Abstracts

Rationale: Outcome data supporting the value of fMRI language mapping prior to epilepsy surgery are limited, and no studies have quantitatively assessed whether surgical resection of ""functionally active"" tissue affects outcome. Moreover, fMRI language mapping methods vary widely across centers and show highly variable ability to detect language-related activation in the anterior temporal lobe. The fMRI in Anterior Temporal Epilepsy Surgery (FATES) study is a 5-year NIH-funded, multi-center, prospective observational study aimed at testing whether preoperative fMRI language lateralization and/or resection of functionally activated tissue predict language and verbal memory outcomes after left anterior temporal lobe surgery. A secondary goal is to assess the feasibility of implementing a standardized language fMRI protocol across multiple centers and imaging platforms.Methods: The study focuses on patients with medication-resistant non-lesional left temporal lobe epilepsy. Enrollment began in mid-2012 at 8 academic epilepsy centers in the U.S. A ninth center was added in 2014. All MRI is performed on 3T scanners, which include 1 GE, 3 Philips, and 5 Siemens platforms. Weekly quality assurance measures are obtained on each scanner using a standard (FBIRN) phantom and acquisition protocol. Prior to surgery, patients undergo structured medical and seizure history interviews, two fMRI protocols optimized for language lateralization and anterior temporal lobe activation, high-resolution structural MRI, resting-state functional connectivity MRI, and a complete neuropsychological battery. Ancillary testing (Wada, MEG, PET, SPECT, cortical stimulation mapping) is encouraged but not required. Surgery must be performed blind to the fMRI data. The neuropsychological battery is repeated at 6 months after surgery. Postoperative structural MRI is obtained for voxel-based mapping of the surgical resection.Results: 108 patients have been enrolled as of 6/1/2015, for an average annual enrollment rate of 4.5 patients per center (range 2.5-8.1). Of the enrolled patients, 92 participated in the preoperative fMRI protocol. Usable fMRI data, defined as successful image acquisition, above-chance task performance, and <20% movement-contaminated images across both task protocols, were obtained in 84/92 or 91.3% of patients. Of these patients, 63 subsequently underwent surgery, and 48 have completed the entire study protocol including 6-month follow-up testing. Preliminary hypothesis testing is currently underway using this dataset.Conclusions: The results demonstrate the feasibility of implementing an optimized, standard fMRI language mapping protocol across multiple centers and imaging platforms. Rates of enrollment, though below expectations, are projected to yield 70-75 complete datasets by the end of the study, which will be adequate for measuring the relationships of primary interest, and for exploring a range of secondary hypotheses concerning predictors of seizure control and quality of life.