Stage 2 Sleep and Intelligence Measures in Autistic Children

Background: Sleep spindles are EEG waves thought to reflect sleep protective mechanisms that inhibit the processing of potentially arousing stimuli. Studies showed that sleep spindle activity correlate with IQ scores in typically developing individuals. Sleep spindles are diminished in children and adults with autism compared to typically developing individuals.

Objectives: We investigated whether the relationship between IQ scores and sleep spindles activity differed in children with autism

Methods:  Thirteen boys with high functioning autism (HFA: 10.2 ± 2.1 years old) and 13 comparison children (COM: 10.2 ± 2.0 years old) were recorded for two consecutive nights. They completed the WISC-III in the morning after night 2. The absolute number of sleep spindles, spindle index (number of spindles /hour) and sigma EEG power (slow: 12-13 Hz, fast 13.25-15.75 Hz) were recorded and computed from frontal (Fp1, Fp2) and central (C3, C4) electrodes during Stage 2 sleep. Results from the two groups were compared using Student t-tests and Mann-Whitney U-tests. Correlations between EEG measures and IQ were tested using Pearson’s rho (alpha=.05).

Results: There were no significant group differences on IQ (HFA: Global =105.2±18.7, Performance=106.2±13.0, Verbal=103.8±22.3; COM: 115.8±10.3, 114.1±12.1, and 115.1±12.8, respectively). Spindle number and density were lower in the HFA group than in the COM group at the Fp2 recording site (669.7±467.3 and 126.8±87.1 vs. 1018.8±466.4 and 216.2±121.2). The HFA group showed a negative correlation between spindle density at the C3 recording site in the first quarter of the night and the WISC global (r=-0.52) and verbal (r=-0.62) IQ scales. Sigma power was significantly lower in the HFA than the COM group for C3 and C4 recording sites in the last quarter of the night (HFA: C3=0.925±0.096, C4=0.650±0.226; COM: C3=1.06±0.18, C4=0.888±0.301). The COM group showed a positive correlation between fast Sigma activity at the C4 recording site in the last quarter of the night and global IQ (r=0.592). There were no significant correlation between IQ and Sigma activity in the HFA group.

Conclusions:  These findings indicate that the relationship between sleep EEG and IQ is different in autistic and typically developing children, both with normal IQ scores. These differences are in terms of scalp location (frontal vs. more posterior), EEG markers (spindle EEG wave density vs. Sigma activity spectral power), time of night (early vs. late night) and, more strikingly, direction of correlation (positive in COM, negative in HFA). Further analyses with more scalp locations and spectral analysis of sleep spindle are under way in order to better characterize the sleep EEG/IQ relationship.