The small number in subgroup analysis could easily skew result.
FWIW and as a summary closure, I believe you and I have two disconnects:
a) I believe the primary function of ph i and ph ii is to weed out drugs. And that the evidence coming from ph iiis is that this isn't done anywhere near enough.
b) I believe that the chance of this particular drug being successful (OS HR<=0.82) in this indication (HG wo TN, refractory BC) is less than 15%.
That said, I could live with the compromise you suggested.
PS Why are you not focused on continuing the TN wo HG population? Statistically they were, essentially, identical results with p values not that much different (both very near 1).
Market Data 
Markets 
AI
