Childhood absence epilepsy (CAE) is a common neurological condition that is associated with loss of awareness and subtle motor features with characteristic spike and wave discharges on the electroencephalogram¹. Morphometric abnormalities have been reported in the thalamus, sub-genual grey matter, and frontal lobe white matter in human CAE subjects². Here we assess white matter connectivity (see below) and grey matter structure (using voxel based morphometry) in patients with refractory CAE.

We assessed white matter fibre density³ and morphology⁴ using a recently developed diffusion MRI analysis method called fixel-based analysis (FBA; the term fixel refers to a specific fibre population within a single voxel)⁵. The benefit of fixel-based analysis compared to more traditional voxel-based analysis methods is that it enables the identification of group differences in specific white matter pathways, including within regions containing multiple overlapping bundles. Furthermore, FBA gives a more complete picture of fibre pathway connectivity by providing information about both axon density and fibre bundle cross-sectional area.

We studied 10 patients with refractory CAE (clinical details in table 1: 3 male; mean age 21.2; age range [16,37]) and forty five age- and gender-matched healthy controls scanned (12 male; mean age 26.5; age range [16,41]). Most (9/10) of the patients had ongoing seizures while on two medications, fulfilling the ILAE criteria for refractory seizures⁶.

Diffusion-weighted images (DWI: 60 directions, b-value 3000 s/mm², and voxel size 2.5 mm isotropic) and 3D T1-weighted data (voxel size 0.9 mm isotropic ) were obtained on a Siemens 3T Trio.

Pre-processing of DWI data was performed as in reference 3. Fibre Orientation Distributions (FODs) were computed using Robust Constrained Spherical Deconvolution (rCSD)⁷ at lmax=8 using MRtrix3. A population-specific FOD template was generated from 12 patients and 12 healthy controls, and all FOD images were registered to that template⁸. During spatial normalization to the template, individual FODs were reoriented using apodised point spread function reorientation³. Three factors were quantified in all white matter fixels: Fibre Density (FD)³, Fibre Cross-sectional Area (FC)⁴ and combined Fibre Density and Cross-sectional area (FDC)³ (Figure 1).

Statistical analysis was done using the recently developed connectivity-based fixel enhancement (CFE)⁵ method. We assigned family-wise error (FWE) corrected p-values to each fixel using non-parametric permutation testing of the CFE enhanced t-statistics (5000 permutations). Significant fixels (FWE p<0.05) are displayed with colour-coding showing fixel orientation (red: R-L, blue: I-S, green: A-P).

For Voxel-Based Morphometry (VBM)⁹ we used Statistical Parametric Mapping (SPM8). A population template was created using Diffeomorphic Anatomical Registration using Exponentiated Lie algebra (DARTEL)¹⁰. Age, gender and total intracranial volume (TIV) were included as covariates-of-no-interest.

FBA showed abnormalities in the right cingulum bundle, left superior longitudinal fasciculus (reduced fibre density, FWE p<0.05; figure 2, column (a)) the corpus callosum, bilateral superior longitudinal fasciculus and superior cerebellar peduncle (reduced Fibre Density and Cross-sectional Area, FWE p<0.05; figure 2, column (c)).

Fibre cross-sectional area was reduced in the right superior longitudinal fasciculus and middle part of corpus callosum in the CAE group (FWE p<0.05; figure 2, column (b)).

VBM showed bilateral increased grey matter volume (GMV) in the anterior cingulate gyrus (FWE p<0.05; figure 3a, 3b; and p<0.001 uncorrected, figure 3c), with a trend for decreased thalamic grey matter, bilateral at uncorrected p<0.005 (figure 3d).

Increased GMV in the frontal lobe and reduction of structural fibre connectivity in associated fibre-tracts, in particular bilateral superior longitudinal fasciculus and right cingulum, describe a critical network associated with higher cognitive functions including consciousness, and extend previous work implicating these structures^1,11. By using the recently developed FBA approach, more extensive structural connectivity decrease were also observed in posterior regions such as the motor area of the corpus callosum and the cerebellar peduncle connecting the thalamus and cerebellum. Our results reinforce the concept that the midline frontal areas are critically involved in the phenotype of generalised spike and wave discharges.

The cerebellar peduncle involvement reflects the role of subcortical structures including thalamus in this form of epilepsy and involving motor associated fibres, most likely secondary to the cortical process.

Lesions in this area of midline frontal cortex may cause bilateral spike wave discharges in a process called secondary bilateral synchrony. The same area is involved without any known lesions in typical CAE which resolves in late adolescence. The imaging findings in the present study of non-lesional patients with intractable CAE suggest that this frontal region and its network connections are critical to epilepsy in patients with spike and wave discharges. This could be either developmental or as a consequence of seizures.