When overall survival fails to confirm event-free survival, should the latter be used to set the standard of care?