Abstracts

Rationale: Rolandic epilepsy (RE) is the most common form of epileptic encephalopathy, characterized by seizures and cognitive deficits in school-age children that spontaneously resolve by adolescence. The pathophysiology of this disease is unknown, but growing evidence suggests that transient disruption of thalamocortical circuits could cause both the stereotyped, focal, sleep-activated epileptiform spikes observed in the inferior Rolandic cortex, and a paucity of sleep spindles, physiological rhythms associated with sleep-dependent learning. We utilized electrical source imaging to study the distribution of spindle deficits in RE and the relationship with cognitive symptoms. In specific, we hypothesized that: 1) spindle rate would be decreased in the inferior Rolandic cortex; 2) spindle deficits would extend beyond the epileptic cortex; and 3) regional spindle deficits would better predict cognitive dysfunction than focal estimates.

Methods: We obtained high-resolution MRI, high-density EEG, and focused neuropsychological assessments in children with RE during active (n = 8, age 9-14.7 years, 3F) and resolved (n = 10, age 10.3-16.7 years, 1F) stages of disease and age-matched controls (n = 8, age 8.9-14.5 years, 5F). We computed spindle rates (spindles/min) using a validated automated spindle detector applied to the source activity in the inferior Rolandic cortex as well as each label in the Desikan-Killiany FreeSurfer cortical atlas during non-rapid eye movement sleep. The results were compared across groups and to measures of fine motor performance, speech-sound processing, attention, and IQ using generalized linear models and a bootstrap analysis of goodness-of-fit. Among detected­ spindles, we also compared spindle features (power, duration, coherence, bilateral synchrony) between groups.

Results: We found that spindle rate was reduced in the inferior Rolandic cortices in active RE (p = 0.007) but not resolved (p = 0.2) compared to controls. We found no difference in spindle features between groups except for decreased bilateral synchrony between spindles in the active group (p = 0.005) but not resolved (p = 0.1) compared to controls. Compared to controls, spindle rate was regionally disrupted beyond the inferior Rolandic cortices in the active group, involving prefrontal, insula, and posterior parietal regions (p < 0.009 for all; Figure 1). Regional spindle deficits positively correlated with cognitive function (p < 1e-4). Regional estimates were better than focal estimates from the inferior Rolandic cortices in predicting cognitive function across all domains tested (p = 0.02) and significantly improved estimates of motor function (p = 0.001; Figure 2).