The conventional wisdom is that because OS events take longer to occur, OS analyses require a longer trial.  I will point out that in the analyses for the F1 subgroup, the OS analysis had a lower p value than the PFS analysis and I assume they were done concurrently based on a single data lock.  Even though the OS had fewer events, the better hazard ratio .62 vs .69 resulted in a lower p value.  So perhaps the conventional wisdom is skewed in this case because of the differing responses of patients to MS vs chemo treatment.  IMGN will have to look at the underlying data from F1 and decide on which endpoint or combination of endpoints best meets their needs.  Longer follow up of F1 patients to see if that survival comparison keeps improving with time may help.