Therapies of last resort in paediatric super-refractory status epilepticus: a case report on combined selective amygdalohippocampectomy, vagus nerve stimulation, and anterior corpus callosotomy

Abstract

Super-refractory status epilepticus (SRSE) is a rare and life-threatening neurological emergency associated with substantial morbidity and mortality. Defined as persistent or recurrent seizures lasting beyond 24 h despite anaesthetic and antiseizure pharmacotherapy, SRSE poses significant therapeutic challenges. Management strategies aim to suppress seizures, provide neuroprotection, and mitigate excitotoxic injury. While emergency epilepsy surgery is traditionally reserved as a last resort due to limited supporting evidence, it has demonstrated efficacy in selected cases. We report the case of an 8-year-old girl with cerebral palsy, drug-resistant epilepsy, cortical visual impairment, and bilateral hearing loss, who developed SRSE unresponsive to maximal antiepileptic and anaesthetic therapy. Following multidisciplinary evaluation, a sequential multimodal neurosurgical approach was implemented, comprising left selective amygdalohippocampectomy (SAH), vagus nerve stimulation (VNS), and anterior corpus callosotomy. SAH targeted the epileptogenic mesial temporal structures while preserving surrounding neocortex and neurovascular anatomy, resulting in immediate seizure cessation. Subsequent VNS implantation and callosotomy were performed following recurrent episodes, achieving sustained reduction in seizure frequency and severity, with Engel class II outcome. This case underscores the potential role of integrating resective, neuromodulatory, and disconnection procedures in paediatric SRSE when conventional treatments fail. Early consideration of sequential surgical strategies may enhance seizure control and improve quality of life in this high-risk patient group.

Introduction

Status epilepticus (SE) is a critical and potentially life-threatening neurological emergency. Its most severe forms, refractory status epilepticus (RSE) and super-refractory status epilepticus (SRSE), are defined by persistence of seizure activity beyond 24 h despite the initiation of anaesthetic treatment [1, 7, 11, 12]. This phenomenon is characterised by abnormally prolonged seizure episodes resulting from impaired mechanisms responsible for seizure termination, spread, or recurrence [1]. The annual incidence of SRSE is estimated at 1.2–3 per 100,000 individuals [12]. Despite maximal antiepileptic therapy and administration of general anaesthetics, 23–43% of patients progress to RSE, and of these, 11–43% evolve into SRSE, in which pharmacological treatment fails to control seizure activity. SRSE is most frequently observed in patients with drug-resistant epilepsy (DRE), affecting approximately 30% of individuals with epilepsy, particularly those with generalised seizures [1]. In the paediatric population, SRSE carries a mortality rate of 30–50%, which may reach 60% in cases of prolonged seizure duration [1]. The risk of recurrence after an initial episode of SE is also considerable, approaching 20% [14], and is highest among patients with underlying structural or metabolic disorders. At present, the management of SRSE remains unstandardised and highly individualised, often relying on high doses of sedatives and antiseizure medications (ASMs), as well as adjunctive strategies such as ketamine, immunotherapy, or ketogenic diet [3, 8, 12]. Although epilepsy surgery is a well-established treatment in chronic drug-resistant epilepsy, it is rarely employed in the acute context of SRSE, owing to concerns regarding feasibility and outcomes under emergency conditions. Nevertheless, growing evidence suggests that early neurosurgical intervention may arrest ongoing seizure activity, limit excitotoxic injury, and reduce intensive care-related morbidity in carefully selected patients [15]. Despite this, surgery is still typically regarded as a therapy of last resort, often undertaken only once pharmacological strategies have failed and when the cascade of cerebral damage—neuronal necrosis, gliosis, and network reorganisation—has already occurred. The case represents one of the few reports worldwide of the management of SRSE employing a combination of advanced neurosurgical techniques: selective amygdalohippocampectomy, VNS, and anterior corpus callosotomy within a single patient, resulting in near-complete seizure remission, corresponding to Engel class II outcome. To our knowledge, the sequential application of all three approaches in a single paediatric SRSE patient represents one of the small number of descriptions in the literature. This case description contributes to the growing body of evidence advocating for earlier consideration of surgical interventions in carefully selected children with SRSE.

Case report

An 8-year-old Caucasian girl with severe neurodevelopmental impairment was admitted to the paediatric neurosurgery unit after transfer from another hospital due to SRSE. Her medical history was significant for spastic quadriplegic cerebral palsy, DRE, cortical visual impairment, and bilateral hearing loss. She was PEG-fed due to dysphagia and malnutrition, and her chronic antiseizure therapy included clobazam, sodium valproate, and vigabatrin. One month prior to admission, she had undergone adenotonsillotomy, after which her parents observed a marked deterioration in seizure control and neurological status.

She was initially brought to intensive care with increasing seizure frequency, characterised by myoclonus, forced left gaze deviation, perioral cyanosis, and hypersalivation. On admission, she was critically ill, cachectic with a BMI of 12.2, and had thoracolumbar neurogenic scoliosis with chest wall deformity. Pulmonary examination revealed coarse crackles in the right lung base suggestive of pneumonia and tachypnoea of 30 breaths per minute. Cardiovascular examination demonstrated tachycardia with hypotension at 75/40 mmHg. On admission she required passive oxygen supplementation for acute respiratory insufficiency. Neurologically, she was unresponsive, reacting only to painful stimuli, with bilaterally dilated pupils that were sluggishly reactive to light. Her gaze was persistently deviated to the left, and she exhibited generalised spastic tetraparesis with superimposed myoclonic jerks most prominent in the lower limbs. Psychomotor development was severely delayed, corresponding to the level of a 5–6-month-old infant.

CT of the head showed no focal lesions or evidence of raised intracranial pressure. Blood and urine cultures were negative, and viral and bacterial panels excluded CMV, EBV, Mycoplasma pneumoniae, and Borrelia burgdorferi. EEG demonstrated diffuse epileptiform activity. MRI was performed on a 3 T system using an epilepsy-specific protocol to facilitate localisation of the epileptogenic zone. In the referring intensive care unit, attempts to terminate seizures with benzodiazepines, dexamethasone, and multiple ASMs had been unsuccessful.

The EEG recording was performed using standard electrode placement according to the international 10–20 system. The background activity showed moderate asymmetry between the hemispheres, with more pronounced abnormalities over the left hemisphere, particularly in the left temporal region, consistent with involvement of the left hippocampal structures. Over the left temporal leads, recurrent epileptiform discharges were observed, predominantly in the form of sharp waves and spike–wave complexes, with maximal amplitude and frequency over the left mesial temporal region. These discharges occur both independently and in short clusters, indicating increased epileptogenicity of the left hippocampus. The right hemisphere also demonstrated intermittent epileptiform activity, although less frequent and less well organised compared to the left side. The abnormalities over the right hemisphere were mainly seen as scattered sharp waves and focal slowing, without a consistent focal maximum. In addition to focal epileptiform discharges, the EEG showed episodes of secondary generalisation, during which epileptiform activity spreads from the left temporal region to involve both hemispheres. These generalised discharges presented as diffuse spike–wave or polyspike–wave complexes, occasionally accompanied by generalised background attenuation, consistent with seizure propagation.

The patient was born prematurely and developed cerebral palsy secondary to bilateral thalamic haemorrhage. As a consequence of early structural brain injury, she subsequently developed hippocampal sclerosis, which remained undetected. The epileptogenic substrate was definitively identified after a 3-Tesla magnetic resonance imaging study in epilepsy protocol during her first hospitalisation for SRSE. The precise age at seizure onset could not be reliably established due to severe early neurodevelopmental impairment; nevertheless, the patient exhibited high baseline seizure burden, with up to 20 seizures per day. Seizures were predominantly focal and frequently progressed to secondary generalised seizures. The characteristic ictal semiology included orofacial automatisms (lip smacking, swallowing) and contralateral right upper limb motor manifestations. EEG generalisation correlated clinically with episodes of apnea. In the past, the patient had been exposed to multiple ASMs, including vigabatrin, valproic acid, lacosamide, nitrazepam, and gabapentin, without achieving adequate seizure control.

Following transfer to the neurosurgical ward, thiopental infusion partially reduced seizure activity but failed to achieve cessation. Progressive respiratory failure and pneumonia necessitated intubation and mechanical ventilation for 23 days, alongside broad-spectrum antibiotic therapy with cephalosporins followed by meropenem, vancomycin, and antifungals. Circulatory collapse required vasopressor infusion. Multidisciplinary review led to modification of antiseizure treatment with the introduction of levetiracetam and lacosamide, resulting in partial reduction in seizure frequency but no neurological improvement. Correlation of the clinical presentation with neuroimaging findings, demonstrating hippocampal sclerosis secondary to long-standing epileptiform activity during years of drug-resistant epilepsy management, together with post-haemorrhagic changes, led to the qualification of the patient for selective amygdalohippocampectomy (Fig. 1).

Fig. 1.

a left, b right Preoperative T1-weighted MRI sequence (a); preoperative T2-weighted MRI sequence (b). MRI demonstrates abnormal morphology of the left hippocampus, which in the differential diagnostic workup raised suspicion of either a neoplastic lesion or hippocampal sclerosis

In the initial phase of management, the patient was scheduled for SAH on the basis of a clearly delineated epileptogenic focus within the left hippocampus. Multimodal evaluation, integrating high-resolution MRI, functional MRI, scalp EEG, and prolonged video-EEG monitoring, provided concordant evidence localising the seizure onset zone. The procedure was conducted under general anaesthesia, with intraoperative neurophysiological monitoring (IONM) supervised by an experienced neurophysiologist to safeguard eloquent cortical and subcortical pathways. IONM consisted of electrocorticography and somatosensory evoked potentials (SEP). SEP monitoring was additionally employed to ensure early detection of potential ischaemia affecting cortical or subcortical structures during the procedure.

Intraoperative electrocorticography was performed during SAH using strip and depth electrodes placed over the mesial temporal structures, including the amygdala and hippocampus. Prior to resection, the ECoG demonstrated frequent epileptiform activity originating from the mesial temporal region, with the highest density of abnormalities recorded from the hippocampal contacts. The epileptiform discharges consisted predominantly of sharp waves and spike–wave complexes, occurring both as isolated events and in repetitive bursts. These discharges showed a clear mesial temporal predominance, consistent with a hippocampal epileptogenic focus. Intermittently, the epileptiform activity exhibited propagation beyond the mesial temporal structures, spreading to adjacent temporal cortical areas, suggesting a network capable of secondary synchronisation. Periods of high-frequency activity were also noted over the hippocampal contacts, further supporting the presence of an active epileptogenic zone. Following resection of the hippocampus and amygdala, repeat intraoperative ECoG demonstrated a marked reduction or complete suppression of epileptiform discharges in the resected mesial temporal region. Residual cortical activity showed improved organisation, with the absence of previously recorded hippocampal spikes. Occasional low-amplitude sharp transients were observed over the lateral temporal cortex, without a consistent epileptiform pattern. Overall, the intraoperative ECoG findings were consistent with a dominant epileptogenic focus within the mesial temporal structures, with effective elimination of the primary epileptiform source following amygdalohippocampectomy.

Resective surgery proceeded without complication, achieving en bloc removal of the hippocampus, uncus, parahippocampal gyrus, and amygdalus. Postoperatively, the patient experienced a single focal seizure; pharmacotherapy was subsequently adjusted by the introduction of phenobarbital and clobazam, with discontinuation of levetiracetam. The girl was monitored for 8 days in an inpatient setting and discharged in a seizure-free condition.

Two weeks after discharge, seizure activity recurred. Given the renewed escalation in frequency and the absence of a clearly delineated epileptogenic focus suitable for further resective intervention, the patient was deemed eligible for second-line therapy with implantation of a vagus nerve stimulator (VNS). The procedure was performed without complications, and the patient achieved sustained seizure freedom for several months thereafter. Vagus nerve stimulation was initiated 2 weeks after device implantation. The stimulation parameters were gradually titrated, with the output current increased by 0.25 mA at intervals of 1–2 weeks. The final stimulation settings were 2.0 mA in normal mode, 2.0 mA for AutoStim, and 2.25 mA for magnet-activated stimulation. The therapy was well tolerated, and no stimulation-related adverse effects were observed.

Four months after her initial admission, she was readmitted with recurrent SRSE, presenting as left-sided tonic–clonic seizures. Prehospital administration of midazolam and clonazepam by her caregivers had failed. On admission, she was unresponsive, tachycardic at 130 bpm, and mildly hypoxic, requiring supplemental oxygen. Her eyes were fixed open with dilated, sluggishly reactive pupils, while neurological status otherwise remained unchanged. Continuous valproate infusion was ineffective, and thiopental was reintroduced. After 3 days, generalised seizures ceased, and status epilepticus resolved. Following the exhaustion of pharmacological strategies and the absence of meaningful clinical improvement, a palliative disconnection procedure in the form of anterior corpus callosotomy was undertaken. The decision was guided by the prior application of neuromodulatory therapy with vagus nerve stimulation and the lack of a well-circumscribed epileptogenic zone precluding further resective surgery. Sectioning of the anterior corpus callosum was performed with the objective of disrupting interhemispheric propagation of epileptiform activity and attenuating progressive involvement of the contralateral limbic circuitry in seizure generation. The postoperative course was uneventful, and a marked reduction in seizure burden was observed, with remaining seizures responding to benzodiazepines. Cenobamate was added to her regimen. She was discharged home in a stable condition with substantial improvement in seizure control and reduced need for emergency interventions. Although she remained severely neurologically impaired due to underlying cerebral palsy, seizure frequency was markedly reduced, achieving Engel II class outcome after 11 months of follow-up, contributing to improved stability in her daily care.

Discussion

This case illustrates the potential role of sequential multimodal neurosurgical strategies in paediatric SRSE, a condition associated with extremely high morbidity and mortality despite aggressive medical treatment. Our patient, who failed maximal antiseizure pharmacotherapy and anaesthetic coma, achieved sustained seizure reduction following a staged approach of selective amygdalohippocampectomy, vagus nerve stimulation, and anterior corpus callosotomy. This appears to be one of the few reported cases combining resective, neuromodulatory, and disconnective procedures in a single patient. Importantly, each intervention provided incremental benefit, resulting in an Engel Class II outcome with substantial improvement in seizure burden and overall clinical stability.

Achieving timely seizure control is crucial for survival and optimal neurological recovery after SRSE [6]. Current treatment remains limited and unstandardised, with robust evidence confined to the early phases of status epilepticus management [15]. First level evidence for refractory or super-refractory stages is lacking, and surgical options are often reserved for selected patients after prolonged disease courses [13, 14]. While epilepsy surgery is well established in chronic drug-resistant epilepsy, its use in the acute SRSE setting remains rare and controversial. Only a minority of 10–20% of patients are considered surgical with even fewer undergoing surgery during SE or SRSE [3]. In two institutional series of 750 and 391 operated paediatric patients with epilepsy, only 15 and 10 patients (2.0 and 2.6%, respectively) underwent epilepsy surgery to treat RSE [5, 8].

Potential neurosurgical strategies for SRSE include focal resections of the epileptogenic zone, neuromodulation—primarily VNS but also deep brain stimulation (DBS) or responsive neurostimulation, and palliative disconnective procedures such as callosotomy or functional hemispherectomy. These interventions may yield significant seizure reduction or seizure freedom in over 90% of paediatric patients [2, 13]. Recent systematic reviews indicate that surgical intervention in refractory and super-refractory status epilepticus can result in immediate termination of ongoing seizure activity in over 80% of cases, with a substantial proportion of patients also achieving sustained long-term seizure improvement [10]. Presurgical evaluation typically involves continuous EEG-video monitoring, high-resolution MRI, neuropsychological testing, and functional MRI. Outcomes are most favourable when imaging reveals a discrete lesion congruent with the electroclinically defined epileptogenic zone, particularly when confined to a single hemisphere [8, 10, 14]. Notably, surgical approach—resective versus non-resective—has not emerged as an independent predictor of outcome, with various strategies demonstrating comparable efficacy [9].

Timing remains a critical determinant of surgical efficacy. Cerebral injury in status epilepticus is primarily driven by excitotoxicity, with neuronal necrosis beginning after hours of sustained activity but potentially occurring rapidly within minutes or taking weeks to take full effect [8]. Emergency neurosurgical intervention may thus exert neuroprotective effects only if delivered before irreversible injury, gliosis, and maladaptive network reorganisation occur. Based on systematic review by Nazi F et al., surgery is recommended after failure of three pharmacological coma trials after 2 weeks [13]. Given the risks of prolonged seizures and intensive care [13], several authors have advocated considering surgery earlier. Preliminary data suggest that timely intervention may improve developmental, neurocognitive, and postoperative outcomes [6, 12].

Resective surgery such as SAH remains the most frequently performed intervention in SRSE. In our patient, correlation between hippocampal abnormality on neuroimaging and clinical semiology justified SAH, which promptly terminated status epilepticus. This supports previous reports that focal resections can achieve immediate seizure control in selected cases [13]. Furthermore, existing studies indicate that resective surgery offers the highest chance of durable seizure control, with seizure freedom rates up to 69% [13]. Such findings underscore the importance of early consideration of surgery when a well-defined epileptogenic focus is present.

The subsequent VNS implantation was guided by recurrent seizures in the absence of a potentially resectable focus. Literature suggests VNS may terminate SRSE in approximately 70% of cases, with some series reporting up to 85% [13, 18]. Despite an incomplete understanding of its mechanisms and optimal stimulation parameters and titration periods, VNS is increasingly recognised as a therapeutic option for SRSE. In our case, VNS was followed by several months of seizure freedom, reinforcing its value as a viable adjunct in multifocal or generalised epilepsies where resection is not feasible. Potential mechanisms may include anti-inflammatory modulation, synchronisation of neuronal networks, and reduction of cortical excitability, thereby restoring more physiological cerebral electrical activity [13].

The final intervention, anterior corpus callosotomy, provided further seizure reduction and improved benzodiazepine responsiveness. As a palliative option, callosotomy can benefit diverse epilepsy syndromes, although its overall efficacy is generally equivalent or slightly inferior to neuromodulation, which may offer a better safety profile [13]. The mechanism of callosotomy is thought to restrict bilateral epileptic synchrony and secondary seizure spread, though the exact pathways remain unclear [9]. The rationale for anterior disconnection sparing the splenium is to preserve interhemispheric transfer of perceptual information while reducing the risk of complications, particularly disconnection syndrome, associated with complete callosotomy [4]. The combined application of resection, neuromodulation, and disconnection, as in this case, is rarely reported but may represent a pragmatic strategy in highly refractory paediatric SRSE.

The therapeutic plan for paediatric SRSE described in our case was devised by an interdisciplinary team of neurosurgeons, a neuropsychologist, and an epileptologist, taking into account seizure semiology, the morphology of the epileptogenic focus, the lack of efficacy of ASMs and analgesics, and the patient’s clinical profile. It represents a patient-oriented therapy, tailored to her clinical characteristics and psychosocial background. An alternative neuromodulatory treatment scheme may involve the use of deep brain stimulation (DBS), either instead of VNS or in cases of its ineffectiveness. DBS can be a plausible option for patients with multifocal or generalised seizures, or for those who are not ideal candidates for resective surgery [13]. It is considered a viable alternative or adjunct to resection in patients with temporal lobe epilepsy. In SE, electrodes are most commonly implanted into the anterior and centromedian nuclei of the thalamus, but also into the subthalamic nucleus, striatum, or globus pallidus [15, 17]. The advantage of DBS over VNS may lie in the ability to stimulate only localised brain regions where electrodes are implanted, as well as the possibility of precisely modulating stimulation parameters in individual areas. In this context, the relative localisation of DBS contrasts with the diffuse neuromodulatory effects of VNS mediated via brainstem pathways, allowing for targeted modulation of specific thalamocortical networks according to epilepsy type and seizure semiology. DBS could therefore be considered a “rescue” therapeutic option. High-frequency stimulation of centromedian nuclei in generalised seizures induces cortical desynchronisation associated with increased vigilance, attentive behaviour, and blocking of epileptic synchronous discharges. Through these mechanisms, centromedian nuclei stimulation may contribute to a reduction in the duration of intensive care treatment, limit ASM associated adverse effects, and potentially decrease the risk of status epilepticus recurrence [16]. However, knowledge regarding the efficacy of neuromodulatory methods, and particularly DBS, remains limited to isolated case reports; therefore, optimistic results should be interpreted with caution. Further research, including randomised trials, is necessary to assess and compare the effectiveness of different neuromodulatory strategies in paediatric SRSE.

Clinically, our case underscores that surgical interventions should not be dismissed as extraordinary or merely palliative in SRSE. In carefully selected children, timely escalation to multimodal neurosurgical strategies may avert irreversible cerebral damage, mitigate iatrogenic complications, and enhance quality of life. Further multicentre collaborations and prospective registries are essential to clarify optimal patient selection, procedural sequencing, and the precise therapeutic window in this devastating condition.

Conclusion

Paediatric super-refractory status epilepticus remains one of the most formidable challenges in acute neurology, with high mortality and profound morbidity despite maximal pharmacological therapy. This report demonstrates that a sequential combination of resective, neuromodulatory, and disconnection procedures can achieve meaningful seizure control when all conventional strategies have failed. Early integration of surgical options may attenuate excitotoxic injury, mitigate intensive care complications, and improve long-term stability, even in patients with severe baseline disability. Systematic evaluation of surgical timing, patient selection criteria, and multimodal treatment pathways is warranted to refine evidence-based algorithms for this critically ill population.

Author contributions

Conceptualization MZ,OM,SK; Validation OM, SK; Formal Analysis OM, ŁK, SK; Writing – Original Draft Preparation - MZ, ŁK; Writing – Review & Editing – SK, OM; Supervision - Project Administration -SK, OM.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethical approval

This is an observational study. The Jagiellonian University Research Ethics Committee has confirmed that no ethical approval is required.

Conflict of interests

The authors declare no competing interests.