A correction to Sunstar's post. The enrollment criterion for the P2 was PS<=2. not PS=2 (Looks like an IHUB software bug on the <= symbol done with a < and a _ under it).
There were only a modest number of PS=2 though, so this is not a large effect on what an expected median would be.
There are other reasons to suspect substantially larger medians (on both arms). Improved supportive care, experimental 3rd line treatments. Look at the median OS numbers in the Doce arm for the recent Opdivo arm trial.
I would speculate 10 vs 15 months (for a modestly working Bavi) would be a better set of numbers to play with than 7 vs 10 or such.
Market Data 
Markets 
AI
