Most of my numbers come out very, very similar to yours but I'm not sure you're accounting for the time distortion at the beginning of the trial or the anticipated enrollment completion. For one thing, you have exactly a 2:1 ratio of treated to control arm patients among the already-enrolled. I do not think they contaminated the ongoing randomization ratio (and blinding) by trying to immediately 'catch up' on the 33 early patients.
Taking into consideration the 33 entirely or mostly experimental patients at the end of 3Q08, and assuming 312 patients are enrolled on August 1st, 2014 (this assumes rapid acceleration from as-yet inactive European sites), and considering the recruiting U.S. sites from 2011 through Sept. 2013, and typical per-site recruitment rates in similar trials, and so on, I think the total number of patients already enrolled by the first interim trigger was significantly higher than your 171, and that a higher number of the total progression events came from treated patients. That's okay though, because the average time those patients had been treated before the trigger was also significantly higher than my inference of your timeline. Even if 25 of the treated patients from 2008 had already progressed (perhaps also 3 controls I'm ignoring), they could first have averaged 18-20 months pfs (depending on whether control lasts 8.1 or 9) without losing consistency with the treatment arm from 2011-2013.
If you weren't considering the early oddities (initial patients, failed blinding, suspended enrollment period) and were assuming that enrollment wouldn't complete until early 2015, then I'd be interested to know how your calculations come out if you do adjust accordingly for the circumstances. Or for that matter, if you were intentionally ignoring those things (e.g. think the company is still being naively optimistic about trial dynamics as in times past), I'd like to know that as well.
In any case, thank you for your contribution.
