How long has the latest and greatest variant of "rPFS" been around?
If you are saying that deciding after initial trial design to add analyzing the data with a modernized rPFS type method would be as improper as adding a subgroup after the fact, then I would disagree, in spite of my respect for your opinion.
It's just not the same thing. The only reason to push that argument would be if you were a competitor saying, "They didn't say mother may I".
Immunotherapy is new. I believe that it is unreasonable to require all aspects of data analysis in a lengthy immunotherapy trial to have been defined up front when the knowledge of what is proper data analysis has mushroomed in recent years. That is the kind of flexibility that should have been added to the FDA decision making as the result of recent changes.
This particular issue of progression vs pseudoprogression is particularly important regarding the use of the most accurate method recognized by the FDA, regardless of when it was recognized relative to the trial design window. It does add another chance for a random good result, but to the extent that it is a better method, that random factor is limited. Combined with the FDA influence in a trial design that confuses OS... they should yield here if this is an issue in contention, and I believe they will; Though it concerns me a great deal that Reefrad seems to also think that late pseudoprogression confused with progression would jeopardize PFS. Still, in spite of you two heavy weights aligning on this, and in spite of the fact that I can't quite wrap my head around how to fix such a situation, I still believe that a recognized improved method for post analysis of progression vs pseudoprogression is a good thing and could some how be utilized to generate a more accurate picture of efficacy. And again, I believe that the FDA would consider such analysis in their consideration of approval.
