>That's a separate question. Your traditional argument has been about the prognostic factors, all of which were prespecified. So was the Cox analysis.<
True, true and unrelated. Two separate issues at play here, imo. Number one is that I don't believe the cox was the prespecified primary analysis to be used to assess survival. It was simply one of the exploratory analyses they had. That we can agree on. Number two is that the procedure for doing the cox analysis was different from 01 to 02a, and that's where the prognostic factors come in.
I certainly agree that the 20 or so prognostics were laid out as the usual suspects ahead of time. I don't think there is much to argue about there, as these types of exploratory analyses are just good ways to extract value from the clinical data. But I'll bet a months salary that the 02a protocol did not prespecify analyzing it using the specific covariates that popped up in a different trial. Had they used 02a to distill the covariates and applied that to 01, you would have gotten different p values / HR all around.
Market Data 
Markets 
AI
