Replies to post #677056 on NorthWest Biotherapeutics Inc (NWBO)

[Doc logic]

exwannabe,

This whole response of yours tries to ignore the advantage given by treatment with L when pseudoprogression occurs whether during the course of treatment with L on the treatment arm or with SOC/placebo crossover when no true progression has occurred. Treatment with L during or after chemo/rad pseudoprogression creates a survival advantage that has been duly noted in previous studies.
The only way to properly measure pseudoprogression advantage is to compare each arm to external controls that never had L after SOC induced pseudoprogression. The data retrieved and analyzed from this trial is built around this very issue so that a greater body of evidence could be used to validate the differences.
The only bone you have to pick is with finding out how many SOC pseudoprogressors are in the comparators and I assume that the independent composers of the ECAs did this to the satisfaction of regulators and JAMA Oncology. So you may claim there is bias but there is no proof of bias. I may work under the assumption that SOC induced pseudoprogression occurred in the ECA groups but I have no proof. The fact that both treatment induced and SOC induced pseudoprogression has had such a big impact on this trial, though, leads me to the assumption that it has been accounted for in the ECAs and if I know Dr. Linda Liau very well at all, this issue would not be ignored or allowed to be challenged because of flagrant misappropriation of the data sets; ). Best wishes.

[sentiment_stocks]

So they could detect the psPD in the 99 but not the psPD in the 232. Sure.

I have to agree with you on this point.

[GermanCol]

So 13 of the 13 early OS censors just happened to be in the SOC arm? Wow. What are the odds. Try ..00000014738810721990. Any RA would have a field day with that one.

That didn't happen by chance, because the odds would have been very low as you mention. Let me explain it again in another way.

During all these years, we have seen the following KM charts, number of censors and events:

1. Final PFS numbers for the 232 patients arm from JAMA publication: See chart and notes below.

191 events, 41 censors (232 - 191 = 41), which correspond to about 20% of the patients.


2. Final PFS numbers for the 99 patients arm from JAMA publication: See chart and notes below.

81 events, 18 censors (99 - 81 = 18) , which correspond to about 20% of the patients.




3. Final OS numbers for the 232 patients arm from JAMA publication: See chart and table below.

There were Zero censors for the first two years, as explained in previous post 479394. That is why the survival % directly read from the chart is almost the same amount as the division of: At risk / Initial At Risk. Because of zero censors. See also notes below the table.




4. Interim OS for all 331 patients blended, from JTM publication: See chart below.

There were 13 censored patients (vertical lines in the chart) in the first 19 months and all corresponded to lost to follow up patients, because of the time of enrollment of the last patient in the trial. All these as explained in more detail in my previous post 479394. https://investorshub.advfn.com/boards/read_msg.aspx?message_id=168995681



So, as you can see, the only KM analysis that doesn’t have censors is number 3 (Final OS numbers for the 232 patients arm from JAMA publication) and all the others have a very high percentage.

The first natural thought that came to my mind was that if there were 13 censors for the first 19 months in the interim blended (all 331 patients) OS KM chart and none for the final OS KM analysis for the 232 DCVax L arm, and those 13 censors corresponded to lost to follow up, then all of the 13 censors for the first 19 months should correspond to the original 99 patients arm. Just simple logic, as the total 331 patients come from the addition of the two arms mentioned (232 patients arm and 99 patients arm).

So now, because I don´t believe these things happen by chance, I started to think what could be the logical reason for that. And finally came to the conclusion that the majority if not all of these censors should be related to pseudo progressors thought to be real progressors, as I will explain.

In the 99 original placebo arm, every pseudo progressor crossed over because it was thought to be a real progressor is classified immediately not only as a PFS censor, but also as an OS censor. On the other hand, in the 232 DCVax arm, every pseudo progressor thought to be a progressor is only a PFS censor, but not an OS censor.

This also explains what you are saying in the following quote:

So they could detect the psPD in the 99 but not the psPD in the 232. Sure.

The reason is related to the cross over in the trial design. For the case of the 99 patients original placebo arm, the patients didn't receive DCVax at the beginning, but at progression or pseudo progression thought to be progression, they were crossed over and started receiving DCVax.

On the other hand, for the case of the 232 patients DCVax arm, all the progressors or pseudo progressors thought to be progressors have been receiving DCvax from the beginning anyway, so there’s is no reason to censor them for OS analysis.

The 1% difference can explain 13 censors in theory. Censors only effect the step size after them. So if most censors have few events after them they have little effect.

Again, there is not a 1% difference as you are trying to say, it is 0.1% and also 1% would not explain 13 censors. Also censors only effect the step size after them as you say and that precisely reinforces what I am saying, because the sooner the censor occurs, the higher the steps from there, so the survival decrease is higher.

And now you are off the deep end. Anybody with the slightest clue knows that the the original OS of 232 vs 99 would be a huge.

No, that is not true and I gave you my detailed reasons. Now give the reason for what you are saying.