Temporal resection

Temporal lobe epilepsy (TLE) occurs in 10%–20% of pediatric cases . In 2017, data from the RCT reported that 14/14 children who underwent temporal lobe resection in their study attained seizure freedom. Metaanalyses of pediatric temporal resections yield consistent results, with overall seizure freedom rates of 73%–74% .

Etiologies contributing to TLE include those of lesional (tumor, FCD) or nonlesional (MTS) origin. Less common etiologies from which TLE may arise are vascular malformations and infection. Lesional substrates of FCD and tumor are the most predominant. Tumor resections for TLE correspond to the greatest seizure freedom rates, reported at 80%–84% Engel I postop . The results are less favorable for FCD at 59%–64% , although the most recent metaanalysis of pediatric TLE found seizure freedom in 79% of children with temporal cortical dysplasia . Gross-total resection in lesional cases is often necessary for optimal seizure outcomes, as the dysplastic tissue is epileptogenic . The consequences of subtotal resection are pronounced, with metaanalysis data reporting 94% versus 38% seizure freedom after gross-total and subtotal resection, respectively .

MTS, while less frequent in children, still represents a significant portion of pediatric epilepsy cases. This etiology is quite amenable to resection and, along with tumors, corresponds to the highest probability of seizure freedom at 78%–85% . Unilateral MTS corresponds to a favorable prognosis whereas bilateral MTS relates worse outcomes. MTS often exists simultaneously with additional pathology/lesion, known as “dual pathology.” In children, the incidence of dual pathology is much higher than in adults . Interestingly, the dual pathology does not correspond to a poorer prognosis compared to MTS alone . In fact, it has been associated with the best seizure outcomes in pediatric epilepsy with 89% achieving Engel I outcomes .

Temporal lobectomy can broadly be divided into anterior (or anteromesial) temporal lobectomy (ATL) and selective amygdalohippocampectomy (SAH). While metaanalyses comparing the two techniques exist for adult cohorts, there is no level 1 evidence for the pediatric population. Furthermore, studies particularly reporting on seizure outcomes in SAH are varied, with seizure freedom rates ranging from 44% to 88% at the last follow-up . Comparative retrospective studies have reported statistically significant improvement in seizure control after ATL . A 2004 study of 89 consecutive children reported 94% versus 77% seizure freedom after ATL and SAH, respectively . While a recent retrospective study reported seizure freedom of 77% versus 44% after ATL and SAH for unilateral TLE . The same study also reported reoperation as significantly more likely after SAH . However, this too is not ubiquitous as a large retrospective series showed nonsignificant difference in seizure freedom for pediatric TLE (73.7% ATL, 76.9% SAH) . It is important to note that ATL cohorts in all studies are considerably larger than SAH. Given the variation of seizure outcomes after surgery it is difficult to draw a firm conclusion regarding the techniques. Ultimately the choice of technique will be determined by institutional experience and individualized presurgical workup.

Complications after temporal resection in children are infrequent and tend to resolve on their own. Looking at data from the RCT, 2/14 patients who achieved seizure freedom after temporal lobectomy reported monoparesis, although these symptoms improved after 12 months . A systematic review of complications after epilepsy surgery found temporary hemiparesis to occur more frequently in pediatric age groups . Two systematic reviews including pediatric patients who underwent temporal lobectomy report an overall morbidity of 5%–16% in children . The most frequently encountered complications included infection, CSF leak, psychiatric complications, and vascular complications. Minor and major (requiring intervention, or not resolving within 3 months) complications have been found to be more common in children, and psychiatric comorbidities occur primarily after temporal resections . The most common consequence of temporal resection is superior quadrantanopia due to the proximity of the resection to Meyer’s loop .

Extratemporal resection

While temporal epilepsy is the most common form in adults, pediatric cases are predominately extratemporal lobe epilepsy (ETLE) . Seizure freedom rates are consistently lower for ETLE than TLE due to the inherent difficulties of the disease. These include a higher proportion of nonlesional and MRI-negative cases, rapid ictal propagation across a wide area of cortex, and a diffuse EZ potentially containing functional and eloquent areas unamenable to resection.

The hunt for the putative resection zone is more elusive in ETLE and thus necessitates an exhaustive use of noninvasive and invasive monitoring techniques. Two-staged procedures using invasive monitoring and subsequent resection have been employed to great success, with up to 60% of patients achieving seizure freedom in extratemporal cases . The use of sEEG, in tandem with a thorough presurgical hypothesis, may mitigate the “inherent” disadvantage usually tied to extratemporal cases . Notably, metaanalysis of nonlesional extratemporal pediatric epilepsy found no association between intracranial monitoring and favorable seizure outcomes . However, the analysis did not make a distinction between specific iEEG modalities (sEEG vs SDE).

A 2013 metaanalyses of pediatric epilepsy surgery reports an overall Engel I outcome in 56% of children undergoing extratemporal resection versus 73% in temporal resection . A more recent metaanalysis in 2020 reports an overall seizure freedom rate of 60% , while other systematic reviews report rates of 34%–70% depending on etiology and the affected lobe . The recent RCT results found considerably higher success with 11/12 patients being seizure-free after extratemporal resection . Factors associated with better seizure outcomes are lack of generalized seizures, earlier age at intervention, lesional epilepsy and extent of lesional resection, and localizing ictal findings . While most of these factors deal with determining a discrete EZ, the presence of MRI abnormalities has not been shown to affect seizure outcomes to statistical significance . However, given that pathology and associated MRI findings are strong predictors of favorable outcomes in most cases, it is likely that this lack of significance arises from underpowered studies .

Frontal lobe epilepsy is the most common type of ETLE in children and is often characterized by having the worst seizure outcomes amongst ETLE resections. Seizure freedom is achieved in approximately 45% of patients postoperatively . A focal versus an extensive resection confers a greater chance of seizure freedom as does gross-total resection of an identified lesion . Caution must be made to spare as much tissue as possible in frontal resections, as closer resection to the precentral and cingulate sulcus has been shown to significantly affect the neurologic deficit. The Rolandic area is of special consideration in pediatric epilepsy. Benign Rolandic epilepsy is a common childhood syndrome and can evolve into a malignant form refractory to medical treatment . Retrospective studies show seizure freedom in 43%–65% after resection of the nonfunctional cortex. Patients are to be more likely in rolandic resection rather than perirolandic (65% vs 45% seizure-free) . Given the functional significance of the Rolandic area, neurologic deficits are common. Permanent sensory or motor deficits have been reported in 28%–43% of selected cohorts, with persistent morbidity progressing to a milder form with time .

Posterior cortex seizures involve temporal, parietal, and occipital lobes, sometimes referred to as the posterior quadrant when containing a single hemisphere. Seizure onset tends to occur before the first year of life and a large proportion of patients will report field defects corresponding to the lesion. A complete temporo-parieto-occipital resection is favored if eloquent areas may be spared, as this is a uniform predictor of better seizure outcomes . However, the resection very often leads to persistent visual field defects like hemianopsia and quadrantanopsia. There is tremendous variation regarding seizure outcome depending on the series, but seizure freedom is reasonably attainable, with a published series reporting upwards of 90% Engel I status postoperatively . Among multilobar resections, resections involving the occipital lobe (temporo-occipital, parieto-occipital, and temporo-parieto-occipital) yielded the most favorable outcomes . Data on seizure outcomes for distinctly parietal and occipital epilepsy in children is modest, but seizure freedom rates in the most recent metaanalysis were observed at 64.1% and 71.1%, respectively .

Hemispheric surgery

Hemispheric surgery is indicated for extensive hemispheric pathology, namely multilobar cortical dysplasia, hemimegalencephaly, hemiconvulsion-hemiplegia epilepsy (HHE) syndrome, Sturge-Weber Syndrome, and Rasmussen’s encephalitis. Acquired etiologies such as trauma, infection, and hemorrhage may also lead to hemispheric epilepsy requiring hemispheric resection. Hemispheric surgery is considerably more likely to be undertaken in children and is one of the most common procedures performed in pediatric epilepsy .

Four metaanalyses examining hemispheric surgery in children found an overall seizure freedom rate of 71.0%–74.7% . RCT findings reported 13/15 patients (87%) having received hemispherectomy to be seizure-free at the conclusion of their study . Hemispheric surgery is often regarded as the most effective procedure in conferring seizure freedom. Metaanalyses comparing multiple procedures in pediatric epilepsy have found hemispheric procedures to achieve greater rates of seizure freedom than temporal or extratemporal procedures . The effectiveness of surgery is influenced by the underlying etiology. A metaanalysis of pediatric hemispherectomy observed 82.8% seizure freedom in Rasmussen’s encephalitis and Sturge-Weber syndrome, but only 61.4% in developmental etiologies such as cortical dysplasia, or hemimegalencephaly . Rasmussen’s encephalitis and Sturge-Weber, along with acquired etiologies such as infection, hemorrhagic events, and HHE, consistently demonstrate better seizure control after hemispherectomy . Historically, most studies have not identified a significant difference between hemispheric techniques in terms of achieving seizure freedom . However, recent evidence suggests that the vertical approach might offer more durable seizure freedom compared to the lateral approach . Yet, this finding was not corroborated in a recent study .

There is evidence to suggest that long-term cessation of seizures may be more attainable in hemispheric surgery than other procedures . A systematic review and metaanalysis of long-term outcomes showed 71% seizure freedom after minimum 5 year follow-up period . Furthermore, the rate of acute postoperative seizures, which independently predict seizure recurrence and poor long-term prognosis, is markedly lower for hemispherectomy than other resections . Preservation of seizure freedom is a desirable advantage unique to extreme surgery, as the same cannot be said for other procedures. Seizure freedom for other procedures diminishes significantly over time . The potential for long-term seizure control, particularly since children referred for the procedure face severe symptoms from catastrophic seizures, should be considered by the treatment team.

Hydrocephalus is the main complication following hemispheric surgery, and approximately 14% of pediatric patients require CSF diversion postoperatively . Other complications that may be encountered are blood loss, infection, and sinus thrombosis . Postoperative hemiparesis is an expected adverse event in patients after hemispheric surgery, though some patients report improvement from their preoperative hemiparesis . Severe complications such as cerebral edema and brainstem shift have also been reported, although these are quite rare.

While all hemispheric procedures are viable, recent series show increased usage of hemispherotomy. The impulse toward hemispherotomy is likely motivated by its superior safety profile. The CSF diversion rate after hemispherotomy is on average 8.3% . Hydrocephalus requiring intervention after anatomical hemispherectomy may be as high as 36.4% . Hemispherotomy is also associated with significantly less intraoperative blood loss, decreased operative time, and greater preservation of tissue integrity . Of note, mortality among all procedures is low, particularly in more recently published series, and does not differ between techniques.

Corpus callosotomy

CC is a palliative treatment for those with generalized or multifocal DRE and is most often employed to reduce injurious drop attacks. The average patient can have up to 500 seizures a day, with concomitant cognitive and developmental delay. Prognostic factors that are likely to confer a better surgical outcome include younger age at surgery (and seizure onset), seizure type (drop attacks) and epilepsy syndrome, slow spike-wave pattern on EEG, lessened synchronicity of postoperative discharges, normal MRI findings, and intelligence quotient >50 .

There has been a lack of consistent classification systems making adequate metaanalyses difficult in this population. On top of this, the assessment of outcomes both short and longer-term, is further complicated due to the severity of neuropsychiatric and neurodevelopmental comorbidities. Thus, previous outcomes studies in CC patients have been difficult to analyze as differing favorable outcome criteria have been used across studies.

Typically, a meaningful improvement is seen in 65%–85% of patients post-CC, with the largest effect being on drop attacks. Complete seizure freedom is seen in fewer than 10%–20% of patients, but the majority experience a significant decrease in drop attacks . The most recent metaanalysis saw seizure freedom rates of 18.8% with positive predictors including the history of infantile spasms, a normal MRI and short epilepsy duration . This same metaanalysis also found drop attack freedom rates of 55.3% with positive predictors including complete CC and idiopathic epilepsy . There remains deliberation when it comes to the extent of a CC. Complete CC has consistently shown better seizure reductions but erred upon due to resultant neurological sequelae. However, with evidence showing that most of these are transient, especially in the pediatric population, most prefer a complete CC on first inspection . Surgical complications comparing total versus anterior CC show no significant difference . Other systematic reviews have also looked to see the differences between CC versus vagus nerve stimulation (VNS). Although VNS is reversible and has fewer complications, it is more expensive and less likely to have a clinically significant reduction in drop attacks compared to CC .

Complication rates are generally low with adverse perioperative events seen in 2%–12% of cases. The most frequent complications include transient lower-extremity weakness, transient aphasia or mutism, infections and hemorrhage . Complications such as transient disconnection syndrome, wherein they are not able to process a unilaterally presented stimulus, are also significantly more likely seen in complete CC versus anterior CC .

Other outcomes may also be more important to the parent/caregiver than seizure reduction in these patients. Quality of life and behavioral outcomes are associated with seizure outcome . However, even with a limited reduction in seizures, parents may rate the outcome highly . Indeed, in one study, all parents reported that they were pleased with the outcome and would recommend the surgery to others .

Multiple subpial transections

Surgical resection is problematic when the seizure focus is in the eloquent cortex and although we now have more modern therapies such as neuromodulation (seen in a later section of this chapter), one of the original techniques for treating seizures in eloquent regions of the brain has been multiple subpial transections (MSTs). This technique severs intracortical fibers whilst sparing subcortical white matter and U-fibers. As such, this technique has been used in augmenting resections for lesions such as cortical dysplasia, tumors, posttraumatic and infectious epilepsy. It has also been utilized for patients with status epilepticus, Rasmussen’s encephalitis and particularly for Landau–Kleffner syndrome (LKS).

MST outcomes are difficult to assess due to the fact that most of these surgeries are performed in tandem with cortical resections. There is also a lack of studies with control groups, thus a direct comparison is challenging.

A recent systematic review on MST found seizure freedom achieved in 55.2% of those undergoing MST combined with resective surgery, and 23.9% of patients who had MST alone . Positive predictors of better seizure freedom rates generally include young age and smaller size of MST with foci in the temporal lobe . Complications are frequent with transient mono- and hemiparesis seen in 19.8%, transient dysphasia in 12.3%, permanent paresis in 6.6% and permanent dysphasia in 1.9% . Other complications include infections, hematoma, CSF leaks, and cerebral edema . Seizure recurrence is also common in this procedure. Thus, with the associated morbidity of MST, the neurosurgical community now considers responsive neurostimulation (RNS) instead of MST when there is an eloquent cortex involved . LKS has a long history associated with MST as it was thought to improve language function in these patients . However, it has been seen that patients with LKS are able to retain their language function better without surgery . Some studies have shown that MST can still improve behavioral status in these patients, nevertheless, evidence remains controversial .

Vagus nerve stimulation

VNS is the most studied form of neuromodulation and the only such procedure with evidence from a pediatric RCT . The 2012 double-blinded trial recruited 41 children with DRE that underwent either high or low-output stimulation. Data from the trial found no significant difference in the 50% response rate (RR) between the two stimulation parameters (High: 14%, Low: 20%) . At the conclusion of the trial, the 50% RR was 26%, while 88% did not respond to treatment, and 15% had a seizure increase of 50% or greater .

Metaanalyses of pediatric VNS data show better outcomes than were reported in the RCT. A 2021 metaanalysis found pooled prevalence for 50% RR to be 56.4% at the last follow-up, with 11.6% of patients achieving seizure freedom . Similarly, a study using the VNS therapy Patient Outcome Registry found 63% of patients responding to treatment with 8.2% seizure-free at the last follow-up .

Factors associated with increased seizure control in VNS vary based on series, including shorter duration of epilepsy and focal motor seizures . However, the aforementioned metaanalyses present interesting deviations from respective surgery that can be appreciated in studies with longer follow-ups. Across most studies, the response to VNS remains stable, even becoming more favorable over time . Responder rate escalated from 40%–49% within 6-month follow-up, to 56%–64% at the last follow-up greater than 2 years . Two metaanalyses reported that responder rate and seizure freedom are greater in those with later age of epilepsy onset, as is typically observed in other procedures .

Regarding complications, the overall rates from the five existing RCTs for VNS which contain both pediatric and adult patients include hoarseness (30%), dyspnea (13%), infection (12%), cough (7%), and throat pain (7%) .

Deep brain stimulation

The SANTE trial highlighted the benefits of the anterior nucleus of the thalamus (ANT) DBS in reducing seizure frequency for adults with DRE, while the ESTEL trial demonstrated the therapeutic potential of centromedian (CM) thalamic DBS in patients with Lennox-Gastaut syndrome . To date, however, no RCT has been conducted in children. Still, the success of these previous trials has paved the way for off-label usage in pediatric populations. The most recent metaanalysis of pediatric DBS patients included 72 patients, of whom 38 underwent centromedian thalamic (CMT) and 14 had ANT electrodes . Other targets included in the study were STN, hippocampus, mammillothalamic tract, posteromedial hypothalamus, caudal zona incerta, and combination electrodes of CMT and ANT. However, these data were considerably less robust. The overall responder rate at the last follow-up was 75%, with 8% achieving seizure freedom . A greater response to treatment was seen in patients with CMT, although this did not reach statistical significance . Other factors, such as MRI findings, etiology, or generalized seizures were not associated with a more favorable prognosis across studies . There is still evidence that DBS of CMT may be particularly effective in the treatment of generalized seizures . Most complications in pediatric DBS seem to occur due to infection .

Responsive neurostimulation

A 2011 multicentered double-blinded RCT found 42% and 9% seizure reduction in treatment and control arms, respectively, at 3 months of blinded evaluation . Further open-label extension studies have shown sustained efficacy with progressive improvements in seizure control over time—the 50% RR improved from 54% at 2 years, 56% at 5 years, to 73% at 9 years .

While only approved for 18 years and older, RNS has seen effective usage in pediatric populations. Two recent metaanalyses of RNS in children report an overall responder rate of 73%–80% at the last follow-up . The safety profile of RNS in children is strong, with no neurologic deficits or hematoma being reported across metaanalyses and recent case series . Infection is the most commonly reported complication, occurring in approximately 6%–11% of pediatric cohorts . Given the paucity of data, no factors related to better prognosis have been strongly presented in RNS studies , although a younger age at implantation may be more beneficial in conferring seizure reduction . Interestingly, recent metaanalysis and retrospective findings show that amongst pediatric patients with a 50% RR to RNS, more than half achieve seizure reduction of greater than 75% .