I'm still not quite sure what to think about "enrollment" vs. "randomization", personally (which I'm perfectly aware doesn't count for anything anywhere) I would consider a patient enrolled if he met the entrance criteria for getting the Lupron-shots in the first place.
You can make a strong case for either definition of 'enrollment'. E.g. there are certainly other trials where there is a run in period and if you don't meet the criteria after the run in period you are not 'enrolled'.
My point isn't that I am certain one way or another which definition Dendreon is using, but that it is a big difference and given the quality of IR I can't know which is correct. So I have to assume worst case - that shortly after enrollment the first LHRH treatment starts.
it would also explain why Dr. Kylstra mentions 160 patients as treated
I presume you are referring to Klystra's presentation on the PROTECT protocol? But that was several years ago. 160 as mentioned in that briefin is the target enrollment. Is there some more recent update????
those who achieved the undectable PSA level - represents a special group that may progress slower.
That is true. But I don't get the pertinence to this trial per se? The protocol does not require undetectable PSA at any point. It only requires that PSA be less than 1 ng/ml after LHRH agonsit therapy, which is way, way above 'undetectable' levels.
I get kind of worried when reading Iwfal's latest guestimates on the chances of stat sig of P11, I would certainly have expected these numbers (chances of stat sig) to be more favourable for a long position...
There are, of course, a bunch of assumptions built into those numbers. E.g. a key assumption is that the required number of events is 75% of the target enrollment. If the number was only, say, 65%, then the chance of stat sig given 18 months from final enrollment to unblinding is significantly greater. But, for instance, I think it very(!) unlikely that the required number of events is much lower than 75%.
Another way to think about this is that 9901, which had about 90 survival events, was about 50-60% powered (don't remember my exact calcs) given the found HR of near 1.8. PROTECT is probably planning on about 120 events. More, but not a lot more. So unless you expect the HR to be a LOT bigger than 9901 (in which case the requisite number of events will take forever) you shouldn't expect a highly powered trial.
Final note - Essentially no one who is LHRH agonist naive fails to respond to their first treatment. Virtually certain to be a lot less than 10% (caveat is that this exact protocol has never been published before - but no published paper of naive patients shows anywhere near 10%). So 10% failure would imply a lot of 2nd and 3rd cycle patients, who will have a PSA recurrence a month or two faster. But again, we don't know.
AI
