Article Text
PDF
Abstract
Objective A major contributor to dementia in Parkinson disease (PD) is degeneration of the cholinergic basal forebrain. This study determined whether cholinergic nucleus 4 (Ch4) density is associated with cognition in early and more advanced PD.
Methods We analysed brain MRIs and neuropsychological test scores for 228 newly diagnosed PD participants from the Parkinson’s Progression Markers Initiative (PPMI), 101 healthy controls from the PPMI and 125 more advanced PD patients from a local retrospective cohort. Cholinergic basal forebrain nuclei densities were determined by applying probabilistic maps to MPRAGE T1 sequences processed using voxel-based morphometry methods. Relationships between grey matter densities and cognitive scores were analysed using correlations and linear regression models.
Results In more advanced PD, greater Ch4 density was associated with Montreal Cognitive Assessment (MoCA) score (β=14.2; 95% CI=1.5 to 27.0; p=0.03), attention domain z-score (β=3.2; 95% CI=0.8 to 5.5; p=0.008) and visuospatial domain z-score (β=7.9; 95% CI=2.0 to 13.8; p=0.009). In the PPMI PD cohort, higher Ch4 was associated with higher scores on MoCA (β=9.2; 95% CI=1.9 to 16.5; p=0.01), Judgement of Line Orientation (β=20.4; 95% CI=13.8 to 27.0; p<0.001), Letter Number Sequencing (β=16.5; 95% CI=9.5 to 23.4; p<0.001) and Symbol Digit Modalities Test (β=41.8; 95% CI=18.7 to 65.0; p<0.001). These same relationships were observed in 97 PPMI PD participants at 4 years. There were no significant associations between Ch4 density and cognitive outcomes in healthy controls.
Conclusion In de novo and more advanced PD, lower Ch4 density is associated with impaired global cognition, attention and visuospatial function.
- Parkinson disease
- basal forebrain
- cognition
- attention
Statistics from Altmetric.com
Introduction
As Parkinson disease (PD) progresses, cognitive impairment and dementia become a major source of morbidity. Dementia in PD (PDD) is associated with greater caregiver burden, decreased quality of life,1 longer hospitalisations2 and increased mortality.3 A major contributor to PDD is degeneration of the cholinergic neurons of the nucleus basalis of Meynert (NBM).4–6
The NBM cannot be directly measured using MRI, so two methods have been applied to assess the volume of this region. Some studies showed that reduced volume of the substantia innominata, a manually defined region containing the NBM and other structures, was associated with cognitive impairment in PD.7–9 An alternative method is to calculate grey matter density within cholinergic nucleus 4 (Ch4), which includes the cholinergic neurons of the NBM, by using probabilistic maps of the basal forebrain nuclei.10 Two recent studies found that in PD reduced Ch4 volume at baseline was associated with greater decline in Montreal Cognitive Assessment (MoCA) score over 5 years11 and greater risk of developing mild cognitive impairment or dementia.11 12
In this study, we applied probabilistic atlases to T1 magnetization-prepared rapid gradient-echo (MPRAGE) MRI sequences to calculate cholinergic basal forebrain nuclei and neocortical densities. The primary objective was to determine whether Ch4 density is associated with cognitive function (1) in newly diagnosed PD participants and healthy controls and (2) in more advanced PD patients undergoing clinical evaluations prior to neurosurgical procedures. To assess whether the relationships between Ch4 density and cognition were specific to Ch4, we also evaluated the relationships between the regional densities for cholinergic nucleus 1, 2 and 3 (Ch123) and neocortex and cognitive outcomes.
Methods
Presurgical PD cohort
This study analysed data from a retrospective cohort of PD patients who underwent a first neurosurgical procedure to treat motor symptoms between 01 January 2010 and 01 July 2016 at the University of Virginia. The procedures included deep brain stimulation, radiofrequency ablation or focused ultrasound lesioning. Only subjects who completed a comprehensive neuropsychological evaluation and brain MRI as part of the presurgical evaluation were considered for this study. The diagnosis of PD was confirmed by a movement disorders neurologist as part of the presurgical evaluation.
Patient characteristics and assessments were collected from a clinical database and chart review. As the comprehensive neuropsychological evaluations were conducted for clinical purposes, not all patients completed the same neuropsychological tests. For the purposes of this study, individual neuropsychological test items were assigned to one of five cognitive domains—attention, which included processing speed, memory, language, visuospatial or executive functioning—and summary domain z-scores were calculated. Most subjects were administered the MoCA as a global measure of cognition. The following test scores were assigned to the attention domain: Trail Making Test A; Symbol Digit Modalities Test (SDMT) written and oral versions; Stroop Color-Word Test (word and colour tasks); Wechsler Adult Intelligence Scale (WAIS), third or fourth edition (WAIS-III or WAIS-IV) Digit Span; WAIS-III Letter Number Sequencing (LNS); WAIS-IV Arithmetic, Symbol Search and Coding; Delis-Kaplan Executive Function System (D-KEFS) Trail Making Test Visual Scanning, Number Sequencing, Letter Sequencing and Motor Speed conditions; Wechsler Memory Scale (WMS)-III Spatial Span; and Repeatable Battery for Neuropsychological Status (RBANS) Digit Span and Coding. The following test scores were assigned to the language domain: Category fluency test (Animals); Controlled Oral Word Association Test (FAS or CFL); Action fluency; RBANS Picture Naming and Semantic Fluency. The following test scores were assigned to the memory domain: Brief Visuospatial Memory Test—Revised Total Recall and Delayed Recall; Hopkins Verbal Learning Test-Revised (HVLT-R) Total Recall, Delayed Recall and Recognition Discrimination Index; WMS-III or WMS-IV Logical Memory I and II; RBANS Story Memory and Recall; RBANS List Learning, Recall and Recognition; RBANS Figure Recall; and California Verbal Learning Test, Second Edition Total Recall, Short Delay Free Recall and Long Delay Free Recall. The following test scores were assigned to the visuospatial domain: Judgement of Line Orientation (JLO); WAIS-III or WAIS-IV Block Design; RBANS Figure Copy; and RBANS Line Orientation. The following test scores were assigned to the executive domain: Trail Making Test B; Wisconsin Card Sorting Test errors, perseverative responses and perseverative errors; Stroop Color-Word Test interference; WAIS-III or WAIS-IV Matrix Reasoning; Short Category Test, Booklet Format; WAIS-IV Similarities; and D-KEFS Trail Making Test Number-Letter Switching condition. Standardised scores for all tests were converted to z-scores, and available z-scores were averaged for each domain for each subject. Only subjects with two or more neuropsychological tests in each domain were included in analyses except for the visuospatial domain, which was derived from one or more neuropsychological test scores. As an additional measure of visuospatial function, Clock Drawing was scored on a 15-point scale.
Early PD and healthy control cohort
We analysed baseline brain MRIs and neuropsychological test scores for 228 PD participants and 101 healthy controls from the Parkinson’s Progression Markers Initiative (PPMI). We also analysed brain MRIs and neuropsychological test scores for the subset of 97 PD participants who had neuroimaging available 4 years after baseline. The PPMI is a prospective, observational, multicenter study, which aims to identify biomarkers of PD progression. At baseline, PD participants are newly diagnosed (≤2 years), not receiving medications for PD, and do not have dementia. Data for this study were downloaded on 31 May 2017. Study visits include assessment of PD motor symptoms, cognition and neuropsychiatric symptoms. Neuropsychological testing included the MoCA, HVLT-R, JLO, WMS-III LNS, SDMT and semantic fluency (animals). Except for the MoCA, age-normed standardised scores were used for all cognitive tests. Other details about study methodology are available (www.ppmi-info.org/study-design).13
Brain MRI
Using the MPRAGE T1 sequence of available MRI scans, we applied voxel-based morphometry methodology.14 15 Images were segmented into grey matter, white matter and cerebrospinal fluid, and high-dimensionally fit to the Montreal Neurological Institute (MNI) standard space using the CAT12 toolbox (http://dbm.neuro.uni-jena.de/cat/) in conjunction with SPM12 (http://www.fil.ion.ucl.ac.uk/spm/software/spm12/) in Matrix Laboratory (MATLAB) (Mathworks, Natwick, Massachusetts, USA). To improve the fidelity of segmentation for low-contrast subcortical regions, we utilised Lorio et al’s enhanced tissue probability map over SPM12’s standard tissue priors.16 Warping of subject images utilised the Diffeomorphic Anatomical Registration Through Exponentiated Lie Algebra algorithm, which is embedded in SPM12. To preserve the absolute volume of grey matter, segmented images were multiplied by the relative voxel volumes contained within the Jacobian determinant matrix of the deformation field.15 Basal forebrain grey matter density was measured according to a probabilistic map of Ch4 for the reference MNI single-subject brain that was derived from the three-dimensional reconstruction of histological sections from 10 postmortem human brains10 (see figure 1). Unilateral neocortical density was measured according to the Harvard-Oxford probabilistic map of the left and right neocortex for the reference MNI single-subject brain derived from 37 individually segmented T1-weighted MRI images.17 Relative Ch4, Ch123 and neocortical densities were calculated with a custom MATLAB script, which multiplied the grey matter density value for each voxel by the weighting contained within the probabilistic map. Weighted grey matter density values for Ch4, Ch123 and neocortex were summed bilaterally.18
Probabilistic map of basal forebrain cholinergic nuclei 1–4. Coronal views show cholinergic nucleus 4 (Ch4) (A) and cholinergic nuclei 1, 2 and 3 (Ch123) (B) overlaid on brain anatomy. The colour scales indicate the probability (range 0.001–1) that a voxel corresponds to Ch4 (red) or Ch123 (blue). Surface mesh reconstructions (C and D) show the location of Ch4 (red) and Ch123 (blue) in Montreal Neurological Institute anatomic space.
Statistical analysis
Clinical characteristics for both cohorts are reported as mean±SD or median with IQR as appropriate. Pairwise comparisons of age and regional densities between cohorts were performed using Student’s t-tests. In all statistical models, regional density (Ch4, Ch123 or neocortex) was the predictor of interest and the cognitive test score was the outcome variable. All hypothesis tests were two-sided. P values<0.05 were considered statistically significant unless a Bonferroni correction is noted. Analyses were performed using Stata V.14.2.
In the presurgical PD cohort, a linear regression model adjusted for age and sex was used to assess the relationship between Ch4 density and MoCA score. Linear regression models adjusted for sex were used to assess the relationships between Ch4, Ch123 and neocortical densities and each cognitive domain z-score. As cognitive domain z-scores were derived from standardised scores normed for age, these linear regression models were not adjusted for age. Bonferroni correction was applied as a multiple outcome adjustment to comparisons of each regional density to five cognitive domains. Tobit regression models were used to evaluate the relationship between regional densities and the latent attributes assessed by Clock Drawing score, as the score was bounded by an upper limit of 15.
For PPMI cohorts, a linear regression model adjusted for age and sex was used to assess the relationship between Ch4 density and MoCA score. Linear regression models adjusted for sex were used to assess the relationships between Ch4 density and age-normed cognitive test scores. As cognitive test scores were standardised scores normed for age, these linear regression models were not adjusted for age. Bonferroni correction was applied as a multiple outcome adjustment to comparisons of each regional density to six age-normed cognitive test scores.
Results
PD participants in the presurgical cohort
Clinical characteristics of the 125 PD participants with neuropsychological testing results and MRI scans of sufficient quality to calculate regional densities are reported in table 1. Ch4 density and age at neuropsychological testing were significantly inversely correlated (r=−0.34, p=0.0001). Thus, we adjusted for the effect of age in the linear regression model for regional densities and MoCA score. As standardised scores for neuropsychological tests are age-normed, we did not adjust for age in the models assessing the relationship between regional densities and domain z-scores. There was no correlation between Ch4 density and years of education (rs=0.13, p=0.15). Ch4 density was moderately to highly correlated with Ch123 (r=0.68, p<0.0001) and neocortical (r=0.72, p<0.0001) densities.
Clinical characteristics and regional density values of Parkinson disease and healthy control cohorts
In a linear regression model adjusted for age and sex, greater Ch4 density was associated with higher MoCA score (β=14.2; 95% CI=1.5 to 27.0; p=0.03). Similarly, greater Ch123 (β=15.9; 95% CI=2.6 to 29.1; p=0.02) and neocortical (β=25.2; 95% CI=9.9 to 40.4; p=0.001) densities were also associated with higher MoCA score. Next, we evaluated the relationship between Ch4, Ch123 and neocortical densities and each of the five cognitive domains in linear regression models adjusted for sex. Ch4, Ch123 and neocortical densities were all significant predictors of the attention and visuospatial domain z-scores (p<0.01; table 2). In Tobit regression models adjusted for age and sex, greater Ch4 (β=20.0; 95% CI=3.1 to 36.9; p=0.02, n=96) and neocortical (β=19.9; 95% CI=0.1 to 39.7; p=0.049) densities were associated with higher values for the latent attributes assessed by the censored variable Clock Drawing score, but Ch123 density was not (β=8.2; 95% CI=−9.8 to 26.2; p=0.37).
Linear regression model results for association between regional densities and cognitive domain z-scores in the presurgical Parkinson disease cohort
PD participants in PPMI cohort at baseline
To further explore the relationship between Ch4 density and cognition in PD, we analysed baseline data for 228 de novo PD participants from the PPMI. There was no significant difference in age between the PPMI PD cohort at baseline and the presurgical PD cohort (p=0.2). Compared with the presurgical PD cohort, the PPMI PD cohort had higher mean Ch4 (p<0.0001), Ch123 (p<0.0001) and neocortical (p=0.0007) densities. Clinical characteristics of the PPMI PD cohort at baseline are reported in table 1.
In a linear regression model adjusted for age and sex, greater Ch4 density was associated with higher MoCA scores in PD patients (β=9.2; 95% CI=1.9 to 16.5; p=0.01). In linear regression models adjusted for sex, greater Ch4 density was associated with higher JLO (β=20.4; 95% CI=13.8 to 27.0; p<0.001), LNS (β=16.5; 95% CI=9.5 to 23.4; p<0.001) and SDMT (β=41.8; 95% CI=18.7 to 65.0; p<0.001) scores. Ch4 was highly correlated with Ch123 (r=0.73, p<0.0001) and neocortical (r=0.68, p<0.0001) densities. In linear regression models adjusted for age and sex, there was neither a significant association between Ch123 and MoCA score (β=5.8; 95% CI=−0.3 to 11.9; p=0.06) nor neocortical and MoCA score (β=6.2; 95% CI=−2.5 to 14.9; p=0.16).
In linear regression models adjusted for sex, greater Ch123 density was associated with higher JLO (β=13.8; 95% CI=8.4 to 19.2; p<0.001) and higher LNS (β=8.9; 95% CI=3.1 to 14.6; p=0.003) scores. Similarly, neocortical density was associated with higher JLO (β=24.5; 95% CI=17.0 to 32.0; p<0.001) and higher LNS (β=16.4; 95% CI=8.3 to 24.4; p<0.001) scores. See table 3 for results of all linear regression models evaluating the relationships between regional densities and cognitive outcomes in this cohort.
Results of linear regression models for association between regional densities and cognitive test scores in the de novo Parkinson disease cohort at baseline
PD participants in PPMI cohort at 4 years
Clinical characteristics of the 97 PPMI PD participants with 4-year data are reported in table 1. In this subcohort, greater Ch4 (β=30.0; 95% CI=18.2 to 41.8; p<0.001), Ch123 (β=11.6; 95% CI=1.8 to 21.4; p=0.02) and neocortical (β=11.5; 95% CI=4.7 to 18.3; p=0.001) densities were significantly associated with higher MoCA scores in linear regression models adjusted for age and sex. In linear regression models adjusted for sex, greater Ch4 (β=65.9; 95% CI=32.0 to 99.8; p<0.001) was significantly associated with higher SDMT scores (n=94). Similar to baseline analyses, Ch4, Ch123 and neocortical densities were significantly associated with JLO and LNS scores (all p<0.008), except that Ch123 was not significantly associated with higher LNS score (β=10.3; 95% CI=0.7 to 19.8; p=0.04). See table 4 for results of all linear regression models in this subcohort.
Results of linear regression models for association between regional densities and cognitive test scores in the de novo Parkinson disease cohort at 4 years
Healthy controls in PPMI cohort at baseline
Clinical characteristics of healthy controls from PPMI are reported in table 1. There were no significant differences in Ch4 (p=0.19), Ch123 (p=0.99) or neocortex (p=0.99) densities between PPMI PD cohort at baseline and controls. In healthy controls, Ch4 was highly correlated with Ch123 (r=0.77, p<0.0001) and moderately correlated with neocortical (r=0.65, p<0.0001). In linear regression models adjusted for age and sex, greater Ch4 (β=0.6; 95% CI=−4.8 to 6.1; p=0.82), Ch123 (β=2.8; 95% CI=−2.7 to 8.3; p=0.33) and neocortical densities (β=2.0; 95% CI=−4.4 to 8.5; p=0.53) were not associated with higher MoCA scores. Except for a significant association between greater neocortical density and higher JLO (β=16.2; 95% CI=5.6 to 26.7; p=0.003) score, there were no associations between regional densities and cognitive test scores. See table 5 for results of linear regression models evaluating the relationship between regional densities and cognitive test scores.
Results of linear regression models for association between Ch4 density and cognitive test scores in healthy controls at baseline
Discussion
In two independent PD cohorts, we found that Ch4 density was associated with global cognition as assessed by the MoCA. These findings are consistent with prior studies implicating cholinergic basal forebrain degeneration in global cognitive impairment in PD. Pathological studies have shown greater neuronal loss in the NBM in PD patients with dementia compared with those without.5 6 As measured by positron emission tomography (PET) and single-photon emission computed tomography (SPECT) imaging, there is reduced cortical cholinergic innervation in PDD compared with PD.19 20 A PET study using a ligand binding acetylcholinesterase showed that greater impairment in global cognition in PD was associated with a greater proportion of individuals having deficient cortical cholinergic activity. In the PD group with a global cognitive z-score 2 SD below the mean, greater than 80% had a significant cortical cholinergic deficit.21
When we analysed the relationships between Ch4 density and individual cognitive domains, we found that Ch4 density was associated with attention in early and more advanced PD. Attentional impairment in PD has previously been associated with cholinergic denervation in acetylcholinesterase PET studies.21 22 Secondary analysis of data from a clinical trial for the cholinesterase inhibitor rivastigmine showed that rivastigmine caused a significant improvement in multiple aspects of attention in PDD compared with placebo.23 In addition to attention, we found an association between Ch4 density and visuospatial function in both the early and more advanced PD cohorts. These findings are consistent with a previously reported association between cortical cholinergic denervation as measured by acetylcholinesterase PET and JLO performance in PD.22 A link between cortical cholinergic denervation and visuospatial function in PD is supported by acetylcholinesterase PET studies showing greater cholinergic denervation in posterior regions in PD, especially the occipital cortex.24 25 In a study of healthy controls, cholinergic augmentation with donepezil was associated with improvement in visuospatial performance.26 When we analysed a subcohort of PD participants from PPMI with MRI at 4 years, we found the same associations between Ch4 density and cognition, that is, lower Ch4 density was associated with worse performance on tests of global cognition (MoCA), attention (LNS and SDMT) and visuospatial function (JLO). These relationships between Ch4 density and cognition appear to be specific to PD, as there were no associations between Ch4 density and global cognition, attention or visuospatial function in healthy controls from the PPMI.
To explore whether the relationships between Ch4 density and cognition were specific to this region, we investigated the relationships between the other cholinergic basal forebrain nuclei, Ch123, and cognition. We also investigated the relationships between the primary target of Ch4, the neocortex, and cognition. Like Ch4, Ch123 and neocortical densities were significantly associated with MoCA score, attention z-score and visuospatial z-score in the more advanced PD cohort. Similarly, Ch123 and neocortical densities were associated with visuospatial function (JLO) and attention (as measured by LNS score) in the PPMI PD cohort. Among healthy controls, the only significant association was between neocortical density and JLO score. This suggests that premorbid cerebral volumes or pathology other than PD-related pathology, for example, Alzheimer disease (AD) pathology, may contribute to associations between regional densities and visuospatial function.
In the PPMI PD cohort, there were some differences in the relationships between regional densities and the MoCA and attentional measures. Unlike Ch4 density, Ch123 and neocortical densities were not associated with MoCA at baseline but were at 4 years. Also in contrast to Ch4 density, Ch123 and neocortical densities were not associated with SDMT at baseline or 4 years. Lastly, Ch123 was not associated with LNS score at 4 years while it was at baseline. One explanation for these discrepant findings between regional densities, global cognition and attention may be that we lacked power to detect relationships that were present. However, there may be a specific relationship between Ch4 density and attention in early PD that informs our findings for SDMT scores and the MoCA. Performance on both the LNS and the SDMT require aspects of attention. Compared with LNS, performance on the SDMT relies more on processing speed. Processing speed is widely distributed and may be affected by neocortical cholinergic deficiency in the absence of significant neocortical atrophy. The apparent relationship between reduced Ch4 density and impaired processing speed may also be mediating the relationship between Ch4 density and MoCA scores. Whether processing speed is preferentially affected by cortical cholinergic deficiency in early PD deserves further investigation.
As cholinergic nuclei 1 and 2 project to hippocampus, and cholinergic nucleus 3 projects to the olfactory bulb,27 we did not expect that there to be a relationship between Ch123 density and the cognitive impairment typically observed in PD. However, Ch4 and Ch123 were highly correlated. Thus, it is not surprising that Ch123 density shared most of the same relationships with cognition as Ch4 density. Even so, there were some relationships between Ch4 and cognition in the early PD cohort specific to Ch4 as discussed above.
Of interest, we found no relationships between cholinergic basal forebrain volumes and memory function. Like our study, an early study using cholinesterase PET found an association between neocortical cholinergic deficiency and attentional impairment but not memory impairment.22 Subsequent larger studies did find an association between cortical cholinergic deficiency and memory impairment.21 28 One explanation for this discrepancy is that the earlier study was not adequately powered to detect a difference between cholinergic deficiency and memory impairment.22 Similarly, the lack of power to detect an association may explain why we did not detect a relationship between cholinergic basal forebrain density and memory function. However, the numbers of PD participants investigated in the current study are similar in size to the larger PET studies.21 28
Another possible explanation for why we did not find a relationship between cholinergic basal forebrain density and memory function is that our PD cohorts are relatively younger, even the more advanced cohort, and therefore are less likely to have co-occurring AD pathology. One study found that, while reduced cholinergic basal forebrain volume predicted entorhinal cortical degeneration, memory impairment was present only when there was also entorhinal cortical involvement.29 It is possible that the reduced likelihood of AD pathology in this study’s PD cohorts make them less likely to have entorhinal cortical involvement and memory impairment as accompaniments of cholinergic basal forebrain degeneration. Our findings of a relationship between Ch4 density and measures of attention and visuospatial function in de novo and more advanced PD differ from a prior study evaluating Ch4 and cognitive measures in the PPMI PD cohort.11 That study found no association between Ch4 and cognitive test scores at baseline, nor any association between cholinergic nucleus 1 and 2 and cognition.11 Potential reasons for the discrepant findings are that the other study used a different probabilistic map of the cholinergic basal forebrain derived from a single brain, dichotomised Ch4 at baseline, used raw scores for cognitive tests and had a smaller sample size. Our study used a probabilistic map of the cholinergic basal forebrain derived from 10 brains and maintained Ch4 density as a continuous predictor.
The limitations of our study apply to the presurgical PD cohort: it is retrospective, not all patients received the same neuropsychological tests, and presurgical patients may be intrinsically different than PD patients who do not undergo surgical interventions for motor symptoms. Even with these limitations, findings in the presurgical PD cohort and de novo PPMI PD cohort were consistent. A strength of this study is that the associations between Ch4 density and cognition—global cognition, attention and visuospatial function—were found in both early and more advanced PD cohorts. A strength of the presurgical cohort is that it is relatively large for a neuroimaging study in PD patients with longer disease duration and motor symptoms requiring surgical treatment.
In summary, we found that reduced Ch4 density in PD, as measured by brain MRI, is associated with impaired global cognition, attention and visuospatial function in early and more advanced PD. Impairment in one aspect of attention, processing speed, may be preferentially associated with reduced Ch4 density, and thus cortical cholinergic denervation in early PD. This study provides further evidence that cholinergic basal forebrain degeneration is an important contributor to cognitive impairment in PD, even soon after diagnosis. Considering the increased morbidity and mortality associated with cognitive impairment in PD, there should be continued research investigating therapies that target the cholinergic basal forebrain, such as cell-based therapies or low-frequency deep brain stimulation, and continued research investigating medications other than cholinesterase inhibitors that target the cholinergic system, such as muscarinic and nicotinic receptor agonists or modulators.
Acknowledgments
Data used in the preparation of this article were obtained from the Parkinson’s Progression Markers Initiative (PPMI) database (www.ppmi-info.org/data). For up-to-date information on the study, visit (www.ppmi-info.org). PPMI—a public–private partnership—is funded by the Michael J Fox Foundation (MJFF) for Parkinson’s Research and funding partners, including Abbvie, Avid Radiopharmaceuticals, Biogen, Bristol-Myers Squibb, Covance, GE Healthcare, Genetech, GlaxoSmithKline, Lilly, Lundbeck, Merck, Meso Scale Discovery, Pfizer, Piramal, Roche, Servier and UCB. The MJFF was not involved in the data analysis for this article. Neither the funding agency nor any of the sponsors of the PPMI were involved in the design and conduct of the study; collection, management, analysis and interpretation of the data; preparation, review or approval of the manuscript; and decision to submit the manuscript for publication.
References
Footnotes
Contributors MJB and TJD contributed to the conception and design of the research; MJB, SAS, JCB, CSF, JLF, MES, CAM and TJD contributed to the acquisition, analysis and interpretation of data; MJB drafted the text; and MJB, SAS, JCB, CSF, JLF, MES, CAM and TJD reviewed and critiqued the text.
Funding This work was supported by the Commonwealth of Virginia's Alzheimer's and Related Diseases Research Award Fund, administered by the Virginia Center on Aging, School of Allied Health Professions, Virginia Commonwealth University; and the Office of the Assistant Secretary of Defense for Health Affairs through the Neurotoxin Exposure Treatment Parkinson's Research (NETPR) under Award No. W81XWH-16-1-0768. Opinions, interpretations, conclusions and recommendations are those of the authors and are not necessarily endorsed by the Department of Defense.
Disclaimer MJB: salary: University of Virginia; clinical trials: National Institutes of Health, Azevan, Axovant, Merck, Eisai, Biogen and Acadia. SAS: salary: University of Virginia; grants: US Department of Health and Human Services—Administration for Community Living and Virginia Dementia Specialized Supportive Services. JCB: grants: UVA Neuroscience Presidential Fellowship and UVA Data Science Presidential Fellowship. CSF: salary: University of Virginia. JLF: salary: University of Virginia; salary support: Acadia. MES: salary: University of Virginia. CAM: salary: University of Virginia; grants: US Department of Health and Human Services—Administration for Community Living and Virginia Dementia Specialized Supportive Services. TJD: salary: University of Virginia; grants: Biocore, LLC, UVA Data Science Presidential Fellowship, UVA Neuroscience Presidential fellowship and UVA Health Sciences.
Competing interests MJB, SAS, JCB, JLF, MES and TJD received grant support from the Department of Defense and the Commonwealth of Virginia’s Alzheimer’s and Related Diseases Research Award Fund.
Patient consent for publication Not required.
Ethics approval The institutional review board at the University of Virginia approved this study.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data are available on reasonable request, and may be obtained from a third party and are not publicly available.