If you look at the second interim analysis on the CLDX GBM vaccine trial, I think you'll get a good flavor of VBLT's. The primary CLDX endpoint was futility(they failed)with both groups(vaccine and non-vaccine)having exactly the same events, median overall survival, ect... there was a pre specified efficacy analysis with a p < than .03 which would have been a fairly hard hurdle to crack. I've read in some patent documents on VB111 they planned a similar futility analysis and also a efficacy package in the interim with a p < than .03.
I have various questions I'm not sure about....
1. Assuming they don't pass futility muster, will they obtain any information... KM curves at the time, or at least how many events occurred in both groups, or any other data... amount of partial/complete responses ext.... Dror may have said no, but I'm not sure this is the total case. The party line is only futility, but I think they will learn more IMHO.
2. The way the initial interim analysis was set up, I would have said there is no chance for success with a strict 91 event cutoff since there was no potential time on drug stipulation, now there is... it's not just 108 events, but also that half of the patients will have to have had potential 12 study months on the drug... this makes it more possible to succeed on an interim assuming biological activity of VB-111. I'm assuming about end of July -- early August is when half of the patients were randomized since the pace of enrollment escalated... this pushes the potential interim analysis at around August 1's? of 2017 at the time about half of the patients are enrolled.
3. This leads to the rationale for the change. VBLT petitioned the FDA for the changes not that long ago... it was disclosed in a conference call, but surely planned out many months in advance. So, IMHO it was not based on "what they knew or even surmised" but based on increasing the chances for hitting the hurdle an ending the trial early by achieving success on the interim analysis. Since they strengthen the overall statistical package, they stated this was the reason.... but IMHO, my hunch, is they are hoping/planning the changes based on hitting the interim hurdle.
Take a look at the KM curve separations in the rGBM VB111, the curves separate nicely at around six months and expand a bit thereafter.... same appears to be true for ovarian and thyroid. On the other hand if you look at the rGBM trials, the partial and complete responders did not occur right away, but after several months. Assuming VB-111 works, it will take time... 6-9 months in my opinion to see clear distinction between the groups. Still it's a hard hurdle to pass with ending the trial early by hitting the interim analysis. I would posit about 10 % chance of flunking futility, 30 % change of ending the trial early, and 60 % chance of continuing the trial; before the changes I would have put it at 10 % flunking futility, 85 % at continuing the trial and 5 % of hitting the interim analysis and ending the trial early.
