Abstracts

Rationale: Understanding how an epileptic “focus” interacts with other brain areas as part of a network can guide therapy, including informing decisions about surgery. Functional connectivity analysis of fMRI allows noninvasive assessment of correlations in the time course of the BOLD signal across brain regions, delineating functional networks. While fMRI is most commonly used in the context of cognitive or sensory activation paradigms, it can also be used without any task, i.e. during the resting state, to learn about interictal aspects of epilepsy. Here, we apply functional connectivity analysis of resting state fMRI to healthy controls and pts with temporal lobe epilepsy (TLE). Based on prior work suggesting an important, dynamic relationship between bilateral hippocampi in TLE, we focus this pilot study on inter-hippocampal connectivity.Methods: 5 pts (3 with severe TLE [> 4 sz/month]; 2 with <1 sz/year; all with normal or near-normal structural MRI) and 25 controls were scanned for 5 min at rest using an EPI sequence with z-shimming to maximize signal in medial temporal regions. Image processing included realignment, extraction of physiological fluctuations, coregistration to anatomic images, normalization to MNI space, and spatial smoothing. The adjusted time series signal from a single “seed” voxel in each anterior hippocampus was entered into a voxel-wise multiple linear regression model. Single-subject maps depicting brain areas in which activity correlated temporally with seed activity were generated. Control maps were entered into a 2nd-level analysis to generate an average map depicting “normal” hippocampal connectivity. For each seed, the number of significantly correlated voxels was computed within the contralateral hippocampus (defined by standard mask) and over the whole brain. The ratio between these values served as an index of relative hippocampal functional connectivity. The normal range for this ratio was defined as 2 std +/- the geometric mean of 24 controls.Results: The normal pattern of strong, symmetric bilateral hippocampal connectivity (likely mediated in part by a direct anatomic connection: the hippocampal commissure) was disrupted in the 3 pts with severe TLE (see figure). When quantified, relative hippocampal functional connectivity was below the normal range on the affected side in 2 of 3 pts with severe TLE (1.75 std below normal mean in the 3rd pt), but was normal in mild TLE. An abnormal ratio can be understood as reflecting two factors: 1) decreased inter-hippocampal connectivity, which likely relates to expected dysfunction of an epileptic hippocampus, as well as active inhibitory processes within and between hippocampi; and 2) increased connectivity of an epileptic hippocampus with the rest of the brain, which may reflect propagation of interictal epileptic activity.Conclusions: Identification of normal patterns of hippocampal functional connectivity and abnormalities associated with severe TLE support use of functional connectivity analysis of resting state fMRI to investigate network aspects of human epilepsy. (supported by Weill-Cornell GCRC, NIH M01 RR00047)