>>> The FDA has a long history of allowing x-over in a trial with a surrogate endpoint after each patient has met the surrogate. This is not much different. So I don't understand your point. <<<
I don't have a problem with crossover designs , per se , just when the crossover experimental drug is not the same as the treatment arm experimental drug.
Imagine a situation where the freezing process generates factors in the crossover Provenge that add to efficacy , or add to toxicity , or both. You would have a hard time sorting it all out in a trial like this. If the crossovers had lived twice as long as those who received fresh Provenge , could you tell whether it was due to the freezing or , instead , to the delayed administration ? You'd have a similar problem if frozen Provenge caused patients to die much quicker than the non-crossovers.
There may be no such confounding in these trials but it's still a bad design for pivotal trials in first-in-class drugs , IMO.