Abstracts

Rationale: We examined whether interictal EEG changes predict ketogenic diet outcome. Background-slowing might predict side-effects/bad outcome. Decreased spike frequency might predict successful treatment, to be used as an argument to continue the diet. The results of the first 7 patients studied are presented.Methods: Each patient underwent three 24-hour EEGs: 1st EEG at baseline, 2nd EEG after 6 weeks of ketogenic diet; 3rd EEG at 1 year treatment or at drop-out because of ineffectiveness/side-effects. Interictal spike counts were obtained from EEG during 4 hours of wakefulness and 4 hours of sleep. Detection was performed using Persyst Spike Detector Reveal version 2004.04.14 and visual artefact-rejection of clustered artefacts. For background analysis, 20 five-second epochs from each EEG were randomly selected during wakefulness (eyes-open). Epochs containing eye-movements, artefacts or paroxysms were excluded. Spectral analysis was performed and power distributions of 20 epochs of each EEG were obtained. Mean absolute powers of 4 leads (F4-C4, C4-O2, F3-C3, C3-O1) were computed for 0.5 Hz bins from 0.5-25 Hz. These bins were analyzed by t-tests for independent samples. Results: Spike counts were performed in all 21 EEGs. In only two patients marked changes in spike counts in the 2nd EEG compared to baseline were noted. The 2nd EEG of patient 1 (non-responder) showed a mean spike count of 1376 spikes/hour during wakefulness (compared to 4 spikes/hour during baseline) and 765 spikes/hour in sleep (362 spikes/hour baseline). The 3rd EEG recorded after 1 year also showed this severe worsening. Patient 6 (responder) had marked improvement of especially his first follow-up wake EEG with mean 36 spikes/hour in wake (633 spikes/hour baseline) and 295 spikes/hour in sleep (493/hour baseline). His 3rd EEG did not show the same improvement. This patient showed clinical improvement within few weeks of treatment, having still daily seizures but less frequent and severe. All other 2nd and 3rd EEGs of the other patients only showed minor changes and these were not predictable of outcome. For spectral analysis 15 EEGs were included. Patient 6 had no spike-free epochs in EEG1 and patient 1 had background slowing related to frequent epileptic paroxysms. The power distributions obtained for three subsequent EEGs indicated stable background activity in 4 out of 5 patients. In patient 7 the maximum of the power distribution increased in the 0.5-8.0 Hz range. Statistical testing of differences in power per 0.5 Hz bin only showed significant t-values in a few bins in this range, because of high standard deviations of the means. This patient was a responder and had no side effects of the diet. Conclusions: The EEGs of 7 patients treated with ketogenic diet show that EEG-background is generally not worsened. In patients with marked changes in interictal spike counts at 6 weeks ketogenic diet, the EEG may predict outcome in terms of seizure frequency or seizure severity. Because of the small group, it is too early to advise to use EEGs to decide about continuing treatment.