According to the CC the delay in operable results is "logistical." The conference in early Oct has an early due date that Roche couldn't meet given they are just seeing the data for the first time this week. Then next conference is Nov and they have submitted an abstract. It sounded like they would present the data there if the abstract is accepted, and later (at the beginning of 14) if not.
So it has, according to Curis, nothing to do with the study or the data.