Abstracts

Rationale: Interictal high frequency oscillations (40-200Hz) measured from scalp EEG of focal epilepsy patients (sHFOs) usually co-occur with spikes (Epilepsia 2011,52:1812), and were shown to be a more specific and accurate marker than spikes for delineating the seizure onset zone (Neurology 2011,77:524). The metabolic correlates of sHFOs as measured by blood oxygenation level-dependent (BOLD) signal are unknown. Measuring sHFOs in EEG acquired inside the MR scanner still poses a remarkable technical challenge because of their very low amplitude. We addressed this issue by comparing BOLD responses to spikes in patients in whom a separate sHFO investigation has demonstrated spikes with frequent sHFOs to responses in patients with spikes having rare sHFOs (gamma 40-80Hz and ripples 80-200Hz). Methods: 28 focal epilepsy patients were studied. 6 were excluded due to negative EEG-fMRI evaluation (4 without spikes, 1 with spikes but no BOLD changes and 1 with persistent artifacts preventing the analysis). The remaining 22 patients were divided into two equal groups based on the frequency of co-occurrence of gamma or ripples with their spikes outside the scanner: high vs. low gamma (HG/LG) and high vs. low ripples (HR/LR), each group having 11 subjects. We looked for significant differences in spatial distribution of cortical and subcortical BOLD changes, spike field- concordant BOLD cluster volume and maximal t-value between the high and the low groups. We took into consideration the number of spikes, occurrence of spike runs and generalized spike and wave (GSW) in each group. Results: The HR group had significantly more thalamic BOLD changes than the LR group (11/11 vs. 2/11, examples in Fig1-2). The difference remained after discarding the three patients with GSW in the HR group. In HR, 5/11 had only activation in the thalamus, 4/11 had only deactivation and 2/11 had activation and deactivation. In the LR group 2/11 had thalamic activation. The lateralization of thalamic BOLD responses was concordant with the lateralization of ripples in 12/13 patients. HR had a larger spike field-concordant BOLD cluster and higher maximal t-value than LR, but these changes could be attributed to higher number of GSW and spike runs in the HR than in the LR group. No significant changes where found between the BOLD characteristics of the HG and LG groups. Conclusions: Interictal spikes measured on scalp EEG, when they are accompanied by high rate of HFOs in the ripple but not in the gamma band, are associated with thalamic metabolic changes that are lateralized to the side of the cortical ripples. This result could be explained by the animal data connecting HFOs generated in the cortex with spike and wave discharges and the thalamo-cortical circuitry [Neuroscience 1997, 17:6783; Neuroscience 2004, 123:299]. This work was supported by grant MOP-38079 of the Canadian Institute of Health Research.