Gnawkz,
---------------------------------
"Just to clarify your comment:
1. You believe that if the crossover group does not perform on par with the test group and shows a statistically significant gap, then the trial should stop.
2. This other individual you mentioned on another message board who has a much larger base of knowledge ... believes otherwise. That the DCVax-L trial designed with cross over in mind will not end early and rather continue till completion."
-------------------------
Looks like Flipper pretty much answered your questions, regarding PFS. The difficulty comes from the final decision on full approval being based on OS. Some thought Northwest would be able to use PFS to the very final approval, while others thought that OS would be required. I think the document that circulated nails that down, showing a plan by Northwest to use OS as the final measure.
Again, it is in OS where things get murky. The cross-over group is not a good representation of either the control group character, nor the experimental group character, and where it lies in between is not possible to nail down. That makes the crossover group OS data only useful in certain circumstances. In other circumstances, they would have to rely on historical norms, but I don't know if that gets done in this type of trial. There are definitely trials where from phase 1, to the end, there is no placebo group in the prior phases before the confirmatory portion or final phase, and the confirmatory also has no placebo. Thus any analysis would be based on historical norms. But I don't know if such has ever been allowed for a trial with a placebo in the earlier phases. DCVax-L had a placebo in the early phases because the evaluation criteria was PFS.
As far as the requirement for OS improvement over historical norms for approval goes... I don't know, but it is possible that even a small improvement would be enough if PFS passes the primary goal. They appear to have set down a criterion for the OS to be examined, but they set down no threshold for ok, nor a means to evaluate it. That is really where we are.
1) Yes. If the crossover group does substantially worse than the experimental group, with statistical significance, then I think that begs termination of the placebo for the sake of patients. Of course the FDA is probably not very concerned with what I think.
2) Regarding the opposing view by an un-named poster: I looked back over that exchange with him, and I think he was referring to a different issue. So, forget I brought that up.
News 
Market Data 
Discover 
AI
