NorthWest Biotherapeutics Inc (NWBO)

[GermanCol]

GermanCol, your argument on censors being discarded fails, badly.

First, Flipper is correct. If the LTFU's are gone, that is because they were found. One does not delete censored data when the SAP says otherwise. This one is trial stats 101.

Ex, no, it doesn’t fail badly. I don’t think the censors for the first two years were discarded, what I think is they were not included in the JAMA publication because they were all part of the 99 arm and none of the 232 arm.

As you can see, the treatment arm in the JAMA publication continued having 232 patients as before, but there were zero censors for the first two years from randomization in the JAMA publication, while in the 2018 JTM interim results publication (that included all 331 patients blended) there were 13 censors for the first 19 months from surgery (16 from randomization) that were necessarily LTFU (lost to follow up) because of the time since enrollment of the last patient when that publication was done. All this is explained in detail in my post from almost 2 years ago that you are replying to, I don't know why after all this time, but I am happy to answer you.

Your key point is the read at 24 months where you assert the K/M shows 35% OS and thus no censors prior to then (when about 8 or 9 would have been expected in the 2018 dataset).

Problem is that 35% read. I put the graph into a digitizer and see the number more like 36% (precision is clearly an issue, even with tech assistance). And that is enough to allow for several censors, even 8 or 9 is possible.

That’s not my only key point. Other of the key points is for example the read at 12 months of 77%, because the earlier the censor effect occurs, the higher the effect. Regarding my 35% read vs. your 36% read it is your eye and word against mine and others can decide who to believe. But in any case, that 1% difference doesn’t allow for 8 or 9 censors as you mention.

Also, following is the Progression Free Survival chart from the results publication.



As you can see from the notes below the chart, there were 81 events in the 99 patients arm. That means there were 18 patients ( 99 – 81 ) that were part of the 99 that arm, but were not PFS events. So they are censors for PFS analysis. I think that the majority if not all of those patients were censored because they were pseudoprogressors, that being initially thought to be progressors, received DCVax after pseudoprogression as part of the crossover design of the trial. As a consequence of that, these patients were not included in the 64 patients rGBM arm, because they were not real progressors when they crossed over.

Following is a Chart that shows graphically what I’m explaining.



As you can see, there are also 17 patients that had their progression event counted but were not included in the 64 patients corresponding to the recurrent GBM arm:

81 (PFS events) - 64 (rGBM arm patients) = 17

I think the majority of these 17 correspond to patients that died before progression or others that were too sick to cross over. Some of these have cases have been mentioned by Linda Liau and commented by Flipper in some of his posts.

I hope this chart also this puts to rest some new FUD narrative that I have seen in new posts mentioning that part of the 64 patients in the rGBM arm are pseudoprogressors that helped having better results. That is not true and that is why of the 81 PFS events only 64 were included in that arm

Also I hope this chart reinforces why I think the censors in the OS charts for the JTM publication for the first 2 years corresponds to the 99 arm and none to the 232 arm. Because the majority if not all of the PFS censors in the 99 arm ended up being censors in the OS chart because they corresponded as I mentioned to pseudoprogressors that started receiving DCVAx, so they needed to be censored also for the OS analysis. On the other hand, for the case of the 232 patients arm, pseudoprogressors didn’t have to be censored for the OS analysis (only for PFS analysis) because they anyway started receiving DCVax since the beginning of the trial. There was no crossover.

What is a fact is that a "win" in the 232 vs 99 OS comp would be huge, even if not stat sig. It is still an endpoint of the trial. And team NWBO has buried it.

That is not a fact at all. I think it is exactly the opposite: The reality is that the less statistically significant the difference between the OS for the 232 arm and the 99 arm, the better the results for DCVax.

I will try to explain why as clear as possible, so we can end up with all this non-sense.

Following are the endpoints we are discussing:

1. The primary objective of this study is to compare overall survival (OS) between patients randomized to DCVax-L and control patients from comparable, contemporaneous trials who received standard of care therapy only, in newly diagnosed glioblastoma.

2. The first secondary objective is to compare overall survival (OS) between patients randomized to placebo who received DCVax-L treatment following disease recurrence, and control patients from comparable, contemporaneous clinical trials, in recurrent GBM.

3. The fourth secondary objective, OS, is overall survival compared between subjects randomized to DCVax®-L and those randomized to Placebo within Study 020221.

** The first one listed above is the one used to prove that DCVax works for nGBM, the second one listed above is the one used to prove that DCVax works for recurrent GBM and the third one listed above is the one that doesn’t have statistical meaning anymore, but you keep bringing up to generate confusion and fear.


And following is the explanation step by step:

- First of all, a more statistically significant endpoint for recurrent GBM (first secondary endpoint or number 2 in the list above), means that the 64 patients that received DCVax after recurrence are living significantly longer than the SOC.
- If this happens, it means that the 64 patients arm is getting closer to the 232 arm because that arm is over the SOC.
- If this happens, it means that the 99 patients arm is getting closer to the 232 arm also, because those 64 patients are a subgroup of the 99 patients arm.

So, as a consequence of this, the fourth secondary endpoint that you keep bringing to the discussion (number 3 in the list above) results being less statistically significant.

Going further, as a consequence of this, the probability of having statistically significant differences for the 3 endpoints on discussion (the primary, the first secondary and the fourth secondary) is very low. And because we know that the primary and the first secondary are statistically significant (which is greater news for DCVax for nGBM and rGBM), the most probable result for the fourth secondary endpoint you keep bringing up is that is not statistically significant.

Other important reason why this fourth secondary endpoint you keep bringing is not valid. is that as you can see from the chart above, only 17 patients or a little bit more would remain as part of the 99 original arm for comparison purposes, because the rest would have been censored for OS analysis after receiving DCVax on progression or pseudo progression, so there is no way to have a reliable statistical analysis.

Regarding what you say about the definition of left censors, it was the translation I found, but you can replace the term for LTFU (lost to follow up). Also when I mention right censored patients, is the translation I found for patients that have been in the trial for the time they are shown as a censor, without dying or progressing when the survival analysis was held, and they will continue in the trial.

*** Just in case, English is not my native language and this is not investing advice.