Authors :
Presenting Author: Brian Ji, BS – Massachusetts General Hospital
Peter Hadar, MD, MS – Massachusetts General Hospital; Harvard Medical School
Angelique Paulk, PhD – Massachusetts General Hospital
Mark Richardson, MD, PhD – Massachusetts General Hospital
Sydney Cash, MD, PhD – Massachusetts General Hospital
Pariya Salami, Ph.D. – Massachusetts General Hospital
Rationale:
Prior work has shown that seizures with broad onset often engage thalamic circuits, and neuromodulation of thalamic targets has emerged as a promising therapy for drug-resistant seizures. However, the thalamus is not homogeneous and is instead composed of multiple nuclei, with the specific contributions of these individual nuclei to seizure dynamics remaining unclear. Therefore, understanding nucleus-specific thalamic involvement in seizures is critical for improving neuromodulation strategies and informing clinicians in devising more targeted treatment plans.
Methods: We analyzed 24 patients with epilepsy who were treated with responsive neurostimulation (RNS) targeting either the centromedian (CM) or pulvinar (PLV) thalamic nuclei and who also had prior stereo EEG recordings from the same nucleus. Patients were classified as responders (Engel class I-III) or non-responders (Engel class IV) based on reduction in seizure frequency. Stimulation parameters were compared between responders and non-responders across both groups to assess their impact on treatment outcomes. In total, 64 seizures met our inclusion criteria for seizure duration and type. For each seizure, we constructed functional connectivity networks spanning seizure onset to termination and evaluated the role of the implanted thalamic nucleus (CM or PLV) by computing its network degree. To account for variability in seizure duration, each seizure was normalized to 100 equal time intervals. Connectivity was then examined across three phases: onset, middle (progression), and termination. Differences between responders and non-responders were assessed using the Mann-Whitney U test at each phase.
Results: Stimulation parameters differed between groups. Among patients with CM stimulation, 85.7% of those receiving high-frequency stimulation were responders. However, in the PLV group, only 40.0% of patients receiving high-frequency stimulation were responders. Consistent with these clinical differences, stereo-EEG recordings revealed distinct patterns of thalamic involvement. In non-responders, the CM exhibited high and sustained connectivity across all seizure phases, whereas responders showed significantly less CM involvement (p < 0.05). This pattern was consistent across different seizure onset zones, including frontal, temporal, and parietal regions. Interestingly, the results were reversed for the PLV group. Responders tended to show greater pulvinar involvement within the seizure network, while non-responders showed lower involvement.