Automated fibre quantification of the fornix predicts outcome after surgery for intractable temporal lobe epilepsy
Russell Glenn1, Leonardo Bonilha1, Barbara Kreilkamp2, Mark P Richardson3, Bernd Weber4, and Simon S Keller2

1Medical University of South Carolina, Charleston, SC, United States, 2University of Liverpool, Liverpool, United Kingdom, 3King's College London, London, United Kingdom, 4University Hospital Bonn, Bonn, Germany

Synopsis

Imaging markers of postoperative seizure control in refractory temporal lobe epilepsy (TLE) would provide a useful clinical tool for surgical decision making. In the present diffusion tensor imaging study, we report that regional tissue characteristics of the fornix ipsilateral to the side of intended resection are related to postoperative seizure control in patients with TLE. Interestingly, areas found to be abnormal only in patients with a suboptimal outcome were located outside the margins of resection. The identification of fornical abnormalities outside the area of intended resection may be an important prognostic marker of suboptimal seizure control after temporal lobe surgery.

Background and Purpose

The reasons for persistent postoperative seizures (PS) in patients with refractory temporal lobe epilepsy (TLE) due to hippocampal sclerosis (HS) remain unknown. Despite few applications, preoperative diffusion tensor imaging (DTI) may provide new insights into the causes of postoperative seizures after temporal lobe surgery. The vast majority of tractography approaches in epilepsy have quantified mean scalar metrics of whole white matter tracts, which can be insensitive to regional tract pathologic alterations. In the present study, we reconstructed the fornix from DTI data and quantified regional tissue properties along the fornical pathway to investigate whether preoperative fornical tract profiles were related to postoperative seizure outcome in patients with refractory TLE. Non-invasive neuroimaging prognostic markers that stratify patients according to likely postoperative outcome would be extremely helpful for clinical decision making and preoperative counselling.

Methods

We recruited 42 patients with TLE due to HS and 44 healthy age- and sex-matched controls into this study. All patients underwent comprehensive preoperative evaluation, preoperative MRI including 60-direction DTI, amygdalaohippocampectomy, and postoperative seizure outcome assessment, as described recently.1 Controls received the same imaging protocol. We reconstructed the left and right fornix and generated fornical tract profiles from the DTI data for each participant using an extension of automated fibre quantification (AFQ) methods.2 Each tract profile was reconstructed for mean diffusivity (MD) and fractional anisotropy (FA) along 100 nodes between two reference regions of interest (ROIs) which were defined based on anatomical landmarks on a T1 template image. Group-wise comparisons were made between controls and patients with an excellent (ILAE 1) or suboptimal (ILAE 2+) postoperative seizure outcome using the standardised International League Against Epilepsy (ILAE) outcome classification.3 Tract profiles for each patient were separated into ipsilateral and contralateral side based on the side of seizure onset determined during presurgical evaluation. For statistical analyses, every 20 consecutive nodes were averaged to generate 5 distinct fornical ROIs and statistical tests were performed between each group using a two-sample t-test at a significance level of α = 0.05. Multiple comparisons were corrected for using false discovery rate (FDR) thresholding with n = 30 total comparisons. Node-wise t-scores were also reconstructed along the length of the tract to demonstrate regional group-wise differences, anatomically. The basic fornix identification and segmentation scheme is illustrated in Figure 1. Tract profiles and ROIs begin with the lowest node or ROI number at the top of the thalamus and continue along fornix into the medial temporal lobe with the most anterior portion of the segmented tracts corresponding to the highest node or ROI number.

Results

22 (52%) patients were found to have an excellent postoperative seizure outcome and 20 (48%) had a suboptimal outcome. There were no significant differences in the FA tract profiles, but significant differences were detected in in the MD tract profiles for both patient groups in the ipsilateral temporal lobe. Significant differences were detected in patients with suboptimal surgical outcomes outside of the ipsilateral temporal lobe up to ROI 2 with a trend towards significant differences in ROI 1, but not in patients with excellent response. Contralateral fornices demonstrated more normal tract profiles although a significant difference was detected in ROI 3 in the patients with excellent outcomes. MD tract profiles are illustrated in Figure 2.

Discussion and Conclusion

We observed that patients with a suboptimal postoperative outcome deviated from healthy controls and patients with an excellent outcome in tissue properties of the ipsilateral fornix extending outside of the temporal lobe. This abnormal tissue is located outside the margins of resection of conventional temporal lobe surgery, and may contribute to support a postoperative epileptogenic network. This extends recent findings indicating that thalamohippocampal pathways – that are largely mediated by the fornix – are preferentially abnormal in patients with persistent postoperative seizures.1 Abnormal tissue characteristics of the ipsilateral fornix outside the margins of resection may be a candidate imaging prognostic marker of continuing postoperative seizures in refractory TLE.

Acknowledgements

This work was funded by a UK Medical Research Council grant awarded to SSK (MR/K023152/1).

References

1. Keller SS, Richardson MP, Schoene-Bake JC, et al. Thalamotemporal alteration and postoperative seizures in temporal lobe epilepsy. Ann Neurol. 2015 May;77(5):760-74.

2. Yeatman JD, Dougherty RF, Myall NJ, Wandell BA, Feldman HM. Tract profiles of white matter properties: automating fiber-tract quantification. PLoS One. 2012;7(11):e49790.

3. Wieser HG, Blume WT, Fish D, et al. ILAE Commission Report. Proposal for a new classification of outcome with respect to epileptic seizures following epilepsy surgery. Epilepsia. 2001 Feb;42(2):282-6.

Figures

Figure 1. Fornix segmentation. (A-B) Left and right fornices were identified by tracts passing through the blue and yellow rectangles, respectively. (C) Example fornices segmented for an individual patient. (D) Tract profile are divided into 5 ROIs beginning at the top of the thalamus and continuing into the temporal lobe.

Figure 2. Preoperative mean (± s.e.m) tract profiles for MD and FA for ipsilateral and contralateral fornices. The * indicates significant differences (p < 0.05) in comparison to tracts in the control group after correcting for multiple comparisons with FDR thresholding. ROI numbers correspond to the regions in Figure 1D.

Figure 3. T-score plots indicate locations with group-wise differences compared to the control group. Ipsilateral fornices demonstrate differences within the temporal lobe in TLE, but the differences extend outside of the temporal lobe for the ILAE 2+ group (arrow). The contralateral fornices are less affected in TLE.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
0778