Abstracts

Rationale: Predicting seizure outcome (SO) after anterior temporal lobectomy (ATL) is a major clinical goal. With clear evidence that even focal epilepsies disrupt brain networks (Tracy et al., 2014), resting-state functional connectivity (rsFC) methods have been increasingly used on a pre-surgical basis to characterize the impact of seizures on brain activity. In this project, we sought to determine whether rsFC measures prior to surgery can discriminate between patients who will obtain good versus poor post-surgical outcome in terms of seizure control. Methods: We studied 36 refractory TLE patients (18 right, 18 left) and 12 healthy controls. All the participants underwent a 5-minute resting-state fMRI session (prior to ATL for patients). All the participants were left-hemisphere dominant for language. The SO was based on seizure status at least one year post-surgery. The patients were identified as either "good" (no seizures since surgery, n=12 in each ATL group) or "poor" outcome (at least one seizure after surgery, n=6 in each group) using adapted Engel classifications (GO and PO, respectively). The cortical brain was parcellated into 90 regions (Tzourio-Mazoyer et al., 2002) and rsFC were individually computed between each pair of regions. We then investigated at the whole brain level, 4 graph-theory properties: modularity, characteristic path length, clustering coefficient and global efficiency. These properties were computed based on binarized maps to maintain the strongest positive connections (from the top 5% to 30%). A bootstrapping strategy (10,000 repetitions in each group; sample with replacement) was applied (Sequeira et al., 2013). Results: The groups did not differ in any clinical or demographic characteristics (Table 1). The data showed significant differences exist between ATL patients on all measures, regardless of the side of the pathology (Fig. 1). Both PO right ATL and GO left ATL patients yielded the more normative data (e.g. closer to controls' values), in comparison to the other two ATL groups (GO left ATL and PO right TLE). Our modularity measure appears to best distinguish the outcome groups, with a GO associated with increased functional segregation in the left TLE group, but reduced segregation in the right TLE group. Conclusions: Our results suggest that the brain organization characteristics associated with a good outcome differs between left and right ATL groups. In our GO left ATL patients, our findings indicate that in the dominant hemisphere greater separation between regions (i.e., a breakdown within more networks) may serve to segregate the epileptogenic zone from the rest of cortex. In the setting of the non-dominant hemisphere, reduced segregation appears adaptive. These distinct patterns imply the neuroplastic responses supporting seizure control and brain recovery differs by hemisphere. Our results suggest that pre-surgery whole-brain rsFC measures may be able to reliably predict seizure outcome following ATL. In the future, we hope to provide specific pre-surgical rsFC algorithms predictive of post-operative seizure status in TLE.