in the trial at that point was less than 2 years. If I model the curves with later separation and declining hazard rates (esp for Stimuvax), I get an HR right around the level you quoted (.725).
I would agree that, if you are skeptical about Stimuvax, and don't believe the immunotherapy story, then you'd stick with a more traditional approach to the modeling, and end up with the MST estimates in the mid-20's or higher, as you say.
I actually think the control arm is performing well. I think more common use of concurrent chemo-radio is producing good results. I wouldn't be surprised if the control MST came out in the 23 to 25 month range. But given the Merck Germany guidance on 2nd interim timing (and they have regular event count updates), you can't make the math work and have Stimuvax fail.
For me, it's an Occam's Razor situation. Is it more likely that chemo-radio is taking a "great leap forward", or is it more likely that Stimuvax is matching its performance in the two earlier Phase II trials. That's a subjective judgment, of course.
I would point out -- and this is something many didn't note, or may still poo-poo after it is noted -- that the MST in the Phase IIB for Stimuvax was one patient short of being around 47 months. The 47-month survival was 49%. Again, I'm not saying this is an enormous trump card, but it's important data. I think we'd be having a very different discussion, at very different price levels, if that statistic had been 51% instead of 49%.
Regards,
TGW
Market Data 
Markets 
AI
