Chapters Transcript Drug-resistant Epilepsy - Interventions beyond Pharmacotherapy Adam S. Greenblatt, MD, outlines the benefits of surgical interventions in DRE. Um I'm gonna talk today about interventions beyond pharmacotherapy and the treatment of drug resistant epilepsy. And so our objectives together are gonna be to distinguish between drug resistant epilepsy and pseudo resistant epilepsy to outline the benefits of surgical interventions in drug resistant epilepsy. And also we're gonna try to appraise the utility of different neurostimulation or neuromodulatory strategies for these medically. Uh I do have one disclosure related to this talk. Um I will be presenting some information about the response of neurostimulation system um from neurop pace. I don't have a longitudinal financial relationship with them. They did give me a stipend to go to a round table um to discuss uh practical use with the device. Um None of my other disclosures are relevant to this talk. Uh Very briefly before we talk about drug resistant epilepsy. I think it's really important to um emphasize distinctions between um seizures and epilepsy. So, seizure is an event, it is a transient occurrence. Whereas epilepsy refers to an enduring predisposition to having recurrent seizures and this is a point that's even lost on my trainees, my residents and my fellows and you can reach a diagnosis of epilepsy classically in one of three ways. One is that you have more than one lifetime provoked seizure. One is that you have a one lifetime seizure and it's part of some syndrome that's known to carry a va vulnerability to having seizures such as having a seizure in the setting of Angelman Syndrome or Rett syndrome or some other primary generalized epilepsy syndrome, classically like juvenile myoclonic epilepsy. But also if you have one lifetime seizure, you can have a diagnosis of epilepsy. And that's if your recurrence risk is greater than 60%. So the question is, how do we get to that recurrence risk of greater than 60%? And it comes in one of two ways. One is if you have a structural lesion, uh that is cla classically implicated in epileptogenic. So, here in this MRI, this patient has left mesiotemporal sclerosis. This finding uh increases their recurrence risk greater than 60%. So even with one lifetime seizure, if you have a finding like this on MRI, the patient has epilepsy. Similarly, if you have an eeg that shows abnormal interact or epileptic form discharges, two increases the recurrence risk to greater than 60%. And that's why it's so important that in any person who has one lifetime seizure, they get an eeg because that can re stratify their risk of recurrence over the next decade and can inform whether or not they should be on an anti seizure medication right away. So just some background epidemiologically. Um There was a study conducted by Alan Hauser uh back in the 19 nineties that explored time trends in the age specific incidents of epilepsy um out in Rochester. And this graph up here looks at different decades between 1935 and 1984. Um And I tried to list those here for you. Um But the, the point is that there is a bimodal distribution in the incidence of epilepsy. We classically see it in very young patients because of vulnerabilities uh that underlie this particular particular demographic like perinatal injury and genetic susceptibilities to epilepsy. And then we kind of see it taper off but it never reaches zero. There is some incidence of epilepsy throughout life and we see it again as patients get older because of changes to the brain that put them at risk of developing epilepsy like strokes and neurodegenerative diseases. A systematic review of the lifetime uh pooled incident of epilepsy across countries uh conducted by Feast and others. More recently in 2017 showed a pulled incidence of 61 patients per 100,000. Um And we also learned from that study um and pretty much confirming something that's more intuitive is that epilepsy rates are higher and lower and middle income countries than they are in high income countries. And that's probably because of the comorbid risk factors for epilepsy that predominate in those particular countries. So there are higher rates of traumatic brain injury, there's poor uh maternal fetal care. Um and there are other things that predispose people to epilepsy. So what is drug resistant epilepsy? There is a very specific definition for this and there hadn't been for many years, but drug resistant epilepsy is defined as the failure of adequate trials of two well tolerated and appropriately selected medications for the type of epilepsy the patient has. And so this um was reaffirmed in a position paper in 2010. And this is what we mean when we say drug resistant epilepsy, there are some exceptions to this rule in people who have cava noma related epilepsy or um certain types of long term epilepsy associated tumors like d nuts. But generally speaking, we're talking about failure of two anti seizure medications. And so the question is why, why did they pick the number two? And we have to go back to a seminal paper that was published in 2000 in the New England Journal of Medicine by Kan and Brodie. And up until this time, we didn't know this information. Um but Quan and Brody looked at 525 patients perspectively between the ages of nine and 93 who are given a diagnosis of epilepsy and treated at their center. And I want to turn your attention to table two. So um after the first seizure medication um was introduced, 47% of patients became seizure free after the second anti seizure medication was introduced in people who continued to have seizures, 13, an additional 13% became seizure free. But after two medications, you can see right here that less than 5% of people became seizure free. So when we never stopped trying to work with anti seizure medications to achieve seizure freedom, the chance of people achieving seizure freedom is less than 5%. After two medications. Following up these initial insights, there was an outcome study that was undertaken in Scotland in an unselected cohort of newly diagnosed epilepsy. Patients to determine the number of anti seizure medications needed to fail before it could be determined drug resistant. This was six years later and basically, they had the same finding. So about 65% of patients became seizure free for at least 12 months. And throughout the course of the study, about 60% remained in remission. Uh These graphs here uh show remission rates according to age uh after an incident diagnosis of epilepsy. And it's stratified by primary generalized epilepsy, focal epilepsy and all epilepsy. And you can see that the patients in midlife um had the lowest chance of being in remission. And this is important, practically speaking, because um interventions that can improve health outcomes and improve quality of life have a greater impact on quality adjusted life years if we can offer interventions earlier. And it seems like these are the patients um who would stand to benefit the most or who are most represented by the, the, by the people who are vulnerable to having drug resistant epilepsy. So this is a timeline of all the anti-seizure medications that have been um developed. Um There's, it's nearly complete. Um There's probably a few that are off here. Um For example, Florine's a new medication that's not on here. Um But we, there, there's a lot of ways to divide anti seizure medications. And as an epileptologist, we classically think about them in, in broad categories of 1st, 2nd and 3rd generation and that's because of the different profiles that each generation has. So older medication typically tend to have more drug drug interactions. Um and they have more narrow therapeutic windows and they require regular blood monitoring. They have um a whole constitution of uh of risk for long term endocrinologic dysfunction, particularly osteopenia and osteoporosis and um increase certain risk factors for um associated with the vascular disease. Um And so we have to be mindful when reflecting medications which medications our patients are on. And if we can further optimize them, you might ask. And rightly, so given the, the very large investment in developing anti seizure medications over the last 25 years, how has that impacted our seizure freedom rates? And so that study was explored in 2018. This is a longitudinal cohort study following a succession of newly diagnosed epilepsy patients who presented at a single health center and there was a significant difference in overall probability of seizure freedom between patients treated with the 1st and 2nd, anti seizure medications and 2nd and 3rd. But after that, still no difference between the third and the fourth and the fourth and the fifth regimens. And if you look at this table over here, which is a little bit more intuitive, looking at the cumulative effect of the previous regimen. Again, you see that there's a, a big difference in the probability of seizure freedom um by introducing one and also adding a second but less. So for three and really no difference between three and 44 and 55 and six and six and seven. So really, despite the very wide armamentarium of medications that we have really, it hasn't made a huge dent in the number of people becoming seizure free. So we have to think of uh why is a patient not becoming seizure free? It's very possible they have drug resistant epilepsy, but it's also possible they have pseudo resistance, meaning the appearance of being drug resistant, but they are having breakthrough seizures or events for some other reason. So, um there's a lot of reasons why someone might continue to have events. One is misdiagnosis, right? So there could be some condition that mimic seizures like pnee or syncope or ti A. Uh and the patient might be having those. There's also the possibility the patient has more than one thing going on. So they could have an accurate diagnosis of epilepsy, but the epilepsy is well controlled and it's the uncontrolled paroxysmal events of another type that are the ones that we're really breaking through. And often times it can be very difficult for us to parse it out. We use an epilepsy monitoring unit to try and tease it out. So we'll refer patients electively after requiring a precertification and a pre authorization to get them into the hospital and we'll take them off their medications and we'll try and characterize their habitual events. Uh Another consideration for breakthrough events is inappropriate medication selection. So there are broad spectrum medications and narrow spectrum medications, narrow spectrum, meaning they really only work for focal epilepsies and broad spectrum, meaning they work for both focal and generalized epilepsies. And so sometimes people have an epilepsy that's responsive to medications, but you need to switch them to the appropriate medication for the type of epilepsy that they have. The, the classic example of this is a patient on carBAMazepine with juvenile myoclonic epilepsy whose seizures are worse, the particularly the myoclonic seizures, then they get switched over to a broad spectrum like Depacote or lamoTRIgine and their seizures improve. We also have to think about drug drug interactions. Is it a possibility that one of our medications is speeding up the clearance of another one? And therefore, it looks like the patients failed to when really they're at a sub therapeutic level of one of them. That's an important consideration and then the most common cause of in of, of breakthrough seizures, which is medication, no adherence. So the medications only work if you take them. Um some medications have a longer half life and that, that's really helpful for patients who tend to miss a dose here and there. Um, but more often than not patients have breakthrough seizures because they forgot to take a dose or two or three. And so, in that case, it's not about pursuing alternative interventions. It's making sure that they can maintain adherence. So you've established that the patient has drug resistant epilepsy, they've failed one, then they failed to, they're still taking the medications, they haven't missed doses, they don't have provocation. What other things can we offer our patients? And so I'm gonna focus the rest of this discussion on um two of these things. One is on receptive surgeries, um both traditional receptive surgeries and more minimally invasive approaches. And the other is gonna be on neuromodulatory and neurostimulation strategies. When we assess the efficacy of a surgical intervention, there's um two scales that we use. One is the international league against epilepsy outcome scale. This is a very simple scale. Um and it's very good in the sense that it's simple, it just classifies outcomes based upon the burden of overall number of seizures and that, that's very easy to use. It's very easy to apply in clinical practice. One is complete seizure freedom, no auras, six is more than 100% increase in baseline seizure days. It's unlikely that any of our interventions worsen people seizures. It's probably just that the natural progression of their drug resistant epilepsy puts them in a position where seizures beget seizures and there's kindling and they get worse over time. Um But, but this is one scale that's used. The problem with this scale is it really lacks nuance and so not all seizures are created the same and seizures have the same impact on our patients. A focal aware seizure doesn't carry the same morbidity and mortality as a bilateral tonic clonic seizure, for example. And so getting rid of six bilateral tonic clonic seizures as a year. But having worsening of auras actually would probably make a very significant improvement in a patient's life by enabling them to be able to drive again and to resume many of their activities of daily living and reducing the risk of sudden unexpected death. But it wouldn't be captured by a scale like this. So more commonly we use the scales, the ingles uh classifications term after jerome Ingle is a neurosurgeon. Um and it's a lot more complicated as you can see. Uh but generally speaking, it's stratified in terms of whether or not seizures are disabling. So if they're free of disabling seizures, that's an Ingle one, if they have rare disabling seizures, that's an Ingle two, if they have some kind of worthwhile improvement in three. And if they don't have a worthwhile improvement. That's a four and we always strive for an angle one, but typically a one or two significantly increases a person's quality of life. Um And so those can both be considered good outcomes. So, this was another seminal paper published in the New England Journal of Medicine. Um In 2000, this was the the first class one evidence for epilepsy surgery, epilepsy, surgery been being practiced um for the prior century, but there wasn't any rigorous data to support its use. And so um in this particular study, there were 80 patients who were included um 40 in the medical group and 40 in the surgical group. And basically, people were randomized to surgery pretty much immediately. Um and this was temporal lobe epilepsy surgery or they were um put on a waitlist for surgery and continued on medications for the for a year essentially. And these are the outcomes. So this is a survival curve on the panel a for the percentage of patients um without seizures, impairing awareness. And you can see the findings are striking people in the medical group. Um about 8% of them achieved seizure remission without an intervention. And people's surgery approached 60% of that of that group in terms of having seizure freedom or I should say, um seizure freedom from disabling seizures. And so the number needed to treat for this was two, which obviously is very high just for a reference to have something to compare it to. And um the audience that I'm talking to might know a actually knows a lot more about this than I do. But for reference, looking at an old study in 2002, about the number needed to treat for aspirin for secondary um cardiovascular and thrombotic prevention. The number needed to treat was 67. Uh that was a paper publishing jam internal medicine. So this is a very, very powerful intervention, not something that we commonly see in medicine. Um Similarly, in panel b, the per looking at percentage without any seizures at all yet, about 40% of that of that surgical group achieve achieving total seizure freedom. In addition to seizure outcomes, we all the the authors also looked at quality of life metrics. And so there was a statistically significant difference in the quality of life as assessed by a quality of life inventory and epilepsy, which is a validated scale for assessing quality of life outcomes. There was a statistically significant difference between the two groups with uh patients undergoing surgery, having improved quality of life and that trended out for the entire year. And even though there was not a statistically significant difference um in the uh between the patients in terms of employment. Um and or being in school, there was a trend towards improvement in surgery which potentially would have continued if there had been longer surveillance. So all of these findings um suggest that surgery um re really improves um patients both from a health standpoint and a life standpoint. So, what does this look like? Um This is an an classic anterior temporal lobectomy. Um There's different margins depending on whether you're using a dominant temporal lobe or the non dominant temporal lobe dominant being the the side reserved for language function. The neurosurgeon will pursue a resection of a maximum of 6 to 656.5 centimeters of the non dominant temporal lobe and include the amygdala and the anterior hippocampus and the hippocampal body. Um and then in the dominant temporal lobe, they'll go to about 4 to 4.5 centimeters and this is what it looks like afterwards. Um It's pretty darn effective, but there are a lot of considerations with this um which are obvious, right? One, it's an open surgery, it has a long time to recovery. Um Both from an inhospitable standpoint. There's like a, you know, patients are typically here for three days after the surgery and then they're recovering at home for some time, they can't return to work right away. Um There's concern for cognitive deficits. We're um taking out a significant portion of brain, it's not eloquent cortex, but it is involved in um different measures of memory and uh language. And so, um there's that important consideration. Uh a lot of these um patients do have behavioral changes and that might be due to release of underlying vulnerabilities related to the resection. There's also the question of the functional reserve of the patient. Certainly a 25 year old can tolerate a temporal lobectomy. In most cases better than an 85 year old, both from a surgical standpoint and from a uh a network disruption standpoint. And then there's also the consideration of presence of a lesion. So if you have a lesion that you know is causing seizures, um or you strongly suspect is causing seizures, you need to take out all of this tissue in order to achieve a good outcome. And so a lot of people thought, well, let's try um a smaller procedure and see how that works. And that's the selective amygdala hippocampi toy. And so this classic selective Amygdala Hipp pampy uses a transcortical approach where patients will um basically have a a surgical incision through the lateral aspect of the temporal lobe. And then the surgeon will come in um from the side and take out the Amygdala and the hippocampus. There's several varieties on this approach. There's a Transylvanian approach that comes through the Sylvian fissure. There's an infra temporal approach and there's also variations on this transcortical theme. Each one of them has different technical skill requirements um and, and, and variables um complication profiles uh and different centers use different approaches. So this study um was done by J Nadal in 2018. This was a meta analysis which made an attempt to compare a TL to sh anterior temporal lobectomy, to selective amygdala hippocampal. Many people have attempted to study this but it's kind of difficult to make comparisons because of the variety of techniques and also heterogeneity in the tools used to appraise um neurocognitive outcomes. But looking at just the surgical outcomes itself. Um This study strongly suggests that there's not a a meaningful difference in outcomes from a seizure standpoint between A TL and selective amygdala hippocampi toy for patients with mesial temporal lobe epilepsy, that's specific to this patient population. Um Also not shown here as a funnel plot showing that there was no uh publication bias or no suggestion of publication bias looking at neuropsychological outcomes or neurocognitive outcomes. Um This particular study published a little over a decade ago now, uh by Tanner Verde um compared memory outcomes at one year follow up in patients with mesial temporal lobe epilepsy, but um specifically who had hippocampal sclerosis. So, um this is a very specific subset of patients with this condition. Um So all patients were treated with a TL in this case, they called it cortico amygdala hippocampal, but this is a TL and this is your selective amygdala hippocampal. Um And the tests on the top are um tests of verbal memory and a test on the bottom are tests of non verbal memory. OK. Um So they had uh two tests here and they stratified also by, by um hemisphere bilaterality of the intervention. So, on tests of the of Wexner memory scale. Um Both interventions caused significant decline from baseline, particularly with immediate recall. Um But as you can see, the emp lobectomy was worse. Um And then on Ray's auditory verbal learning test, there was a significant impairment in the total score, immediate and delayed recall as well as recognition um with the, with uh the selective Amygdala Hipp camp toy on the left, whereas only the total score and delayed recall was impaired with a TL on the left. You don't see any statistically significant differences on the right. Uh On test of verbal, I'm sorry, on test of non verbal memory, excuse me. Um such as simple geometric drawings, there were statistically significant improvements with the anterior temporal lobectomy on the left that were not seen on the right. However, on this alternative test, um the ray Oster complex figure test, uh the selective resulted in significant improvements that were not seen with a TL. So it's kind of difficult to know what to do with this information. There's a variety of reasons why the information arose this way. One consideration is that these tests are simply not the appropriate indices for measuring um the the the the translational neurocognitive effects of these interventions. Um That's one thought. Um Another alternative explanation um potentially is that uh this is just something specific to patients who have hippocampal sclerosis because that was the specific type of people with measle temporal lobe epilepsy. Um who are being studied in this study design. And then um also there, like I said before, there are different techniques for doing selectives. And this was a hetero you know, a heterogeneous consortium of different selective procedures that were involved. So that kind of complicates our understanding of. Um if, if, if that played a role in the development of these outcomes, I think suffice to say, you know, selective is not conclusively better than a TL for neurocognitive outcomes in these particular patients. Um But the jury is still out. So one other thing we can do is laser ablation or laser interstitial thermal therapy. And that's typically what we do here at our center. And so this is a more minimally invasive procedure, but it attempts to capture what a selective does um with less burden to the patient. So you don't have to go in and open the skull in order to get the tissue out. Basically, the, the lit takes light energy emitted by the laser and converts it into thermal energy in the surrounding tissue, thereby destroying the volume of tissue um in a in a relationship that correlates with the amount of power of the laser that's applied. Um It's perform in the or it's guided by MRI thermometry, it's performed by the neurosurgeon. Uh and oftentimes it can be accomplished in one pass sometimes because of the trajectory of the laser, it requires two passes that can often be done in one setting. Um Usually it's only in one setting. Um here in this image, you can see the trajectory of the probe and how it burns the tissue here. And it does the, the same thing in principle that the selective would do without having to remove the skull. There's two FDA approved systems for this in the United States. One involves a 12 watt neodymium doped atrium aluminum Garnet laser. The other is a 15 watt diode laser cooled with saline. There's no known difference between these two symptoms. It's just a matter of the uh preferences of the institution. Um and, and, and the neurosurgeons experience. And so, uh this has been around for a long time in the context of other conditions. Um We use it to burn um tumors, um both both malignant and benign, certainly get after getting biopsy for malignant ones. Um and there's a variety of other uses for it. But um over the past decade or so, it's really uh picked up in its use for epilepsy. Several series in the literature reported successful outcomes with laser ablation with varying degrees of efficacy. It's, it's usually been estimated to achieve 50 to 60% of seizure freedom. But there were no rigorous studies really examining this. There is a um a prospective trial um called slate that's going to look at this, but the data is not yet available. It hasn't been published that being said there was a study, a multi center study um published by Brett Youngerman and colleagues, looking at 268 consecutively treated patients across 11 centers um with, with lit and 50% of patients had durable seizure freedom. At last follow up, consistent with what had been seen in case series. And about one in seven patients had rare disabling seizures at last follow up. Uh And among patients uh with seizure recursions, two thirds became seizure free after subsequent A TL and about 50% actually became seizure free with repeat passes with the laser. So if the, if the physician had a very strong hypothesis that we were in the right area and there was a significant improvement in the seizure burden, but it was felt on the imaging, for example, that maybe the procedure didn't capture all of the tissue that um that the, that they wanted to target, they could take the patient back to the operating room at some point in the future and complete the procedure and have very high rates of seizure freedom with uh another go adverse events with laser are pretty rare. Um The most common one is a visual field deficit, but that's still quite uncommon. Um And much less commonly, it carries the risk of any other neurosurgical procedure. So, um uh in very, very rare cases, um if the there's a problem with the trajectory, it could cause a hemorrhage, that's exceedingly rare. I've only seen that once. Um the most common side effect is an a analgesic responsive um, headache, um, which goes away over time. And patients really like this procedure, you don't have to shave their head and they, their skull doesn't come off and it's, and it's pretty effective. And the other important thing about it is that, um, it doesn't obviate the, um or I should say it doesn't preclude the opportunity to pursue a wider and more definitive intervention in the future. Um So there's this other tool called radio frequency ablation and this has been around for a while. Um but isn't as using co commonly in practice, we do use it here. Um Early series of this technique showed about 16% of patients experiencing seizure freedom with about half experiencing seizure reduction. But as people have become more comfortable with the technique, uh we've seen an improvement in overall seizure um freedom with this particular technique. What this is is um you take an electrode and you put it in the target that you want to ablate and you provide a medium frequency alternating current in such a way that it heats up the uh the surrounding tissue. And basically you burn the tissue with that current. Uh And the way you know that you've successively burn the tissue is that the electrode impedances will skyrocket at the point of successful ablation. And the patient will typically report hearing a crack or pop at that point. And then, you know, you can turn it off. Um later series with this technique have shown that um if you have really good targets, um you can get and you and you have really well selected patients, you can have seizure freedom rates as high as 76%. Um So, in, in that particular study, um there was ablation of the enter Rhino and perirenal cortices which are neocortical areas that are immediately adjacent to the amygdala and the hippocampus. But it, it's important to note that it wasn't overall volume, the overall volume of ablation that um contributed to the good seizure outcomes. It was really knowing what the targets were. And so, um I should emphasize that we don't want to burn anything and everything. We just want to have a very good, we want to test a very good hypothesis. And so if you feel like you're in the right area based upon a concordance of the eeg data, the semi technology um and the structural imaging then um it's a reasonable thing to do this larger series um conducted by the group at Emory looked retrospectively at 57 patients who underwent radio frequency ablation via stereo eeg uh and 30% of patients in this particular series with this group who has a lot of experience with the radio frequency ablation um achieve seizure freedom on the first pass with R FA which is higher than what had been traditionally reported. Um And again, this did not preclude the opportunity to go for delayed surgery. Um The other 70% of these patients went for a secondary surgical procedure. Um 26 of those were laser ablations. Uh And you can see that there was a pretty high success rate with that. So um almost 80% had a favorable outcome based upon Ingle one or Ingle two classification. That's, that's pretty high. And this is an example of stereo eg this is something else we do at our center. Um I I in my personal practice, radio frequency ablation is, is a wonderful by product of an opportunity more than a technique that we typically use as a way to control seizures. So oftentimes patients come to us with epilepsy um without a clear lesion on their MRI or their pet and they have a seizure semi technology that suggests that the seizure is coming from a specific place in their brain. But we can't say with certainty. And what we do is a surgical exploration of the hypothesized epileptic network. So based upon all of the foundational data, the neuropsych testing the pet, the MRI, the eeg um and any ancillary test that we might perform in anticipation like a spect or electrical source imaging. Um We formulate hypothesis about where the seizure starts and where it spreads. And then we test that hypothesis. Historically, in the United States, this was done with grids and strips. Um But this is a less invasive procedure you can see um it's planted, we plan it in events and we put it in a variety of different targets um that look at not only where it's starting but where we think it might be spreading to understand the electro anatomical clinical correlations. And if we have tested the hypothesis and it comes back um accurately. In other words, our hypothesis is correct, we found the seizures exactly where we thought they were based on the data. We can radio frequency ablate the target with the electrode that's already in the place that we're looking for. And so this does two things for us. One, it gives us an opportunity to treat the patient but also gives us an opportunity to further test the hypothesis because if we can perform the radio frequency ablation in the right area and the patient has an improvement in their seizure freedom. It can act as a bridge to a more definitive intervention somewhere down the road like a laser ablation or a traditional resection if they do indeed relapse. So, there are other considerations. Um again, with these techniques, one is the location of the seizure onset zone. Is it an eloquent cortex? Is this a part of the brain that I can and want to destroy or is it a part of the brain that subserve a fundamental function um that a patient cannot do without, I don't wanna abl calque cortex and make a patient blind. I don't want to um destroy motor cortex and leave them. He metic don't want to destroy language cortex, et cetera, et cetera. Um also vascular supplies. So some parts of the brain are um are more rich in terms of their vascular supply. The insula is a primary example and it could be very hard to resect certain parts of the insular or even burn them because of all of the vessels around them um because you can cause a devastating stroke. Um There's also considerations for the functional connectivity of the brain. So certain parts of the brain are um linked, they are major hubs to everywhere else in the brain. Um A good example is of this is the precuneus, which is in the parietal lobe, but has connections to occipital cortex and parietal cortex, frontal cortex, temporal cortex. Um and in injury to the precuneus can be devastating for a patient because it can cause a very significant disconnection syndrome. Um There's also um thought about the functional reserve of the patient, their cognitive reserve, their, their independence in a patient who's not walking at baseline, um who's intellectually disabled. I'm less worried about them taking a hit, they're already not functioning, but in someone who is a physician or a lawyer or an engineer or um some, someone who juggles a lot of responsibilities. Uh One of these procedures, um particularly a large receptive procedure could have a very significant impact on their ability to carry out their functions. And then finally, there's also epilepsies that are vocal, there is generalized epilepsies or combined epilepsies um in which the seizures start appearing everywhere in the brain all at once. And certainly we can't resect the entirety of the brain. So we have to have other interventions beyond these. That's where neurostimulation comes in. So there are three FDA approved neurostimulation devices and we're gonna talk a little bit about each of them. This diagram is very funny to me because um I hope to never see a patient who has all three of these devices implanted at once. Um But I do like the schematic as, as an, as an instructional cartoon as to um which devices exist and kind of where they go. So uh the vagus nerve stimulator was the um the earliest FDA uh electrostimulation device for epilepsy. It was first implanted in 1988 but it wasn't FDA approved until 1997. There have been several clinical trials looking at VNS. Basically what it is is there's an implantable pulse generator that goes into the chest, it's connected to an extension or lead that then um basically gets situated on the vagus nerve. Um And no one really understands um all of the nuances on the mechanisms of how VNS exerts its benefit. We just know that it does. Um There is a hypothesis that there is a Vegas network and there are broad connections um um starting um in the brain stem that then produce to a specific subcortical structures such as the thalamus and, and in addition to sub cortex cortical targets um that create the effect we see in seizure reduction, but it's really not known how it exerts its benefit. Um The f the first device was this one here. Um The NC PM 100 it was a dual pin open loop generator and it had a wider range of stimulation parameters. Um But it was considered a pioneering device for people who previously did not have um another intervention available to them. Um It also featured a read switch to allow on demand stimulation. If you swipe the therapy's block or horseshoe magnet over the generator. Uh Then with this uh pulse M 102, the system changed from a dual pin to a polarized single pin system. And the relevance of that is there were less, there was less potential of lead communication issues with the generator. So that was a, a big improvement and there was further innovation to the, the generators design um because the Demi pulse was extremely small. But finally, the actual, the, the biggest change since that time was this aspire SRM 106 model. Uh Because in this particular model, they um instituted AAA type of closed loop uh feedback system um called auto simulation where the VNS not only delivered therapy but detected changes in heart rate. And for people who have icy tachycardia or increases at heart rate, um associated with their seizures. This particular feedback mechanism was valuable because the device would automatically deliver additional stimulation at the time of seizures. Uh In addition to its regular cyclical stimulation cycle, uh and you can set that any number of different ways and we'll, we'll talk about that. So these are the results from the VNS study group. Um And because you can't blind patients to VNS, they feel when they're getting stimulated, there's no way to have a sham group. So what they did with the way they designed it was they had a low stimulation group and a high stimulation group. And here are the, here are the parameters for those groups. And here are the um the baseline characteristics of the study groups. And looking here you see that first that the mean percent change in seizure frequency was more significant in the high stimulation group relative to the low stimulation group with a difference of approximately 18% between the groups um in in the mean seizure percent reduction. Secondly, we also see that the high stimulation group exhibited a statistically significant change in the median number of seizures when comparing baseline seizure frequency to frequency at 12 weeks. Whereas we saw no change in the median number of seizures at 12 weeks in the control group, the low stimulation group. So there is a modest effect. So how does this work? Um Typically the device will be in its normal mode, it will deliver stimulation it cycles, um, it'll go off for a minute and then on for a minute and then off for five minutes. But you could set it for any duty cycle that you want. You could do 30 seconds in two minutes. Um, and, and it just delivers that stimulation in regular cycles. Um If you swipe the magnet over the device, it will automatically go into a, a magnet mode and then it will deliver simulation again at pre specified settings, whatever you choose, typically, that's a current that's slightly higher than um whatever is in normal mode. Um And I would emphasize that the there is a therapeutic um window for a current and, and so it's not that like all currents are equal. As you saw with the low stimulation, the high stimulation, we want to get people to a higher stimulation as high stimulation as they can tolerate. Because real world data has shown us that higher stimulations I improve outcome. Um And here's how the auto stimulation works. So, um during the normal regular cyclical stimulation cycle or during a magnet, the device is detecting any changes in the baseline heart rate, um it basically counts the heart rate for a certain epic and then it will um count it again over another epic. Um And you set it for a threshold. So it's not like it's always gonna go off at 100 20 beats per minute. It's going to set the new baseline at 100 20 then you will set it at 30% of that or 40% of that or 50% of that. And then if it crosses that. So for example, using easy numbers, if the heart rate is 80 I said it for 50% threshold, then it will go into auto stem mode. Um if the patient jumps to 100 20. Um so it's not like they're always going off. And so that's how that works. OK. What about um deep brain simulation? So this is a newer technology um And it's been being used for Parkinson's and essential tremor. But more recently, it's been um adapted into the epilepsy domain. Um And for good reason. So similar to the VNS, the D BS has an implantable pulse generator, it has an extension lead. But instead of attaching the vagus nerve stimulator, it goes into the brain and then it attaches to these electrodes that are situated into the thalamus. And classically, it's FDA approved designation is for the anterior nucleus of the thalamus. It's a very specific target that requires um very specialized MRI sequences. Um And this is the the interface for that device. So here's what the generator looks like, the patient gets an application on their phone and this is what the interface looks like for the programming physician. So, what's the evidence for this? So, in 2010, um there was um a randomized control trial um called the Sante trial, it was a prospective multi center double blind randomized trial. Um looking at 100 10 adults with severe drug resistant focal epilepsy. Uh and the implantation procedure was standardized across centers. Half received stimulation and half um had it implanted but received no stimulation during a three month blinded phase. And one month after stimulation, patients were either randomized to the stimulation at five volts, 100 45 pulses per second or it wasn't turned on, they would cycle again at one minute on five minutes off. Um And the and the investigators were not allowed to change any medications during the three month blinded phase or the nine month unblinded phase. And so what you see here is initially after implantation, both patients had a reduction in their seizures. But when stimulation was turned on, the patients who underwent stimulation continued to have a decrease in their um median percent seizure frequency relative to baseline. Whereas this was not sustained for the control group. Ok. This is a histogram looking at the seizure frequency changes from baseline to 25 months of stimulation. So this is every patient who was enrolled. And you can see that um if you're looking at seizure responders, seizure responders are people who arbitrarily have been defined as people who have at least a 50% reduction in the uh frequency of seizures. So, um everyone at who is here and above is considered a seizure responder. You have a little bit over 50% who, who are responders, but much closer to 90% of patients had some improvement with implantation of the device. Whereas a very small segment of people got worse. And again, I'd like to emphasize that this is the natural course of drug resistant epilepsy. This is what I would expect if patients didn't have any intervention. So I don't think that it was the device that was causing these problems. I think this is just the natural course of their disease. Um Looking at adverse events in the clinical trial, there was concern that um D BS caused um some memory impairment and depression. However, the investigators didn't control for who had these um pre uh pre pretrial comorbid conditions. And so actually, there were differences um in, in between the groups and who had these conditions prior to randomization. What we've learned longitudinally is that these difference that were seen during the controlled phase just haven't, didn't pan out. So we don't typically counsel patients that um D BS causes worsening, depression or memory impairment. On the contrary, the idea that their seizure um that their seizures have improved and that they have increased. These reduction often um seems to help with their mood symptoms. Um Neuropsychological test, composite scores which aren't showed here um also showed statistic statistically significant gains from baseline to five years in um in all domains. So, attention, executive function um and um and subjective cognitive function we also have longitudinal data. Now, um longitudinal outcome data at seven years and 10 years from the original sane participants. All of this is open label at this juncture. Um But uh we've learned a couple of things. One is that these patients continue to demonstrate a gradual and sustained improvement in seizure reduction over time. Uh And you can see it two years, uh It's about 60% at, you could say it's at least 50% at four years. It's at least 60% and at six years, it's at least 70%. So that's kind of how I counsel patients that those are, those are the expectations with this device. Um And sub analysis by type of epilepsy, if you look down here showed that patients with temporal lobe seizures and frontal lobe seizures stand to gain the most benefit patients with other types of sei other types of um distributions for icy onset seem to do better and then um did not sustain for reasons unclear. Fortunately, most patients have frontal lobe epilepsy and temporal lobe epilepsy. And it's important for me to point out that this could be a product of the stimulation target. So again, we're targeting the anterior nucleus of the thalamus, which is involved in peep circuit and has extensive connections to the frontal lobe and the temporal lobe and not as many connections to the um the parietal lobe and the occipital lobe. And so this just may be a product of what we're targeting and other types of epilepsies require different targets. Another really important thing is that the overall mortality and so up rates reported in the study, um which respectively were 6.9 and two deaths per 1000 person years um are much better than what we see in a classic drug resistant epilepsy population where suit up rates are typically 7 to 9 per 1000 patient years. So the data suggests that D BS may have an impact on the risk of a sudden unexpected death in epilepsy. And last but not least, the newest player of the game, responsive neurostimulation. So um responsive neuro stimulation is really unique, not in that, it not only offers a therapy for patients. Um but it also offers real time monitoring um of patients brain waves while they're on medications out in the real world. So this is called an um electrocorticography epoch or E cog for short. Um And what it does is it looks at the pe person's brain waves over time and the device does not capture all of the information because it simply doesn't have that storage capacity. But what it does capture is whatever you tell it to capture. And so um there is a detection algorithm that you set up based upon your patient. Um And there's a number of different features you use that um to, to, to set up those detections, you can use area under the curve if you're interested in the amplitudes of these waves, you can set up line length detection. So changes in the, in the length of these underlying waves. Um and you can use a, a band pass filter. So if there's a change in the frequency of the waves, and so again, the RNS has an implantable pulse generator, except this goes under the skull, it's connected to two extensions and um those extensions then can go to a depth lead where there's electrodes on the depth of lead or they can go to a cortical strip. So, um for all intents and purposes, a hypothetical um, one dimension strip that lies on the convexity of the brain. So you can put it on the surface of the brain or have it crawl around the, the brain, for example, here in the temporal lobe. And then what happens is the patient uploads the data every day. So they send it to a data monitoring system. Um And it goes to a, a central server and then the physician can review it and the engineer can review it. And so in addition to setting up all of these detections, the device will de deliver therapies again based upon how you choose to program it. And so there's lots of options, the device stores up to 30.5 channel minutes. Um And that's, you can select 30 60 90 100 80 or 242nd windows. It'll capture two thirds like right before the detection. And then one third, after the detection, with the idea being, you want to see what the kind of seizure onset looks like. Um But it's, there's a lot of flexibility to do what you want with it. So these are the various stimulation settings that are available on the device. We typically monitor charge density, which is a measure of charge over area measured in micro cools per centimeter squared. And we increase at at regular intervals by about 0.5. Um not to get um um too much in the weeds with all of this. But there's different stimulation strategies that we choose based upon the type of setup we're looking for. And the type of setup we're looking for is dependent on the type of epilepsy um that the patient has. So just for example, a very common scenario is a patient who has bilateral mesial temporal lobe epilepsy. So they have seizures, not just coming from the left hippocampus and not just coming from the right hippocampus, they have it coming from both. Certainly, we can't resect both hippocampi. So what we can do is we can put electrodes into both hippocampi and we can set them up to stimulate both hippocampi. Um And in addition to doing so, we can detect seizures and make a determination about the overall burden of seizures coming from each um each hippocampus. And that's actually super interesting because what we've begun to learn is that um what we see in an epilepsy monitoring unit isn't always the same thing as what's going on out in the real world, particularly when a patient's on medications. And so we might find out that they have a ratio of seizures of 7 to 1 or 8 to 1 or 9 to 1. And actually, if all of their seizures are, or I should say, 90% of their seizures are coming from the left hippocampus and not the right hippocampus that actually might make them a candidate for a more traditional surgical approach in the future. And so it actually opened the doors to alternative um longitudinal strategies for patients that they didn't previously have while offering them a therapy during the interval period. So this was the pivotal RNS trial. Um It was done um In 2014, they looked at 100 91 subjects with intractable focal seizures from one or two foci uh And what they saw at least in the blinded phase was a percent change in seizures of 38% in the stimulated arm and 17% in the sham stimulation group. Um But during the open label period, um you can see that the the median percent reduction seizures continues to go up and that's a product of two things. One is that um you let time do it have its course, right? So more neurostimulation over time leads to um putative plastic changes that um help with seizure burden. Uh But also um we've gotten better at knowing what kinds of stimulation parameters we want for different types of epilepsies. And we can change our stimulation parameters on a patient by patient basis to target a AAA more um precision oriented approach for each patient. This is again, is the individual histogram for um all of the patients in the pivotal trial. And you can see that about um 4/5 here uh had some kind of improvement. And if you're looking here, about half are considered responders. And again, uh we also have longitudinal data uh for these patients um at seven years. Uh I'm sorry, over nine years. So um each one of these uh bars represents a different um uh a different group. Uh One was um the constant cohort population. One was the last observation carried forward population um And one was the 91 minimum diary. So this is the 91 minimum dia diary. This is the constant cohort and this is the last observation carried forward. And you can see over time um there is an increase in the median percentage change in seizure frequency. Um And this is not the natural course of drug resistant epilepsy. Typically, seizures get worse over time. So, um this would suggest that the RNS is having an impact in conjunction with changes um to seizure medications over time. Um um In panel B, um approximately three out of five patients are responders and in panel c um the percentage of patients with um meso temporal lobe epilepsy, nonmeal lobe epilepsy, achieve seizure freedom for a period of three months, six months in a year shown here. OK. Um very quickly in the interest of time future directions. So, um there are a lot of interesting directions um uh both from a clinical research and a translational research standpoint. So um the fact that we're able to monitor all these patients in real time has led to new and interesting discoveries. One is that um seizures um believe it or not um in many patients have cyclical patterns. So they might have uh circadian patterns, modan patterns or circ annual patterns um that we're now parsing out because we have chronic e cog recordings of patients um and the ability to figure out what these cyclical patterns look like, offer promise of um seizure forecasting in the future. Um with. So with a probability distribution that fits each patient individually, um We're also looking at local field potentials with brain sense technology. In the deep brain stimulation, we could see what local field potentials look like around the time of seizures. And that may lead to opportunities to um target specific programming um strategies for patients. Um Another thing um that's really taken off is a change in pharmacotherapy trials um and specifically novel trial designs where we do time to event um clinical trials. So instead of having a patient sit in the placebo arm and not get any benefit. If a patient fails in the placebo arm, they can then um they can then switch over to the treatment arm. Um And this offers the opportunity to um uh help patients um get an experimental treatment more quickly. And then as I alluded to, there are new targets um being explored, um we don't have randomized controlled trial data for that, but we do have case series um using some of these technologies in central media nucleus and pulvinar nucleus. Um but particularly in Linux gusto patients for central media nucleus. Um And so there's opportunities to leverage these technologies to other types of epilepsy syndromes where people didn't um have any options previously and then from a translational research standpoint, um There are uh there's a prospect of precision medicine approaches. So, gene editing technologies, we're really good about knowing um which medications not to give in certain um epilepsy conditions. Um But we don't know, don't know yet if certain epilepsies would uh respond more favorably to specific combinations of uh seizure medications. Uh So, um that's an exciting avenue. Um There is a trial going on right now. Um looking at interneuron transplantation. So, basically uh plugging um inhibitory interneurons um into the brain and seeing how they reduce seizures over time, which is really exciting. Um And then uh optogenetics. So, basically using CRISPR technology um to insert um light sensitive proteins called options um into, into different neurons and then um turning the lights on to activate um certain interneurons and study brain networks that way to determine a causal effect between um specific networks and seizure development. Uh Very, very quickly. We are um having um a symposium um that's specifically targeted for primary care for physicians as well as general neurologists and advanced practice providers in the community. It's going to be on November 8th. There's going to be a panel of speakers. We have a, a keynote speaker coming in from the University of Pittsburgh. Uh Professor Alwa is gonna talk about electro electrical stimulation, epilepsy, but we'll also have the former president of the American Epilepsy Society giving talks on anti seizure medications. Um uh Professor Hogan, uh and we'll have um a variety of other amazing speakers. Um And so if you guys are interested, um we'd love to have you love to introduce you um and meet you all in person. Created by Presenters Adam S. Greenblatt, MD Assistant Professor, Department of Neurology, Division of Epilepsy View full profile
