Cognitive dysfunction is a common non-motor symptom of Parkinson’s disease (PD). Cognitive decline in PD is likely associated with dysfunction in the cholinergic system, which is affected by synuclein pathology early in the disease course. Recent studies have shown an association between reduced integrity of the basal forebrain (BF), which provides cholinergic innervation to most of cortex, and diminished cognitive functioning in PD. Specifically, those with PD and reduced cholinergic innervation also have higher rates of cognitive impairment. However, no study has directly investigated the relationship between basal forebrain integrity and cortical cholinergic levels. In the present study, we examined this relationship through measures of basal forebrain microstructural integrity and cholinergic nerve terminal density in cortical and subcortical gray matter.

Participants included 92 non-demented individuals with idiopathic PD (M:F=64:28; Age=67.0±7.1 yrs) who underwent structural MRI, diffusion MRI, and [18F] fluoroethoxybenzovesamicol (FEOBV) cholinergic PET imaging. We used a basal forebrain and region of interest defined by AssemblyNet, which uses ensembles of pretrained convolutional neural networks to create a full brain segmentation. Bilateral putamen from this atlas was also included as a control region. We measured microstructural integrity using free water fraction (FWF), a diffusion MRI-derived metric of extracellular water that associates with cellular density and neuroinflammation. For PET data, we computed the distribution volume ratio (DVR) by regions as defined by FreeSurfer. A factor analysis of DVR in all 88 FreeSurfer ROIs resulted in seven clusters of ROIs covering 1) widespread bilateral cortical regions (PC1); 2) subcortical and limbic regions (PC2); 3) bilateral cingulate regions (PC3); 4) left frontal regions (PC4); 5) right frontal and temporal regions (PC5); 6) cerebellum (PC6); and 7) bilateral entorhinal cortex and left temporal cortex (PC7). We performed seven separate regression analyses per ROI (controlling for age and disease duration) to evaluate the association between BF FWF and cholinergic levels in these regions. To determine if these ROIs showed unique associations with BF FWF, we then entered ROIs with a significant association with BF FWF as independent variables in a stepwise regression with forward selection with BF FWF as the dependent variable.

BF FWF was significantly and negatively associated with cholinergic levels in PC1 (AR2=.042, ß=-0.208, p=.04), PC3 (AR2=.044, ß=-0.206, p=.03), PC4 (AR2=.056, ß=-0.239, p=.02), and PC7 (ß=-0.215, p=.04). BF FWF trended towards a negative association with cholinergic levels in PC5 (AR2=.045, ß=-0.168, p=.09) and PC6 (ß=-0.188, p=.09). Putamen FWF did not significantly associate with any of the ROIs. In the follow-up stepwise regression, only PC4 contributed significantly to the overall model (AR2=.061, ß=-0.261, p=.02).

Basal forebrain FWF was inversely related to cholinergic levels in regions that are directly innervated by the basal forebrain (e.g., cingulate cortex, left frontal cortex, and bilateral entorhinal cortex). Future research should directly investigate the relationship between basal forebrain integrity, cortical cholinergic levels, and cognition. Separating the basal forebrain into specific nuclei would also be beneficial, as different nuclei may have differing associations with specific hemispheric cholinergic pathways and cognition.