I suspect a big part of Provenge's problem is that many oncologists just aren't convinced in their hearts that it really works. The notion of a drug that doesn't impact any of the measures that they normally use but mysteriously prolongs lives I suspect produces some cognitive dissonance for them.
I'm guessing that by the time the patient has progressed on Zytiga many oncologists will think it's too later for Provenge. But if you had a drug that provided marked symptomatic relief for bone mets, they would still use it in a late-stage patient.