Abstract
Bullinger KL, Crudele AN, Morrell MJ, Johnson L. Epilepsia Open. 2025; 10: 1099-1110. https://doi.org/10.1002/epi4.70069 Objective: To leverage RNS® System chronic thalamic EEG recordings to compare ictal and interictal thalamic electrophysiology in idiopathic generalized epilepsy (IGE) and Lennox–Gastaut Syndrome (LGS). Methods: As part of ongoing clinical trials in IGE and LGS, intracranial EEG data were acquired from the centromedian nucleus (CM) bilaterally. Ictal recordings were evaluated by trained epileptologists for preictal and ictal onset patterns. The number of detections and the frequency characteristics of interictal and ictal activity were compared across groups. Results: Ictal activity was clearly observed in the CM in IGE and LGS. Detections of interictal epileptiform activity and electrographic seizures were more frequent in LGS. Interictal frequency peaks had similar characteristics across cohorts, although low frequencies were more frequent in LGS. All participants with LGS had 2 or more ictal onset patterns. Most often, onset patterns were similar bilaterally. However, 20% of participants had some seizures with different onset patterns on the right and left, and 50% had some seizures that were unilateral. Low voltage fast activity was the most common onset pattern, followed by bursts of polyspikes and then rhythmic <13 Hz activity. In IGE participants, rhythmic <13 Hz activity was the most common ictal onset pattern; spike and wave was the next most common onset pattern. Two additional onset types were observed: polyspike onset and high amplitude delta followed by low amplitude fast activity onset. Onset patterns were generally consistent within an individual; only 20% of participants had multiple seizure onset types. Significance: Differences in interictal background are consistent with the known phenotypes of IGE and LGS. Furthermore, CM recordings of ictal activity in both cohorts support the feasibility of closed-loop stimulation, with detection settings programmed according to each individual's seizure onset pattern.
Commentary
The centromedian (CM) nucleus of the thalamus is a highly interconnected hub linking widespread cortical and subcortical networks. 1 With its growing role as a therapeutic target for multifocal, diffuse, frontal, and generalized epilepsies, understanding its electrophysiological activity in the context of epilepsy is critical.2,3 The paper by Bullinger et al. 4 reports recordings from patients enrolled in the NAUTILUS IGE study or the LGS study with RNS devices targeting the CM, aimed at identifying background and ictal signatures in generalized epilepsies. The recordings were included from 15 patients with Lennox–Gastaut syndrome (LGS) and 16 with idiopathic generalized epilepsy (IGE) for background activity, and 10 with LGS and 10 with IGE for ictal activity, with each patient contributing up to 10 seizures (or long episodes when fewer were available). In total, 96 seizures from LGS and 57 from IGE were analyzed.
Background Recordings
Baseline recordings showed that the CM exhibits oscillatory peaks across delta, theta, sigma, beta, and gamma bands, with circadian modulation. Overall, there were no significant differences in frequency, normalized power, or bandwidth of the identified peaks between the two epilepsy groups. This suggests that the CM has inherent oscillatory properties relatively independent of epilepsy subtype.
In both IGE and LGS, beta peaks were more common during the day while sigma peaks at night. Given the CM's inputs from the brainstem reticular activating system, circadian modulation is not unexpected, and its connections to frontal and prefrontal cortices may explain beta and gamma peaks. Importantly, background activity was averaged over 90-second windows at scheduled times (2
Ictal Recordings
The authors appropriately address the fundamental question of whether seizure-related activity can be observed in the CM in generalized epilepsy which is an important consideration for closed-loop systems, as stimulation is delivered only after detection of a “clear change” in pattern. Their findings confirm that CM activity does indeed change with cortical seizure onset patterns. This question had been difficult to answer in the era before thalamic stereo-EEG and CM implants, but their results align with what is increasingly being reported.
Seizure signatures reflected in the thalamus differed between LGS and IGE. In LGS, more than two seizure onset patterns were observed, most commonly low-voltage fast activity followed by polyspikes, consistent with the multiple seizure types and onset patterns characteristic of this syndrome. Five of the ten patients had at least one unilateral seizure, and two had undergone prior callosotomy. LGS is considered a combined focal and generalized epilepsy. If lateralization within the CM reflects a dominant epileptogenic network, it raises the question of whether some patients might also be candidates for resective surgery or hybrid procedures, provided the involved regions are amenable to resection.
In IGE, the most common ictal onset pattern was <13 Hz sharp or rhythmic activity in 53% of seizures, typically followed by spike-wave discharges. Because such low-frequency activity is not a common cortical onset pattern, one might hypothesize that in IGE the CM primarily serves an oscillatory pacemaker role, whereas in LGS it may be directly involved early in seizure initiation and propagation. If the study presented cortical signatures alongside the thalamic signature in LGS, this would have been useful to the reader, especially since strip recordings were available. This would have added to the already growing body of data from thalamic EEG and signatures from different thalamic nuclei as they related to cortical seizure patterns.
Electroclinical Dissociation
A longstanding challenge in generalized epilepsy is the abundance of interictal discharges without clear behavioral correlates, making them difficult to differentiate from ictal events. In this study, magnet swipes rarely confirmed seizures, resulting in a degree of electroclinical dissociation. Some thalamic detections may therefore represent subtle absences or myoclonic seizures that patients do not recognize, or alternatively, represent interictal activity. However, restricting therapy only to clearly identified ictal events risks undertreatment when considering the potential network-level effects of chronic neuromodulation. With the advent of other ultra–long-term ambulatory EEG, such as sub-scalp, sub-galeal systems and trials designed to study them, the question of whether EEG as an objective measure alone is sufficient to gauge outcomes is often raised. However, it remains important to correlate patient-reported events with objectively detected ictal activity and to view these measures as complementary rather than singular indicators of outcomes. This approach is critical for accurately characterizing treatment effects and seizure burden.
Limitations and Future Directions
The authors acknowledge that the study design did not permit interpretation of synchronicity between the left and right CM in LGS, as these patients had two devices, one controlling the cortical strip and one the thalamic strip, operating independently.
Because stimulation is delivered in response to detections, patients with LGS are likely to receive stimulation more frequently than those with IGE. The implications of this differential stimulation burden for seizure outcomes warrant investigation. Preictal- or ictal-triggered stimulation targets pathological activity more selectively, and because the CM may play a role in regulating and coordinating normal cortical and subcortical functions, responsive or adaptive neurostimulation may be preferable if comparable therapeutic benefits can be achieved while minimizing disruption of essential physiological rhythms.
Interpretation of these findings should account for the limitations of anatomical targeting. Although the authors used the THOMAS atlas to verify CM localization, the CM is a heterogeneous structure with functionally and histologically distinct parvocellular and magnocellular regions. 1 While at least one contact per patient was confirmed within the CM, recordings were obtained from multiple contacts, making it likely that adjacent non-CM thalamic nuclei, were also sampled. Moreover, the contact providing the clearest oscillatory signals may not correspond to the most effective stimulation site. 5 Future studies integrating multimodal data including individualized imaging, structural connectivity, and electrophysiological analyses will be essential for linking these neural signatures to precise anatomical targets and network-level dynamics, ultimately guiding optimal outcomes.
This paper confirms the presence of ictal signatures at seizure onset. However, the presence of a signature does not always translate into clinical response, underscoring the complexity of thalamocortical networks and the need for further study. 5 It will be particularly important to determine whether specific signatures correlate with outcomes, which is a key question from the larger trials. Nonetheless, the central message is clear: the CM consistently exhibits both interictal and ictal signatures in generalized epilepsies. These findings further support its role as a network hub and its continued use as a neuromodulation target.
