In 43 SEEG patients, spindle deficits at SOZ and resected pairs did not differ by ILAE outcome (180 s NREM; Holm p = 1.0)

We assessed whether local spindle deficit in non-rapid eye movement (NREM) sleep related to surgical outcome in 43 patients with stereo-electroencephalography (SEEG) data. The analysis used archived per-pair spindle features and clinical seizure-onset-zone (SOZ), resection, and ILAE outcome labels for 728 bipolar pairs from 180 seconds of NREM sleep per patient. We compared the per-patient area under the curve (AUC) for spindle density in SOZ versus non-SOZ pairs (H1) and resected versus non-resected pairs (H2) between seizure-free (ILAE 1) and not seizure-free (ILAE 2-6) groups, using two-sided label permutation tests and Holm correction. Neither hypothesis was supported: H1 means were 0.341 and 0.356 (24 versus 19 patients; permutation p = 0.83; Holm p = 1.0; outcome-prediction AUC 0.53, 95% CI 0.35-0.71); H2 means were 0.330 and 0.335 (10 versus 11; p = 0.96; Holm p = 1.0; outcome-prediction AUC 0.47, 95% CI 0.20-0.75). Across outcome groups, mean spindle AUC was 0.348 at SOZ pairs (n = 43, p = 0.0001) and 0.332 at resected pairs (n = 21, p = 0.0004), indicating a spindle deficit in both groups. The fraction of analyzed pairs resected differed between outcome groups (0.716 versus 0.362, p = 0.0035). The relative-sigma control had p = 0.061; it was not corrected or prespecified and is not evidence. The small H2 groups and limited 180-second recordings leave uncertainty, including an H2 outcome-AUC 95% CI of 0.20-0.75. Clinical labels, pooled centres, and subset contact selection further limit interpretation.

SEEG · NREM sleep · sleep spindles · epilepsy surgery · ILAE outcome · resection · spindle deficit