Abstracts

Rationale: A recent study by Weiss, et al.[1] supports the utility of phase-locked high gamma (PLHG), the envelope of the high gamma range (HG, 80-150 Hz) of the EEG scaled by its phase locking to lower frequencies (4-30 Hz), in localizing the seizure onset zone (SOZ) in patients undergoing invasive monitoring for intractable epilepsy. We have recorded intracranial EEG (iEEG) from three subjects before and after corpus callosotomy for poorly-lateralized intractable epilepsy. Since the procedure was successful in isolating the pathological hemisphere, these recordings offer a unique opportunity to test the accuracy of the PLHG biomarker against a ground truth.[1] Weiss SA, Lemesiou A, Connors R, Banks GP, McKhann GM, Goodman RR, Zhao B, Filippi CG, Nowell M, Rodionov R, Diehl B. Seizure localization using ictal phase-locked high gamma A retrospective surgical outcome study. Neurology. 2015 Jun 9;84(23):2320-8. Methods: Three subjects with intractable epilepsy poorly lateralized on iEEG, but strongly suspected to be focal from imaging and clinical investigations, underwent palliative laser corpus callosotomy at the Northwell Health Comprehensive Epilepsy Center. iEEG placed prior to the procedure captured at least one seizure before and after the operation per subject. PLHG, the HG envelope and line length (LL), a proxy for human-readable changes in EEG morphology, were calculated for each seizure, before and after the operation, and for two to four 30-minute interictal pre-operative baseline segments from each subject, matched for sleep state during that subject's seizures. A channel was regarded as having been recruited into the seizure when its value exceeded a variable threshold between 2 and 5 standard deviations above an average of data for that metric pooled across all channels from 6.7 seconds of the immediate preictal period. Data from the different thresholds were combined and the first 8 recruited channels were examined for a preference in laterality. Results: All subjects' post-operative iEEGs were determined to have lateralized by a neurologist experienced in interpreting these studies. In Subject 1, none of the three metrics significantly implicated either hemisphere in the two pre-operative seizures, while PLHG significantly implicated the pathologic hemisphere in the two post-operative seizures (Bernoulli distribution fit, factor compared to more rigorous of 0.5 and fraction of functioning channels on each respective side, p < .05, Bonferroni corrected at n = 3). The other two metrics showed a trend toward significance. In Subject 2, none of the criteria successfully lateralized the 6 pre-operative seizures, while only LL implicated the pathologic hemisphere in the four post-operative seizures. In Subject 3, all three metrics implicated the pathologic hemisphere in the two pre-operative seizures. The one post-operative seizure showed significant lateralization to the pathologic hemisphere by PLHG, while the other two measures indicated a strong trend toward the pathologic hemisphere. Conclusions: In agreement with prior work, in our series of three subjects with poorly-lateralized epilepsy, PLHG appears to be an informative marker in identifying the pathologic hemisphere, however it is not uniformly useful across patients. More studies should be performed to evaluate the accuracy of PLHG relative to other metrics, like line length. Funding: S.K. is supported by the Feinstein MD Scholarship at Northwell Health. A.M. is supported by US NIH grant 1R01MH111439-01.