REVIEW URRENT C OPINION
Functional brain networks in epilepsy: recent advances in noninvasive mapping Francesca Pittau and Serge Vulliemoz
Purpose of review Epilepsy is one of the most frequent chronic neurological disorders. Recent evidences strongly suggest that epilepsy is due to a dysfunction within an epileptic network, rather than due to the pathological activity of single sources. The aim of this article is to review the recent advances on functional connectivity revealed by noninvasive neuroimaging techniques. Recent findings Functional connectivity detected through hemodynamic [functional MRI (fMRI)] and electro-magnetic techniques (EEG/MEG) in patients with epilepsy gives an insight into the physiopathogenesis of epileptic network underlying focal epilepsies and specific epileptic syndromes. An increasing number of studies suggest a relevance for surgical cases, both for localizing the focus and for predicting postsurgical cognitive impairment, based on the interactions between pathological and physiological networks. Summary fMRI and EEG/MEG functional connectivity are complementary techniques and help in identifying the interactions between epileptic activity and physiological networks at different scales. Neuropsychological and neuropsychiatric impairment can be explained by such interactions. fMRI and EEG/MEG functional connectivity help in localizing important drivers of epileptic activity and can also help in predicting postsurgical outcome. Given the large number of methods applied, strict validation, mostly obtained in surgical series, is of utmost importance to understand the benefits and limitations of each technique. Keywords electroencephalography, epilepsy, functional MRI, functional connectivity, magnetoencephalography, network
INTRODUCTION Epilepsy is one of the most frequent chronic neurological disorders [1], and almost one-third of patients suffer from intractable seizures (drug-resistant epilepsy) despite adequate medical treatment [2]. The notion of large-scale brain network has become a key concept in underlying epilepsy, in the study of seizure generation and propagation, disease evolution [3], and cognitive impairment before and after surgery. There are different ways to characterize functional interactions among brain regions. Functional connectivity measures the statistical dependency between signals (e.g. electromagnetic or hemodynamic) recorded from different brain areas. This approach cannot infer causality, because it does not account for the direction of flow information. Effective or directed connectivity refers explicitly to the influence that one neural system exerts over another, and aims at describing causal influences. This review will focus on functional connectivity revealed by noninvasive techniques, such as simultaneous www.co-neurology.com
functional MRI (fMRI) and electroencephalography (EEG) (EEG-fMRI), and electric source imaging (ESI) or magnetic source imaging (MSI). Functional connectivity estimated using intracranial EEG, and structural connectivity measures will not be covered.
FOCAL EPILEPSIES: IRRITATIVE ZONE AND EPILEPTIC NETWORK Neuroimaging and electrophysiology findings increasingly suggest that epileptic seizures and epilepsies are related to abnormal brain function within a network of cortical and subcortical structures Presurgical Epilepsy Evaluation Unit, Neurology Department, University Hospital of Geneva, Geneva, Switzerland Correspondence to Dr Serge Vulliemoz, Epilepsy Unit, Neurology Department, University Hospital of Geneva, 4 rue Gabrielle-Perret-Gentil, 1211 Geneva 4, Switzerland. Tel: +0041795533823; e-mail: serge.vulliemoz @hcuge.ch Curr Opin Neurol 2015, 28:338–343 DOI:10.1097/WCO.0000000000000221 Volume 28 Number 4 August 2015
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Functional brain networks in epilepsy Pittau and Vulliemoz
KEY POINTS fMRI and EEG/MEG functional connectivity are complementary techniques and help in identifying the interactions between epileptic activity and physiological networks at different scales. fMRI and EEG/MEG functional connectivity help in localizing the important drivers of epileptic activity and can also help in predicting postsurgical outcome. Given the large number of strategies applied, strict validation, mostly obtained in surgical series, is of utmost importance to understand the benefits and limitations of each technique.
(epileptic network) rather than to the dysfunction of a single epileptogenic region [4].
Functional MRI fMRI detects blood oxygen level dependent (BOLD) signal change reflecting metabolically active brain areas not only in relation to a specific physiologic or pathologic event, but also in resting state conditions [resting state-fMRI (RS-fMRI)]. Simultaneous EEG and fMRI (EEG-fMRI) recordings map focal hemodynamic changes, correlated to events of interest identified on the EEG [5]. EEG-fMRI has proven useful in characterizing focal and generalized epilepsies. BOLD responses to focal interictal epileptic discharges (IEDs) can be multifocal, also present at a distance from the presumed focus, in regions without any detectable EEG changes, strengthening the concept of epileptic network [6]. These IED-related hemodynamic changes can help in predicting postsurgical outcome, with a high-positive predictive value in the case of focal responses concordant to the resected region and negative predictive value in the case of diffuse/not-concordant BOLD responses [7,8]. Connectivity fMRI studies investigate the relationships between a set of predefined relevant brain regions (seed) selected by the investigator or between one seed region and the rest of the brain (i.e. requiring an ‘a priori’ hypothesis) or at the whole brain scale. Using the maximum value of the IEDrelated BOLD map as seed, a study in 23 patients with frontal lobe epilepsy found an increased functional connectivity in the neighborhood of the seed and a decrease in regions remote to the seed compared with controls [9]. Patient-specific connectivity patterns were not significantly changed when comparing fMRI runs with spikes to fMRI without any spike detectable on the simultaneous EEG. Another study, using an fMRI independent component analysis approach in patients with drug-resistant epilepsy,
showed that BOLD components related to the individual epileptogenic region are similar in EEG-fMRI with versus without detectable IED in scalp EEG [10]. This suggests that, in patients with pharmaco-resistant epilepsy, brain regions involved in the epileptic network show coherent hemodynamic fluctuations independently from the epileptic activity detectable on scalp EEG. Animal models, even if for absence epilepsy, corroborate this finding [11]. Understanding the functional connectivity of epileptic network with and without IED is preliminary and fundamental to better understand the interaction between physiological and epileptic networks. Connectivity analysis can indeed help in localizing the epileptic focus [12,13] and give useful information regarding postoperative outcome (for a recent review, see [14]). A method based on Bootstrap Analysis of Stable Clusters [15] in individual patients found alteration in resting state network (RSN) spatial organization in five out of six patients who were postoperatively seizure-free. The abnormalities were lateralized to RSN spatially related to the epileptic focus. Remote changes in other RSNs were also found suggesting both local and remote reorganization in the epileptic brain [16]. A study investigated the spatial relationship between ictal and interictal epileptic networks in 28 patients with drug-resistant focal epilepsy who underwent both subtraction ictal-single-photonemission-computed-tomography coregistered to MRI (SISCOM) and interictal EEG-fMRI in different sessions. The spatial overlap between hyperperfusion on SISCOM and positive BOLD responses on EEGfMRI, and between hypoperfusion and negative BOLD responses within the same network suggests a good concordance between the irritative and seizure onset zone. Furthermore, it indicates that the perfusion and the BOLD response vary in the same direction, although lack of metabolic data prevents inferring about neurovascular coupling [17]. Concerning negative BOLD, a case report [18] reported ictal EEG-fMRI and stereo-EEG (SEEG), recorded in independent sessions, in a patient with drug-resistant frontal lobe epilepsy and negative motor seizures. Ictal BOLD response demonstrated a left dorso-lateral frontal cortex positive BOLD response (concordant with IED analysis) and a simultaneous negative BOLD response in the bilateral central cortex (remote from the epileptogenic focus). SEEG recording in the central cortex containing negative BOLD showed a clear b rhythm (
Functional brain networks in epilepsy: recent advances in noninvasive mapping Francesca Pittau and Serge Vulliemoz
Purpose of review Epilepsy is one of the most frequent chronic neurological disorders. Recent evidences strongly suggest that epilepsy is due to a dysfunction within an epileptic network, rather than due to the pathological activity of single sources. The aim of this article is to review the recent advances on functional connectivity revealed by noninvasive neuroimaging techniques. Recent findings Functional connectivity detected through hemodynamic [functional MRI (fMRI)] and electro-magnetic techniques (EEG/MEG) in patients with epilepsy gives an insight into the physiopathogenesis of epileptic network underlying focal epilepsies and specific epileptic syndromes. An increasing number of studies suggest a relevance for surgical cases, both for localizing the focus and for predicting postsurgical cognitive impairment, based on the interactions between pathological and physiological networks. Summary fMRI and EEG/MEG functional connectivity are complementary techniques and help in identifying the interactions between epileptic activity and physiological networks at different scales. Neuropsychological and neuropsychiatric impairment can be explained by such interactions. fMRI and EEG/MEG functional connectivity help in localizing important drivers of epileptic activity and can also help in predicting postsurgical outcome. Given the large number of methods applied, strict validation, mostly obtained in surgical series, is of utmost importance to understand the benefits and limitations of each technique. Keywords electroencephalography, epilepsy, functional MRI, functional connectivity, magnetoencephalography, network
INTRODUCTION Epilepsy is one of the most frequent chronic neurological disorders [1], and almost one-third of patients suffer from intractable seizures (drug-resistant epilepsy) despite adequate medical treatment [2]. The notion of large-scale brain network has become a key concept in underlying epilepsy, in the study of seizure generation and propagation, disease evolution [3], and cognitive impairment before and after surgery. There are different ways to characterize functional interactions among brain regions. Functional connectivity measures the statistical dependency between signals (e.g. electromagnetic or hemodynamic) recorded from different brain areas. This approach cannot infer causality, because it does not account for the direction of flow information. Effective or directed connectivity refers explicitly to the influence that one neural system exerts over another, and aims at describing causal influences. This review will focus on functional connectivity revealed by noninvasive techniques, such as simultaneous www.co-neurology.com
functional MRI (fMRI) and electroencephalography (EEG) (EEG-fMRI), and electric source imaging (ESI) or magnetic source imaging (MSI). Functional connectivity estimated using intracranial EEG, and structural connectivity measures will not be covered.
FOCAL EPILEPSIES: IRRITATIVE ZONE AND EPILEPTIC NETWORK Neuroimaging and electrophysiology findings increasingly suggest that epileptic seizures and epilepsies are related to abnormal brain function within a network of cortical and subcortical structures Presurgical Epilepsy Evaluation Unit, Neurology Department, University Hospital of Geneva, Geneva, Switzerland Correspondence to Dr Serge Vulliemoz, Epilepsy Unit, Neurology Department, University Hospital of Geneva, 4 rue Gabrielle-Perret-Gentil, 1211 Geneva 4, Switzerland. Tel: +0041795533823; e-mail: serge.vulliemoz @hcuge.ch Curr Opin Neurol 2015, 28:338–343 DOI:10.1097/WCO.0000000000000221 Volume 28 Number 4 August 2015
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Functional brain networks in epilepsy Pittau and Vulliemoz
KEY POINTS fMRI and EEG/MEG functional connectivity are complementary techniques and help in identifying the interactions between epileptic activity and physiological networks at different scales. fMRI and EEG/MEG functional connectivity help in localizing the important drivers of epileptic activity and can also help in predicting postsurgical outcome. Given the large number of strategies applied, strict validation, mostly obtained in surgical series, is of utmost importance to understand the benefits and limitations of each technique.
(epileptic network) rather than to the dysfunction of a single epileptogenic region [4].
Functional MRI fMRI detects blood oxygen level dependent (BOLD) signal change reflecting metabolically active brain areas not only in relation to a specific physiologic or pathologic event, but also in resting state conditions [resting state-fMRI (RS-fMRI)]. Simultaneous EEG and fMRI (EEG-fMRI) recordings map focal hemodynamic changes, correlated to events of interest identified on the EEG [5]. EEG-fMRI has proven useful in characterizing focal and generalized epilepsies. BOLD responses to focal interictal epileptic discharges (IEDs) can be multifocal, also present at a distance from the presumed focus, in regions without any detectable EEG changes, strengthening the concept of epileptic network [6]. These IED-related hemodynamic changes can help in predicting postsurgical outcome, with a high-positive predictive value in the case of focal responses concordant to the resected region and negative predictive value in the case of diffuse/not-concordant BOLD responses [7,8]. Connectivity fMRI studies investigate the relationships between a set of predefined relevant brain regions (seed) selected by the investigator or between one seed region and the rest of the brain (i.e. requiring an ‘a priori’ hypothesis) or at the whole brain scale. Using the maximum value of the IEDrelated BOLD map as seed, a study in 23 patients with frontal lobe epilepsy found an increased functional connectivity in the neighborhood of the seed and a decrease in regions remote to the seed compared with controls [9]. Patient-specific connectivity patterns were not significantly changed when comparing fMRI runs with spikes to fMRI without any spike detectable on the simultaneous EEG. Another study, using an fMRI independent component analysis approach in patients with drug-resistant epilepsy,
showed that BOLD components related to the individual epileptogenic region are similar in EEG-fMRI with versus without detectable IED in scalp EEG [10]. This suggests that, in patients with pharmaco-resistant epilepsy, brain regions involved in the epileptic network show coherent hemodynamic fluctuations independently from the epileptic activity detectable on scalp EEG. Animal models, even if for absence epilepsy, corroborate this finding [11]. Understanding the functional connectivity of epileptic network with and without IED is preliminary and fundamental to better understand the interaction between physiological and epileptic networks. Connectivity analysis can indeed help in localizing the epileptic focus [12,13] and give useful information regarding postoperative outcome (for a recent review, see [14]). A method based on Bootstrap Analysis of Stable Clusters [15] in individual patients found alteration in resting state network (RSN) spatial organization in five out of six patients who were postoperatively seizure-free. The abnormalities were lateralized to RSN spatially related to the epileptic focus. Remote changes in other RSNs were also found suggesting both local and remote reorganization in the epileptic brain [16]. A study investigated the spatial relationship between ictal and interictal epileptic networks in 28 patients with drug-resistant focal epilepsy who underwent both subtraction ictal-single-photonemission-computed-tomography coregistered to MRI (SISCOM) and interictal EEG-fMRI in different sessions. The spatial overlap between hyperperfusion on SISCOM and positive BOLD responses on EEGfMRI, and between hypoperfusion and negative BOLD responses within the same network suggests a good concordance between the irritative and seizure onset zone. Furthermore, it indicates that the perfusion and the BOLD response vary in the same direction, although lack of metabolic data prevents inferring about neurovascular coupling [17]. Concerning negative BOLD, a case report [18] reported ictal EEG-fMRI and stereo-EEG (SEEG), recorded in independent sessions, in a patient with drug-resistant frontal lobe epilepsy and negative motor seizures. Ictal BOLD response demonstrated a left dorso-lateral frontal cortex positive BOLD response (concordant with IED analysis) and a simultaneous negative BOLD response in the bilateral central cortex (remote from the epileptogenic focus). SEEG recording in the central cortex containing negative BOLD showed a clear b rhythm (
