The study is called
Clonal expansion and epigenetic reprogramming following deletion or amplification of mutant IDH1. View abstract
here. I've also uploaded this to the Brain Tumor Library,
Pathology folder,
IDH-mutant glioma subfolder. This data was presented at the SNO conference in Phoenix last year. I'll be following up with comments, and perhaps an Astrocytoma Options update on this.
Didn't yet understand, why is it so important?
ReplyDeleteThis study is quite important because it shows that in IDH1-mutant gliomas, when the mutant copy of IDH1 is deleted, those cells seem to gain a growth advantage. The same thing is true with gain/amplification of the mutant IDH1 copy.
DeleteAs it says in the discussion "the unifying feature is a loss of the heterozygous mutant state". What this means is that in newly diagnosed IDH1-mutant gliomas, there is one mutant copy of IDH1 and one non-mutant (wild-type) copy. This is called a heterozygous mutant state (one mutant and one non-mutant copy). This balance between mutant and non-mutant IDH1 is necessary for maximum 2-hydroxyglutarate production. When there is imbalance between mutant and non-mutant IDH1, then 2-HG production decreases. This is the case when the mutant or wild-type copy of IDH1 is deleted, or it could be the case when the mutant IDH1 copy is amplified.
In this study, out of 50 IDH1-mutant low grade gliomas, only 6 had copy number alterations in IDH1 in the tumor recurrences. Only three of these were clonal (found throughout the recurrent tumor, as opposed to subclonal which means found in only some part of the recurrent tumor). So this is a rare occurence, but the important part is that these tumors with copy number alterations for IDH1 (deletions, gains or amplification) actually have a growth advantage.
Firstly, when initial tumors had no copy number alterations for IDH1, but the recurrent tumor had deletion of IDH1 found all throughout the tumor, this implies that the clone with the loss of mutant IDH1 had a growth advantage over the cells with the normal heterozygous mutant state. The authors speculate that "TMZ treatment may contribute to
outgrowth of the CNA [copy number altered] cells, perhaps by preferentially killing the tumor cells that retain heterozygous IDH1 mutations".
Secondly, when they examined Ki-67 in the tumors, a measurement of cell proliferation, the areas of the tumor with altered IDH1 levels (indicating copy number alterations) had increased Ki-67 positivity.
Thirdly, cells tend to lose either the mutant or wild-type copy of IDH1 when cultured in vitro, suggesting that these cells have a growth advantage in cell culture conditions.
Fourthly, the tumors with IDH1 copy number alterations (including deletion of mutant IDH1) recurred as higher grade tumors.
Fifthly, the three cases with clonal copy number alterations (two cases with deletion of mutant IDH1, and one case with copy number gain of mutant IDH1) retained G-CIMP (CpG Island hypermethylation), they were classed as G-CIMP-low, rather than G-CIMP-high. G-CIMP low is associated with worse survival outcomes. There was also massive hypomethlation outside of CpG islands in these cases.
The conclusion of all this data is that deletion or copy number gain of mutant IDH1 leads to decreased 2-hydroxyglutarate production and major changes in DNA methylation outside of CpG islands, and these tumors gain a growth advantage. If mutant IDH1 is not required for tumor progression and deletion of mutant IDH1 gives a growth advantage: "A central question is whether mutant IDH is a reasonable target for therapy."
Small molecule IDH inhibitors have been shown to work quite well in IDH-mutant leukemia, leading to FDA approval of enasidenib, (AG-221, mutant IDH2 inhibitor), but the authors of this study explain the differences between IDH mutant leukemia and glioma, and why what is true for IDH mutant leukemia might not be true for IDH mutant glioma.
DeleteThank you Stephen, as usual, for providing such a clear and readable summary of that study. The findings are very disappointing (although foretold is forewarned and it is good to know) as my partner has applied to the Bayer pan-mutant IDH1 trial and looks like she may actually get accepted.
ReplyDeleteShe has just undergone a de-bulking surgery last week and is due to go back in for a second surgery this coming week - the surgery was complex so they split it in 2, the first part asleep, the second awake. We have not yet had any pathology back from this latest surgery, however results from her last surgery 2 years ago showed she has a diffuse Astrocytoma – IDH1 mutated, TP53 mutated only (no ATRX mutation) and has a low mutational burden of 4 Muts/Mb - so the Bayer trial is one of the most exciting for us.
As you mention, the study deals with tumours that have acquired a Copy Number Alteration (most likely to TMZ treatment) and subsequent dramatic reduction of 2HG levels and these cells had a growth advantage over their heterozygous mutant counterparts.
I just wanted to ask if you believe that if a tumour does not have CNA’s (detected with epigenetic testing, FoundationOne etc) it would be worthwhile still taking part in the trial – given the significant OS improvement that Bayer showed in their in vitro and in vivo study? Or would it be too risky given the fact that there may be undetected cells with CAN’s that would then thrive in the low 2HG environment created by the IDH inhibitor and therefore clonally spread/expand to occupy a larger portion of the tumour and in doing so, make it more aggressive?
Also, do you have any thoughts as to why Bayer would not have detected such a dramatic clonal increase in CNA cells during their extensive in vitro/vivo study? is it perhaps that they used engineered cell lines that were uniform and had no chance of containing any CNA cells?
Thanks,
Ryan.
Theoretically, the mutant IDH1 inhibitor drugs should have no effect on cells with deletion of mutant IDH1. These drugs would only be active in cells where the mutant IDH1 enzyme was active (and the activity of the mutant IDH1 enzyme is to produce 2-hydroxyglutarate).
ReplyDeleteWhat is more worrisome is that use of the IDH1 inhibitor drugs could mimic the effect of these IDH1 copy number alterations, by shutting down the activity of mutant IDH1 and 2-hydroxygluratate production.
However, given the fact that some IDHmut glioma patients, especially low grade gliomas, do seem to have had positive response to the mut-IDH1 inhibitor drugs, with disease stabilization and some objective responses, the effects of inhibiting mutant IDH1 activity pharmacologically doesn't seem to be identical to the effects of having deletions or other copy number alterations of IDH1. There are still a lot of unknowns though, given that it's still early days in the investigation of these drugs.
One question that the study brought up is how chemotherapy relates to all of this. Did the clones with deletion of mutant IDH1 take over because they were less sensitive to TMZ treatment than the clones with heterozygous IDH1 mutation?
Also what is the specific mechanism that makes the cells with the copy number alterations more proliferative or more resistant? Is it the changes in DNA methylation patterns that this study described? If so then there is also a time factor, because it may take some time for these methylation changes to come into effect. As the study mentioned, it will be necessary to track tumors being treated in these mutant IDH1 inhibitor trials very closely, not only to look for the outgrowth of subclones with copy number alterations in IDH1, but also to look for the changes in DNA methylation patterns described in this study (transition from G-CIMP high to G-CIMP low, and hypomethylation/demethylation outside of CpG islands). A lot of our questions won't be answered until we start getting this kind of data back from the ongoing clinical trials.
To answer your question about the Bayer study, they would not detect copy number alterations unless they were specifically looking for them, but that's not what the study was designed for.
Stephen maybe I am reading too much into your tone, but this seems like an overly pessimistic view of IDH1 inhibitors from just this one study. Some thoughts on it:
Delete1) The study had about 10-15% of patients exhibit some kind of copy number alteration and within that had a mixture of clonal and subclonal. So that seems to indicate that not only is it not that common, it may only produce a moderate growth advantage.
2) The study notes TMZ may cause copy number alterations, but also subclonal CNA's have been found in patients at first diagnosis. So it is not exclusively therapy driven, and again it seems to indicate that the growth advantage is not that extreme because otherwise it would be more common to see patients with tumors where the primary colony has a CNA at initial diagnosis.
3) As noted It doesn't seem like the study clearly showed that the drop in 2-hg levels caused the change in methylation patterns. Additionally, studies of IDH inhibitors have shown that contrary to expectation the reduction in 2-hg levels do not seem to change methylation patterns, and that there are even suggestions to investigation demethylation drugs to combine with IDH inhibitors (though that seems questionable in light of this study).
4) Some of the data on IDH inhibitors is indicating that the reduced 2-hg allows cellular differentiation and also restores apoptosis. Plus the cellular microenvironment improves which may allow for more of an immune response. That should thus reduce risks of acquiring a CNA.
5) In patient studies of AG-120 look fairly promising, but certainly with only a few years of patient data and almost no detailed analysis of inhibitor treated tumors after therapeutic failure, this new study indicates some risks. It doesn't seem to provide that much of a linkage though showing that taking an IDH inhibitor will drive tumorigenisis, but it certainly shows a risk factor there.
Overall it seems like it should dampen the enthusiasm around IDH inhibitors a bit as far as them being long term cures across the board, but if they can change OS significantly in many cases then that is still pretty promising. It seems fairly straightforward that recurrent patients will see benefit, though this study does add some concerns about treating patients with IDH inhibitors earlier on.
Maybe a good comparison would be to put this study in the same framing as Costello lab's studies in regards to TMZ driven hypermutation. Basically that for IDHmut patients TMZ has clearly shown a benefit to both PFS and OS, but Costello's genetics showed tmz drives hypermutation and upgrade in maybe half of patients. So CNA's may be the way the cancer evades IDH inhibitors, but if PFS/OS is improved it still may be the best option.
I didn't mean to sound pessimistic, but the study under discussion does raise some questions about the proper role of the IDH inhibitors.
DeleteMost of the research I've come across seems to point to an essential role of mutant IDH1 in tumorigenesis, but as tumors progress and gain further mutations, they become less dependent on mutant IDH1 as a driver and can actually grow and proliferate quite well as secondary GBMs that have lost the IDH1 mutation (although this happens in only a minority of cases).
I certainly think the inhibitors will play a role in IDH1-mutant glioma therapy, but it still isn't that clear what that role will be. I think the clinical data we've seen so far (for AG-120) shows a higher chance of benefit in lower grade gliomas versus high grade.
I think it's important to recognize that some tumors may still depend on mutant IDH1 and its downstream effects and would be responsive to the inhibitors, while more advanced tumors may have evolved to a point where they are no longer dependent on mutant IDH1 or its effects and could even gain a growth advantage in its absence. My feeling (which is backed up by the little bit of clinical data we've seen) is that the inhibitors will be useful for some, but not for others.
It's still early days in the clinical investigation of these drugs and there's far more questions than answers at this point.
As you touched on, I think an ideal use of the mut-IDH inhibitors would be in combination with immunotherapies, as studies have shown the immunosuppressive effects created by the accumulation of 2-HG in and around these tumors.
To go back to your second point, there is no mechanism whereby TMZ could cause copy number alterations. Instead it could provide a selective advantage to those cells that have randomly undergone such alterations, and if so we could expect this subclone to outcompete its neighbours and eventually become the dominant clone over time.
The changes in DNA methylation patterns as IDH1-mutant tumors evolve is quite interesting. These tumors typically always display the G-CIMP hypermethylation phenotype at diagnosis. But it seems like as the tumors progress and evolve, sometimes they begin to lose methylation at certain locations. Again the Costello lab at UCSF is responsible for most our knowledge of this phenomenon.
https://www.ncbi.nlm.nih.gov/pubmed/26373278
"Genes transcriptionally upregulated through promoter hypomethylation during malignant progression to high-grade glioblastoma were enriched in cell cycle function, evolving in parallel with genetic alterations that deregulate the G1/S cell cycle checkpoint."
This is another example of how low grade IDH-mutant tumors change and evolve over time as they progress to higher grades, with the implication that sensitivity to various therapies probably also changes and evolves over time, as genetic or epigenetic alterations that were critical to the genesis of these tumors, become less important as tumors evolve and gain new genetic and epigenetic drivers (often picking up alterations more characteristic of IDH wild-type GBM, such as CDKN2A/B deletions).
To be clear, my position is not only that IDH-mutant gliomas will require different therpeutic strategies versus non-mutant, but also that low grade IDH-mut gliomas will probably need different strategies from higher grade tumors, that are further along in their evolution. This may apply to strategies such as the mutant IDH1 inhibitors we've been discussing, and may also apply to strategies involving demethylating agents such as 5-azacytidine or decitabine. Again, it is early days in this research and there is a great deal still to be researched with more unknowns than knowns.
I appreciate your critical questioning, this is the kind of inquisition necessary to drive further research and discovery.
I was just reading something about IDH1 inhibitors and I'm a bit confused...IDH1 inhibitor is not the same as IDH1 peptide vaccine right? Inhibitor just inhibits 2-HG production, which was shown is not enough once the glioma cells already present and IDH1 peptide vaccine helps immune system target those cells?
ReplyDeleteThanks
Correct, they are two completely different things. Small molecule drugs like ivosidenib (AG-120) inhibit the mutant IDH1 enzyme, and thus 2-HG production.
DeleteThe peptide vaccines are completely different and are composed of a sequence of amino acids that include the IDH1 R132H mutation where an arginine has been replaced with a histidine amino acid. This should prime the immune system to recognize this mutation. The peptide vaccines do not directly interact with mutant IDH1 glioma cells, but instead interact with the immune system.
Just doing some reading on Pubmed...found this e-pub ahead of print.
ReplyDeleteIDH1 mutation diminishes aggressive phenotype in glioma stem cells.
https://www.ncbi.nlm.nih.gov/pubmed/29115585
"The results showed that the positive rate of GSCs in patients with IDH1-R132H was significantly less than that in patients with IDH1-wt. The positive rate of GSCs was correlated with IDH1 mutation, TNM stage and poor overall survive. After transfection in vitro, IDH1-R132H overexpression led to reduced GSCs proliferation, migration and invasion, inducing apoptosis and improving GSC differentiation, accompanied by a significant reduction in activity of β-catenin. Several mediators, effectors and targets of the Wnt/β-catenin signaling were downregulated. The data demonstrate that IDH1 mutation reduces the malignant progression of glioma by causing a less aggressive phenotype of GSCs which are involved in the Wnt/β‑catenin signaling."