Abstracts

Rationale: Multiple seizure foci are an increasingly recognized cause of epilepsy surgery failures. Identification of the precise localization of seizure onset and propagation is critical to determining management strategies and guiding resection. Previously, we demonstrated that phase-locked high-frequency oscillations (HFOs) are a sensitive and specific indicator of ictal invasion. Here we use ictal HFOs and termination patterns to characterize the evolving spatial structures of seizures and provide a framework for identification of multifocal epileptogenic zones. Methods: Consecutive patients at Columbia University Medical Center undergoing phase II evaluation using stereotactic depth arrays between 2014-2016 were included. Ictal HFOs were defined as previously described using visual identification in Insight (Persyst Inc) and a validated custom filter (FIR, order 90, cutoff 80 Hz, Gibbs 15). Contact localization was performed by coregistration of the pre-implant T1 volumetric MRI with the post-implant volumetric CT using FSL. Distinct HFO clusters were defined as sites separated by at least 2 cm in different anatomic structures. Asynchronous termination was defined by difference in offset time and morphology. Channels were accordingly divided into non-overlapping offset groups. Statistical comparisons were conducted using nonparametric tests of significance. Results: Standard presurgical evaluation suggested multifocal syndromes in 5/13 patients. Ictal HFOs were identified in 50/74 (68%) seizures from 12/13 (92%) patients. HFOs were recorded exclusively from gray matter structures with the exception of one case from adjacent CSF. Multiple HFO clusters were detected in 7/13 cases (absent in 8%, one cluster in 38%, two in 15%, three in 23%, four in 8%, five in 8%). Multiple HFO clusters were associated with asynchronous terminations (OR 7.1, 95% CI 1.57-45.8), with a strong correlation between the number of clusters and the number of offset groups for focal unaware seizures (r_s=0.81, p=4.8x10^-7). No correlation was observed for focal aware seizures, which all had one or two clusters in the same offset group. Pairs of contacts within the same HFO cluster were more likely to share offset groups than pairs from different clusters (median proportion per seizure 1.0 vs 0.33, p=5.0x10^-8). Among the 29 seizures with HFOs in at least two clusters, 14/16 with noncontiguous sites demonstrated asynchronous termination, compared to only 2/13 with contiguous sites. Conclusions: Many focal epilepsies may in fact have multiple distinct seizure foci. Ictal HFOs can identify independent epileptogenic zones and propagation patterns that may be difficult to assess with standard EEG. The association of distinct HFO clusters with asynchronous terminations suggests independent seizure foci and a model of seizures involving multiple independent but interlinked cores. The combined analysis of ictal HFOs and termination patterns is helpful in identifying cases where epileptogenic zones are undersampled and electrode revision may be appropriate. Funding: NIH R01 NS084142 and R01 NS095368