Abstracts

Rationale: Symptomatic epilepsy associated with hypothalamic hamartoma (HH) is characterized by gelastic seizures (GS) as well as other seizure types and behavioral problems. The unique entity of HH provides a disease model for subcortical epileptogenesis and epileptic encephalopathy, but its epileptic network remains elusive. By means of simultaneous recording of EEG with functional MRI (EEG-fMRI), we attempted to investigate brain networks involved in epileptogenesis/encephalopathy associated with HH. Since stereotactic radiofrequency thermocoagulation (SRT) to HH interface has been successful as epilepsy surgery (Kameyama et al., 2010), we aimed to evaluate the efficacy of EEG-fMRI to locate/lateralize the interface in comparison with subtraction ictal SPECT coregistered to MRI (SISCOM).Methods: Eight patients with HH (age: 1-27 years) underwent EEG-fMRI (IRB#E217). All had GS and 7 developed other seizure types. A certified electroencephalographer off-line identified interictal spikes (9-451 / 30-90 minutes recording). Spike-related activation and deactivation were analyzed using general linear model (FSL software). First, the spatio-temporal dynamics of the epileptic network were examined by applying the hemodynamic response function at, before and after the spike onset (time-shift model = -8, -6, -4, -2, 0, +2, +4) in each patient. Clinical utility of EEG-fMRI for identifying the HH interface was compared with SISCOM in patients with HH having unilateral hypothalamic attachment (N =5). Second, group analysis (N = 7) was performed after excluding 1 year-old patient (p < 0.05 after cluster thresholding was defined significant).Results: Spike-triggered fMRI revealed epileptic networks consisting of activation and/or deactivation in all patients. In 6 of 8 patients, the hypothalamus ipsilateral to HH showed activation with the time-shift model before the spike onset. In addition, activation involved the thalamus, anterior cingulate cortex, hippocampus, and neocortices to a variable degree. Group analysis showed common activation in the ipsilateral hypothalamus and midbrain, and bilateral cerebellum (contralateral dominancy). Deactivation was observed with the models after spike onset in the bilateral thalami, putamens, hippocampi and the default mode network (DMN). Among 5 patients with unilateral HH attachment, activation was seen in 3/5 patients by EEG-fMRI while hyperperfusion was seen in 1/3 patients by SISCOM (no seizures captured by ictal SPECT in 2).Conclusions: Among the HH-related seizure network, the ipsilateral hypothalamus, ipsilateral midbrain and contralateral cerebellum were commonly activated in group analysis, indicating their role in generation of GS. Commonly observed DMN deactivation may be related with impaired consciousness during seizures and resultant epileptic encephalopathy. Inter-individual variance for activation in neocortices may explain variable seizure types besides GS in each patient. Sensitivity to identify the HH surface by EEG-fMRI was not less than SISCOM in this study, and further case accumulation is warranted to establish its utility for presurgical evaluation of SRT.