NorthWest Biotherapeutics Inc (NWBO)

[sentiment_stocks]

Yes 50% os events occurred a long time ago. So the median is set,

I don't believe that is correct.

Just because 50% of the events occurred, doesn't mean we've reached the median. If these patients had all enrolled on the same day, then yes, it would.

But their enrollment was staggered from 2008 to 2015. So when 50% of the OS events were reached in real time, they were not reached as they will be examined in this trial.

Say there were 5 people in the trial and they were enrolled over 5 years.

Well the first person enrolled in year one might have died in 2 months.

The second person enrolled in year two might have died in three years.

The third person enrolled in year three might have died in one year.

The fourth person enrolled in year four might have died in two years.

And the fifth person enrolled in year five might still be alive.

So that curve would plot like this...

Patient 1 - two months
Patient 3 - one year
Patient 4 - two year
Patient 2 - three years

Patient 5 is still alive and so is not on the curve yet. So right now, the median is 1.5 years because it's between one to two years.

So let's say that patient 5 dies at six months, and they stop the trial. Then the median for the trial becomes one year.

Patient 1 - two months
Patient 5 - six months
Patient 3 - one year
Patient 4 - two year
Patient 2 - three years

But let's say that patient 5 has been alive for two years and six months, and then he finally dies. By waiting (remember, this patient was the last to enroll in year five) the median becomes two years.

Patient 1 - two months
Patient 5 - six months
Patient 3 - one year
Patient 4 - two years
Patient 5 - two years and six months
Patient 2 - three years

So that is how the patients who actually die would count in the medianOS. Hopefully, this also illustrates how those final two OS events can move the curve, although at this point, it's probably a negligible amount.

So while I'm not sure if a censored patient that plots on a curve can be the actual patient that IS represents the median, I do know that the censored patients count towards determining the hazard ratio. And with almost 30% of them still being alive, the longer they live, the more the hazard ratio is improved. That may have been what Larry Smith actually meant.

But Linda Liau didn't think medians were as relevant when examining the efficacy of a treatment. She did say in her now deleted video from December 2016...

But also there is a need for relevant endpoints. You know, median survival, overall survival at one year, or even two years, is really… may not be the meaningful endpoint for these trials. It’s really that long term tail end. The hazard ratio at three years or four years is really what we should be computing for these trials.