Reduced volume of neuromodulatory brainstem nuclei is associated with early hallucination-related vulnerability in Parkinson's disease.

Abstract

Early hallucination-related phenomena, including minor visual hallucinations (mVHs), are clinically relevant features of Parkinson's disease (PD), but their neuroanatomical substrates remain unclear. Brainstem nuclei supporting sensory gating, together with cholinergic and noradrenergic dysregulation, are proposed modulators of hallucinations. The objective of the present study was to test whether PD participants with early hallucination-related phenomena show reduced volumes of brainstem nuclei with cholinergic and noradrenergic relevance, using cross-sectional analysis of baseline structural MRI, clinical, and neuropsychological data from 641 PD participants in the Parkinson's Progression Markers Initiative (PPMI). Participants scoring ≥1 at screening, baseline, or during the first nine months of follow-up were categorized as PD-mVHs⁺, whereas those who consistently scored 0 throughout the first year were categorized as PD-mVHs^-. Histology-informed probabilistic atlases were used to segment brainstem, subcortical, and cortical regions. Analyses of covariance (ANCOVAs) compared regional volumes across groups, with group and sex as fixed factors, the group × sex interaction, and age, education, standardized estimated intracranial volume (eTIV), Montreal Cognitive Assessment (MoCA) score, and disease duration as continuous covariates. Hierarchical logistic regression tested whether brainstem volumes added explanatory value for hallucination-related status beyond clinical, neuropsychological, and other volumetric predictors. PD-mVHs⁺ participants showed reduced left locus coeruleus (LC) and left laterodorsal tegmental nucleus (LDTg) volumes, with no significant differences in pedunculopontine tegmental nucleus or dorsal raphe nuclei. After covariate adjustment, the left LDTg effect survived both Bonferroni and false discovery rate (FDR) correction, whereas the left LC effect survived FDR correction but not the Bonferroni-adjusted threshold. Adding LC and LDTg volumes to the hierarchical logistic regression improved fit beyond disease duration, depressive symptoms, immediate verbal recall, and semantic fluency. These findings suggest that reduced MRI-derived volume of the left LDTg, and possibly of the left LC, is associated with early hallucination-related vulnerability in PD.