Abstracts

Rationale: Single-Photon Emission Computed Tomography (SPECT) provides one of the only alternatives to scalp and invasive EEG in the localization of the seizure onset zone in epilepsy surgery evaluations. Traditional visual analysis has been shown to be inferior to ictal-interictal SPECT subtraction with coregistration to MRI (SISCOM) (O’Brien et al, 1998), and to Ictal-interictal SPECT Analysis using Statistical Parametric Mapping (ISAS) (McNally et al, 2005). ISAS compares the patient’s ictal-interictal subtraction images to a group of 14 normal (nonictal) scan pairs to identify statistically significant perfusion differences. We sought to estimate the frequency and type of disparity in localization using the two methods in a series of pediatric epilepsy surgery candidates. Methods: Ictal-interictal SPECT scan pairs were analyzed using SISCOM performed with Analyze software (O’Brien et al, 1998), and ISAS using published methods (McNally et al, 2005, http://spect.yale.edu) with described parameters. For each method, regions of interest (ROI) were characterized as localizing if there was a convincingly dominant region of hyperperfusion. Agreement in localization was characterized as concordant when both methods demonstrated similar ROI (localizing) or if neither method revealed significant ROI (nonlocalizing). Results were discordant when ROIs were demonstrated by only one method or if the methods differed in localization. Results: 25 ictal-interictal SPECT scan pairs were reviewed. Ages ranged from 13 months to 16 years (mean = 9.5 years). SISCOM and ISAS in most cases agreed. Studies were concordant but nonlocalizing in 11 (42%), and both concordant and localizing in 5 cases (25%). 8 cases (32%) demonstrated clear disparities in localization. Of these, 4 cases localized only with SISCOM, 3 cases only with ISAS, and 1 cases by both but with differing localization. Conclusions: In this study, a clearly dominant hyperperfusion localization was identified in over half of the cases by either ISAS or SISCOM, but results are disparate in about a third of the cases. Both SISCOM and ISAS have previously been shown to be helpful in presurgical localization of the seizure onset zone. Previous work (Paige 2007) suggested ISAS is capable of producing results which are more often conclusively localizing with a higher inter-rater reliability than SISCOM in a larger dataset of pediatric and adult presurgical evaluations. The results of this study provide evidence that these two important methodologies using the same dataset can produce results that might provide differing presurgical recommendations in up to a third of cases. No existing study has compared the true positive and false positive localizations of each method in crucial epilepsy surgery invasive EEG and outcome results. Given the high rate of disparity of ISAS and SISCOM localization, there is a compelling need both for further study into the methodological reasons of the disparity and for clinical testing of the localization accuracy of each method for surgical planning and resection.