Focal epilepsies are the most common form of epilepsy, with seizures originating in one brain region. While some focal epilepsies are caused by structural brain lesions, many affected individuals have normal brain imaging using conventional MRI. Familial Focal Epilepsy with Variable Foci (FFEVF) is of particular interest because family members have seizures originating from different cortical regions¹. Recent evidence suggests that approximately 12% of families with Familial Focal Epilepsy contain mutations in the gene DEPDC5², which is now known to be critical for cell growth^3,4.

To investigate how white matter connectivity might be altered in DEPDC5 related epilepsy, we applied a fixel-based analysis (FBA)⁵ to diffusion weighted MRI data. The benefit of FBA is its ability to detect group differences in specific fibre populations within a voxel (i.e. a fixel), even in regions containing crossing fibres⁵, as well as to identify altered white matter connectivity as manifested by both fibre density⁶ and morphology⁷ differences.

We recruited 15 subjects that were positive for DEPDC5 mutations, including 10 cases with frontal lobe epilepsy (FLE), and 5 unaffected individuals (who did not have seizures). We compared the 10 DEPDC5+ve FLE subjects to matched controls (Table 1a). To further investigate if the unaffected DEPDC5+ve subjects had a different pattern of altered white matter disruption, we also compared the 5 unaffected subjects to controls (Table 1b).

To investigate if white matter differences between DEPDC5+ve and control subjects were unique to DEPDC5 or a consequence of FFEVF in general, we recruited 6 subjects with FLE that were DEPDC5 negative. All 6 patients were known to be positive for mutations in the gene KCNT1 . We compared the 6 KCNT1+ve FLE patients with controls (Table 1c).

DWIs were acquired on a Siemens 3T Trio (60 directions, b-value=3000 s/mm², voxel size 2.5 × 2.5 × 2.5 mm³). Pre-processing was performed as described in ⁶. FODs were computed by Robust Spherical Deconvolution⁸ using MRtrix (www.mrtrix.org). A population-specific FOD template was generated from a subset of data (12 patients and 12 controls) and all FOD images were registered to that template⁹.

To identify WM differences between patients and controls, three factors were quantified in all white matter fixels: Fibre Density (FD)⁶, Fibre Cross-section (FC)⁷ and also combined Fibre Density and Cross-section (FDC)⁶ (Fig. 1).

Statistical analysis was performed using connectivity-based fixel enhancement (CFE)⁵. We assigned family-wise error corrected p-values to each fixel using permutation testing of the CFE enhanced t-statistics (5000 permutations). Significant fixels (p < 0.05) were displayed using MRtrix3, where each fixel was colour-coded according to the fixel orientation (red: R-L, blue: I-S, green: A-P).

When comparing the 10 FLE DEPDC5+ve subjects to controls, we identified several pathways with a significant reduction in FD compared to controls (Fig. 2a). Affected fibre pathways include the genu and splenium of corpus callosum, cingulum, inferior frontal occipital fasciculus, and superior longitudinal fasciculi (SLF). As shown by Fig. 2b, there were very few significant decreases in FC. In the combined FDC analysis fewer pathways were significant, suggesting that combining FC with FD dilutes the statistical power of FC alone (Fig. 2c).

In the analysis of unaffected DEPDC5+ve subjects (Table 1b), no significant fixels were identified in the FD, FC or FDC (not shown). While this group contains only 5 subjects and therefore statistical power is limited, it nevertheless suggests that WM pathways are only affected in DEPDC5+ve individuals with Epilepsy.

Figure 3 shows fixels with reduced FD, FC and FDC in KCNT1+ve FLE subjects. Aside from the cingulum, alterations were observed in similar pathways to those observed in the DEPDC5+ve FLE analysis. The principal point of difference, however, is that the most extensive reductions are observed in the FC analysis (Fig.3 b).

The decreases in FD, FC and FDC can be interpreted as a reduction in the number of axons in the affected pathways, which presumably implies a reduction in connectivity. While there are similar pathways affected in both DEPDC5+ FLE and KCNT1+ FLE, the different manifestation of the WM changes (FD vs FC) suggests differences in brain development rather than a consequence of seizures.

This study demonstrates the benefit of using FBA to identify specific fibre pathway alterations even in regions with crossing fibres (e.g. SLF).