The study team analyzed intraoperative electrocorticography to link very high-frequency activity (vHFA, 500–2,000 Hz) with interictal spikes. They identified 132,359 spikes across 56 patients and 1,925 vHFA-positive events. Short, reproducible timing differences emerged: the vHFA peak preceded the associated spike peak by a median of 8.0 ms (95% CI: 5.0–16.0). That temporal lead suggests high-frequency activity is not merely coincident noise but a discrete signal temporally related to the spike event. The investigators reported that vHFA-positive spikes were substantially more likely to be classified as co-occurring (spikes occurring across contacts within 50 ms), and this relationship remained after statistical adjustment for both anatomy and inter-electrode distance (OR: 3.43; 95% CI: 2.82–4.17). Short latency, spatial distribution across contacts, and robust adjusted odds ratios together indicate a distinct electrophysiologic signature linked to propagated interictal activity.
The authors then compared spatial distributions of vHFA-positive co-occurring spikes with resection maps in the subset of patients who achieved postoperative seizure freedom (n reported as 35). Contacts with higher rates of vHFA-positive co-occurring spikes were far more likely to lie inside the resected region. Quantitatively, the association was large: odds ratio 11.92 per occurrence/min (95% CI: 5.62–25.32). The same comparison using all spikes yielded a much smaller effect (OR: 1.09 per occurrence/min; 95% CI: 1.06–1.11). These contrasting effect sizes indicate that the vHFA-positive subset of spikes concentrated in resected tissue more strongly than the overall spike population, suggesting added localizing specificity when vHFA is considered.
Across the entire cohort, 20 patients experienced seizure recurrence after resection. The investigators quantified the proportion of vHFA-positive co-occurring spikes that remained outside the resected area and tested its relationship with outcomes. A higher proportion of preserved vHFA-positive co-occurring spikes correlated with seizure recurrence (OR: 3.22 per 0.01 increase in proportion; 95% CI: 1.14–9.12). In practical terms, sites showing these high-frequency–linked, spatially distributed spikes that were not removed appeared more likely to be associated with ongoing seizures. The data therefore connect a specific intraoperative electrophysiologic marker with clinical endpoints.
The team developed seizure outcome prediction models incorporating the rate or proportion of vHFA-positive co-occurring spikes and evaluated model performance across centers using bidirectional external validation. Models that included vHFA-positive co-occurring spikes achieved area under the ROC curves of 0.80 and 0.69; the model developed on the larger sample performed better when tested on the other center. The report provides these AUC values but does not present additional model calibration metrics or decision-threshold analyses in the abstract. These cross-center results support some degree of generalizability, while also revealing variability in discrimination depending on the development sample size.
This retrospective, two-center cohort comprised patients with drug-resistant focal epilepsy who underwent intraoperative electrocorticography followed by resection between January 2016 and May 2025. Spikes were detected and classified: events across contacts within 50 ms were labeled co-occurring, and all other spikes were classified as isolated. The presence of vHFA (500–2,000 Hz) associated with spikes was determined and temporally compared to spike peaks. The investigators adjusted key analyses for anatomy and inter-electrode distance. The authors reported ethics approvals from the Institutional Review Board of Wayne State University and the Ethics Committee of the National Epilepsy Center, NHO Shizuoka Institute of Epilepsy and Neurological Disorders. They also stated that data are available upon reasonable request to the authors.
The report is a retrospective analysis from two centers; the source did not report prospective validation or multicenter randomized assessment. The methods for automated vHFA detection, signal preprocessing, and specific electrode coverage patterns were not detailed in the abstract; the source did not report those procedural specifics in the provided text. The study relied on intraoperative electrocorticography and resection mapping performed at expert centers, which may limit direct extrapolation to different surgical workflows or electrode configurations. The authors noted cross-center variability in model performance, and they made data available on request, enabling independent replication. Remaining questions that the source identified implicitly include whether standardized vHFA detection pipelines will reproduce these associations across broader clinical settings, how intraoperative decision-making would change when vHFA-positive contacts disagree with other localization data, and what incremental benefit vHFA provides relative to established presurgical evaluations. The article ends with these practical limitations and open questions rather than a broad concluding statement.