In this cross-sectional cohort study, we provided a comprehensive description of the clinical and structural characteristics of SCA3. Our results demonstrated that the spatiotemporal decline in the SCA3 brain began in the cerebellum, progressed to the spinal cord, brainstem, and extensive subcortex. Correlation analysis revealed that the expanded CAG repeat size was associated with volume loss in regional cerebellum and mean upper cervical cord area (MUCCA). Notably, our mediation analysis revealed that the association between CAG repeat size and ICARS was significantly mediated by MUCCA measures. In addition, our VBM and ROI approaches allowed the availability of individual single-point measures of these regional volumes, which can serve as valuable outcome biomarkers and derive sample size estimates for future interventional trials.
Previous postmortem and MRI studies have reported severe and widespread brain pathology in SCA3, involving neuronal loss in the cerebellum (both cerebellar lobules and WM), spinal cord, brainstem, thalamus, basal ganglia, and cerebral cortex [7, 17–21]. We here used VBM- and ROI-based approaches to evaluate brain and spinal cord volume loss in patients with SCA3. Our results closely reflected the neurodegeneration pattern observed in autopsy and smaller-sample MRI investigations. Additionally, a neuropathological study of 12 genetically confirmed autopsy cases of SCA3 revealed an interesting pattern of neuronal loss, with less prominence in the cerebellar cortex but more pronounced in the dentate nucleus [21]. An MRI study involving 38 SCA3 patients reported a more pronounced WM volume loss in the cerebellum and brainstem compared to GM structures [22]. Our findings in a prospective study, which involved 92 SCA3 patients, observed significant volume loss primarily affecting WM in the cerebellum, brainstem, and thalami. This consistency across different methodologies underscores the robustness of these observations and points towards a distinctive pattern of neurodegeneration in SCA3. The differential vulnerability of WM versus GM structures suggests that therapeutic strategies aimed at preserving or restoring WM integrity may hold promise in mitigating disease progression in SCA3.
Previous MRI studies have reported that structural damage preceded the onset of clinical manifestations in SCA3 [8, 9], as well as in other poly(Q) diseases, including Huntington’s disease and familial amyotrophic lateral sclerosis [23, 24]. In presymptomatic SCA3 patients, volume loss has been identified in the spinal cord, cerebellar WM, and pons [8, 9, 25, 26]. Our findings align with these observations, indicating cerebellar volume loss in the presymptomatic stage. Moreover, these studies and ours all implicated that structural damage of SCA3 followed a specific temporal-dependent process. As the disease duration increased, volume loss extended from the cerebellum to the brainstem and spinal cord, and to specific regions of subcortical white matter. Of noteworthy was a discrepant observation in the cerebral cortex, which may be explained by differences in the potential effect of geographic and ethnic variability. While these findings contribute to the understanding of stage-dependent biomarkers reflecting the cascade of clinical and pathological events, it is crucial to acknowledge that validation with pathological data is necessary for a comprehensive interpretation.
Beyond SCA3, pathologic and imaging assessments assert that both Alzheimer's disease and Parkinson’s disease evolve along different stages (Braak’s staging) [27–30]. A recent study examining Alzheimer's disease clinical spectrum reported that microglial activation plays a key role in linking the effects of amyloid beta to tau pathology spread across Braak’s stages [30]. Postmortem studies on Parkinson’s disease brains have also identified a close association between microglial activation and Braak’s staging [31]. Consistent with these findings, previous studies on patients with SCA3 reported an increase in activated microglia in the brain [32, 33]. In addition, Bonanomi et al. reported in a yeast model of SCA3 that expanded ataxin-3 toxicity was mediated by prion protein [34]. Collectively, these studies and ours suggest the possibility that an interaction between expanded ataxin-3 and activated microglia leads to the trans-synaptic spread of expanded ataxin-3 and a cascade of events culminating in the death of nearby neurons.
Pathological studies showed that larger poly(Q) stretches in the corresponding ataxin-3 protein enhance its aggregation tendency and toxicity to neurons [35]. However, previous imaging studies with small-sample SCA3 patients showed conflicting results regarding the associations between expanded CAG repeat size and brain structure changes [36, 37]. A potential association between the CAG repeat size and hypothalamic volume was observed in one report [36]. In contrast, another study reported the CAG repeat size had no significant influences on the degeneration rate of cerebellar function and structure based on N-acetylaspartate/creatine ratios and 3-dimensional fractal dimension values [37]. In the present study, we recruited a much larger cohort and our MRI data enabled VBM and ROI analyses, revealing that a larger CAG repeat size could accelerate volume loss in the right cerebellar lobule IV and cervical spinal cord (as analyzed by MUCCA). The cerebellum and spinal cord are usually considered vulnerable tissues in patients with SCA3, and these findings suggest that the right cerebellar lobule IV and cervical spinal cord could be more sensitive to mutant ataxin-3 toxicity in SCA3.
Several studies have indicated that the expanded CAG repeat size in ATXN3 was a distinct determinant of disease progression (as analyzed by ICARS) in SCA3 [6, 38, 39]. However, the underlying mechanism by which expanded CAG repeat size influences ICARS progression in SCA3 is unclear. The present study not only confirmed the negative association between expanded CAG repeat size and ICARS in SCA3 patients but provided further evidence that this association was partly mediated by MUCCA measures. This mediation effect suggests that cervical spinal cord atrophy may be related to mutant ataxin-3 toxicity per se, the predominant pathology caused by CAG expansion. Pathological data have showed that the number of lower motor neurons in the spinal cord was inversely associated with the size of the expanded CAG repeats [40]. Additionally, several studies reported the associations between cervical cord atrophy and clinical parameters, including disease duration and severity [8, 20]. Taken together, these findings indicate that the detrimental effect of expanded CAG repeat size on ICARS in SCA3 is partly associated with volume loss in cervical spinal cord.
As targeted therapies for SCA3 are in development, with the initiation of the first safety trials involving antisense oligonucleotides (NCT05160558, NCT05822908). Future, preventive trials are a realistic option and a comprehensive understanding of biomarkers that reflect the cascade of pathological events associated with SCA3 is a crucial prerequisite. Here, we found that both right cerebellar lobule IV volume and MUCCA were associated with disease duration or ICARS progression in SCA3 patients. This suggests that quantitative analyses of the right cerebellar lobule IV and spinal cord might be valuable disease progressive biomarkers in SCA3. Furthermore, by setting the median disease duration of 7 years as a cut-off value to monitor disease progression under the assumption of a given treatment, we assessed potential biomarkers including total ICARS, right cerebellar lobule IV volume, and MUCCA for sample size estimation. The results suggested that MUCCA required the minimum participants, followed by total ICARS, and the right cerebellar lobule IV volume requiring highest sample size. Therefore, we propose MUCCA as potential candidate for clinical trial endpoints. Such estimations are crucial for therapeutic trials against rare disease such as SCA3, where patient recruit and follow-up are challenging, making large sample sizes impractical.
Several limitations should be noted in our study. First, our findings were based on cross-sectional data, which do not allow an accurate estimation of the time-dependent evolution of brain and spinal cord degeneration. Second, our study focused solely on measuring cervical spinal cord cross-sectional area. Future investigations could enhance the understanding of the association between expanded CAG repeat size, spinal cord degeneration, and ICARS by examining the entire length of the spinal cord. Third, the reliability of these biomarkers in reflecting disease progression needs confirmation through longitudinal studies. Also, since our study was conducted in a single center, the results should be replicated in other studies for broader validation. Finally, given that our study was based on neuroimaging of structural MRI, it would be ideal if future investigations can seek confirmation based on postmortem histopathology.