Juvenile neuronal ceroid lipofuscinosis (CLN3) is a neurodegenerative autosomal recessive lysosomal storage disease with a reported incidence of 0.2-7.0 per 100,000 births^1-2. It is among the most common progressive childhood encephalopathies, leading to death. First clinical symptoms appear around the age of four to seven. Symptoms include loss of vision, seizures, loss of cognitive and motor functions, ultimately followed by premature death at an age of 16 to 35.

Previous magnetic resonance imaging (MRI) studies have reported progressive hippocampal atrophy³, decreased gray matter volume in the dorsomedial part of the thalamus and decreased white matter volume in the corona radiata⁴. A diffusion MRI (dMRI) study reported increased apparent diffusion coefficient (ADC) in late infantile neuronal ceroid lipofuscinosis⁵.

DMRI can be used to noninvasively probe white matter microstructure and connectivity⁶. Recent advances, such as constrained spherical deconvolution (CSD)⁷, have enabled the reliable reconstruction of neural tracts through regions with complex (e.g. crossing) fiber configurations^8-10, present in the majority of white matter¹¹.

Here, we investigated global white matter microstructure in CLN3 using whole-brain CSD-based tractography^12-13 and white matter tract skeleton¹⁴. In addition, local microstructural abnormalities were investigated using tract-based spatial statistics (TBSS)¹⁴.

Material and preprocessing

We acquired dMRI and T1 data from 14 patients with CLN3 and 14 age-matched controls in 32 gradient orientations using b=1000 s/mm² and 2 mm isotropic voxel size with a Philips 3.0T machine. Patients were 9.6±3.4 years during the first acquisition and 11.4±3.2 years during the second acquisition (N=12). The age of control subjects was 11.2±2.3 years. The difference in age was not statistically significant.

Global microstructural analyses

Whole-brain probabilistic tractography was performed to reconstruct fiber tracts with CSD in ExploreDTI^7,12-13. Subject motion¹⁵, eddy current and echo-planar imaging induced distortions¹⁶ were corrected. In addition, fractional anisotropy skeleton was reconstructed as published in ¹⁴. Mean values for fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD), radial diffusivity (RD), and coefficient of planarity (CP)¹⁷ were calculated across the whole tractogram and skeleton. Statistical analyses were performed with general linear model by using age and gender as covariates.

Local microstructural analyses

Local microstructural analyses were performed with TBSS¹⁴ in FMRIB Software Library (FSL)¹⁸. Subject motion and eddy current induced distortions were corrected with FSL’s EDDY-tool¹⁹. In TBSS, mean FA skeleton of the whole sample is reconstructed, onto which individual subjects’ skeletons are then projected. Statistical analyses were performed using permutation tests and threshold-free cluster enhancement with FSL’s randomize tool²⁰.

Global microstructural analyses

We found significantly decreased FA and CP values, and significantly increased AD, RD, and MD values in subjects with CLN3. The results were similar for the first and second acquisition. The largest relative difference was in the FA values using the tractography approach (-15% difference) (p=0.000001). There were no significant differences between the two acquisitions in subjects with CLN3. The analyses between the two acquisitions were repeated without using age as a covariate, resulting in no significant differences. Results are presented in Table 1.

Local microstructural analyses

With TBSS, we found widespread voxel-wise decreases in FA as shown in Fig. 1, for example in corona radiata (P=0.006) and posterior thalamic radiation (P=0.002) (Fig. 2). In addition, MD (Fig. 3), AD, and RD were increased and CP was decreased in many regions.

Highly decreased FA values were found in subjects with CLN3 consistently using all approaches and the differences were distributed across the whole brain. In addition, diffusivity metrics were increased and CP was decreased in children with CLN3. The diffusivity results are in concordance with the previous dMRI study⁵.

We showed, by investigating CP¹⁷, that the decrease in FA was not caused by an increase in the complexity of fiber organization²¹. As CP was decreased in children with CLN3, its effect to FA would be to the opposite direction. Possible remaining causes for the FA difference include decreased fiber coherence²² and decreased myelination²³.

The skeleton-based approach²⁴ and CSD-based tractography^25-26 both have their limitations. Therefore, we performed analyses with both methods, differing in their limitations. However, a limitation is that the acquisition was suboptimal for CSD²⁷.

As there were no significant differences between the two acquisitions in children with CLN3, the microstructural white matter abnormalities may be present in early childhood or even in infancy.