Lot of assuming going on here. No more likely than them changing the primary endpoint when the IA showed likely failure and the DSMB provided a futility rec.
They changed the primary endpoint to the secondary of OS (which everyone on the planet agrees is the gold standard) when they suspected that the vaccine might be causing psPD,
The fact that they admitted an inability to identify psPD applies to their inability to determine it in the new bullshit rGBM arm too.
They changed the primary endpoint to the secondary of OS ...
No, the secondary was OS between the 232 randomized to -L vs the 99 randomized to SOC. If the had changed to that it would have caused far less statistical concerns that the change they made.
... (which everyone on the planet agrees is the gold standard) ...
The gold standard is OS between randomized arms. OTOH, OS to to external controls has issues.
...when they suspected that the vaccine might be causing psPD...
They changed it after they knew the results of the IA that informed them that the PFS endpoint was not going to be met,
... , and while they were still completelyblinded to the randomized data.
Being blinded to the patient level data does not mean they do not know of information derived from it It also was not a very high quality blind given Cognate (then owned by LP) had access to the unblinded shipping records.