Authors :
Presenting Author: Michele Jackson, BA – Boston Children's Hospital, 300 Longwood Ave, Boston, MA 02115, USA
Navaneethakrishna Makaram, PhD, MS – Boston Children's Hospital, Harvard Medical School, Boston, MA 02115, USA
Edeline Jean Baptiste, BS – Boston Children's Hospital, 300 Longwood Ave, Boston, MA 02115, USA
Doroteja Dragovic, MS – Boston Children's Hospital, 300 Longwood Ave, Boston, MA 02115, USA
Tanuj Hasija, PhD, MSc – Paderborn University, Paderborn, Germany
Paulina Moehrle, MD Candidate – Boston Children's Hospital, 300 Longwood Ave, Boston, MA 02115, USA
Maurice Kuschel, MSc – Paderborn University, Paderborn, Germany
Saeid Sadeghian, MD – University of Minnesota, Minneapolis, MN 55455, USA
Stephanie Dailey, BA – Boston Children's Hospital, 300 Longwood Ave, Boston, MA 02115, USA
William Bosl, PhD – Boston Children's Hospital, 300 Longwood Ave, Boston, MA 02115, USA
Eleonora Tamilia, PhD – Boston Children’s Hospital, Harvard Medical School, Boston, MA, USA
Tobias Loddenkemper, MD – Boston Children's Hospital, 300 Longwood Ave, Boston, MA 02115, USA
Solveig Vieluf, PhD – LMU University Hospital, LMU Munich, Germany
Rationale:
Seizure forecasting may enable seizure risk monitoring and improved seizure preventive care. We aim to advance the development of prediction biomarkers by evaluating brain connectivity strength across electroencephalogram (EEG) frequency bands in patients with generalized onset motor (GOM) and focal onset motor (FOM) seizures during preictal periods.
Methods:
We included patients with epilepsy, aged 1 month to 21 years who were enrolled (2015-2021) in the long-term monitoring unit at Boston Children’s Hospital and presented with seizures. Seizure onset, offset, and semiology were determined through video-EEG review.
We analyzed standard 19-channel EEG (10-20 montage) and evaluated three 5-min periods before seizure onset: 1) distant preictal (120min before onset), 2) intermediate preictal (62.5min before onset), 3) immediate preictal (5.5min before onset). We filtered the signal (Notch: 60Hz and bandpass: 1-100Hz), detected and excluded artifacts with the cleanEEG1 pipeline. EEG data were average referenced and segmented into non-overlapping 5-s epochs. Functional connectivity (FC) matrices were computed through orthogonalized amplitude envelope correlation between each pair of EEG channels in 5 frequency bands: delta δ [0.5–4Hz]; theta θ (4–8Hz]; alpha α (8–14Hz]; beta β (14-38Hz] and gamma γ (38-80Hz]. FC matrix was averaged across epochs to obtain interval FC for each frequency band. To assess overall connectivity strength, we computed the averages of the network degree distribution to obtain Mean Functional Connectivity (MFC). We performed a two-way mixed ANOVA with preictal periods as within-patient and seizure type as between-patient factors to compare frequency-specific MFC across preictal periods for GOM and FOM seizure groups, followed by post-hoc Bonferroni adjustments.
Results:
From 450 patients enrolled (900 seizures), we included 94 seizures (49 patients, 51% male, median age: 10.9 yrs) with 36 (22 patients) generalized and 58 (30 patients) focal onset motor seizures.
Results revealed a main effect of seizure group in MFC for theta, alpha, beta and gamma (F(1, 92)=7.36, p=0.008, ƞ2=0.074; F(1, 92)=5.28, p=0.024, ƞ2=0.054; F(1, 92)=7.59, p=0.007, ƞ2=0.076; F(1, 92)=6.18, p=0.015, ƞ2=0.063) and a trend towards a seizure group main effect in delta MFC (F(1, 92)=3.76, p=0.056, ƞ2=0.039). There was a trend towards a preictal period main effect in MFC for beta (F(1.73,159.36)=2.63, p=0.083, ƞ2=0.028).
MFC for theta, alpha, beta and gamma were higher in patients with GOM seizures as compared to FOM seizures at all three preictal periods (p=0.008 [CI:0.006-0.041], p=0.024 [CI:0.004-0.051], p=0.007 [CI:0.009-0.058], p=0.015 [CI: 0.005-0.045], Table 1). Interactions between seizure group and preictal period were not significant.
Conclusions:
EEG MFC strength in theta, alpha, beta and gamma frequencies is higher in GOM seizures as compared to FOM seizures and warrants further research into EEG-based biomarkers for seizure prediction.
1Delorme A et al. EEGLAB. J Neurosci Methods. 2004 Mar 15;134(1):9-21.
Funding:
The Epilepsy Research Fund supported this study.