BACKGROUND: Parkinson's disease with mild cognitive impairment (PD-MCI) shows notable heterogeneity in cognitive presentation and progression. Current diagnostic frameworks assume biological homogeneity. Identifying biologically defined subtypes using neuroimaging-based individual biomarkers, cognitive profiles, and molecular signatures can clarify disease heterogeneity and advance precision medicine approaches.
METHODS: The study recruited 104 drug-naïve patients with PD-MCI and 107 healthy controls. T1-weighted MRI, diffusion tensor imaging, comprehensive neuropsychological assessments, and transcriptomic data were obtained. Quantile regression normative modeling was applied to create age- and sex-adjusted curves of gray matter volume maturation. Individual deviations were calculated to identify subtypes using the K-means clustering method. Differences in gray matter volume deviations, free water alterations, cognitive performance, transcriptomic associations, and network connectivity patterns were compared across subtypes. External validation was performed in an independent PPMI cohort (n = 115).
RESULTS: Two distinct subtypes emerged. Subtype 1 (n = 63) features localized supra-normal gray matter volume deviations in dorsolateral prefrontal regions, elevated free water indicating microstructural disruption, executive dysfunction, and transcriptional enrichment in metabolic dysfunction and neurodegenerative pathways. Subtype 2 (n = 41) displays widespread infra-normal gray matter volume deviations across cortical-limbic areas. Microstructural integrity is preserved despite volumetric reduction, reflecting successful adaptive compensation. Cognitive network connectivity is enhanced (density: 0.95 vs 0.90). Molecular signatures occur in cellular organization and signal transduction pathways. Subtypes show distinct vulnerability patterns in frontal-executive versus limbic-memory networks. External validation in an independent PPMI cohort (n = 115) confirms the two-subtype structure, with consistent reproduction of the fundamental degenerative-adaptive dichotomy despite some regional variability in the validation cohort.
CONCLUSIONS: Study findings provide evidence that PD-MCI encompasses biologically distinct subtypes. These subtypes have divergent neuroanatomical patterns, microstructural properties, cognitive profiles, and molecular mechanisms. Multimodal approach reveals fundamental disease heterogeneity and establishes a framework for precision medicine in Parkinson's disease cognitive impairment. Sex-specific distribution across subtypes indicates biological sex should be considered an integral component of PD-MCI heterogeneity rather than a confounding factor. This distribution has important implications for developing sex-specific therapeutic strategies.
