Avoiding Insularity: Insights in Insular Semiology From sEEG

Abstract

Anterior Versus Posterior Insular Epilepsy: Correlations With Semiology Based on Stereoelectroencephalography

Chisholm J, Krishnan B, Khoury J, Alexopoulos A, Bingaman W, Serletis D, Najm I, Bulacio JC. Ann Neurol. 2025 Nov;98(5):1111-1124. doi: 10.1002/ana.70015. Epub 2025 Aug 13. PMID: 40799141; PMCID: PMC12577678.

Objective: To characterize insular seizure semiology and correlate with stereoelectroencephalography (SEEG) seizure onset in a well-defined cohort, in particular examining differences between anterior and posterior insular seizures. Methods: We documented all semiological signs and the timing of emergence for 45 patients with SEEG-confirmed insular epilepsy, along with the precise location of the seizure onset zone (SOZ) within the insula. Semiological signs and other noninvasive data were compared between those with anterior and posterior insula SOZ, with more detailed insular subregion description when appropriate. Co-occurrence patterns of insular semiological signs were also investigated. Results: A total of 87% reported auras, and the corresponding insular SOZ demonstrated an anterior-to-posterior gradient by aura type. The absence of aura was significantly associated with an anterior insula SOZ (p = 0.01). Late grunting/moaning (p = 0.07), symmetric mouth elementary motor (p = 0.1), blinking (p = 0.05), and chewing/repetitive swallowing (p = 0.09) also suggested an anterior insula SOZ. Unilateral nonpainful somatosensory aura (p = 0.07) and early arm/hand elementary motor (p = 0.1) suggested a posterior insular SOZ. Interpretation: We report the characteristics and semiological features in a large cohort of primary insular epilepsy patients. The type of aura is crucial for identifying the SOZ, and the timing of semiological emergence contributes to the anatomo-electro-clinical localization of an insular SOZ. These findings can enhance the identification of insular epilepsy, making targeted curative treatment feasible.

Commentary

Over the past 20 years (reviewed by Jobst et al ¹ ), there has been increasing recognition that the insula is not an island in the brain, but has dense connections to other cortical regions. A seizure starting in the insula can appear to arise from a connected structure, convincing the surgical team to intervene by resecting an area of seizure propagation and not generation, with predictably poor results. With the advent of improved imaging and robotic stereotactic electroencephalography (sEEG),^2,3 surgical groups have become more willing to explore deep structures like the insula, which were previously more perilous to sample with grid and strip electrodes. This has led to the identification of insular onsets in people with epilepsy who have previously failed prior resection, with improved results.

In spite of the improved safety profile associated with sEEG, there is still a risk of intracranial hemorrhage. There is often some degree of reluctance in sampling deeper targets like the insula, especially when traversing the Sylvian fissure with its dense vascular tree with an orthogonal approach. In addition, sEEG electrode placement is limited by the number of recording channels available in the epilepsy monitoring unit. Thus, surgical teams need to be prudent in which areas of the brain to sample. As a result, insular sampling has often been limited to patients with a high suspicion of insular onset, either because of imaging or semiology. Classically, the semiologies associated with insular onset that have prompted sEEG sampling have been autonomic and gustatory for the anterior insula and somatosensory, including pain, from more posterior onset.^4,5

Chisholm and colleagues ⁶ have made a significant contribution to the literature by analyzing the contact locations of insular electrodes and correlating electrographic properties to rigorously documented semiology in 45 patients. They used only spontaneous recordings, not stimulation-induced seizures. They not only looked at the semiology of seizure onset of different seizure onset zones (SOZs) within the insula, but also semiologies associated with seizure spread by looking at the evolution of seizure characteristics within seizures from different parts of the insula. Many of the patients had classic semiologies, with laryngeal symptoms, early tachycardia, and unpleasant facial expression from SOZs in different regions of the anterior insula, and unilateral nonpainful somatosensory aura and arm/hand elementary motor signs from SOZs in different areas of the posterior insula. However, some had no aura (13%), if localized to the anterior insula, or other signs that were not consistently localized to one area of the insula (although this may be due to the small n leading to inadequate power).

One concern is for sampling error: is the area of first electrographic appearance the SOZ or an area of early spread? One way to address this is to limit the analysis to those patients who had Engel I outcomes after resection or laser ablation. The authors state that the results were similar when limiting their analysis to only those patients with Engel I outcomes (20/39 who underwent surgery). However, the raw data for such patients are not presented, presumably because the study would have been underpowered. In addition, postoperative MRIs for even the Engel I patients show that more than just the SOZ was removed in most cases, raising the concern that some of these patients may have had their SOZ along the surgical approach.

So what are readers to take from this paper and apply to the person with epilepsy in front of them in clinic? It is challenging, since the decision making that led to insular sampling in this patient cohort is unclear. In addition, there is no analysis of whether the contacts were on the dominant or nondominant side. The most obvious take-away is that we should be sampling the insula more often in our explorations to define the epileptogenic zone, and not just in cases of classic insular semiology. Lateralization on the basis of PET, MRI, MEG, and ictal SPECT was false in this cohort up to 25% of the time, depending upon the test and the subgroup. Scalp EEG findings can also lead to incorrect lateralization. In fact, 31% did not have epileptiform discharges on scalp EEG, supporting the idea that some people with negative ictal video EEG may still have electrographic seizures. Thus, patients with concern for an insular SOZ should be investigated bilaterally if sEEG is considered.

Like many excellent papers, this study raises additional questions. As different groups assemble larger numbers of patients with good surgical outcomes, there is an opportunity to better refine which patients are likely to have good results from sampling of different subregions of the insula.