Stereo electroencephalography–guided radiofrequency ablation in focal epilepsia partialis continua: illustrative case

Abstract

BACKGROUND

Epilepsia partialis continua (EPC) is a variant of focal motor status epilepticus that can occur as a single or repetitive episode with progressive or nonprogressive characteristics.

OBSERVATIONS

The authors describe the feasibility of identifying focal EPC in a 33-year-old woman using video electroencephalography (VEEG), electroencephalography source localization, [¹⁸F]fluorodeoxyglucose positron emission tomography, magnetic resonance imaging, and psychiatric and neuropsychological assessments and of treating it with stereo electroencephalography–guided radiofrequency (SEEG-RF) ablation. EPC comprised recurrent myoclonus of the right thigh and iliopsoas with a progressive pain syndrome after left anterior-temporo-mesial resection. Switching between VEEG under regular and epidural block helped to define myoclonus as the presenting ictal symptom with a suspected seizure onset zone in the left parietal paramedian lobule. After the epileptic network was identified, SEEG-RF ablation abolished all seizures. No correlation was found between pain and VEEG/SEEG abnormalities. Rehabilitation began 3 days after the SEEG-RF ablation. By 1 year of follow-up, the patient had no EPC and could walk with assistance in rehabilitation; however, due to the abrupt abolishment of EPC and underlying psychological factors, the patient perceived her pain as overriding, which prevented her from walking.

LESSONS

The application of SEEG-RF ablation is an efficient therapeutic option for focal EPC with special concerns regarding concurrent nonepileptic pain.

https://thejns.org/doi/10.3171/CASE23611

Epilepsia partialis continua (EPC) is a variant of focal motor status epilepticus. The International League Against Epilepsy defined EPC as “a condition of continuously repeated fragments of epileptic motor/sensory seizures with preserved consciousness.”¹ EPC can occur as a single episode or as repetitive episodes and can have progressive or nonprogressive characteristics.² The prevalence of EPC is less than 1 case per million persons,³ with a debatable higher incidence in males.⁴ Electroencephalography (EEG) in adults and children with EPC is usually unaltered by sleep-wakefulness.^{4, 5} Moreover, 11% of patients have a normal EEG.³ EPC can present as somatosensory and/or psychocognitive aura continua.¹ Its etiology can be systemic (metabolic, mitochondrial, autoimmune), focal (stroke, lesional, inflammatory, traumatic), attributable to tick-borne encephalitis, or idiopathic.²

Drugs (steroids, benzodiazepines, topiramate, levetiracetam), immunomodulation, and plasmapheresis have variable effectiveness.^{2, 6, 7} Vagus nerve stimulation,⁸ transcranial magnetic stimulation,⁹ and cortical stimulation¹⁰ have limited roles in treating EPC. Surgery can be performed in patients with EPC to limit generalization.¹¹ It is preferable to perform a hemispherectomy in childhood because of a lower complication rate, brain compensation, and favorable long-term outcomes.¹²

Two case studies have reported the use of invasive recordings for EPC diagnosis.^{13, 14} In the first case, a 14-year-old patient with upper-limb and facial EPC underwent stereo EEG (SEEG). Myoclonus served as the marker for analyzing the SEEG-averaged activity and identifying the precentral-premotor gyri as the seizure onset zone (SOZ).¹³ In the second case, electrocorticography in a patient with Rasmussen-related EPC showed multiple and independent interictal abnormalities.¹⁴ The use of radiofrequency (RF) ablation as a treatment for EPC has never been reported.

Here, we demonstrate the feasibility of performing SEEG to identify the SOZ in a patient with EPC and to exclude pain related to seizures. In addition, SEEG-guided RF (SEEG-RF) ablation was used for treating focal EPC.

Illustrative Case

A blind 33-year-old right-handed woman, who had been diagnosed with drug-refractory EPC 4 years prior, abruptly developed recurrent myoclonus of the right thigh and iliopsoas with seizure-independent progressive pain syndrome after undergoing a left anterior-temporo-mesial resection (LtATMR) at another center. The patient described the pain as deep and severe crushing, twisting, and pinching in the right lower limb. Drugs and LtATMR had no effect on her seizures, which occurred 3–6 times daily. They included an aura of disconnection or confusion, déjà vu, and “feeling distant,” which evolved into right limb and orofacial automatism with altered speech. The seizure was almost constant and repetitive. At the beginning of an event, close observation showed gentle twitching of the anterolateral and medial right proximal quadriceps, which was apparent under the skin. This lasted from 10ths of a second to 1 minute and was accompanied by a typical pattern on EEG. Thereafter, the seizure pattern on the EEG stopped, and small clonic leg movements were observed visually and on electromyography (EMG). The movements showed a correlation with the EMG activity, though this activity did not propagate to other regions. The clonic activity was not considered a seizure but rather a response to the previous seizure stimulation. Despite the patient’s blindness, dark surroundings triggered her seizures.

A comparison of evaluations (video EEG [VEEG] for several days, magnetic resonance imaging [MRI], and a psychiatric evaluation) before LtATMR with those performed 2 years after LtATMR, as part of the pre-SEEG evaluations, showed a significant decline in learning and verbal memory. Intelligence and language remained normal. An adjustment disorder with depressive symptoms was noted. The patient reported severe pain that was both physical and psychological.

A fully digitized VEEG recording (Polaris, NDI) was performed with 32 electrodes (10–10 based). SEEG (DIXI Medical) was done several weeks later with postoperative reconstruction (CURRY version 9.0.0.36, Compumedics Neuroscan). The recordings were amplified by 20,000 and filtered during VEEG and SEEG (low pass [LP] 70 Hz, high pass [HP] 1 Hz, notch 50 Hz; LP 500 Hz, HP 1, notch off) and sampled at 1000 Hz (Neuvo, Compumedics Neuroscan).

EEG was band-pass filtered (1–38 Hz) and downsampled to 250 Hz. Noisy segments were excluded. We identified T5 electrical polarity reversal events and created epochs of −500 to +500 msec with the epoch center (time 0) at the peak of the maximum power of the normalized general field potential (GFP).¹⁵ All epochs were then averaged relative to the peak of activity. The signal-to-noise ratio (SNR) was automatically defined.¹⁶ Independent component analysis (ICA)¹⁷ was presented on a spatial scalp map and as a filtered signal in the time domain. Selected components were fitted with rotating dipole models.¹⁶ A finite element model (FEM; 75,968 triangles, side length 2.0 mm, with normal) was computed for source reconstruction.¹⁸ A source current density (standardized low-resolution brain electromagnetic tomography [sLORETA]) with a 38,025 sources lead field was then constructed.¹⁹

Positron emission tomography (PET) and MRI (Biograph mMR, Siemens Healthineers) were performed while the patient was under general anesthesia following an injection of 5 mCi [¹⁸F]fluorodeoxyglucose (FDG; exposure radiation 3.5 mSv). Anatomical scans were acquired with T1-weighted structural MRI (magnetization-prepared rapid acquisition gradient echo) sequence, 192 slices along the transverse axis, 1 mm isometric, slice resolution 256 × 256).

SEEG was conducted with the patient under general anesthesia using a Leksell G-frame (Elekta; frame registration <1 mm) and the S8 Medtronic navigation system. Five orthogonal electrodes and 3 oblique electrodes were oriented to the paracentral lobule (PCL). Two oblique electrodes were oriented toward the posterior insula (INS) with their external contacts overlapping the paracentral ones. The postsurgical computed tomography scan was co-registered with the presurgical MRI plan, with the least involved electrode within the white matter serving as the reference. SEEG was performed for 14 days. On the 10th day, sensorimotor and speech mapping, as well as provocation stimulations (5 and 50 Hz, 200–500 msec, 6- to 10-second trains with 0.5- to 6-mA intensity), were performed while the patient was taking her regular antiepileptic drugs. Once the epileptic network was identified, RF ablation was done using a bipolar current (120 mA [50 V] applied to adjacent SEEG contacts along a single electrode for 50 seconds, or less if the impedance collapsed to 0). Forty-eight hours later, the SOZ was identified in the PCL with highly involved posterior borders of the previous LtATMR site. The 3rd and 4th conventional stimulations were done for sensorimotor and speech mapping (50 Hz, 500 msec, 6- to 8-second trains with 0.5- to 6-mA intensity) and provocation (5 Hz, 200–500 msec, 8- to 15-second trains with 0.5- to 6-mA intensity).²⁰ Deep brain stimulation (DBS) for alleviation of pain (130 Hz, 60 msec, 1-minute trains with 0.5- to 4-mA intensity) was done in the anterior cingulum and posterior INS. Lastly, to isolate its impact, RF ablation was performed²¹ without antiepileptic drugs.

The patient began rehabilitation 72 hours later. Posterior basic rhythms were normal with no asymmetries. Hyperventilation increased the bilateral fronto/parieto-central rhythmic runs. Intermittent photic stimulation showed no driving responses. Abnormal activity consisted of maximal left posterotemporal slowing.

Epileptiform activity included interictal activity in the left posterotemporal, parietocentral, and frontocentral reversals (P7/T5, P3, T3, CP5, TP9, CP1, and O1), which were propagated rostrally. Ictal activity showed left temporal reversals before quadriceps twitching. Once these had commenced, left pericentral reversals started at the same time along with changes in low-voltage EMG that were locked to the rhythmic movement of the leg. To isolate the centrally driven activity from the potentially chronic myoclonus, an epidural block was administered. We concluded that the twitching was the seizure: it was time-locked to parieto/pericentral reversals (CP5, CP1, C3, P3), which propagated quickly from a previous posterotemporal discharge centered at T5 (Fig. 1A). These discharges resolved concomitantly with the end of the quadriceps twitches, after which nonepileptic clonic activity began (Fig. 1B and C). Pain was independent of any seizure.

FIG. 1.

Seizure onset and end once P3–C3 are drafted and disengaged, respectively. A: Typical onset of a seizure once P3–C3 engages. When T5 reversals spread to P3, showing fast rhythmic P3–C3 activity, a concomitant EMG signal indicates right quadriceps twitches, that is, the onset of a typical seizure (green continuous squares). B: The disappearance of C3 discharges/fast activity occurs in parallel to the disappearance of the twitches, which are followed by the repetitive jerks of the right leg, as can be seen on EMG (green dashed squares). C: The jerks evolve into clonic activity, which is independent of EEG abnormalities (green dashed squares).

Left temporomesial malacia and hypometabolism were seen on FDG-PET, as expected from the previous LtATMR. Additionally, left pericentral hypermetabolism was observed.

The time lock average of 14 reversals suggested a T5-centered source, which was maintained over 44 msec and spread to T3, CP5, and C3 and further to F7, F9, and FC5 (Fig. 1A). The first 2 ICA components depicted the source at T5 (explainability 55.8%; Fig. 2B). A rotating dipole was fit to these components (Fig. 2B) and together with sLORETA (SNR 6.1, explained signal 74.2%) these were superimposed on T1 images (Fig. 2B) and on three-dimensional (3D) cortex reconstruction from T1 scans (Fig. 2D). Overall, a P7(T5)-CP5–suspected SOZ led to a CP5-centered SEEG (Fig. 3B and C) with sentinels to the pain matrix areas (anterior cingulum and posterior INS; Fig. 3D–F). The meticulous lateral entry points of the INS electrodes and the medial entry points of the PCL electrodes enabled minimizing the proximity to the pyramidal tract (Fig. 3A) along Talairach’s E2-3 fields (Fig. 3B and C). A single contact deviation of 2.64 mm from the plan was observed.

FIG. 2.

Interictal source analyses suggest a P7(T5)–CP5 source. A 1-second peak of general field potential time-locked averaged EEG activity (A, upper; based on 14 T5 reversals). Spatial scalp voltage distribution evolution (A, right) suggests a T5-centered source, maintained over 44 msec, propagating to T3, CP5, and C3 and further to F7, F9, and FC5 (red, positive voltage; blue, negative voltage). Morlet wavelet time-frequency power decomposition of the averaged signal (A, lower left) trimmed 1.63–16 Hz, globally normalized) supports relative theta band activity increase at averaged activity peak selectively at T5 but to some extent delayed at TP9. ICA decomposition of the signal (B). The first 2 components depict 55.8% of the entire activity at T5. Source current density (sLORETA) and ICA-computed dipole model (C) fit with an SNR of 6.1 and an explained signal of 74.2%. The dipole colors correspond to the matching colors of the dipole traces in panel B. Plotting of sLORETA current distribution (D) on top of the 3D-reconstructed FEM cortex surface suggests an interictal source at the P7(T5)–CP5 area.

FIG. 3.

The previously resected zone (A, gray area; also seen on the MRI scan) and the contour of the craniotomy for subdural grids (dashed line) are marked. A color gradient scale demonstrates the degree of electrode involvement. Red indicates the primary SOZ (CP5–P3–CP1–C3); orange, secondary SOZ/highly irritative zone (TP9, P7); green, propagation zones (T3, O1). Avoidance of the pyramidal tract (B). Lateral/medial entry points of the INS/PCL electrodes, respectively, were chosen to avoid the pyramidal tract. The main PCL-oriented hypothesis electrodes: PCL SEEG (C) and insular-cingular SEEG (D). Overlapping oblique (ob; FC3/C3/CP3-PCL-ob) and orthogonal (CP3) contacts to optimize diagnosis and RF ablation if suitable. Suspected pain-oriented electrodes: PCL SEEG (E) and pain matrix SEEG (F). Overlapping oblique insular (INS-4-ob and INS-5-ob) and anterior cingulum orthogonal contacts (FC3 [CingAnt] and C5 [Post-INS]). Note that the external contacts of INS-4 and INS-5 are practically overlapping with C3-PCL (asterisk, D) and have further enabled us to merge RF fields. Before (G) versus after (H) SEEG-RF ablation: the 10-electrode SEEG exploration allowed identification of the epileptic network. An interplay between 2 highly active zones was observed. Both had distinctive EEG abnormalities but were time-locked and always preceded the right leg twitches. The primary SOZ shows a recording of the PCL (first 3 traces): a rhythmic and sharply contoured activity involving contacts in both M1 (C3 and FC3 contacts) and S1 (CP3-PCL). The secondary SOZ/highly irritative zone shows a recording of the posterior border of the previous craniotomy (last 2 traces): an almost continuous sharp activity was seen interictally, which further increased before each seizure onset. Both zones were ablated (once functional mapping excluded eloquent faculties) under awake conditions, leading to clinical, EEG, and EMG abolishment of EPC. AG = angular gyrus; Ant = anterior; Cing = cingulum; FG = fusiform gyrus; IZ = irritative zone; PCl = PCL; PHG = para-hippocampal gyrus; Post = posterior.

Two days after the surgery, the epileptic network included 2 SOZs with distinctive EEG abnormalities that were time-locked to the leg twitches (Fig. 3G). The primary SOZ appeared as rhythmic and sharply contoured activity in the PCL (upper 3 traces) and involved contacts in both M1 (C3 and FC3 contacts) and S1 (CP3-PCL). The secondary SOZ/highly irritative zone was identified in the posterior border of the craniotomy (lower 2 traces) and appeared as an almost continuous, sharp interictal activity that increased before each seizure.

Stimulation in the SOZ-involved contacts imposed no restraints concerning eloquence. Both the primary SOZ contacts (FC3-PCL [mostly 8–10], C3-PCL [5–8], CP3-PCL [5–9], external-INS4 [16–18]) and the secondary SOZ contacts (T3/parahippocampal gyrus and TP7/fusiform gyrus) were ablated under awake continuous EEG monitoring. EPC immediately resolved (Fig. 3H) with an expected flaccid dorsal and plantar flexion weakness.

Pain was independent of seizures. Stimulation of pain-related regions showed distinctive responses: posterior INS stimulation (INS-4 and INS-5) reduced pain at 130 Hz only (DBS-like frequencies) following transient increased pain lasting several seconds and an increased heart rate (>100 beats per minute). Anterior cingulum stimulation had no impact.

By 1 year of follow-up, the pre-LtATMR and EPC seizures had disappeared completely, and the patient expressed her satisfaction with this outcome. Although she could walk with assistance and change positions while lying in bed for the first time in 4 years, she feared going home. Both the patient and her caregivers noted significant pain reduction at night; however, while she needed fewer analgesics and had fewer sleepless nights, her pain perception had not changed, probably due to the abrupt abolishment of EPC and other psychological factors. Despite the patient’s request to undergo DBS to alleviate her pain, we concluded that her return home would clarify the need for further intervention.

Patient Informed Consent

The necessary patient informed consent was obtained in this study.

Discussion

Observations

To the best of our knowledge, this is the first report on the use of SEEG-RF for the treatment of focal EPC. We have shown that focal EPC can abruptly occur remotely from surgical sites and in otherwise normal brain regions. An epidural block enabled the dissociation of the right quadriceps myoclonus, which was time-locked to ictal P3 discharges from the clonic nonictal jerks. SEEG also played an essential role in ruling out the possibility that the patient’s concurrent pain was associated with seizures. SEEG exploration confirmed a pericentral SOZ, which was immediately abolished following SEEG-RF ablation.

Evaluation of outcomes in severe cases, such as the one described here, must include epilepsy, rehabilitation, and pain aspects. An Engel epilepsy surgery outcome class 1A was achieved compared to the pre-LtATMR evaluations and EPC seizures. Concerning pain, 2 aspects must be considered: 1) there was a discrepancy between the patient’s subjective feeling that pain levels were unchanged despite her ability to sleep better at night and her decreased use of analgesics. This may be explained by the acute disappearance of all seizures, which renders pain a major role. 2) Despite the patient’s insistence on undergoing DBS for pain, the proximity to the SEEG-RF, her complex comorbidities, and insufficient time within her natural ecosystem have led us to postpone any further intervention at this stage.

Our conclusions are limited to this single case study. Future reports in the literature may improve our understanding of the treatment of EPC using SEEG-RF.

Lessons

EPC can occur abruptly in otherwise normal brain regions and remotely from surgical sites. Focal EPC may be diagnosed using SEEG and treated with SEEG-RF ablation. The investigation of EPC with concomitant pain should address the question of whether these symptoms have an underlying common network or separate neural/epileptic networks. In the case described here, SEEG exploration of EPC with concurrent pain proved that the pain was independent of the seizures, preventing futile extensive RF ablation. The clinical evaluation and follow-up of such complex patients must be holistic and multidisciplinary.

Disclosures

Dr. Getter reported personal fees from Neurohelp outside the submitted work.

Author Contributions

Conception and design: Levy, Abu Arisheh, Lorberboym, Sepkuty. Acquisition of data: Getter, Zer-Zion, Abu Arisheh, Kilani, Lorberboym, Shemesh, Sepkuty. Analysis and interpretation of data: Levy, Getter, Abu Arisheh, Sepkuty. Drafting the article: Levy, Abu Arisheh, Lorberboym, Sepkuty. Critically revising the article: Levy, Getter, Zer-Zion, Abu Arisheh, Lorberboym, Sepkuty. Reviewed submitted version of manuscript: Abu Arisheh, Kilani, Lorberboym, Shemesh , Sepkuty. Approved the final version of the manuscript on behalf of all authors: Levy. Statistical analysis: Abu Arisheh. Administrative/technical/material support: Abu Arisheh, Kilani, Madar, Lorberboym, Sepkuty. Study supervision: Levy, Abu Arisheh, Lorberboym, Shemesh, Sepkuty. Database creation: Mirson.

Correspondence

Mikael Levy: Jabotinsky 115, Tel Aviv, Israel. miko1levy@gmail.com.