Prognostic Value of High-Frequency Oscillations Combined with Multimodal Imaging Methods for Epilepsy Surgery
Epilepsy surgery remains a critical intervention for patients with medically refractory epilepsy, aiming to achieve seizure freedom through the precise localization and complete removal of epileptogenic zones (EZs). Traditional noninvasive imaging techniques, such as magnetic resonance imaging (MRI), positron emission tomography (PET), and magnetoencephalography (MEG), play pivotal roles in preoperative evaluation. However, challenges arise in cases with non-lesional MRI or discordant findings across modalities, necessitating invasive monitoring via stereo-electroencephalography (SEEG). High-frequency oscillations (HFOs), particularly ripples (80–200 Hz), have emerged as promising biomarkers for EZ localization. This study evaluates the prognostic value of combining ripples with multimodal imaging (PET and MEG) to improve surgical outcomes.
The study cohort comprised 21 patients (13 men, 8 women) with drug-resistant epilepsy who underwent SEEG-guided resection surgery. The mean age at seizure onset was 9.6 years, with a mean disease duration of 13.0 years. All patients underwent PET, while 11 also underwent MEG. Postoperative outcomes were assessed using the Engel classification system, with a follow-up period ranging from 13.37 to 42.70 months. Eleven patients (52.4%) achieved seizure freedom (Engel class I), highlighting the heterogeneity of surgical outcomes in this population.
Methodological Framework and Key Findings
SEEG and Ripple Analysis
Intracranial SEEG recordings were performed using electrodes with 8–16 contacts, covering putative EZs identified during preoperative evaluations. HFOs were analyzed from 5-minute interictal segments during slow-wave sleep, focusing on ripples due to sampling rate constraints. Ripples were considered clinically significant if they occurred densely (>10 events in 5 minutes). Regions generating ripples were compared against resection areas to determine prognostic accuracy.
Ripples alone demonstrated moderate accuracy (42.9%) in predicting surgical outcomes, with sensitivity and specificity of 50.0% and 36.4%, respectively. Resection of ripple-generating areas did not significantly correlate with seizure freedom (P > 0.05). However, the spatial distribution of ripples often extended beyond the eventual resection zone, suggesting physiological or propagated activity in non-epileptogenic regions.
PET and MEG Contributions
PET identified hypometabolic regions in all patients, but its accuracy (42.9%) mirrored that of ripples. Only seven patients had resection areas overlapping with PET findings, underscoring limitations in extratemporal lobe epilepsy localization. MEG, performed in 11 patients, demonstrated superior accuracy (81.8%), with sensitivity and specificity of 85.7% and 75.0%, respectively. MEG dipoles localized clusters of interictal spikes, but resection of these regions alone did not guarantee seizure freedom (P = 0.157).
Multimodal Integration
Combining ripples with PET (Group 1, n=21) marginally improved accuracy (38.1%) but lacked statistical significance (P = 0.259). In contrast, combining ripples with MEG (Group 2, n=11) yielded higher specificity (100%) and positive predictive value (PPV: 100%), though limited by sample size. Strikingly, the triad of ripples, PET, and MEG (Group 4, n=11) achieved the highest accuracy (90.9%), with sensitivity and specificity of 85.7% and 100%, respectively. Patients with complete resection of regions concordant across all three modalities exhibited significantly better outcomes (P = 0.008).
Case Illustrations
Two cases exemplify these findings. A 6-year-old boy with nocturnal seizures underwent MEG, which localized interictal spikes to the right frontal lobe. SEEG ripples overlapped with this region, and resection led to sustained seizure freedom (44-month follow-up). Another patient with insular epilepsy showed hypometabolism on PET and ripples in the left temporal lobe. Resection of the PET- and ripple-concordant zone resulted in seizure control, emphasizing the synergy of multimodal data.
Clinical Implications and Limitations
The study underscores the necessity of integrating HFOs with noninvasive imaging to optimize SEEG electrode placement and resection planning. While MEG and PET provide critical preoperative insights, ripples offer complementary electrophysiological data, particularly in MRI-negative cases. The superior prognostic value of multimodal integration (ripples + PET + MEG) suggests a paradigm shift toward leveraging diverse biomarkers for precision surgery.
However, limitations include the small cohort size and exclusion of fast ripples (200–500 Hz) due to technical constraints. Future studies should explore noninvasive HFO detection via scalp EEG or MEG to enhance accessibility. Additionally, larger cohorts and longitudinal data are needed to validate these findings.
Conclusion
This study establishes that combining ripples with PET and MEG significantly enhances the accuracy of EZ localization and surgical outcome prediction. Intracranial HFO recording remains indispensable, but preoperative imaging guides electrode implantation, improving diagnostic yield. As epilepsy surgery evolves, multimodal approaches integrating electrophysiological and metabolic data will be essential for achieving optimal patient outcomes.
doi.org/10.1097/CM9.0000000000001909
Was this helpful?
0 / 0