NorthWest Biotherapeutics Inc (NWBO)

[AVII77]

The reason patients agree to join trials in the first place is to have the opportunity to perhaps receive a better treatment than SOC. As AVII77 says, futility does not necessarily mean that the treatment does not work but I believe that regulators want patients to always have the right to know, at the earliest appropriate time, what the futility situation is so that they can make their own best informed choice. I believe that clinical trial guidelines specifically point to this and that this is the most ethical way to proceed.

I generally agree with what you say above with the exception of the bolded "futility situation".

One reason these DSMB's are constituted is because they are composed of members with the scientific and statistical background to understand the nature of evolving data. They are supposed to be impartial and unbiased. One of their roles is to sort of enforce a Ulysses Contract; tying the researchers to the mast to resist the sirens song of promising (but not conclusive) data as the trial evolves (Sorry, I love that analogy, it's not mine but I don't remember where I got it).

It is only if overwhelming early data (positive or negative) is seen that they can "untie" the trial before its' conclusion.

And they use statistical tests to guide them (they are not bound by the tests, they evaluate everything in context).

So, I prepared the figure below (from the SPRINT trial investigating aggressive BP lowering) for something else but further edited it for this discussion (all the colored text is mine overlain on a figure from the paper buried in a supplemental appendix).

It shows the stopping boundaries for the trial along with results from the first 5 interim analysis's. The trial stopped at the 5th interim (it crossed above the efficacy boundary).

I also sketched in blue a hypothetical futility boundary. Notice it asymptotically approaches the efficacy boundary at further time points (100% information fraction=100% expected events).

If they were to do an interim with just 5 more events still to acrue, it had better be damn close to p=0.05 because if it's not, the chances of those 5 events bringing it better than 0.05 is very small. Conversely, earlier in the trial the futility boundary is close to the what I labelled the "harm" boundary which is the line below which Placebo is convincingly better.

I ask you:

What would you do if you were on the DSMB and saw results as shown with the red dot (and green arrow)?

I would also ask you: what would you do if the red dot was below the 0 line (not conclusive "harm", just placebo "trending" better) and you also saw the OS data trending positive?

Would your knowledge of response dynamics in immunotherapies affect your decision considering the endpoint is PFS?

These are serious questions intended to illuminate my theory here and I hope you give it some thought.