The trial with Avastin was statistically significant with only 22 patients. In that trial, median PSA doubling time went from around 7 months to 12 months. We are now 16 months past the end of trial enrollment.
While I do not remember the exact statistical tests used in the PSADT trial, they undoubtedly used a paired test. (I.e. they had a 'before' measurement and an 'after' measurement). Paired tests are often much more powerful than population tests (e.g. where you only have an 'after' state) so stat sig is possible even in a small trial.
Simply intrigued by the amount of time its taking P-11 to reach the first two PSA endpoints and how many participants have not yet evented.
The tails on the curve of 'time to PSA recurrence' are very long. Ok, so it takes only, say, 7 months for the median time to recurrence. But it might take (don't have my calculator here) 14 months for 75%. It is the tails that are holding things up.
FWIW If they announced today that the trial would likely end in Dec sometime I would lay the odds of stat sig as something like the following:
a) If the TRUE placebo time to recurrence under this protocol is 6 month, then they have a 85% chance of stat sig.
b) If th TRUE placebo time to recurrence under this protocol is 7 months, then they have a 60% chance of success.
c) If the TRUE placebo time to recurrence under this protocol is 8 months then they have a 40% chance of stat sig.
The problem is that this protocol is moderately different than the protocols published in the literature. I'd bet the TRUE placebo time to recurrence is under 8 mo and over 6 mo. But where? As you can see above it makes a difference.
could you explain "the number of events and the HR that matters."
If you tell me the Hazard Ratio and the number of events then I can tell you the p value (or pretty close to it). So that is how trials are designed: "Assume an HR of x, what number of events do we need such that 9 out of 10 identical trails would be stat sig?"
Market Data 
Markets 
AI
