Cholinergic Denervation Patterns Across Cognitive Domains in Parkinson’s Disease

Participants

A total of 86 patients with PD (67 men and 19 women) were included in this cross-sectional study. Patients had a mean ± SD age of 67.8 ± 7.6 years and motor disease duration of 5.8 ± 4.6 years. Inclusion criteria consisted of a clinical PD diagnosis in accordance with the UK PD Society Brain Bank clinical diagnostic criteria.²³ Exclusion criteria included evidence of large-vessel stroke or mass lesions on anatomic imaging, the use of anticholinergic or cholinesterase inhibitor drugs, presence of severe depression as measured using the Geriatric Depression Scale,²⁴ and presence of PD dementia. A healthy control (HC) group consisting of 5 men and 7 women with a mean age of 67.8 ± 7.8 years was included for normative PET imaging data. This study was approved by the institutional review boards of the University of Michigan School of Medicine and Veterans Affairs Ann Arbor Healthcare System. Written informed consent was obtained from all participants prior to any study procedures and conducted in accordance with the Declaration of Helsinki.

All patients with PD underwent motor examinations using the Movement Disorder Society Revised Unified PD Rating Scale Part III (MDS-UPDRS-III), with a mean score of 34.3 ± 12.1 and a Hoehn and Yahr score of 2.4 ± 0.6. Motor assessment was performed in the morning in the dopaminergic off state, that is, after overnight withdrawal of dopaminergic medication. Mean levodopa equivalent dose (LED)²⁵ was 647 ± 410 mg. More details of the clinical and demographic characteristics are described in Table 1.

Neuropsychological Assessment

All patients with PD underwent a detailed neuropsychological assessment including at least 2 neuropsychological tests for each cognitive domain, in line with recommendations of the Movement Disorder Society (MDS) Task Force.^26,27 The cognitive test battery consisted of the California Verbal Learning Test; Wechsler Memory Scale; Stroop Color Word Test; Delis-Kaplan Executive Function System Trail Making Test; the Wechsler Adult Intelligence Scale Digit Span, Matrix Reasoning Task, and Digit-Symbol Modalities Tests; the Letter and Semantic Verbal Fluency; Boston Naming Test; Benton Judgment of Line Orientation Test; and the Clock Copy Test of the PD–Cognitive Rating Scale. Conditions possibly influencing neuropsychological test performance, including visual problems, color blindness, dysarthria, and dyskinesia were taken into consideration when interpreting cognitive performance, excluding (sub)tasks if needed. Patients were considered PD-MCI based on MDS PD-MCI level II criteria.²⁶ Neuropsychological tests and subtasks represented specific cognitive domains as shown in Table 2.

A z score for every participant on specific tests was calculated based on a data set of a healthy control group of 77 older participants. Data from these controls were collected in our laboratory to provide normative data of a group with similar geographical and social–economic backgrounds and ensure the use of identical assessment techniques. The healthy control group was of similar age, gender, and educational-level distributions as the patient population. The z scores of the PD–Cognitive Rating Scale Clock Copy Test were based on the normative data described previously, correcting for age, gender, and educational level.²⁸ By averaging all z scores on the tests or subtasks, an average z score for each cognitive domain was obtained (see Table 1). A global cognitive z score was computed as the average of all cognitive domains. A higher z score reflects better cognitive task performance. Neuropsychological testing of patients was performed while they were on their usual dopaminergic medications.

Imaging Acquisition and Analysis

All participants underwent brain magnetic resonance imaging (MRI) and VAChT [¹⁸F]FEOBV PET imaging. T1-weighted MRI was performed on a 3 Tesla Philips Achieva system (Philips, Best, The Netherlands). A 3-dimensional inversion recovery-prepared turbo field echo was performed in the sagittal plane using repetition time/echo time/inversion time = 9.8/4.6/1041 milliseconds, turbo factor = 200, single average, field of view = 240 × 200 × 160 mm, and acquired matrix = 240 × 200 × 160 slices and reconstructed to 1 mm isotropic resolution.

[¹⁸F]FEOBV was prepared as described previously.²⁹ [¹⁸F]FEOBV delayed dynamic imaging was performed over 30 minutes (in six 5-minute frames) starting 3 hours after an intravenous bolus dose injection of 8 millicurie. [¹⁸F]FEOBV²⁰ PET imaging was performed in 3-dimensional imaging mode using an ECAT Exact HR+ tomograph (Siemens Molecular Imaging, Inc., Knoxville, TN), which acquires 63 transaxial slices (slice thickness, 2.4 mm; intrinsic in-plane resolution, 4.1 mm full width at half maximum over a 15.2 cm axial field of view). [¹⁸F]FEOBV PET imaging was performed while patients were on their usual dopaminergic medication.

The PET imaging frames were spatially coregistered within subjects with a rigid-body transformation to reduce the effects of subject motion during the imaging session.³⁰ Statistical parametric mapping software (SPM12; Wellcome Trust Centre for Neuroimaging, London, UK) was used for PET-MRI registration using the cropped T1-weighted magnetic resonance volumetric scan. Freesurfer software (Laboratory for Computational Neuroimaging, Athinoula A. Martinos Center for Biomedical Imaging, Boston, MA) was used to define cortical and subcortical magnetic resonance gray matter volume of interest (VOI). MRI-based partial volume correction of the PET data was performed.³¹

Cortical VOI labels from the Mindboggle-101 data set segmented in FreeSurfer were used to identify gray and white matter VOI.³² A white matter reference tissue approach was used to determine VAChT binding as previously reported.^21,33 Distribution volume ratios were calculated from the ratio of averaged frames for gray matter targets and supratentorial white matter reference tissue.²¹ A single total neocortical VOI was created for the VOI-based statistical analysis.

Voxel-Based PET Analysis

Voxel-based PET analysis was performed as previously described.³⁴ All brain images were spatially normalized to Montreal Neurological Institute template space using the diffeomorphic anatomical registration through exponentiated lie algebra (DARTEL) normalization protocol³⁵ and smoothed with a Gaussian kernel of 8 mm full width half maximum to adjust the anatomical variability between the individual brains and to enhance the signal-to-noise ratio. The relevant brain areas were displayed in Montreal Neurological Institute atlas coordinates (in millimeters) in the stereotactic space using the automated anatomical labeling toolbox.

Statistics

Analyses were performed using IBM SPSS Statistics for Windows, version 24.0 (IBM Corp., Armonk, NY). Correlations between VOI-based whole-brain neocortical VAChT binding and performance on specific cognitive domain z scores were analyzed using a partial Pearson correlation coefficient controlling for MDS-UPDRS-III scores and LED levels. The Holm-Bonferroni method was used to correct for effects of multiple testing.

Voxel-wise statistical analysis was performed using SPM12 software using the parametric [¹⁸F]FEOBV distribution volume ratio images of all patients to assess both positive and negative topographic correlations between specific cognitive domain z scores and whole-brain cholinergic binding. In addition, a voxel-wise group comparison between the PD and the HC group was performed. The result of this analysis, a topographic profile showing regions significantly different between PD and HC, was transformed into a mask. The domain-specific voxel-wise analyses were then repeated using this mask to identify PD-specific versus aging-related regions. Both domain-specific voxel-based analyses were controlled for MDS-UPDRS-III scores and LED levels. The false discovery rate (FDR) approach was used for correction for multiple testing effects in the voxel-based analysis.

Cognitive Functioning

Based on MDS PD-MCI level II criteria, 39 (45.3%) participants in the PD group were classified as PD-MCI (Table 1). Only 3 of the 39 patients with PD-MCI presented with single-domain PD-MCI; all other patients with PD-MCI showed multidomain impairments. Attention was the most commonly affected domain (n = 30), followed by executive functions (n = 28) and memory (n = 27) in the 39 patients with PD-MCI. The language and visual domains were affected in 18 and 14 of the 39 patients, respectively, with PD-MCI.

Domain-Specific Cognitive Correlates of Global Neocortical [¹⁸F]FEOBV Distribution Volume Ratios

Significant correlations were present between VOI-based global cortical VAChT binding and the memory (r = 0.423, P < 0.001), executive function (r = 0.352, P < 0.001), and attention (r = 0.321, P = 0.003) domains (Table 3).

Voxel-Based Regional Cerebral [¹⁸F]FEOBV Binding Correlates of Different Cognitive Domains

Whole-brain voxel-based analyses were performed to explore the correlation between regional brain VAChT binding and the different cognitive domains, controlling for LED levels and parkinsonian motor impairment as measured by the MDS-UPDRS-III. Only positive correlations were found across domains with no negative correlations observed. Positive correlations indicate that a higher cholinergic binding is associated with better cognitive performance.

For the memory domain z score, regional cerebral cholinergic correlations (P < 0.01 FDR corrected) were seen in widespread cortical and subcortical brain regions (Fig. 1A). The topographic profile included the cingulate cortex (anterior, mid, posterior, and retrosplenial regions), prefrontal cortex (dorsolateral prefrontal cortex, orbitofrontal cortex, and gyrus rectus regions), insula and operculum, thalamus and visual thalamus with the pulvinar and lateral geniculate nucleus (LGN), caudate nucleus, hippocampus and parahippocampal regions, temporal lobe (superior, middle, and inferior temporal gyrus and temporal pole) bilaterally, and left lingual gyrus.

Results found for the executive function (Fig. 1B) and attention domains (Fig. 1C) showed partially overlapping topography with the memory domain. Overlapping regions across all 3 domains included the cingulate cortex, insula/operculum and visual thalamus (especially the LGN), and hippocampal region (P < 0.01 FDR corrected). Compared with the memory domain, involvement of the temporal lobe and prefrontal cortex was more limited for executive functions. The spatial extent of overlap was less for the attention domain but included the cingulate cortex, insula/operculum, temporal pole, right LGN, and LGN–fimbria transitional area (P < 0.01 FDR corrected).

The topographic profile for the language domain showed limited regional VAChT binding correlates, mainly including the right LGN, hippocampal fimbria, and LGN–fimbria transitional region (P < 0.01 FDR corrected, Fig. 1D).

Whole-brain voxel-based analysis for the visuospatial domain did not show significant voxels after correction for multiple comparisons.

Sensitivity Analysis of Voxel-Based Regional Cerebral [¹⁸F]FEOBV Binding Correlates of Different Cognitive Domains Superimposed on PD-Related Versus HC-Related Cholinergic Innervation Changes

Whole-brain voxel-based group comparisons between the PD and the HC group were first performed to identify disease-specific VAChT binding differences (P < 0.05 FDR corrected; Fig. S1). Widespread predominant posterior cortical and subcortical differences were found. Subcortical regions included the thalamus and pallidum. Analyses in the opposite direction showed no significant higher cholinergic binding in the PD group.

The regional topography was then used as a mask to repeat the voxel-based cognitive domain-specific analyses. After correcting for multiple comparisons, the memory and executive function domains showed significant regional correlates with overlapping topography across both domains (Fig. 2; P < 0.05 FDR corrected). The significant regions for the memory domain (Fig. 2A) were less extensive but topographically comparable with the correlates in the non-PD-specific analysis (Fig. 1). The most prominent regions included the superior and medial temporal lobe, hippocampus and parahippocampal region, thalamus (including the left visual thalamus), insula, operculum, superior frontal region, postcentral gyrus, the anterior and posterior cingulum, and the precuneus. The cholinergic correlates of the executive function domain (Fig. 2B) show overlapping regions with the memory domain, including the temporal lobe, parahippocampal regions and left hippocampus, thalamus and left visual thalamus, insula and operculum, cingulum, and postcentral gyrus. The significant regions for the attention domain (Fig. 2C) included the insula, operculum, (para)hippocampal region, and the middle and anterior cingulate cortex. The language domain (Fig. 2D) showed significant regions in the (para)hippocampal region and LGN left more than right.