Using a Large Control Cohort to Benchmark EEG Spectral Abnormalities in SCN2A Developmental And Epileptic Encephalopathy
Abstract number :
3.09
Submission category :
2. Translational Research / 2C. Biomarkers
Year :
2023
Submission ID :
660
Source :
www.aesnet.org
Presentation date :
12/4/2023 12:00:00 AM
Published date :
Authors :
Presenting Author: Melina Tsitsiklis, PhD – Beacon Biosignals
Patricia Fogerson, PhD – Beacon Biosignals; Elise Brimble, MS – Invitae Corp.; Alex Arslan, MA – Beacon Biosignals; Jayne Nerrie, MA – Beacon Biosignals; Kim Laberinto, BSc – Beacon Biosignals; Jay Pathmanathan, MD, PhD – Beacon Biosignals; Nasha Fitter, MBA – Invitae Corp.; Jacob Donoghue, MD, PhD – Beacon Biosignals
Rationale: Recent therapeutic trials for genetic epilepsies have reported treatment-induced changes to background spectral features to indicate efficacy. However, interpreting background spectral patterns requires an understanding of both disease-specific changes and expected developmental changes in background EEG rhythms. We hypothesize that when normalized to age-matched control data, background spectral features could serve as an EEG biomarker in SCN2A developmental and epileptic encephalopathy (DEE).
Methods: Longitudinal EEG and clinical data from individuals with SCN2A variants were collected by the Invitae Ciitizen® platform. 471 recordings from 31 early onset (EO) and late onset/late onset with infantile spasms (LO/LOIS) subjects (ages one day to 16 years) were suitable for spectral analysis. Normative data was assembled from 1247 children without an epilepsy diagnosis or epileptiform features based on neurologists’ review, resulting in 1725 control EEG recordings (ages zero days to 16 years), selected from an EEG database of over 70,000 subjects (Beacon Platform). Relative spectral power was computed via the multitaper method within standard frequency bands. To capture differences from age-matched controls, median relative power was computed across all control recordings, for each band and within age bins. SCN2A spectral feature values using age-matched control median and median absolute deviation.
Results: Differences in SCN2A background spectral features compared to controls were notable in all frequency bands (Figure 1). LO/LOIS SCN2A subjects’ spectral features differed markedly from controls, with greater relative delta power across all ages, and reduced relative power in all other bands compared to both controls and EO subjects. EO subjects exhibited more subtle deviations. Classifying SCN2A vs control was best accomplished with relative delta power (maximum AUC of 0.77). Absence of frontal-occipital alpha asymmetry strongly predicted SCN2A status in older subjects. A mixed effects model including z-scored relative delta power and SCN2A phenotype revealed a positive association between z-scored relative delta power and the probability of reporting gross motor developmental delay (Figure 2).
Figure 1. Relative band-wise spectral power by age for SCN2A and controls.
Figure 2. Effect of deviation from control (z-scored) relative delta power on gross motor developmental delay for SCN2A subjects. represent model predictions and shaded bars represent model uncertainty, with other fixed effects set to their mean value.
Conclusions: We quantified background spectral power changes in SCN2A-DEE and their relationship to gross motor development. SCN2A-DEE appears to disrupt the expected pattern of healthy EEG spectral development. The more abnormally elevated a subject’s delta power is with respect to neurotypical controls, the more likely they are to report a gross motor delay, consistent with the persistence of slower background features correlating with worse clinical outcome in DEE.
Funding: N/A
Translational Research