Avid Bioservices Inc (CDMO)

[CuresForHumanity]

CP, thanks. First, just want to remind everyone this is all speculation. Basically, If we assume that N=582 and first look-in occurs at .8*.33*582 ( which is approximately 154) and we hinder the simulation at 3*the n=8 dropout – I believe JQ mentioned ( so we have 24 dropouts) and we further hinder the simulation by giving the control arm a MOS of 10 versus 9.4 in other trials mentioned. Then, if we use CP’s enrollment data and use a MOS of 15 for the Bavi arm vs MOS of 10 for the control arm – we may have already triggered a first look-in. So either we are doing around 15 MOS or better or look-in has occurred and data is being evaluated. When I ran this simulation as of today we would have 164 events even with the 24 dropouts. I really did this randomly using modulus on the primary key – so the buckets aren’t perfect or completely realistic although they do produce the 10 vs 15 MOS.
BAVI events as of today in this simulation would be 55 ( took out the dropouts) vs the control arm of 109 events. I am also questioning if enrollment was really this slow – only 27 enrolled in the first half of 2014. This makes a big difference on the simulation, if enrollment was not this slow – then I might think that they already had first look-in and are reviewing data – assuming they need 154 events to trigger first look-in.