are we somehow under the impression that we know what the control group MST is in START?
Actually we do (with some caveats - see bottom) - because:
1) we know they used O'Brien Fleming to set the interim alpha at the interims
2) the first interim was 50% information fraction (and the combo of the above 2 implies we know that the p value had to be worse than about 0.003 and HR had to be greater than about 0.725 (from memory))
3) we know the rough blended median survival time at the time of the interim because we know the enrollment curve
4) therefore we know the placebo MST has to be greater than x (about 26 months if I remember correctly) or the trial would have been stat sig at the first interim.
Caveats to the above and its utility in estimating the placebo group MST for the original ph ii subgroup carried forward:
1) It is possible that they used an mITT for this first interim by excluding some of the patients not allowed to follow protocol due to FDA hold. But this doesn't seem likely for the first interim.
2) There are some protocol differences between the first and second trials that might, in aggregate, make enough difference to make a meaningful difference - e.g. the first trial required responsiveness to only chemo, but the second required responsiveness to combo chemo/rad.
Market Data 
Markets 
AI
