Good morning everybody. I'd like to thank the organizing group for inviting me to speak and share a little bit of information about use of stereo EEG in infants and very young children. I have no disclosures to discuss. In my talk, I hope that I will convince you that epilepsy surgery is a worthwhile endeavor to explore in the young children with drug-resistant epilepsy. We'll talk a little bit about the different modalities of invasive monitoring and the risks and benefits of those procedures. Then I'll provide examples that you can hopefully bring back home to your institutions about how you can use your pre-surgical information to guide an SEEG implantation. Rohini already talked a little bit about the scope of the problem, the prevalence of infantile epilepsy within the United States, as well as the fact that a third of individuals, specifically children under the age of 36 months, are drug-resistant. The ILEE has put out a recommendation that epilepsy surgery absolutely should be considered in infants for drug-resistant epilepsy, which is not surprising. We consider that for older children and adults. But they also emphasize that for infants with focal onset seizures, particularly those with lesional epilepsy, that should be considered an early process. And yesterday's Hoyer Lecture was a nice introduction about how early epilepsy surgery can help to reduce the morbidity of seizures, but can also prove to provide developmental improvements so that these kids can reach their best developmental potential. Now, infants, of course, are not small adults, and there are distinct challenges associated with epilepsy surgery evaluations in the very young. Clinical semiology and clinical onset can be challenging because these infants are unable to describe their auras, and it's hard to determine when, of course, that starts. It may also be difficult to assess awareness in these individuals and to test them. These challenges with the semiology can compromise our ability to look for a preceding electrographic pattern and may delay an ictal injection for spec'd. Now we know that in infants, the ictal pattern actually can be challenging as well. Multiple seizure types can have a very diffuse EEG signature despite having a focal onset. And numerous studies have shown that seizures with a diffuse electrographic pattern can be successfully treated with a focal resective surgery. From an imaging perspective, since children are incompletely myelinated, it can be very difficult to identify lesions as well as to determine lesional boundaries. And, of course, infants are not hardheaded like adults. They have very soft skulls. Now that limitation has classically led to us considering epilepsy surgery and invasive monitoring using subdural electrodes. Now the advantage of this, of course, is you've got this beautiful 2D regular array for gyro resolution. On the flip side, however, this does require a large craniotomy, and there's certain morbidity associated with that procedure. In addition, there are certain brain regions such as the orbital frontal region which can be very difficult to sample using subdural electrodes. Now Rohini talked a little bit about how the field is moving towards less use of subdural electrodes and more towards stereo EEG monitoring in general. And there are distinct advantages to this approach. Specifically, you can sample anatomically disparate areas. So areas in both hemispheres, areas in those difficult to reach areas like I mentioned the orbital frontal region. For certain conditions like tuberous sclerosis where they have multifocal tubers, this offers a distinct advantage. This technique also allows you to sample various areas of the ictal network so you can have a better understanding of the whole ictal anatomic electroclinical relationship that's occurring. It also allows for repeat epilepsy surgery evaluations. This, again, is particularly useful in an infant population where, for instance, again with tuberous sclerosis, they may develop additional seizure types later on in life that may be amenable to repeat epilepsy surgery. And using stereo EEG, you don't have as many adhesions as you would with a craniotomy. There are, of course, this doesn't require a craniotomy so that reduces the risk of infection and may have comparable or lower risk of hemorrhage. Now the disadvantage of stereo EEG is how myopic it is. You really are only seeing the brain where you implant it. So it's crucially important that you have a good implantation strategy because if you're just not covering it, you're not going to see ictal onset and not going to be able to find the epileptogenic zone. There are lots of different techniques about how you implant a stereo EEG. I'm going to show you what we do at our institution, but of course there are a lot of other places that do it very well and may do it slightly differently. The approach that we use is that when we implant, we are implanting everything as a potential treatment corridor. Specifically, if we are going to do LIT in the future, each of these trajectories can be used for that modality of surgical treatment. In addition, the way we implant our electrodes, we are aiming not necessarily just for the distal source of where that electrode is implanted, but we will monitor along the entire trajectory. So in some instances, we're gathering useful information both from the distal context of that electrode as well as the proximal context of that electrode. And again, every institution is going to do that a little bit differently. Now let's review a little bit about the process of how a stereo EEG is implanted, for instance, using robotic assistance. Now this all starts with a very high resolution MRI and vascular imaging, which allows our neurosurgical colleagues in conjunction with the epileptologist to make a trajectory plan. Now using either a frameless approach with bony fiducials or a frame-based approach with pediatric pins, you can do final trajectory planning, you can co-register your robot. Now the robot is really good at making things go a little bit faster to navigate to that electrode trajectory. You're going to make a stab incision, a small skull defect with, for example, a twist drill, and then you secure an anchor bolt into the skull. Then a stylet is introduced, that is removed, and the stereo EEG electrode is advanced through that track and then ultimately secured to the bolt. This is an iterative process with the robot, and you can repeat that as many times as needed, depending on how many electrodes you're implanting. Now the critical part for infants is this step at the bottom in the middle, which is the anchor bolt. And this has been a stumbling block that people have identified as being potentially problematic in the infantile population. So what exactly is the issue? Well, we were worried about skull thickness, because you're securing an ankle bolt into the skull, and you really have desired, reported in the literature, a thickness of at least two millimeters. And there are serious concerns about if that bolt is not secure. If that bolt slips, you can have mis-targeting of your electrode. That can lead to hemorrhage. The other possibility is if that bolt slips, it can create fractures, which again, can lead to hemorrhage. And we want to avoid those complications if at all possible. Now we know that skull thickness is an age and location developmental process. So obviously, as you get older, with older pediatric kids or adults, this becomes less of an issue. But for infants, this is a serious concern. Now, interestingly, a group in Southern California has studied the skull thickness of healthy infants under the age of 36 months. And what they did was they did three TMRIs, and they segmented out the bone, as shown in the image to the right. And they measured the thickness over time in these individuals. And what they did is they then plotted their data out in a color map. So you can see here from their image, focus first on the color map. So thin thicknesses, so thinner than what we would like, less than two millimeters, are going to be in your blue color range. Whereas the more desired thicknesses, what we would like, two or above, are going to be in your teal, yellow, orange range. And you can see from this data taken at 12 months, 24 months, and 36 months, a fair amount of the skull actually is in our desired range. So there's a lot of teal and a lot of yellow-orange being displayed. So this suggests that while we have classically been very conservative and concerned about skull thickness and using bolts in infants, this suggests that a bolted stereo EEG implantation approach is actually feasible based on skull thickness. Now of course we're going to give you a little bit more hands-on information about that. I'm going to focus on the information from my institution. Now planning an SEEG for us is not significantly different in an infant from an older individual. So we'll run through this fairly quickly. It of course is a multimodal presurgical evaluation which includes history and exam, a neurophysiologic evaluation, imaging evaluation, and a functional evaluation with for us the more common modalities indicated here in yellow and less commonly used modalities indicated in orange. Now all of this data is gathered, discussed in a multidisciplinary surgical conference. And then based on that we make a personalized recommendation for each individual. There are some individuals that go straight to surgery. So they can go straight to either a surgical resection or an ablative procedure. And depending on the circumstances we may choose to do post-operative invasive monitoring to fine-tune the surgical corridors to again aim for greater surgical resolution, seizure resolution. Another option is that if we have good anatomic correlation we may choose to do classic subdural monitoring with grids and strips. This of course can lead to a surgical resection, again with the possibility of post-operative repeat subdural monitoring or could lead to a neuromodulatory approach with closed-loop stimulation in an older individual. Bear in mind of course closed-loop stimulation is not approved in the pediatric population. So that would be off-label. If we have anatomic disparate regions that we're interested in of course we're going to consider stereo EEG. And this can again lead to the various approaches of surgical intervention with the possibility of post-ablative stereo EEG monitoring. Now SEEG also can be fine-tuned. For instance if we gather some information about regionalization of the ictal pattern we may choose to implant further electrodes in that region for greater resolution. Or we may find that based on a desire for functional mapping we'll actually convert into a subdural electrode invasive monitoring strategy. So those are generally the paradigms that will be considered in surgical conference. Now in terms of planning a stereo EEG specifically we're going to go back to that pre-surgical evaluation and look at those specific modalities, each of those little pieces of data that we've gathered in that process. And just to give you a couple of examples of how we would plan that, we have an individual with a vague somatosensory aura that then develops a focal impaired awareness seizure with autonomic features that led us to be a little bit suspicious about the cingulate. And we look at the cingulate we find pointing at the big red arrow a nice juicy tuber located in the cingulate. And when you overlay that with the PET you can see this very nice area of hypometabolism as well. And so hopefully you can see in the image to the right that electrode traversing right through that tuber, again as a potential ablative corridor if we need to use it in the future. Here's another example. We have an aura with a head rush sensation followed by an inability to speak. We were interested in implantation in the right insular region. It was a non-lesional area but it lit up on an ictal subtracted spect. There was a very robust med cluster in that region and it was also a region of pet hypometabolism. Maybe a little bit more challenging to see but in the image to the far right with the big red arrow pointing at it is a small purple electrode that is again traversing that area of interest. Using this we developed a rationale for each stereo electrode implant. So semiology, spect, pet, med cluster. You can generate then a list of electrodes using this rationale. Now what I'm showing here is an example of one of our implantations for a child with TSC who had a very high tuber burden and we were simultaneously characterizing three distinct seizure types. So there are by necessity, unfortunately, a lot of electrodes implanted because we are trying to characterize three different seizure networks. Now in addition to planning the stereo G, one of the important parts of this process is actually just wrapping the head. Each of these bolts is individually wrapped with gauze. The cables are bundled, folded and taped to the vertex of the head with additional padding so that minimizes the amount of tension on the electrodes and on the bolts. And then beyond that there is a large circumferential head wrap so that this child now looks like a cone head. There are still instances of kids that are very active. They are doing alligator rolls in bed and we do sometimes lose electrodes, sometimes necessitating going back to the OR and replacing them. And that is something that we do warn families of in advance, especially infants, that we may need to go back to the OR just to reimplant an electrode because that's a critical electrode and we haven't gathered the necessary information as of yet. So let's quickly do a case example. This is a 3-year-old with TSC and refractory epilepsy. He had already had one epilepsy surgery. At one month of age he was having 100 seizures a day which necessitated an emergent right frontal tuberectomy. He was briefly seizure-free but rapidly had emergence of additional seizure types. He was medically refractory having failed multiple anti-seizure medications as well as dietary therapy. He did have developmental delay in the language, social interaction and motor spheres. His typical debilitating seizure was a focal seizure with alteration of awareness and asymmetric tonic posturing with a left effect and more than the right. Now going through our different pre-surgical evaluation, imaging is going to be our next spot. And you can see from the CT and the MRI, he actually had multifocal tubers with the largest being this calcified tuber in the right parietal region. Now interestingly we found a nice meg cluster that located to that region and not surprisingly that tuber was associated with pet hypometabolism. So we have very nice imaging and neurophysiologic concordance with that area being suspicious. Now in terms of his EEG, this is what it looks like. Good sensitivity, standard page. Don't expect you to read it other than saying it's very disorganized with very high amplitude electrographic discharges with the spike wave index really being 80-90% both in wakefulness and sleep. If we turn it way down so we can actually see something, you'll see in the first few seconds of the page that you can see multifocal epileptiform discharges as well as multiple regions of FAST activity. Now with the ictal event itself, what we found is that there was repetitive spiking in the right posterior region which is followed by some subtle rhythmic theta activity primarily in the right central region. There was then a small amount of FAST activity in the right temporal parietal region before this large delta wave complex which corresponded to clinical onset. Then you can see here the evolution of that into spike and wave activity which starts in the right posterior temporal region. So after reviewing all of this child's data in conference, what our group decided to do was go ahead and do a surgery-first approach. So underwent laser ablation of that parietal tuber followed by post-ablative monitoring by stereo EEG. So this is an image of the lit of that particular tuber. Now in terms of his post-ablative monitoring, we've put in five electrodes as well as covered with a surface 1020. That is one potential approach to give you a little bit more geographic representation of the electrographic activity. We monitored him for three days and we captured 12 of his typical seizures. What you can see here is the depth electrodes on the top followed by the surface electrodes on the bottom. What the first arrow is pointing at is a spike complex in these purple electrodes, PP4, which is shown in the inset image. Then the second arrow in the image actually shows a buildup of FAST activity both in that purple region, again, that PP electrode chain, but also in the green chain which is an electrode in the right central region. Now this is all subclinical at this point and you can see based off the surface EEG, this buildup of delta activity that is happening followed by a decrement that happens at the last second of this page. Out of this decrement, we now have clinical onset on the next page and then evolution, again with a fairly diffuse signature on surface EEG, but a robust change in the temporal and central regions on the implanted electrodes. So ultimately, based on this data, what we did was an extension of the previous parietal ablation, so we went a little bit more inferiorly than we were before. So getting a little bit more perituberal tissue as well as ablating an independent small tuber in the right temporal region. At one year follow-up, we were pleased to see that his EEG had dramatically improved. This is again at standard sensitivities on the bottom. So you can see the dichotomy between that early EEG and how it looked afterwards. He's angle one, thankfully. At a year of follow-up and developmentally, he had already been starting to show improvements. He remained nonverbal, but he was moving more, he was walking independently, he was now using both hands equally. And then thankfully at two years follow-up, he remained seizure-free. Now that's one case example. Let's expand it to a larger population. This is a cohort of 38 patients with TSC that we treated at Texas Children's, aged between a year and a half and 13 years of age. They were having frequent daily seizures. From this group, 26% went a subdural route for invasive monitoring, whereas two-thirds, we did stereo EEG. And the diagram to the right shows an age distribution of those individuals that underwent stereo EEG with age plotted on the y-axis. Now if we do a subset of this, because obviously this SIG is all about the young children, if we look at just those individuals that are less than three and a half years of age, we've got 23 individuals in this cohort. Looking specifically at the younger cohort, there is a little bit more subdural monitoring that occurred in this population. So we're at 39% now compared to 26. But still a significant portion of this population underwent SEEG monitoring. And again, I think that's related to the underlying condition of TSC and the fact that tubers can be multifocal and in both hemispheres. If we look specifically at the stereo EEG younger cohort, you can see that again the age distribution with our youngest individual being a year and a half old, these individuals had 5-18 electrodes implanted and they were monitored 2-10 days to capture their daily seizures. Now out of this monitoring, what was the surgical recommendation? If you look at the entire cohort, laser ablation was the more commonly recommended procedure as the first intervention with 24 of those individuals undergoing LIT. Now some of these individuals did go on to have a second surgical procedure. Because as I said, sometimes we did do post-operative monitoring and would repeat the process. So in the lighter bars you can see those that had a second surgical procedure. So LIT more common in the total cohort. Looking again at that younger population, less than 3-1⁄2, we've now got purple bars. You can see that actually it's fairly equal. We now don't have a dramatic skew towards ablation. It's actually fairly even with a slight prevalence, a preference to resection versus ablation. And again, some of these individuals did have a second surgical procedure. Now we've surgically treated them, but how did they ultimately do? So if you look at the total cohort, we actually found that 80-90% of these individuals had a greater than 50% reduction in seizures and had a familial reported developmental improvement. And there was no statistical significant difference between those that were treated with resection first or with an ablative approach first. I mentioned that some of these individuals did have iterative surgeries. And in fact, those that had a second or a third surgical procedure did ultimately again have a greater than 50% reduction in seizures. Looking at the younger cohort, we can actually see that these outcomes are actually fairly similar in this group as well. Now of course, our institution is not the only institution doing stereo EEG in young individuals. This is a nice case report out of CHOP describing a 17-month-old with TSC where they placed 10 electrodes again using anchor bolts. And they actually described that they went in skull as thin as a millimeter. And they didn't have any problems with anchor bolts and stereo EEG even with that thin of a skull. They were able to identify an epileptogenic zone and ultimately the patient underwent a tuberectomy. Now if you're still a little gun-shy about the bolts and the thin skull, there is another potential option and that's a boltless approach. So there's another technique here described, a paper in an older individual with a dramatic lack of bony skull in a large region as well as in a younger cohort down to the age of three years where they used a split cannula approach. And what happens in this approach is they're actually, rather than securing an anchor bolt, they're using a robot or a frame to guide a cannula along the electrode trajectory. They then advance the stereo EEG electrode through that and then remove the cannula. They can then secure the SEEG electrode to the scalp itself. So this does not require a bolt. But theoretically, hypothetically, if you have a lot of scalp edema, that will actually shift the position of your electrode, so something to bear in mind. So hopefully I've shown you over the course of this talk that seizures in young children are a problem to be taken seriously as it can adversely affect development. Younger children do have thinner skulls, but this should not stop us from considering stereo EEG implantation. And hopefully I've shown you how multimodal data in a presurgical evaluation can guide implantation of a stereo EEG approach and ultimately be used to guide a surgical intervention and lead to seizure reduction. I'd like to acknowledge the fantastic group of epileptologists, neurosurgeons, psychologists, APPs, radiologists that I work with at Texas Children's Hospital, all under the guidance of Dr. Riviello and Dr. Weiner, as well as to show my appreciation to all the referring providers out there who send us our patients and let us take care of them. Thank you.

epilepsy surgery

stereo EEG

drug-resistant epilepsy

infant challenges

seizure reduction

multimodal evaluation