Obviously, the p value in 9902A as a stand-alone measurement is not supportive, but IMO the hazard ratio, the 3.3-month median survival advantage, and the ~50% three-year survival advantage are.
Just a FWIW - the p value, the 3 year survival advantage, the HR, ... are not independent. I.e. you should expect that about 16% of the time in a trial of a worthless drug that you will get a 3.3 month median survival advantage and a 50% 3 year survival advantage.
Perhaps the AC will not view it this way - somehow thinking that the clinical advantage is independent of the p value for a given trial size - but I would count on the one or two statisticians to make it clear that they are not independent.
IMPNHEO (inre stats anyway) 9902a being supportive turns on how prespecified the 9902a analysis was. I think that the very size of the improvement indicates a valid Regression, but you never know... .
Market Data 
Markets 
AI
