IDH Status.
All GBM's are either IDH mutated or IDH wildtype.
Subgrouping based on IDH status isn't viable in statistical terms.
Because:-
90% of GBM's are primary / de novo GBM's.
And 95% of primary GBM's are IDH wild-type, some report it at 99 or 100%.
Secondary GBM's (preceded by a lower grade glioma) make up about 10% of all GBM's.
And something like 40 or 50% of secondary GBM's are IDH mutated.
Thus in any broad GBM population, only about 5% will be IDH mutated.
Translate that to our trial and that gives you about 16 of the 331 being IDH mutated.
Perhaps 5 in the control arm, and 11 in the treatment arm.
Straight away, you can see that the numbers are so small, that arm by arm comparison based on IDH mutation is a statistical non-starter.
There is a distinct group of GBM patients that have a few common characteristics. These are secondary GBM, much younger age, Proneural gene expression, and often IDH mutation. And if you fit into that group, you have a considerably better prognosis, perhaps 50% longer OS expectation.
(Kat Deeley was likely one of these.)
The prognostic importance of this grouping rivals the prognostic importance of methylation status.
But as stated, the numbers in our trial with IDH mutation will be small; in the region of 5%.
So why does it matter at all, and why did LP flag it up?
Well, I actually don't know why LP flagged it up, seeing as it was somewhat downplayed in JTM.
But the reason it could be of a bit of importance, is that it is a worthwhile exercise to check that IDH mutated patients are in the same proportion in the control arm as the treatment arm.
It's almost implausible, but what if a large proportion of the IDH mutated patients by a quirk of fate ended up in the control arm?
It would skew your results somewhat.
So if you can check it, it's worthwhile to do so.
In practice, because of the age stratification that ensures median age is basically the same across the two arms, it's very unlikely that one arm would have significantly more of the handful of IDH mutated than they should.
Now you could say that because the trial population of 331 is a few years younger than 331 patients in the general GBM population, that our trial might just have 2 or 3 more patients with this prognostic advantage than you would normally expect.
But that really would be a very small factor.
Of course (as I've said previously), if you were running a single arm trial with just historical control, you could pull a fast one and stack your trial with loads of secondary GBM patients who will also likely be IDH mutated. But not really advisable to do that, because it's (a) unethical and (b) you might get rumbled...
I only know the above because I've read and distilled dozens of papers on the subject over the last couple of years.
Any one individual paper might give slightly different percentages to mine, but I've tried to give average figures based on quite a few papers.
AI
