David Miller (subs reqd) has a lengthy report / speculation out today on what enabled Pazdur to trash Provenge. If it's right, the chances of such circs repeating next time around might be slim, so greater FDA flexibility about data purity might be a fair assumption.