Mobilizing a New Era in Lennox-Gastaut Syndrome Treatment and Prevention

Abstract

This review summarizes content presented at the Pediatric State of the Art Symposium held during the American Epilepsy Society's annual meeting in December 2024. The symposium focused on Lennox-Gastaut syndrome (LGS), a severe developmental and epileptic encephalopathy that emerges in childhood. Despite its diverse etiologies, LGS is defined by a convergent constellation of electroclinical features: multiple seizure types including tonic seizures, slow spike-wave and generalized paroxysmal fast activity on EEG, and intellectual disability. LGS is almost always refractory to available therapies and accounts for substantial costs—in healthcare spending and in quality of life for affected individuals, their families, and caregivers. The symposium highlighted recent breakthroughs in research, clinical care, and outcome measurement that have positioned the clinical, scientific, and patient advocacy communities to usher in a new, more hopeful era of treatment and prevention.

Introduction

We are nearing a century since the first descriptions of Lennox-Gastaut syndrome (LGS).^1-3 This milestone invites reflection on the progress made and the challenges that remain in understanding and treating this complex epilepsy syndrome. ⁴

We begin by reviewing recent clinical and preclinical research breakthroughs that have transformed our understanding of LGS, including the discovery that diverse etiologies causing LGS converge on shared brain circuits, the utilization of electronic health record data to identify computable phenotypes and study natural history, and the development of new animal models to study pathogenesis and treatment.

We then highlight new and emerging medical, surgical, and neuromodulation therapies for LGS, as well as gaps that should be prioritized for future study, including the need for greater standardization of medical approaches and the importance of defining the “pre-LGS” state to facilitate preventive clinical trials.

Finally, we explore the critical role of addressing non-seizure outcomes—such as communication, motor skills, and sleep—and the development of fit-for-purpose assessment tools that prioritize the experiences of caregivers and people with LGS by capturing incremental yet meaningful improvements in quality of life.

Research Breakthroughs and Research Priorities for LGS

Conceptualizing LGS as a Secondary Network Epilepsy

In 2022, ⁵ the International League Against Epilepsy (ILAE) revised the definition of LGS to require the following: (a) the onset of seizures before 18 years of age, (b) tonic seizures in addition to at least one other seizure type (e.g., atonic, atypical absence, epileptic spasms, or others), (c) slow spike-wave (SSW) and generalized paroxysmal fast activity (GPFA) on current or historical interictal EEG recordings, and (d) intellectual disability. Despite this clarity, accurately identifying LGS can be challenging due to its dynamic and evolving clinical presentation,^6-8 especially in early stages or as symptoms fluctuate with development or treatment.

Further complicating this picture is how diverse etiologies converge to produce the same electroclinical phenotype. In short, almost any pathology can lead to LGS, spanning the full spectrum of etiological subgroups encountered in epilepsy ⁹ : genetic, structural, infectious, immune, metabolic, and unknown causes. This heterogeneity is often a point of contention between those who argue the focus should be on treating individual etiologies and those who embrace the diversity as a chance to uncover shared mechanisms and develop broadly applicable therapies.^10,11

However, neuroimaging studies may offer a potential resolution. One recent proposal is the conceptualization of LGS as a disorder of convergent, “secondary” brain networks.^11,12 The hypothesis is that diverse etiologies lead to convergent alterations in the developing brain, which give rise to shared patterns of pathological activity that manifest as the cardinal electroclinical features of LGS. For example, studies using simultaneous EEG with functional MRI (EEG-fMRI) have shown that bursts of GPFA do not engage all brain areas synchronously—as might be expected from their generalized EEG appearance—but rather involve increased activity in the frontal and parietal “association” cortex and the thalamus, basal ganglia, cerebellum, and brainstem.^12-14 This pattern is similar across individuals with structural, genetic, and other causes of LGS.^12-14

The concept of secondary networks in LGS distinguishes two, complementary treatment goals. The first, which we might call etiology-based or “precision” therapy, aims to intervene as early as possible, before the secondary network process has been established. This might include prompt removal of a focal structural brain abnormality^15,16 or genetic testing to guide the selection of anti-seizure medications that target the specific molecular defects.^17,18 The second approach, we might call network-based therapy, instead aims to treat the shared brain circuits through techniques such as neuromodulation^19,20 or disconnection procedures like corpus callosotomy. ²¹

However, several key knowledge gaps remain in this framework. For instance, we still lack a complete understanding of the risk factors that drive secondary network epileptogenesis; the same etiology can variably lead to LGS, other epilepsy syndromes, or even no epilepsy at all. While neuroimaging studies have begun to uncover the shared networks through which LGS arises from different etiologies, the specific neurobiological mechanisms mediating this convergence are unclear. Addressing these gaps will require longitudinal and preclinical studies to identify the structural and functional brain changes involved in the evolution of LGS. This may provide potential therapeutic targets for steering neurodevelopment toward more favorable outcomes.

Leveraging Electronic Medical Data for LGS Research

Clinical trials and research on LGS depend on the identification of individuals who meet diagnostic criteria, ⁵ with studies often performed across multiple centers due to small participant numbers. However, accurate and efficient identification of individuals with LGS is challenging for several reasons. Definitions of the syndrome have fluctuated over time,^5,9 leading to confusion about when the diagnosis should be applied and inconsistent documentation of key diagnostic terms such as “GPFA,” “SSW,” and “tonic seizures” in electronic health records (EHRs). Additional challenges include substantial diagnostic overlap with other epilepsy syndromes, the shifting appearance of LGS over time, and the reported phenomenon of “strategic” diagnosis, ²² where individuals who do not meet all criteria for LGS are nevertheless labelled as such to facilitate access to restricted therapies.

A potential solution is the development of computable phenotypes ²³ from EHRs or other data sources: these are combinations of search terms that, even in the absence of an explicit diagnosis, can identify LGS cases with high sensitivity and specificity. An ongoing, Patient-Centered Outcomes Research Institute (PCORI)-funded study (ClinicalTrials.gov ID: NCT05374824) aims to develop such computable phenotypes for LGS from EHR data. These phenotypes could then be used to mine existing, high-volume electronic clinical data across multiple sites, potentially allowing studies on natural history, social and geographical disparities in treatment utilization, and comparative effectiveness studies across different treatment modalities and etiologies of LGS.

Preclinical Research to Understand LGS Pathogenesis and Advance Treatment

Current therapies for LGS are rarely sufficient: more than 90% of individuals have medically refractory epilepsy.^8,24 The lack of effective therapies reflects a poor understanding of the fundamental mechanisms underlying LGS pathogenesis.

Preclinical models provide a powerful means to address this gap, but few currently exist for LGS. What might an informative preclinical model of LGS look like? It would recapitulate known etiologies (e.g., a specific gene or structural brain abnormality), core electroclinical criteria (e.g., early-life onset, hallmark EEG patterns and seizure types), comorbidities (e.g., cognitive and sleep disturbances), and scalability for high-throughput therapeutic screening. Although it is unlikely that any single model organism would embody all these characteristics, a combination of models may provide a comprehensive strategy to simulate LGS pathogenesis and test new treatments.

Improved understanding of genetic causes of LGS, together with the development of gene-editing techniques, are expanding options for preclinical models in LGS. A recent example is the Gabrb3^+/D120N knock-in mouse model,^25,26 which recapitulates multiple aspects of the syndrome including progressive atypical absence, tonic, and other seizure types; runs of spike-wave activity on EEG; and cognitive deficits.

How might preclinical models inform our understanding of LGS pathogenesis and treatment? Promising examples include the recent discovery that increased myelin plasticity exacerbates epilepsy progression in an SCN8A Scn8a+/- mouse model of generalized epilepsy, and that this epilepsy progression can be prevented by suppressing activity-dependent myelination. ²⁷ In zebrafish, high-throughput drug screens in models of Dravet syndrome identified Clemizole, a modifier of serotonin signaling, with potent anti-seizure effects ²⁸ ; this has led to ongoing human trials for Clemizole as an adjunctive treatment for Dravet syndrome and LGS (ClinicalTrials.gov IDs: NCT04462770, NCT05066217). Recent studies in which other genes, including some linked to LGS, are engineered into zebrafish have demonstrated their utility as epilepsy models.^29,30

Improving Clinical Care for LGS

Standardized Medical Treatment for LGS: State of the Art

Anti-seizure medications (ASMs) remain the cornerstone of LGS treatment. At present, eight ASMs have received Food and Drug Administration approval in the USA for the treatment of LGS: clonazepam, felbamate, lamotrigine, topiramate, rufinamide, clobazam, cannabidiol, and most recently fenfluramine. ³¹ Others are commonly used off-label for LGS, including valproic acid.

However, despite an increasing number of options, the evidence base for treatment selection is limited for several reasons. First, individual centers are likely to see only a small number of patients with LGS, making it challenging to identify sufficient candidates for early-stage trials. Second, LGS evolves over time, often from other medically refractory syndromes such as infantile epileptic spasms syndrome (IESS), making monotherapy studies difficult or impossible: ASMs are typically evaluated as add-on therapies for LGS, hindering evaluation of their standalone efficacy. Third, there is significant heterogeneity in trial designs, including variability in inclusion criteria, outcome measures, and follow-up times, which limits comparability of results between trials and is a barrier to high-quality meta-analyses.^32-35 Fourth, few studies have explored the impact of treatments on outcomes beyond seizures, such as cognitive and behavioral changes or quality of life. Lastly, there is a lack of head-to-head comparisons among different ASMs (or between ASMs and non-pharmacological options), which prevents the development of evidence-based, standardized treatment algorithms.

In their absence, several consensus-based treatment algorithms for LGS have been developed to assist clinical decision-making,^31,36-39 typically designed by an expert panel after reviewing the available literature. These algorithms typically support valproic acid as a first-line agent for LGS, given that 25–30% of patients attain >50% seizure reduction. ⁴⁰ However, there are concerns that while valproic acid is efficacious for reducing atypical absence and myoclonic seizures,^41,42 efficacy for tonic or generalized tonic clonic seizures is less clear. ⁴² Second-line medications (add-ons to valproic acid) in expert recommendations often include lamotrigine and rufinamide.^31,36-39 The quality of evidence for other ASMs remains low or very low.^32-35

Beyond ASMs, other medical therapies include the ketogenic diet, a high-fat diet designed to mimic the metabolic and anti-seizure effects of starvation. ⁴³ Evidence for the ketogenic diet as an LGS therapy includes several clinical trials in drug-resistant epilepsy comparing the ketogenic diet to standard care, considered together in a 2020 Cochrane review. ⁴⁴ This led to the recommendation that the diet can be helpful, consistent with findings from open-label cohort studies showing that ∼30–50% of individuals with LGS experience >50% seizure reduction after six months or more on the diet.^45-47 The limitations of the diet as a therapy include the need for a restrictive diet and considerable side effects.

A recently developed consensus-based algorithm, integrating new evidence, was developed by Auvin and colleagues in 2024. ³⁹ This algorithm recommends first-line treatment with valproic acid, followed by second-line adjunctive treatment with lamotrigine, then third-line rufinamide, and finally fourth-line options such as clobazam, cannabidiol, or fenfluramine. Felbamate may be an effective fourth-line option, but the risk of aplastic anemia must be considered. The algorithm recommends that, alongside medical treatments, resective surgery, neuromodulation, ketogenic diet and corpus callosotomy should be considered with a personalized approach. Available treatments should not necessarily be considered in sequence, but in parallel.

Principles that can guide LGS treatment include conducting methodical treatment trials, avoidance of excessive polytherapy (goal of no more than two medications at a time), and maintaining vigilance for subtle/subclinical seizures and status epilepticus.

Addressing the Under-Utilization of Surgery for LGS

Surgery can be a highly effective treatment for medically refractory epilepsy, but there is increasing recognition that it is underutilized and often delayed in LGS,^48,49 particularly in those who carry etiologies traditionally perceived as surgically non-remediable (e.g., non-structural genetic causes). In published studies of LGS, the lag time between seizure onset and surgery can be 20 years or more, often after numerous ASM attempts.^16,50 This contrasts with ILAE guidelines stating that referral for surgical evaluation should occur as soon as drug resistance is ascertained, ⁵¹ defined as the failure of adequate trials of two tolerated and appropriately chosen ASMs. ⁵²

In addition to invasive neuromodulation, which is discussed below, the main surgical options for LGS include resective or ablative surgery and disconnection procedures including corpus callosotomy. A recent meta-analysis found that 76.1% (105/138) of individuals with LGS undergoing resective surgery had a ≥ 50% seizure reduction at ≥2 years follow-up, with 47% (65/138) becoming seizure-free. ²¹ Regarding corpus callosotomy, the same meta-analysis found that 66.3% (193/291) of patients had a “worthwhile” (typically ≥50%) seizure reduction at ≥2 years of follow-up. Cognitive and quality-of-life improvements were described following both surgery types; no changes or occasional declines in these measures were also reported. ²¹

The choice and extent of surgery are guided by multiple considerations including etiology, age, and baseline function before surgery. Regarding etiology, persons with LGS secondary to focal cortical lesions should be referred for prompt resective/ablative surgery, given that the likelihood of seizure freedom is highest in this scenario. ²¹ Focal features on examination or seizure semiology (e.g., hemi-tonic seizures or focality preceding seizure spread and generalization) may indicate a potential focal origin and should prompt further evaluation for surgical candidacy, including neurophysiological and functional studies to determine focality and preserve post-surgical function. In the case of diffuse or multi-focal lesions (e.g., tuberous sclerosis complex), where complete removal of abnormal brain tissue may not be achievable, selective removal of maximally involved areas (e.g., defined by stereoencephalography [SEEG]) can still provide improved quality of life, including seizure freedom in some cases. ⁵³ Additionally, while patients with non-structural genetic etiologies are typically not candidates for resections, recent studies suggest the benefits of corpus callosotomy or neuromodulation. ⁵⁴ These findings warrant a reappraisal of surgery's role in improving quality of life for people with LGS, even if the intent is not always curative.

Advances in Neuromodulation for LGS

Neuromodulation is changing the treatment landscape for epilepsy. Common forms include vagus nerve stimulation (VNS), deep brain stimulation (DBS), and responsive neurostimulation (RNS). These treatments carry Food and Drug Administration approvals in the USA for focal epilepsy, but off-label use in LGS is growing.

DBS has emerged as a particularly promising approach for LGS, as it provides the ability to modulate bilateral epileptic networks by directly targeting key nodes involved in seizure propagation, such as the thalamus. Early pioneering work by Fisher ⁵⁵ and Velasco^56,57 demonstrated the anti-seizure effects of bilateral DBS of the centromedian nucleus of the thalamus for LGS, with some individuals experiencing prolonged seizure freedom. This was followed by the Stimulation of the Anterior Nucleus of the Thalamus for Epilepsy (SANTÉ) study, a larger, randomized, double-blind, placebo-controlled trial of 110 participants, most with focal epilepsy. ⁵⁸ SANTÉ showed a median 40.4% reduction in seizures after 3 months of blinded DBS to the bilateral anterior thalamic nuclei, increasing to a 56% reduction after two years of unblinded follow-up, and then to 75% after ten years. ⁵⁸

In 2022, the efficacy of bilateral thalamic centromedian DBS for LGS was evaluated in a randomized, double-blind, placebo-controlled trial, called the Electrical Stimulation of Thalamus for Epilepsy of Lennox-Gastaut phenotype (ESTEL) trial. ¹⁹ The trial enrolled 20 young adults with LGS, of whom 19 completed all study phases (one participant was explanted due to an infection). The primary outcome was the proportion of participants who showed a reduction of 50% or more in diary-recorded seizures after 3 months of stimulation, in the treatment (DBS active) versus control (DBS inactive) groups. Fifty percent of participants in the treatment group achieved ≥50% seizure reduction, compared with 22% of controls, which was not statistically significant. However, a secondary outcome defined by electrographic—as opposed to diary—seizures showed a significant effect of stimulation, with 89% of the treatment group showing a ≥ 50% reduction compared with none of the controls. Across all participants, the median seizure reduction at study exit (baseline vs. 3‒6 months active stimulation/9 months post-implantation) was 46.7% for diary-recorded seizures and 53.8% for electrographic seizures. There was also an improvement in caregiver-reported measures of epilepsy severity and disability, but no changes in cognitive or behavioral measures. ⁵⁹ Ongoing work is examining longer-term outcomes, with 5-year follow-up data from ESTEL expected in late 2025.

Other neuromodulation studies on the horizon for LGS include the Children's Adaptive Deep Brain Stimulation for Epilepsy (CADET) trial, which will assess the efficacy of thalamic centromedian DBS for children (5‒14 years) with LGS ⁶⁰ (ClinicalTrials.gov ID: NCT05437393), and the RNS System LGS Feasibility Study (NCT05339126), which is investigating the preliminary safety and effectiveness of closed-loop, dual thalamic and cortical RNS delivered bilaterally to the thalamic centromedian nucleus and a recently identified “epileptic hotspot” in the premotor frontal cortex. ⁶¹

Important next steps toward optimizing neuromodulation for LGS include identifying the most effective stimulation locations and paradigms^20,62; developing robust biomarkers of seizure burden to allow rapid stimulation programming and to improve the reliability of outcome measurement beyond seizure diaries ⁶³ ; and performing comparative effectiveness studies to guide evidence-based treatment selection among an increasing number of options. ⁶⁴

Transforming Clinical Trials for LGS

Defining “Pre-LGS” for Preventive Clinical Trials

LGS is diagnosed in early childhood, often in children who already have a history of seizures and epilepsy as infants, particularly IESS. ⁶⁵ Anecdotally, seasoned pediatric epileptologists sometimes say that a child is “heading towards LGS” when an EEG begins to show early changes associated with refractory epilepsy, like multifocal spikes or runs of diffuse spike-wave discharges. This clinical intuition raises the possibility that there are early biomarkers that precede a diagnosis of LGS. Such a biomarker might then serve as an entry criterion for a clinical trial of LGS prevention. Viewed from this angle, the ILAE criteria for LGS describe the end-stage of a progressive disease, suggesting there may be windows of opportunity to intervene to prevent LGS evolution.

However, the details of this evolution are not well described. One retrospective study found that the average duration from first seizure to LGS diagnosis is two years. ⁶ Another study found that progression from IESS to LGS was more likely in those with a history of developmental delay and seizures prior to the onset of spasms, and when seizures were not controlled by first-line treatment of spasms. ⁶⁵

Using the Rare Epilepsies in New York City (RENYC) database, which includes neurology records across five academic medical centers in New York City, Deering and colleagues ⁶⁶ performed a retrospective cohort study of 33 children who met most but not all ILAE criteria for LGS ⁵ : tonic seizures were not specifically required due to uncertainty about when these begin, although they were seen in 25/33 (76%) of cases. ⁶⁶ Two-thirds of children had early-life seizures (median onset age of 5 months) that were followed by a period of seizure freedom—a “gap”—before the diagnosis of LGS. In most children, SSW appeared prior to or during this gap. The authors posited that SSW may therefore herald the onset of LGS, suggesting its potential utility as an EEG marker to guide preventive treatment.

Recent clinical trials in early life epilepsies provide some precedent for such preventive intervention. The EPISTOP ⁶⁷ and PREVeNT ⁶⁸ trials assessed whether early vigabatrin treatment can prevent seizures in at-risk infants with tuberous sclerosis and abnormal EEG findings. Both trials found that preventive vigabatrin reduced the incidence of infantile spasms. However, they disagreed on other outcomes: the EPISTOP trial reported that vigabatrin also reduced the risk of drug-resistant epilepsy and seizures other than spasms, whereas the PREVeNT trial found no preventive effect on drug-resistant epilepsy or focal seizures.

Work remains to translate these findings into actionable strategies to prevent LGS. Validating the utility of SSW as a biomarker will require studies to assess its inter-rater reliability and confirm its predictive value prospectively. Equally important will be understanding whether families and caregivers will accept early, potentially aggressive treatment in a child who does not yet have LGS. Finally, identifying the optimal preventive treatment modality will likely require a program of clinical trials to identify the right therapy at the right dose to give at the right time. High-dose steroids may be a reasonable approach, given their effectiveness for epileptic encephalopathies like IESS, ⁶⁹ epileptic encephalopathy with spike-wave activation in sleep (EE-SWAS), ⁷⁰ and LGS.^71-74

Measuring What Is Important to Caregivers and People with LGS

In clinical trials and routine management of LGS, the primary focus is typically on stopping seizures, particularly “drop” seizures, given their acute risk of injury. However, there is growing recognition—among caregivers and clinicians—that non-seizure consequences can have similarly severe, or even greater, impacts on people with LGS and their families. ⁷⁵

In 2023, the DEE Parents Speak survey collected information from nearly 270 families and caregivers of people with developmental and epileptic encephalopathies (DEEs), including 51 with LGS (http://inchstoneproject.org). Respondents were asked to rank their priorities for meaningful change, highlighting areas of concern beyond seizure control. The top-ranked priorities included improvements in expressive communication (expressed by 78% of respondents), gross motor skills (49%), sleep (29%), and receptive communication (20%). The survey also highlighted areas where patients find fulfillment, like listening to music (expressed by 86% of respondents) and cuddling (76%), which are important for developing comprehensive care strategies that not only address challenges but also promote quality of life and well-being. The findings were similar across patient ages, syndromes (e.g., LGS vs non-LGS), specific genetic etiologies, and levels of severity.

However, despite increasing awareness of these priorities, they are rarely addressed adequately in real-world settings. For example, another survey conducted with the LGS Foundation, involving 245 caregivers of people with DEEs (54% with LGS), found that over 90% of respondents emphasized the importance of measuring behavior, communication, and quality of life, yet fewer than 50% felt that their healthcare providers evaluated these aspects effectively. ⁷⁶ When queried about barriers to addressing these needs, most respondents (51%) reported a lack of developmentally appropriate instruments as a significant obstacle.

Accurately measuring non-seizure challenges and detecting progress requires fit-for-purpose tools that overcome floor effects and capture small improvements—that is, the “inchstones,” not the milestones. The Inchstone Project is one initiative aimed at addressing this gap by designing and evaluating tools sensitive to functional abilities in people with DEEs (https://inchstoneproject.org). For example, a pilot study of The Inchstone Project evaluated 10 individuals with SCN2A-related DEE using five measures of development and behavior, ⁷⁷ including norm-referenced tools like the Developmental Profile, Fourth Edition (DP4), and measures of responsivity used in the context of coma and severe brain injury, including the Coma Recovery Scale, Pediatric Version (CRS-P). Norm-referenced measures often showed floor effects, with individuals commonly receiving the lowest attainable scores. However, raw scores showed a greater range and variability, suggesting they may have utility for detecting subtle progress. Similarly, measures of responsivity like the CRS-P did not show floor effects when using raw scores, suggesting the potential to repurpose tools from other clinical domains to capture meaningful changes in people with DEEs.

Another area of unmet need that has emerged from The Inchstone Project is the impact and frequency of cortical visual impairment (CVI) in individuals with DEE. The DEE Parents Speak survey found that a diagnosis of CVI was confirmed or suspected in nearly half (47%) of the children surveyed, and that greater CVI severity was related to more severe cognitive impairment. Additionally, a recent analysis estimates that less than 14% of those with CVI receive a diagnosis, delaying access to appropriate care and treatment. ⁷⁸ This highlights the need for targeted interventions and assessments to recognize and manage CVI in those with DEEs, and the need to develop tools that measure one function while accommodating unrelated impairments in other domains (e.g., cognitive tests that are not dependent on vision impairments due to CVI).

Conclusions

LGS is regularly described as “catastrophic” or “devastating.” While these terms capture the real and profound challenges faced by individuals and caregivers, they risk overshadowing the substantial progress made in understanding and managing LGS. This symposium highlighted recent advances across eight key areas, including the conceptualization of LGS as a secondary network epilepsy, the innovative use of electronic medical data to perform natural history studies and identify gaps in treatment coverage, and the development of preclinical models to explore pathogenesis and treatment. It also reviewed improvements in evidence-based treatment strategies, spanning pharmacotherapies, neuromodulation, surgery, and preventive approaches, alongside a growing emphasis on non-seizure outcomes and tailored assessment tools to measure meaningful, incremental progress. Together, these efforts mark a turning point toward more comprehensive and patient- and caregiver-centered care.