Interictal PAC separated clinical SOZ channels after band-power control in 35 SEEG patients, but less than power or spike rate alone

We tested interictal phase-amplitude coupling (PAC) as a marker of the clinical seizure-onset zone (SOZ) in public SEEG data (OpenNeuro ds004100, HUP). We used one interictal segment of 20–90 s for each patient. We analyzed 35 patients, 3,571 bipolar channels, and 383 SOZ channels. PAC was the Tort modulation index between 3–4 Hz phase and 65–115 Hz amplitude. The analysis plan was written before the statistics. For each patient, we calculated the AUC for SOZ against non-SOZ channels, then tested the mean AUC over patients with a sign-flip test. PAC was higher on SOZ channels (mean AUC 0.641, 95% CI 0.586–0.695, 28 of 35 patients above 0.5, Holm p = 1e-5). After removing the linear effects of 3–4 Hz and 65–115 Hz power within each patient, PAC still separated the channels (mean AUC 0.594, CI 0.544–0.643, Holm p = 8.6e-4). After also removing spike-like transient rate, the effect was small (mean AUC 0.551, CI 0.508–0.596, p = 0.029). PAC alone separated SOZ channels less well than 3–4 Hz power (0.731), 65–115 Hz power (0.693), or spike rate (0.710). The result applies to clinical SOZ labels, short segments of unknown sleep state, one band pair, and linear covariate control. The package includes the analysis code and plan, byte ranges and hashes for the fetched data, channel and patient results, and reproduction instructions.

phase-amplitude coupling · seizure-onset zone · stereo-EEG · intracranial EEG · epilepsy · OpenNeuro · AUC · interictal