Abstracts

Rationale: The cumulative risk of relapse after initially successful treatment for infantile spasms (IS) approaches 40-50%. Although risk of relapse has been linked to the emergence of multifocal spikes, weeks to months after initial response (Hayashi 2016), there are no standard EEG features at the time of response that predict relapse. Methods: We studied 50 patients with IS at the UCLA Mattel Children's Hospital who received a first-line treatment (prednisolone, ACTH, or vigabatrin) and underwent follow-up video-EEG within one month of treatment initiation. Each follow-up EEG was sampled 4 times (2 awake and 2 sleep) and the following quantitative metrics were derived in a blinded fashion: functional connectivity (calculated via cross-correlation), the Detrended Fluctuation Analysis (DFA) exponent and intercept, amplitude (defined as the range of broadband bandpass-filtered data within one-second windows), spectral edge frequency (SEF, calculated by fast Fourier transform), and Shannon entropy (a measure of signal disorder). Time to relapse was evaluated using Cox proportional hazards regression. Results: Among a cohort of 50 children with IS, 32 exhibited 'initial response' defined as freedom from epileptic spasms and hypsarrhythmia on follow-up overnight EEG. Among these 32, 10 exhibited relapse of epileptic spasms. Four relapsed within 1 month and eight relapsed within 1 year of initial response. In sequential univariate Cox proportional hazards regression, time to relapse was associated with higher post-treatment DFA intercept (p = 0.009), higher amplitude (p = 0.02), and higher SEF (p = 0.033). In multivariate analysis, time to relapse was associated with DFA intercept (p = 0.04) and there was a trend for SEF (p = 0.09). As a point of reference, DFA intercept greater than the median and SEF greater than the median were independently associated with an estimated 4-fold and 7-fold increased risk of relapse, respectively. Conclusions: This study suggests that quantitative attributes on EEG at the time of response to treatment-in the absence of clinically-identified hypsarrhythmia-have robust prognostic value and may specifically identify patients at highest risk for relapse. However, our conclusions are limited by the retrospective design and relatively small sample size. Prospective study in a larger cohort is warranted. Funding: This study was accomplished with support from UCB Biopharma, the Elsie and Isaac Fogelman Endowment, the Hughes Family Foundation, and the UCLA Children's Discovery and Innovation Institute