Dendreon Corp fka DNDNQ

[rancherho]

Thank you for an interesting refrerence. Appendix 1 as well as Question 17 is a worthwhile read.The whole approach to stopping rules used by DSMB's can be quite complex and are all based on that grey area when alpha is missed but statistical significance at the end of the trial is assured. If you eliminate the futility stopping rules since all Provenge and GVAX patients will have completed primary treatment before the interim look in each (and the CEGE CEO specifically stated that Vital 1 would have no futility analysis) that suggests that DSMB stopping rules (other than for safety problems) would only apply to efficacy stopiing. Perhaps this is why Gold has given mixed signals (eg, no p value given at interim) while appearing optimistic that final results will be known well before 2010. I would assume that a contributing factor will be that (1) it is a non subjective survival trial, (2) in a terminal (and painful disease) and (3)experienced investigators (including Drs. Small, Higano and Petrylak) are involved in large, multi institutional trials.

Any Bayesian analysis might use the Taxotere TAX 327 as a reference dataset wrt treatment effect, required events to stop the trial(both absolute and %), hazard ratio and p value at any potential stopping point. IAW, given the likely subsequent use of Taxotere after both Provenge and GVAX, if the prospective stopping point would not be likely to result in clearly superior results, let the trial continue. A summary of that data from NEJM is as follows:
"From March 2000 through June 2002, 1006 men with metastatic hormone-refractory prostate cancer received 5 mg of prednisone twice daily and were randomly assigned to receive 12 mg of mitoxantrone per square meter of body-surface area every three weeks, 75 mg of docetaxel per square meter every three weeks, or 30 mg of docetaxel per square meter weekly for five of every six weeks." "There were three comparisons of interest between the docetaxel and mitoxantrone groups: docetaxel given every three weeks was compared with mitoxantrone, weekly docetaxel was compared with mitoxantrone, and the combined docetaxel groups were compared with mitoxantrone. The study was designed to detect with 90 percent power a hazard ratio of 0.75 for death in the docetaxel groups as compared with the mitoxantrone group, with a two-sided type I error of 0.05 and with the data analyzed according to the intention to treat. The sample size was established as 1002 patients, and analysis was planned after 535 deaths had occurred. To allow for multiple comparisons, a P value of 0.04 was considered to indicate statistical significance for the comparison of the combined docetaxel groups with the mitoxantrone group, and a P value of 0.0175 was considered to indicate statistical significance for the comparison of each docetaxel group with the mitoxantrone group (all P values were two-sided), thus ensuring an overall significance level of 0.05." "One planned interim analysis of safety was conducted after the recruitment of 120 patients. No interim analysis for efficacy was performed." There were 672 men in the FDA approved three weekly docetaxel dosing and control arms in TAX 327. There were 166 deaths in the d3w arm and 201 in the mitantrone + prednisone (M_P) control arm (total: 367) (55% of the enrollees in the two arms had died.) The hazard ratio three weekly docetaxel was 0.78 and the Kaplan Meier stratified log rank p value for survival was 0.009, Confidence Interval 0.62, 0.94). The reported treatment effect was 2.4 months. http://content.nejm.org/cgi/content/full/351/15/1502

JMHO.