This explanation comes closest to what I think actually happened. But subsequent events are still quite puzzling, and the silence from the company regarding many of the details surrounding them makes it difficult to assess the prospects for approval.
If there was a futility finding on PFS, NWBO did not reveal it, which I have read they did not have to. I presume the secondary endpoint at that time, Overall Survival, was next in line, but the company decided to wait for several more years in order to take into account delayed treatment effects. What is not clear is what triggered the decision to switch to external controls and why the SAP was changed just before data lock. I know the difficulty posed by the crossover design is cited as the reason, but that was known even before the interim PFS analysis. Clearly, at some point the external controls option was deemed superior to staying with the randomized control design, even though some of the external control trials had no data on several variables used for the matching, and there was no patient level data at all.
At this point it seems it comes down to how regulators view the much less than perfect match in the controls, and the relatively small size of the survival advantage.