Abstracts

This is a Late Breaking abstract

Rationale: Although deep brain stimulation of the centromedian (CM) and anterior (ANT) nuclei of the thalamus are increasingly commonplace in children, it is unclear whether local field potentials (LFPs) from chronically implanted electrodes are associated with seizure burden in the real-world setting. Here, we describe associations between chronic LFPs and seizure burden using the Percept device, which permits long-term recordings.

Methods: Eleven patients with medically refractory epilepsy (aged 5-17) underwent deep brain stimulation (DBS) from February 2021 to May 2022. Each patient was implanted with a Percept PC B35200 neurostimulator. Bilateral leads were implanted in the CM (nine patients) or ANT (two patients) using stereotactic guidance. The patients’ clinical characteristics are presented in Table 1. _x000D_ _x000D_ Following surgery, caregivers were instructed to record seizures based on button presses. In addition, every ten minutes the device saved to memory the spectral power contained in a narrow, 5-Hz band, calculated from the most recent ten minutes of an LFP signal. The center frequency of the 5-Hz band was pre-selected by a clinician. _x000D_ _x000D_ To begin processing the narrowband-power recordings, values above a per-patient threshold were considered artifactual and removed. The logarithm of each recording was high-pass filtered with a cutoff of one week to remove long-term drift. Data were baselined to the daily minimum point. Finally, each minimum point per hour was plotted as a light-colored point (Figure 1C). Using the hourly minimum activity mitigated spike artifacts. _x000D_ _x000D_ This study has been reviewed and accepted by the institutional Research Ethics Board (REB) of the Hospital for Sick Children. Patient consent/assent was obtained in accordance with institutional guidelines (Declaration of Helsinki)._x000D_
Results: A circadian pattern was apparent in the LFP-power time series of both hemispheres of all 11 patients. For illustration, an LFP-power time series covering five days for a single patient is shown in Figure 1A-B. Figure 1C-D show these data plus 23 additional surrounding days, after preprocessing and superposition. Comparing the activity across hemispheres, the LFP activity graphs usually appeared similar but occasionally were quite different, as in the case of Figure 1. _x000D_ _x000D_ In nine of the 11 patients, the LFP power in at least one hemisphere correlated significantly (p < 0.05) with the histogram of counts of symptomatic events. In three such patients, both hemispheres showed positive significant correlation (mean Kendall’s tau-b correlation: 0.46). In two patients, the correlation was positive in the left and negative in the right hemisphere (e.g., Figure 1). In one patient, both hemispheres were negatively correlated with event counts, and in three patients there was exactly one significant correlation._x000D_