Abstracts

Rationale: Infantile spasms is a potentially devastating form of epilepsy that typically strikes in the first year of life and is classically accompanied by a chaotic, high-voltage, and asynchronous electroencephalographic (EEG) pattern known as hypsarrhythmia. As the brain networks that mediate hypsarrhythmia are unknown, we set out to use scalp EEG measurements to assess connectivity across brain regions in infantile spasms patients with and without hypsarrhythmia.Methods: Twenty-minute segments of scalp EEG data were collected from 14 patients with infantile spasms (10 with hypsarrhythmia) during waking and sleep, both before and after treatment. Connectivity was measured between pairs of electrodes by identifying the maximum cross correlation within successive one-second epochs of data. Maximum values occurring at zero time lag were removed to reduce the effects of volume conduction, and a partial correlation identified cases where the common reference may have inflated the correlation. Significance was assessed by normalizing the cross correlation value by the expected variance of the calculation and comparing to a baseline distribution generated via permutation resampling (Chu et al. 2012). The strength of the connection between a pair of electrodes was defined as the percentage of one-second epochs that were significant.Results: When the connectivity was assessed independently on the first and second halves of each dataset, the two patterns were consistent with one another, suggesting that stable networks underlie the EEG activity, even in those with hypsarrhythmia. The mean connection strength was greater in infantile spasms patients with hypsarrhythmia, as compared to those without hypsarrhythmia, in the pre-treatment awake (ANOVAN, p<1e-7) and asleep (p<0.05) and post-treatment asleep data (p<1e-6). A paired t-test indicated that the connection strength of individual electrode pairs was reduced following successful treatment (p<1e-11 awake, p<1e-10 asleep, hypsarrhythmia; p<1e-20 asleep, non-hypsarrhythmia). A smaller reduction was seen in subjects that were treated but did not respond (p<1e-3 awake, p<1e-5 asleep, hypsarrhythmia).Conclusions: Overall, this analysis enables us to quantify characteristics of scalp EEG that are not evident through visual inspection, including an assessment of the connectivity between brain regions. The results show higher levels of correlated activity in infantile spasms patients with hypsarrhythmia compared to those without hypsarrhythmia, as well as the presence of persistent brain networks in children with infantile spasms. These findings are contrary to the empirical clinical description of this background pattern as “chaotic,” suggesting that this type of analysis may shed light on the physiological mechanisms underlying hypsarrhythmia and infantile spasms. Additionally, the significant decrease in connection strength seen after treatment suggests a relationship between quantitative measurements of connectivity and the pathophysiological state leading to infantile spasms.