Dendreon Corp fka DNDNQ

[nerdseeksblonde]

I think my earlier suggestions on
triple-blind analysis is still worthy
of consideration at DNDN and the FDA. Even if
FDA doesn't by it right away, they could publish
papers and get people interested in working with them.
<sarcasm>
While I'm not sure this will work as good as
threats against opposing oncologists,
</sarcasm>
if you had a blinded party that could analyse the data with
prognostic factors and predict which were drug
and placebo patients based on excess survival,
that would eliminate all the questions about
after-the-fact ad hoc rationalization. My wavelet
technique would be one way to separate them once
you have predictive model ( based on casuality as
much as possible rather than fitting).

To reiterate another point, if they had an MOA and all the gene chip results from prior patients and had calibrated the results from pure cultures of things like DC's or NK's they wouldn't have to guess on endpoints. They would probably
be able to tell confidence in PSA dynamics versus someother
endpoint and impress everyone when they got stat sig on
the chosen endpoint.