So the genetic testing came back on my partner’s tumor (GBM, first recurrence), and while researching the mutations, I found Stephen W’s account of the hyperprogression study linking use of checkpoint inhibitors with accelerated progression in people with certain mutations, including amplification in MDM2 or MDM4. All 6 of the people in the cohort with MDM2/MDM4 amplifications experienced hyperprogression (across 5 different cancer types, none of which were primary CNS, although several had metastases in the brain; 5 were MDM2 amplified, 1 was MDM4).
My partner’s genetic report includes copy number gain in MDM4. No indication of the size of the gain; I’ve asked the company if they can clarify.
As it happens, we started pembrolizumab 2 weeks ago. I figured the worst that would happen was that it was ineffective; it never occurred to me that it might accelerate progression.
The study doesn’t prove that checkpoint inhibitors caused the hyperprogression. It’s possible that MDM2/MDM4 amplification caused the hyperprogression on its own. The possible mechanism seems understood: MDM2 suppresses p53, a major tumor suppressor (p53 is mutated in over 50% of all cancers). I found one study that showed MDM2 amplification is prognostic for poor survival in breast cancer. On other hand, the chart at the end of the hyperprogression study sure makes it look like acceleration began with the onset of checkpoint therapy.
My partner’s tumor showed high expression of PD-L1 (50%), and also has mutations in NF1 and TERT. All three of those are correlated with better response to PD-1 inhibitors in other cancers.
The next pembro infusion is scheduled for March 19. We have to decide by then whether to continue with it or not. Thankfully we’ll have an MRI that morning. The tough call will be if there’s no evidence of progression. I know there’s no clearcut rationale for making a decision either way, but I’m curious if anyone here has any advice.
Other biologically relevant mutations found are TERT, PTPN11, NFE2L2, PTEN, and CDKN2A. IDH1 is wildtype. MGMT is reported as unmethylated by sequencing. The tumor mutational burden increased from 2.05 to 5, and the MSI status is stable (i.e., no mismatch repair defect).
Our current regimen: CCNU, pembrolizumab, Vitamin D, melatonin, PSK, curcumin (also levetiracetam, Vimpat, Xarelto)
Planned additions: boswellia
Actively considering: ECGC, THC/CBD, Hydro+ALA, celexocib, metformin, everolimus
Thanks,
Brandon
What company did the sequencing? In the FoundationOne technical information it says copy number gains detected if there are at least 8 copies (normal copy number = 2), so low level copy number gains wouldn't necessarily show up in the report.
ReplyDeleteTempus did the sequencing. Haven't heard back from them yet.
DeleteUnfortunately we don't yet know whether the MDM amplifications play a causative role in the hyperprogression phenomenon following checkpoint inhibitor therapy, or if this is only an association without a causative role. The mechanisms haven't been worked out yet. About a third of GBMs have TP53 mutation, but TP53 mutations weren't connected with hyperprogression, so I'd guess there are other factors involved besides MDM influence on p53.
ReplyDeleteOh that's a really good point. In the original Clin Canc Res paper, 42% of the tumors had alterations in TP53, but the authors didn't report a correlation. So the hyperprogression is due either to some other activity of the MDM2 family or to some other alteration happening in tandem with the MDM2 family amplification (whether the checkpoint inhibitors are part of the causal chain or not). Or the result is spurious due to small sample size.
DeleteWith PTEN mutation, PARP inhibitors may be worth considering.
ReplyDeletehttps://btcocktails.blogspot.com/2018/01/parp-inhibitors-for-pten-mutant-cancer.html
Mutations in the TERT promoter are the most common mutations in GBM, found in over 80% of all GBM. I doubt TERT promoter mutations are predictive of response to checkpoint inhibitors in GBM, given that most GBMs have this mutation, but only a minority respond to checkpoint inhibitors alone.
ReplyDeleteNF1 mutation is much less common (around 10% of GBMs) and usually indicate a GBM in the mesenchymal subcategory. NF1 mutation is rarely found in the classical and proneural subtypes. NF1 is a tumor suppressor that inactivates RAS, so the RAS-RAF-MEK-ERK pathway could be overactive in these tumors.
I can't recall if I've pointed out this study or not, but one study showed mTOR complex 1 and MEK as key therapeutic targets in NF1-mutant cancers. Everolimus (and rapamycin) inhibits mTOR complex 1 and trametinib (MEK1/2) and cobimetinib (MEK1 selective) are FDA approved MEK inhibitors (though likely have suboptimal BBB penetration).
https://www.ncbi.nlm.nih.gov/pubmed/24913553
Thanks for this, Stephen. If I read the literature correctly, there’s good evidence that everolimus and rapamycin pass the BBB, but not so for the MEK inhibitors. And in murine models, the MEK inhibitors were shown to have poor BBB penetration (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4109207/).
ReplyDeleteI found an article that reviews toxicities of MEK inhibitors in combination with BRAF inhibitors in pediatric glioma (https://www.ncbi.nlm.nih.gov/pubmed/27494648), but my university doesn’t subscribe to the journal, so I wasn’t able to look at the full text. There’s no indication that they discuss BBB penetration, so it may not be useful, anyway.
On the other hand, I did find a case study of someone with refractory GBM who had a significant objective response to everolimus combined with trametinib (https://www.ncbi.nlm.nih.gov/pubmed/26936308), meaning at least in his case there was therapeutic penetration. The patient was NF1-deficient himself (i.e., germline NF1-mutant), which may or may not be relevant.
Everolimus and rapamycin probably have suboptimal brain penetration as well.
Deletehttps://www.ncbi.nlm.nih.gov/pubmed/24727322
See my discussion of P-glycoprotein and other ABC transporters here:
http://astrocytomaoptions.com/the-blood-brain-barrier/
The contrast enhancing component of tumors have disruption of the blood brain barrier, though this is the part of the tumor most likely to be resected, with the non-enhancing parts of the tumor protected by an intact BBB. The best recent review of this subject is here (this study reflects and informs my views on the subject):
https://academic.oup.com/neuro-oncology/article-abstract/20/2/184/4107399
(full study available in the Brain Tumor Library, Blood-brain barrier folder)
I can access most journal articles and upload them to the library on request.
A germline NF1 mutation is relevant, as all the tumor cells would therefore have the mutation as well. Germline BRCA mutations are a fundamental drug target (with PARP inhibitors) in ovarian and breast cancer.
Brandon,
Deleteyou can access a lot of articles via sci-hub. Here is full article, which you couldn't open:
http://sci-hub.tw/10.1080/17425255.2016.1214710
Everolimus achieves penetration of the BBB at therapeutic levels. It has FDA approval for the treatment of a specific low-grade glioma (subependymal giant cell astrocytoma). The response rate in the phase III trial was 58%, with 88% of responders achieving durable response (https://www.ncbi.nlm.nih.gov/pubmed/27351628). Brain levels may not reach plasma levels, but it doesn’t appear they need to.
DeleteThere’s a case study in which 4 children with primary CNS melanoma were treated with trametinib. All 4 showed some response, meaning there was therapeutic penetration of the BBB (https://www.nature.com/articles/bjc201749). (Side note: Fuck these are hard to read. The oldest patient died at age 4.)
Of course, greater therapeutic effect could still be achieved by increasing BBB penetration. Stephen, your page on the BBB is very interesting. I’m adding SSRIs and ED meds to my list to consider. I found your page from last year reevaluating the use of PDE5 inhibitors. How are you feeling about them at the moment?
Also, thanks, Matjaz. The article itself wasn’t useful but it did lead me to some other articles that were.
True enough, I'm not trying to discourage everolimus, just noting its BBB crossing could be improved by P-glycoprotein inhibitors etc.
DeleteUnfortunately, in terms of temporary BBB disruption, what works in rodents doesn't always translate to humans (this is true in general as well):
https://www.ncbi.nlm.nih.gov/pubmed/29332604
For other readers, my review of the PDE5 study referred to is here:
http://btcocktails.blogspot.com/2017/01/re-evaluating-sildenafil-viagra-in-gbm.html
I would probably be more comfortable with a several day pulse of sildenafil or other PDE5 inhibitors on the days of chemotherapy, as opposed to taking it every day, due to the prognostic associations discovered in this study.
This study shows everolimus probably shouldn't be used in the general GBM population, unless biomarkers are identified that are predictive of benefit or response (NF1 mutation could be one of these?).
Deletehttps://www.ncbi.nlm.nih.gov/pubmed/29126203
Damn. That does give me pause. 171 patients and a control arm, that’s hard to dismiss. It’s possible that a small subpopulation would benefit, but that would mean that the rest are harmed even more. I'll bring it up with the oncologist on Monday. Thanks for finding this.
DeleteFrom what I’ve gathered, the inclusion criteria for the cocktail approach are a mixture of theory and empirical evidence, both clinical and preclinical. If a treatment has a low side-effect profile and no evidence of harm, then a decent (or even weak) theoretical argument can justify inclusion. That holds even if there’s evidence of lack of positive effect (when tested as a monotherapy, for example). But it’s hard to imagine a theoretical argument good enough to justify use in the face of decent evidence of harm. Evidence really ought to supersede theory, in general.
I suppose there’s a possibility that everolimus triggered hyperprogression in a subset of patients and had no effect in the others, leading to poorer survival curves. In that case, you could argue for including everolimus in a custom cocktail, if you had confidence that you’re in the right subgroup. But there’s no subgroup analysis this study.
The situation is different with pembro. I have reasons to think my partner is in a subgroup that might benefit from it (NF1-mutant and high PD-L1 expression) and a reason to think he’s in a subgroup that might be harmed by it (MDM4 amplification), although none of those reasons are particularly strong. What I really need is clinical evidence that pembro has positive effect against NF1-mutant GBM.
Great news: my partner's scans today showed only apparent treatment effects. We'll have to keep an eye on those areas, of course, but the radiologist and doctor expressed confidence about them. My partner had some worsening of symptoms over the weekend, and the word "hyperprogression" was ringing pretty loudly in my thoughts. It appears we've dodged that bullet, at least for now.
ReplyDeleteThat's great, thanks for sharing!
Delete