Abstracts

Rationale: Infantile spasms is a severe form of epileptic encephalopathy. Its diagnosis is traditionally based on the presence of epileptic spasms and hypsarrhythmia on EEG, though the determination of hypsarrhythmia is imprecise (Hussain et al., 2015). The presence or absence of hypsarrhythmia does not predict response to treatment (Demarest et al. 2017). Recent studies reported that high frequency oscillations (HFOs) were observable on scalp EEG, particularly in young children with drug-resistant epilepsy (Kobayashi et al., 2015; Bernardo et al., 2018). Other studies reported that a higher degree of cross-frequency coupling between HFOs and delta frequency on invasive recordings frequently localized the seizure-onset zones (Nariai et al., 2011; Nonoda et al., 2016). This study determined if scalp-EEG measured interictal HFO rate and modulation indices (MI, reflecting the strength of cross-frequency coupling) were associated with (1) presence or absence of active epileptic spasms, and (2) response among children with epileptic spasms.  Methods: We retrospectively identified children who underwent overnight video-EEG evaluation of suspected infantile spasms based on clinical history. For each subject, a single 10-minute epoch of sleep EEG sampled with 2000 Hz and devoid of obvious artifact, was extracted. Interictal HFO (80-500Hz) rate (/min) and MIs (quantifying the coupling between HFO and 3-4Hz and that between HFO and 0.5-1Hz) were determined at the whole brain level using an automated HFO detector using the RMS method (Staba et al., 2002) and MI algorithm (Canolty et al., 2006). We compared HFO rate as well as MIs between (1) children with and without epileptic spasms, and (2) responders and non-responders, with response defined as short-term resolution of epileptic spasms. Results: We identified 30 children, of whom 23 exhibited active epileptic spasms during the overnight recording. Seven children exhibited no epileptic spasms, one of whom had a previous history of epileptic spasms. The remaining six had non-epileptic events alone (e.g., gastroesophageal reflux) and were neurologically normal. Children with epileptic spasms had higher HFO rate and MIs (p < 0.05 for all indices; Table 1) compared to those without spasms. Using a logistic regression model, HFO rate turned out to be the model with the highest predictive performance for active epileptic spasms (R square: 0.60; AUC: 0.96) with sensitivity of 0.83, specificity of 1.00, and accuracy of 0.87 with a cut-off value of 1.13/min or higher (Table 2). Eight children with active epileptic spasms (2 responders and 6 non-responders) had serial EEG before and after treatment (ACTH, prednisolone, and/or vigabatrin). Responders showed a reduction in the HFO rate and MIs on the follow-up EEGs. Conversely, non-responders showed variable changes (reduction in 2; increase in 2; no consistent changes in 2). Conclusions: Children with active epileptic spasms may show increased interictal HFO rate and cross-frequency coupling between HFO and slow wave. Automatic measurement of these HFO-related measures is feasible on scalp EEG and has the potential to provide a useful biomarker to monitor treatment response in children with infantile spasms. Funding: Susan S. Spencer Clinical Research Training Scholarship in Epilepsy