Longitudinal hippocampal and extra-hippocampal microstructural and macrostructural changes following temporal lobe epilepsy surgery

Abstract Objectives 1) Characterize the evolution of microstructural changes in the contralateral, non-operated hippocampus—using longitudinal diffusion tensor imaging (DTI)—following surgery for temporal lobe epilepsy (TLE). 2) Characterize the downstream extra-hippocampal volumetric changes of the fornix and mammillary bodies after TLE surgery. 3) Examine the relationship between these measures and seizure/cognitive outcome. Methods Serial structural and DTI brain MRI scans were collected in 25 TLE patients pre- and post-surgery (anterior temporal lobectomy, ATL – 13; selective amygdalohippocampectomy, SelAH – 12) and in 12 healthy controls. Contralateral hippocampal fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD) and radial diffusivity (RD) were computed with manual hippocampal tracings as volumes of interest following co-registration to anatomical images. Fornix and mammillary body volumetry was performed by manual segmentation. Results After surgery, the non-resected hippocampus showed significant postoperative decline in FA (p = 0.0001), with increase of MD (p = 0.01) and RD (p = 0.0001). In contrast to the timing of our previously reported volume changes where atrophy is observed in the first week, diffusion changes occurred late, taking 1–3 years to develop and are not significant at one week after surgery. Diffusion changes are accompanied by delayed limbic circuit volume loss in the mammillary bodies (35%; p  Significance Differences observed in the timing of postoperative volume (first week) and FA/MD (one year) changes would suggest that early contralateral hippocampal atrophy is not secondary to fluid shifts (dehydration) while the late DTI changes suggest ongoing microstructural changes extending beyond the early postoperative period. Postoperative hippocampal diffusion changes are accompanied by delayed mammillary body and fornix volume loss which did not differ when stratified by seizure outcome nor was correlated with degree of hippocampal diffusion change. Finally, we did not identify any significant correlation between postoperative diffusion parameter change and memory performance.