I have spent time running MANY different curves with different tails and median OS’s based on numerous available curves and others that I modified/blended – I have not “settled” on a final set of curves yet but would like to post my observations so far (note that all will be in ranges since I ran each model ~ 200 times and the ranges that I discuss are the 80% ranges – given the fact that the tails are so long on some of the curves – you can understand how some of the possible results out of the same curve can have a 95% range of 3-4 months.)
1 If we assume no benefit at ~17.5 months median OS (NWBO curve), the trial should have ended sometime in Jan. At ~19.3 months OS (ratio-ed UCLA curves) with no benefit, sometime between mid Mar and mid Apr (makes sense so far…remember the curves do have slightly different shapes.)
2 If I assume at 19 month OS for the control and 21.5 OS for treatment I get an 80% range from mid Feb through mid May (so far this is my preferred curve – gives some safety margin assuming a 2.5 month benefit with an extended control OS).
3 At 19 OS and 23.8 OS (a modified IMUC curve that has a shorten 100% survival period and same shape through the majority of the curve until the tail - dropped this down to 40% survival at 30 months and then drop off slowly.) I get an 80% range from end of Mar through mid Jun.
4 At 17.5 OS and 20.3 OS I get an 80% range from very end of Feb through mid Apr.
5 If I run 19 OS and the P1 results (38 OS) and then we don’t see the 32nd event until end of this year, beginning of next!
….and many, MANY more curves and runs – I guess bottom line, I would feel best with the 32nd event coming in Apr-May time frame – this is where I got the most hits with the different curves/OS’s so this is my prediction for now. When I get a chance I will plug in numbers that Disco says are sfnsurgsen’s in his post today and compare.
....and remember this is only a model with assumptions that may or may not be correct but at least it is better than having nothing =)
(4 At 17.5 OS and xxxx I get an 80% range from very end of Feb xxxxxxx.)
Being a poor dumb biochemist and claiming no expertise in modeling survival curves, medical expertise in treating GMB, dendritic cell science, etc. Having done a lot of reading and following NWBO, IMUC, CLDX, DNDN........ I BOLDLY predict that this is the results we will see! Why ..........because I believe mangement let slip that they are very confident the 32 death event will occur this 1st QTR. It came out too easy ,and to shrug it off as a mispoken answer by a new management member is ........well its very naive.
This is half the number of events required to be able to unblind the trial. At this point, there is a scheduled review by an independent committee to see if the trial is 1, safe and 2, not a waste of time lives and money. If they determine that both conditions are good, the trial is allowed to continue. The timing of this milestone can shed some light on how well the treatment is going based on a range of ASSUMPTIONS as to how the control group should progress.(based on historical results). This is more clearly discernable in this case because imuc published their enrollment curve as part of a presentation earlier. That takes away some of many variables.
BTW - I should add that when I say the "Apr to May time frame" that is the soonest that I would want to see the 32nd event and still feel comfortable investing more (anything sooner and I'll have to re-evaluate my current position). If we get to mid May with no news then I'll feel very good.