Abstracts

Rationale: Infantile spasms is a potentially devastating form of epilepsy characterized by clinical spasms and a unique electroencephalographic (EEG) pattern known as hypsarrhythmia. Classic hypsarrhythmia is defined by multi-focal, independent epileptiform discharges on a disorganized background activity with asynchronous large amplitude slow waves. Visual identification of hypsarrhythmia is challenging due to wide variability in temporal and spectral characteristics and the existence of several variants of the classic pattern. Quantitative measurements of hypsarrhythmia have the potential to improve the accuracy and objectivity of initial diagnostic testing and to expedite successful treatment, but basic amplitude and spectral characteristics have never been reported. Methods: We quantified EEGs of 21 infantile spasms patients, both before and after treatment, and 21 control patients using four basic measurements: (1) Shannon entropy was calculated for the delta (1-3 Hz), theta (4-7 Hz), alpha (8-12 Hz), and beta (13-30 Hz) frequency bands to describe the irregularity of the time series. (2) The broadband amplitude was assessed by calculating the range in 1-second windows and constructing empirical cumulative distribution functions. (3) The power spectrum of the EEG was calculated via Morlet wavelet convolution and was used to determine the spectral edge frequency-95%. (4) The temporal structure of the data was analyzed by determining the time lag at which the autocorrelation of the amplitude envelope became insignificant. Results: Signal entropy in the delta band was significantly higher in data with hypsarrhythmia when compared to control data (FDR: adj. p < 0.05), with a median value of 7.2 bits. The median amplitudes in hypsarrhythmia were significantly higher than control data and post-treatment data without hypsarrhythmia (p < 1e-6), and the mode amplitude range in hypsarrhythmia was 120-130 µV. Spectral power was greater in the delta and alpha frequency bands in hypsarrhythmia when compared to control data, while post-treatment data without hypsarrhythmia showed suppression of activity in the theta frequency band. The median spectral edge frequency in hypsarrhythmia was significantly lower than post-treatment data in 10/19 channels and control data in all channels (FDR: adj. p < 0.05), with values between 6.5 and 8.5 Hz. Lastly, in the beta frequency band, we found significant temporal correlations at longer time lags for post-treatment data when compared to data with hypsarrhythmia (p < 0.01), which reached insignificance at a median time lag of 30 seconds. Conclusions: We quantified basic characteristics of hypsarrhythmia with four EEG measurements and identified significant differences when compared to control and post-treatment data. These quantitative descriptors of hypsarrhythmia may aid clinicians in diagnostic evaluation by providing objective ranges of values for the amplitude, power spectrum, and temporal properties. This will better inform the patient’s therapeutic plan and therefore has the potential to improve treatment outcome. Funding: This study was funded in part by an ICTS UCI-CHOC Collaborative Grant.