Abstract
Background:
Brain diffusion tensor imaging (DTI) has been shown to reflect cognitive changes in early Parkinson’s disease (PD) but the diffusion-based measure free water (FW) has not been previously assessed.
Objectives:
To assess if FW in the thalamic nuclei primarily involved with cognition (i.e., the dorsomedial (DMN) and anterior (AN) nuclei), the nucleus basalis of Meynert (nbM) and the hippocampus correlates with and is associated with longitudinal cognitive decline and distinguishes cognitive status at baseline in early PD. Also, to explore how FW compares with conventional DTI, FW-corrected DTI and volumetric assessments for these outcomes.
Methods:
Imaging data and MoCA scores from the Parkinson’s Progression Markers Initiative database were analyzed using partial correlations and ANCOVA. Primary outcome multiple comparisons were corrected for false discovery rate (q-value).
Results:
Thalamic DMN FW changes over 1 year correlated with MoCA changes over both 1 and 3 years (partial correlations −0.222, q=0.040, n=130; and −0.229, q=0.040, n=123, respectively; mean PD duration at baseline=6.85 months). NbM FW changes over 1 year only correlated with MoCA changes over 3 years (−0.222, q=0.040). Baseline hippocampal FW was associated with cognitive impairment at 3 years (q=0.040) and baseline nbM FW distinguished PD-normal cognition (MoCA≥26) from PD-cognitive impairment (MoCA≤25), (q=0.008). The exploratory comparisons showed FW to be the most robust assessment modality for all outcomes.
Conclusion:
Thalamic DMN FW is a promising cognition progression biomarker in early PD that may assist in identifying cognition protective therapies in clinical trials. FW is a robust assessment modality for these outcomes.
Keywords: Free water, biomarker, cognition, progression, dementia
Introduction:
Cognitive impairment is common in Parkinson’s disease (PD) and progresses to dementia in the majority of patients.1 Unlike PD motor symptoms, there are limited treatments for cognitive impairment and no therapies exist for preventing or delaying the onset of dementia. Because dementia occurs in >80% of PD within 20 years of diagnosis, is highly disabling, increases mortality and is a primary contributor to nursing home placement in PD, there is a critical need for cognition protective therapies that can be administered early in PD prior to the onset of dementia.2–7 As it is unrealistic to perform clinical trials 5–10 years in duration to determine if a therapy can delay the onset of dementia, use of biomarkers that reflect both early and later longitudinal cognitive decline, i.e. cognition progression biomarkers, in PD clinical trials will be essential towards the goal of identifying cognition protective therapies. Use of such biomarkers will greatly assist in distinguishing a therapy’s long-term, disease-modifying effect on the biological underpinnings of progressive cognitive decline versus simply a short-term, symptomatic benefit on cognition.
The Parkinson’s Progression Markers Initiative (PPMI) is an observational, longitudinal clinical study to identify disease progression biomarkers in early, de novo (not receiving PD medications at baseline) PD. As disease-modifying clinical trials typically enroll early, de novo PD, biomarker findings from PPMI are highly relevant for clinical trial design. Recent PPMI studies have shown structural and microstructural changes in the nucleus basalis of Meynert (nbM), hippocampus and whole thalamus to be associated with cognitive changes in PD.8–10 Specifically, reduced nbM volume and increased nbM mean diffusivity (MD), a sign of tissue microstructural damage, distinguished PD with cognitive impairment (PD-CI) from PD with normal cognition (PD-NC) at baseline and predicted more rapid longitudinal cognitive decline.8,9 In addition to the nbM, increased MD in the hippocampus, thalamus, entorhinal cortex (EC) and amygdala distinguished PD-CI from PD-NC with the hippocampus showing the strongest effect after partial volume correction.9 Reduced hippocampal volume was previously shown to be associated with longitudinal cognitive decline in PD while reduced nbM volume was found in PD dementia (PDD) and shown to correlate with cognition.11–15 Finally, an additional PPMI study by Zhang et al. found 1-year changes in whole thalamic radial diffusivity and axial diffusivity, but not MD, to inversely correlate with 1-year changes in Montreal Cognitive Assessment (MoCA) scores highlighting the thalamus as a promising cognition progression biomarker in early PD.10
As cognition progression biomarkers have the potential to identify cognition protective therapies, our main objective in the present analyses was to extend the findings by Zhang et al. by examining for longitudinal correlations between neuroimaging and cognition changes in early PD from PPMI over both 1 and 3 years. By including the 3-year timepoint, we could determine whether short-term correlations persisted in the longer term. We also expanded the number of regions of interest (ROIs) in our primary analyses to include the nbM, hippocampus and the thalamic nuclei primarily involved with cognitive processing: the anterior nucleus (AN) and dorsomedial nucleus (DMN).16,17 In addition, baseline imaging assessments were examined to determine if they were associated with longitudinal cognitive decline and if they could distinguish PD-CI from PD-NC. As exploratory outcomes, we compared volume, conventional diffusion tensor imaging (DTI), free water (FW) and FW-corrected DTI (DTI-t) assessments of these ROIs in addition to the whole thalamus and the EC for the above outcomes. In order to determine if significant findings were unique to PD as opposed to age-related changes, we also repeated most of these analyses comparing PD to aged-matched, healthy controls (HCs) from PPMI.
Because the bi-tensor, diffusion-based assessment of free water (FW) has been shown to be more sensitive than conventional single tensor DTI for detecting longitudinal changes in the substantia nigra in PD,18–20 we hypothesized that the same would be true for ROIs related to cognition and, therefore, chose FW as our primary outcome imaging assessment modality for these comparisons. Specifically, we hypothesized that 1-year longitudinal changes in FW in one or more of the primary ROIs would inversely correlate with 1-year and 3-year longitudinal changes in MoCA scores. Such findings would support FW in these ROIs as cognition progression biomarkers. FW corresponds to water molecules within a voxel that are not hindered or restricted by the cellular environment and therefore originate from extracellular water.21 There is pathological evidence supporting FW to reflect tissue atrophy and inflammation,22,23 which are known to occur in PD. Thus, higher FW levels are believed to reflect more advanced tissue degeneration.18–20,24,25
Methods:
Participants:
Clinical and magnetic resonance imaging (MRI) data from the PPMI database (https://www.ppmi-info.org/access-data-specimens/download-data/) were downloaded in April 2018. Patients were included if they had PD confirmed in May 2021 by the PPMI Consensus Committee (personal communication with committee). PD patients and healthy controls (HCs) were included if they had diffusion-weighted imaging (DWI) and T1 imaging passing quality control analysis and MoCA scores at baseline and either year 1 or year 3. We were primarily interested in assessing longitudinal correlations between changes in imaging and cognition as these analyses can identify cognition progression biomarkers. Because the MoCA is recommended by the Movement Disorder Society (MDS) Task Force for assessing global cognition in PD, is able to predict future PDD, has demonstrated sensitivity to longitudinal change in PPMI and has been used in recent PPMI imaging studies; 8–10,26–28 we used the MoCA for the longitudinal correlative analyses. PD-CI at baseline was defined as MoCA≤25, which has been shown to be an optimal screening cutoff for detecting PD-CI29,30 and was previously used by Schulz et al.9 to define PD-CI in the PPMI cohort. Only PD-NC patients (MoCA≥26 at baseline) were considered for assessing associations between baseline imaging and longitudinal conversion to cognitive impairment. Conversion to Level I mild cognitive impairment (MCI) was defined per MDS-Task Force criteria.27 This definition of cognitive impairment was used to simulate the methodology used in previous PPMI predictive analyses to define PD-MCI.8,9
MRI Acquisition:
MRI scans were acquired at 11 PPMI sites on 3 Tesla machines (all Siemens) using standardized acquisition protocols at each site. The MRI protocols included a three dimensional magnetization prepared rapid gradient echo (MPRAGE) sequence for structural T1-weighted imaging (T1WI, TR/TR/TI = 2300/3/900ms; 1 mm isotropic resolution; twofold acceleration; sagittal-oblique angulation) and a cardiac-gated two-dimensional single-shot echo-planar sequence for mapping brain water diffusion (TR ranged from 8,400 to 8,800 depending on subjects’ heart rate, TE = 88ms, 2 mm isotropic resolution; 72 contiguous slices, twofold acceleration, axial-oblique aligned along the anterior-posterior commissure, with diffusion-weighted gradients along 64 sensitization directions and a b factor of 1000s/mm2).
MRI Assessment:
The raw diffusion weighted imaging (DWI) data were first corrected for head motion, eddy-current effects and geometric distortions. The diffusion tensor was calculated using both single tensor and bi-tensor models, with the later providing an estimate of FW and FW-corrected DTI metrics (DTI-t).21
3D T1-weighted images were corrected for intensity inhomogeneity using the N4 tool and subsequently segmented using the SIENAX tool31 to obtain partial volume estimate maps of the grey matter. The nbM was then segmented as previously described.9 Briefly, a histologically defined map of the nbM32 was brought into the native 3D T1-weighted image space for each scan via non-linearly warping with Advanced Normalization Tools.33 Then, the gray matter partial volume map was used to select nbM voxels with values greater than 0.5, as previously done in other PPMI studies.9
In addition, each 3D T1-weighted image was processed using the FreeSurfer 6.0 pipeline34 along with the thalamic nuclei segmentation submodule.35 Outputs were visually inspected for errors and misclassification. Manual corrections to the FreeSurfer output (e.g., introduction of white matter control points, editing of brain mask, white matter mask) were made as appropriate. The resulting whole thalamic, hippocampal, and entorhinal cortex segmentations were not manually edited (Figure 1) as they were found to be accurately segmented during visual quality control. For the AN, the anteroventral segmentation was utilized. For the DMN, the following individual nuclei were combined into a single segmentation: paratenial, reuniens (medial ventral), mediodorsal medial magnocellular and mediodorsal lateral parvocellular.
Figure 1:
A coronal (top image) and axial (bottom image) slice from a 63-year-old male Parkinson’s disease patient. Except for the nucleus basalis of Meynert, all segmentations are shown with a single-color stemming from the FreeSurfer-derived processing. The nucleus basalis of Meynert is shown with a non-linearly registered probabilistic atlas, with yellow indicating greater probability of corresponding to the structure.
DWI b=0 data was brought into 3D T1-weighted space using boundary-based registration36 as implemented both in the FSL software package as well as FreeSurfer. Average DTI values were then extracted for each ROI after bringing the quantitative maps into 3D T1-weighted space using the b=0-derived registration parameters.
Statistical Analyses:
Unless otherwise specified, statistical analyses were conducted using Statistical Package for Social Sciences (SPSS 26.0) software (Armonk, NY: IBM Corp.). Normality of data was tested using the Kolmogorov-Smirnov test along with inspection of histograms and Q-Q plots.
To assess for longitudinal associations between changes in MoCA scores and FW, we used partial correlations, controlling for age, sex and education, separately in PD patients and HCs. Specifically, we investigated the relationship between 1-year changes in imaging parameters and MoCA changes from baseline to year 1 and baseline to year 3. Bootstrapping with 1000 samples was performed to obtain 95% confidence intervals for each correlation coefficient. With Waikato Environment for Knowledge Analysis (Weka version 3.8.1) software (University of Waikato, New Zealand), we also performed cross-validation using FW changes between baseline and year 1 to predict changes in MoCA score from baseline to year 1 and baseline to year 3. Linear regression models were constructed. We ran the analysis with 10 folds and report the results in terms of correlation coefficients (CC) and root mean squared error (RMSE). Analysis of covariance (ANCOVA) was run to compare baseline imaging parameters between PD-NC (MoCA ≥26) patients at baseline who converted to Level 1 MCI at years 1 and 3 versus those that did not. As in the partial correlation analyses, age, sex and education were controlled for in addition to FreeSurfer-derived estimated total intracranial volume for volumetric measures. Baseline imaging parameters were compared between PD patients having MoCA scores ≥26 and those with scores ≤25 using ANCOVA models with the same covariates as described previously. For the HC analyses, HCs were compared to all PD patients as well as just PD with MoCA ≤25 at baseline using ANCOVA.
A linear mixed model (LMM) was analyzed to assess the longitudinal effect on MoCA (dependent variable) score changes over time (considered as a categorical variable due to pre-planned time points as per the study protocol) comparing PD patients versus HCs (“group x time” interaction effect) while controlling for age, sex and education. We then assessed the effect of FW on MoCA over time independent of subject groupings. In this analysis, we were primarily interested in the “FW x time” interaction effects. LMMs were also used to assess the longitudinal effect on FW (dependent variable) in our four primary ROIs (nbM, thalamic DMN, thalamic AN and hippocampus) over time in both groups while controlling for age and sex. We were primarily interested in the “group x time” interaction effects. Finally, we constructed four separate LMMs to investigate the longitudinal effect on MoCA (dependent variable) for each of the four ROIs while controlling for age, sex and education. Our primary interest in these models was the “group x time x FW” interaction effect.
The primary analyses investigating FW associations within the nbM, thalamic DMN, thalamic AN and hippocampus were corrected for false discovery rate using the Benjamini-Hochberg procedure (q-value).37 A q-value<0.05 was considered to be significant. Exploratory analyses in the PD group also assessed conventional DTI and FW-corrected DTI (DTI-t) measures as well as volume for the same regions as in the primary analyses in addition to the whole thalamus and EC. Exploratory analyses were not corrected for multiple comparisons and were expressed as p-values. The robustness of exploratory assessments was determined by comparing the numerical partial correlation coefficient strength or partial eta squared effect size, as appropriate. A partial eta squared effect size of 0.01 and 0.05 were considered small and medium effect sizes, respectively.
Results:
Participant demographics are detailed in Table 1.
Table 1:
Healthy control and Parkinson’s disease patients’ characteristics
| Healthy Controls | Parkinson’s disease | |||||
|---|---|---|---|---|---|---|
| Baseline to Year 1 (n=58) |
Baseline to Year 1 (n=130) |
Baseline to Year 3 (n=123) |
||||
| Baseline | Year 1 | Baseline | Year 1 | Baseline | Year 3 | |
| Age (years) | 60.7 (10.4) | -- | 60.3 (9.3) | -- | 60.4 (9.2) | -- |
| No. male sex (%) | 20 (34.5%) | 20 (34.5%) | 44 (33.8%) | 44 (33.8%) | 44 (35.8%) | 44 (35.8%) |
| Disease duration (months) | -- | -- | 6.8 (7.1) | -- | 6.9 (7.2) | -- |
| No. on PD medication (%) | -- | -- | 0 | 91 (70.0%) | 0 | 114 (92.7%) |
| Total MDS-UPDRS | -- | -- | 31.5 (13.6) | 37.5 (16.9) | 30.8 (13.5) | 41.4 (21.1) |
| MDS-UPDRS (part III) | -- | -- | 20.8 (9.0) | 21.9 (10.8) | 20.6 (9.1) | 23.3 (13.1) |
| MoCA | 28.2 (1.1) | 27.4 (2.1) | 27.5 (2.0) | 26.9 (2.8) | 27.6 (2.0) | 26.7 (3.2) |
| No. MoCA≤25 (%) | 0 (0.0%) | 9 (15.5%) | 23 (17.7%) | 35 (26.9 %) | 20 (16.3%) | 36 (29.3%) |
MDS-UPDRS: Movement Disorder Society-Unified Parkinson’s Disease Rating Scale, MoCA: Montreal Cognitive Assessment, PD: Parkinson’s disease. Values are mean (SD) unless otherwise noted.
Imaging comparisons between PD patients and HCs:
Of the 4 ROIs in the primary analyses, only thalamic DMN FW significantly increased over 1 year in PD patients compared to HCs (0.02 ± 0.04 vs. 0.00 ± 0.02, respectively, q=0.016). There were no significant imaging differences at baseline between HCs and PD even when limiting the PD patient group to those with MoCA ≤25.
Associations with cognition in PD patients:
Of 4 ROIs included in the primary analyses, only thalamic DMN FW changes over 1-year showed significant negative correlations with both short-term (1-year partial correlation: −0.222, q=0.040) and longer-term (3-year partial correlation: −0.229, q=0.040) changes in MoCA scores (Table 2a) in early PD. NbM FW changes over 1 year negatively correlated with MoCA changes over 3 years (partial correlation: −0.222, q=0.040) but not over 1 year (partial correlation: −0.151, q=0.120, Table 2a). Only baseline hippocampal FW was associated with cognitive impairment at 3 years (q=0.040, Table 2b). Only baseline nbM FW distinguished PD-NC from PD-CI (q=0.008, Table 2c).
Table 2:
Primary Outcome Results of Associations between Free Water Assessments and Cognition in PD
| A) Longitudinal correlations between one-year changes in free water and changes in MoCA scores over 1 and 3 years. |
| ROI | MoCA change: Baseline to Year 1 |
MoCA change: Baseline to Year 3 |
||||
|---|---|---|---|---|---|---|
| PD patients (n = 130) |
PD patients (n = 123) |
|||||
| Correlation [95% CI] |
p-value | q-value | Correlation [95% CI] |
p-value | q-value | |
| nbM | −0.151 [−0.289, 0.038] |
0.090 | 0.120 |
−0.222
[−0.367, 0.026] |
0.015 | 0.040 |
| DMN-T |
−0.222
[−0.416, −0.027] |
0.013 | 0.040 |
−0.229
[−0.386, −0.067] |
0.013 | 0.040 |
| AN-T | −0.114 [−0.297, 0.060] |
0.204 | 0.233 | −0.178 [−0.352, −0.005] |
0.054 | 0.086 |
| Hipp | −0.098 [−0.293, 0.105] |
0.275 | 0.275 | −0.199 [−0.348, −0.022] |
0.030 | 0.060 |
| B) Baseline free water’s association with conversion to Level I Mild Cognitive Impairment in PD at Years 1 and 3. |
| ROI | Year 1 | Year 3 | ||||||
|---|---|---|---|---|---|---|---|---|
| Convert (n=22) | No-Convert (n=91) | p-value | q-value | Convert (n=19) | No-Convert (n=92) | p-value | q-value | |
| nbM | 0.233 (0.036) |
0.226 (0.027) |
0.410 | 0.656 | 0.239 (0.037) |
0.225 (0.026) |
0.186 | 0.496 |
| DMN-T | 0.273 (0.031) |
0.260 (0.040) |
0.643 | 0.735 | 0.283 (0.038) |
0.261 (0.038) |
0.108 | 0.432 |
| AN-T | 0.314 (0.069) |
0.315 (0.060) |
0.304 | 0.608 | 0.336 (0.061) |
0.313 (0.062) |
0.515 | 0.687 |
| Hipp | 0.350 (0.037) |
0.358 (0.032) |
0.929 | 0.929 |
0.380
(0.034) |
0.348
(0.035) |
0.005 | 0.040 |
| C) Baseline free water in PD with and without cognitive impairment |
| ROI | MoCA ≤25 (n=25) |
MoCA ≥26 (n=122) |
p-value | q-value |
|---|---|---|---|---|
| nbM |
0.264
(0.086) |
0.228
(0.029) |
0.002 | 0.008 |
| DMN-T | 0.274 (0.027) |
0.264 (0.038) |
0.820 | 0.877 |
| AN-T | 0.322 (0.067) |
0.317 (0.061) |
0.349 | 0.698 |
| Hipp | 0.364 (0.040) |
0.354 (0.038) |
0.877 | 0.877 |
Group-wise comparisons were analyzed using A) partial correlations or B) and C) using a general linear model correcting for age, sex and education; B) only included patients with MoCA≥26 at baseline; CI-confidence interval; nbM-nucleus basalis of Meynert; DMN-T- thalamic dorsomedial nucleus; AN-T- thalamic anterior nucleus; Hipp-hippocampus; MoCA–Montreal Cognitive Assessment. Q-values are p-values corrected for false discovery rate from multiple comparisons. Values are mean (SD) unless otherwise noted.
Longitudinal comparisons between PD and HCs to assess for distinguishing variables:
Using linear mixed models with MoCA as the dependent variable, the group x time interaction effect was not significant (p=0.571, F=0.544, df=436.0). We then assessed the changes over time in the four ROIs with respect to time and MoCA as the dependent variable, independent of the group. In this regard, the FW x time interaction effect was significant for thalamic AN (p=0.046, F=2.710, df=261) but not for hippocampus (p=0.081, F=2.269, df=299), thalamic DMN (p=0.112, F=2.016, df=276) nor nbM (p=0.164, F=1.720, df=201).
When FW was the dependent variable, the group x time interaction effects were as follows: nbM (p=0.065, F=3.439, df=291); thalamic DMN (p=0.048, F=2.703, df=151); thalamic AN (p=0.726, F=0.414, df=197) and hippocampus (p=0.686, F=0.498, df=165). When MoCA was set to the dependent variable and FW measures were included, the group x time x FW interaction effects were as follows: nbM (p=0.223, F=1.381, df=230); thalamic DMN (p=0.020, F=2.187, df=227), thalamic AN (p=0.025 F=2.753); df=207) and hippocampus (p=0.024, F=3.732, df=196).
Cross-fold validation of associations between FW and MoCA scores in PD:
Considering MoCA score changes from baseline to year 3, individual models including FW changes from baseline to year 1 were as follows: nbM – CC = 0.208, RMSE = 2.863 ; DMN – CC = 0.283, RMSE = 2.804 ; AN – CC = 0.235, RMSE = 2.847; Hipp – CC = 0.217, RMSE = 2.860. When including all four ROIs as predictors, the final model included nbM and DMN, although results were not stronger than DMN alone (CC = 0.273, RMSE = 2.821).
Exploratory outcomes:
For the exploratory comparisons, FW was the most robust assessment modality for all outcomes based on partial correlation strength (Table 3a) or partial eta squared effect size (Table 3b and 3c). There were several occasions when a DTI assessment showed an effect with a p-value<0.05 while the corresponding DTI-t did not (Table 3a, b, c) indicating that the DTI effect was being driven by FW contamination.
Table 3:
Exploratory Comparisons of Diffusion-Based and Volumetric Assessments’ Associations with Cognitive Outcomes
| A) Longitudinal correlations between one-year changes in imaging assessments and changes in MoCA scores over 1 and 3 years. |
| Measure | ROI | MoCA change: Baseline to Year 1 (n=130) | MoCA change: Baseline to Year 3 (n=123) | ||
|---|---|---|---|---|---|
| Correlation | p-value | Correlation | p-value | ||
| MD | nbM | −0.034 | 0.702 | −0.135 | 0.143 |
| Thalm | −0.156 | 0.081 | −0.168 | 0.070 | |
| DMN-T | −0.215 | 0.016 | −0.265 | 0.004 | |
| AN-T | −0.068 | 0.448 | −0.112 | 0.227 | |
| Hipp | −0.151 | 0.092 | −0.215 | 0.020 | |
| EC | −0.065 | 0.469 | −0.174 | 0.059 | |
| MD-t | nbM | 0.094 | 0.292 | 0.095 | 0.303 |
| Thalm | −0.060 | 0.503 | 0.031 | 0.738 | |
| DMN-T | −0.068 | 0.453 | −0.102 | 0.273 | |
| AN-T | 0.035 | 0.702 | 0.054 | 0.565 | |
| Hipp | −0.126 | 0.161 | −0.056 | 0.546 | |
| EC | 0.105 | 0.243 | 0.012 | 0.895 | |
| aD | nbM | −0.031 | 0.729 | −0.097 | 0.293 |
| Thalm | −0.178 | 0.047 | −0.175 | 0.058 | |
| DMN-T | −0.234 | 0.009 | −0.303 | 0.001 | |
| AN-T | −0.062 | 0.495 | −0.077 | 0.408 | |
| Hipp | −0.139 | 0.123 | −0.220 | 0.017 | |
| EC | −0.069 | 0.448 | −0.163 | 0.078 | |
| aD-t | nbM | 0.049 | 0.584 | 0.000 | 0.999 |
| Thalm | −0.109 | 0.228 | 0.006 | 0.951 | |
| DMN-T | −0.095 | 0.294 | −0.145 | 0.116 | |
| AN-T | −.003 | 0.972 | 0.058 | 0.532 | |
| Hipp | −0.120 | 0.184 | −0.165 | 0.074 | |
| EC | 0.003 | 0.977 | −0.014 | 0.878 | |
| rD | nbM | −0.033 | 0.712 | −0.143 | 0.120 |
| Thalm | −0.143 | 0.111 | −0.161 | 0.081 | |
| DMN-T | −0.202 | 0.024 | −0.237 | 0.010 | |
| AN-T | −0.068 | 0.454 | −0.128 | 0.168 | |
| Hipp | −0.153 | 0.088 | −0.205 | 0.026 | |
| EC | −0.063 | 0.485 | −0.178 | 0.053 | |
| rD-t | nbM | 0.049 | 0.584 | 0.000 | 0.999 |
| Thalm | −0.019 | 0.832 | 0.045 | 0.631 | |
| DMN-T | −0.029 | 0.744 | −0.042 | 0.649 | |
| AN-T | 0.065 | 0.469 | 0.041 | 0.657 | |
| Hipp | −.073 | 0.418 | 0.029 | 0.753 | |
| EC | 0.146 | 0.105 | 0.029 | 0.758 | |
| FW | nbM | −0.151 | 0.090 | −0.222 | 0.015 |
| Thalm | −0.175 | 0.050 | −0.199 | 0.031 | |
| DMN-T | −0.222 | 0.013 | −0.229 | 0.013 | |
| AN-T | −0.114 | 0.204 | −0.178 | 0.054 | |
| Hipp | −0.098 | 0.275 | −0.199 | 0.030 | |
| EC | −0.088 | 0.328 | −0.198 | 0.032 | |
| Volume | nbM | −0.019 | 0.829 | −0.035 | 0.706 |
| Thalm | 0.015 | 0.865 | 0.094 | 0.312 | |
| DMN-T | 0.011 | 0.899 | 0.136 | 0.143 | |
| AN-T | −0.013 | 0.885 | −0.059 | 0.528 | |
| Hipp | 0.025 | 0.781 | −0.015 | 0.875 | |
| EC | −0.058 | 0.521 | 0.006 | 0.950 | |
| B) Baseline imaging assessments’ associations with conversion to Level I Mild Cognitive Impairment at Years 1 and 3. |
| Measure | ROI | Year 1 | Year 3 | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Convert (n=22) | No-Convert (n=91) | p-value | Partial eta2 | Convert (n=19) | No-Convert (n=92) | p-value | Partial eta2 | ||
| MD | nbM | 0.818 (0.061) | 0.804 (0.049) | 0.252 | 0.012 | 0.821 (0.064) | 0.804 (0.048) | 0.514 | 0.004 |
| Thalm | 0.846 (0.049) | 0.836 (0.64) | 0.983 | 0.000 | 0.862 (0.064) | 0.837 (0.061) | 0.544 | 0.004 | |
| DMN-T | 0.897 (0.060) | 0.888 (0.076) | 0.759 | 0.001 | 0.921 (0.082) | 0.886 (0.072) | 0.302 | 0.010 | |
| AN-T | 0.949 (0.123) | 0.966 (0.114) | 0.194 | 0.016 | 0.983 (0.117) | 0.964 (0.117) | 0.923 | 0.000 | |
| Hipp | 1.061 (0.064) | 1.048 (0.070) | 0.750 | 0.001 | 1.102 (0.064) | 1.044 (0.068) | 0.005 | 0.071 | |
| EC | 1.050 (0.111) | 1.0845 (0.105) | 0.153 | 0.019 | 1.110 (0.123) | 1.077 (0.106) | 0.415 | 0.006 | |
| MD-t | nbM | 0.564 (0.013) | 0.561 (0.014) | 0.230 | 0.013 | 0.563 (0.014) | 0.561 (0.014) | 1.000 | 0.000 |
| Thalm | 0.557 (0.017) | 0.558 (0.013) | 0.833 | 0.000 | 0.555 (0.016) | 0.560 (0.013) | 0.103 | 0.025 | |
| DMN-T | 0.544 (0.020) | 0.547 (0.015) | 0.757 | 0.001 | 0.539 (0.019) | 0.548 (0.015) | 0.046 | 0.037 | |
| AN-T | 0.568 (0.019) | 0.577 (0.023) | 0.211 | 0.015 | 0.565 (0.017) | 0.577 (0.024) | 0.023 | 0.048 | |
| Hipp | 0.577 (0.009) | 0.581 (0.008) | 0.489 | 0.004 | 0.576 (0.006) | 0.581 (0.008) | 0.088 | 0.028 | |
| EC | 0.571 (0.014) | 0.577 (0.013) | 0.195 | 0.016 | 0.576 (0.014) | 0.572 (0.012) | 0.421 | 0.006 | |
| aD | nbM | 1.120 (0.054) | 1.120 (0.056) | 0.595 | 0.003 | 1.120 (0.063) | 1.119 (0.055) | 0.848 | 0.000 |
| Thalm | 1.113 (0.053) | 1.115 (0.068) | 0.997 | 0.000 | 1.142 (0.064) | 1.117 (0.066) | 0.606 | 0.003 | |
| DMN-T | 1.168 (0.067) | 1.159 (0.084) | 0.546 | 0.003 | 1.196 (0.082) | 1.157 (0.081) | 0.290 | 0.011 | |
| AN-T | 1.209 (0.122) | 1.234 (0.122) | 0.136 | 0.021 | 1.251 (0.121) | 1.230 (0.123) | 0.983 | 0.000 | |
| Hipp | 1.239 (0.058) | 1.1226 (0.072) | 0.781 | 0.001 | 1.280 (0.066) | 1.222 (0.069) | 0.007 | 0.067 | |
| EC | 1.213 (0.112) | 1.243 (0.109) | 0.242 | 0.013 | 1.271 (0.130) | 1.236 (0.108) | 0.341 | 0.009 | |
| aD-t | nbM | 0.945 (0.020) | 0.944 (0.031) | 0.858 | 0.000 | 0.940 (0.030) | 0.945 (0.032) | 0.213 | 0.015 |
| Thalm | 0.850 (0.027) | 0.846 (0.019) | 0.903 | 0.000 | 0.845 (0.025) | 0.848 (0.020) | 0.095 | 0.026 | |
| DMN-T | 0.823 (0.039) | 0.823 (0.028) | 0.661 | 0.002 | 0.820 (0.034) | 0.824 (0.029) | 0.360 | 0.008 | |
| AN-T | 0.847 (0.032) | 0.867 (0.050) | 0.098 | 0.025 | 0.853 (0.034) | 0.865 (0.050) | 0.223 | 0.014 | |
| Hipp | 0.764 (0.024) | 0.768 (0.016) | 0.786 | 0.001 | 0.764 (0.027) | 0.768 (0.016) | 0.773 | 0.001 | |
| EC | 0.735 (0.029) | 0.739 (0.025) | 0.928 | 0.000 | 0.740 (0.023) | 0.737 (0.026) | 0.481 | 0.005 | |
| rD | nbM | 0.627 (0.066) | 0.611 (0.049) | 0.167 | 0.018 | 0.632 (0.068) | 0.610 (0.049) | 0.292 | 0.010 |
| Thalm | 0.705 (0.049) | 0.696 (0.062) | 0.973 | 0.000 | 0.721 (0.065) | 0.698 (0.060) | 0.521 | 0.004 | |
| DMN-T | 0.762 (0.062) | 0.752 (0.076) | 0.875 | 0.000 | 0.784 (0.085) | 0.752 (0.071) | 0.349 | 0.008 | |
| AN-T | 0.820 (0.126) | 0.831 (0.114) | 0.259 | 0.012 | 0.850 (0.116) | 0.830 (0.117) | 0.900 | 0.000 | |
| Hipp | 0.972 (0.069) | 0.958 (0.070) | 0.740 | 0.001 | 1.014 (0.065) | 0.954 (0.068) | 0.006 | 0.070 | |
| EC | 0.969 (0.112) | 1.006 (0.103) | 0.120 | 0.022 | 1.029 (0.121) | 0.998 (0.106) | 0.462 | 0.005 | |
| rD-t | nbM | 0.374 (0.018) | 0.370 (0.016) | 0.091 | 0.026 | 0.375 (0.017) | 0.369 (0.016) | 0.204 | 0.015 |
| Thalm | 0.410 (0.016) | 0.414 (0.015) | 0.715 | 0.001 | 0.410 (0.016) | 0.415 (0.015) | 0.255 | 0.012 | |
| DMN-T | 0.404 (0.022) | 0.408 (0.020) | 0.977 | 0.000 | 0.399 (0.022) | 0.409 (0.020) | 0.074 | 0.030 | |
| AN-T | 0.429 (0.027) | 0.431 (0.026) | 0.879 | 0.000 | 0.420 (0.022) | 0.432 (0.026) | 0.056 | 0.034 | |
| Hipp | 0.484 (0.013) | 0.487 (0.010) | 0.637 | 0.002 | 0.483 (0.011) | 0.487 (0.011) | 0.137 | 0.021 | |
| EC | 0.489 (0.015) | 0.495 (0.015) | 0.103 | 0.025 | 0.488 (0.017) | 0.495 (0.015) | 0.104 | 0.025 | |
| FW | nbM | 0.233 (0.036) | 0.226 (0.027) | 0.410 | 0.006 | 0.239 (0.037) | 0.225 (0.026) | 0.186 | 0.016 |
| Thalm | 0.231 (0.025) | 0.222 (0.032) | 0.674 | 0.002 | 0.239 (0.031) | 0.223 (0.032) | 0.327 | 0.009 | |
| DMN-T | 0.273 (0.031) | 0.260 (0.039) | 0.643 | 0.002 | 0.283 (0.038) | 0.261 (0.038) | 0.108 | 0.024 | |
| AN-T | 0.314 (0.069) | 0.315 (0.060) | 0.304 | 0.010 | 0.336 (0.061) | 0.313 (0.062) | 0.515 | 0.004 | |
| Hipp | 0.350 (0.037) | 0.358 (0.032) | 0.929 | 0.000 | 0.380 (0.034) | 0.348 (0.035) | 0.005 | 0.074 | |
| EC | 0.353 (0.059) | 0.370 (0.054) | 0.149 | 0.019 | 0.384 (0.062) | 0.365 (0.054) | 0.340 | 0.009 | |
| Volume | nbM | 0.289 (0.104) | 0.302 (0.115) | 0.696 | 0.001 | 0.281 (0.100) | 0.307 (0.116) | 0.554 | 0.003 |
| Thalm | 11761 (1217) | 12242 (1288) | 0.147 | 0.020 | 11667 (1247) | 12206 (1245) | 0.104 | 0.025 | |
| DMN-T | 1808 (204) | 1924 (205) | 0.081 | 0.028 | 1810 (202) | 1909 (200) | 0.180 | 0.017 | |
| AN-T | 263 (42) | 274 (46) | 0.422 | 0.006 | 259 (42) | 275 (45) | 0.182 | 0.017 | |
| Hipp | 7017 (790) | 7115 (851) | 0.957 | 0.000 | 6732 (983) | 7153 (760) | 0.071 | 0.031 | |
| EC | 1898 (434) | 1916 (361) | 0.739 | 0.001 | 1859 (548) | 1929 (336) | 0.597 | 0.070 | |
| C) Baseline imaging assessments with and without cognitive impairment |
| Measure | ROI | MoCA ≤25 (n=25) | MoCA ≥26 (n=122) | p-value | Partial eta2 |
|---|---|---|---|---|---|
| MD | nbM | 0.844 (0.076) | 0.807 (0.050) | 0.019 | 0.038 |
| Thalm | 0.862 (0.056) | 0.840 (0.062) | 0.761 | 0.001 | |
| DMN-T | 0.904 (0.052) | 0.891 (0.073) | 0.623 | 0.002 | |
| AN-T | 0.975 (0.118) | 0.966 (0.114) | 0.454 | 0.004 | |
| Hipp | 1.075 (0.078) | 1.054 (0.071) | 0.940 | 0.001 | |
| EC | 1.129 (0.0873) | 1.082 (0.107) | 0.117 | 0.017 | |
| MD-t | nbM | 0.556 (0.0560 | 0.562 (0.014) | 0.169 | 0.013 |
| Thalm | 0.560 (0.019) | 0.558 (0.014) | 0.742 | 0.001 | |
| DMN-T | 0.548 (0.020) | 0.546 (0.016) | 0.412 | 0.005 | |
| AN-T | 0.570 (0.020) | 0.575 (0.023) | 0.318 | 0.007 | |
| Hipp | 0.578 (0.010) | 0.580 (0.008) | 0.673 | 0.001 | |
| EC | 0.571 (0.018) | 0.576 (0.013) | 0.281 | 0.008 | |
| aD | nbM | 1.230 (0.080) | 0.119 (0.055) | 0.027 | 0.034 |
| Thalm | 1.145 (0.062) | 1.120 (0.066) | 0.647 | 0.002 | |
| DMN-T | 1.174 (0.055) | 1.162 (0.081) | 0.475 | 0.004 | |
| AN-T | 1.124 (0.114) | 1.233 (0.119) | 0.577 | 0.002 | |
| Hipp | 1.252 (0.076) | 1.233 (0.072) | 0.945 | 0.000 | |
| EC | 1.288 (0.095) | 1.241 (0.111) | 0.119 | 0.017 | |
| aD-t | nbM | 0.935 (1.00) | 0.944 (0.031) | 0.215 | 0.011 |
| Thalm | 0.855 (0.028) | 0.847 (0.021) | 0.459 | 0.004 | |
| DMN-T | 0.827 (0.028) | 0.823 (0.030) | 0.842 | 0.000 | |
| AN-T | 0.862 (0.029) | 0.863 (0.047) | 0.786 | 0.001 | |
| Hipp | 0.764 (0.020) | 0.769 (0.018) | 0.825 | 0.000 | |
| EC | 0.737 (0.032) | 0.738 (0.025) | 0.940 | 0.000 | |
| rD | nbM | 0.651 (0.078) | 0.614 (0.052) | 0.024 | 0.036 |
| Thalm | 0.720 (0.054) | 0.700 (0.061) | 0.828 | 0.000 | |
| DMN-T | 0.769 (0.053) | 0.760 (0.073) | 0.716 | 0.001 | |
| AN-T | 0.840 (0.122) | 0.833 (0.115) | 0.414 | 0.005 | |
| Hipp | 0.987 (0.080) | 0.965 (0.072) | 0.939 | 0.000 | |
| EC | 1.049 (0.084) | 1.003 (0.107) | 0.120 | 0.017 | |
| rD-t | nbM | 0.367 (0.038) | 0.371 (0.016) | 0.268 | 0.009 |
| Thalm | 0.413 (0.019) | 0.414 (0.015) | 0.969 | 0.000 | |
| DMN-T | 0.409 (0.021) | 0.407 (0.020) | 0.378 | 0.006 | |
| AN-T | 0.424 (0.025) | 0.431 (0.025) | 0.284 | 0.008 | |
| Hipp | 0.485 (0.012) | 0.486 (0.011) | 0.763 | 0.001 | |
| EC | 0.488 (0.016) | 0.494 (0.015) | 0.103 | 0.019 | |
| FW | nbM | 0.264 (0.086) | 0.228 (0.029) | 0.002 | 0.067 |
| Thalm | 0.237 (0.028) | 0.225 (0.032) | 0.724 | 0.001 | |
| DMN-T | 0.274 (0.027) | 0.264 (0.038) | 0.820 | 0.000 | |
| AN-T | 0.322 (0.067) | 0.317 (0.061) | 0.349 | 0.006 | |
| Hipp | 0.364 (0.040) | 0.354 (0.038) | 0.877 | 0.000 | |
| EC | 0.387 (0.049) | 0.368 (0.055) | 0.278 | 0.008 | |
| Volume | nbM | 0.317 (0.095) | 0.300 (0.111) | 0.197 | 0.012 |
| Thalm | 11703 (819) | 12103 (1281) | 0.073 | 0.023 | |
| DMN-T | 1755 (203) | 1893 (208) | 0.005 | 0.054 | |
| AN-T | 253 (36) | 272 (46) | 0.027 | 0.034 | |
| Hipp | 6842 (521) | 7054 (839) | 0.281 | 0.008 | |
| EC | 1774 (216) | 1910 (366) | 0.054 | 0.026 |
Group-wise comparisons were analyzed using A) partial correlations or B) and C) using a general linear model correcting for age, sex and education; B) only included patients with MoCA≥26 at baseline; Partial eta2-Partial eta squared effect size; MD, aD, rD-mean, axial and radial diffusivity, respectively; t-free water corrected; FW-free water; ROI-region of interest; nbM-nucleus basalis of Meynert; Thalm-whole thalamus; DMN-T-dorsomedial thalamic nucleus; AN-T-anterior thalamic nucleus; Hipp-hippocampus; EC-entorhinal cortex; MoCA–Montreal Cognitive Assessment. Values are mean (SD) unless otherwise noted.
Discussion:
In early PD, we found 1-year longitudinal increases in thalamic DMN FW to significantly correlate with longitudinal decreases in MoCA scores over both 1 and 3-years. In addition, 1-year longitudinal increases in nbM FW correlated with longitudinal decreases in MoCA scores over 3-years but not over 1-year (Table 2a). Thus, in our primary analyses corrected for multiple comparisons, 1-year change in thalamic DMN FW was the only measure to show both short-term (1-year) and longer-term (3-year) significant correlations with changes in cognition in early PD. This is the first report of such findings and supports thalamic DMN FW as a promising cognition progression biomarker in early PD that may assist in identifying cognition protective therapies in clinical trials.
Longitudinal changes in DMN FW were found to be specific to PD and not due to age-related changes as no such changes were found in age-matched HCs. The linear mixed model analyses between PD and HC further supported this finding. Also, only FW changes in the DMN, but not the other 3 ROIs, showed longitudinal differences between PD and HCs. We provide a theory in the Supplemental Data file explaining these findings.
In our primary analyses, we also found elevated baseline hippocampal FW to be associated with cognitive impairment at 3 years (Table 2b) and elevated baseline nbM FW to distinguish PD-CI from PD-NC (Table 2c). It is not clear why distinct ROIs showed various sensitivities to these different cognitive outcomes. A recent PPMI study showed that widespread atrophy across a PD-specific network including many subcortical and cortical ROIs in one measure was highly sensitive for predicting both motor and cognitive impairment in early PD after 4.5 years.38 Such findings support PD pathology to be diffuse even at the time of diagnosis making it less likely for any single ROI to reflect several different clinical outcomes.
In our exploratory analyses, FW was consistently the most robust assessment modality for all cognitive outcomes (Table 3). In addition, the exploratory analyses revealed how DTI assessments that are not corrected for FW contamination can be misleading. There were several occasions of significant MD, axial diffusivity (aD) or radial diffusivity (rD) outcomes that were no longer significant after correcting for FW contamination. In all of these occasions, it can be seen that FW itself was driving the significant DTI findings. For example, increased MD, aD and rD in the nbM at baseline distinguished PD-CI from PD-NC; however, these discriminatory abilities were no longer apparent after correction for FW contamination (MD-t, aD-t and rD-t) while nbM FW strongly showed this distinction (Table 3c). On two occasions, a DTI-t value showed a p-value<0.05 while the corresponding uncorrected DTI did not (Table 3b). When examining the raw data for these 2 occasions (DMN-T and AN-T, MD-t vs. MD, Table 3b), it can be seen that relatively higher baseline values were associated with opposite cognitive outcomes at year 3 indicating strong contributions of FW contamination to these DTI assessments. These findings demonstrate the importance of correcting for FW contamination in DTI assessments and show FW itself to be a robust diffusion-based assessment for these outcomes.
Longitudinal increases in free water was previously shown to be more sensitive than conventional DTI or DTI-t for detecting longitudinal structural changes in the substantia nigra among three different PD cohorts including PPMI.18–20 Furthermore, posterior substantia nigra (pSN) FW has been shown to predict future motor abilities, inversely correlate with striatal dopaminergic terminal integrity and to not be significantly affected by acute dopaminergic medication administration in PD.18,20,24,39 Similar to the SN,18–20 we found longitudinal increases in DMN FW to be PD specific and not secondary to age-related changes. It is noteworthy that these longitudinal correlations, whether in the SN or DMN, have been inversely related to adverse clinical outcomes such as motor or cognitive dysfunction suggesting that FW reflects underlying tissue pathology associated with these PD symptoms.
Because clinical trials assessing a therapeutic’s disease-modifying actions typically enroll early PD patients, our findings utilizing the PPMI PD cohort showing 1-year increases in thalamic DMN FW to correlate with both 1 and 3-year declines in cognition are highly relevant. If a therapy was shown to reduce thalamic DMN FW accumulation in a clinical trial of 1-year duration in early PD, our current findings suggest that this effect would be expected to persist for at least 3 years. As the MoCA is known to be a sensitive longitudinal predictor of future PDD,28 longitudinal imaging correlates to the MoCA, such as thalamic DMN FW, may also reflect risk for future PDD. Thus, therapies shown to reduce longitudinal DMN FW accumulation may also potentially delay the onset of PDD, although further research is needed to confirm and extend our findings. Our findings in addition to the previous pSN FW findings18,20 in early PD support the use of MRI in disease-modifying clinical trials to assess longitudinal changes in both thalamic DMN and pSN FW. Such a design would enable objective assessment of a therapy’s effects on tissue integrity in these ROIs relevant for cognitive and motor functions that are known to progressively deteriorate in PD and lead to mounting disability.
We were unable to replicate previous findings of reduced nbM volume being associated with cognitive changes in PD from PPMI.8,9 Our greatly reduced sample size available for volumetric analyses compared to these previous studies likely contributed to our different results. Because we only performed volumetric analyses on patients who also had valid diffusion scans in order to allow direct comparisons among these various imaging assessments, this greatly limited the number of scans available for volumetric analyses. In addition, our PD cohort was recently confirmed by the PPMI Steering Committee to consist of only PD patients, which is unique to our study compared to previous PPMI studies8–10 and may also have contributed to our different results. Because this is the first study showing these findings, it is important for these results to be replicated in other early PD longitudinal cohorts.
The main strengths of this study are utilization of the large, early-stage PPMI PD cohort recently confirmed to be free of non-PD patients; use of advanced MRI mapping and thalamic segmentation techniques; and use of FW, DTI, DTI-t and volumetric assessments across multiple ROIs using the same set of images allowing for direct comparisons. One weakness of our study was including longitudinal follow-up of only 3 years. Although PPMI captured MRI and cognition data for 4 years; only fifty-eight, 4-year diffusion scans were available for analyses, which was too small of a sample to provide sufficient power to assess for the proposed outcomes at year 4. Although acute administration of dopaminergic medication was previously shown not to influence pSN FW,39 a potential medication effect cannot be excluded for the ROIs examined here considering that 70% and 93% of patients were receiving PD medications at years 1 and 3, respectively (Table 1). We did not examine lobar cortical thickness or caudate FW, which may have been weaknesses considering that diffuse cortical thinning and increased caudate FW have previously been associated with cognitive impairment in later-stage PD.24,40 Also, PPMI included only single-shell diffusion MRI data whereas multi-shell data would have made the FW estimation more robust.
Conclusion:
Thalamic DMN FW longitudinally correlates with cognitive decline in early PD and may represent a cognition progression biomarker that could assist in identifying cognition protective therapies if used in clinical trials. FW is a robust assessment modality for the cognitive outcomes evaluated.
Supplementary Material
Acknowledgements:
We thank all of the Parkinson’s disease patients and healthy controls who participated in PPMI.
Funding sources for the study: This study utilized data from the Parkinson’s Progression Markers Initiative, which is a public-private partnership funded by the Michael J. Fox Foundation for Parkinson’s Research and funding partners including Abbvie, Avid, Biogen, Bristol-Myers Squibb, Covance, GE Healthcare, Genentech, GlaxoSmithKline, Lilly, Lundbeek, Merck, Meso Scale Discovery, Pfizer, Piramal, Roche, Servier, Teva, UCB, and Golub Capital. An additional funding source was NIH P41EB015902.
Authors’ financial disclosures related to the research reported in the manuscript: None.
Author’s Roles:
Thomas Guttuso, Jr.: 1. Research project: A. Conception, B. Organization, C. Execution; 2. Statistical Analysis: A. Design, C. Review and Critique; 3. Manuscript Preparation: A. Writing of the first draft, B. Review and Critique.
Daniel Sirica: 1. Research project: C. Execution; 3. Manuscript Preparation: B. Review and Critique.
Duygu Tosun: 1. Research project: C. Execution; 2. Statistical Analysis: C. Review and Critique; 3. Manuscript Preparation: B. Review and Critique.
Robert Zivadinov: 1. Research project: C. Execution; 2. Statistical Analysis: C. Review and Critique; 3. Manuscript Preparation: B. Review and Critique.
Ofer Pasternak: 1. Research project: C. Execution; 2. Statistical Analysis: C. Review and Critique; 3. Manuscript Preparation: B. Review and Critique.
Daniel Weintraub: 1. Research project: C. Execution; 2. Statistical Analysis: C. Review and Critique; 3. Manuscript Preparation: B. Review and Critique.
Francesca Baglio: 1. Research project: C. Execution; 2. Statistical Analysis: C. Review and Critique; 3. Manuscript Preparation: B. Review and Critique.
Niels Bergsland: 1. Research project: B. Organization, C. Execution; 2. Statistical Analysis: A. Design, B. Execution, C. Review and Critique; 3. Manuscript Preparation: B. Review and Critique.
Footnotes
Financial Disclosures for previous 12 months:
Thomas Guttuso, Jr.: President of e3 Pharmaceuticals, Inc. Support from Pharma 2B, Inc. for clinical trial patient enrollment.
Daniel Sirica: None.
Duygu Tosun: None.
Robert Zivadinov: Received personal compensation from Bristol Myers Squibb, EMD Serono, Sanofi, Novartis and Keystone Heart for speaking and consultant fees. He received financial support for research activities from Bristol Myers Squibb, Sanofi, Novartis, Keystone Heart, V-WAVE Medical, Mapi Pharma and Protembis.
Ofer Pasternak: None.
Daniel Weintraub: Dr. Weintraub has received research funding or support from Michael J. Fox Foundation for Parkinson’s Research, Alzheimer’s Therapeutic Research Initiative (ATRI), Alzheimer’s Disease Cooperative Study (ADCS), the International Parkinson and Movement Disorder Society (IPMDS) the National Institute on Aging (NIA) and the U.S. Department of Veterans Affairs; honoraria for consultancy from Acadia, Aptinyx, CHDI Foundation, Clintrex LLC (Otsuka), Eisai, Great Lake Neurotechnologies, Janssen, Sage, Scion, Signant Health, Sunovion and Vanda; and license fee payments from the University of Pennsylvania for the QUIP and QUIP-RS.
Francesca Baglio: None.
Niels Bergsland: None.
References:
- 1.Aarsland D, Kurz MW. The epidemiology of dementia associated with Parkinson disease. J Neurol Sci 2010;289:18–22. [DOI] [PubMed] [Google Scholar]
- 2.Hely MA, Reid WG, Adena MA, Halliday GM, Morris JG. The Sydney multicenter study of Parkinson’s disease: the inevitability of dementia at 20 years. Mov Disord 2008;23:837–44. [DOI] [PubMed] [Google Scholar]
- 3.Aarsland D, Larsen JP, Tandberg E, Laake K. Predictors of nursing home placement in Parkinson’s disease: a population-based, prospective study. J Am Geriatr Soc 2000;48:938–42. [DOI] [PubMed] [Google Scholar]
- 4.Schrag A, Jahanshahi M, Quinn N. What contributes to quality of life in patients with Parkinson’s disease? J Neurol Neurosurg Psychiatry 2000;69:308–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Lawson RA, Yarnall AJ, Duncan GW, et al. Cognitive decline and quality of life in incident Parkinson’s disease: The role of attention. Parkinsonism Relat Disord 2016;27:47–53. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Forsaa EB, Larsen JP, Wentzel-Larsen T, Alves G. What predicts mortality in Parkinson disease?: a prospective population-based long-term study. Neurology 2010;75:1270–6. [DOI] [PubMed] [Google Scholar]
- 7.Coelho M, Marti MJ, Sampaio C, et al. Dementia and severity of parkinsonism determines the handicap of patients in late-stage Parkinson’s disease: the Barcelona-Lisbon cohort. Eur J Neurol 2015;22:305–12. [DOI] [PubMed] [Google Scholar]
- 8.Ray NJ, Bradburn S, Murgatroyd C, et al. In vivo cholinergic basal forebrain atrophy predicts cognitive decline in de novo Parkinson’s disease. Brain 2018;141:165–76. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Schulz J, Pagano G, Fernandez Bonfante JA, Wilson H, Politis M. Nucleus basalis of Meynert degeneration precedes and predicts cognitive impairment in Parkinson’s disease. Brain 2018;141:1501–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Zhang Y, Wu IW, Tosun D, Foster E, Schuff N, Parkinson’s Progression Markers I. Progression of Regional Microstructural Degeneration in Parkinson’s Disease: A Multicenter Diffusion Tensor Imaging Study. PLoS One 2016;11:e0165540. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Kandiah N, Zainal NH, Narasimhalu K, et al. Hippocampal volume and white matter disease in the prediction of dementia in Parkinson’s disease. Parkinsonism Relat Disord 2014;20:1203–8. [DOI] [PubMed] [Google Scholar]
- 12.Mak E, Su L, Williams GB, et al. Baseline and longitudinal grey matter changes in newly diagnosed Parkinson’s disease: ICICLE-PD study. Brain 2015;138:2974–86. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Hanyu H, Asano T, Sakurai H, Tanaka Y, Takasaki M, Abe K. MR analysis of the substantia innominata in normal aging, Alzheimer disease, and other types of dementia. AJNR Am J Neuroradiol 2002;23:27–32. [PMC free article] [PubMed] [Google Scholar]
- 14.Oikawa H, Sasaki M, Ehara S, Abe T. Substantia innominata: MR findings in Parkinson’s disease. Neuroradiology 2004;46:817–21. [DOI] [PubMed] [Google Scholar]
- 15.Choi SH, Jung TM, Lee JE, Lee SK, Sohn YH, Lee PH. Volumetric analysis of the substantia innominata in patients with Parkinson’s disease according to cognitive status. Neurobiol Aging 2012;33:1265–72. [DOI] [PubMed] [Google Scholar]
- 16.Van der Werf YD, Jolles J, Witter MP, Uylings HB. Contributions of thalamic nuclei to declarative memory functioning. Cortex; a journal devoted to the study of the nervous system and behavior 2003;39:1047–62. [DOI] [PubMed] [Google Scholar]
- 17.Fama R, Sullivan EV. Thalamic structures and associated cognitive functions: Relations with age and aging. Neurosci Biobehav Rev 2015;54:29–37. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Burciu RG, Ofori E, Archer DB, et al. Progression marker of Parkinson’s disease: A 4-year multisite imaging study. Brain 2017;140:2183–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Guttuso T Jr., Bergsland N, Hagemeier J, Lichter DG, Pasternak O, Zivadinov R. Substantia Nigra Free Water Increases Longitudinally in Parkinson Disease. AJNR Am J Neuroradiol 2018;39:479–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Ofori E, Pasternak O, Planetta PJ, et al. Longitudinal changes in free-water within the substantia nigra of Parkinson’s disease. Brain 2015;138:2322–31. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Pasternak O, Sochen N, Gur Y, Intrator N, Assaf Y. Free water elimination and mapping from diffusion MRI. Magn Reson Med 2009;62:717–30. [DOI] [PubMed] [Google Scholar]
- 22.Pasternak O, Shenton ME, Westin CF. Estimation of extracellular volume from regularized multi-shell diffusion MRI. Med Image Comput Comput Assist Interv 2012;15:305–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Febo M, Perez PD, Ceballos-Diaz C, et al. Diffusion magnetic resonance imaging-derived free water detects neurodegenerative pattern induced by interferon-gamma. Brain structure & function 2020;225:427–39. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Yang J, Archer DB, Burciu RG, et al. Multimodal dopaminergic and free-water imaging in Parkinson’s disease. Parkinsonism Relat Disord 2019;62:10–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Wang Y, Wang Q, Haldar JP, et al. Quantification of increased cellularity during inflammatory demyelination. Brain 2011;134:3590–601. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Caspell-Garcia C, Simuni T, Tosun-Turgut D, et al. Multiple modality biomarker prediction of cognitive impairment in prospectively followed de novo Parkinson disease. PLoS One 2017;12:e0175674. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Litvan I, Goldman JG, Troster AI, et al. Diagnostic criteria for mild cognitive impairment in Parkinson’s disease: Movement Disorder Society Task Force guidelines. Mov Disord 2012;27:349–56. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Kim HM, Nazor C, Zabetian CP, et al. Prediction of cognitive progression in Parkinson’s disease using three cognitive screening measures. Clin Park Relat Disord 2019;1:91–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Hoops S, Nazem S, Siderowf AD, et al. Validity of the MoCA and MMSE in the detection of MCI and dementia in Parkinson disease. Neurology 2009;73:1738–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Nazem S, Siderowf AD, Duda JE, et al. Montreal cognitive assessment performance in patients with Parkinson’s disease with “normal” global cognition according to mini-mental state examination score. J Am Geriatr Soc 2009;57:304–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Smith SM, De Stefano N, Jenkinson M, Matthews PM. Normalized accurate measurement of longitudinal brain change. J Comput Assist Tomogr 2001;25:466–75. [DOI] [PubMed] [Google Scholar]
- 32.Eickhoff SB, Stephan KE, Mohlberg H, et al. A new SPM toolbox for combining probabilistic cytoarchitectonic maps and functional imaging data. Neuroimage 2005;25:1325–35. [DOI] [PubMed] [Google Scholar]
- 33.Klein A, Andersson J, Ardekani BA, et al. Evaluation of 14 nonlinear deformation algorithms applied to human brain MRI registration. Neuroimage 2009;46:786–802. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Fischl B, van der Kouwe A, Destrieux C, et al. Automatically parcellating the human cerebral cortex. Cerebral cortex 2004;14:11–22. [DOI] [PubMed] [Google Scholar]
- 35.Iglesias JE, Insausti R, Lerma-Usabiaga G, et al. A probabilistic atlas of the human thalamic nuclei combining ex vivo MRI and histology. Neuroimage 2018;183:314–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Greve DN, Fischl B. Accurate and robust brain image alignment using boundary-based registration. Neuroimage 2009;48:63–72. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J Roy Stat Soc Ser B (Methodological) 1995;57:289–300. [Google Scholar]
- 38.Zeighami Y, Fereshtehnejad SM, Dadar M, Collins DL, Postuma RB, Dagher A. Assessment of a prognostic MRI biomarker in early de novo Parkinson’s disease. NeuroImage Clinical 2019;24:101986. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Chung JW, Burciu RG, Ofori E, et al. Parkinson’s disease diffusion MRI is not affected by acute antiparkinsonian medication. NeuroImage Clinical 2017;14:417–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Hanganu A, Bedetti C, Degroot C, et al. Mild cognitive impairment is linked with faster rate of cortical thinning in patients with Parkinson’s disease longitudinally. Brain 2014;137:1120–9. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.