NorthWest Biotherapeutics Inc (NWBO)

[GermanCol]

Lykiri is correct. That information came from NYAS data:

Here are the NYAS DATA:

64 rGBM DCVax-L patients (crossover group)
- 44% methylated (28 rGBM DCVax-L patients)
- 52% unmethylated (33 rGBM DCVax-L patients)
- 5% missing (3 rGBM DCVax-L patients)

So that actually means that there are about 23 methylated and 10 unmethylated patients in the group of 35 patients that weren't included in any of the nGBM and rGBM arms. And that, together with other factors, accounts for the success of this group. This is actually not a negative and could be fully explained by reasons different to DCVaxL, and reasons related to the effect of DCVaxL, as follows:

1. MGMT Methylated patients have much better prognostic, so it is expected that from the 99 original placebo group there was less recurrence for them than it was for the MGMT unmethylated patients. And that was the reason why at the end this group of 35 had a higher proportion of MGMT methylated patients.

2. I think also that in the group of 35 there is a higher percentage of IDH-1 mutated patients (who have much better prognostic than wildtype) than in the nGBM and rGBM arms because of the following reasons:

- The numbers highlighted in yellow in the final row of the following figure, taken from slide 19 of 48 of the final results presentation, are the adjusted for rounding percentages of IDH-1 Mutated (3%), IDH-1 Wildtype (88%) and Missing (9%) for the nGBM DCVaxL treatment arm and this corresponds to 7 IDH-1 Mutated patients (3% x 232 = 6.96)






- The numbers highlighted in yellow in the final row of the following figure, taken from slide 21 of 48 of the final results presentation, are the adjusted for rounding percentages of IDH-1 Mutated (3%), IDH-1 Wildtype (84%) and Missing (12%) for the rGBM DCVaxL treatment arm and this corresponds to 2 IDH-1 Mutated patients (3% x 64 = 1.92)




- Normally the percentage of IDH-1 mutated patients in the previous GBM definition (the one used for this trial) is between 7 and 10%, so in the whole trial there would be expected between 23 (7% x 331) and 33 (10% x 331) IDH-1 mutated patients.

As calculated above, there are 7 IDH-1 mutated patients in the nGBM treatment arm and 2 in the rGBM treatment arm.

So in the group of 35 non crossovers we could expect between 24 (33 - 7 -2) and 14 (23 - 7 - 2) IDH - mutated.

A fair question would be why could be such a difference in distribution of the IDH-1 mutated patients between the arms? I think the answer could be that when the patients were randomized, IDH-1 was not known as a confounding factor, so the patients were not statistically stratified by that variable and end up in different proportions in the original arms. So the majority of patients could have fallen at the beginning in the 99 original placebo group.

And then, after that, the explanation is easier. It is absolutely expected that from the 99 original placebo group, the ones with best prognostic factors (IDH-1 mutated) didn't progress as the wildtype ones, so they mainly stayed in the 35 non cross over group.

Anyway, such a different proportion between the two initial arms of the trial is not very probable, so it could have been the case that between 7 and 10 patients were IDH-1 mutated in the original placebo group (7 to 10%), so at the end there would be between 5 (7 - 2) and 8 (10 - 2) in the 35 non crossover group.

So it is probable that between 5 and 24 patients ended in the 35 non crossover group.

3. I also think there is a possibility that some of those 35 patients received DCVaxL that actually helped them, but were not real progressors (could have been pseudoprogressors from SOC initially thought to be progressors) or didn't classify to the rGBM arm for any other reason.

So I'm not worried at all if the 35 patients lived longer than any other group. It is actually positive. That's the counter intuitive part.


* Sorry if some parts of my post are not clear, English is not my native language.