Open Access

Hannah BaumeisterORCiD, Philipp Wegner, Mónica Ferreira, Tamara Schaprian, Marcondes França, Thiago Junqueira Ribeiro Rezende, Alberto Rolim Muro Martinez, Hong Jiang, Zhao Chen, Liao Weihua, Marcus Grobe-Einsler, Berkan Koyak, Demet Önder, Bart van de Warrenburg, Judith van Gaalen, Alexandra Durr, Giulia Coarelli, Matthis SynofzikORCiD, Ludger Schöls, Paola Giunti, Hector Garcia-Moreno, Gülin ÖzORCiD, James Joers, Dagmar Timmann, Andreas ThiemeORCiD, Heike Jacobi, Jeroen J. de Vries, Peter Barker, Chiadikaobi Onyike, Eva-Maria Ratai, Jeremy Schmahmann, Kathrin ReetzORCiD, Jon Infante, Jeannette Huebener-Schmid, David KueglerORCiD, Thomas Klockgether, David Berron, Jennifer FaberORCiD, Falk Lüsebrink, Stefan Hetzer, Michael Ewers, Julian Hellmann-Regen, Eike Spruth, Daniel Janowitz, Ingo Kilimann, Marie Kronmüller, Annika Spottke, Oliver Peters, Josef Priller, Tony Stoecker

• Background Spinocerebellar ataxia type 3 (SCA3) is characterised by progressive brain atrophy, with regional volume loss detectable via MRI prior to clinical manifestation. We aimed to identify the previously unknown sequence of brain atrophy in SCA3 and evaluate whether this sequence can be translated into an atrophy staging framework to enable accurate clinical prognosis and trial enrichment. Methods We included data from 322 SCA3 mutation carriers, enrolled in observational studies conducted across Europe, the Americas, and Asia. Participants underwent follow-up assessments up to five years after baseline. The Subtype and Stage Inference machine learning algorithm was applied to estimate the most likely atrophy sequence(s) from baseline anatomical MRI. The Scale for the Assessment and Rating of Ataxia (SARA) was used to capture ataxia severity. Atrophy stages were analysed in relation to SARA and time from disease onset. Interventional trials were simulated to estimate required sample sizes under different atrophy stage eligibility criteria. Findings We identified a uniform sequence of brain atrophy in SCA3, characterised by earliest volumetric decline in the caudal brainstem and substantial involvement of the white matter. Atrophy stage was associated with both SARA and time from disease onset. Atrophy staging outperformed single-region volumetrics in predicting SARA over time. Applying atrophy stage cut-offs substantially reduced the sample sizes needed to adequately power hypothetical clinical trials. Interpretation These findings yield mechanistic insights into the progression of neurodegeneration in SCA3 and possess immediate translational relevance, facilitating patient stratification and sample enrichment for interventional trials. Funding National Ataxia Foundation (NAF).