>In a case like this it is helpful to see the K-M curves themselves. Comparing only the median values of the active-arm PFS and OS to their placebo-arm counterparts can be misleading.<
For overall survival i definitely agree. I've already seen the k-m curves for the finished phase III. In the subgroup of interest, there is no evident treatment influence on the k-m until 8 months. Also from the look of the k-m curves, it looks clear that the lack of PFS improvement isn't simply due to a consideration of a single point.
Heck what do i know? I'm sure the trial will work.