Cytoarchitectonic abnormalities along white matter pathways in temporal lobe epilepsy: Combining diffusional kurtosis imaging and automated fiber quantification
Russell Glenn1, Jens H Jensen1, Simon S Keller2, Joseph A Helpern1, and Leonardo Bonilha3

1Medical University of South Carolina, Charleston, SC, United States, 2University of Liverpool, Liverpool, United Kingdom, 3Charleston, SC, United States

Synopsis

Temporal lobe epilepsy (TLE) is the most common form of medically refractory epilepsy and is associated with focal brain abnormalities causing recurrent, unprovoked seizures originating from the temporal lobe. However, cytoarchitectronic changes can be detected outside of the temporal lobe and may be associated with the clinical course of the disease. We implement a novel neuroimaging approach which combines the strengths of diffusional kurtosis imaging and automated fiber quantification for the non-invasive characterization of white matter pathways and demonstrate its sensitivity to detect pathological alterations associated with TLE. The proposed technique may provide further insights into the clinicopathology of TLE.

Background and Purpose

Temporal lobe epilepsy (TLE) is associated with seizure onset and accompanying structural abnormalities in the temporal lobe of the brain, with epileptogenesis often being associated with mesial hippocampal sclerosis. However, a growing body of evidence suggests that structural changes in TLE extend beyond the temporal lobe, and the characterization of phenotypical abnormalities in TLE may provide important insight into the largely variable clinical courses of the disorder. In the present study, we implement a novel neuroimaging approach which combines the strengths of diffusional kurtosis imaging (DKI)1,2 and automated fiber quantification (AFQ)3 for characterizing pathological changes in specific white matter (WM) pathways that constitute important conduits of information to and from the temporal lobe.

Methods

Study Participants: A cohort of 32 adult subjects with left TLE (mean age 44.8, SD 16.7; 22 females) and 36 age and gender matched controls (mean age 40.4, SD 11.6; 24 females) were included in this study. All subjects were diagnosed with left TLE in concordance with the diagnostic criteria proposed by the International League Against Epilepsy (ILAE), including a comprehensive medical history and a full neurological evaluation.4 The subject population had varying disease severity and included subjects with recently diagnosed, well-controlled, and medication refractory TLE. Image Acquisition: DKI datasets were acquired with a 3T Magnetom Verio MRI scanner (Siemens Medical, Erlangen, Germany) using a vendor-supplied, single-shot diffusion-weighted EPI sequence with a twice refocused spin echo and a 12-channel head coil. The protocol included 3 diffusion weightings of b = 0, 1000, and 2000 s/mm2, 30 isotropically distributed diffusion encoding directions, a total of 10 b = 0 images, TR / TE = 8500 / 98 ms, voxel dimensions = 3.0×3.0×3.0 mm3, matrix size × number of slices = 74×74×40, and a parallel imaging factor = 2. Image Analysis: DKI analysis included the estimation of mean diffusivity (MD), fractional anisotropy (FA), mean kurtosis (MK), axonal water fraction (AWF),5 and DKI-derived tractography6,7 and was performed with diffusional kurtosis estimator (https://www.nitrc.org/projects/dke/). DKI was incorporated into AFQ software (https://github.com/jyeatman/AFQ) using fully automated in-house scripts, including in-house algorithms to automatically segment the fornix. WM tracts included were left and right fornix, parahippocampal white matter (PWM), arcuate fasciculus (AF), inferior longitudinal fasciculus (ILF), uncinated fasciculus (UF), and cingulum bundle. Statistical Analysis: Along-the-tract profiles were created for each fiber group along 100 nodes and each tract was divided into 5 ROIs by averaging every 20 consecutive nodes. In all cases, tract node and ROI numbers increase closer to or further anteriorly within the temporal lobe. Statistical analysis was performed using repeated two-sample t-tests and significance was determined at p < 0.05 after correcting for multiple comparisons with false discovery rate (FDR) thresholding (n = 240 total comparisons). Effect size was calculated with Cohen’s d parameter.

Results

Tract profiles are summarized in Figures 1-2 and Tables 1-3. In general, MD tended to be higher and FA, MK, and AWF tended to be lower in subjects with TLE; MK and AWF showed more significant changes, larger effect sizes, and more expansive group-wise differences compared to MD and FA; and the abnormal tract profiles demonstrated a qualitative crescendo effect, increasing in significance into the temporal lobe. Significant reductions in MK and AWF were found in the left fornix and PHW in multiple ROIs in subjects relative to controls, consistent with clinical observations of mesial hippocampal sclerosis in TLE. MK and AWF also demonstrated significant changes in multiple ROIs in the left and right AF and ILF indicating the presence of extrahippocampal pathology.

Discussion and Conclusion

By combining the strengths of DKI and AFQ, we have implemented a novel diffusion MRI-based image analysis technique, which can quantify microstructural abnormalities in specific, functionally important WM fiber pathways. The along-the-tract diffusion profiles were demonstrated to identify group-wise pathological changes, with the largest effect sizes lateralizing to the left temporal lobe and extending along the tracts beyond the left temporal lobe and including the right side of the brain. By implementing a specific model of WM microstructure, some of the observed changes may be attributed to neuronal loss via reduction in AWF, which estimates the fraction of water that is confined inside WM axons. Previous DTI studies average mean MD and FA across whole tracts, which may be insensitive to regional tract alterations and overlook significant tissue pathology. This is the first study to combine improved diffusion characterization with DKI and along-the-tract profiles quantified by AFQ. This technique may provide further insights into structural pathology underlying TLE as well as other disorders affecting WM pathways.

Acknowledgements

This work was supported in part by National Institutes of Health research grant T32GM008716 (to P. Halushka) and by the Litwin Foundation (to J.A.H.).

References

1. Jensen JH, Helpern JA, Ramani A, Lu H, Kaczynski K (2005): Diffusional kurtosis imaging: the quantification of non-Gaussian water diffusion by means of magnetic resonance imaging. Magn Reson Med. 53:1432-40.

2. Jensen JH, Helpern JA (2010): MRI quantification of non-Gaussian water diffusion by kurtosis analysis. NMR Biomed. 23:698-710.

3. 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.

4. Proposal for revised classification of epilepsies and epileptic syndromes: Commission on Classification and Terminology of the International League Against Epilepsy. Epilepsia. 1989;30:389–99.

5. Fieremans E, Jensen JH, Helpern JA. White matter characterization with diffusional kurtosis imaging. Neuroimage 2011;58:177–188.

6. Jensen JH, Helpern JA, Tabesh A (2014): Leading non-Gaussian corrections for diffusion orientation distribution functions. NMR Biomed. 27:202-11.

7. Glenn GR, Helpern JA, Tabesh A, Jensen JH. Optimization of white matter fiber tractography with diffusional kurtosis imaging NMR in biomedicine, 2015, 28, 1245-1256.

Figures

Figure 1. Mean, group-wise along-the-tract tract profiles for left and right fornix and PWM and tract ROI histograms with error bars indicating standard error of the mean. Statistical tests indicate group-wise differences in patients vs controls with significance levels adjusted for multiple comparisons using FDR.

Figure 2. Mean, group-wise along-the-tract tract profiles for left and right AF and ILF and tract ROI histograms with error bars indicating standard error of the mean. Statistical tests indicate group-wise differences in patients vs controls with significance levels adjusted for multiple comparisons using FDR.

Table 1. Summary statistics for left and right sided Fornix and PWM ROIs. Parameter values represent group-wise mean (± sd) for each ROI and p-values are corrected for multiple comparisons using FDR with * indicating p < 0.05 and ** indicating p < 0.005.

Table 2. Summary statistics for left and right sided AF and ILF ROIs. Parameter values represent group-wise mean (± sd) for each ROI and p-values are corrected for multiple comparisons using FDR with * indicating p < 0.05 and ** indicating p < 0.005.

Table 3. Summary statistics for left and right sided CB and UF ROIs. Parameter values represent group-wise mean (± sd) for each ROI and p-values are corrected for multiple comparisons using FDR with * indicating p < 0.05 and ** indicating p < 0.005.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
3093