Given the list of all-star pathologists and neuro-oncologists behind this paper (including Andreas von Deimling, David Louis, Wolfgang Wick, Michael Weller, Roger Stupp, Monika Hegi, Martin van den Bent, Michael Platten, and many more), this work will very likely get translated into actual changes of the grading system for IDH-mutant astrocytomas.
"Thus a genetic approach to taxonomy would favor a term such as “high-grade A IDHmut” rather than the current term of GBM-IDHmut." "One possible solution, which we have considered implementing in Heidelberg, would be to restrict classification to the term A[stroctyoma] IDHmut and then adapt grading according to molecular lesions, thereby omitting the term GBM-IDHmut. The term “glioblastoma” would be reserved for those histologically defined glioblastomas lacking IDH mutation or not having had adequate (not otherwise specified, NOS) or diagnostic (not elsewhere classified, NEC) work-ups"
Nice to finally see the above quotes in print, this has been my opinion for years.
This is an excellent study. My only criticism is that hypermutated tumors with very high mutational load were not evaluated as a distinct group. This is different than the high or low copy number variant load (CNVL) considered in the paper. Mutational load was only mentioned once, in the discussion:
"An interesting parameter emerging from our analysis was CNVL [copy number variant load]. A proportion higher than 350 Mb either lost or gained in the areas covered by the methylation arrays correlated with poorer OS. Previously, the mutational load has been shown to correlate with tumor grade in IDH-mutant glioma [9]. While our data more reflect genomic instability and those data rely on an accumulation of mutations they also support a quantitative approach to tumor grading."
Uploaded to the Brain Tumor Library, Pathology folder, IDH-mutant glioma subfolder.
ReplyDeleteOne of the findings is that it appears to be better to have a grade IV IDH-mutant astrocytoma without homozygous deletion of CDKN2A/B than a grade III IDH-mutant astrocytoma with homozygous deletion of CDKN2A/B. In other words CDKN2A/B status is more important than the more traditional histopathological determiners of tumor grade.
ReplyDeleteIn the simplest most widely applicable version of the new tumor grading system for IDH-mutant astrocytomas, the new grading was performed as follows:
ReplyDeleteIDH-mutant astrocytomas with homozygous CDKN2A/B deletion: astrocytoma, IDH-mutant, grade 4
Tumors with necrosis, but no homozygous CDKN2A/B deletion: astroctyoma, IDH-mutant, grade 3
Tumors with no necrosis and no homozygous CDKN2A/B deletion: astrocytoma, IDH-mutant, grade 2
That is an extremely interesting study, thanks very much for posting it Stephen!
ReplyDeleteRegarding the CDKN2A/B status that they have determined as a strong prognostic factor, is this different to CDKN2A ? In my partner's Caris report, they have CDKN2A under the heading "GENES TESTED WITH NO AMPLIFICATION DETECTED" - it does not mention anything about deletion, could this mean the same thing?
Also, it is very encouraging how it shows in Fig 1(b) a median OS for A2(IDHmut) of 19 years!
CDKN2A and CDKN2B are two different genes, but they are located adjacent to each other on chromosome 9p, so they are often both deleted together.
DeleteNormally there are two copies of every gene per cell (one copy from the mother, one from the father). Gene amplification is when there are more than two copies (sometimes many copies) of a gene in a cell. Deletion is when one or both copies of a gene are deleted. Homozygous deletion is when both copies of a gene are deleted.
It doesn't appear that Caris reports include gene deletions. They include instances of amplification equal to or >8 copies. FoundationOne reports include copy number alterations when copy number is either >8 or 0, in other words, gene amplifications, or homozygous deletions.
That survival curve is encouraging for A2-IDHmut, however the data isn't very mature considering how few patients were actually followed up for that long (19 years). The size of that last "stair step" is very large, showing very few patients were actually followed up that long (most of the surviving patients were "censored" before that time due to limited follow up, and the little nicks on the survival curves show the time of last follow-up for the surviving patients). This isn't surprising considering the IDH1 mutation was only discovered less than 10 years ago. However I wouldn't be surprised if the mature data eventually shows a median of 15 years or more for A2, IDHmut. It's a moving target though as new therapies come online, improving survival even further.
Thanks for the detailed explanation Stephen. I wish we had opted for the FoundationOne report instead of Caris!
ReplyDeleteI evaluated all the major tumor sequencing services on behalf of my friend, and we decided on Caris, and I was very happy with the quality of the report we got back. The Caris Next Generation sequencing was more cost effective than Foundation (we're Canadian, so paying out-of-pocket either way). Also, Caris reported the mutant allele frequency for every mutation, which Foundation doesn't, and that information is helpful. Caris also included immunohistochemistry testing for the four main mismatch repair genes (MSH2, MSH6, MLH1, PMS2) which isn't included in Foundation reports, and this was also valuable to us. I didn't notice at the time that Caris didn't include information on homozygous deletions. If detecting CDKN2A/B deletion were the top priority, Foundation may have been the better choice. I'm not sure I would choose differently if I had to choose again.
DeleteSo, I had my tumor material tested for the CDKN2A/B deletions (MLPA analysis) and unfortunately I don't think it came back with 'good' news, i.e. there are definitely deletions there across the board. I'm trying to figure out how these link back to the IDH1 classification study, especially if these should be interpreted as a 'homogyous codeletion'. I've tried to get to grips with the definitions, but I'm struggling a bit - e.g. difference between heterogeous and homogyous deletions. My NO says my results should probably be interpreted as 'homogyous' deletion...
ReplyDeleteThe methodology used in my test seems to have been a bit different from the study. I get the following ratios for the various Genes:
CDKN2A-3 0.47
CDKN2A-2 0.38
CDKN2A-1 0.37
CDKN2B-2 0.36
CDKN2A-1 0.65
I was wondering if you could help me shed some light on this. Firstly, to confirm this is indeed a case of a 'homogynous codeletion', and secondly, if so, I was curious if there are any medical implications you are aware of beyond being a possible prognostic factor alone... i.e. are there any obvious therapy implications? I heard in some other cancers there may be some meds being tested for this purpose...
Would be very grateful for any guidance. Thanks!
John
Hi John,
DeleteThe terms are "heterozygous" and "homozygous".
Everyone has two copies of the 23 chromosomes: everyone inherits one set of the chromosomes from the mother, and another set from the father, making a total of 46 (23 + 23). Because you have two sets of chromosomes, you also (normally) have two copies of every gene. A heterozygous or hemizygous gene deletion means one copy is deleted. A homozygous deletion means both copies are deleted.
I'm not sure what the numbers after CDKN2A refer to. Perhaps if you email me the report I'll be able to help shed some light on it.
A homozygous deletion of CDKN2A and CDKN2B would explain why your tumor would be behaving as a grade IV rather than a lower grade. These genes are inhibitors of cyclin-dependent kinases 4 and 6, which negatively regulate RB1. RB1 is a tumor suppressor that prevents the cell from moving to the DNA synthesis stage of the cell cycle. So, when CDKN2A/B are deleted, CDK4/6 are overactive, inhibiting RB1 and allowing cells to freely move through the cell cycle, increasing proliferation.
CDK inhibitors such as palbociclib, ribociclib and abemaciclib are approved for breast cancer. Abemaciclib is even shown to be brain penetrant in mouse models, and has been in GBM trials. However, in a trial that included 17 GBM patients, the two patients with the best response to abemaciclib did not have CDKN2A/B loss, many patients with CDKN2A/B loss did not respond to abemaciclib, and there seemed to be a connection between TP53 mutation and response to abemaciclib.
See my comments here:
http://btcocktails.blogspot.com/2017/11/surgery-avastin-or-study-drug.html#comment-form
and the study here:
http://cancerdiscovery.aacrjournals.org/content/early/2016/05/18/2159-8290.CD-16-0095.full-text.pdf
In my explanation of people having two copies of each chromosome I should have said "most people" rather than "everyone". Some people have extra copies of chromosomes. For example, Down Syndrome is caused by trisomy 21, or three copies of chromosome 21 rather than two.
DeleteThanks a lot for your comments and explanations, Stephen. I have emailed you with the full data. Would be very grateful if you can make some sense of it in the context of the IDH1 classification study.
DeleteBest,
John
Hi John,
DeleteThe interpretation of the numbers can be found in a short video here:
https://support.mlpa.com/kb/articles/video-how-does-mlpa-work
See especially minute 4:18 - 5:57
The test sample (tumor sample) is normalized to the average of a set of reference samples. The ratio is test sample: reference sample. A ratio of 1:1 means the test sample has the same number of copes (usually 2) as the reference samples. A ratio of 0.5 means heterozygous deletion.
In your case the ratios for CDKN2A (exons 1 and 2) and CDKN2B (exon 2) were around 0.35 - 0.38.
If I had to attempt an interpretation of these numbers, I would say that there is at least a subclone of the tumor with homozygous deletion of CDKN2A/B, because of the ratios being significantly lower than 1 (no deletions) or 0.5 (heterozygous deletion).
There are clearly some cells without homozygous deletion (because the ratio is higher than 0), and this may include non-tumor cells (infiltrating macrophages can make up a large proportion of a GBM tumor, or tumor subclones without CDKN2A/B deletions, or with only heterozygous deletion. But there is at least a subpopulation in the tumor with homozygous deletion at CDKN2A/B.
That would be my best guess.
Hi Stephen,
DeleteThanks a lot for your explanation, which is extremely useful. I guess my confusion was how you tell a homozygous deletion from a heterozygous deletion, which isnt addressed in the video, but I presume then it's just a value significantly below 0.5 then.
However, 2 exons (CDKN2 A exon 3, and CDKN2 B exon 1) do not seem to fall into this category then. I suppose either way, the takewaway is that I have significant deletions in this area as a whole, anyway.
Again, many thanks for taking the time to look at my results and explain.
Hope you have a nice weekend
John
For the sake of hypothesis, let's assume that the sample was completely composed of tumor cells (no non tumor cells), and that all the cells had a heterozygous deletion of CDKN2A/B. In that case the ratio would be no lower than 0.5. The fact that there were ratios below 0.5 shows that at least some of the cells have a homozygous deletion.
DeleteIt is rather messy, with some exons having different ratios. But yes, the takeaway is that there are significant deletions in this region, and the cells with the most significant deletions (homozygous deletion in at least some exons of both CDKN2A and CDKN2B) probably have the proliferative growth advantage over the cells with less deletion and intact copies of these genes.
It's not the best news, but it does explain why the tumor would be a grade 4 rather than a lower grade.
Hope you have a nice weekend too in spite of all this.
Thanks a lot, Stephen. As you mentioned in another post somewhere, it's probably more important to try to make it such that the theoretical curves becomes less meaningful, than to dwell too long on prognostic markers... Still, I think it's good to know.
DeleteBest regards,
John
Another study from a few years back supporting the new approach to grading and potentally treatment approach...
ReplyDeletehttps://academic.oup.com/jnen/article/74/5/442/2614357
IDH1-R132H acts as a tumor suppressor in glioma via epigenetic up-regulation of the DNA damage response
ReplyDeletehttp://stm.sciencemag.org/content/11/479/eaaq1427
"Mutations in isocitrate dehydrogenase 1 (IDH1) are frequently found in gliomas and are associated with better outcomes. Núñez et al. discovered that, in addition to its roles in metabolism and epigenetics, mutant IDH1 also helps maintain genomic stability in tumors by enhancing the DNA damage response. This finding helps explain why patients with IDH1-mutant tumors have better survival despite their tumors being less sensitive to radiation than other gliomas. The authors also examined the mechanism for this phenomenon in mouse models and demonstrated that pharmacological inhibition of the DNA damage response sensitizes IDH1-mutant tumors to radiation, suggesting a potential direction for further therapeutic advances."
Ops, maybe posted this in wrong topic...misread the title for new approaches to *treating* of IDH-mut astrocytomas. Sorry
Delete