Abstracts

Rationale: Resective surgery is an effective treatment option for patients with pharmacoresistant seizures. Often, temporal lobe seizures are localized using scalp electroencephalography (EEG); however, in some cases invasive EEG monitoring is needed to adequately localize the seizure onset zone prior to resection. The objective of this study was to determine if invasive EEG monitoring conducted prior to left TL is associated with greater decline in verbal memory and language than left TL alone. Methods: Data for this retrospective study were obtained from an IRB-approved, neuropsychology registry for patients who underwent epilepsy surgery at Cleveland Clinic. Inclusion criteria were: 1) age 16 or older, 2) had left TL that included mesial structures, 3) no evidence for right hemispheric language dominance; 4) no prior brain surgery; and 5) completed pre and postoperative measures of memory (Wechsler Memory Scale) and language (Boston Naming Test). A total of 177 patients (46 with and 131 without invasive monitoring) met these criteria. Each outcome was measured using two methods: 1) difference in standard score from before to after surgery and 2) presence or absence of clinically meaningful decline using established cutoffs from published reliable change indices (RCIs). Seventeen covariates (e.g., cognitive, EEG, MRI, demographic, epilepsy-related) were identified as likely related to either the likelihood of invasive monitoring or to cognitive outcome following epilepsy surgery and were thus included in a propensity score model. The effect of invasive monitoring on cognitive outcome was then estimated, using the propensity score (both via weighting and direct adjustment - the double robust approach) to account for differences in baseline characteristics among patients receiving and not receiving invasive monitoring. Linear and logistic regression models were then used to compare the surgical groups on the cognitive outcomes. Results: Unadjusted estimates suggested greater memory decline in patients who underwent invasive monitoring; however, after weighting and adjustment for the propensity score, these effects were no longer apparent. In contrast, after weighting and adjustment, the estimate for the language effect indicated greater naming decline among those who had invasive monitoring (Score Difference = -4.50; 95% CI -8.50, -0.50). However, when RCIs were used to define clinically meaningful cognitive change, there was no significant treatment effect on either cognitive outcome. Conclusions: Patients with temporal lobe epilepsy who undergo invasive monitoring prior to left TL demonstrate greater declines in naming than those who do not. However, these differences were not clinically meaningful as determined by RCIs. Regardless of how postoperative change is measured (i.e., change score or RCI cutoff), there are no differences in verbal memory outcome as a function of invasive monitoring after adjusting for relevant covariates.