Resident Physician University of Pittsburgh Pittsburgh, Pennsylvania, United States
Introduction: The stereo EEG (sEEG) methodology requires the formulation of specific anatomo-clinico-electrical hypotheses prior to electrode implantation. The purpose of implantation is to interrogate these hypotheses to guide subsequent surgical intervention for drug-resistant epilepsy. However, formulating specific hypotheses and implantations in temporal lobe epilepsy (TLE) can be challenging. Accordingly, utilizing our large surgical team’s experience of over 800 sEEG cases, we developed a classification scheme for sEEG exploration in TLE.
Methods: We identified N=59 patients (ages 19-77) with TLE who underwent sEEG (2020-2022). After reviewing each patient’s multidisciplinary discussions, four implantation hypotheses were proposed – temporal, temporal/basal/occipital, anterior perisylvian, and perisylvian. Retrospective chart review of each patient’s history, implantation strategy/rationale, and outcomes was performed. This classification was quantitatively validated against semiology and anatomic coverage using K-modes/means clustering with the Rand Index (RI). The clinical validity of this classification scheme was assessed by comparing classification to successful localization of the epileptogenic zone (EZ) and seizure freedom.
Results: Classification results were – temporal: 13.6%, temporal/basal/occipital: 25.4%, anterior perisylvian: 30.5%, perisylvian: 30.5%. Age at sEEG (p=0.32), sex (p=0.67), and handedness (p=0.33) did not vary by classification. The median number of electrodes varied (temporal: 11, temporal/basal/occipital: 15, anterior perisylvian: 14.5, perisylvian: 14, p=0.039). Quantitative validation against clustering by semiology (RI=0.63) and lobar coverage (RI=0.73) showed moderate agreement. The EZ was successfully localized in 79.7% of cases and did not vary by classification (p=0.22). The presence of an MRI lesion did not associate with hypothesis (p=0.30) or successful EZ localization (p=1.00). Seizure freedom (Engel Ia) at 6 months was 71% and did not differ by classification (p=0.29).
Conclusion : Our proposed “pattern of sEEG implantation” classification of TLE hypotheses is well-validated against semiology and anatomic coverage. This classification provides reproducible and consistent anatomical framework in sEEG, guiding future sEEG explorations and classifying TLE into meaningful categories to facilitate clinical/translational research.
