Posterior cortical gray matter heterogeneity between hyposmia- and RBD-defined prodromal phenotypes in Parkinson's disease: a PPMI voxel-based morphometry study

Abstract Background Prodromal Parkinson's disease (PD) includes clinically distinct enrollment phenotypes. We examined whole-brain gray matter volume (GMV) in Parkinson's Progression Markers Initiative (PPMI) participants recruited through hyposmia-only or REM sleep behavior disorder (RBD)-only pathways and assessed the stability of the findings. Methods Baseline T1-weighted MRI from 465 hyposmia-only and 149 RBD-only participants (N = 614) was analyzed with CAT12/SPM25 voxel-based morphometry (VBM). Models included age, sex, intracranial volume, and scanner family. Whole-brain clusters were defined at voxel p < 0.001 and tested using family-wise error (FWE) correction. We examined scanner-restricted samples, 123 sex/scanner-matched pairs, and 10,000-sign-flip permutation analysis of linear models (PALM). Bilateral cuneus/calcarine estimates and clinical assessments were exploratory. Results Whole-brain analysis identified a posterior phenotype effect (F cluster: k = 4,499; pFWE = 4.15×10⁻⁷), with higher GMV in hyposmia-only participants. In 123 matched pairs, a posterior H > RBD T cluster survived correction (pFWE = 0.0266), whereas the F contrast did not. PALM also identified a posterior effect (two-sided threshold-free cluster enhancement [TFCE] FWE p = 0.0027; voxelwise FWE p = 0.0277). Adjusted bilateral cuneus differences were + 0.211 mL (95% CI 0.095–0.327) in the full cohort and + 0.187 mL (0.041–0.333) after matching. MRI-proximal RBD screening questionnaire scores showed no significant association with regional GMV. Conclusions Posterior cortical GMV differed between these enrollment-defined prodromal phenotypes. The anatomical finding was supported by nonparametric inference, although the matched F test and strictest scanner restriction were nonsignificant. Longitudinal studies are needed to establish its clinical relevance.