Abstracts

Rationale: The vast majority of histopathological findings in children who are surgically treated for drug-resistant epilepsy include focal cortical dysplasia (FCD), hippocampal sclerosis (HS), nonspecific gliosis (NG), and low-grade epilepsy-associated tumors (LGT). Modern neuroimaging techniques have greatly enhanced the detection of these epileptogenic lesions, but significant uncertainty persists, and many lesions remain difficult to characterize by imaging. Scalp EEG is a well-established and widely available diagnostic tool for the non-invasive estimation of the epileptogenic zone (EZ). Yet, it is unclear whether it could also be utilized to better understand the altered brain pathology underlying the patient’s seizure generation.

We hypothesize that the combined use of scalp EEG source imaging (ESI) and functional connectivity (FC) techniques during interictal epileptiform discharges can help reveal neurophysiological signatures of pathologic tissue.

In this study, we aim to non-invasively estimate the connectivity profiles of the resected brain tissue in 52 children who had epilepsy surgery and to identify whether these connectivity features can help differentiate between various epileptogenic pathologies.


Methods: We analyzed presurgical EEGs (19-24 channels) from 52 children with DRE who had histopathological confirmation of different brain lesions after epilepsy surgery. Each child's resected tissue (RT) was delineated manually after co-registering their pre- and post-op MRI. For each patient, we reconstructed their full-brain cortical activity during spikes (ESI) and estimated the FC of all cortical points in the RT (Fig 1A-B). We quantified the strength of the RT connections with the rest of the brain (outside-FC) as well as the connections within the RT (inside-FC) (Fig 1C) and computed their FC ratio (inside/outside; see Fig 1D) in 5 frequency bands (delta to gamma).

We compared FC ratio values across different epileptogenic pathologies (Fig 1E) by repeated measure ANOVA and MANOVA. We also employed multinomial logistic regression and ROC curves to assess the performance of the FC ratio in distinguishing epileptogenic pathologies.


Results: We included 52 children: 15 had FCD2-3, six FCD1, nine HS, 14 LGT and eight NS (Fig 2A). The repeated measure ANOVA revealed that FC ratio differs between epileptogenic pathologies (F=7.34; p< 0.001) (Fig 2B). The MANOVA analysis revealed a significant difference in the theta FC ratio across pathologies. We found that FCD1 was associated with a higher FC ratio in the theta band than FCD2 or 3 (FCD2+) and HS (p< 0.05). Similarly, HS presented higher FC ratio values compared to NG (p< 0.05; Fig 2C). Theta FC ratio distinguished FCD type 1 from other pathologies with a mean AUC of 0.81 (mean OR [95% C.I.] = -6.2 [-0.9; -11.4]; p< 0.05; Fig 2D)


Conclusions: Our findings indicate that distinct EEG connectivity profiles accompany specific epileptogenic brain pathologies, with FCD1 lesions being particularly distinguishable from others. These correlations between histopathology and neurophysiological measures underscore the potential diagnostic, and perhaps prognostic, utility of preoperative FC source analysis.


Funding: R03NS127044 by the NINDS of NIH