Abstracts

Rationale: Temporal lobe epilepsy (TLE) has been associated with widespread gray matter atrophy determined by MRI [1]. In this study we quantified regional gray matter volumetric covariance [2] across a population of TLE subjects, similar to that used by Bernhardt et al. [3] with cortical thickness. In this context, volumetric covariance identifies regions throughout the brain that are similarly affected by TLE. This analysis may be sensitive to effects not detected by standard volumetric group difference comparisons, and so may provide complementary evidence of the structural impact of TLE.Methods: Structural T1 weighted 3T MRI (3DTFE, TR=8.9 ms, TE=4.6 ms, flip=8°, 1mm3) was collected on 44 TLE patients (18 LTLE, 26 RTLE) and 44 healthy controls individually age and gender matched to each patient (mean age difference 0.3 ±1.6 yrs, LCON, RCON). Images were preprocessed via bias field correction and denoising, gray matter segmentation with SPM12, and non-linear normalization and modulation. Ninety AAL atlas regions were identified. After linear regression of age, age2, and total brain volume, Spearman correlations between all pairs of regional volumes were used to create a 90 by 90 correlation matrix for each group (LTLE, RTLE, LCON, RCON). Then Fisher Z difference maps LTLE-LCON and RTLE-RCON were computed. A pair permutation with 50000 trials was used to determine 95% (two-tailed) confidence of the difference for each cell in the Z difference matrices. This method creates a null distribution by randomly assigning a group to each subject of a TLE/CON pair. In this way, variance due to age and gender across the groups in each permutation are the same.Results: The Z-Difference maps are shown in Figure 1 with red indicating increased correlation in TLE, and blue indicating decreased. The percentage of total pairs possible per hemisphere with significant difference from controls is shown in Figure 2. TLE patients had unusually high covariance among structures in the hemisphere contralateral to the seizures. In LTLE this increase is most concentrated in right temporal to right subcortical regions, right frontal to right temporal, and within right frontal regions. In RTLE the increases are most prominent between left limbic, occipital and parietal regions.Conclusions: This study identifies covariance of gray matter volumes across TLE, while minimizing the effect of age and gender. Our results suggest that some characteristics of epilepsy such as duration, seizure frequency and medication are causing synchronous volumetric changes across networks of contralateral regions that are normally unrelated by heredity, age, etc., with LTLE and RTLE primarily affecting different regions. While it is known that TLE results in ipsilateral atrophy, these results imply that atrophy occurs across ipsilateral regions in the same relative manner as in controls, even if the magnitude in TLE is greater. [1] Coan et al. PlosOne 2014;9(1) [2] Alexander-Bloch et al. Nature Reviews Neuroscience 2013;14:322. [3] Bernhardt et al. Cerebral Cortex 2011;21:2147. [Funded by NIH R01 NS75270 – VLM]