Abstracts

Rationale: Predictors of good surgical outcome in medically refractory temporal lobe epilepsy include: MTS on MRI, MRI concordant ictal EEG, and unitemporal interictal epileptiform discharges (IEDs). Predictors of bad outcome include: discordant IEDs and multiple seizure foci. Though EEG monitoring weighs heavily in surgical decision-making, precise guidelines have yet to be set. Specifically, it is unknown how many seizures are required to ensure that discordant seizures are not missed. We evaluated whether the number of seizures and poorly-localized seizures (PLSz) recorded during pre-surgical VEEG monitoring for rTLE were associated with surgical outcome, and explored which factors determine how many of each are captured.Methods: We reviewed 32 patients with VEEG and temporal lobe resection and at least one year of follow up. Clinical characteristics were evaluated for association with surgical outcome, and number of seizures and PLSz recorded in the EMU. Specific subject characteristics were evaluated for effect on seizure freedom, total number of seizures recorded and total number of poorly localized seizure recorded. Predictor associations with seizure freedom were examined with Fisher s exact tests and Student s t-test. Generalized linear regression was used to examine predictor associations with number of seizures and PLSz recorded in the EMU.Results: Only longer duration of epilepsy was associated with seizure freedom (mean 27.5 vs 14.9 yrs, p=0.007) (Table 1). Number of seizures recorded during EMU was not associated with seizure freedom (p=0.475). There was a trend of subjects with fewer recorded PLSz having better surgical outcome (p=0.0523). Higher number of recorded seizures was associated with discordant EEG findings (p=0.015), PLSz (p=0.006), and seizure clusters during VEEG (p<0.001) (Table 2). There was no correlation between patient-reported seizure frequency and seizures recorded. Poorly localized seizures were associated with multifocal ictal pattern (p=0.0077), longer duration of epilepsy (p<0.0001), and higher number of failed AEDs (p=0.002). Number of pre-EMU seizure types reported by the subject did not correlate to capturing PLSz (p=0.0755) but confirmation of multiple seizure semiologies during EMU did (p=0.0011). Seizure clusters were associated with higher number of PLSz (p=0.0453). Conclusions: We did not observe a relationship between the total number of seizures in the EMU and surgical outcome, however, there was a trend indicating that a larger number of poorly localized seizures is associated with worse surgical outcome. We found that more seizures recorded, more days with seizures and more seizure clusters were all associated with higher numbers of PLSz. However, a prior discordant EEG patterns were associated with reduced total seizures observed in EMU. Our results suggest that there is a not a magical number of recorded seizures to ensure good outcome, but the presence or absence of PLSz may affect both how many seizures are recorded and subsequent surgical decision-making.