Controversies in the timing of pediatric epilepsy surgery: is earlier better?

Introduction

A widely accepted indication to consider neurosurgical intervention for drug-resistant epilepsy (DRE) is after an adequate trial of two antiseizure medications (ASMs) has not achieved seizure control. This guideline has been used in multiple clinical studies and randomized surgical trials . The two ASM thresholds are based on data demonstrating profoundly diminishing returns in seizure response with each subsequent medication . Rates of seizure control have not improved in recent decades despite modern drug development . The average rate of achieving seizure control when moving on to a third medication after adequate trials of two well-selected ASMs is low, therefore consideration for nonpharmacologic interventions at this threshold is a recommendation without controversy.

This guideline permits flexibility in interpreting several factors, such as what defines an adequate trial, if the specific ASM was an appropriate choice for the specific patient and seizure type, and if any seizure frequency less than complete seizure freedom is considered well controlled. For example, a medication may have been discontinued early prior to reaching a full dose escalation due to an adverse side effect and therefore may not be considered an adequate trial. Although the patient’s seizures were not controlled by that medication trial, because a maximal dose was not achieved it can be reasonably substituted for another ASM without necessarily expecting a lower probability of response. This manner of counting medication trials is consistent with the cohort studies that defined seizure response by only considering the first “well-tolerated” medication . Despite such allowances in medication trials, some patients find themselves meeting the definition of medically refractory epilepsy very quickly when seizures fail to respond to medication at the very onset.

Patients for whom medications fail to control their seizures from initial treatment can rapidly progress through each drug trial. Guidance on the speed of titrating an ASM can vary based on the mechanism and side effect profile of the specific medication. Gentle titration is often necessary to minimize the risk of adverse side effects . It should not be the goal to simply progress through medication expeditiously simply to complete trials of two medications without reasonable opportunity for success. However, there are reports of specific medication titration schedules that can permit more rapid dose escalation with good results . Irrespective of the rate at which medications are trialed, 22%–36% of individuals become refractory to medication. This reported range varies depending on the specific definition of DRE used in respective studies .

Decreasing rates of seizure control with subsequent medication trials is a finding strongly supported by longitudinal data following patients treated for newly diagnosed epilepsy. For example, two well-designed prospective studies followed a cohort of adult and adolescent patients in Scotland. The second expanded the first and reached a total of 1795 newly diagnosed patients to track seizure control with successive medication trials over a 30-year period from 1982 to 2012. About half of the cohort achieved seizure control with the first ASM (50.5%). The second, third, and all subsequent regimens yield only 11.6%, 4.4%, and 2.1% rates of control, respectively . Therefore, the logic follows that due to the low likelihood of achieving seizure freedom with continuing medication trials, it is appropriate to consider nonpharmacologic options that may carry higher risk but also a greater potential benefit.

Collectively, this background provides the bare minimum of information to guide those who treat individuals with epilepsy for when to consider nonpharmacologic intervention. In this chapter, we will discuss clinical conditions in which it may be appropriate to proceed with surgical intervention without necessarily delaying a complete trial of two medications as suggested by the previously mentioned guideline. Medical guidelines, by definition, are intended to provide generalized guidance for a representative population. Guidelines are not intended to dictate individual treatment decisions, particularly when unique circumstances vary from the presumed population-level characteristics. Therefore, this chapter will consider various specific scenarios where the likelihood of achieving seizure control with medications is suspected to be significantly lower than in the general epilepsy population. For these groups, the high rates of response to the initial medication among the general population studies in clinical trials do not apply. In such a scenario, the guidelines proposing waiting for a lack of medication response may not be necessary. Patients, families, and their healthcare providers can elect a more personalized treatment plan specific to the patient’s unique condition, which may differ from the guidelines derived from the more general epilepsy population.

Motivation for early intervention

Better seizure response with early intervention

Multiple clinical studies have documented a significant relationship between seizure response and the time interval from seizure onset to time of surgical intervention . Despite the heterogeneity in format for reporting seizure duration, these studies repeatedly show that individuals whose seizures are treated sooner have a greater probability of a good seizure response when compared to those with a longer time interval until surgical intervention. A related metric addressed in several of these studies is the potential for better seizure response for individuals treated in the pediatric age range . It should be noted that the duration of epilepsy until treatment is a modifiable risk factor that can be improved with early intervention and increased awareness by providers and caregivers. In contrast, age at seizure onset is not a modifiable risk factor for most patients.

Improved seizure response with earlier intervention has not been a universal finding across all studies. For example, Spencer and colleagues reported a prospective multicenter study aimed at predicting seizure outcomes after resective epilepsy surgery . In their multivariate analysis, only the absence of generalized tonic-clonic seizures and the presence of hippocampal atrophy were found to be independently associated with seizure freedom at 2 years after surgery. Specifically, neither age at onset (grouped by age <5, 5–9, 10–19, 20–29, 30–39, and ≥40 years) nor duration of epilepsy (grouped by <5, 5–9, 10–14, 15–19, and ≥20 years) revealed a statistically significant association to seizure freedom at 2 years.

However, a systematic review of the literature and metaanalysis did find a significant association between shorter epilepsy duration with postoperative seizure freedom . Bjellvi and colleagues provided a narrative review of 25 observational studies from the literature and included data from 12 studies representing 1545 patients for metaanalysis. They grouped outcomes by duration of epilepsy at 2-, 5-, 10-, and 20-year intervals as well as a dichotomized grouping of less than 5 compared to greater than 10 years. They acknowledge the published literature has a significant heterogeneity and that evaluation by Grading of Recommendations Assessment, Development and Evaluation (GRADE) criteria suggest a low certainty.

Progressive cognitive decline while seizures are uncontrolled

In addition to the association with better seizure response, early intervention may be able to decrease the severity of cognitive decline associated with ongoing seizures. Some cognitive dysfunction can be temporary and reversible, such as epileptic encephalopathy that improves after achieving seizure reduction . A striking example of such is reported by Pizoli and colleagues where a patient with epileptic encephalopathy was examined clinically and with resting state functional MRI (rs-fMRI) before and after corpus callosotomy. They found significantly altered resting state networks (RSNs) consistent with the patient’s global cognitive decline prior to surgery. An early surgery was performed to intervene in the epileptic encephalopathy. After surgery the severe seizure burden was reduced, and cognitive function returned towards baseline. This cognitive recovery was supported by clinical neuropsychiatric evaluation as well as rs-fMRI functional connectivity (FC) metrics showing recovery of a more typical appearing network organization after surgery. FC analysis included a comparison of covariance between various RSN hubs in the brain suggesting a low level of covariance during the state of epileptic encephalopathy, which can be interpreted as a dramatic reduction of organized brain activity. This clinical and translational research data exemplifies the brain’s restricted ability to develop normally in a state of epileptic encephalopathy that can be reversed by a palliative procedure to reduce the seizure burden.

The exemplar case above describes a 5-year-old whose development was significantly altered by a state of epileptic encephalopathy. Childhood and adolescence represent a time of tremendous cognitive and social development that is critical to later life achievements. Critical developmental periods can be drastically interrupted or delayed due to the ongoing disruption of frequent seizures. In cases where surgical intervention is ultimately necessary for a reasonable chance at seizure freedom, there is no benefit to the patient for delaying surgical evaluation. Furthermore, neurologic deterioration tends to progress over time. While an acute encephalopathy may be recoverable, other forms of chronic cognitive decline become static. Regaining developmental potential and stopping neurologic decline is a strong motivator to undertake the risks of surgical intervention at a young age. It is logical to suppose that halting the progressive decline sooner should result in greater potential to recover previously lost function. However, the clinical evidence to support this hypothesis is difficult to ascertain.

Autism spectrum disorder (ASD) is a diagnostic category with symptomatology that fundamentally includes impaired social communication. ASD has a high co-occurrence with epilepsy upwards of 30% . These rates are higher than each respective condition in the general population . The pathophysiologic mechanism that links epilepsy and ASD is not clear. A significant hurdle in disentangling these two pathologies is the difficulty in defining childhood autism, as described by Kanner in 1943 , and similar conditions said to be on a diagnostic spectrum. The DSM-IV previously included five subtypes to pervasive developmental disorder (PDD), including: (1) autistic disorder, (2) Asperger syndrome, (3) PDD not otherwise specified, (4) childhood disintegrative disorder, and (5) Rett disorder. Subsequently, the DSM-V was published in 2013 and removed Rett syndrome, which now has a well-defined genetic mutation, and now combined the remaining four conditions under a single label of ASD . Understanding these definitions helps to appreciate the literature as it has evolved over time . Retrospective cohort studies have described an association with early age of onset in the co-occurrence of ASD and seizures . While some hypothesize a causal relationship, it is an ongoing controversy to decipher if seizures lead to developing ASD or vice versa. As such, it is even less clear if early intervention for seizures helps to lower the likelihood of ASD with epilepsy.

Landau–Kleffner syndrome (LKS) is a childhood-onset condition characterized by acute regression in language and an acquired verbal auditory agnosia. Language regression as seen in LKS does not fall under the umbrella definition of ASDs, yet it is a cognitive deficit often described in ASD cohorts reported in the literature. The term autistic epileptiform regression represents this association . With LKS, most patients have seizures. Case series on language recovery after seizure-directed treatment have varied with some reporting no improvement after treatment with ASM and others reporting improvement after surgical intervention via multiple subpial transections .

The impact of early seizure intervention on cognitive regression syndromes remains difficult to characterize. This is largely owing to challenges defining syndromes and heterogeneity across small cohorts reported in the literature. Progress has clearly been made in improving these definitions but will require further directed effort and a thorough understanding of the evolution of these diagnoses with respect to epilepsy. Aside from the above-mentioned syndromes, evidence for cognitive improvement with early seizure intervention has been described in other clinical settings.

Basic neuroscience of critical developmental periods leads to the hypothesis that reducing seizures during development may improve cognitive performance. Outside of the syndromes mentioned above, clinical evidence supports this hypothesis by demonstrating lower cognitive function associated with earlier age at seizure onset and longer duration of uncontrolled epilepsy . Case series have reported improved cognitive function after both medical and neurosurgical treatment that improved seizure control . The term “catch-up” development has been used to describe a related phenomenon when cognition is impaired by the presence of frequent seizures and once those seizures resolve the child’s cognitive function rapidly improves towards an age-appropriate level.

Epilepsy surgery in early life

Seizures can disrupt cognitive function at any stage of development, however, seizures during infancy and early childhood have been found to cause particularly severe neurologic decline. Such devastating impact has led many authors to conclude special attention is warranted for patients in infancy and early childhood to consider early intervention. The young age, small physical size, and associated low circulating blood volume demand special consideration for the risks of surgical intervention in infants. Clinical evidence in this vulnerable population is limited, but several case series suggest the potential for good seizure outcomes that permit recovery of cognitive development .

In a retrospective cohort, Ramantani and colleagues set out to evaluate the effect on cognitive development from early life surgery for medically refractory epilepsy. Their cohort included 30 children under the age of 3 years old who were followed longitudinally pre- and postoperatively. They measured developmental metrics using the Bayley Scales of Infant Development, Kaufmann Assessment Battery for Children, the Vineland Adaptive Behavior Scales, among others and in addition to clinical assessments by psychologists and pediatric neurologists. The use of standardized cognitive assessments allowed longitudinal assessments of developmental progress as well as comparison to typically developing control subjects. In this manner, the authors were able to document developmental progress relative to their presurgical baseline in all but one of their patients. However, relative to controls, 75% remained at their relative level of cognitive deficit and the remaining 25% showed a relative decrease in the developmental level .

Most case series do not have a high level of standardized longitudinal cognitive developmental measures as reported by Ramantani and colleagues. Seizure outcomes are more readily assessed across institutions. A large multinational retrospective study of the impact of epilepsy surgery in infancy was conducted across 19 centers spanning the years 1999–2020. This study included infants under 3 months of age undergoing resective or disconnective surgery. Outcomes were excellent with seizure freedom reaching 45% and 70% for focal resection and hemispheric surgery, respectively . For such a vulnerable population of very young individuals, these data represent a large cohort with excellent results indicating early intervention during infancy has great seizure response and is safe to perform by experienced clinical teams.

Cognitive outcomes are similarly important to seizure outcomes and may also be impacted by intervention early in life. A study by Kadish and colleagues addressed this concern in a cohort of patients who underwent epilepsy surgery in the first 3 years of life. Fig. 2.1 summarizes a key result of their study demonstrating the severity of cognitive impairment in this cohort (compared to the typical distribution shown as dashed curve). The second part of this figure visualizes cognitive impairment as a function of duration of epilepsy .

Surgery prior to completing trials of two medications

Despite strong motivating factors to seek early surgical intervention from the time of seizure onset to surgical evaluation remains unacceptably long. While there are likely a multitude of causes for such delays that are undoubtedly multifactorial, one decision to evaluate is what scenarios are appropriate to consider surgery prior to completing two adequate medication trials. A neurosurgeon can be easily biased by the mentality that “to cut is to heal.” We should exercise appropriate judgment in balancing the risks of surgery that may be obviated by effective medical therapy. However, a risk-benefit calculus should include the consequences of ongoing seizures that continue to impact the patient during medication trials, which may be lessened by surgical intervention, as well as other benefits of surgery that may coincide with treating seizures.

Focal epilepsy associated with a definable lesion on clinical imaging has an excellent response to surgical treatment. When clinical evidence is concordant with a focal lesion on brain MRI, the likelihood of reaching seizure freedom from surgical excision of the structural abnormality identified on imaging is exceptionally high. Concordant clinical evidence comes in several forms, including seizure semiology that may suggest a specific hemisphere or functional region, routine video electroencephalography (EEG) that suggests a region of onset, metabolic imaging such as positron emission tomography and single-photon emission computed tomography that can show regions of aberrant metabolic demand associated with a seizure focus, and additional modalities to further refine electrophysiologic data such as high-density EEG and magnetoencephalography. Even focal lesional epilepsy can be controlled with medication, but early surgical consideration for lesional epilepsy is warranted in some scenarios.

Tumor-associated epilepsy

New onset seizure is an often observed presenting symptom of a brain tumor. In this setting, it is common to grapple with the conceptual differentiation of a “tumor surgery” versus an “epilepsy surgery.” The implication of these two approaches concerns their indications and goals of the surgery. The clinical scenarios where tumor and epilepsy concerns overlap reveal critical differences in clinical decision-making that follow different guidelines. For a tumor surgery, often the indication for intervention is evident by symptoms related to mass effect and the potential for progressive growth without treatment. Even in situations where complete tumor excision is not feasible, cytoreduction can have a significant impact on adjuvant therapies. These goals are not typically shared when considering epilepsy surgery. In the absence of neoplasia, lesional epilepsy most often represents dysplasia where concern for growth or malignant progression is not typical. Similarly, cytoreduction is not an objective in epilepsy surgery because the volume of a seizure focus does not correlate with its epileptogenic potential. Despite the diverging surgical indications, there is often overlap in consideration that may partly represent oncologic interests, such as tissue diagnosis and tumor grading to guide further treatment of neoplasia, or epilepsy interests, such as to relieve symptoms that may solely be the presenting seizure.

Routine MRI can characterize lesions by their magnetic resonance signal characteristics. These characteristics allow neuroradiologists to generate a differential diagnosis. In best-case scenarios, the differential may be a short list with a strong leading diagnosis. Other times the differential is broad and may include neoplasia and dysplasia with equal likelihood. It is not until tissue is obtained and histology analyzed that a definitive diagnosis can be reliably obtained. Therefore, lesional epilepsy cases should always have some level of judicious suspicion for neoplastic potential. In cases where seizures are the primary manifestations of symptoms, rapid progression of seizure frequency and severity may be a sign of less benign pathology. This is in contrast with a patient whose seizures remain at a constant frequency with unwavering semiology; more typical primary epilepsy that should undergo standard medication trials. Similarly, gadolinium uptake is more suggestive of aggressive pathology and warrants consideration of neoplasia as opposed to dysplasia.

In the absence of aggressive features such as rapidly progressing seizures and concerning imaging, low-grade tumors may be otherwise difficult to distinguish from dysplasia, both clinically and radiographically, and can be associated with seizures over a longer period. Long-term Epilepsy Associated Tumors (LEATs) is a term that has been coined to describe this clinical entity . In line with the overlapping oncologic and epilepsy indications, a case series of 154 epilepsy-associated tumors reported improved outcomes associated with a shorter duration between seizure onset and surgery as well as with gross total resection , each being a well-described prognostic factor for epilepsy and oncology outcomes, respectively. In their series, Luyken and colleagues encourage early surgical intervention due to the high rate of seizure freedom and the potential for malignant transformation. While most LEATs are low grade, progression and malignant transformation have been described .

In the absence of high-grade histology, rates of seizure freedom from surgical resection of LEATs are exceptionally high, warranting early intervention irrespective of medication trials . In some case series greater than 70%–80% of patients achieve seizure freedom after surgical resection of LEATs . Such seizure freedom rates are greater than that achieved after an adequate trial of two ASMs.

Vascular lesions

Vascular malformations are another unique consideration for lesional epilepsy. Cavernous malformations (cavernomas) and arteriovenous malformations (AVMs), among other variations of cerebral vascular malformations, are relatively common causes of seizures. Like the overlapping oncologic indications discussed above, vascular indications for intervention are similarly disparate at times and shared in others. The indications for treating AVMs have been questioned extensively in recent literature following the ARUBA trial . These concerns arise primarily regarding the natural history of spontaneous hemorrhage of a known AVM and the risk of neurologic deficit incurred by surgical treatment. The impact of an AVM on seizures may be related to the mass effect or secondary to hemorrhagic stroke after rupture. Cavernomas are a related vascular malformation characterized by lower rates of blood flow relative to an AVM. Hemorrhage of a cerebral cavernoma often carries less risk for significant morbidity. However, chronic effects on the adjacent cortex causing seizures is a common presenting symptom of a cavernoma . Similar to low-grade tumors, a focal vascular lesion carries very good outcomes for seizure control after surgical resection in well-selected patients . Yet, in opposition to LEATs, the natural history and lifetime risk of these vascular lesions relative to treatment morbidity has led to controversy regarding the timing of surgical intervention .

Conclusion

The key advantages and disadvantages of early surgical intervention is summarized in Table 2.1 . This chapter reviewed several clinical scenarios where early surgical intervention for epilepsy may be warranted. It is critical to understand the data that lead to the currently accepted definition of medically refractory epilepsy in order to appropriately apply guidelines to the practitioner’s patients. However, epilepsy neurosurgeons and neurologists should also be aware of the possibility of late seizure remissions in patients who may be able to achieve good seizure control without invasive procedures. Intractability and response to medication do not happen at a uniform time for all patients and all seizures. While sufficient research to define those populations who are expected to become refractory is limited, nonidiopathic localization-related epilepsy in children is more often refractory with a history of a higher rate of seizures at onset, status epilepticus, and focal EEG slowing . A well-informed provider may combine this information regarding natural history, predictive factors, age, and developmental stage with a nuanced approach to individualized patient care. In doing so, patients for whom the likelihood of achieving seizure control is low and the probability for good surgical outcomes is high can be appropriately selected for early neurosurgical intervention.

Table 2.1

Only gold members can continue reading. Log In or Register to continue