Background: Direct in vivo measurement of α-synuclein pathology remains a major unmet need in multiple system atrophy (MSA), constraining biologically informed diagnosis, disease staging and therapeutic development.

Methods: Forty-five patients with clinically diagnosed MSA and 9 healthy controls underwent 18F-FD4 PET at Huashan Hospital, Fudan University. Images were analyzed visually and semi-quantitatively. Spatiotemporal progression patterns (via event-based modeling) were assessed. Eight patients completed one-year follow-up scans.

Results: Here, we evaluated the α-syn PET tracer 18F-FD4 in 45 patients with MSA. 18F-FD4 recapitulated the expected anatomical distribution of pathology and showed 97.8% concordance between tracer topography and clinical syndrome, supporting its ability to map disease-relevant α-syn burden in vivo. In around one-third of the patients, tracer-positive regions were detected before the emergence of the corresponding overt motor syndrome, suggesting that 18F-FD4 PET can reveal clinically silent pathology. Unsupervised machine-learning-based progression modeling further identified three trajectories of regional involvement—cerebellar-origin, striatal-origin and a previously unrecognized synchronous multi-regional pattern—thereby extending beyond the conventional MSA-C/MSA-P classification. Integration of 18F-FD4 PET with dopaminergic and metabolic imaging enabled a biologically grounded staging framework that refined patient stratification. Finally, longitudinal increases in regional 18F-FD4 uptake correlated with clinical worsening, supporting its potential as a progression biomarker.

Conclusion: Together, these findings provide proof of concept that 18F-FD4 PET can visualize pathology, model progression and track disease severity in vivo, positioning it as a promising biomarker for patient stratification and therapeutic monitoring in MSA clinical trials.