DAT-SPECT-based subtype and stage inference in Parkinson’s disease

Abstract Heterogeneity in the pattern of nigrostriatal degeneration can complicate the prognosis and treatment of Parkinson’s disease. We analysed dopamine transporter single-photon emission computed tomography with 123 I-ioflupane (DAT-SPECT) using atlas-based 12-region segmentation in 636 drug-naive patients with sporadic Parkinson’s disease and 126 healthy controls. Subtype and Stage Inference (SuStaIn) algorithm was applied to multi-regional DAT-SPECT data at baseline, 1, 2, and 4 years. SuStaIn identified three reproducible subtypes of nigrostriatal dopaminergic degeneration: (S1) left posterior putamen, (S2) right posterior putamen, and (S3) bilateral caudate, with high longitudinal stability. At baseline, S3 patients were older, exhibited greater cognitive and psychiatric burden, and demonstrated lower cerebrospinal fluid α-synuclein seeding positivity. Longitudinally, motor symptoms improved more in S1 and S2 than in S3 upon initiation of treatment, whereas depression and anxiety worsened preferentially in S2 and impulsive–compulsive behaviours increased in S3. DAT-SPECT-based SuStaIn thus reconstructs biologically plausible dopaminergic progression patterns and yields clinically meaningful subtypes with short-term prognostic relevance.