Nexavar and Sutent come to mind as both trials allowed crossover after the interim, thereby negating the survival endpoint.
Interesting point - the FDA documents for approval of Nexavar based upon the PFS indicate that they were still supposed to have a final OS analysis at 540 events (with assigned p value around 0.04). And Bayer's 2006 presentation of interim OS results (the last results that would exist before contamination by earlier x-over from the passed PFS trigger) also indicate that there was still supposed to be a final analysis at 540 events. It seems to be been lost that that final OS was nowhere near stat sig - the HR was 0.88 and the p value was 0.14. And I had to hunt fairly hard to find that data since it didn't get much airplay.)
Some data that the FDA plays favorites wrt big pharma
I don't believe any small biotech company has been granted accelerated approval by Pazdur w/o a confirmatory trial started. One can argue this is more due to problems with the initial application, but it's a coincidence.
Both Allos's Folotyn and Gloucester's Istodax got accelerated approval without confirmatory trials started.