If you read the paper, it was disclosed that an IA happed in '15. Given the PFS results, I can't see it being anything other than a futility rec, and reason for the FDA hold. So, they changed the endpoints and cherry-picked data to support them. However, nothing has changed. The trial failed*.
* For PFS (original primary endpoint), the hazard ratio was 1.1. So those in the vaccine arm did worse than placebo. They did not report the hazard ratio for OS (original secondary endpoint), but the blended mOS was 23.1 months for placebo, while for the vaccine arm it was 22.4 months. This means the vaccine did worse than placebo. So, the trend was to harm, not benefit.