Huntington’s disease is a monogenetic, neurodegenerative movement disorder, considered to be a model disorder in that it is amenable to predictive genetic testing, thereby enabling the natural history of neurodegeneration to be studied using non-invasive neuroimaging techniques. Here, we investigated whether longitudinal diffusion-weighted imaging of the basal ganglia could be used to detect early microstructural changes in participants with presymptomatic Huntington’s disease (preHD).

Diffusion-weighted imaging scans and behavioural measures were obtained twice one year apart on 35 participants with preHD (>38 mutant CAG repeats) and 19 age- and gender-matched healthy controls¹.

The basal ganglia, comprising here of the striatum and pallidum identified at a single-subject level using FIRST² in FSL³, served as a priori regions of interest (ROIs) and were subdivided in 2 anterior and posterior subregions following a plane including the anterior commissure. This gross subdivision was used to follow the established differences in connectivity profile for each of these regions, as the anterior basal ganglia are connected to prefrontal areas associated with cognitive/limbic processing, whereas the posterior basal ganglia are connected with sensorimotor brain regions⁴.

Diffusion tensor indices (FA and MD) averaged in these two subregions were compared between participants with preHD and healthy controls. We further carried out the same analyses after dividing the preHD group into 3 preHD subgroups according to the years left to the predicted onset of their disease according to the formula in Aylward et al.⁵: far from onset (FAR, N = 11), midway to onset (MID, N = 11), and close to onset of symptoms (CLOSE, N = 13). All results were corrected for multiple comparisons.

When comparing all of the participants with preHD and healthy controls, no significant longitudinal change was found in FA or MD. Independent samples t-test revealed however significantly higher FA and MD in the anterior and posterior basal ganglia of the preHD participants compared with healthy controls both at baseline and at the follow-up visit separately (Fig. 1).

A further analysis, after splitting the preHD group into the 3 subgroups, revealed a significant F effect on the change of MD over time in the posterior basal ganglia across the 4 groups. Post-hoc t-tests contrasting each pair of groups showed a significant change between CLOSE and FAR subgroups in the posterior basal ganglia (Fig. 2). Interestingly, it seemed that this was mainly driven by an increase of MD in the CLOSE group, compared with a trend towards decrease of MD in the FAR group.

We found no significant change in FA over time across the 4 groups (Fig. 2).

To the best of our knowledge, these are the first results showing significant longitudinal changes in the microstructure of the basal ganglia within a preclinical HD population. Diffusion-weighted imaging therefore has the potential to be used as a longitudinal biomarker in preHD. It is necessary however to divide preHD participants into meaningful subgroups according to their predicted age at onset to detect any effect. Without such a subdivision, longitudinal changes can be averaged out due to very distinct patterns of changes in FA, and particularly MD, between the subgroups.

Of note, a k-means clustering analysis (k=2), based on the changes in clinical and behavioural measures, did not reveal a clear distinction between healthy participants and the age-of-onset subgroups, or within the preHD subgroups (Fig. 3). Diffusion imaging might therefore provide additional diagnostic information when compared with clinical testing. In particular, atrophy patterns in the basal ganglia are monotonic over time and across groups with increased disease severity¹. On the contrary, FA changes, while not significant across groups, seem to plateau in CLOSE after an increase in FAR, a less strong increase in MID, in line with findings comparing preHD and HD participants⁶. Similarly, changes in MD are highly non-monotonic, with a strong increase of MD in CLOSE, but a trend towards decrease in FAR. This decrease of MD, which might seem counter-intuitive at first, might hint at an initial, early stage neuroinflammation process followed by neurodegeneration leading to an increase of MD. This result is similar to what has been observed in preclinical and probable Alzheimer's disease⁶.

In conclusion, while FA and MD might be less sensitive to longitudinal changes than volumetric measures¹, they provide mechanistic insights into the underlying physiopathological process that are complementary to the monotonic, non-specific changes in the volume of the basal ganglia.