Impairment of synaptic homeostasis in ESES (adapted from Bolsterli et al., 2011). (A) Slow wave detection method in ESES tracing and extraction of the slope of the slow wave. Below: the tracing reproduces 15 s of the raw signal from one EEG lead during NREM sleep. Circles indicate detected waves, bold red circles indicate waves entering the analysis and bold green circles indicate waves that were excluded because they were part of a spike wave complex (spikes are indicated by small blue dots). Above: 5 s of band-pass filtered signal of the same EEG lead are enlarged. Slow waves identified according to the method proposed by Riedner et al. (2007) are indicated by bold red circles. In the inlet on the right, the extraction of the slope for the first of the two waves is illustrated. The slope was defined as the amplitude of the negative peak divided by the time interval between the negative peak and the following zero crossing (light grey circles). (B) Ascending slope as a function of amplitude in a control subject and in an ESES patient. Scatter plots of the ascending slope of all selected slow waves against their amplitude in the first (blue dots for control, red dots for ESES child) and last hour of sleep (light blue for control, pink for ESES child). For each hour, a regression line was fitted to the data (blue and light blue for the control; red and pink for the ESES patient). Note that in the ESES patient, the line corresponding to the last hour (pink) is hidden behind the red line corresponding to the first hour, indicating no changes in slope between the first and last hour. On the contrary, a shift of the regression line across the night is observed in the control subject. The vertical black line indicates the voltage (75 uV) at which the ascending slopes were compared between the first and last hour. (C) Changes of the slope of slow waves during the night. Ascending slopes at an amplitude of 75 uV in the first and last hour of sleep are shown for control subjects (left) and patients (middle). Solid lines indicate the group mean. Error bars illustrate s.e.m. In the right panel, the superimposition of the mean slope (± s.e.m.) of control subjects (blue line) and patients (red line) shows that the slope of slow waves did not differ between patients and control subjects in the first hour of sleep, but was significantly steeper in patients compared to control subjects in the last hour of sleep, suggesting an impairment of the homeostatic recovery of sleep in ESES (**p<0.001, paired t-test; *p<0.01, unpaired t-test).