The following are just my opinion.
It depends.
PFS by itself provides clinical benefit. If that benefit is large enough, your question might be moot.
If the clinical benefit is not large enough to make your question moot, but nonetheless statistically significant, then the concept is that upon the final PFS analysis (248 events in this case), the FDA can use the same trial for confirmation. The general regulatory idea under this scenario is that OS matures to such an extent, in this case by 248 PFS events, that OS is nearly complete. However, according to the European/German website, that analysis is not done until 233 events.
Beartrap alerted us in December 2015 that LP stated at the annual meeting if AA were to be sought, it would be done following the primary completion date.
To me, that possible regulatory pathway would ultimately require unblinding prior to 233 OS events. The OS data would then be followed unblinded thereafter. (Remember, the trial was randomized.) There are a few alternative things that could develop over that time. 1. statistical significance. 2. No statistical significance but trending toward treatment benefit. 3. treatment trending toward control benefit. 4. SS for control.
The biggest conundrum is the third alternative. There was once a trial for another therapy that ultimately showed clear although admittedly small benefit for the surrogate PFS in another indication, but slight benefit for control regarding OS. That case also involved a crossover, so the slight benefit for control was thought to be due to crossover treatment, yet the FDA was harsh, requiring another trial. In our case, the FDA required the crossover, so there is that if this third alternative were to play out -- hopefully we won't need to go down that third path. The degrees of results also seem to matter, so does safety as well as other endpoints.
OK, I'm going to try and make that my only post today.
Market Data 
Markets 
AI
