Hypothesis-driven patterns of intracranial exploration in temporal lobe epilepsies

Abstract

AbstractBackgroundSurgical treatment is a highly effective option for drug-resistant temporal lobe epilepsy (TLE). When non-invasive exploration is insufficient to localize the epileptogenic zone, anatomo- electroclinical (AEC) hypotheses can be interrogated by stereoelectroencephalography (SEEG) methodology. To facilitate more effective exploration and intervention, we developed a classification system of AEC hypotheses for temporal lobe SEEG explorations: Mesial lateral temporal (MLT), temporal basal occipital (TBO), anterior perisylvian (AP), and perisylvian (P).MethodsIn a cohort of 60 patients with drug-resistant TLE who underwent SEEG, we collected demographic data, clinical/epilepsy history, neuroimaging, seizure semiology, and the analysis of the multidisciplinary epilepsy patient management conference. We identify and describe the 4 patterns of hypotheses, analyze the semiological and structural features, and validate the proposed classification system using a random forest classifier machine learning algorithm.FindingsUsing this approach, we identify the features most predictive of each classification, and demonstrate an overall 90% classification accuracy using semiological features and 97% accuracy using electrode location. Age, sex, and the presence of an MRI abnormality did not vary by classification. We qualitatively explain the features of each classification with case examples. Finally, we specifically provide indirect targeting coordinates for each electrode to facilitate reproducible SEEG explorations. Following SEEG exploration, 94% patients underwent surgical intervention (82% selective resection, 12% neuromodulation). In resection patients, one-year seizure freedom rate was 73% and did not vary significantly by classification (MLT: 86%, TBO: 71%, AP: 75%, P: 68%; p=0.9) with overall 26% verbal memory decline.InterpretationThe patterns of explorations, grounded in semiology and cortical cytoarchitectonic and functional features, guided volumetrically-restricted resections that resulted in a high rate of seizure freedom. This approach guides both a theoretical approach to TLE and a practical realization that can be tailored to the individual patient.Research in contextEvidence before this studyWe searched the MEDLINE database from inception to June 1, 2024 using the terms (“temporal lobe epilepsy”, “mesial temporal lobe epilepsy”, “anatomo-electroclinical hypothesis”, “stereo EEG”, “neuropsychological outcome”, “surgical strategy”, “intracranial exploration”, and/or “seizure freedom”. We supplemented this with search of Google Scholar and reference list. After review, we selected 1 randomized controlled trial (RCT), 17 cohort studies, 9 technical reports, 1 epidemiological report, and 2 systematic reviews. The RCT compared surgical treatment of temporal lobe epilepsy vs. medical management using a standardized temporal lobectomy approach in all patients and demonstrated significant benefit for surgical treatment. However, it did not address the process of anatomo-electroclinical hypotheses to guide intracranial exploration and selective temporal resections nor did it address neuropsychological outcomes.The cohort studies varied in focus, but described various semiological features, surgical approaches, and neuropsychological outcomes. Despite these reports, a systematic approach to intracranial exploration using SEEG and detailed analysis of seizure freedom and neuropsychological outcomes is lacking.Added value of this studyOur study addresses this limitation in one of the largest cohorts of TLE patients with SEEG explorations to date. Using semiological observations and a detailed cytoarchitectonic analysis, we propose a novel hypothesis classification system of four hypotheses in TLE and describe this in detail. We quantitatively validate this approach using a machine learning based approach and provide details to facilitate a practical implementation. Finally, we demonstrate a high rate of seizure freedom with excellent neuropsychological outcomes using this approach.Implications of all of the available evidenceIn this study, we demonstrate that that an appropriately designed intracranial exploration (based on the proposed hypothesis classification) can provide excellent results without necessitating a standard temporal lobectomy. This challenges the prior selective vs. standard temporal dichotomy in the literature but builds on several themes previously reported. This precision and patient-centered approach integrates multimodal information and multidisciplinary discussion resulting in more selective surgical interventions that maximize seizure freedom while minimizing neuropsychological morbidity.