The 33% improvement in PFS is not an actual requirement for success, but rather the modeling assumption used for powering the trial. Any statsig PFS benefit relative to the control arm will technically be a success, which is why the sell-side analyst’s comments about expecting a modest PFS benefit and a 60% probability of success are not contradictory.
The problem with a small absolute PFS benefit may come into play during the ODAC panel, however, so the absolute size of the PFS benefit does matter apart from whether or not the data are statsig. Regards, Dew