The issue is that the median TTP & PFS are ~6wks. RECIST by CT / MRI is also every patient ~6wks. So any patient who did not progress by 6wks checkup is bumped to 12wks. So the HR numbers are artificially inflated by the near binary nature of the measurements. You can easily visualize how the KM graphs look like in both cases. OS wont have this peculiarity. Thus, I am worried about the final OS especially since cross over is also allowed. So the question is what others will think when OS is not impressive. This became too risky to hold till asco IMO.