Friday, 27 January 2017

Hypermutation Risks of Temodar

Hi all,
I've found great value in this blog and am posting for the first time.

I was diagnosed in march 2016 with grade 2 oligo.  Its a somewhat larger tumor, about 7x6x5cm and unlike most oligo's is not in the frontal lobe, but is left parietal and overlapping motor, sensory and speech areas and nearing midline crossover.  Because of all that it was considered a high risk surgery zone and so I have so far only had a biopsy and have been doing chemo.   I'm IDH1 mt, mgmt meth, 1p19q codel, ATRX normal, p53 normal.

So the broad plan was chemo then RT.  The hope was to shrink the tumor to be able to reduce the RT zone. So far 6 rounds of Temodar but I have now switched over and done 1 round of CCNU after having learned more about hypermutation risks with TMZ.

I've dug thru what published data there is and its all clear as mud.  On the one hand the 2014 Science paper (Johnson et al from the Costello UCSF lab) that seems to have really launched this topic found very clear and scary linkages showing hypermutation and upgrading at recurrance to GBM in about half of the patients treated with TMZ vs none with only RT, on the other hand it was a very small data set, had mixed genetics, was clearly a selected dataset and not randomized, and only looked at TMZ alone or RT alone, not the more common RT+TMZ.  The few other studies I have found add some implications that mutations in TP53 and mismatch repair genes  increase the risk, though some data also shows TMZ driving mutations of TP53 and MSHx which then creates the path to hypermutation.

In simplified terms if chemo mutates the cancer to a point where the 'self check' mechanisms during cell division stop working then the next time the cell gets hit with chemo it will go ahead and replicate in spite of the genetic damage from chemo and create new set of mutations.

To further complicate the picture most of the historical data looking at different chemo options is comparing PCV to TMZ, but procarbazine is a monoalkylating agent that is quite similar to TMZ.  So I've gotten some input from Doc's that CCNU alone may be less mutagenic because it is a bi-alkylating agent and will more effectively stop DNA replication thru double strand breaks.

So Steven or others who are into the science side of cancer, do you have any thoughts on how to unravel all this?  Recommendations on other approaches to consider, or ways to enhance effectiveness of chemo for low grade cases?


SW edited this post to include the image below, from a presentation at the 2016 SNO conference in Phoenix Arizona:


  1. Mutations in mismatch repair genes (like MSH6) do more than increase the risk of hypermutation - they are the root cause of hypermutation. I've also seen the hypothesis that p53 mutant tumors are more susceptible to hypermutation, but I don't believe that's the case. In the expanded dataset (yet to be published) from the Costello lab, 1p/19q codeleted oligodendrogliomas had an equally high rate of hypermutated recurrence compared to p53 mutated astrocytomas. All the patients in these studies had the IDH1 mutation. I believe the risk of hypermutation has more to do with MGMT methylation status, but the root cause is defective DNA mismatch repair in at least one cell, or a small subclone of the tumor, in which case TMZ becomes a potent mutagen for those cells. TMZ may also be reponsible for the initiating mutation in one of the mismatch repair genes which sets the stage for hypermutation following further TMZ treatment.

    A 2009 study by Stephen Yip et al. documented 4 cases of hypermutated GBM, all having an MSH6 mutation in the recurrent tumor. Prior treatment in these four cases was TMZ in one case, CCNU in one case, and PCV in 2 cases. Therefore CCNU is not risk-free in terms of hypermutation. What we don't know is how great the risk is with each of these treatments. It has been most well studied by the Costello group in the context of TMZ monotherapy for lower grade IDH1-mutant gliomas. We can't really estimate the rate of hypermutated recurrence following PCV or CCNU, this hasn't been studied systematically.

    There was some compelling data from the randomized NOA-04 trial in Germany that PCV was superior to TMZ for your tumor type (IDHmut, 1p/19q codeleted oligodendroglioma).

    On this basis I'd probably pick PC over TMZ for IDHmut oligodendroglioma, but I would probably try to drop the vincristine.

  2. Thanks Stephen!

    Wow that is good info, and the NOA-04 is super helpful. I had also found another paper indicating possible benefit of PCV over TMZ (lassman 2011 neuro-oncology), but it is a retrospective study that has some definite study group biasing in it.

    I certainly agree that vincristine seems to just add risk and not add benefit and am feeling good that procarb&CCNU or ccnu alone is possibly more effective than TMZ. I had gotten input from a leading Doc who specialized in pharmacology that CCNU is a bi-alkylate that cross links the dna strands, vs procarb and tmz are mono and thus are more mutagenic. So he recommended CCNU alone, but he is doing more of a leadership role these days vs studies so not sure how current that info is in regards to hypermutation.

    Do you have a any sense of what ratio of MMR mutation is likely pre-existing vs actually driven by chemo? Part of the treatment plan variations seem to tie in with how long chemo should be done. I've gotten very conflicting messages. On the one hand I have been told that limiting to 6-8 cycles will significantly reduce the risk of hypermutation, on the other hand have been told that in low grade the division rate is low enough that its necessary to do 12-24 cycles of tmz to really get significant benefit (because too few cells will be undergoing division during each round). If the MMR mutation is pre-existing it seems like it may not make much of a difference, vs if its driven by the chemo than it seems like the # of rounds would significantly raise risk.

    From the prelim data from the Costello lab do you have a sense of what percentage of people will have hypermutation? From the published data it was pretty scary to see 5 of 8 tmz cases got hypermutation and upgraded to GBM, vs more like 2 out of 15 RT cases had significant mutation. Are you thinking that MGMT meth poses high risks and so maybe chemo should be limited or skipped?


  3. Bryan great questions. I'll start by saying that I edited your post to include an image of a slide presented at the SNO 2016 conference during a talk by Joe Costello. As you can see in the slide, 12/19 (63%) of IDHmut astrocytomas treated with TMZ alone had hypermutated recurrences, and 10/16 (62.5%) of IDHmut oligodendrogliomas.

    1. At one point I was told the hypermutation rate for IDHmut astrocytomas treated with TMZ alone was 11/21 (52%) which differs only slightly from these figures.

    2. As for your question about mismatch repair mutations pre-existing prior to TMZ treatment, versus being induced by TMZ treatment, that is still an open question. This was also discussed by Joe Costello at the SNO conference. He presented a case of one patient whose tumor did not have a mismatch repair mutation at first diagnosis. Later (I think it was after treatment), an MSH6 mutation was found, but the tumor was still not hypermutated. The first and second recurrences were hypermutated. The slide was titled "An early stage of hypermutation" and may be a case in which the MSH6 mutation was caused by TMZ treatment. But the jury is still out on this question.

      Tumor cells with defective mismatch repair are resistant to TMZ, because mismatch repair mechanisms are part of how TMZ is thought to be cytotoxic to cells. TMZ inserts methyl groups onto DNA, which leads to mismatches in the DNA code. Mismatch repair mechanisms then kick in to correct the mismatch, but if the methyl group is not removed (by MGMT) then continuous futile cycles of mismatch repair ensue, eventually leading to irreparable DNA damage and cell death. So mismatch repair defects through mutation in mismatch repair genes can be thought of as a form of TMZ resistance, with hypermutation being a byproduct (without functional mismatch repair, the DNA mismatches caused by TMZ neither get repaired nor cause the cells to die, leading to a rapid accumulation of new mutations).

      Tumor cells with active MGMT (or an unmethylated MGMT gene promoter) tend to be resistant to TMZ to begin with, so there would be far less evolutionary selection pressure to acquire new resistance mechanisms (such as mismatch repair mutations).

      The catch-22 is that TMZ is most effective for patients with methylated MGMT promoter ("methylated MGMT"), but in theory such tumors would also be at higher risk for the development of other resistance mechanisms such as mismatch repair gene mutations. The link between MGMT methylation status and risk of hypermutation is only theory at this point, and hasn't been demonstrated by evidence.

      One upside of hypermutation is that it makes a tumor more immunogenic and probably more likely to respond to immunotherapies. This is documented for PD-1 inhibitors in other cancers (they are more likely to work in hypermutated tumors with mismatch repair defects).

    3. Sorry Stephen, I mean to respond back to this and got sidetracked (after 1 round of CCNU&Procarb I dipped into grade3 neutropenia. argh).

      So better late then never, thank you SO MUCH for snagging that picture from the SNO conference. Its some pretty useful info and will be very interesting to see what the next paper is that comes out on this.

      While seeing it was basically 50/50 to have hypermutation even with Oligo, it was at least reassuring that it is less likely to promote to GBM.

      Its still certainly unclear if TMZ is more mutagenic than CCNU or not and will probably be a while before that gets worked out, if ever.

      From a study standpoint though it seems like different geographic regions have different approaches to SOC (like US vs German vs Brazil) so I wonder if there are some clean population data sets that could be looked at retrospectively to see what and when hypermutation occurs ( chemo then RT, vs RT then chemo, vs chemo only till progression, vs RT only till progression).

      Also a good comment on the possibility that hypermutation could aid in immunotherapies, though seems like its countered a bit with the likelihood that hypermutation would reduce the effectiveness of targetted pathway inhibitors (like rapamycin analogs/MTor).

    4. It's okay Bryan, I never expect quick replies. I know how much most people on here are going through.

      We can expect another paper on this sometime in the not-too-distant future.

      Oligos are much less likely to progress to grade 4, with or without hypermutation. But yes it was good to see two hypermutated oligo cases that didn't progress to grade 3/4.

      We might not find out how risky hypermutation is following CCNU, since almost all glioma patients get treated with TMZ first. It would be possible in the lab though to knock out a mismatch repair gene in a glioma cell line, apply CCNU, and see if hypermutation results. It would be another question how likely CCNU is to initiate a mismatch repair mutation in the first place, the way TMZ most likely can.

      I'm with you, we could learn a lot about hypermutation by sequencing tumors retrospectively and then looking at what treatments were received. The main barrier is that most centers aren't doing the extensive genetic sequencing they would need to do to determine if a tumor is hypermutated or not.

      I'm not certain hypermutation would reduce the efficacy of mTOR inhibitors. One of the common pathways that picks up mutations in hypermutated tumors is the AKT/mTOR pathway (one example of which could be an activating mutation in the mTOR gene itself, or mutations in PTEN, PIK3CA, AKT), so you could argue that mTOR inhibitors are more likely to have an effect in those cases, but of course it's better not to have such a mutation in the first place. It's likely these new "tertiary" mutations in pathways like Akt/mTOR or CDKN2A/CDK/RB1 that cause the tumor to progress from grade 2 to 4.

      It's better not to have a hypermutated tumor (which usually occurs as a grade 4), but if that's what you're dealing with, PD-1 or PD-L1 antibodies and other immunotherapy is probably a very good investment.

  4. In one recent study of IDH mutant, 1p/19q codeleted oligodendrogliomas, there was no case of hypermutation following temozolomide (3 cases), procarbazine+ACNU+vincristine (7 cases) or ACNU+vincristine (1 case).

    Although based on previous studies, I suspect hypermutation would have been seen if more cases had been treated with TMZ, zero out of 8 cases of hypermutation following ACNU (with or without procarbazine) leads me to believe hypermutation is much less likely to occur with PAV chemotherapy (and by extension PCV or other nitrosourea based regimens) compared to TMZ.

  5. I don't want to open a new topic...

    Repurposing Mebendazole as a Replacement for Vincristine for the Treatment of Brain Tumors.

    But at the moment, the article is not yet shown on PubMed (only abstract) - I guess it is still to be uploaded.
    "We also compared the therapeutic efficacy of mebendazole and vincristine against GL261 orthotopic tumors. We found that mebendazole showed a significant increase in animal survival time, whereas vincristine, even at a dose close to its maximum tolerated dose, failed to show any efficacy. In conclusion, our results strongly support the clinical use of mebendazole as a replacement for vincristine for the treatment of brain tumors."

    1. Thanks Matjaz, you can download the full study from this page:

  6. What would be the recommonded dose of Mebendazole for glioma in humans?Thanks..