Replies to post #643158 on NorthWest Biotherapeutics Inc (NWBO)

[flipper44]

Continued:

The Kaplan-Meier (KM) curve assumes that censored subjects have the same probability of survival as those who are not censored, up until their censoring time. In other words, the KM curve considers that the probability of survival for censored subjects is representative of the entire group at risk, assuming no systematic differences in the risk of the event between censored and uncensored subjects.

[exwannabe]

There was a km calculator someone provided on ihub years ago, and if you added censors prior to the median, your survival improved.

Hardly need a K/M calculator to do the high school math.

Of course if you just add censors the survival will improves because you are adding patients who lived for some period of time w/o dying. But that is not the same as what happens when you censor patients that are in the trial.

When you take a dataset and randomly censor patients the K/M will on average not change. That is the reason why it is nearly universally used in situations where the data set is incomplete. Every OS number LL is mentioned in this trial is a K.M calculation despite the rants of longs here asserting it is not real.

As far as maturity, the data presented by LL at SNO was fully mature to about 34 months. That means the median and 2 year OS were 100% mature and the 3 year was damn close.

And you are wrong about 3 censors in the JAMA paper. If you think you can assume the difference between at risk and %OS is numerically the censors, bad math, it can be as much as twice that, So you have 5-10 censors in the JAMA OS. There may have been a few more early censors in the '18 data, but not some huge difference that makes the comp bogus.

Regardless of all, you still have the core fact that this is still an endpoint of the trial (no matter how many times Hoffy and Bala deny it). Anybody with a brain and some basic understanding knows it is important. And NWBO elects to keep it hidden.

As always, whenever NWBO fails to disclose something, it is almost certainly not good news.

[iclight]

You can’t legitimately compare an immature line with dozens of censors with a mature line with three censors.

You do know that they actually have the mature data for that comparison, they just won't release it, right? It would need to show major separation after 36 months in the survival curves but there simply aren't enough patients who made it past 36 months to show efficacy or they would have overlaid those data sets and set things straight.

Moreover, the argument of comparison is for not, where NWBO’s position is that there is crossover efficacy.

Which they haven't proven as they won't release the complete data set for that either. Just like the don't have the complete datasets for the ECAs.

LL: "Everyone is living longer and it's not helping our trial". Selection bias is a bitch.