Hemodynamic changes preceding the interictal EEG spike in patients with focal epilepsy investigated using simultaneous EEG-fMRI

Abstract

EEG-fMRI is a non-invasive technique that allows the investigation of epileptogenic networks in patients with epilepsy. Lately, BOLD changes occurring before the spike were found in patients with generalized epilepsy. The study of metabolic changes preceding spikes might improve our knowledge of spike generation. We tested this hypothesis in patients with idiopathic and symptomatic focal epilepsy.

Eleven consecutive patients were recorded at 3 T: five with idiopathic focal and 6 with symptomatic focal epilepsy. Thirteen spike types were analyzed separately. Statistical analysis was performed using the timing of spikes as events, modeled with HRFs peaking between − 9 s and + 9 s around the spike. HRFs were calculated the most focal BOLD response. Eleven of the thirteen studies showed prespike BOLD responses. Prespike responses were more focal than postspike responses. Three studies showed early positive followed by later negative BOLD responses in the spike field. Three had early positive BOLD responses in the spike field, which remained visible in the later maps. Three others had positive BOLD responses in the spike field, later propagating to surrounding areas. HRFs peaked between − 5 and + 6 s around the spike timing. No significant EEG changes could be identified prior to the spike.

BOLD changes prior to the spike frequently occur in focal epilepsies. They are more focal than later BOLD changes and strongly related to the spike field. Early changes may result from increased neuronal activity in the spike field prior to the EEG spike and reflect an event more localized than the spike itself.

Introduction

Epilepsy is one of the most common neurological diseases, characterized by spontaneous seizures and EEG abnormalities. While various clinical symptoms, such as loss of consciousness and uncontrolled motor activity, occur during seizures, EEG changes in the period between seizures, interictally, are not accompanied by clinical symptoms. These interictal epileptiform discharges can occur in various forms, such as spikes, poly-spikes and sharp waves and are believed to be the result of summated membrane events from abnormally hypersynchronous neurons within epileptic tissue (Matsumoto and Ajmonemarsan, 1964). As the spatial resolution of surface EEG recording is limited, it is not always easy to define the area that generates interictal spikes, also called the irritative zone (Rosenow and Lüeders, 2001).

Simultaneous electroencephalogram (EEG) and functional magnetic resonance imaging (fMRI), EEG-fMRI, is a non-invasive tool with which hemodynamic changes in the brain following interictal spikes detected on scalp EEG can be detected. Several studies demonstrated that this method provides valuable information about the generators of interictal spikes in epilepsies of variable aetiology, including focal and generalized types (for review see Gotman et al., 2006; Gotman, 2008). EEG-fMRI studies allow us to look at positive changes in the blood oxygenation level dependent (BOLD) signal, as well as negative BOLD changes. Most fMRI studies use a model of the hemodynamic response function (HRF) derived from the auditory response (or Glover response) in healthy subjects (Glover, 1999), with a time-to-peak of 5.4 s. This model proved adequate to disclose BOLD responses to external stimuli in healthy volunteers, but it is still uncertain how appropriate it is to study patients with epilepsy. Even if it is believed that the HRF to epileptic spikes does not vary significantly from those to external stimuli (Lemieux et al., 2001, Bénar et al., 2002), in some studies HRFs could be identified which showed different peak times (Gotman et al., 2006, Jacobs et al., 2007) or even non-canonical shape (Lemieux et al., 2008). Additionally, the sensitivity of EEG-fMRI studies in epilepsy could be improved with the use of four different HRFs peaking at 3, 5, 7 and 9 s after the spike (Bagshaw et al., 2004, Bénar et al., 2002, Jacobs et al., 2007).

Up to now it was thought that the BOLD changes resulted from synchronized neuronal activity arising at the time of the spike and detected with a delay of approximately 6 s (Benar et al., 2002). For the first time, Makiranta et al. (2005) showed in an epilepsy model in pigs that BOLD changes could precede the occurrence of interictal spikes after penicillin injection. Similarly a study by Diehl et al. (1998) indicated changes in middle cerebral artery blood flow prior to generalized spike and slow waves. Another study showed HRFs peaking between − 1 and − 3 s prior to the spike and suggested that changes in neuronal activity prior to spikes and invisible in the EEG could lead to early BOLD responses (Hawco et al., 2007). The changes in this study however were not specific to the localization of the EEG spike. BOLD changes prior to the spike were also identified systematically in patients with generalized epilepsies (Moeller et al., 2008). In generalized epilepsies, subcortical structures are most likely involved in the generation of spikes and positive BOLD changes were found in all patients in the thalamus with an onset of up to 9 s before polyspike wave discharges.

Many EEG-fMRI studies still struggle with a low sensitivity and specificity. While early EEG-fMRI studies using one standard HRF and performed at 1.5 T only observed BOLD responses in 40–60% of patients (Krakow et al., 2001, Boor et al., 2007), sensitivity could be improved to 80–90% using variable HRF models and higher field strength (Jacobs et al., 2007). The specificity of EEG-fMRI is more difficult to assess as many patients show rather widespread BOLD responses (Kobayashi et al., 2006a, Lemieux et al., 2008) and their importance and physiological correlate is uncertain. It has to be noted that the correspondence of the BOLD response with the spike field varies between 40% and 80% (Al-Asmi et al., 2003, Kobayashi et al., 2006a) and similar numbers are observed for the concordance with epileptogenic lesions (Kobayashi et al., 2006b, Jacobs et al., 2007). The majority of studies also show additional distant BOLD changes, which are hard to explain.

These problems may derive from HRFs, which are unable to model the actual hemodynamic changes appropriately, or from a misconception of the timing that assumes that the most focal energy change should result directly from the spike seen on surface EEG. In the present study, we analyzed BOLD changes prior to the spike, starting 9 s before the spike, in two groups of patients with focal epilepsy of different aetiology: idiopathic (rolandic) epilepsy and symptomatic epilepsy.

Rolandic epilepsy, also known as benign epilepsy with centro-temporal spikes (BECTS), is, with 15% of all childhood epilepsies, the most frequent epilepsy in school-aged children and most likely genetically determined (Bray and Wiser, 1964, Doose et al., 1997). The interictal EEG shows high voltage centro-temporal spikes or sharp-waves followed by a slow wave, which show a centro-temporal negativity followed by a frontal positivity (Baumgartner et al., 1996, Minami et al., 1996). This very typical spike configuration facilitates the localization of the irritative zone and EEG and MEG findings locate it in the inferior (face) and superior (hand) sensori-motor cortex (Baumgartner et al., 1996, Kamada et al., 1998). EEG-fMRI studies found a high agreement between this localization and BOLD changes, which were mostly present in the perisylvian, central and premotor regions (Archer et al., 2003, Boor et al., 2003, Boor et al., 2007, Lengler et al., 2007).

Symptomatic epilepsies are characterized by lesional pathological changes, which can be identified in structural MRI studies and are potentially epileptogenic (Engel, 2001). Seizures are hypothesized to originate within or in part of these lesions (Rosenow and Lueders, 2001). Even if spikes can be multifocal, the localization of the irritative zone in most patients is within the lesion. Several EEG-fMRI studies provided evidence that the irritative zone can be localized within lesions using this technique (Kobayashi et al., 2005, Jacobs et al., 2007).

These two patient groups were selected as they most likely have very different pathophysiological mechanisms for their epilepsy. In contrast to patients with symptomatic epilepsy, no major tissue changes are expected in rolandic epilepsy and there has been no evidence that there could be changes in blood flow, influencing the BOLD response. However, in both patient groups, the localization of the irritative zone is more precise than in other types of epilepsy, which results in a better possibility to judge the concordance between interictal EEG spike and BOLD change. We hypothesize that changes in BOLD prior to the spike can occur in focal epilepsies and may be more focal than changes following the spike.