Website Edition: April/May 2013

Journal Featured Article

Movement Disorders
DOI: 10.1002/mds.25282

Cerebrospinal fluid Aβ levels correlate with structural brain changes in Parkinson's disease

Authors: Mona K. Beyer MD, PhD1,2,*, Guido Alves MD, PhD1,3, Kristy S. Hwang BS4,5, Sona Babakchanian BS4,5, Kolbjorn S. Bronnick PhD1, Yi-Yu Chou MS5, Turi O. Dalaker MD, PhD1, Martin W. Kurz MD, PhD1,3, Jan P. Larsen MD, PhD1, Johanne H. Somme MD6, Paul M. Thompson PhD4,5, Ole-Bjørn Tysnes MD, PhD7, Liana G. Apostolova MD, MSCR4,5

1 The Norwegian Center for Movement Disorders, Stavanger University Hospital, Stavanger, Norway
2 Department of Radiology and Nuclear Medicine, Oslo University Hospital, Oslo, Norway
3 Department of Neurology, Stavanger University Hospital, Stavanger, Norway
4 Department of Neurology, David Geffen School of Medicine, University of California Los Angeles, Los Angeles, California, USA
5 Laboratory of Neuro Imaging, David Geffen School of Medicine, University of California Los Angeles, Los Angeles, California, USA
6 Department of Neurology, Cruces University Hospital, Baracaldo, Spain
7 Department of Neurology, Haukeland University Hospital, Bergen, Norway

*Correspondence to: Dr. Mona K. Beyer, Department of Radiology and Nuclear Medicine, Oslo University Hospital, Sognsvannsvn 20, 02770 Oslo, Norway;

Relevant conflicts of interest/financial disclosures: Nothing to report.

Full financial disclosures and author roles may be found in the online version of this article.

Article first published online: 13 FEB 2013


ParkWest is a large Norwegian multicenter study of newly diagnosed drug-naïve subjects with Parkinson's disease (PD). Cognitively normal PD subjects (PDCN) and PD subjects with mild cognitive impairment (PDMCI) from this cohort have significant hippocampal atrophy and ventricular enlargement, compared to normal controls. Here, we aimed to investigate whether the same structural changes are associated with cerebrospinal fluid (CSF) levels of amyloid beta (Aβ)38, Aβ40, Aβ42, total tau (t-tau), and phosphorylated tau (p-tau). We performed three-dimensional radial distance analyses of the hippocampi and lateral ventricles using the MRI data from ParkWest subjects who provided CSF at baseline. Our sample consisted of 73 PDCN and 18 PDMCI subjects. We found significant associations between levels of all three CSF Aβ analytes and t-tau and lateral ventricular enlargement in the pooled sample. In the PDCN sample, all three amyloid analytes showed significant associations with the radial distance of the occipital and frontal horns of the lateral ventricles. CSF Aβ38 and Aβ42 showed negative associations, with enlargement in occipital and frontal horns of the lateral ventricles in the pooled sample, and a negative association with the occipital horns in PDMCI. CSF Aβ levels in early PD correlate with ventricular enlargement, previously associated with PD dementia. Therefore, CSF and MRI markers may help identify PD patients at high risk for developing cognitive decline and dementia in the course of their illness. Contrary to Alzheimer's disease, we found no associations between CSF t-tau and p-tau and hippocampal atrophy. © 2013 Movement Disorder Society

Volume 28, Issue 3, pages 302-310, March 2013

Read more

Article Text

Mild cognitive impairment (MCI) is often present in Parkinson's disease (PD) patients at the time of diagnosis.[1] PD subjects with MCI (PDMCI) are approximately five times more likely to receive a diagnosis of dementia over the subsequent 4 years.[2] Pathologically, PD is characterized by progressive degeneration of dopaminergic neurons, Lewy bodies (LBs), and Lewy neurites. Cortical LBs are characteristic for PD dementia (PDD).[3] Diffuse cortical amyloid deposition has been reported postmortem in PDD,[4] suggesting that the interface between PD and Alzheimer's disease (AD) needs to be more thoroughly investigated. Coexisting AD pathology promotes more-rapid cognitive decline and early PDD.[5]

Amyloid beta 1-42 (Aβ42), total tau (t-tau), and phosphorylated tau (p-tau) are the most established cerebrospinal fluid (CSF) biomarkers of AD. Low CSF Aβ42 and high CSF t-tau and p-tau levels have been reported in AD and in MCI subjects who later converted to AD.[6] Investigations into levels of traditional CSF AD biomarkers in PD have been conflicting.[7-9] Yet, recent large-scale studies have reported reduced CSF levels of Aβ42 in both cognitively normal and cognitively impaired PD patients.[10, 11] We recently reported reduced CSF Aβ42, but normal CSF t-tau and p-tau levels in our newly diagnosed drug-naïve ParkWest PD cohort.[10]

CSF Aβ42, t-tau, and p-tau levels have been linked to hippocampal atrophy and ventricular enlargement in AD.[12-14] Ventricular enlargement, hippocampal atrophy, and white-matter hyperintensities (WMHs) are commonly observed in PD and have been associated with cognitive decline.[15-18] Previous ParkWest analyses showed larger ventricles, but no hippocampal atrophy or greater WMH burden in PDMCI, relative to PDCN.[15, 19] The associations between white-matter changes and CSF markers in the ParkWest sample is currently being investigated. A previously published ParkWest analysis failed to find an association between WMH and cognitive performance.[19] However, as other research groups have reported potential associations between cognitive function and CSF markers in PD,[20] and that white-matter change may contribute to cognitive deficits associated with PD,[16] we have included WMH as a covariate in our analyses.

Whether CSF Aβ and tau markers are associated with ventricular enlargement and hippocampal atrophy in PD remains to be determined. The aim of this study was to investigate, in a three-dimensional (3D) model, whether these MRI measures are associated with CSF levels of amyloid beta 1-38 (Aβ38), amyloid beta 1-40 (Aβ40),42, t-tau, and p-tau in early, untreated PD patients with and without MCI.

Materials and Methods


All subjects are participants of the ongoing Norwegian ParkWest study.[21] ParkWest is a population-based, multicenter, prospective, longitudinal cohort study aiming to improve our understanding of motor and nonmotor PD progression and develop promising biomarkers for PDMCI and PDD. Only newly diagnosed drug-naïve PD subjects meeting the Gelb et al.'s clinical criteria for PD[22] were eligible to participate.

Diagnostic procedures included evaluation of medical history, standardized clinical, neuropsychiatric, and neuropsychological examinations at study entry, radiological examination with 3D T1-weighted brain MRI, dopamine transporter imaging when available (n = 31), and repeated clinical assessments to evaluate drug response. Patients with atypical parkinsonism or dementia during the first year of motor onset were excluded, according to the revised dementia with LB criteria by McKeith et al.[23] Details of the diagnostic ascertainment are reported by Alves et al.[24]

Of the 207 drug-naïve incident PD subjects who agreed to longitudinal participation, 91 provided CSF and MRI data at baseline. Seventy-three subjects were cognitively normal (PDCN), whereas 18 met criteria for PDMCI. All participants signed informed consent. The study protocol was approved by the Regional Committee for Medical Research Ethics of Western Norway.

Clinical and Neuropsychological Examination

Severity of parkinsonian symptoms was assessed with the Unified Parkinsons Disease Rating Scale (UPDRS)[25] and the modified Hoehn & Yahr (H & Y) scale.[26] All PD subjects underwent neuropsychological assessments of memory, executive and visuospatial function,[21] which consisted of the Mini–Mental State Examination (MMSE) scale [27] verbal memory assessment with the California Verbal Learning Test II,[28] visuospatial assessment with the Silhouettes and Cube subtests from the Visual Object and the Space Perception Battery,[29] and assessment of attention and executive functioning with a semantic verbal fluency test (animal fluency),[30] the serial 7 test from the MMSE,[27] and the Stroop test.[31]

Cognitive data for all PD subjects were converted to z-scores using the mean and standard deviations (SDs) of an age-matched cognitively normal control group, as previously described.[15, 21] PD subjects with cognitive performance more than 1.5 SDs below the predicted level by an age-, education-, and sex-corrected linear regression model in one or more cognitive domains were classified as having PDMCI.


MRI was performed at four of the five study sites. A 1.5-T MRI scanner was used at three centers (Philips Intera, Best, The Netherlands, at Stavanger and Haugesund, and Siemens Symphony, Erlangen, Germany, in Bergen). A 1.0-T Philips Intera system was used at Arendal.

The following protocols were used: (1) at Stavanger: repetition time/echo time (TR/TE) 10.0/4.6 msec, flip angle 30.0 degrees, 1-mm slices with no gap, number of excitations (NEX) 2, and matrix 256 × 256; at Haugesund: TR/TE 20.0/4.6 msec, flip angle 30.0 degrees, 1-mm slice thickness with no gap, NEX 1, and matrix 256 × 256; at Bergen: TR/TE 2130.0/3.9 msec, flip angle 15.0 degrees, 1-mm slice thickness with no gap, NEX 1, and matrix 256 × 256; and at Arendal: TR/TE 25/6.9 msec, flip angle 30.0 degrees, 2-mm slice thickness with no gap, NEX 1, and matrix 256 × 256.

All baseline images were inspected by two neuroradiologists. Scans with large-vessel cortical infarcts or other structural lesions that could result in parkinsonian symptoms (N = 7) were excluded. Twelve subjects with scan or movement artifacts and 5 with baseline dementia were also excluded from the study.

Imaging Analyses

3D T1-weighted MRI data were subjected to intensity[32] and spatial normalization to the International Consortium for Brain Mapping ICBM53 brain atlas using a nine-parameter transformation (three translations, three rotations, and three scales).[33] Hippocampal formations of a randomly selected ParkWest training data set were manually segmented on gapless coronal slices by one experienced rater (M.K.B.) blinded to subjects' age, sex, and diagnosis after a detailed, well-established protocol. Traces were closely inspected for accuracy by a second experienced hippocampal rater (L.G.A.). The training sample composition was proportionate to the ratio of subject enrollment across the four imaging centers to prevent, as far as possible, potential center bias in statistical sampling. The hippocampal traces included hippocampus proper, dentate gyrus and subiculum.

Hippocampal Segmentation

Hippocampi of the full data set were segmented with our automated machine-learning hippocampal segmentation approach, based on statistical adaptive boosting.[34] Using the manually traced hippocampal training set, AdaBoost develops statistical rules for classifying each voxel in a new image as belonging to the hippocampus or not. Based on the feature information contained in the positive and negative voxels of the training data set, AdaBoost developed a set of rules and computed the optimal combination and weighting of these features for accurate segmentation of unknown images.[34]

Ventricular Segmentation

We employed a previously validated semiautomated ventricular segmentation approach.[35] Briefly, a human rater (M.K.B.) first traced the lateral ventricles of 4 subjects. These traces were converted into 3D parametric ventricular mesh models, termed atlases. Using fluid registration, each atlas was separately warped to match, and thereby extract, the shape of the lateral ventricle of each new subject's scan. This resulted in four lateral ventricle segmentations per subject, which were then averaged to create one final ventricular model. Averaging four separate segmentations minimizes automated labeling errors that occur when only one atlas is used.

After converting the segmented hippocampi and lateral ventricles into 3D parametric meshes, we computed the medial core (a medial curve threading down the center of each structure) and the radial distance from the medial core to each surface point for each structure in each subject.[36] Radial distance provides an intuitive measure of the thickness of the structure from its core to each point on its boundary.

For analysis of WMH lesion volumes, masks of WMH lesions were prepared using a semiautomated local threshold technique on axial FLAIR images, as previously described.[19] One single rater created the masks blinded to diagnosis and test results, with excellent reproducibility. Normalized volumes of WMH (to total brain volume) were used in the further analyses.

Interscanner Reliability Analyses

To test scanner reliability, we conducted human phantom scanning. Three cognitively normal volunteers were imaged twice on the same day on each scanner. Hippocampi were manually traced and volumes were obtained as previously described. Whole-brain volumes were obtained using Sienax.[37] We used PASW software (PASW for Windows Release 18.0.1; SPSS, Inc., Chicago, IL) to obtain Cronbach's alpha reliability coefficients on whole-brain and hippocampal volumetric measurements between scanning sites. Intersite Cronbach's alpha for whole-brain volume was 0.96 and for hippocampal volume was 0.97.

CSF Collection and Analyses

All lumbar punctures were performed between 7 and 10 a.m. after overnight fasting to minimize diurnal variation of the level of CSF Aβ. CSF levels of Aβ38, Aβ40, and Aβ42 were quantified by an Aβ triplex assay (Human Aβ peptide Ultra-Sensitive Kit; Meso Scale Discovery, Gaithersburg, MD). CSF levels of t-tau and p-tau were analyzed using a commercial sandwich enzyme-linked immunosorbent assay (INNOTEST® hTAU-Ag and INNOTEST® PHOSPHO-TAU (181P); Innogenetics). CSF samples were randomized and run in duplicates, blinded for clinical diagnosis, according to the manufacturer's instructions.

Apolipoprotein E Analysis

DNA was extracted from 200 µL of ethylenediaminetetraacetic acid blood using the QIAamp 96 DNA Blood Kit (Qiagen, Hilden, Germany). Apolipoprotein E (ApoE) genotype was determined using the LightCycler Apo E Mutation Detection Kit (Roche Diagnostics, Mannheim, Germany), according to the manufacturer's instructions.

Statistical Analysis

Demographic and clinical between-group comparisons were conducted with the two-sample t test using PASW software. Differences in sex and apolipoprotein E4 genotype (ApoΕ4) distributions were assessed with a chi-square test. The presence of at least one ApoΕ4 allele was coded as 1, and absence was coded as 0. Three PDCNs with missing ApoΕ4 data were coded as 0.5. Variables with non-normal distribution (i.e., MMSE) were analyzed using Mann-Whitney U test. We used univariate statistics to examine for an effect of age, ApoE4, WMH, and scanning site on CSF analytes and ventricular volume in our sample.

Main analyses were conducted with linear regression. Models tested associations between hippocampal and ventricular radial distance (or volume) and CSF variables while adjusting for scanning site, age, WMH, and ApoΕ4. Partial correlations between ventricular volumes and CSF variables while adjusting for age, scanning site, WMH, and ApoE4 were also sought.

For 3D mapwise multiple comparisons correction, we ran 100,000 permutations thresholded at P < 0.01. After completion of these random permutations, a final corrected P value was computed indicating in what proportion of iterations the experimental P value beat the P value of the original experiment.


CSF Aβ, t-tau, and p-tau were available for 91, 84, and 89 subjects, respectively. There were no differences in age, education, motor subscale of UPDRS or H & Y, ApoE4 genotype, or MMSE between subjects who agreed or refused to undergo a lumbar puncture. A significantly greater proportion of men agreed to a lumbar puncture, compared to women (68% versus 43%; P = 0.015).

PDMCI subjects were, on average, older and had a lower mean MMSE score than PDCNs. There were no significant differences in CSF analyte levels between diagnostic groups or between ApoE4 carriers and noncarriers (Table 1). PDMCI subjects had significantly larger left and trend larger right ventricular volumes (left: P = 0.027; right: P = 0.083), but comparable hippocampal volumes, compared to PDCNs (Table 1).

Table 1. Comparison of demographic, clinical, and CSF variables in PD controls and PDMCI (top) as well as ApoE4 carriers and noncarriers (bottom)

Variable PDMCI (N = 18) PDCN (N = 73) P Value
Sex (M/F) 10/8 50/23 0.410
Age, years 71.7 (6.4) 65.1 (9.9) 0.008
Education mean, years 11.4 (3.4) 11.4 (3.5) 0.940
APOΕ4 allele present, N (%) 4 (22) 28 (38.4) 0.250
PD side, R/L/both 14/10/7 61/53/13 0.170
UPDRS motor score 20.8 (9.0) 20.7 (10.3) 0.720
H & Y on state score 1.9 (0.4) 1.8 (0.6) 0.360
MMSE 26.7 (3.0) 28.3 (1.8) 0.035
CSF Aβ38, pg/mL 417.2 (262.9) 467.5 (277.3) 0.430
CSF Aβ40, pg/mL 5,545.9 (1,422.6) 5,820.2 (2,214.0) 0.840
CSF Aβ42, pg/mL 363.0 (169.4) 363.3 (208.2) 0.700
CSF t-tau, pg/mL 227.3 (84.6) 221.2 (127.5) 0.260
CSF p-tau, pg/mL 53.1 (24.6) 58.3 (37.4) 0.860
Left hippocampus, mm3 4,603 (909) 4,897 (866) 0.210
Right hippocampus, mm3 4,654 (992) 4,868 (791) 0.330
Left frontal horn, mm3 15,461 (1,217) 14,939 (1,316) 0.130
Left temporal horn, mm3 265 (63) 237 (45) 0.180
Left occipital horn, mm3 6,251 (954) 5,371 (1,348) 0.011
Left whole ventricle, mm3 21,976 (1,588) 20,546 (2341) 0.016
Right frontal horn, mm3 17,553 (2,666) 15,821 (2,989) 0.027
Right temporal horn, mm3 404 (66) 392 (62) 0.480
Right occipital horn, mm3 3,696 (1,169) 4,089 (1,050) 0.170
Right whole ventricle, mm3 21,653 (2,967) 20,302 (2,919) 0.083


  ApoΕ4 carriers (N = 32) ApoΕ4 noncarriers (N = 59) PValue
1.      Values denote means (SD). P values were computed using a two-sample t test for continuous variables and a chi-square test for sex. For tests with categorical variables or non-normal distribution, Kruskal-Wallis' test with Mann-Whitney's test was used. Significant P values are in bold and show group differences.
CSF Aβ38 pg/mL 483.5 (255.5) 433.3 (267.8) 0.240
CSF Aβ40 pg/mL 6,082.1 (2,018.4) 5,550.3 (2,036.9) 0.391
CSF Aβ42 pg/mL 350.1 (195.7) 358.4 (181.8) 0.842
CSF t-tau pg/mL 231.5 (111.4) 211.8 (118.1) 0.386
CSF p-tau pg/mL 61.4 (33.6) 54.6 (36.4) 0.275

3D univariate age, center, ApoE4, and WMH statistical maps are provided in Figure 1. Only age and WMH showed significant association with ventricular radial distance (age left: pcorrected = 0.04; right: pcorrected = 0.0006; WMH left: pcorrected = 0.045; right: pcorrected = 0.053). Age was also correlated with hippocampal volume (left: r = −0.17, P = 0.019; right: r = −0.36, P < 0.001), Aβ40 (r = 0.25; P = 0.016), t-tau (r = 0.44; P < 0.001), and p-tau (r = 0.23; P = 0.03). WMH volume was correlated with hippocampal volume (left: r = −0.38, P < 0.001; right: r = −0.24, P = 0.03) and t-tau (r = 0.4; P < 0.001). One-way analysis of variance showed no significant differences in mean CSF analyte levels between centers.

image Figure 1. 3D significance (left panel) and correlation maps (right panel) showing associations of ventricular radial distance with age, center, ApoE4, and WMH in the pooled sample. Areas in red and white in statistical maps show statistical significance (P < 0.01).

It is worth noting that our results were not driven by outliers, that the regression standardized residuals had a normal distribution, and that we excluded the possibility of multicolinearity using the colinearity diagnostics option procedure recommended by Belsey et al.[38]

Multivariate Analyses

3D regression analyses were conducted in the pooled sample and then separately in PDCNs and PDMCIs. ApoΕ4 status, WMH and scanning site were used as covariates. We found no significant associations between any of our CSF analytes and hippocampal radial distance (maps not shown).

In contrast, we found significant associations with ventricular radial distance. In the pooled sample (Fig. 2 ), all three amyloid measures showed significant negative associations with the occipital (Aβ38: left: pcorrected = 0.0023; right: pcorrected = 0.0034; Aβ40: left: pcorrected = 0.025; right: pcorrected = 0.043; Aβ42: left: pcorrected = 0.022; right: pcorrected = 0.01) and frontal horns (Aβ38: left: pcorrected = 0.0022; right: pcorrected = 0.0027; Aβ40: left: pcorrected = 0.016; right: pcorrected = 0.01; Aβ42: left: pcorrected = 0.0054; right: pcorrected = 0.0045). CSF t-tau showed significant negative associations with the left occipital (pcorrected = 0.046) and frontal horns (pcorrected = 0.034).

image Figure 2. 3D significance (left panel) and correlation maps (right panel) of associations between CSF Aβ and tau species and ventricular radial distance in the pooled sample. Areas in red and white in statistical maps show statistical significance (P < 0.01). All results are corrected for center, age, and ApoE4.

In PDMCIs, only Aβ40 showed a trend-level significant negative association with right occipital horn enlargement (right pcorrected = 0.063) (see Supporting Fig. 3).

In PDCNs, all three amyloid measures showed significant associations with radial distance of the occipital and frontal horns of the lateral ventricles (occipital horn Aβ38: left: pcorrected = 0.01; right: pcorrected = 0.0088; Aβ40: left: pcorrected = 0.07; right: pcorrected = 0.029; Aβ42: left: pcorrected = 0.057; right: pcorrected = 0.046; frontal horn Aβ38: left: pcorrected = 0.0027; right: pcorrected = 0.0033; Aβ40: left: pcorrected = 0.059; right: pcorrected = 0.018; Aβ42: left: pcorrected = 0.015; right: pcorrected = 0.0052) (see Supporting Fig. 4). T-tau and p-tau showed no significant associations with morphology of the lateral ventricles in PDCNs.

We repeated linear regression analyses with ventricular volumes as the dependent variable. In all cases, the direction of the observed associations remained unchanged, yet some of the effects became trend level or not significant (Supporting Table 1). We also reversed the regression models, taking CSF analytes as the dependent variables. Mean hippocampal volume was not a significant predictor of any CSF measure. Mean ventricular volume was a significant predictor of CSF Aβ38 (P = 0.03), Aβ40 (P = 0.016), and Aβ42 (P = 0.007). Finally, we repeated analyses after removing all subjects scanned on the 1-T MRI instrument. Results remained largely unchanged (Supporting Table 2; Supporting Figs. 3–5).


Using the baseline data from a large, population-based prospective study of incident PD, we found significant associations between several CSF Aβ species and lateral ventricular enlargement in PDCNs and in the pooled sample of PD subjects. No significant correlation was detected between ventricular radial distance and CSF p-tau, whereas t-tau showed significant associations with ventricular size only in the pooled sample. There were no significant associations between any CSF analyte and hippocampal radial distance.

42, Aβ40, and Aβ38 are the three most common Aβ peptides generated in the processing of amyloid precursor protein.[39] Aβ42 is particularly prone to aggregate and sequester in the form of amyloid plaques in the human brain.[40] Low CSF Aβ has been reported on in both AD and PD.[10] The significance of the other two abundant Aβ species—Aβ40 and Aβ38—on cognitive impairment in PD remains poorly understood. A recent study reported on the CSF Aβ42/Aβ38 ratio as the measure that best discriminates between dementia with LBs and AD.[41] Ventricular expansion predicts future cognitive decline[42] and often coincides with low levels of CSF Aβ42 in cognitively normal persons.[13, 43] All three Aβ species were associated with memory impairment in our research cohort.[10] Taken together, these data suggest a role of these CSF biomarkers for cognitive decline in PD.

We found a significant negative association between CSF t-tau with the left occipital and frontal horns in the pooled sample. CSF tau—a marker of axonal neuronal degeneration[44]—has been reported to be normal[10] or decreased in PDCNs.[11] Yet, increased CSF tau levels were evident once cognitive impairment was clinically manifest.[9] As we follow the ParkWest cohort, we will be able to investigate the development of late-occurring CSF tau changes and their relation to structural brain changes.

To our knowledge, this is the first study to investigate associations between hippocampal and lateral ventricle structural differences and neurodegenerative CSF biomarkers in PD. Our results support an association between CSF Aβ measures and structural brain changes in newly diagnosed drug-naïve PD patients. The observed associations in the cognitively normal PD group agree with other reports,[10, 45] and suggests that both CSF and structural brain changes develop before the manifestation of cognitive decline. In the PDMCI group, we found a trend-level significant negative association with right occipital horn enlargement and Aβ40. Yet, PDMCI subjects were, on average, older than the PDCN subjects, and although we adjusted for age, one cannot fully exclude some residual effect. Thus, one must consider the possibility that the observed ventricular changes are also related to aging. Future studies and our longitudinal analyses of the ParkWest data will be well positioned to address this issue.

In this study, we did not find a significant correlation between WMH volumes and CSF Aβ markers in PD. This is in contrast to, for example, findings of a reported inverse relationship between the volume of chronic white-matter change and CSF Aβ markers in a study of subjects with subjective cognitive impairment and MCI.[20]

We failed to find an association between brain white-matter change and cognition in this baseline ParkWest cohort, who had generally little WMH load.[19] Others have found a possible association between white-matter change and PDD[17, 18] as well as an independent relation between WMH and cognitive impairment in PD.[17] White-matter change has also been associated with cognitive impairment in patients with stroke.[46] A recent study showed that increased WMH was associated with worsened motor performance[47] and may contribute to cognitive deficits in PD.[16] It remains to be seen whether WMH will have more affect on cognition in the longitudinal follow-up of our cohort, as has been suggested.[16]

A strong association between the presence of ApoE4 and AD exists,[48] and ApoE4 was higher in subjects with severe WMH after a lacunar infarct.[49] We did not find significant differences between groups with or without cognitive impairment regarding the presence of the ApoE4 allele in the ParkWest study[15] or in a 10-year longitudinal study.[50] This could be partly related to the low WMH burden found in our cohort and the fact that they had no infarcts. Another possible explanation is that ApoE4 does not act through vascular pathology in the PD population. In this study, we found a lower percentage of ApoE4 carriers in the cognitively impaired subjects (22%) than in the cognitively normal group (38%) (Table 1).

Thus, it is possible that factors other than ApoE4 might drive neurodegeneration, leading to atrophy in early PD.

Our study has both strengths and limitations. A major strength is its longitudinal population-based prospective cohort design and its focus on newly diagnosed PD. Our research subjects are well characterized. The use of state-of-the-art imaging analysis techniques to identify regionally specific disease-associated changes, which are difficult to detect with more-conventional analyses, is another strength of this study.

One limitation of our analyses is the small sample size of the PDMCI group with available CSF (N = 18), which restricts our power to detect statistically significant correlations in this group. Without postmortem brain examination, the absence or concomitant AD pathology in our subjects cannot be ascertained. However, many consider low CSF Aβ42, a widely accepted amyloid marker in AD, suggestive of concomitant AD pathology. Postmortem pathologic examinations of the ParkWest cohort are planned.

Our data show that CSF amyloid pathology correlates with ventricular expansion in PD, even in the earliest disease stages. The significance of the structural and CSF findings with respect to PD progression and cognitive decline will be ascertained as we proceed with the longitudinal analyses of the rich ParkWest data set. The time course and sequence of the development of clinical, biochemical, structural, and neuropathologic abnormalities in PD warrants further investigation.


  1. Muslimovic D, Post B, Speelman JD, Schmand B. Cognitive profile of patients with newly diagnosed Parkinson disease. Neurology2005;65:1239–1245.
  2. Janvin CC, Larsen JP, Aarsland D, Hugdahl K. Subtypes of mild cognitive impairment in Parkinson's disease: progression to dementia. Mov Disord2006;21:1343–1349.
    Direct Link:
  3. Hurtig HI, Trojanowski JQ, Galvin J, et al. Alpha-synuclein cortical Lewy bodies correlate with dementia in Parkinson's disease. Neurology2000;54:1916–1921.
  4. Burack MA, Hartlein J, Flores HP, Taylor-Reinwald L, Perlmutter JS, CairnsNJ. In vivoamyloid imaging in autopsy-confirmed Parkinson disease with dementia. Neurology2010;74:77–84. doi: 10.1212/WNL.0b013e3181c7da8e.
  5. Halliday GM, McCann H. The progression of pathology in Parkinson's disease. Ann N Y Acad Sci2010;1184:188–195.
    Direct Link:
  6. Mattsson N, Zetterberg H, Hansson O, et al. CSF biomarkers and incipient Alzheimer disease in patients with mild cognitive impairment. JAMA2009;302:385–393.
  7. Holmberg B, Johnels B, Blennow K, Rosengren L. Cerebrospinal fluid Abeta42 is reduced in multiple system atrophy but normal in Parkinson's disease and progressive supranuclear palsy. Mov Disord2003;18:186–190.
    Direct Link:
  8. Mollenhauer B, Trenkwalder C. Neurochemical biomarkers in the differential diagnosis of movement disorders. Mov Disord2009;24:1411–1426.
    Direct Link:
  9. Compta Y, Marti MJ, Ibarretxe-Bilbao N, et al. Cerebrospinal tau, phospho-tau, and beta-amyloid and neuropsychological functions in Parkinson's disease. Mov Disord2009;24:2203–2210.
    Direct Link:
  10. Alves G, Bronnick K, Aarsland D, et al. CSF amyloid-beta and tau proteins, and cognitive performance, in early and untreated Parkinson's disease: the Norwegian ParkWest study. J Neurol Neurosurg Psychiatry2010;81:1080–1086.
  11. Montine TJ, Shi M, Quinn JF, et al. CSF Abeta(42) and tau in Parkinson's disease with cognitive impairment. Mov Disord2010;25:2682–2685.
    Direct Link:
  12. Apostolova LG, Hwang KS, Andrawis JP, et al. 3D PIB and CSF biomarker associations with hippocampal atrophy in ADNI subjects. Neurobiol Aging2010;31:1284–1303.
  13. Chou YY, Lepore N, Saharan P, et al. Ventricular maps in 804 ADNI subjects: correlations with CSF biomarkers and clinical decline. Neurobiol Aging2010;31:1386–1400.
  14. Fagan AM, Head D, Shah AR, et al. Decreased cerebrospinal fluid Abeta(42) correlates with brain atrophy in cognitively normal elderly. Ann Neurol2009;65:176–183.
    Direct Link:
  15. Apostolova L, Alves G, Hwang KS, et al. Hippocampal and ventricular changes in Parkinson's disease mild cognitive impairment. Neurobiol Aging2012;33:2113–2124.
  16. Bohnen NI, Albin RL. White matter lesions in Parkinson disease. Nat Rev Neurol2011;7:229–236.
  17. Lee SJ, Kim JS, Yoo JY, et al. Influence of white matter hyperintensities on the cognition of patients with Parkinson disease. Alzheimer Dis Assoc Disord2010;24:227–233.
  18. Beyer MK, Aarsland D, Greve OJ, Larsen JP. Visual rating of white matter hyperintensities in Parkinson's disease. Mov Disord2006;21:223–229.
    Direct Link:
  19. Dalaker TO, Larsen JP, Dwyer MG, et al. White matter hyperintensities do not impact cognitive function in patients with newly diagnosed Parkinson's disease. Neuroimage2009;47:2083–2089.
  20. Selnes P, Blennow K, Zetterberg H, et al. Effects of cerebrovascular disease on amyloid precursor protein metabolites in cerebrospinal fluid. Cerebrospinal Fluid Res2010;7:10.
  21. Aarsland D, Bronnick K, Larsen JP, Tysnes OB, Alves G. Cognitive impairment in incident, untreated Parkinson disease: the Norwegian ParkWest study. Neurology2009;72:1121–1126.
  22. Gelb DJ, Oliver E, Gilman S. Diagnostic criteria for Parkinson disease. Arch Neurol1999;56:33–39.
  23. McKeith IG, Dickson DW, Lowe J, et al. Diagnosis and management of dementia with Lewy bodies: third report of the DLB Consortium. Neurology2005;65:1863–1872.
  24. Alves G, Muller B, Herlofson K, et al. Incidence of Parkinson's disease in Norway: the Norwegian ParkWest study. J Neurol Neurosurg Psychiatry2009;80:851–857.
  25. Fahn S, Elton RL. Unified Parkinson's Disease Rating Scale. In: Fahn S, Marsden CD, Calne DM, Lieberman A, eds. Recent Development in Parkinson's Disease. Florham Park, NJ: MacMillan Health Care Information; 1987:153–163.
  26. Hoehn MM, Yahr MD. Parkinsonism: onset, progression, and mortality. Neurology1967;17:427–442.
  27. Folstein MF, Folstein SE, McHugh PR. “Mini-mental state.” A practical method for grading the mental state of patients for the clinician. J Psychiatr Res1975;12:189–198.
  28. Delis DC, Kramer JHEK, Ober BA. CVLT II. California Verbal Learning Test. Second Edition. Adult version. San Antonio, TX: The Psychological Corporation: Harcourt Assessment; 2000.
  29. WarringtonEK, James M. The Visual Object and Space Perception Battery. Bury St Edmunds: Thames Valley Test Company; 1991.
  30. BentonAL, Varney NR, Hamsher KD. Visuospatial judgment. A clinical test. Arch Neurol1978;35:364–367.
  31. Stroop JR. Studies in interference of serial verbal reactions. J Exper Psychol1935;18:643–662.
  32. Shattuck DW, Sandor-Leahy SR, Schaper KA, Rottenberg DA, Leahy RM. Magnetic resonance image tissue classification using a partial volume model. Neuroimage2001;13:856–876.
  33. Collins DL, Neelin P, Peters TM, Evans AC. Automatic 3D intersubject registration of MR volumetric data in standardized Talairach space. J Comput Assist Tomogr1994;18:192–205.
  34. Morra JH, Tu Z, Apostolova LG, Green AE, Toga AW, Thompson PM. Automatic subcortical segmentation using a contextual model. Med Image Comput Comput Assist Interv2008;11:194–201.
  35. Chou YY, Lepore N, deZubicaray GI, et al. Automated ventricular mapping with multi-atlas fluid image alignment reveals genetic effects in Alzheimer's disease. Neuroimage2008;40:615–630.
  36. Thompson PM, Hayashi KM, DeZubicaray GI, et al. Mapping hippocampal and ventricular change in Alzheimer disease. Neuroimage2004;22:1754–1766.
  37. Smith SM, Zhang Y, Jenkinson M, et al. Accurate, robust, and automated longitudinal and cross-sectional brain change analysis. Neuroimage2002;17:479–489.
  38. Belsey DA, Kuh E, Welsch R. Regression Diagnostics: Identifying Influential Data and Sources of Cillinearity. New York, NY: Wiley; 1980.
  39. Gabelle A, Roche S, Geny C, et al. Correlations between soluble alpha/beta forms of amyloid precursor protein and Abeta38, 40, and 42 in human cerebrospinal fluid. Brain Res1357:175–183.
  40. Blennow K, deLeon MJ, Zetterberg H. Alzheimer's disease. Lancet2006;368:387–403.
  41. Mulugeta E, Londos E, Ballard C, et al. CSF amyloid beta38 as a novel diagnostic marker for dementia with Lewy bodies. J Neurol Neurosurg Psychiatry2011;82:160–164.
  42. Carlson NE, Moore MM, DameA, et al. Trajectories of brain loss in aging and the development of cognitive impairment. Neurology2008;70:828–833.
  43. Fjell AM, Walhovd KB, Fennema-Notestine C, et al. Brain atrophy in healthy aging is related to CSF levels of A{beta}1-42. Cereb Cortex2010;20:2069–2079.
  44. Tosun D, Schuff N, Truran-Sacrey D, et al. Relations between brain tissue loss, CSF biomarkers, and the ApoE genetic profile: a longitudinal MRI study. Neurobiol Aging2010;31:1340–1354.
  45. Siderowf A, Xie SX, Hurtig H, et al. CSF amyloid {beta} 1-42 predicts cognitive decline in Parkinson disease. Neurology2010;75:1055–1061.
  46. Xiong Y, Mok V, Wong A, et al. The age-related white matter changes scale correlates with cognitive impairment. Eur J Neurol2010;17:1451–1456.
    Direct Link:
  47. Bohnen NI, Muller ML, Zarzhevsky N, et al. Leucoaraiosis, nigrostriatal denervation and motor symptoms in Parkinson's disease. Brain2011;134:2358–2365.
  48. Wider C, Ross OA, Nishioka K, et al. An evaluation of the impact of MAPT, SNCA, and APOE on the burden of Alzheimer's and Lewy body pathology. J Neurol Neurosurg Psychiatry2012;83:424–429.
  49. Wen HM, Baum L, Cheung WS, et al. Apolipoprotein E epsilon4 allele is associated with the volume of white matter changes in patients with lacunar infarcts. Eur J Neurol2006;13:1216–1220.
    Direct Link:
  50. Kurz MW, Dekomien G, Nilsen OB, Larsen JP, Aarsland D, Alves G. APOE alleles in Parkinson disease and their relationship to cognitive decline: a population-based, longitudinal study. J Geriatr Psychiatry Neurol2009;22:166–170.