Epilepsy Essentials: Novel Interictal Biomarkers for Delineating the Epileptogenic Zone in Children With Drug-Resistant Epilepsy

Approximately 20% to 30% of people with epilepsy have seizures that cannot be controlled with antiseizure medications.¹ For these individuals with drug-resistant epilepsy (DRE), epilepsy surgery, which offers an approximately 50% chance to achieve seizure freedom,² is the most effective therapeutic approach. Epilepsy surgery allows the reversal of psychologic and social comorbidities as well as decreased dependence on family and society.³ The efficacy and safety of epilepsy surgery has improved substantially in recent decades, offering a favorable risk–benefit balance.⁴ Recent advances in neuroimaging and electrophysiology have assisted in the early diagnosis of epilepsy and allow a comprehensive surgical evaluation for children with DRE. Epilepsy surgery for children with DRE offers the best opportunity to prevent a lifetime of disability, recover function, and improve cognition and quality of life,⁵ because the pediatric brain possesses extensive neural plasticity. However, epilepsy surgery remains an underutilized treatment, particularly for children.⁶

The success of epilepsy surgery depends heavily on precision and accuracy in defining and resecting the epileptogenic zone (EZ): the brain area responsible for the generation of seizures. No clinical examination exists to delineate the EZ unambiguously. Instead, the EZ is defined on the basis of multiple noninvasive diagnostic tests that provide a first attempt to identify its anatomic location.⁷ However, the results of these tests are often inconclusive, and extraoperative intracranial EEG (iEEG) monitoring may be needed, in which depth or subdural electrodes are implanted inside the brain. The goal of iEEG monitoring is to localize the seizure onset zone (SOZ)—the brain area where clinical seizures are initiated—because the SOZ is regarded as the best approximator of the EZ. However, the SOZ may not represent the full extent of the EZ, and its delineation requires the recording of several stereotyped seizures at the expense of considerable human and financial resources.⁸ Moreover, iEEG presents limitations because of its invasiveness and offers limited spatial coverage. As a result, the actual focus and extent of the SOZ may be missed, leading to unsuccessful surgery results. In addition, the interpretation of invasive ictal studies requires the recording of multiple stereotyped clinical seizures, which may require days to weeks of hospitalization. A noninvasive biomarker for the EZ that can improve presurgical planning and potentially replace extended extraoperative iEEG monitoring is of great importance to public health.

Electric and Magnetic Source Imaging for Irritative Zone Localization

Noninvasive electrophysiologic methods, such as magnetoencephalography (MEG) and high-density EEG (HD-EEG), are used in the presurgical evaluation of children with DRE. These tests allow the localization of the irritative zone—the brain area that generates interictal epileptiform discharges—through electric and magnetic source imaging (ESI/MSI). A growing body of literature has shown that ESI and MSI can localize the irritative zone with an accuracy that ranges from ~10 to 20 mm.^9-12 ESI performed even in the conventional low-density EEG (up to 20 channels with no coregistration) can identify the irritative zone with a localization accuracy of ~20 to 25 mm.¹⁰ Some studies have extended these findings by showing that MSI can change the clinical plan in 21% to 33% of people with DRE.¹³ The rate of seizure freedom after resective surgery has been shown to be higher if MSI findings are considered.¹⁴ In a series of 1000 cases, MSI findings predicted good surgical outcome (ie, Engel score I) at least 5 years after surgery with a sensitivity and specificity of ~75%.¹⁵ By using more elaborate source localization methods (ie, dipole clusterness), the localization accuracy of ESI/MSI can be improved, augmenting the presurgical evaluation of children and young adults with normal, nonfocal, or subtly abnormal MRI results, who may be regarded as ineligible for focal surgical resection with minimal or no functional loss.¹¹ By using this dipole-based clustering technique, our team rendered seizure-free an individual who had DRE for more than 16 years and who had had previous presurgical evaluation with stereotaxic EEG that did not localize the epileptogenic focus.¹⁶ ESI and MSI can obviate the need for invasive procedures, or, in more complex clinical cases, can optimize their planning.¹⁷

Despite this evidence, ESI and MSI are performed in few tertiary epilepsy centers on a regular basis mostly due to lack of personnel expertise and high maintenance cost. Furthermore, HD-EEG and MEG recordings are rarely performed simultaneously, although they offer both confirmatory and complementary information. MSI is blind to cortical sources with radial orientation (ie, at the crown of the gyri or at the bottom of sulci) and deep sources in the brain but offers better spatial resolution than ESI.¹⁸ ESI is sensitive to sources with radial or tangential orientation but is susceptible to tissue conductivity (especially skull conductivity), resulting in deflected voltage field distribution. Theoretical, modeling, and stimulation studies indicate that the combination of these 2 modalities into a single solution—electromagnetic source imaging—provides improved localization accuracy of the epileptogenic focus compared with individual modalities. Our group recently showed that combined electromagnetic source imaging can provide superior localization and improved predictive performance than individual modalities in a cohort of children with DRE caused by focal cortical dysplasia.¹⁹ Electromagnetic source imaging combined with a dipole clustering method¹¹ presented localization abilities similar to invasive recordings with iEEG.¹⁹ ESI and MSI are underutilized powerful tools that can improve the presurgical evaluation of people with DRE, thereby improving their surgical outcomes.

ESI and MSI for SOZ Localization

ESI and MSI can be useful not only in localizing the irritative zone, but also in localizing the SOZ. However, the localization of ictal activity with MSI or ESI may present some challenges. At seizure onset, the signal-to-noise ratio may be relatively too low to offer a reliable source localization solution, or biologic artifacts may obscure the onset of the seizure. Moreover, MEG recording time is often limited to up to 1 hour, hampering the ability to capture a seizure in people with infrequent seizures. However, despite these difficulties, an increasing number of studies have reported that ictal MSI or ESI activity can help localize the EZ. Ictal MSI results may be more focal and closer to the invasively determined SOZ. In some cases where focal interictal MSI findings cannot be obtained, ictal MSI could provide correct localization of the EZ. ESI can be helpful in localization of the SOZ even when applied to recordings from the standard 10-20 conventional EEG. Our group has shown that ESI performed on scalp standard EEG recordings²⁰ and the virtual implantation of “sensors” inside the brain estimated from ictal scalp EEG²¹ can approximate the SOZ and predict the surgical outcome in a cohort of children with DRE.

High-Frequency Oscillations as a Biomarker of Epilepsy

High-frequency oscillations (HFOs)—brain activity with frequencies above 80 Hz—have been proposed as alternative (to spikes) interictal biomarkers of epilepsy. The HFOs can be recorded both invasively with iEEG and noninvasively with EEG or MEG. They are classified into ripples (>80 Hz) and fast ripples (>250 Hz). Ripples have high levels of sensitivity, because they can be recorded easily in most people with epilepsy. Ripples can be recorded both invasively with iEEG as well as noninvasively with EEG and MEG. However, ripples have low specificity, because they can be seen in physiologic brain areas that need to be preserved during surgery. Fast ripples are more specific biomarkers of the EZ, but are difficult to detect with conventional macroelectrodes, and thus cannot be seen in all people with DRE. Recent MEG studies have shown that interictal ripples can be localized noninvasively with high precision in children with DRE.²² Ripples overlying interictal spikes detected with MEG or HD-EEG seem to be prognostic, noninvasive biomarkers of epileptogenicity, because removing their cortical generators predicts good surgical outcome.²² Conversely, scalp ripples alone are most likely generated by non-epileptogenic areas that should not be resected. Similar to seizures, ripples propagate from more to less epileptogenic areas of spread. In a recent iEEG study, our group showed that surgical resection of the onset of these ripple-propagating areas is sufficient for the individual to become seizure-free (Figure 1).²³ This propagating phenomenon was also observed noninvasively with MEG and HD-EEG through virtual sensors implantation over specific brain areas.²⁴ Noninvasive localization of the ripple onset, as defined by the virtual implantation, predicted good surgical outcome in children with DRE, more precisely than the entire area generating ripples or interictal spikes.²⁴

Functional Connectivity

In cases where ESI or MSI fail to localize the epileptogenic foci, an entire brain network may be responsible for the generation of seizures. Even in cases of focal epilepsy, seizure-generating tissues are thought to represent epileptogenic focus, which are involved in microscale to macroscale pathophysiologic mechanisms. Thus, epileptogenic activity is possibly generated by brain networks, and spreads across them, rather than by single foci.²⁵ Studying these networks with suitable tools and methods is thus necessary to enable the identification of pathologic hubs that play critical roles in the propagation of ictal activity in this epileptogenic network.

Our group recently showed that functional connectivity measures, applied on pairs of electrodes from iEEG recordings, can discriminate among different epileptogenic states.²⁶ We computed functional connectivity among iEEG electrodes in different states (ie, interictal without spikes, interictal with spikes, preictal, ictal, postictal) and frequency bands, and observed a hierarchical epileptogenic organization among states for functional connectivity: lower functional connectivity during interictal and preictal states followed by higher functional connectivity during ictal and postictal states (Figure 2). We also observed higher functional connectivity inside the resection in individuals with good outcome in different frequency bands, but no differences in individuals with poor outcome (Figure 2). Our findings are in line with other studies in the field showing higher connectivity for iEEG electrodes localized within the resection for people with good outcome compared with individuals with poorer postoperative seizure control.²⁷

These findings indicate that high functional connectivity values assessed with iEEG recordings are biomarkers of highly epileptogenic tissue and predictors of good surgical outcome if these areas are resected. However, these studies were performed on iEEG, which has limited field of view and may fail to identify the EZ. By using noninvasive methods, such as MEG and HD-EEG, our team localized pathologic hubs having high functional connectivity values noninvasively; surgical resection of these pathologic hubs renders people with DRE seizure-free (Figure 3).^28,29

Conclusion

Emerging technologic developments in neuroimaging tools and signal processing and source localization methods enhance precise localization of the EZ in people with DRE. These recent developments may result in a more efficient, less invasive, and less time-intensive presurgical evaluation process. Such advancements in epilepsy research may lead to effective clinical tools that can be used in the near future to improve surgical outcomes in children with DRE.