Connectome-based models of the epileptogenic network: a step towards epileptomics?

This scientific commentary refers to ‘Anatomic consistencies across epilepsies: a stereotactic-EEG informed high-resolution structural connectivity study’, by Besson et al. (doi:10.1093/brain/awx181). Epilepsy is broadly defined as a state of recurrent spontaneous seizures that arise when the balance between neuronal excitation and inhibition is disrupted. Epileptogenesis can be examined at different ‘levels’ of the nervous system: first as ions and membranes, then cells and circuits/synapses, and finally large-scale neuronal networks. Compelling evidence from animal models, experimental paradigms and humans indicates that specific cortical and subcortical networks play a fundamental role in the genesis and expression of seizures (Spencer, 2002). In this issue of Brain, Besson and co-authors (2017) tackle the fascinating question: is there a link between structural connectivity and the organization of the epileptogenic network? In an attempt to unravel this conundrum, the authors have designed an original diffusion tensor imaging (DTI) connectivity analysis informed by stereoelectroencephalography (SEEG). They demonstrate, at least at a mesoscopic group level (i.e. an intermediate view between the micro- and macroscopic scales), relative maintenance of structural connectivity within nodes of the epileptogenic network, together with diffuse loss of connectivity between these nodes and other brain regions. Epilepsy is a serious and prevalent condition affecting approximately 65 million people worldwide. More than one-third of patients suffer from seizures resistant to antiepileptic drugs (Wiebe and Jette, 2012). Uncontrolled epilepsy is harmful to the brain, has devastating socioeconomic consequences, and is associated with increased risk of injury and sudden death. About half of patients with drug-resistant seizures have focal epilepsy that is potentially amenable to neurosurgical treatment. The objective of epilepsy surgery is to eliminate brain regions involved in the genesis of seizures. In this context, MRI has become of fundamental importance due to its unmatched spatial resolution and whole-brain coverage. Specifically, the increasing availability of contrasts allowing for flexible assessment of morphology, tissue microstructure and connectivity, make MRI the quintessential tool not only to unveil epileptogenic lesions, but also to characterize the systemic impact of the disease. Indeed, MRI quantitative analysis lends non-invasive markers that have substantially increased the success rate of epilepsy surgery. Yet, up to 50% of operated patients continue to experience seizures. In some cases, this may be due to insufficient resection of the structural lesion causing the seizures. In others, emerging imaging data suggest that anomalies extending beyond the lesion may negatively impact surgical outcome; these observations challenge the conventional model of focal epilepsy and have revived the concept of distributed neural epileptogenic networks. The hypothesis that focal epilepsy may be more adequately described as a system-level disorder is a burgeoning research area fuelled by steady advances in MRI techniques that allow probing connectivity in vivo. Notably, DTI, together with resting state functional MRI and structural covariance, are key modalities at the heart of the field of connectomics, the analysis of large-scale brain networks (Bullmore and Sporns, 2009). The network concept naturally lends itself to describe the epileptogenic process as a hierarchical model (Bartolomei et al., 2017) through the analysis of intracerebral (SEEG) data, in which focal seizures are generated in localized networks, placed within the so-called epileptogenic zone, before recruiting other regions, as shown in Fig. 1. Organization of the epileptogenic networks in focal epilepsy. The epileptogenic zone network includes brain regions (A, B, C, and D) that may generate seizures. During seizures, the epileptogenic zone triggers a second set of regions, which form the propagation zone network (E and F). SC = subcortical regions, e.g. the thalamus. Adapted from Bartolomei et al. (2013). Notably, a structural brain lesion within the epileptogenic zone, which may or may not be visible to the naked eye on the MRI, is an integral part of this hierarchical model. The epileptogenic zone may be limited to a unique dysfunctional area, paralleling the classical notion of the seizure focus. Alternatively, the seizure onset may be characterized by discharges that simultaneously (or very rapidly) involve several other brain regions; in this scenario, a single focus cannot accurately describe the spatial organization of the epileptogenic zone, supporting the idea of epileptogenic networks. Importantly, from both a theoretical and practical standpoint, a structural brain lesion is considered at the core of the epileptogenic zone. The findings of Besson et al. are in line with recent graph-theoretical analysis in temporal lobe epilepsy showing increased within-structure covariance (likely representing hyper-connectivity), but decreases between structures, particularly with regards to the hippocampus and amygdala (Bernhardt et al., 2016). In this context, increased local connectivity may reflect axonal sprouting, while decreases in interstructure covariance likely represent disconnection within the limbic circuitry. Besson and co-workers also show preferential loss of structural connectivity within the attentional networks, particularly the salience network (dorsal anterior cingulate and insular cortex). Although the examined cohort is very small and the underlying aetiologies are heterogeneous, findings seem consistent across epileptic syndromes. The proposed framework, which combines advanced DTI analysis with SEEG, is an important step towards the characterization of the epileptic process. We should, however, keep in mind that the definition of epileptogenic networks largely depends on the accuracy of the initial hypotheses and precision of electrode placement. In other words, although many consider SEEG as the ‘gold-standard’ marker of the epileptogenic network, attempts to map invasively this complex landscape are intrinsically limited by the low spatial resolution of this electrophysiological method, which typically relies on 8–15 implanted electrodes. The access to brain regions other than those suspected to be involved in the epileptic process is restricted, inevitably leading to spatial under-sampling. Moreover, as the electrode placement is driven by clinical needs and is not standardized, the topological signature varies across subjects, as shown by the heterogeneous single-subject distribution in the study by Besson and co-workers. The inconsistent number of electrodes across subjects also leads to networks of different sizes being mapped, making comparisons between local and large-scale characteristics even harder. Finally, the modulating effect of antiepileptic drugs remains unclear, though it is admittedly difficult to address this issue in a systematic fashion. Over the past decade, SEEG has been increasingly used across the world, with significant heterogeneity of practice among centres. This trend likely reflects a rise in the number of patients with so-called MRI-negative epilepsy, particularly extra-temporal syndromes. Importantly, and maybe paradoxically, it is more difficult to identify the epileptogenic zone when the MRI is unrevealing (Tellez-Zenteno et al., 2010). Consequently, in many cases, uncertainty over the location of the epileptogenic zone warrants considering a primary hypothesis together with one or several alternatives. Clearly, however, SEEG cannot offer optimal sampling of all, which may lead to partial identification of the epileptogenic network, the worse scenario being that the seizure focus may be missed altogether. Chances of postoperative seizure freedom, a recognized indicator of identification of the epileptogenic zone, are thus significantly lower in patients with normal-appearing MRI compared to those with documented structural lesions, as confirmed by this study. Thus, in this day and age when many centres opt for SEEG investigations to ‘reveal’ an MRI-occult lesion, clinicians should be mindful of the cumulative effects of these factors when making therapeutic decisions. A large number of patients with normal-appearing MRI who benefit from epilepsy surgery present with histological signs of subtle focal cortical dysplasia type II or mesiotemporal lobe sclerosis. This is an incentive to push the limits of non-invasive resources, particularly given that only a handful will become seizure-free after surgery if SEEG is the primary mode of investigation, for reasons discussed above. Indeed, constant advances in MRI acquisition and postprocessing analysis have dramatically increased the sensitivity to detect subtle epileptogenic lesions (Bernasconi et al., 2011), to the point that it can be argued that modern epileptology must include responsibility for advanced imaging optimized for this challenging group of patients. Thus, both clinical and scientific investigations should focus as much on the search for MRI evidence of an underlying structural lesion as on identifying the electrophysiological signature of the epileptogenic zone. In this context, SEEG procedures would greatly benefit from guidance provided by advanced MRI methods. Whether diffusion-derived alterations in the attentional networks observed by Besson et al. (2017) relate to electrophysiological anomalies, the underlying structural brain pathology, or cognitive metrics, remains unclear. Notably, data in healthy controls have provided evidence for substantial overlap between structural and functional domains (Honey et al., 2007). In focal epilepsy, although impairments in resting state functional coupling seem to parallel morphological disruptions, only very few multi-modal MRI studies have specifically addressed this issue (Voets et al., 2012). In this context, polysynaptic functional coupling, the putative basis of connectivity in the absence of direct structural connections, may complicate the interpretation of findings. The study by Besson and co-workers, which attempts to quantify the complex phenomena at the basis of the spatiotemporal organization of the epileptogenic zone, is admirable and forms a strong methodological framework. Future efforts in the field of ‘epileptomics’ should aim to validate these models with non-invasive, whole-brain electrophysiological techniques, such as magnetoencephalography, and to integrate them with advanced structural and functional MRI. A coherent multidisciplinary approach will help determine whether connectome-based analysis of the epileptogenic network can be used to improve surgical procedures, including minimally invasive neuroablative methods, and to refine current MRI-based predictors of surgical outcome.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.