Utilizing thalamic deep brain stimulation for an electronic seizure diary in a definite sudden unexpected death in epilepsy case

Abstract

Objective

Sudden unexpected death in epilepsy (SUDEP) is a serious threat to individuals with intractable epilepsies, contributing to premature mortality. Understanding the elusive pathophysiological mechanisms of SUDEP, especially in cases without observable terminal events, remains a crucial area for investigation. This study aimed to shed light on the burden of epileptiform activity preceding SUDEP by utilizing an automated electronic seizure diary derived from a sensing‐enabled thalamic deep brain stimulator (DBS).

Methods

Herein, we present the case of a 57‐year‐old man afflicted with intractable multifocal epilepsy secondary to cortical dysplasia and encephalomalacia resulting from severe traumatic brain injury. Despite an initial successful resection and subsequent resurgence of seizures necessitating DBS treatment, the patient tragically succumbed to SUDEP.

Results

In‐depth analysis of the patient's electronic seizure diary, complemented by data from the sensing‐enabled DBS, unveiled a terminal electrographic seizure. Notably, we observed a significant increase in power within specific frequency bands recorded from the thalamus preceding the terminal event. Furthermore, these heightened band power events displayed a discernible temporal clustering pattern, primarily manifesting during specific morning and evening hours. An autopsy conclusively confirmed the diagnosis of definite SUDEP.

Interpretation

This unique case report underscores the feasibility of harnessing thalamic DBS sensing capabilities to monitor seizure burden and, potentially, to tailor interventions aimed at reducing seizure frequency and associated mortality risks.

Introduction

The leading cause of premature mortality in patients with refractory epilepsy is sudden unexpected death in epilepsy (SUDEP). ¹ The pathophysiological mechanisms underlying SUDEP remain a subject of debate, primarily because terminal events often occur without witnesses and typically take place during nighttime. ² However, insights gleaned from witnessed and monitored cases suggest that SUDEP frequently ensues in the postictal phase following tonic–clonic seizures, with apnea, bradyarrhythmia, and asystole emerging as plausible mechanisms contributing to this tragic outcome. Nevertheless, there have also been reported instances of SUDEP occurring without the presence of a terminal seizure, further underscoring the need for additional observational studies to comprehensively delineate the events preceding SUDEP. ³

In the realm of epilepsy management, intracranial neuromodulation techniques such as responsive neurostimulation (RNS) and deep brain stimulation (DBS) are increasingly being recognized as established therapies for individuals with intractable epilepsies. It is important to note that this particular subgroup carries the highest risk of SUDEP due to the higher burden of seizures that failed multiple antiseizure medications and surgery. Leveraging the capabilities of intracranial EEG sensing through devices like the RNS, previous studies have shed light on the occurrence of seizures before SUDEP.4, 5 In this case report, we present the first‐ever documented instance of changes in seizure burden, as recorded through sensing‐enabled thalamic DBS preceding SUDEP. This case was confirmed as a definite SUDEP through postmortem analysis. ⁶ Our findings hold the potential to significantly impact our understanding of SUDEP and provide valuable insights into the role of thalamic EEG changes as a precursor to this devastating condition.

Case Presentation

A 57‐year‐old ambidextrous gentleman presented with intractable epilepsy since he was 12 years old due to structural etiology (cortical dysplasia) that was complicated by severe traumatic brain injury that required emergency left craniotomy for intracranial hemorrhage. He has two types of seizure: (a) focal motor seizures with semiology of left arm tonic contraction, right hand dystonia, and chapeau de gendarme; and (b) focal‐to‐bilateral tonic–clonic seizures. A presurgical workup in 2020 including a 3 tesla MRI revealed extensive cortical abnormality that was suspected as dysplasia extending the left superior frontal gyrus, insula, operculum, and encephalomalacia in the left superior and middle temporal gyrus. Wada test indicated predominantly right hemisphere language representation. The patient was then discussed at the multidisciplinary epilepsy patient management conference, and the consensus was to perform left‐sided invasive stereo‐EEG (sEEG) intracranial monitoring with the hypothesis of left frontal operculum, insular, and cingulate regions.

The sEEG procedure (5 days) involved the implantation of electrodes in the targeted areas, including the anterior, middle, and posterior superior insula, posterior prefrontal, anterior and middle cingulate, presupplementary motor, supplementary motor area, anterior insula, temporal pole, and middle and lateral orbitofrontal regions (Fig. 1A). During the interictal period, there were abundant synchronous interictals distributed in the left frontal, middle cingulate, and lateral frontal areas, and independent interictals in the anterior cingulate and posterior prefrontal regions. Additionally, abundant interictals were noted in the insular‐frontal opercular network and mesial and lateral temporal regions. The patient had 55 discrete seizures, captured during the sEEG, and one status epilepticus lasting over 15 min. The consistent sEEG onset pattern for the discrete seizures includes a high‐amplitude spike over a broad region involving the left anterior and middle superior insula, anterior and middle cingulate, and lateral orbitofrontal regions followed by low amplitude fast activity. Notably, the involvement of the orbitofrontal region was rapid within a second (Fig. 1B). The sEEG onset pattern for status epilepticus was distinct, with periodic discharges on the middle basal temporal region, amygdala, mid‐hippocampus, temporal pole, and middle inferior insula. Overall, the study indicated a broad insular‐frontal opercular ictal network, corresponding to extensive brain abnormality observed on MRI, with fast orbitofrontal recruitment and the temporal region as additional epileptogenic region.

Figure 1.

(A) Electrode placement plan diagram for intracranial phase 2 sEEG. AC – anterior cingulate; AIN, anterior insula; AMY, amygdala; ASIN, anterior superior insula; LOF, lateral orbitofrontal; MBT, middle basal temporal; MC, middle cingulate; MH, mid‐hippocampus; MIIN, middle inferior insula; MOF, middle orbitofrontal; MSIN, middle superior insula; PPF, posterior prefrontal; PSIN, posterior superior insula; PSM, presupplementary motor; SMA, supplementary motor area; TP, temporal pole. (B) An example of seizure onset captured during the intracranial phase 2 sEEG with the initial spike on MSIN and PSIN (red arrows), and early involvement of MOF and LOF (blue arrows).

He was offered palliative resection involving the left frontal operculum and anterior insular region, and pathology confirmed focal cortical dysplasia (FCD I). Before the surgery, the patient experienced multiple focal seizures every day. Following the resection, there was a notable improvement, with only five seizures reported at the 6‐month postresection follow‐up visit. His antiseizure medications were continued with minor changes. However, almost 1.5 years postsurgery, his seizures returned and gradually increased to every 1–2 weeks focal seizures with impaired awareness. He lives alone, prompting him to set up safety measures, including a Life Alert system for seizures and consideration of alternative treatment options. A 72‐h ambulatory EEG recorded the occurrence of two prolonged electrographic status epilepticus episodes, each with a left temporal onset, lasting 20 min and occurring during sleep. Based on these findings, in our multidisciplinary patient management conference, a consensus was reached to offer DBS targeting the left pulvinar (Pul) and left centromedian (CMN) nuclei. The rationale for selecting the thalamic targets was driven by an epileptogenic network involving temporal and frontal‐cingulate regions lateralized to the left hemisphere only and the history of motor seizures, including bilateral tonic–clonic seizures. DBS initial parameters were set to a frequency of 145 Hz, a pulse width of 90 μs, and a current amplitude of 2.0 mA, delivered continuously. DBS sensing (Medtronic Percept) was enabled with power in a band (PIB) 12.15–17.15 Hz (Pul) and 16.05–21.05 Hz (CMN) (Fig. 2A), which confirmed recording seizures as reported by the patient and ambulatory scalp EEG.

Figure 2.

(A) The left pulvinar (Pul, power in the band [PIB] 12.15–17.15 Hz) and centromedian (CMN, PIB 16.05–21.05 Hz) nuclei recorded events including the terminal event (TE) which is characterized by a noticeably elevated band power, measured in Arbitrary Units (AU). (B) The scatter plot illustrates the hourly distribution of increased band power events throughout the day. Notably, distinct clusters are observed during the morning hours of 4–5 AM and the evening hours of 4–6 PM, and an arrow points to the TE. (C) The event count histogram shows a significant increase in events/seizure clusters in the weeks preceding the terminal seizure (SUDEP). (D) Polar histogram showing the probability density of seizure occurrence with respect to the hour of the day. p‐value shows quantifies significance of seizure clustering and is calculated using Hodges‐Ajne test for nonuniformity of circular data.

Analysis of his phone seizure record indicated frequent seizures in clusters, with the DBS local field potential PIB peaks around 4 to 6 PM and 4 to 5 AM (Fig. 2B). Unfortunately, 2 months postimplantation, the patient was found dead, prone on the bathroom floor, with his head propped on the bathtub during a welfare check. A review of the automated seizure diary from the DBS sensing confirmed a terminal seizure recorded at 1:44 PM. The manifold increase in the power observed in both centromedian and pulvinar thalamic recordings suggests a potential motor seizure (Fig. 2A). Autopsy confirmed the diagnosis of definite SUDEP by excluding other possible causes. ⁷

Discussion

This case study represents a unique and significant contribution to the field of SUDEP research, employing ambulatory electrophysiological data obtained from thalamic DBS. To define these events, we used a threshold criterion: power within the frequency band of 12.15–17.15 Hz exceeding 4,011.1 Arbitrary Units (AU) (z‐score = 3.2) recorded from the left Pul nucleus (Fig. 2A). The threshold was selected as 3.2 z‐score, with respect to the means and standard deviation computed from the entire 62 days of available recording. Notably, we found terminal seizure recorded from the left pulvinar and centromedian thalamic nuclei, as evidenced by significantly increased PIB 12.15–17.15 Hz and 16.05–21.05 Hz, respectively.

Our patient did not use the magnet to mark events because he lived alone and relied on a phone diary instead. Throughout the 3‐week baseline period (with sensing activated but stimulation turned off) and the subsequent treatment period leading up to the fatal seizure, we recorded the occurrences of increased band power events (Fig. 2A). A noteworthy observation was the significant rise in the number of these events leading up to the terminal event (Fig. 2A–C). Furthermore, we observed that these increased band power events exhibited a significant temporal clustering pattern (Hodges‐Ajne p‐value <2 × 10⁻⁵), predominantly occurring during the morning hours of 4–5 AM and the evening hours of 4–6 PM, as illustrated in Figure 2D. Recognizing these temporal patterns could potentially pave the way for tailored therapeutic interventions aimed at preventing SUDEP.

Similar to our data, an analysis of SUDEP cases utilizing RNS electrocorticograms (ECoGs) revealed an escalating burden of epileptiform activity or seizures occurring within days to months preceding SUDEP, albeit in a limited number of cases. ⁴ Increased seizure burden and clustering of seizures have been identified as a risk factor for SUDEP. ⁸ Likewise, in our patient's case, there was clear evidence of, multiple seizures occurring within a condensed timeframe (a seizure cluster) and a higher overall seizure burden in the weeks leading up to SUDEP (Fig. 2C). The findings also emphasize the importance of monitoring seizure burden and patterns as potential risk factors for SUDEP.

This case study underscores the potential value of utilizing an automated ambulatory electronic seizure diary obtained from thalamic DBS in guiding the adjustment of antiseizure medications to minimize or prevent subsequent seizures. It is important to note that, unlike the RNS device where ECoGs can be readily visualized, the current Percept DBS records power in band data that may not be easily interpretable for practicing epileptologists. Previous studies involving thalamic stereo‐EEG and thalamic DBS have confirmed that recordings of both focal seizures and focal‐to‐bilateral tonic–clonic seizures are attainable from the centromedian and pulvinar thalamic regions. ⁹ Our study adds to the growing body of evidence demonstrating the benefits of using DBS data to manage epilepsy.10, 11

In summary, this case report demonstrates the viability of utilizing an electronic seizure diary derived from thalamic DBS to guide therapy. The observed temporal clustering of seizures underscores the potential for tailoring therapeutic interventions to prevent future seizures in a personalized manner.

Author Contributions

SP was involved in original draft preparation, conceptualization, visualization and supervision. SL was involved in visualization, review and editing, and supervision. SS was involved in original draft preparation, data curation and visualization. YV and VV were involved in conceptualization and methodology.

Funding Information

No funding information to disclose.

Conflicts of Interest

No conflicts of interest to disclose.

DATA AVAILABILITY STATEMENT

Data has been peer reviewed.