Authors :
Presenting Author: Ionuț-Flavius Bratu, MD, PhD candidate – Hôpital de la Timone
Isabelle Lambert, MD, PhD – Hôpital de la Timone
Eve Tramoni-Negre, PhD – Hôpital de la Timone
Samuel Medina Villalon, Eng – Hôpital de la Timone
Agnès Trébuchon, MD, PhD – Hôpital de la Timone
Fabrice Bartolomei, MD, PhD – Hôpital de la Timone
Rationale:
Memory consolidation relies on hippocampal–neocortical transfers to stabilize new information, a process that can be disrupted by epileptic activity. The postictal state, the longest continuous seizure-induced dysfunction, remains little studied in this context. Recent postictal stereo-EEG (SEEG) evidence suggests that preserved cerebral complexity is a prerequisite for optimal cognitive function (Bratu at al, Ann Neurol. 2025). We hypothesized that postictal hippocampal complexity disruption would be associated with impaired long-term memory consolidation.
Methods:
We prospectively included 20 patients explored by SEEG with hippocampal sampling and cognitive–behavioral status adequate for memory assessment (Lambert et al, Ann Neurol. 2020). Verbal memory was assessed with RL/RI-16 (French Free and Cued Selective Reminding Test) and the Wechsler Memory Scale (WMS) III – Logical Memory subtest (history recall-recognition); visual memory with a modified Delayed Matching to Sample test and the WMS III abstract drawings subtest. Memory performance was evaluated at immediate, 30-minute, and 1-week delays. 1-week retention was calculated as (performance at 1 week)/(performance at 30 min)×100. Postictal hippocampal complexity disruption duration was quantified using the Postictal Alteration Time (PAT) tool (Bratu et al. Ann Clin Transl Neurol. 2024), a permutation-entropy–derived measure, across all ictal events (electro-clinical seizures and subclinical discharges with clear-cut non-interictal spatio-temporal evolution) between the 30-min and 1-week tests. For PE analysis we used an embedding dimension of 3 samples, lag of 1, and 5-second windows with 2.5-second overlap. All ictal–perictal traces were downsampled to 256 Hz, with a 30-min baseline taken before spontaneous seizures or stimulation sessions. The maximum PAT per seizure across hippocampal contacts was summed to yield a cumulative postictal burden (c-PAT). Left hippocampal PATc was correlated with verbal memory, right hippocampal c-PAT with visual memory, using Spearman’s ρ with Benjamini–Hochberg False Discovery Rate (FDR) correction. Hippocampal epileptogenicity was quantified using the (Connectivity) Epileptogenicity Indices and validated thresholds (Makhalova et al, Ann Clin Transl Neurol. 2023).
Results:
Of 30 analyzed hippocampi (15 left, 15 right), 8 left and 11 right were quantified as epileptogenic. The mean number of ictal events per patient was 7.5 (1-28 range). Visual memory assessments could be performed in 19 patients and verbal memory in 17. Left hippocampal c-PAT correlated with poorer 1-week verbal consolidation (FDR-p< 0.05):