Avid Bioservices Inc (CDMO)

[tradero]

I believe that 1st lookin has NOT taken place yet… I have done another simulation that estimates possible dates for the lookins and they seem to be quite later than many here have expected. Actually 1st lookin will not happen until October 2015 (if both Arms BAVI and Placebo perform poorly) and Decer 2015 if BAVI and Placebo Arm perform really well (according to past records)

My approach goes as follows… I have used the excellent spreadsheet that Golfho prepared and in particular one of his 4 Scenarios that assumes that a patient is enrolled 4 months after site opening and 4 months between enrollments

ENROLLMENT
These results gave 595 patients enrolled by November 2015 and that fits very well with the comments from PPHM: “Enrollment is going according to plan (end of 2015)”


So I took this numbers for this new approach.
Also any patient should belong to any of 8 groups according to the time they event : 3 , 5 , 7, 9, 11, 13, 15 months or they are survivors. Patients will be grouped by the months they are still alive after enrollment.
Let’s also consider 9 Cases (Scenarios) depending how good/bad the Bavi & Placebo Arm work. The percentage of “eventings” for each group tries to have a Median OS of 7,9 or 11 for the Placebo Arm and a MOS of 10,11 or 13 for the BAVI Arm

EVENTING GROUPS AT DIFFERENT SCENARIOS & PERCENTAGES OF OCURRENCE



So again any patient should belong to any of these 8 groups (according to the time they event : 3, 5, 7 months etc.) and Patients will only start “eventing” after at least 3 months after they have been enrolled. The calculations of “events” will be done according to the forementioned percentages. Do not mistake Enrollment Groups with Eventing Groups.
Calculations were made and These are the results per Case of number of Patients who had “evented” at the beginning of each month in 2015

RESULTS PER CASE (SCENARIO)



So not until next October 2015 we find 169 events… and that corresponds to a bad performance of both the Bavi and the Placebo Arms.
For a GOOD Bavi (and Placebo Arm) performance (Case 1) we would have to wait until December 2015 for 1st lookin
If you have read this far you deserve some more detailed explanation on how the calculations were made.
Let’s find out how many events took place beginning of this month (July 2015) and for Case 1. We should start by looking from 3 months behind (April 2015) backwards till 15 months behind (April 2014) to come to the following distribution of patients according to the time of enrollment:

These numbers can be checked, for instance in April 2015 (3 months before July) the total number of enrollments is 308 (see at the top), etc.
So we start with the group that has been enrolled more than 15 months: 1 person.
Survivors: 1 x 0.04 = 0.04 (not considered) Group of Eventing in 15 months = 1 x 0.05 = 0.05 (not considered)
“Total to exclude” = 0 + 0 = 0
We follow to the next Enrolled Group: 13 months =< enrolled <15 months. Number 9 – “Total to exclude” = 9
Group of Eventing in 13 months = 9 x 0.18 = 1.62 (we consider 1)
“Total to exclude” from this group and prior = 0 + 1 = 1
We follow to the next Enrolled Group: 11 months =<enrolled<13 months. Number 14 – “Total to exclude”=14–1=13
Group of Eventing in 11 months = 14 x 0.23 = 3.22 (we consider 3)
“Total to exclude” from this group and prior = 0 + 1 + 3= 4
We follow to the next Enrolled Group: 9months =<enrolled<11months. Number 77 – “Total to exclude”=77–4=73
Group of Eventing in 9 months = 73 x 0.23 = 16.7 (we consider 16)
“Total to exclude” from this group and prior = 0 + 1 + 3 +16 = 20
And so on and so forth… until we come to the number of 70 events by July 1st

CONCLUSIONS
If the enrollment model is relatively accurate (which I think it is) we can conclude from the results of this approach that even with a bad performance for the Bavi or Placebo trials the 1st lookin does not seem to happen before October 2015. From October 2015 onwards seem to be reasonable time frames for the 1st lookin to happen. However 1st lookin and 2nd lookin are very close to each other.

I will be happy to listen to any arguments against this approach. Unfortunately I have had no one checking the hypothesis I have taken nor my calculations, so I admit errors are possible… Yet the results say we will still have to wait a while until the lookins take place. No problem for me.
If anyone is interested in this spreadshit and wants to play around with different strategies please send me an email (my email is [e-mail]). The spreadshit is based on an impeccable Excel that Golfho made.