Absence seizures: individual patterns revealed by EEG-fMRI

Abstract

Purpose: Absences are characterized by an abrupt onset and end of generalized 3-4 Hz spike and wave discharges (GSWs), accompanied by unresponsiveness. Although previous electroencephalography-functional magnetic resonance imaging (EEG-fMRI) studies showed that thalamus, default mode areas, and caudate nuclei are involved in absence seizures, the contribution of these regions throughout the ictal evolution of absences remains unclear. Furthermore, animal models provide evidence that absences are initiated by a cortical focus with a secondary involvement of the thalamus. The aim of this study was to investigate dynamic changes during absences.

Methods: Seventeen absences from nine patients with absence epilepsy and classical pattern of 3-4 Hz GSWs during EEG-fMRI recording were included in the study. The absences were studied in a sliding window analysis, providing a temporal sequence of blood oxygen-level dependent (BOLD) response maps.

Results: Thalamic activation was found in 16 absences (94%), deactivation in default mode areas in 15 (88%), deactivation of the caudate nuclei in 10 (59%), and cortical activation in patient-specific areas in 10 (59%) of the absences. Cortical activations and deactivations in default mode areas and caudate nucleus occurred significantly earlier than thalamic responses.

Discussion: Like a fingerprint, patient-specific BOLD signal changes were remarkably consistent in space and time across different absences of one patient but were quite different from patient to patient, despite having similar EEG pattern and clinical semiology. Early frontal activations could support the cortical focus theory, but with an addition: This early activation is patient specific.

Figure 1

Sliding window analysis of absences. Successive gamma function regressors having a full-width at half-maximum of 2 s and separated by 2.5 s were used as separate regressors. The number of regressors applied in the general linear model analysis depended on the duration of each absence. All these regressors were merged in the same general linear model analysis. Successive F-tests were applied on t-value results obtained for three consecutive 2.5 s models. Each F-test reflected results in a 7.5 s window. These windows overlapped by 5 s, providing an F-value every 2.5 s. Windows before the onset and after the end of absences were included until no further BOLD signal changes were observed. Epilepsia © ILAE

Figure 2

Sliding window analysis for both absences of Patient 1 (windows 0 to 52.5 s). First activation consistent for both absences was found in the right inferior frontal cortex (white arrows). In the first absence this activation started 2.5 s after the onset, for the second absence it coincided with the onset. The thalamic activation and deactivation in default mode areas and caudate nucleus started between 5 s and 7.5 s after onset. BOLD signal changes in the thalamus and default mode areas exceeded the duration of the absence by several windows. Thalamic activation was followed by thalamic deactivation in both absences. Common remarks for Figures 2, 3, and Supplementary Figures 1–7 mentioned in Figure 2. The duration of the absence is indicated by one EEG-channel (Fp2 with average reference); the onset is indicated by a black arrow. Please note that the EEG onset of the absence is shifted 5 s to account for the hemodynamic delay of the BOLD response and allow a direct comparison of absence duration and associated BOLD signal changes. White arrows indicate areas of the earliest cortical activation which were consistent if more than one absence was recorded. Regions of interest for cortical BOLD signal extractions (see Fig. 4) were placed in these areas. Epilepsia © ILAE

Figure 3

Sliding window analysis for both absences of Patient 2 (windows −5 to 40 s). First activation consistent for both absences was found in the left frontopolar cortex (white arrows). In the first absence this activation started 10 s after the onset, for the second it started 5 s after onset. The thalamic activation started 7.5 s after onset in both, deactivation in the caudate nucleus 5 s after onset, and deactivation in default mode areas 5 s after onset in the first and 2.5 s before onset in the second. BOLD signal changes in the thalamus and default mode areas exceeded the duration of the absence by several windows. In the second absence thalamic activation was followed by thalamic deactivation. In addition, see common remarks for Figures 2, 3, and Supplementary Figures 1–7 mentioned in Figure 2. Epilepsia © ILAE

Figure 4

BOLD signal changes from region of interest (ROI) analyses. All absences are displayed showing the percentage BOLD signal changes in ROIs in the thalamus, areas of cortical increase, caudate nucleus, and default mode areas. BOLD signal changes for three scans before the onset of the absence are displayed and continue until the BOLD signal changes reaches baseline after the absence. The pattern of BOLD signal changes for the different absences one of each patient is very similar. Patients 1–3 show long-lasting BOLD signal changes in the thalamus and default mode areas that exceed the duration of the absences by several seconds. Positive BOLD signal changes in cortical areas—which were consistent for several absences of each patient—are found in 6 patients; in Patients 2, 4, and 6 these BOLD signal changes exceed the percentage BOLD signal changes observed in the thalamus, caudate nucleus, and default mode areas. Epilepsia © ILAE