Somatic genetic epilepsies: Dr. Christian Bosselmann

Somatic genetic epilepsies arise from mutations that occur early in fetal development. They are usually only detectable by genetic sequencing of tissue. For these epilepsies, the timing of the mutation is key: For example, research has shown that focal cortical dysplasia type IIB and hemimegaloencephaly are genetically the same disease, but arise from somatic mutations at different developmental time points. This relatively new area of research is discussed by Dr. Alina Ivaniuk and Dr. Christian Bosselmann.

Resources:

Analysis of 1,386 epileptogenic brain lesions reveals association with DYRK1A and EGFR (Nature Communications 2024 - C. Bosselmann et al.)

Neocortical development and epilepsy: Insights from focal cortical dysplasia and brain tumours (The Lancet Neurology 2021 - I. Blumcke et al.)

Contribution of somatic Ras/Raf/Mitogen-activated protein kinase variants in the hippocampus in drug-resistant mesial temporal lobe epilepsy (JAMA Neurology 2023 - S. Khoshkhoo et al.)

Somatic mosaicism and neurodevelopmental disease (Nature Neuroscience 2018 - AM D'Gama and CA Walsh)

BRAF somatic mutation contributes to intrinsic epileptogenicity in pediatric brain tumors (Nature Medicine 2018 - HY Koh et al.)

SLC35A2 loss-of-function variants affect glycomic signatures, neuronal fate and network dynamics (Brain 2025 - D Lai et al.)

Precise detection of low-level somatic mutation in resected epilepsy brain tissue (Acta Neuropathologica 2019 - NS Sim et al.)

Toward a better definition of focal cortical dysplasia: An iterative histopathological and genetic agreement trial (Epilepsia 2021 - I Blumcke et al.)

Seizure outcome and use of antiepileptic drugs after epilepsy surgery according to histopathological diagnosis: A retrospective multicentre cohort study (The Lancet Neurology 2020 - HJ Lamberink et al.)

Sharp Waves episodes are meant for informational purposes only, and not as clinical or medical advice.

Let us know how we're doing: podcast@ilae.org.

The International League Against Epilepsy is the world's preeminent association of health professionals and scientists, working toward a world where no person's life is limited by epilepsy. Visit us on Facebook, Instagram, and LinkedIn.

Transcript

[00:00:00] 

Dr. Alina Ivaniuk: Hello everyone and welcome to another episode of the Sharp Waves podcast. My name is Alina, I'm with YES-ILAE, and today we're continuing our exploration of epilepsy genetics. This time we'll talk about the topic that is really on the crossroads of surgical epilepsy, genetics, and neuropathology as well.

Namely, we'll be diving into somatic variation and its role in epilepsy. Today I'm joined by Dr. Christian Bosselman, who is an adult neurologist, epileptologist and clinician scientist at University Hospital Tubingen. His clinical and research work focuses on genetic epilepsies, rare epilepsies, but also somatic genetics of epigenic lesions, which was recently reflected in his recent Nature Communications paper that has already been making an impact in the field.

This is actually the second time Dr. Bosselmann is on Sharp Waves. You may remember his very insightful contributions to our episode of large language models in epilepsy. Dr. Bossman, thank you again for joining us and [00:01:00] it's a pleasure to have you back.

Dr. Christian Bosselmann: Thank you so much, Alina, and good to be back. 

Dr. Alina Ivaniuk: To get us started, could you briefly talk about the concept of somatic variants overall and how they're different from germline variants? 

Dr. Christian Bosselmann: Sure, absolutely. So when we're talking about in sequencing and epilepsy genetics, we are mostly used to be talking about detecting germline variants. These are genetic variants that are present in the gamete, and then in every cell of the body. And these are detected by usually sequencing methods from blood. So targeted deep panel sequencing, exome and genome sequencing. 

Conversely, somatic variants are absent in the gamete. They happen as post-zygotic mutation events, and they're usually only detectable by tissue sequencing.

And when we're talking about epilepsy, specifically, the type of somatic variants we are most interested in are those in [00:02:00] neuro progenitor cells, which interestingly have a much high mutation rate than embryonic progenitor cells.

Dr. Alina Ivaniuk: Okay, I see. So we are talking about the changes in genetic makeup of particular cells, not in the whole body per se, but in particular cells. Let's talk about what kinds of epileptogenic lesions are driven by those somatic variants. What can we expect from those post-zygotic changes? 

Dr. Christian Bosselmann: Of course. So we are usually talking about two broad categories of epileptogenic lesions: malformations of cortical development, MCDs, and long-term or low-grade epilepsy associated tumors (LEAT).

Among these, the most genetic entities, that is, those with the highest diagnostic yield in somatic variant detection for MCDs are focal cortical dysplasias, especially type 2B and mild malformations of cortical development with oligodendroglial hyperplasia and epilepsy: MOGHE for short. And in [00:03:00] LEAT we are mostly talking about ganglioglioma, GG, and dysembryoplastic neuroepithelial tumors, or DNETs.

And in these particular lesions, we would expect a diagnostic yield of around 35% to up to 85% depending on the type of sample and sequencing technology. And one point that I think that is really interesting is both MCDs and LEATs, this whole group of epileptogenic lesions are really best thought of as neurodevelopmental disorders.

They have a shared genetic and mechanistic background. You would think that, you know, malformations of cortical development and these tumors are very different, but in a lot of ways they're not. And there's a great perspective by Ingmar Blumcke on this. Of course we have to acknowledge that there are differences in the distributions of genes mutated, the transcriptomic patterns and so on.

So those are MCDs and LEATs. There is [00:04:00] very recent work by Sattar Khoshkhoo and colleagues who have shown that mesial temporal sclerosis may also have some somatic contributions, which is really exciting because that's one of the type of lesions that we always would've thought to be more complex in its genetic background.

Then there's a large group of epileptogenic lesions that are probably under-recognized as being genetic in origin, but we still encounter them a lot in all surgical decision making. And these are cerebral cavernomas, right. And we have to appreciate that these have a very high yield for somatic variants in PIK3CA, CCM1 (KRIT1), CCM2, and CCM3.

And they have a very interesting oligogenic sort of pathway mechanism behind that. And then lastly, just to differentiate that those were somatic variant driven lesions. And then there are of course many epileptogenic lesions arising from germline [00:05:00] syndromes, which clinicians are usually quite familiar with, like TC1 and 2, NF1, Filamin A variants in periventricular nodular heterotopia, and so on.

And finally, there's one group that is currently understood to be non-genetic, or at least non-monogenic, which are stroke, trauma, post infectious after encephalitis and including Rasmussen syndrome. So these are, you would be very unlikely to detect any genetic origin. 

Dr. Alina Ivaniuk: Well, that's interesting.

If you think about this, all epilepsy, any etiology can be broken down to a genetic cause if you think broadly about that, that's exciting. It’s also exciting how you highlighted that different, distinct epileptogenic lesions traditionally thought to be epileptogenic lesions like focal cortical dysplasia, malformations of cortical development, they're [00:06:00] really close relatives of tumors, epileptogenic tumors and the biology there is really similar in a sense. That's very interesting. Could you highlight or explain to us, how does the timing of the mutation or somatic variation during brain development influence the phenotype or lesion type from focal cortical dysplasia that may affect only a small chunk of the brain tissue to full blown hemimegaloencephaly?

Dr. Christian Bosselmann: Absolutely. So in somatic variants, timing is everything. And the best way to visualize this is thinking of the lineage that is the cells that are the progeny of a single neuro progenitor cell, you know, lineages branch out. And the point at which you have that single mutational event defines the part of the lineage that [00:07:00] is affected and how many cells are therefore affected.

Now, if you have a single mutational event early on, after or in the very first neuro progenitor cell, then the entire large lineage, that entire tree really is carrying that mutation. Whereas if you have a, a very late post-zygotic mutational event, it may only be a very small fraction of that lineage. You know, one cell and the next two or three generations, but a tiny part of what came out of that first neuro progenitor cell. 

So you can see that the timing of this mutation dries the size and location of that lesion entirely. Focal cortical dysplasia type 2B and hemimegaloencephaly are genetically speaking essentially the same disease entity—only the timing is different and this is very nicely demonstrated and illustrated in a review by Chris Walsh in Nature Neuroscience, 2022.

Dr. Alina Ivaniuk: That's fascinating. So the timing is everything. Yeah, that's, that's very interesting. Could you talk a bit about the top genes that everybody needs to know about?

Because the field is so rapidly developing, there are so many new developments and it's so hard to keep on track with everything. But is there, or are there any genes? And maybe you could also talk about biological implications of those genes that everybody sort of needs to be aware of because they are probably more meaningful or more commonly encountered.

Dr. Christian Bosselmann: Sure, happy to. So as you said, you know, epilepsy genetics is a rapidly evolving field. Every year there are new gene associations out there. Fortunately in somatic variant detection, we are at a time and place where the field is still rather good to oversee, right? It's only 19 to about 21 genes that are currently understood to [00:09:00] be associated with lesional focal epilepsy.

So you can be familiar with most of them at this point. There are, in my opinion, three to maybe four must-know genes that clinicians or researchers should be familiar with because they make up the lion's share of diagnostic yield in brain tissue sequencing and they're also most clinically relevant.

Spot number one would be BRAF and ganglioglioma, especially V600E. That's a highly recurrent single variant. And Koh and colleagues have demonstrated that this variant confers intrinsic epileptogenicity. Then you have, of course mTOR and FCD type 2B, a classic; FGFR1 in DNET also has a very strong gene-disease association for that particular lesion. And the same goes for SLC35A2 and MOGHE.

Dr. Alina Ivaniuk: [00:10:00] Interesting. So there are some gene-phenotype correlations even in somatic epilepsy. So people wonder sometimes when they read the papers on focal cortical dysplasia type 2, especially. We know that there is a particular histological phenotype or balloon cells, dysplastic cells, and we are still in the process of understanding of how those cells interact and how they incur physiology of the lesion. Is there anything that we know about which kinds of cells are mutated in those lesions and if the variants or mutations somehow influence the architecture of the lesion?

Dr. Christian Bosselmann: 

 I think the best way to answer this would be to circle back slightly and talk a bit on what exactly is going on in focal cortical dysplasia type 2B. As I mentioned, there are currently about 19 to 21 genes in lesional focal epilepsy. And the vast majority of these are either in [00:11:00] the RAS-RAF-MAP kinase pathway or the mTOR pathway, and that applies to focal cortical dysplasia type 2B.

As you're aware, the mTOR pathway is a critical junction for growth signaling. And briefly, the idea is that pathway activation leads to an overgrowth phenotype, neuronal hypertrophy, increased dendritic branching, synaptic changes. And there's also an element of dysfunctional neuronal migration.

When I do counseling for my patients and explain to them what their focal cortical dysplasia is like, I usually like to talk about these neurons as being stripped electrical wires that end up in the wrong place and cause local shortcuts, really, and that gets the point across very nicely.

And another way to memorize this concept of an overgrowth phenotype is tuberous sclerosis complex. As you know, TC1 and 2, loss of [00:12:00] function through germline variants, for example, or two-hit mechanism, reduce the inhibition of Rheb, upregulate mTORC1, and you have a pathway activation, right? And then these patients, we see dysregulated growth everywhere, you know, in cardiac, in renal, in pulmonary tissues.

And this is for us key information for disease surveillance. So that's what we understand by an overgrowth phenotype. The RAS/MAP kinase pathway largely overlaps with the mTOR pathway. There's a lot of inter talk going on. The dysregulation is more about cell cycle control and proliferation timing. And then before we move back to focal cortical dysplasia, type 2B, as we're talking about pathways I want to highlight one exception which is SLC35A2. That's a galactose transporter important for glycosylation, and the mechanism really wasn't understood for a long time because everything else is [00:13:00] either one of these two major pathways, SLC35A2 stands alone in a sense. 

And what always struck me is how different the germline and somatic disorders associated with SLC35A2 are. The germline phenotype is a congenital disorder of glycosylation and X-linked and extremely severely epileptogenic, but the somatic disease usually only leads to very subtle, but still highly epileptogenic lesions in an otherwise neurologically healthy person. And there’s a very recent paper by Dulcie Lai and Heinzen’s group that goes into much more depth on the probable underlying mechanism for that.

But moving back finally to focal cortical dysplasia type 2B, the tissues arise from your progenitor cells that further differentiate, right? And histopathologists will tell you that they observe dysmorphic neurons and balloon cells. [00:14:00] And both of these are the result of that overgrowth phenotype that we just talked about. Both of these cell types carry the pathogenic somatic variants that lead to this upregulation of the mTOR pathway. 

In terms of the lesion architecture, I think we have to appreciate that focal cortical dysplasias are heterogeneous. You know, it's not just it's not a block of mutated tissue. There's an unequal distribution of these mutated cells throughout. Indeed, if we sample the same lesion multiple times, the diagnostic yield from somatic sequencing usually increases. And we can also see a gradient as we approach the center. The most mutated part of that lesion, the number of mutated cells increases, and the epileptogenicity on sEEG, for example, increases. [00:15:00] 

So they are complex and heterogeneous lesions in a sense. 

Dr. Alina Ivaniuk: Well, that's interesting and thank you for circling back to other types of lesions as well and sort of correlating the biology with the phenotype that we see in comparing the germline and somatic phenotypes. I think it's also useful to think about that in that way, to try to conceptualize how different sorts of pathological mechanisms can lead to epileptogenicity in different ways.

And I think that the last point that you mentioned sort of segues to my next question pretty well. There may be some challenges in detecting those variants by just the nature of how heterogeneous those lesions are in their histological composition and also genetic makeup.

Could you walk us through some technical and biological challenges that arise with attempts to sequence and detect somatic variants in those lesions? [00:16:00] 

Dr. Christian Bosselmann: Yeah, I mean, it's a tough problem. So in somatic variant detection, epilepsy, t's a needle in a haystack problem. Our colleagues in oncology have it easy by comparison, because in cancer, the majority of these oncogenic variants confer a strong selective advantage, resulting in clonal lesions.

That's not usually the case in epileptogenic lesions. They don't have a strong growth. You know, gangliogliomas, DNETs, they are very static. Focal cortical dysplasias don't change much. And therefore, the number of mutated cells in these epileptogenic lesions, the variant allelic frequency, in a sense, the VAF, is usually below 5%. That is, less than 1 in 20 cells carries that pathogenic mutation. In some lesions, less than 1 in 100 cells carry the somatic mutation, [00:17:00] and that's still enough to cause severe epileptic seizures. And this was best shown by Nam Suk Sim and colleagues in a large cohort where they did size specific amplicon sequencing.

So the lesions are, for the most part, normal tissue. And there's a few, very few of these, as I said, these stripped wires, these dysmorphic neurons, balloon cells or BRAF-carrying cells and so on that cause the actual seizures. 

If we want to detect somatic variants from this haystack of normal tissue, we need a high sequencing depth. So we need to obtain enough reads of the same segments to have a chance to read the mutant allele. We are mostly doing bio brain tissue sequencing, so the tissue is homogenized and then sequenced. So the number of mutated alleles as a proportion of wild type and the percentage of mutated cells [00:18:00] becomes essentially the same.

There are more elegant ways of doing this, target enrichment and single cell approaches, but bio tissue sequencing is a lot more clinically feasible at scale, but still we need a high sequencing depth.

And this quickly gets extremely expensive. But I'm optimistic that this will get better. The trend we've seen over the last 10, 15 years is that sequencing costs always decrease. So this technology will only become more readily available. Then we have a lot of software limitations because we're still actively figuring out on how to best analyze the sequencing results. There are really no good standard established pipelines for this, because most of this is from oncology. And their pipeline settings, the software they use is usually not fully applicable or optimal because they have a very [00:19:00] different set of genes that they're interested in. And as I mentioned, a very different level of variant allelic fraction where they would expect their variance to be.

So for the study we did, the one you mentioned, we needed to develop custom in-house pipelines to make this happen. There are some specialized tools developed for epilepsy, like Mosaic Forecast, and especially Deep Mosaic. And there's thankfully some additional information and guidance by the Brain Somatic Mosaicism Network.

There was software, and lastly, technological limitations. And these are real world concerns. Optimally, we'd like to work with fresh frozen tissue. But that's not usually all easily available sometimes. We have to make do with paraffin embedded tissues, which come with a whole set of challenges in the way they are prepared for sequencing and the amount of sequencing artifacts you would [00:20:00] expect afterwards.

So we have to spend a lot more time making sure that we minimize our false-positive rates, you know, accounting with duplex sequencing and optimizing our pipelines in a way that we discard as many sequencing artifacts as possible, then checking the rest. 

It's a matter of sample availability. Very few centers operate enough where you could power large gene discovery studies. So this usually needs biobanks. And a lot of what we did was driven by the Cleveland Clinic and the European Epilepsy Brain Bank. And without these two centers coming together and pooling their data and resources, our large-scale study really wouldn't have been feasible.

But if you're working in biobanks, you are usually restricted to brain samples only because they're from histopathologists. Histopathologists save brain. And there are few biobanks that have [00:21:00] enough paired samples, because we would usually optimally like to sequence both blood and brain. So looking towards the future, I think we've become a lot more aware that paired samples are superior. And we are making rapid progress in terms of the pipelines and the sequencing technology that will inform further studies. 

Dr. Alina Ivaniuk: Thank you so much for breaking down all those challenges at different levels. And it seems that there is, there are so many things that could be improved, but developments are fairly exciting and rapid and hopefully we'll get there.

And let me segue to another question that I think I heard quite a few times. And that's people's speculations or rather people questioning whether or not MRI-negative epilepsies could be caused by somatic variation. And if so, are we yet there to say or to try to understand if that could be true? Can you comment on that? [00:22:00] 

Dr. Christian Bosselmann: That's a very good question because as you know, MRI-negative cases in our surgical decision making are particularly challenging, and that's really where a lot of our know-how and resources are pulled to figure out a way to make these cases lesional in a sense, you know, being sure that we have the right target for the resective approach.

We've already talked about the concept of lesions with a low variant allelic fraction, right? Samples where 1 in 100 or 1 in 20 cells carried the mutation. We've also talked a lot about the timing of the post-zygotic mutational event. Epileptogenic lesions that are MRI negative are often very late post-zygotic mutational events with a low VAF and very, very subtle histopathological changes.

In my opinion, this is one of the best use cases for somatic variant detection. Consider a sample [00:23:00] where you see very subtle changes in the Olig2 staining, but then you get the additional information that there is a pathogenic somatic variant in SSC35A2 that's present. That would make you as a histopathologist in that case, much more confident that this is a MOGHE, you know, a known epileptogenic lesion.

And therefore the whole epilepsy surgery team can be certain that an epileptogenic lesion was resected and that this wasn't just healthy cortical tissue. And that's key for the postoperative prognosis and counseling. 

Detecting somatic variants in MRI-negative cases before resection is another topic entirely. Previously we were talking about the time after resection and what we do with the tissue afterwards. Now we're talking about the time spent in phase one, phase two, and informing the patient management conference decision making. Before you have access to brain tissue, you can't really [00:24:00] do much because these variants we were talking about are private to the brain. They are from neuro progenitor cells. 

There are some groups looking at sequencing from CSF or cell-free DNA and blood. And one thing that has made a large impact over the last two or three years was sequencing from depth electrodes. That's difficult to implement. The groups that have put the work into it, the centers have gone through a ton of technological advancement. But this is a really exciting way to bring this whole concept of somatic variants to the presurgical planning as well. So, yeah, I think in MRI-negative cases, it's one of the best possible additional diagnostic modalities.

Dr. Alina Ivaniuk: That's so very interesting. And I think you also jumped a bit into the next question that I was about to ask, what the clinical relevance of identifying somatic variants, and you talked about how it could inform our surgical planning, [00:25:00] prognostication, counseling of the patients. If you could expand on that more and probably also talk about your own practices? If this knowledge somehow impacted the way you consult patients, or you guide the discussions in the patient management conferences, share your experience. 

Dr. Christian Bosselmann: Sure. So as you said, I already picked up one example where histopathologists benefit from genetic information especially in these hard-to-diagnose cases with rare histopathologies or very subtle changes that are prone to different diagnosis, depending on what sort of immunohistochemistry you'd have access to, for example.

And that was very nicely illustrated by Ingmar Blumcke in an iterative agreement trial in Epilepsia a few years ago. So we've seen that genetic information yields a more accurate histopathological diagnosis. And we also know from work by Lamberink and the European Epilepsy Brain Bank [00:26:00] that a good histopathological diagnosis directly informs prognosis because the rate of postoperative seizure freedom directly depends on the type of lesion that was present in the first place, but that's a rather indirect use of the genetic information. 

We're also see starting to see specific subgroups beyond histopathology. There's work by a group in Bonn that have looked at the transcriptomic profiles of BRAF, altered gangliogliomas, for example. And there are specific patterns of these gangliogliomas that are more or less prone to seizure recurrence postoperatively. Similarly the Erlangen group has looked at PTPN11 altered ganglioglioma, where you could see that there are gangliogliomas that carry just V600E, this highly recurrent epileptogenic variant that we talked about earlier, who usually do quite well after surgery, and then there are [00:27:00] these gangliogliomas that have two or three or four additional somatic variants that have a much higher chance of a worse surgical outcome. 

So I think as we become more aware of these genetic biomarkers, now this patient stratification this becomes much more valuable information for postoperative patient counseling.

When we're talking about potential clinical impact, we're also usually talking about precision medicine, genetics, right? In somatic variants, it's still very early days. Of course, if you have a lesion where you know that there's an mTOR pathway activation, it would be tempting to use an mTOR inhibitor. And after the EXIST-3 trial, they are, you know have been shown to be quite effective and safe in other mTOR pathway disorders. We don't quite know how well they do in epileptogenic lesions [00:28:00] because we don't have access to larger well-designed studies yet. That's work that's ongoing. So these are currently still firmly off label.

Other antagonists, for example, for oncogenes in LEAT, are even more preliminary. So that's hopefully something left for the future. 

Lastly you mentioned my own clinic experience, and I think that that's important to sort of put this into perspective. The study we did was on, you know, 1,400 biobank samples and that's data and tissue that's obtained long after the initial epilepsy surgery. And that's important because you need to have some frame of reference for the outcome and the data collection. In my own personal practice, this hasn't made an impact yet, but I firmly believe that we are rapidly moving towards a point where this concept of somatic variant detection of sequencing brain tissue after epilepsy surgery becomes [00:29:00] technological and computationally feasible for clinical implementation.

To really make this happen, we need an actual prospective clinical implementation trial because as I mentioned earlier, there's a lot of technological limitations that we need to work through.

Dr. Alina Ivaniuk: How far in the future do you think it's going to happen? Do we talk about matters of years or dozens of years? How far away do you think it is from the actual bedside? 

Dr. Christian Bosselmann: So I think the most encouraging thing is that there are already some very first academic and industry providers for epilepsy somatic variant detection. So there are already places where you can send in brain tissue and get a panel sequencing result.

Clinicians are becoming more aware of the diagnostic modality not just of somatic variant detection, but epilepsy genetics in the [00:30:00] context of epilepsy surgery as well, hopefully also through this podcast as well. At this point, you still need a lot of institutional support and expertise because this is a team effort. You need a good collaboration between epileptologists, epilepsy surgeons, geneticists, bioinformaticians. So it takes a village. But I really want to underline the point I made earlier. I firmly believe that both germline and somatic variant detection and sequencing from both blood and brain will become standard modalities of the pre- and post-surgical evaluation of epilepsy patients at some point in the near future.