J Magn Reson Imaging. 2026 Jul 20. doi: 10.1002/jmri.70459. Online ahead of print.
ABSTRACT
BACKGROUND: Progressive supranuclear palsy (PSP) is both an underdiagnosed and a frequently misdiagnosed disorder. To remedy these diagnostic limitations, MRI has been used to investigate brain morphology to differentiate PSP from Parkinson’s disease (PD) and multiple system atrophy (MSA). However, while nigrostriatal degeneration and tau aggregation are prominent in PSP, and result in iron accumulation and atrophy in the region, substantia nigra (SN) atrophy remains an underexplored diagnostic marker.
PURPOSE: To investigate the diagnostic utility of QSM-derived SN parameters for PSP.
STUDY TYPE: Retrospective.
POPULATION: 123 (59 Males/64 Females) PD patients, 48 (26 M/22 F) MSA patients, and 22 (11 M/11 F) PSP patients were included in the main dataset. MSA patients include 20 parkinsonian type (MSA-P) (11 M/9 F) and 18 cerebellar type (MSA-C) (9 M/9 F) of MSA with 10 undetermined subtype. The external validation set included 12 (6 M/6 F) healthy controls, 13 PD (7 M/6 F), and 10 PSP (6 M/4 F) patients.
FIELD STRENGTH AND SEQUENCE: 3 T, MPRAGE T1-weighted imaging and multi-echo gradient echo imaging (mGRE) for QSM.
ASSESSMENT: Group level differences of SN volume, magnetic susceptibility, and susceptibility-to-volume ratio (SVR) among PSP, PD, and MSA groups, and these metrics’ differentiating power are measured. Bivariate logistic regression with T1-based morphological markers alongside SN metrics is performed.
STATISTICAL TESTS: Mann-Whitney U test for group comparisons. Receiver operating characteristic analysis with bootstrapping. p < 0.05 after Bonferroni correction is defined as statistically significant result.
RESULTS: The SN SVR was significantly higher in the PSP group compared to the MSA group and the MSA-P group. Moreover, using pons measurements with SN SVR in a bivariate model resulted in the highest differentiation between the PSP and MSA groups (AUC = 0.88, 95% CI: [0.78-0.95]), particularly driven by the increased differentiation between the PSP and MSA-P groups (AUC = 0.88, 95% CI: [0.75-0.98]). External validation supported the generalizability of SN SVR, yielding 100% sensitivity and 80% specificity for differentiating PSP from HC and PD.
DATA CONCLUSION: QSM-based SN morphometry can complement T1-weighted imaging for Parkinsonism assessment.
EVIDENCE LEVEL: 3.
TECHNICAL EFFICACY: Stage 2.
PMID:42475080 | DOI:10.1002/jmri.70459
