" ..the FDA would allow them to pre-screen responders.."
In general, the FDA has no problem with the concept.
In this case you have issues though. First, I doubt there exists a way to screen w/o providing several rounds of treatment. This kind of invalidates the trial, yes?
Second, as has been pointed out, the evidence so far does not support the claim the drug works in the subgroup (just that there is a corrollation, which could very well be that the drug and IR identify survivor types). Thus, you would certainly have to go back to P2 tests and we are now well over 10 years out or so.
Maybe there is some magic data out there that escapes this trap, but you still have 6-7 years or so to run another trial.
Another one bites the dust.
