Replies to post #1418 on Dendreon Corp fka DNDNQ

[p3analyze]

GOod observation, having KM curve beyond median for both treatment arms would be kosher to allow estimate of median OS. I agree, 275 probably is a better event goal.

Just realized my estimate did not factor in the 2:1 ratio. The updated calculations are as follows:

End of Deaths
06/2007 196
12/2007 248
06/2008 291
12/2008 325
06/2009 355
12/2009 377
06/2010 397

We should have the data out by June 08.

[iwfal]

the target of 75% of events is the extreme end of the spectrum, and usually just past the median is reasonable for a survival trial.

This is plain not true. Almost all cancer trials with relatively short time to event use near 75%. Trials with longer times to event can use substantially lower fractions (e.g. MACE trials), but cancer trials are almost always between high 60's and high 70's - with most clustered right near 75%. If you have an exception I'd like to see it.

9902a is among the longer times to event for a cancer trial (median SoC survival of probably 18 months with the new population vs typical cancer trial survival times of 4-12 months), so it is possible the target number of events is as low as high 60's (say 67%). But I'd be very surprised if it were much lower.