Abstract
Introduction:
To identify clinically implementable biomarkers of cognitive impairment in Parkinson’s Disease (PD) derived from resting state-functional MRI (rs-fMRI) and CSF protein analysis.
Methods:
In this single-center longitudinal cohort study, we analyzed rs-fMRI and CSF biomarkers from 50 PD patients (23 cognitively normal, 18 mild cognitive impairment, 9 dementia) and 19 controls, who completed comprehensive neuropsychological testing. A subgroup of participants returned for follow-up cognitive assessments three years later. From rs-fMRI, we studied the connectivity within two distinct Default Mode Network subsystems: left-to-right hippocampus (LHC-RHC) and medial prefrontal cortex-to-posterior cingulate cortex (mPFC-PCC). We used regression analyses to determine whether imaging (LHC-RHC, mPFC-PCC), clinical (CSF Aβ-42:40, disease duration), and demographic (age, sex, education) variables were associated with global and domain-specific cognitive impairments.
Results:
LHC-RHC (F3,67=3.41,p=0.023) and CSF Aβ-42:40 (χ2(3)=8.77,p=0.033) were reduced across more cognitively impaired PD groups. Notably, LHC-RHC connectivity was significantly associated with all global and domain-specific cognitive impairments (attention/executive, episodic memory, visuospatial, and language) at the baseline visit. In an exploratory longitudinal analysis, mPFC-PCC was associated with future global and episodic memory impairment.
Conclusion:
We used biomarker techniques that are readily available in clinical and research facilities to shed light on the pathophysiologic basis of cognitive impairment in PD. Our findings suggest that there is a functionally distinct role of the hippocampal subsystem within the DMN resting state network, and that intrinsic connectivity between the hippocampi is critically related to a broad range of cognitive functions in PD.
Keywords: Mild cognitive impairment, Parkinson's disease/Parkinsonism, Dementia, rs-fMRI, Default Mode Network, CSF, Biomarkers
Introduction
The heterogeneity of cognitive impairments in Parkinson’s disease (PD) impedes therapeutic development, and there is a need for clinically implementable biomarkers associated with the varying decline[1, 2]. Across the literature, resting state-functional MRI (rs-fMRI) and CSF Aβ-42 are consistently identified as potential candidates[2-4]. Specifically, a recent meta-analysis of rs-fMRI studies in PD showed that cognitive impairment is associated with Default Mode Network (DMN) connectivity[3, 5]. The DMN is commonly considered a single network where reduced connectivity is associated with decreased memory, slower processing speed, and decreased executive function[3, 5, 6]. However, recent studies suggest that the DMN is organized around a set of interacting subsystems with differential implications in cognitive processes[6, 7]. Anatomically, the central nodes of the DMN are the medial prefrontal cortex (mPFC), posterior cingulate cortex (PCC), and left and right hippocampi and inferior parietal lobes. In the midline, the mPFC and PCC form the core subsystem at the center of many integrated networks[6]. Laterally, within the temporal lobes, the hippocampi form a subsystem that is intrinsically connected but distinct from other brain networks[6, 7]. Together these may function to process past experiences that are adaptive for future use[6].
In the CSF, the most replicated finding is the association between low Aβ-42 and longitudinal PD cognitive decline[2, 4]. While abnormal Aβ in the CSF is traditionally associated with Alzheimer’s disease-related pathology, recent studies demonstrate a third of newly diagnosed PD patients have amyloidosis, without elevated tau[8], and that overexpression of α-synuclein synergistically induces increased production of Aβ[9].
This study combined DMN subsystem connectivity and CSF Aβ, along with other clinical and demographic factors, to determine which variables are associated with cross-sectional and longitudinal global cognitive impairment in PD. We then used comprehensive neuropsychological assessments to underpin the association with domain-specific cognitive impairments.
Methods
Participants
We recruited PD and healthy control (HC) participants from the Stanford Movement Disorders Clinic and the surrounding community between 2012 and 2014 according to inclusion and exclusion criteria detailed in Supplemental Methods. We collected imaging, clinical, and demographic data from 52 participants with PD and 21 age- and education-matched HC at the baseline visit. Of these, 21 participants with PD and 11 HC returned for cognitive assessment at the follow-up visit three years later (see Supplemental Figure).
Motor and cognitive assessments
The movement disorders neurologist performed a comprehensive research assessment that included medical history, general neurological exam, and the MDS-UPDRS-III[10] in the off-medication state according to published criteria[11].
PD participants underwent comprehensive neuropsychological testing[12] in the on-medication state to assess cognitive function without interference by motor deficits [13].
We classified PD participants as cognitively normal (PD-CN) or with mild cognitive impairment (PD-MCI) according to criteria proposed by the Movement Disorder Society commissioned task force[13], or dementia (PDD) as determined by the movement disorders neurologist’s evaluation and the Clinical Dementia Rating scale[14]. HC was defined as cognitive performance within 1.5 standard deviations of age- and education-matched normative values.
This study’s primary goal was to determine imaging, clinical, and demographic variables associated with global cognitive function in PD. For this purpose, we used the Montreal Cognitive Assessment (MoCA), which assesses multiple cognitive domains that may be impaired in PD. The secondary goal was to determine the relationship between these variables and domain-specific cognitive impairments. Therefore, we calculated a standardized composite z-score from neuropsychological test outcomes that best represented attention/executive, episodic memory, visuospatial, and language domains (see Supplemental Methods and Supplemental Table 2).
CSF Protein Measurements
The movement disorders neurologist performed lumbar punctures to collect CSF samples on all PD and HC participants at baseline according to standardized procedures (see Supplemental Methods).
Structural MRI acquisition
We acquired structural MRI and rs-fMRI scans on a GE Discovery MR 750 3.0 T scanner (General Electric Healthcare, Milwaukee, WI) using a custom-built head coil at Stanford University. For each participant, we acquired a high-resolution T1-weighted spoiled grass gradient recalled inversion recovery 3D MRI sequence to facilitate anatomical localization of functional data[11].
rs-fMRI acquisition
As previously published[11], we acquired a 10-minute rs-fMRI scan in the off-medication state using a T2*-weighted gradient-echo spiral in-out pulse sequence (TR=2s, TE=30ms, and flip angle=80°). The field of view was 20cm, and the matrix size was 64x64, which provided an in-plane spatial resolution of 3.125mm. Measures taken to optimize rs-fMRI acquisition are detailed in Supplemental Methods.
rs-fMRI preprocessing
According to standard protocols, we pre-processed rs-fMRI images using SPM-12 (Wellcome Trust Centre for Neuroimaging, University College London). To allow for T1 equilibration, we discarded the first five volumes. During reconstruction, we applied a linear shim correction to each slice[15]. We realigned the images to the first scan to correct for motion and slice acquisition timing and used ArtRepair software (Toolbox for SPM-12, Stanford University) to correct for deviant volumes resulting from head motion and transient artifacts. We excluded two HC, one PD-CN, and one PD-MCI with more than 5% of volumes corrected and more than 2 mm maximum scan-to-scan movement (see Table 1 for details), leaving a final cohort of 50 PD and 19 HC. After the artifacts repair, we co-registered scans to the corresponding participant’s high-resolution T1 structural image. We then spatially normalized images to standard Montreal Neurological Institute space for group-level analyses. Images were resampled to 2mm isotropic voxels and smoothed with a 4mm full-width at half-maximum Gaussian Kernel to decrease spatial noise prior to statistical analysis.
Table 1:
Demographic and clinical characteristics at the baseline and follow-up visits.
| Baseline Visit | Follow-up Visit | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| HC | PD-CN | PD-MCI | PDD | p-value | HC | PD-CN | PD-MCI/PDD | p-value | |
| N | 19 | 23 | 18 | 9 | 11 | 9 | 12 | ||
| Baseline cognitive classification | -- | -- | -- | -- | 11 HC | 9 PD-CN | 5 PD-CN; 6 PD-MCI; 1 PDD | ||
| Time between baseline and follow-up visit, y | -- | -- | -- | -- | 3.3 ± 0.4 | 3.2 ± 0.3 | 3.3 ± 0.3 | ||
| Age, y | 65.1 ± 7.2 (56-79) | 65.3 ± 7.7 (50-79) | 70.0 ± 7.0 (59-85) | 72.1 ± 7.6 (50-85) | 0.025a | 67.6 ± 7.9 (59.5-81) | 67.8 ± 7.5 (54-76.5) | 69.7 ± 6.49 (62-83) | 0.726 |
| Education, y | 16.8 ± 2.0 | 16.2 ± 2.5 | 16.6 ± 2.8 | 16.2 ± 2.5 | 0.866 | 16.8 ± 2.2 | 16.4 ± 2.5 | 16.5 ± 2.8 | 0.350 |
| Female | 13 (68.4%) | 12 (52.2%) | 6 (33.3%) | 3 (33.3%) | 0.137 | 8 (72.7%) | 5 (55.6%) | 4 (33.3%) | 0.177 |
| MoCA | 27.8 ± 1.9 (23-30) | 27.6 ± 2.1 (24-30) | 22.8 ± 3.0 (17-28) | 15.2 ± 4.8 (8-23) | < 0.0001b | 27.5 ± 1.6 (26-30) | 27.0 ± 1.9 (25-30) | 23.6 ± 4.1 (13-27) (n=11) | 0.001c |
| Movement Disorders Society-Unified Parkinson’s Disease Rating Scale Part III off-medication | -- | 36.0 ± 10.7 (20-55) | 35.6 ± 10.0 (20-53) | 43.1 ± 12.8 (18-59) | 0.199 | -- | 39.6 ± 14.8 (19-71) | 41.5 ± 17.1 (23-71) (n=10) | 0.796 |
| Disease Duration, y | -- | 4.9 ± 3.6 (0.5-15) | 6.11 ± 5.0 (1-21) | 7.7 ± 5.2 (1-16) | 0.302 | -- | 7.1 ± 2.3 (3.5-9.5) | 8.3 ± 4.1 (4-11) | 0.372 |
| Levodopa equivalent daily dose | -- | 638.8 ± 400.5 (100-1580) | 783.4 ± 357.0 (260-1450) | 660.8 ± 384.6 (300-1547) | 0.358 | -- | 426.6 ± 247.7 (95.5-720) | 476.8 ± 222.7 (112.5-860) | 0.710 |
| Maximum scan-to-scan displacement | 0.44 ± 0.65 | 0.60 ± 0.45 | 0.69 ± 0.39 | 0.72 ± 0.41 | 0.100 | -- | -- | -- | -- |
| % Volumes repaired | 0.48 ± 0.38 | 0.61 ± 0.58 | 0.90 ± 0.87 | 1.01 ± 1.16 | 0.004d | -- | -- | -- | -- |
The table presents mean ± Standard Deviation (range) or n (%) with bolded values indicating p < 0.05.
post-hoc comparisons revealed no significant difference in age between groups.
post-hoc comparisons revealed significant differences in MoCA between HC and PD-MCI, HC and PDD, PD-CN and PD-MCI, PD-CN and PDD, PD-MCI and PDD.
post-hoc comparisons revealed a significant difference in MoCA between HC and PD-CN, and between PD-CN and PD-MCI/PDD.
post-hoc comparisons revealed no significant difference in the % of volumes repaired between groups.
Abbreviations: CN, Cognitively Normal; HC, Healthy Controls; MCI, Mild Cognitive Impairment; MoCA, Montreal Cognitive Assessment; PD, Parkinson’s Disease; PDD; PD Dementia.
Seed ROI Selection
We used previously published atlases[16, 17] to select regions-of-interest (ROI) within the DMN: left hippocampus (LHC), right hippocampus (RHC), mPFC, and PCC. We used Automated Anatomical Labelling Atlas[16] to select the LHC and RHC as structural ROIs based on the macroanatomy (Figure 1A). Unlike the hippocampi, the mPFC does not have a clear anatomical demarcation; therefore, we defined the mPFC and PCC using a previously published functional atlas based on whole-brain connectivity patterns (Figure 1A)[17].
Figure 1. The Default Mode Network and CSF Aβ-42:40 in PD across worsening cognitive impairments.
(Intended for color reproduction) A) Displayed Regions of Interest used for resting state function MRI connectivity analysis. Left and right hippocampal regions (LHC, RHC) in blue, medial prefrontal cortex (mPFC) region in orange, and the posterior cingulate cortex (PCC) region in yellow. B) The box and whisker plots depict between hippocampal connectivity and C) mPFC-PCC connectivity. The mean is represented with a black circle (•) with median, interquartile range and range (minimum to maximum) with the box. There were no differences in between-group D) hippocampal volumes or E) total grey matter and intracranial volumes; however CSF AB-42:40 concentrations were lowest in the Parkinson’s disease with Dementia (PDD). The bar graphs depict mean with Standard Error.
Abbreviations: CN, Cognitively Normal; HC, Healthy Control; MCI, Mild Cognitive Impairment; PD, Parkinson’s Disease.
To address potential atrophy, we calculated LHC, RHC, gray matter, and intracranial volumes using Freesurfer software (http://surfer.nmr.mgh.harvard.edu). We conducted whole-brain segmentation based on neuroanatomical labels assigned to each voxel in the MR images and manually inspected the labels for accuracy.
ROI-to-ROI Connectivity
We conducted ROI-to-ROI connectivity analysis on the rs-fMRI scans and determined the connectivity within the lateral (LHC-RHC) and midline (mPFC-PCC) DMN regions using CONN Toolbox (www.nitrc.org/projects/conn). First, we linear detrended and bandpass filtered time series to 0.009 - 0.08 Hz. Next, we regressed CSF and white matter signals as confounders. To further address the potential effects of head motion, we added 12-element motion parameters as first-level covariates in the connectivity analysis. To determine the ROI-to-ROI connectivity, we extracted average BOLD time series for all voxels in each ROI and computed the Pearson correlations of each time course from one ROI to another. Finally, we normalized the findings using the Fisher R-Z transformation.
Previous studies have demonstrated that increasing age, male sex, and fewer years of education are associated with and predictive of cognitive impairment in PD[18]. While it remains unclear, some studies suggest that there is hippocampal atrophy in PD and that this is associated with cognitive impairment[19]. To account for these possible confounders, we compared ROI-to-ROI connectivity across groups (HC, PD-CN, PD-MCI, and PDD) and added age, sex, education, and hippocampal volume as second-level covariates. In the regression models, we used hippocampal volume as a single second-level covariate in the ROI-to-ROI analyses.
Statistical analyses
To determine between-group differences in imaging (LHC-RHC and mPFC-PCC), clinical (Aβ-42:40, disease duration), or demographic (age, sex, education) variables, we conducted a one-way analysis of variance (ANOVA). A post-hoc Tukey and Games-Howell adjustment was used for between-group pair-wise comparisons as appropriate. To determine the relationship between hippocampal volume and LHC-RHC, and between Aβ-42:40 and DMN subsystem connectivity, we estimated Pearson’s and Spearman’s correlation coefficients as appropriate.
To determine which imaging (LHC-RHC, mPFC-PCC), clinical (Aβ-42:40 and disease duration), or demographic variables (age, sex, education) are associated with global cognitive function in PD we conducted a series of univariable linear regression analyses. We used this to select the independent variables for the multivariable linear regression with global cognitive function (MoCA score) as the dependent variable for the baseline visit. For the domain-specific cognitive function analysis, we repeated the univariable and multivariable linear regression with the standardized composite z-score for each of the cognitive domains (attention/executive, episodic memory, visuospatial, and language scores) as the dependent variables.
To study the association with global and domain-specific cognitive functions at the follow-up visit, we first used the lasso regression to select the independent variables (sex, imaging, and Aβ-42:40 at the baseline visit, and age and disease duration at the follow-up visit) and then conducted the regular multivariable regression analysis with selected variables as covariates. The penalty parameter in the lasso regression was selected via 10-fold cross-validation. In these analyses, we used cognitive function from the follow-up visit as the dependent variable. We used the variable selection via lasso before fitting the multivariable regression model because of the small sample size for the cognitive functions at the follow-up visit. Finally, we repeated the multivariable linear regression models for the follow-up visit while adjusting for the cognitive scores at the baseline visit to account for the change in cognition rather than merely cognitive performance at follow-up. We conducted all statistical analyses using SPSS 23.0, R and GraphPad Prism 6.00.
All participants provided written informed consent to participate in the study following protocols approved by the Stanford Institutional Review Board.
Results
Participant characteristics
Demographic and clinical characteristics from the baseline and follow-up visits are in Table 1. There were no between-group demographic differences. At the follow-up visit, 5 PD-CN at baseline progressed to PD-MCI, and 1 PD-MCI progressed to PDD. No PD-MCI participants reverted to PD-CN at follow-up.
Hippocampal connectivity and volume
At the baseline visit, we found a significant difference in LHC-RHC between HC, PD-CN, PD-MCI, and PDD groups (F3,67=3.41, p=0.023). There were no differences in LHC-RHC between HC and PD-CN, and there was a declining trend in connectivity across more cognitively impaired PD groups (Figure 1B). Specifically, PDD (mean±SD, 0.69±0.13) had reduced connectivity compared to HC and PD-CN (0.98±0.27; p=0.018 and 0.92±0.24; p=0.069, respectively). There were no between-group differences in left hippocampal (F3,67=0.84, p=0.475), right hippocampal (F3,67=1.64, p=0.188), mean hippocampal (F2,49=1.26, p=0.296), total gray matter (χ2(3)=5.72, p=0.126), or intracranial (F3,67=1.155, p=0.344) volumes across groups (Figure 1D, 1E). Similarly, there were no between-group differences in the ratios between mean hippocampal volumes and either total gray matter or intracranial volumes (F3,65=0.440, p=0.719; F3,65=1.80, p=0.156). There was no correlation between LHC-RHC and hippocampal volume in PD (r=−0.06, p=0.694). There was a trend difference in mPFC-PCC between groups (F3,67=2.48, p=0.069, Figure 1C).
CSF Aβ
At the baseline visit, we found a significant difference in Aβ-42:40 between groups (χ2(3)=8.77, p=0.033), with lower Aβ-42:40 in more cognitively impaired PD (Figure 1F). Aβ-42:40 was associated with performance in global cognitive, executive, and language domains in PD, but not in HC (Table 2). There were no associations between Aβ-42:40 and DMN functional connectivity in PD or HC (See Supplemental Results).
Table 2:
Univariable linear regressions of global and domain-specific cognitive scores at baseline and follow-up visits across all PD participants.
| Global | Attention/ Executive |
Episodic Memory |
Visuospatial | Language | |
|---|---|---|---|---|---|
| Baseline Visit | |||||
| LHC-RHC | 0.39 (0.006) | 0.37 (0.008) | 0.35 (0.014) | 0.43 (0.002) | 0.33 (0.021) |
| mPFC-PCC | 0.25 (0.080) | 0.13 (0.394) | 0.27 (0.059) | 0.08 (0.597) | 0.02 (0.917) |
| Aβ-42:40 | 0.25 (0.079) | 0.48 (<0.0001) | 0.17 (0.252) | 0.22 (0.107) | 0.47 (0.001) |
| Disease Duration | −0.24 (0.088) | −0.38 (0.007) | −0.11 (0.466) | −0.25 (0.086) | −0.01 (0.954) |
| Age | −0.31 (0.028) | −0.51 (<0.0001) | −0.25 (0.082) | −0.145 (0.316) | −0.38 (0.006) |
| Sex | 0.22 (0.120) | 0.23 (0.103) | 0.33 (0.019) | 0.05 (0.756) | 0.05 (0.722) |
| Education | −0.10 (0.477) | 0.06 (0.657) | −0.19 (0.182) | −0.07 (0.625) | −0.22 (0.131) |
| Follow-up Visit | |||||
| LHC-RHC | 0.12 (0.625) | −0.38 (0.099) | 0.04 (0.862) | 0.08 (0.736) | −0.19 (0.399) |
| mPFC-PCC | 0.45 (0.045) | 0.051 (0.829) | 0.51 (0.018) | −0.06 (0.812) | 0.15 (0.522) |
| Aβ-42:40 | 0.67 (0.001) | 0.31 (0.177) | 0.23 (0.321) | 0.16 (0.498) | −0.08 (0.723) |
| Disease Duration | 0.11 (0.640) | −0.23 (0.338) | −0.27 (0.242) | 0.14 (0.552) | 0.07 (0.757) |
| Age | −0.33 (0.155) | −0.19 (0.424) | −0.14 (0.560) | 0.03 (0.899) | −0.16 (0.496) |
| Sex | 0.33 (0.154) | 0.51 (0.021) | 0.25 (0.276) | −0.02 (0.922) | 0.13 (0.582) |
| Education | 0.39 (0.090) | 0.29 (0.222) | 0.13 (0.574) | 0.16 (0.498) | −0.13 (0.568) |
Values represent β (p-value) with bolded values representing significant correlations (p<0.05). Abbreviations: LHC-RHC, left to right hippocampal connectivity; mPFC-PCC, medial prefrontal cortex-posterior cingulate cortex connectivity.
Predicting global cognitive function in PD
The univariable linear regression analyses at the baseline visit showed that LHC-RHC and age were significantly associated with global cognitive function, while mPFC-PCC, Aβ-42:40, and disease duration showed a trend towards significance (Table 2). In the multivariable regression analysis, we found imaging, clinical, and demographic variables explained 19% of the MoCA variance (p=0.024), and LHC-RHC was the only significant variable in the model (Table 3). The univariable linear regression analyses at the follow-up visit showed that mPFC-PCC and Aβ-42:40 were significantly associated with global cognitive function, while education showed a trend towards significance. In contrast to baseline cross-sectional results, where LHC-RHC was associated with global cognitive function, results from the multivariable regression analysis showed that mPFC-PCC and education significantly predicted follow-up cognitive function (Table 4).
Table 3:
Multivariable regression analyses of global and domain-specific cognitive functions at the baseline visit across all PD participants.
| Baseline Outcome |
Global | Attention/ Executive |
Episodic Memory |
Visuospatial | Language |
|---|---|---|---|---|---|
| LHC-RHC | |||||
| B | 7.184 | 0.311 | 0.296 | 0.356 | 0.253 |
| (95% CI) | (0.97, 13.4) | (0.10, 0.52) | (−0.03, 0.625) | (0.12, 0.59) | (0.01, 0.49) |
| p-value | 0.029 | 0.005 | 0.084 | 0.004 | 0.046 |
| mPFC-PCC | |||||
| B | 2.010 | −0.128 | 0.093 | 0.020 | −0.044 |
| (95% CI) | (−3.57, 7.59) | (−0.31, 0.06) | (−0.20, 0.39) | (−0.19, 0.23) | (−0.26, 0.17) |
| p-value | 0.484 | 0.185 | 0.538 | 0.854 | 0.691 |
| Aβ-42:40 | |||||
| B | 29.82 | 2.048 | 0.382 | 2.212 | 6.863 |
| (95% CI) | (−93.0, 152.7) | (−2.05, 6.14) | (−6.11, 6.87) | (−2.36, 6.78) | (2.10, 11.63) |
| p-value | 0.637 | 0.333 | 0.909 | 0.349 | 0.007 |
| Disease Duration | |||||
| B | −0.297 | −0.018 | −0.007 | −0.010 | 0.001 |
| (95% CI) | (−0.61, 0.02) | (−0.03, −0.01) | (−0.02, 0.01) | (−0.02, 0.002) | (−0.012, 0.013) |
| p-value | 0.072 | 0.002 | 0.392 | 0.099 | 0.931 |
| Age | |||||
| B | −0.139 | −0.014 | −0.005 | 0.0004 | −0.003 |
| (95% CI) | (−0.38, 0.11) | (−20.02, −0.01) | (−0.02, 0.01) | (−0.009, 0.009) | (−0.01, 0.01) |
| p-value | 0.270 | 0.002 | 0.433 | 0.930 | 0.584 |
| Sex | |||||
| B | 0.650 | 0.081 | 0.105 | −0.019 | −0.022 |
| (95% CI) | (−2.52, 3.82) | (−0.02, 0.19) | (−0.06, 0.27) | (−0.137, 0.099) | (−0.15, 010) |
| p-value | 0.690 | 0.139 | 0.227 | 0.752 | 0.726 |
| Education | |||||
| B | −0.125 | 0.012 | −0.013 | −0.004 | −0.019 |
| (95% CI) | (−0.71, 0.46) | (−0.008, 0.03) | (−0.044, 0.018) | (−0.025, 0.018) | (−0.04, 0.003) |
| p-value | 0.675 | 0.241 | 0.419 | 0.748 | 0.103 |
Abbreviations: B (95% CI), Unstandardized Beta-coefficient (95% Confidence Interval); LHC-RHC, left to right hippocampal connectivity; mPFC-PCC, medial prefrontal cortex-posterior cingulate cortex connectivity. Variables significantly contributing to the model (p<0.05) are bolded.
Table 4:
Lasso-estimate of the regression coefficient followed by ANCOVA of global and domain-specific cognitive functions at the follow-up visit across all PD participants
| Follow-up Outcome |
Global | Attention/ Executive | Episodic Memory | Visuospatial | Language |
|---|---|---|---|---|---|
| LHC-RHC | |||||
| lasso-estimate | 0 | −0.04 | 0 | 0 | 0 |
| ANCOVA | −0.164 | ||||
| 95% CI | (−0.66, 0.33) | ||||
| p-value | 0.529 | ||||
| mPFC-PCC | |||||
| lasso-estimate | 3.122 | 0 | 0.126 | 0 | 0 |
| ANCOVA | 4.088 | 0.205 | |||
| 95% CI | (0.96,7.22) | (0.03, 0.38) | |||
| p-value | 0.023 | 0.032 | |||
| Aβ-42:40 | |||||
| lasso-estimate | 0 | 5.11 | 0 | 0 | 0 |
| ANCOVA | 5.431 | ||||
| 95% CI | (−8.19, 19.1) | ||||
| p-value | 0.448 | ||||
| Disease Duration | |||||
| lasso-estimate | −0.024 | 0 | −0.007 | 0 | 0 |
| ANCOVA | −0.141 | −0.017 | |||
| 95% CI | (−0.50,0.22) | (−0.04, 0.00) | |||
| p-value | 0.460 | 0.143 | |||
| Age | |||||
| lasso-estimate | 0 | 0 | 0 | 0 | 0 |
| Sex | |||||
| lasso-estimate | 0.045 | 0.148 | 0 | 0 | 0 |
| ANCOVA | 0.468 | 0.219 | |||
| 95% CI | (−1.39, 2.32) | (0.048, 0.39) | |||
| p-value | 0.629 | 0.025 | |||
| Education | |||||
| lasso-estimate | 0.306 | 0.012 | 0 | 0 | 0 |
| ANCOVA | 0.465 | 0.025 | |||
| 95% CI | (0.16, 0.77) | (−0.012, 0.06) | |||
| p-value | 0.009 | 0.206 |
Abbreviations: 95% CI, 95% Confidence Interval; LHC-RHC, left to right hippocampal connectivity; mPFC-PCC, medial prefrontal cortex-posterior cingulate cortex connectivity. Variables significantly contributing to the ANCOVA model (p<0.05) are bolded.
Domain-specific cognitive impairment in PD
Results from the univariable linear regression analyses at the baseline visit are depicted in Table 2. LHC-RHC, Aβ-42:40, disease duration, and age significantly correlated with attention/executive performance; LHC-RHC and sex significantly correlated with episodic memory performance; LHC-RHC significantly correlated with visuospatial performance; and LHC-RHC, Aβ-42:40, and age significantly correlated with language performance.
Results from the multivariable linear regression analyses are in Table 3. We found that reduced LHC-RHC was significantly associated with cognitive impairments in attention/executive, visuospatial, and language, and trended toward significantly associating with episodic memory. In addition, we found that longer disease duration and higher age were associated with attention/executive impairment, and low Aβ-42:40 was associated with language impairment.
Results from the univariable linear regression analyses at the follow-up visit are depicted in Table 2. mPFC-PCC significantly correlated with episodic memory performance, and sex significantly correlated with attention/executive performance.
Results from the multivariable linear regression analyses of the follow-up visit, after variable selection via lasso regression, are depicted in Table 4. In contrast to the baseline results, reduced mPFC-PCC significantly predicted episodic memory, and male sex predicted attention/executive impairment. Accounting for cognitive function at the baseline visit, we found that only education remained significantly associated with future global cognitive function.
Discussion
We studied rs-fMRI of the DMN and CSF Aβ-42:40 to determine which imaging, clinical and demographic biomarkers are associated with cognitive impairments in PD. Within the DMN, we studied two functionally connected and distinct subsystems. Notably, we discovered that intrinsic connectivity between the hippocampi was associated with all global and domain-specific cognitive impairments. Extending these findings to exploratory follow-up data, we found that intrinsic connectivity between the mPFC and PCC was associated with future global cognitive function and episodic memory.
Hippocampal connectivity and cognitive impairment in PD
While a recent meta-analysis showed that cognitive impairment in PD is associated with disrupted DMN connectivity[3], a more recent cohort study did not find reduced DMN connectivity in cognitively impaired PD[20]. These discordant findings could be because unique DMN subsystems contribute differentially to cognitive processing in PD[7]. Only a few prior studies have considered the hippocampal subsystem independently in PD[3, 21]. We found similar hippocampal connectivity between healthy older adults and cognitively normal PD, but reduced connectivity in PD with cognitive impairment, and specifically those with dementia. Further, we found that hippocampal connectivity was associated with global cognitive function and episodic memory, as we expected, but also with attention/executive, visuospatial, and language dysfunctions. While reduced DMN connectivity is commonly associated with impaired memory performance[5], past studies have found associations with executive dysfunction and slower cognitive processing speed in healthy older populations[22]. We extended these findings to PD and observed an association between intrinsic hippocampal connectivity and multi-domain cognitive dysfunctions.
Hippocampal atrophy is previously described in PD and can be associated with cognitive impairment [19]. In this study, structural MRI scans showed no obvious cerebral white matter lesions, and volumetric analyses of the potentially vulnerable hippocampi showed no significant differences between groups. This suggests that in our cohort, a mechanism other than neuronal loss with subsequent atrophy accounts for the association between reduced hippocampal connectivity and PD-related cognitive impairment. It is, however, possible that clinically meaningful hippocampal atrophy was undetectable due to our limited sample size.
mPFC-PCC is associated with future cognitive impairment in PD
While LHC-RHC was associated with baseline cognitive impairments in PD, we found that mPFC-PCC was associated with cognitive impairments at the follow-up visit. This finding is supported by past cross-sectional studies that reported a functional disconnection of the mPFC[23] and showed decreased PCC connectivity most prominently distinguishes PD with versus without mild cognitive impairment[3]. Only one longitudinal study investigated and reported associations between DMN and frontoparietal control networks with PD-related cognitive decline[24]. The follow-up period was, however, limited to one year, and within-network connectivity was not reported. While the mPFC remains unaffected by Lewy bodies before Braak stage 5, it is suggested that early dysfunction of this region arises as a result of alterations in neurotransmitter interactions[25]. Interestingly, we did not find an association between reduced mPFC-PCC connectivity and cognitive impairments in the cross-sectional analysis, where the PDD group showed the largest range of mPFC-PCC connectivity values. This suggests that the relationship between the connectivity in these brain regions and the development of dementia is complex and likely non-linear, warranting further investigation.
Combining biomarkers of PD cognitive impairments
We combined rs-fMRI with CSF analyses to create predictive models of PD-related cognitive impairments. In the past, numerous CSF proteins have been proposed as predictive biomarkers of PD-related cognitive impairment, with the most studied being Aβ, tau, α-synuclein, and neurofilament light chain. Of these, there is agreement across studies that low Aβ-42 predicts the future development of cognitive impairment in PD[2]. However, Aβ-42:40 may be a more accurate measure of abnormal amyloid by accounting for inter-individual variabilities in Aβ production, reducing CSF signal-to-noise ratio, and attenuating biases in inconsistent sample handling. While correlated in the univariable analysis, we did not find Aβ-42:20 associated with future cognitive impairment in the multivariable model. We also found that CSF Aβ-42:40 is associated with attention/executive and language functions in the univariable analysis but was only associated with language in the multivariable model. In light of previous studies, the association with language may in part be explained by the accompanying executive dysfunction. Language impairments may however also be present in the absence hereof[26] and even in PD without MCI or dementia[27]. In our cohort, the relationship between CSF Aβ-42:40 and language was significant only in cognitively impaired PD (r=0.42;p=0.028) but not in cognitively normal PD (r=0.23;p=0.313). Alternatively, the relationship between CSF Aβ-42:40 and language could be attributable to concomitant Alzheimer’s disease-related pathology in some PD patients. Finally, similar to two previous studies of DMN connectivity[20, 28], we found no association between CSF Aβ-42:40 and DMN connectivity. This is in contrast to one study showing an association between CSF Aβ-42 and PD between-network connectivity[29]. However, this later study incorporated connectivity data from 6 intrinsic brain networks, which could explain their results. Together, our findings suggest that the association between DMN connectivity and PD-related cognitive impairment is an independent biomarker from CSF Aβ-42:40.
Methodologic considerations
The empirical results from the analyses at the follow-up visit should be considered in the light of limitations imposed by the low rate of participant return at follow-up. The demographic and clinical characteristics of participants lost to follow-up (see Supplemental Table 1 and Results) show this cohort started at baseline with more severe executive and visuospatial impairments and were more likely to have a baseline diagnosis of dementia. This is unsurprising, considering the rapid decline of executive and visuospatial cognitive functions in PD[1]. In addition, the confidence interval and p-values from the multivariable linear regression post lasso selection may not be valid due to the selection bias. Thus, we urge cautious interpretation that we only identified predictors of follow-up global and memory impairments. Furthermore, cognitive function at the follow-up visit was associated with cognitive function at the baseline visit. When accounting for baseline cognitive function, only education remained significantly associated with follow-up-global cognitive function.
Conclusion
Using biomarker techniques that are available in regular clinics and research facilities, our findings suggest that reduced intrinsic connectivity between the hippocampi is associated with global and multi-domain cognitive impairments in PD. In light of the macroscopic integrity of the hippocampi in PD, these findings provide a unique insight into the pathophysiologic basis of cognitive impairments in PD. Extending these findings to exploratory longitudinal data, our results indicate a dissociation of the DMN in PD, where reduced connectivity between the mPFC and PCC are associated with future cognitive impairments. 3T MRI is available in many facilities, and automated segmentation software is readily available, making this approach suitable for broad, clinically implementable applications. We encourage future largescale studies to assess the utility of these findings as biomarkers of disease progression and therapeutic intervention.
Supplementary Material
Supplemental Figure
Flow-chart of baseline and follow-up participants.
Abbreviations: CN, Cognitively Normal; HC, Healthy Controls; MCI, Mild Cognitive Impairment; PD, Parkinson’s Disease; PDD, PD Dementia.
Highlights.
Resting state-fMRI and CSF biomarkers are associated with PD cognitive impairments
Hippocampal connectivity associates with global and multi-domain cognitive impairments
mPFC to PCC connectivity best predicts future cognitive impairment in PD
Acknowledgement
This research was supported by grants from the Michael J. Fox Foundation for Parkinson’s disease research (grant 6440), the NIH/NINDS (NS075097, NS062684), NIH/NIA (AG047366), and Novo Nordisk.
Full funding sources over the past 12 months:
Pardis Zarifkar has received research support from Novo Nordisk, the foundation of 17-12-1981, John Palle Buhl Law Firm and the Lundbeck Foundation-Danish Society for Neuroscience.
Jeehyun Kim has nothing to disclose
Dr. Christian La is now an employee of LVIS Corporation
Dr. Kai Zhang is now an employee of Nines Inc.
Sophie YorkWilliams has nothing to disclose
Taylor Levine has nothing to disclose
Dr. Lu Tian has nothing to disclose
Dr. Per Borghammer has received research grants from the Lundbeck Foundation, Novo Nordisk Foundation, Jascha Foundation, Danish Research Council, Danish Parkinson Foundation.
Dr. Kathleen Poston has received consulting fees from Allergan and Curasen, and research grants from Sanofi, the Michael J Fox Foundation for Parkinson’s Research and the NIH.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Declaration of interest: The authors declare no conflicts of interest in relation to this study.
Data Availability Statement:
Data not published in this article may be shared by request from qualified investigators.
References
- [1].Aarsland D, Creese B, Politis M, Chaudhuri KR, Ffytche DH, Weintraub D, Ballard C. Cognitive decline in Parkinson disease. Nat Rev Neurol. 2017;13(4):217–31. Epub 2017/03/04. doi: 10.1038/nrneurol.2017.27. PubMed PMID: 28257128; PMCID: PMC5643027. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [2].Leaver K, Poston KL. Do CSF Biomarkers Predict Progression to Cognitive Impairment in Parkinson's disease patients? A Systematic Review. Neuropsychol Rev. 2015;25(4):411–23. doi: 10.1007/s11065-015-9307-8. PubMed PMID: 26626621. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [3].Wolters AF, van de Weijer SCF, Leentjens AFG, Duits AA, Jacobs HIL, Kuijf ML. Resting-state fMRI in Parkinson's disease patients with cognitive impairment: A meta-analysis. Parkinsonism Relat Disord. 2019;62:16–27. Epub 2018/12/26. doi: 10.1016/j.parkreldis.2018.12.016. PubMed PMID: 30580907. [DOI] [PubMed] [Google Scholar]
- [4].Shahid M, Kim J, Leaver K, Hendershott T, Zhu D, Cholerton B, Henderson VW, Tian L, Poston KL. An increased rate of longitudinal cognitive decline is observed in Parkinson's disease patients with low CSF Ass42 and an APOE epsilon4 allele. Neurobiol Dis. 2019;127:278–86. Epub 2019/03/04. doi: 10.1016/j.nbd.2019.02.023. PubMed PMID: 30826425; PMCID: PMC6588475. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [5].Greicius MD, Srivastava G, Reiss AL, Menon V. Default-mode network activity distinguishes Alzheimer's disease from healthy aging: evidence from functional MRI. Proc Natl Acad Sci U S A. 2004;101(13):4637–42. Epub 2004/04/09. doi: 10.1073/pnas.0308627101. PubMed PMID: 15070770; PMCID: PMC384799. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [6].Buckner RL, Andrews-Hanna JR, Schacter DL. The brain's default network: anatomy, function, and relevance to disease. Ann N Y Acad Sci. 2008;1124:1–38. Epub 2008/04/11. doi: 10.1196/annals.1440.011. PubMed PMID: 18400922. [DOI] [PubMed] [Google Scholar]
- [7].Andrews-Hanna JR, Reidler JS, Sepulcre J, Poulin R, Buckner RL. Functional-anatomic fractionation of the brain's default network. Neuron. 2010;65(4):550–62. doi: 10.1016/j.neuron.2010.02.005. PubMed PMID: 20188659; PMCID: PMC2848443. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [8].Irwin DJ, Felder J, Coffey CS, Caspell-Garcia C, Kang JH, Simuni T, Foroud T, Toga AW, Tanner CM, Kieburtz K, Chahine LM, Reimer A, Hutten S, Weintraub D, Mollenhauer B, Galasko DR, Siderowf A, Marek K, Trojanowski JQ, Shaw LM, Parkinson's Progression Marker I. Evolution of Alzheimer's Disease Cerebrospinal Fluid Biomarkers in Early Parkinson's Disease. Ann Neurol. 2020. Epub 2020/06/17. doi: 10.1002/ana.25811. PubMed PMID: 32542885. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [9].Irwin DJ, Lee VM, Trojanowski JQ. Parkinson's disease dementia: convergence of alpha-synuclein, tau and amyloid-beta pathologies. Nat Rev Neurosci. 2013;14(9):626–36. Epub 2013/08/01. doi: 10.1038/nrn3549. PubMed PMID: 23900411; PMCID: PMC4017235. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [10].Goetz CG, Tilley BC, Shaftman SR, Stebbins GT, Fahn S, Martinez-Martin P, Poewe W, Sampaio C, Stern MB, Dodel R, Dubois B, Holloway R, Jankovic J, Kulisevsky J, Lang AE, Lees A, Leurgans S, LeWitt PA, Nyenhuis D, Olanow CW, Rascol O, Schrag A, Teresi JA, van Hilten JJ, LaPelle N, Movement Disorder Society URTF. Movement Disorder Society-sponsored revision of the Unified Parkinson's Disease Rating Scale (MDS-UPDRS): scale presentation and clinimetric testing results. Mov Disord. 2008;23(15):2129–70. Epub 2008/11/26. doi: 10.1002/mds.22340. PubMed PMID: 19025984. [DOI] [PubMed] [Google Scholar]
- [11].Ng B, Varoquaux G, Poline JB, Thirion B, Greicius MD, Poston KL. Distinct alterations in Parkinson's medication-state and disease-state connectivity. NeuroImage Clinical. 2017;16:575–85. Epub 2017/10/04. doi: 10.1016/j.nicl.2017.09.004. PubMed PMID: 28971008; PMCID: PMC5608603. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [12].Hendershott TR, Zhu D, Llanes S, Zabetian CP, Quinn J, Edwards KL, Leverenz JB, Montine T, Cholerton B, Poston KL. Comparative sensitivity of the MoCA and Mattis Dementia Rating Scale-2 in Parkinson's disease. Mov Disord. 2019;34(2):285–91. Epub 2019/02/19. doi: 10.1002/mds.27575. PubMed PMID: 30776152; PMCID: PMC6532623. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [13].Litvan I, Goldman JG, Troster AI, Schmand BA, Weintraub D, Petersen RC, Mollenhauer B, Adler CH, Marder K, Williams-Gray CH, Aarsland D, Kulisevsky J, Rodriguez-Oroz MC, Burn DJ, Barker RA, Emre M. Diagnostic criteria for mild cognitive impairment in Parkinson's disease: Movement Disorder Society Task Force guidelines. Mov Disord. 2012;27(3):349–56. Epub 2012/01/26. doi: 10.1002/mds.24893. PubMed PMID: 22275317; PMCID: PMC3641655. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [14].Emre M, Aarsland D, Brown R, Burn DJ, Duyckaerts C, Mizuno Y, Broe GA, Cummings J, Dickson DW, Gauthier S, Goldman J, Goetz C, Korczyn A, Lees A, Levy R, Litvan I, McKeith I, Olanow W, Poewe W, Quinn N, Sampaio C, Tolosa E, Dubois B. Clinical diagnostic criteria for dementia associated with Parkinson's disease. Mov Disord. 2007;22(12):1689–707; quiz 837. Epub 2007/06/02. doi: 10.1002/mds.21507. PubMed PMID: 17542011. [DOI] [PubMed] [Google Scholar]
- [15].Glover GH, Lai S. Self-navigated spiral fMRI: interleaved versus single-shot. Magn Reson Med. 1998;39(3):361–8. Epub 1998/03/14. PubMed PMID: 9498591. [DOI] [PubMed] [Google Scholar]
- [16].Tzourio-Mazoyer N, Landeau B, Papathanassiou D, Crivello F, Etard O, Delcroix N, Mazoyer B, Joliot M. Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain. Neuroimage. 2002;15(1):273–89. Epub 2002/01/05. doi: 10.1006/nimg.2001.0978. PubMed PMID: 11771995. [DOI] [PubMed] [Google Scholar]
- [17].Shirer WR, Ryali S, Rykhlevskaia E, Menon V, Greicius MD. Decoding subject-driven cognitive states with whole-brain connectivity patterns. Cereb Cortex. 2012;22(1):158–65. Epub 2011/05/28. doi: 10.1093/cercor/bhr099. PubMed PMID: 21616982; PMCID: 3236795. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [18].Cholerton BA, Zabetian CP, Quinn JF, Chung KA, Peterson A, Espay AJ, Revilla FJ, Devoto J, Watson GS, Hu SC, Edwards KL, Montine TJ, Leverenz JB. Pacific Northwest Udall Center of excellence clinical consortium: study design and baseline cohort characteristics. J Parkinsons Dis. 2013;3(2):205–14. Epub 2013/08/14. doi: 10.3233/JPD-130189. PubMed PMID: 23938350; PMCID: PMC3779428. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [19].La C, Linortner P, Bernstein JD, Ua Cruadhlaoich MAI, Fenesy M, Deutsch GK, Rutt BK, Tian L, Wagner AD, Zeineh M, Kerchner GA, Poston KL. Hippocampal CA1 subfield predicts episodic memory impairment in Parkinson's disease. NeuroImage Clinical. 2019;23:101824. Epub 2019/05/06. doi: 10.1016/j.nicl.2019.101824. PubMed PMID: 31054380; PMCID: PMC6500913. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [20].Campbell MC, Jackson JJ, Koller JM, Snyder AZ, Kotzbauer PT, Perlmutter JS. Proteinopathy and longitudinal changes in functional connectivity networks in Parkinson disease. Neurology. 2020;94(7):e718–e28. Epub 2019/12/20. doi: 10.1212/WNL.0000000000008677. PubMed PMID: 31852813; PMCID: PMC7176296. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [21].Das T, Hwang JJ, Poston KL. Episodic recognition memory and the hippocampus in Parkinson's disease: A review. Cortex; a journal devoted to the study of the nervous system and behavior. 2019;113:191–209. Epub 2019/01/21. doi: 10.1016/j.cortex.2018.11.021. PubMed PMID: 30660957; PMCID: PMC6445686. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [22].Andrews-Hanna JR, Snyder AZ, Vincent JL, Lustig C, Head D, Raichle ME, Buckner RL. Disruption of large-scale brain systems in advanced aging. Neuron. 2007;56(5):924–35. Epub 2007/12/07. doi: 10.1016/j.neuron.2007.10.038. PubMed PMID: 18054866; PMCID: PMC2709284. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [23].van Eimeren T, Monchi O, Ballanger B, Strafella AP. Dysfunction of the default mode network in Parkinson disease: a functional magnetic resonance imaging study. Arch Neurol. 2009;66(7):877–83. Epub 2009/07/15. doi: 10.1001/archneurol.2009.97. PubMed PMID: 19597090; PMCID: PMC2972248. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [24].Klobusiakova P, Marecek R, Fousek J, Vytvarova E, Rektorova I. Connectivity Between Brain Networks Dynamically Reflects Cognitive Status of Parkinson's Disease: A Longitudinal Study. J Alzheimers Dis. 2019;67(3):971–84. Epub 2019/02/19. doi: 10.3233/JAD-180834. PubMed PMID: 30776007; PMCID: PMC6398554. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [25].Kehagia AA, Barker RA, Robbins TW. Neuropsychological and clinical heterogeneity of cognitive impairment and dementia in patients with Parkinson's disease. Lancet Neurol. 2010;9(12):1200–13. Epub 2010/10/01. doi: 10.1016/S1474-4422(10)70212-X. PubMed PMID: 20880750. [DOI] [PubMed] [Google Scholar]
- [26].Ibanez A, Cardona JF, Dos Santos YV, Blenkmann A, Aravena P, Roca M, Hurtado E, Nerguizian M, Amoruso L, Gomez-Arevalo G, Chade A, Dubrovsky A, Gershanik O, Kochen S, Glenberg A, Manes F, Bekinschtein T. Motor-language coupling: direct evidence from early Parkinson's disease and intracranial cortical recordings. Cortex. 2013;49(4):968–84. Epub 2012/04/10. doi: 10.1016/j.cortex.2012.02.014. PubMed PMID: 22482695. [DOI] [PubMed] [Google Scholar]
- [27].Bocanegra Y, Garcia AM, Pineda D, Buritica O, Villegas A, Lopera F, Gomez D, Gomez-Arias C, Cardona JF, Trujillo N, Ibanez A. Syntax, action verbs, action semantics, and object semantics in Parkinson's disease: Dissociability, progression, and executive influences. Cortex. 2015;69:237–54. Epub 2015/06/24. doi: 10.1016/j.cortex.2015.05.022. PubMed PMID: 26103601. [DOI] [PubMed] [Google Scholar]
- [28].Campbell MC, Koller JM, Snyder AZ, Buddhala C, Kotzbauer PT, Perlmutter JS. CSF proteins and resting-state functional connectivity in Parkinson disease. Neurology. 2015;84(24):2413–21. Epub 2015/05/17. doi: 10.1212/WNL.0000000000001681. PubMed PMID: 25979701; PMCID: PMC4478033. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [29].Madhyastha TM, Askren MK, Zhang J, Leverenz JB, Montine TJ, Grabowski TJ. Group comparison of spatiotemporal dynamics of intrinsic networks in Parkinson's disease. Brain. 2015;138(Pt 9):2672–86. Epub 2015/07/16. doi: 10.1093/brain/awv189. PubMed PMID: 26173859; PMCID: PMC4643623. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplemental Figure
Flow-chart of baseline and follow-up participants.
Abbreviations: CN, Cognitively Normal; HC, Healthy Controls; MCI, Mild Cognitive Impairment; PD, Parkinson’s Disease; PDD, PD Dementia.
Data Availability Statement
Data not published in this article may be shared by request from qualified investigators.