Changing endpoints from the failed PFS to the secondary OS between the arms would have been stat-wise a no-no, but would have drawn some support. I said, and still maintain, that has the original trial showed a PFS fail but solid OS it would be a very interesting play and could well be approved.
It is the change from OS between the arms to an ECA where this goes off the tracks. Formally, that also was stat wise invalid as it was admittedly based on information from the trial.
Speculatively, I will go farther and say I am 90+% confident that NWBO was actually running IAs on the OS in 17-18. They formally stayed blind, but had advice on if the long tail could actually work against the comparator arm. But that is just speculation.
But past the frequentist sycophants, the ECAs are clearly poor quality. Not just the extent of resection and steroid use, but several other issues all of which cannot possibly be adjusted for as they have no patient level data.
The NWBO SMA tries to spin this as a move to "the gold standard", far from reality.
The SMA and NWBO do look to be the right rack now though. Try to manipulate the price up by using the anti-MM crowd. That will let Bigger et. all. sell what is left to the greatest.