Skip to main content

Full text of "Genetic impact on cognition and brain function in newly diagnosed Parkinson's disease: ICICLE-PD study."

See other formats


doi:10.1093/brain/awu201 Brain 2014: 137; 2743-2758 I 2743 

BRAIN 

A JOURNAL OF NEUROLOGY 

Genetic impact on cognition and brain function 
in newly diagnosed Parkinson's disease: 
ICICLE-PD study 

Cristina Nombela, 1 '* James B. Rowe, 2,3,4 '* Sophie E. Winder-Rhodes, 1 Adam Hampshire, 5 
Adrian M. Owen, 6,7 David P. Breen, 1 Gordon W. Duncan, 8 Tien K. Khoo, 9 Alison J. Yarnall, 8 
Michael J. Firbank, 8 Patrick F. Chinnery, 10 Trevor W. Robbins, 4 John T. O'Brien, 11 
David J. Brooks, 12,13 David J. Burn, 8 the ICICLE-PD study group and Roger A. Barker 1 



1 John van Geest Centre for Brain Repair, University of Cambridge, Cambridge, UK 

2 Department of Clinical Neurosciences, University of Cambridge, Cambridge, UK 

3 Medical Research Council, Cognition and Brain Sciences Unit, Cambridge, UK 

4 Behavioural and Clinical Neuroscience Institute, University of Cambridge, UK 

5 Computational, Cognitive and Clinical Neuroscience Laboratory, Imperial College London, London, UK 

6 Brain and Mind Institute, University of Western Ontario, London, Canada 

7 Department of Psychology, University of Western Ontario, London, Canada 

8 Institute for Ageing and Health, Newcastle University, Newcastle, UK 

9 Griffith Health Institute and School of Medicine, Griffith University, Gold Coast, Australia 

10 Institute of Genetic Medicine, Newcastle University, Newcastle, UK 

11 Department of Psychiatry, University of Cambridge, Cambridge, UK 

12 Imperial College London, London, UK 

13 Department of Clinical Medicine, Positron Emission Tomography Centre, Aarhus University, Denmark 

*These authors contributed equally to this work. 

Correspondence to: Cristina Nombela, 

John van Geest Centre for Brain Repair, 

University of Cambridge, Cambridge Biomedical Campus, 

CB2 OPY, UK 

E-mail: dra.cristinanombela@gmail.com 

See Dujardin (doi:10.1093/brain/awu218) for a scientific commentary on this article. 

Parkinson's disease is associated with multiple cognitive impairments and increased risk of dementia, but the extent of these deficits 
varies widely among patients. The ICICLE-PD study was established to define the characteristics and prevalence of cognitive change 
soon after diagnosis, in a representative cohort of patients, using a multimodal approach. Specifically, we tested the 'Dual 
Syndrome' hypothesis for cognitive impairment in Parkinson's disease, which distinguishes an executive syndrome (affecting the 
frontostriatal regions due to dopaminergic deficits) from a posterior cortical syndrome (affecting visuospatial, mnemonic and 
semantic functions related to Lewy body pathology and secondary cholinergic loss). An incident Parkinson's disease cohort 
(n = 168, median 8 months from diagnosis to participation) and matched control group (n = 85) were recruited to a neuroimaging 
study at two sites in the UK. All participants underwent clinical, neuropsychological and functional magnetic resonance imaging 
assessments. The three neuroimaging tasks (Tower of London, Spatial Rotations and Memory Encoding Tasks) were designed to 
probe executive, visuospatial and memory encoding domains, respectively. Patients were also genotyped for three polymorphisms 
associated with cognitive change in Parkinson's disease and related disorders: (i) rs4680 for COMT Val158Met polymorphism; (ii) 
rs9468 for MAPT H1 versus H2 haplotype; and (iii) rs429358 for APOE-e2, 3, 4. We identified performance deficits in all three 



Received January 20, 2014. Revised May 11, 2014. Accepted June 14, 2014. Advance Access publication July 30, 2014 
© The Author (2014). Published by Oxford University Press on behalf of the Guarantors of Brain. 

This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.Org/licenses/by/3.0/), which permits unrestricted reuse, 
distribution, and reproduction in any medium, provided the original work is properly cited. 



2744 I Brain 2014: 137; 2743-2758 



C. Nombela et al. 



cognitive domains, which were associated with regionally specific changes in cortical activation. Task-specific regional activations 
in Parkinson's disease were linked with genetic variation: the rs4680 polymorphism modulated the effect of levodopa therapy on 
planning-related activations in the frontoparietal network; the MAPT haplotype modulated parietal activations associated with 
spatial rotations; and APOE allelic variation influenced the magnitude of activation associated with memory encoding. This study 
demonstrates that neurocognitive deficits are common even in recently diagnosed patients with Parkinson's disease, and that the 
associated regional brain activations are influenced by genotype. These data further support the dual syndrome hypothesis of 
cognitive change in Parkinson's disease. Longitudinal data will confirm the extent to which these early neurocognitive changes, 
and their genetic factors, influence the long-term risk of dementia in Parkinson's disease. The combination of genetics and func- 
tional neuroimaging provides a potentially useful method for stratification and identification of candidate markers, in future clinical 
trials against cognitive decline in Parkinson's disease. 

Keywords: Parkinson's disease; cognition; functional MRI; genetics 

Abbreviations: DLPFC = dorsolateral prefrontal cortex; ICICLE-PD = Incidence of Cognitive Impairment in Cohorts with Longitudinal 
Evaluation - Parkinson's Disease; LEDD = levodopa equivalent daily dose; MMSE = Mini-Mental State Examination; 
MOCA = Montreal Cognitive Assessment; NART = National Adult Reading Test 



Introduction 



Parkinson's disease was often considered to be primarily a motor 
disorder although dementia has long been recognized as a feature 
of the condition (Gowers, 1893). More recently the early onset 
and heterogeneity of cognitive impairments in Parkinson's disease 
have been recognized, even in the absence of dementia 
(Muslimovic et al., 2005). The cognitive deficits of Parkinson's 
disease affect visuospatial, attentional, executive and memory 
functions (Janvin ef al., 2006; Hely et al., 2008; Elgh et al., 
2009; Aarsland and Kurz, 2010; Pedersen et al., 2013) due to 
the combination of abnormal neurotransmitter systems (e.g. dopa- 
minergic and cholinergic) and both cortical and subcortical Lewy 
body pathology (Kehagia et al., 2010). We have proposed two 
facets of cognitive deficits in Parkinson's disease, in a 'Dual 
Syndrome' hypothesis: (i) changes in dopaminergic transmission 
through the corticostriatal networks leading to deficits in planning, 
working memory, response inhibition and attentional control; and 
(ii) posterior cortical Lewy body pathology and secondary cholin- 
ergic loss affecting visuospatial, mnemonic and semantic functions 
(Kehagia et al., 2013). 

Cognitive impairments are present at diagnosis in a significant 
proportion of affected individuals with between 24% and 62% of 
newly diagnosed patients with Parkinson's disease having deficits 
in executive (e.g. Tower of London Task), visuospatial (e.g. Spatial 
Rotations Task) or memory (e.g. Memory Encoding Task) perform- 
ance compared to healthy controls (Foltynie et al., 2004a; 
Williams-Gray et al., 2007a; Elgh et al., 2009; Yarnall et al., 
2014). By 3 years after diagnosis up to 57% of patients have 
frontostriatal or visuospatial deficits and 10% have Parkinson's 
disease dementia (Williams-Gray et al., 2007a) rising to 17% by 
5 years (Williams-Gray et al., 2007a), 26% by 8 years (Aarsland 
et al., 2003), 46% by 10 years (Williams-Gray et al., 2013) and 
83% by 20 years (Hely et al., 2008). Thus only -15% of patients 
with Parkinson's disease remain cognitively intact in the long 
term (Aarsland et al., 2011). It is therefore important to ascertain 
what determines cognitive decline, and how it relates to subse- 
quent dementia. 



Genetic factors are implicated in Parkinson's disease cognitive 
impairments (Goldberg and Weinberger 2004; Morley et al., 
2012). For example, catechol-O-methyl transferase (COMT) is 
involved in cortical dopamine degradation. A common polymorph- 
ism at codon 158 (Val158Met) affects its enzymatic activity 4-fold 
(Chen ef al., 2004), and influences executive task performance in 
healthy individuals (Stokes et al., 2011; Fallon et al., 2013) and 
patients with Parkinson's disease (Foltynie et al., 2004b; Williams- 
Gray et al., 2007b). The way in which the polymorphism affects 
cortical dopamine levels suggests that either too little or too much 
dopamine worsens task performance, in accordance with an in- 
verted U-shaped curve (Goldberg and Weinberger, 2004; 
Williams-Gray et al., 2007b, 2009b; Rowe et al., 2008). Our hy- 
pothesis was that the COMT polymorphism would affect dopa- 
mine-dependent working memory and planning systems in 
frontostriatal networks, and introduce a non-linear (U-shaped) re- 
lationship between neurocognitive function and levodopa dose. 

A second gene linked to cognitive performance and dementia in 
Parkinson's disease is the microtubule-associated protein tau 
(MAPT). The MAPT haplotype H1 (versus H2) not only predis- 
poses to Parkinson's disease but also Parkinson's disease dementia 
(Goris et al., 2007), possibly by altering the cortical expression of 
4- versus 3-repeat isoforms of tau (Williams-Gray et al., 2009a). 
Our hypothesis was that fronto-parietal systems for visuospatial 
function, related to dementia with Parkinson's disease, would be 
relatively preserved in H2 carriers versus H1 carriers. 

Finally, apolipoprotein E (APOE) has been proposed to alter the 
risk of Parkinson's disease dementia (Chen et al., 2004; Huang 
et al., 2006; Goris et al., 2007; Williams-Gray et al., 2009b; 
Chung et al., 2012; Gomperts et al., 2012, 2013) as well as 
being a risk factor for Alzheimer's disease, even if it does not 
significantly alter the risk of developing Parkinson's disease with- 
out Parkinson's disease dementia (Peplonska et al., 2013; 
Multhammer et al., 2014). APOE has three allelic variants 
(APOE2, 3 and 4), and APOE4 carries the highest risk for 
Alzheimer's dementia (Corder et al., 1993) with APOE2 carrying 
the lowest. Our hypothesis was that memory systems centred 



Early diagnosis of cognitive decline in Parkinson's disease 



Brain 2014: 137; 2743-2758 I 2745 



on the temporal lobe and hippocampus in particular would be 
most impaired in APOE4 carriers. 

In this study we examined the impact of these genetic factors 
on cognitive function in a large cohort of patients with newly 
diagnosed Parkinson's disease. We used functional AARI to meas- 
ure regional brain functions during a range of tasks that encom- 
pass the main cognitive deficits reported in Parkinson's disease 
(Williams-Gray ef al., 2007a; Barone ef al., 2011; Ekman et al., 
2012; Hampshire et al., 2012; Winder-Rhodes et al., 2013; 
Nagano-Saito et al., 2014; Yarnall et al., 2014). The results of 
the comprehensive neuropsychological assessment undertaken by 
the participants are published elsewhere (Yarnall et al. 2014). This 
neuroimaging study focuses on a set of three tasks that provide 
robust experimental models of important cognitive functions af- 
fected by Parkinson's disease, including planning and working 
memory (Tower of London Task), visuospatial function (Spatial 
Rotations Task) and memory (abstract image encoding and recog- 
nition). We sought to define how the early cognitive deficits in 
newly diagnosed patients with Parkinson's disease map onto 
changes in brain activation, and how these activations in patients 
varied as a function of the common genetic variations in COMT, 
MAPT and APOE. 

Materials and methods 

Subjects 

The Incidence of Cognitive Impairment in Cohorts with Longitudinal 
Evaluation - Parkinson's Disease (ICICLE-PD) study recruited a cohort 
of 219 patients with incident Parkinson's disease from community and 
outpatient clinics at the John van Geest Centre for Brain Repair, 
Cambridge, UK (n = 49) and Parkinson's Disease clinics in 
Newcastle-upon-Tyne/Gateshead, UK (n = 119) [from the ICICLE- 
PD cohort, 169 patients agreed to participate in the functional MRI 
study (separate day within 4 months from initial assessment)]. We 
used the United Kingdom Parkinson's Disease Society (UKPDS) Brain 
Bank diagnostic criteria (Hughes et al., 2002), with reconfirmation 
after 18 months, to diagnose Parkinson's disease. Full inclusion and 
exclusion criteria are described in Yarnall et al. (2014). In brief, exclu- 
sion criteria were: parkinsonism diagnosed before the onset of the 
incidence study; insufficient working knowledge of English to perform 
the neuropsychological assessment; dementia at presentation [defined 
as Mini-Mental State Examination (MMSE) score < 24 or Diagnostic 
and Statistical Manual of Mental Disorders, Fourth Edition (DSM IV) 
criteria for dementia or Movement Disorder Society criteria for demen- 
tia]; lack of mental capacity to give informed consent under UK legis- 
lation; history of parkinsonism following the onset of cognitive 
impairment; history or examination suggestive of dementia with 
Lewy bodies, multiple system atrophy, progressive supranuclear 
palsy, repeated strokes or stepwise progression of symptoms indicative 
of 'vascular parkinsonism'; and, exposure to dopamine receptor block- 
ing agents at the onset of symptoms. 

Unrelated age- and sex-matched controls were recruited from the 
MRC Cognition and Brain Sciences Unit volunteer panel in Cambridge, 
UK (n = 50) and from community sources at the Newcastle site 
(n = 35). The Local Research Ethics Committee approved the study, 
performed according to the Declaration of Helsinki, with all partici- 
pants providing written consent. 



Participants undertook a battery of standardized clinical and neuro- 
psychological assessments including: the Unified Parkinson's Disease 
Rating Scale (MDS-UPDRS) (Goetz et al., 2008); MMSE (Folstein 
ef al., 1975); Montreal Cognitive Assessment (MOCA) (Nasreddine 
ef al., 2005); National Adult Reading Test (NART) (Nelson and 
O'Connell, 1978) estimate of premorbid IQ; verbal fluency for 
words starting with the letter P/F (60s) (Benton, 1968) and semantic 
fluency for animals (90s) (Goodglass ef al., 1972). Levodopa equiva- 
lent daily dose (LEDD) value was calculated according to Tomlinson 
ef al. (2010). Patients were assessed ON their usual dopaminergic 
medication (Williams-Gray ef al., 2007b). Additional neuropsycho- 
logical tests and the Geriatric Depression Scale-15 for depression are 
reported by Yarnall ef al. (2014). 

DNA was extracted from peripheral blood using standard phenol/ 
chloroform techniques. Genotyping for rs4680 (COMT Val158Met), 
rs9468 (MAPT H1 versus H2 haplotype) and rs429358 plus rs7412 
(APOE genotype 1-4) was performed using an allelic discrimination 
assay and run on an HT7000 detection system (Applied Biosystems). 

Experimental design 

On the scanning day participants were trained for 30min to perform 
the tasks and practice keyboard responses. Participants lay supine in 
the MRI scanner, with auditory protection and head fixation using 
foam-rubber pads. Stimuli were back-projected onto a screen, and 
viewed via a mirror on the headcoil. Three functional MRI experiments 
were performed. 

Tower of London Task 

We used a 'one-touch' modified version of the Tower of London Task 
(Shallice, 1982; Williams-Gray ef al., 2007b), as a model of prefrontal 
executive function in Parkinson's disease (Rowe ef al., 2001; Lewis 
ef al., 2003). The task presented with two racks of three coloured 
balls in different pockets. Participants determined the minimum 
number of moves to rearrange the balls to match the racks (Owen 
ef al., 1995; Baker ef al., 1996). The control task required one to 
count the difference in the number of balls between the two displays. 
Responses were made with a right hand button-box. The paradigm 
lasted for 10min46s, with intermixed presentations of experimental 
and control items, cued on the screen before each trial as 'plan' or 
'substract', respectively, with three levels of difficulty (levels 2, 3 and 4 
according to the number of moves or number of differences in the ball 
count-dependent variable 2) and intertrial intervals of 5-1 5 s. No feed- 
back was provided. The dependent variables were the latency of re- 
sponse (including mainly the thinking time plus a small contribution 
from the motor response time for the one-touch version of this task) 
and accuracy. 

Spatial Rotations Task 

Spatial impairments in Parkinson's disease are independent of execu- 
tive skills (Cronin-Golomb and Braun, 1997) and depend on the integ- 
rity of posterior parietal cortex and a fronto-parietal network (Cohen 
ef al., 1996; Zacks, 2008). We used this task to probe posterior cor- 
tical function, analogous to earlier studies of Alzheimer's disease 
(Jacobs ef al., 2012). Each item consisted of a reference pattern 
(5x5 grid, top) and four response patterns (bottom). One response 
pattern corresponded to the reference, after rotation by ±90° or 
180°. Three randomized levels of difficulty (levels dependent variable 
2, 3, 4) were defined by the complexity of the pattern. The control 
condition required matching the reference and unrotated response 



2746 I Brain 2014: 137; 2743-2758 



C. Nombela et al. 



patterns. The task lasted 10min46s, with alternate experimental and 
control items (cued on screen by 'rotate' or 'match', respectively) and 
intervening rest intervals of 5-15 s (cued by 'rotate' or 'match' for 
experimental and control items, respectively). No feedback was pro- 
vided. The dependent variables were the latency of response and 
number of accurate responses. 

Memory Encoding Task 

Memory deficits in Parkinson's disease are most related to encoding 
rather than impairments in retention or retrieval processes (Bronnick 
et al., 2011). Encoding deficits generally have a different aetiology to 
executive impairments (Kehagia et al., 2010), and are linked to hip- 
pocampal function (Aarsland ef al., 2011). The Memory Encoding Task 
was selected accordingly. Subjects viewed abstract pictures organized 
in seven blocks (displayed alternatively with intertrial intervals of 5- 
15s) of six images for 4s each, with a 1 s cross-hair fixation between, 
and were asked to memorize them. Participants saw 30 different 
images in the scanner; 18 of them appeared once (exposition fol- 
d = once), 12 appeared twice (exposition fold = twice). After scanning 
(20-min delay), participants completed a recognition test of these 30 
images, intermixed with 32 lures. They reported whether they had 
seen each picture before by two-alternate forced-choice button re- 
sponses. No feedback was provided. The dependent variables were 
the response latency, the number of accurate responses and the d' 
score of hit rate versus false alarms. 

MRI acquisition processing and analysis 

A Siemens TIM Trio 3 T scanner (Siemens Medical Systems) was used 
at one site and a 3 T Philips Intera Achieva scanner at the other. 
Participants underwent high resolution magnetization prepared rapid 
gradient echo scanning (MP-RAGE: repetition time = 2250 ms, echo 
time = 2.98 ms, flip angle = 9°, inversion time = 900 ms, 
256 x 256 x 192 isotropic 1 mm voxels). During functional MRI, 
'BOLD-sensitive' T 2 * weighted echo-planar images were acquired 
(repetition time = 2000 ms, echo time = 30 ms, flip angle = 78°, 
32 x 3 mm sequential descending slices, in-plane resolution 
3 x 3 mm, slice separation 0.75 mm) with 320 volumes for Tower of 
London and Spatial Rotations Tasks and 250 volumes for Memory 
Encoding excluding 10 initial dummy volumes. 

MRI data were processed using Statistical Parametric Mapping 
(SPM8, www.fil.ion.ucl.ac.uk/spm). Functional MRI data were con- 
verted from DICOM to NIFTII format, spatially realigned to the first 
image, and corrected for acquisition delay by sine interpolation with 
reference to the middle slice. The mean functional MRI volume and 
MP-RAGE were co-registered using mutual information, and the MP- 
RAGE segmented and normalized to the Montreal Neurological 
Institute (MNI) Ti template. The normalization parameters were 
applied to all spatiotemporally realigned functional images and 
upsampled to 2x2x2mm, before smoothing with an isotropic 
Gaussian kernel with full-width half-maximum of 5 mm. 

Individual analysis of all three tasks was modelled with the stimulus 
onset times and durations per item. First level general linear modelling 
included six regressors: stimuli were modelled as a boxcar function per 
condition (experimental or control condition) and level of difficulty (2, 
3 and 4) for Tower of London and Spatial Rotations Task whilst 
Memory Encoding was modelled including all stimulus category (pic- 
tures seen once and pictures seen twice, independently of encoding 
success). A parametric modulator for each trial, value 1 / reaction 
time, was included separately for each trial type and condition. Error 
trials were modelled separately. Regressors were convolved with a 



canonical haemodynamic response function and its first temporal de- 
rivative. Six rigid-body motion correction parameters were included as 
nuisance covariates. Contrast images were extracted for individuals 
and entered into a second level region of interest analyses. For the 
Tower of London and Spatial Rotations tasks, subjects were excluded if 
they performed below threshold, as defined by two criteria: (i) long 
thinking time to solve an item, defined as a latency of response >17s 
[response time average + 2.5 standard deviations (SD) in the sample]; 
(ii) <1 correct answer per type of item (control and experimental task) 
and level of difficulty (2, 3 and 4 in both Tower of London and Spatial 
Rotations Tasks). 

Functional MRI data were analysed by region of interest analysis at 
the group level (see 'Results' section and Fig. 1) and corrected for 
multiple comparisons [2-tailed significance level was set at P < 0.05 
cluster-based false discovery rate (FDR)]. Then, region of interest ana- 
lyses were performed using individual measures of averaged effect size 
('beta' parameter estimates) for each region of interest, extracted 
using MarsBaR (MARSeille Boite A Region d'lnteret) toolbox (http:// 
marsbar.sourceforge.net). 

The independent a priori specification of regions of interest was 
based on previous studies of Tower of London and Spatial Rotations 
Tasks (Williams-Gray et al., 2007b; Hampshire et al., 2012). Beta 
values in eight a priori regions of interest were extracted: right dorso- 
lateral prefrontal cortex (DLPFC), left DLPFC, right frontopolar cortex, 
bilateral posterior parietal cortex and precuneus. Additionally, caudate 
nuclei (right caudate: x= -10, y=15, z = 2; left caudate: x= -10, 
y = 15, z = 2, 10mm radius sphere) were included because of their 
high relevance within the frontostriatal network in mediating executive 
functions in healthy controls (Owen et al., 1996) and Parkinson's dis- 
ease (Lewis ef al., 2003). A task-specific region of interest template for 
Memory Encoding Task (Hampshire et al., 2012) was based on inde- 
pendent 60 healthy control data: bilateral hippocampus (right 




Figure 1 Statistical parametric maps contrasting activity in 
active versus baseline conditions rendered into a canonical brain 
in standard anatomic space. (A) Activity during planning minus 
control condition on Tower of London Task across all groups. 
(B) Activity during rotations minus baseline on Spatial Rotations 
Task across all groups. (C) Activity during encoding (pictures 
seen once) minus baseline on the Encoding Memory Task across 
all groups. Figures show areas of signal change above a 
threshold of P = 0.05 after FDR correction for the whole brain 
volume. 



Early diagnosis of cognitive decline in Parkinson's disease Brain 2014: 137; 2743-2758 I 2747 



Table 1 Demographic and clinical variables for participants in each group and site 





Contml Sitp 1 

V_*_l I I LI \J I JILC 1 


Pflrkin^nn'^ 

disease Site 1 


fontrnl ^Jitp 5 


ParkinQnn'Q 

disease Site 2 


P Cirnim 

r v_r i uu u 


P Site 


P flrmm v litp 

interaction 


Gender (M/F) 


27/22 


25/24 


17/18 


57/45 


0.549 


0.407 


0.697 


Age (years) 


63.83 ± 5.8 


65.36 ± 7.9 


66.23 ± 8.4 


64.81 ±11.1 


0.84 


0.94 


0.77 


Years of education 


15.30 ± 6.1 


14.02 ± 2.6 


13.1 ± 3.9 


13.03 ± 3.8 


0.001 


0.001 


0.009 


MMSE 


29.48 ± 0.7 


29.10 ± 0.9 


29.16 ± 1.05 


28.94 ± 1.1 


0.193 


0.193 


0.074 


MOCA 


27.69 ± 1.7 


26.06 ± 2.2 


26 ± 5.9 


26.07 ± 2.7 


0.278 


0.183 


0.146 


NART 


121.85 ± 5.2 


114.58 ± 8.6 


113.5 ± 25.5 


116.69 ± 9.4 


0.448 


0.845 


0.639 


Semantic fluency 


18.53 ± 5.4 


14.38 ±4.2 


12.37 ± 5.7 


11.33 ±4.6 


0.054 


0.001 


0.493 


Category fluency 


31.95 ± 7.5 


22.04 ± 6.2 


23.12 ± 8.4 


21.21 ± 8.1 


0.001 


0.001 


0.001 


MDS-UPDRS-III 




29.28 ± 11.02 




25.36 ± 10.7 




0.06 




LEDD 




484.56 ± 369 




167.69 ± 129 




0.001 




Duration (months) Mean 


21.1 ± 13.2 




6.11 ±4.7 




0.001 






Median 


19.2 ± 13.2 




4.7 ±4.7 




0.001 





P-values are presented separately for comparisons of group (Parkinson's disease versus control), site (1 versus 2) and the interaction between site and disease, using 
ANOVAs (except chi-squared tests of gender). Data are shown without correction for multiple comparisons (values in bold are significant after Bonferroni correction). In 
view of the skewed distribution of symptom duration (Shapiro-Wilk test P < 0.001), the median values for duration are also show (*Mann-Whitney test P-value). 



hippocampus, left hippocampus), left superior parietal gyrus, right 
inferior frontal gyri pars triangularis and pars opercularis, left inferior 
frontal gyrus, left occipital and a large region of interest including 
posterior temporo-parieto-occipital area. 

The region of interest and behavioural analyses used SPSS (version 
21). The first set of analyses used initial parsimonious ANOVAs in 
which categorical variables were run, including: region of interest, 
task condition and difficulty as within-subject factors and disease 
group (patients versus controls) and site (Cambridge versus 
Newcastle) as between-subject factors. However, several continuous 
cognitive and clinical variables have been shown in previous studies to 
affect brain function (e.g. age, disease progression, levodopa doses) 
(Williams-Gray etal., 2007a; Barone et a/., 2011; Taylor etal., 2013). 
We therefore ran secondary ANCOVAs to control for the possible 
effects of these variables. As there were many candidate variables, 
the optimal approach we used was a stepwise multiple linear regres- 
sion approach, progressively excluding variables, variables which ex- 
plained minimal variance. We started each model with entry variables 
of: age, sex, years of education, MMSE, MOCA, NART, letter and 
category fluency UPDRS-III, LEDD and duration of disease. We 
report both the significant contributory variables/covariates and the 
percentage of variance they explained. 



and an interaction between disease and site for category fluency 
[F(1, 1 72) = 10.735; P< 0.001], with relatively higher scores in 
controls at Site 1. Patients at Site 1 had longer duration of disease 
[F(1, 107) = 100.624; P< 0.001], and were on a higher dose of 
levodopa [F(1 ,107) = 48.402; P< 0.001] but were similar in their 
motor severity (UPDRS-III subscale). 

Table 2 compares key clinical and demographic markers for par- 
ticipants in the main ICICLE-PD study and those completing the 
functional MRI studies investigation, confirming that there were 
no significant differences. Table 3 shows the COMT, AAAPT and 
APOE genotype distributions among patients with Parkinson's 
disease. 

Tower of London Task 

Across the two sites the number of patient participants 
(Cambridge/Newcastle) completing the Tower of London task 
was "patient = 1 1 7 (41/76) and the number of healthy controls 
"control = 69 (43/16). Behavioural performance is illustrated in 
Fig. 2. 



Results 

Demographics and neuropsychology 

Gender, age, MOCA and MMSE scores were matched between 
groups and sites (Table 1), with no significant interactions be- 
tween these factors and site. For years of education there was a 
main effect of site [F(1 ,172) = 23.431 ; P< 0.001] with fewer 
years at Site 2, and a main effect of disease group 
[F(1,172) = 19.760; P < 0.001] with controls having spent longer 
in formal education, but no significant interaction. There were 
corresponding differences between groups (higher score in 
controls) and sites (higher scores at Site 1) in terms of category 
fluency [disease: F(1, 172) = 15.544; P< 0.001, site: F(1,172) = 
12.392; P< 0.001] and letter fluency scores [disease: 
F(1 ,172) = 3.754; P < 0.054, site: F(1 ,172) = 10.735; P< 0.001] 



Latency of response 

The within subject factors of condition [control versus plan, 
F(1, 186) = 497.432; P< 0.001] and difficulty [F(2,372) = 63.762; 
P< 0.001] were significant with an interaction effect between 
condition and difficulty on latency of response [F(2,370) = 
73.744; P< 0.001], confirming that more difficult planning 
items required longer response times. Repeated-measures 
ANOVA confirmed an effect of site [F(1, 1 86) = 7.278, 
P < 0.008, shorter at Site 1]. There was no main effect of disease 
(F < 1) or interaction effect between disease and site (F < 1). 

The addition of between subject demographic and neuropsy- 
chological variables (age, years of education, MMSE, MOCA, 
NART, semantic and category fluency scores) in separate re- 
peated-measures ANCOVAs revealed a shorter response latency 
in younger subjects [F(1, 1 86) = 6.090; P< 0.015], with a higher 
number of years of education [F(1 ,186) = 4.033, P < 0.046], 



2748 I Brain 2014: 137; 2743-2758 



C. Nombela et al. 



Table 2 Clinical and demographic values of the ICICLE- 
Parkinson's disease (Yarnall et al., 2014) cohort and sub- 
group participating in this functional MRI study 





ICICLE-PD 


Functional 






MRI-ICICLE 


n 


219 


141 


Mean age 


65.9 


65.08 


MDS UPDRS-IM severity 


28.32 


27.34 


MOCA 


25.70 


26.06 


Maleifemale 


140:79 


82:59 



No differences were significant (x 2 and f-test contrasts between groups as 
appropriate). 



higher MMSE [F(1, 186) = 19.152; P< 0.001], higher MOCA 
[F(1,186) = 5.378; P < 0.021] and greater letter fluency 
[F(1, 186) = 48.06; P < 0.03]. No significant effects were found 
for NART [F(1 ,186) = 1.290; not significant] or category fluency 
[F(1,186) = 3.551; not significant]. 

In a separate analysis of patients with Parkinson's disease only, the 
addition of disease-specific between -subject variables (UPDRS-III, LEDD 
and duration) in repeated measures ANCOVA indicated that pa- 
tients with higher UPDRS-III score took marginally longer to respond 
[F(1,117) = 3.827, P< 0.05]. Neither LEDD nor duration had a signifi- 
cant effect. Separate repeated-measures ANOVAs with COMT, 
AAAPT and APOF genotype indicated no effect on latency (F < 1). 

A stepwise multiple regression analysis in patients with 
Parkinson's disease indicated that the MMSE explained significant 
variance in latency in the resulting model [model F(4,117) = 8.395, 
P < 0.004, MMSE f(116) = -2.897, P < 0.004, 6.7% of the vari- 
ance explained r = 0.26]. 

Accuracy 

Task condition [F(1 ,186) = 71 .414; P< 0.001] and difficulty 
[F(2,372) = 47.9.3; P < 0.001] effects were significant with a sig- 
nificant interaction between condition and difficulty [F(2,370) = 
16.473; P< 0.001] confirming that more difficult planning items 
were less likely to be completed. Repeated-measures ANOVA con- 
firmed an effect of site [F(1, 1 86) = 20.586, P< 0.001, higher in 
Site 1] but there was no effect of disease (F < 1) or interaction 
between disease and site on accuracy [F(1, 186) = 2.353, not 
significant]. 

The addition of between subject demographic and neuropsy- 
chological variables in separate repeated-measures ANCOVAs 
showed higher accuracy scores in younger participants 
[F(1, 1 86) = 26.075; P < 0.001], with more years of education 
[F(1, 186) = 9.601; P < 0.002], higher MMSE [F(1 ,186) = 19.331 ; 
P < 0.001], higher MOCA [F(1,186) = 14.011; P < 0.001], higher 
letter fluency score [F(1,186) = 14.725; P < 0.001] and higher cat- 
egory fluency score [F(1,186) = 11.176; P < 0.001]. No significant 
effects were found for NART [F(1,186) = 1.216; not significant]. 

In a separate analysis of patients with Parkinson's disease only, 
the addition of between subject clinical variables revealed no sig- 
nificant effect of UPDRS-III [F(3,1 17) = 3.827; not significant], 
LEDD, duration, COMT, MAPT or APOF genotype (all F < 1). 



Table 3 The distribution of the different polymorphisms of 
the studied genes (COMT, MAPT and APOE) Parkinson's 
disease participants per site 



Genes 


Polymorphism 


Site 1 


Site 2 


Total 


COMT 


Met/Met 


15 


32 


47 




Met/Val 


22 


59 


81 




Val/Val 


7 


30 


37 


MAPT 


H1/H1 


26 


85 


111 




H1/H2 


16 


34 


50 




H2/H2 


2 


2 


4 


APOE 


APOE2 


28 


66 


94 




APOE3 


11 


50 


61 




APOE4 


5 


5 


10 



The stepwise multiple regression in the Parkinson's disease 
group revealed a significant model [F(1, 1 17) = 12.298, P < 0.00 
1] of explanatory variables that included years of education [f(1 
16) = 4.224, P< 0.001], MOCA [f(1 16) = 3.321 , P < 0.001] and 
NART [f(116)= -2.089, P < 0.039] explaining a total 24.5% of 
the variance. 

Functional MRI regional activity 

The activity in regions of interest associated with planning was 
estimated from the contrast of 'all planning tasks minus all con- 
trol conditions'. Repeated-measures ANOVA showed no main 
effect of disease [F(1 ,186) = 1 .353; not significant], site 
[F(1, 1 86) = 1.723; not significant] or interaction (F<1). 
There was a main effect of region of interest [Fig. 3; 
F(7,1309) = 130.196; P < 0.001] and a significant interaction be- 
tween region of interest and disease group [F(7,1 309) = 2.244; 
P < 0.029]. Post hoc contrast indicated that the control 
group had greater activation of the right frontopolar 
[F(1,186) = 6.658; P< 0.011], right caudate [F(1, 1 86) = 11.368; 
P< 0.001] and left caudate [F(1 ,186) = 5.081 ; P < 0.025] com- 
pared to patients. 

In the Parkinson's disease group, there was a trend towards an 
effect of higher UPDRS-III score [F(1,134) = 3.359, P < 0.069] but 
no effect of LEDD (F<1) or duration (F<1) on activation. 
COMT genotype (contrasting Met/Met = 30 and Val/Val = 29), 
site and LEDD intake (median split: high LEDD >275mg = 34, 
low LEDD <275mg = 25) were used as between-subjects factors. 
There was no significant effect of site (F<1), COMT, APOE 
genotypes (F<1), or LEDD [F(1,117) = 1.387; not significant]. 
However, there was a significant interaction between genotype 
and LEDD [LEDD x COMT, F(1 ,1 17) = 5.732; P < 0.020] with 
post hoc f-tests confirming higher beta values in both Met/Met 
homozygotes at low LEDD and Val/Val homozygotes at high 
LEDD compared to Val/Val homozygotes at low LEDD and 
Met/Met homozygotes at high LEDD within the right DLPFC 
[f(58) = 2.530; P < 0.014], left DLPFC [f(58) = 2.050; 
P< 0.045], right frontopolar [f(58) = 2.040; P < 0.008], right 
caudate K(58) = 2.089; P < 0.045] and left caudate 
[f(58) = 2.087; P < 0.040] (Fig. 3). There was no effect of 



Early diagnosis of cognitive decline in Parkinson's disease 



Brain 2014: 137; 2743-2758 I 2749 



Al 

15 
13 

£11 

| 9 

H 7 

5 
3 



ci 

3.5 



g 2.5 



1.5 



Planning task - thinking time 



II 



M2 M3 M4 

Level of dfflculty 



Spatial rotations task - thinking time 




A2 

6 

OB ' 
= 

a 

t i 

L. 

; 



Planning task - Accuracy 




B2 



Hi 



M2 M3 M4 

Level of <fIGculty 

Spatial rotations task - Accuracy 




jiJ 




6 5 



° 4 
- 1 * 



I I I 



R2 R3 R4 

Level ofolffkulty 

Encoding memoiy - thinking time 



C2 

33 

9 23 



R2 R3 R4 

l*vel of (iffkulty 

Encoding memorv - Accuracy 




I -i 1 i|L_ii 




Once Twice Unseen 

Exposition fold 

Control 



Once Twice Unseen 

Exposition fold 



PD 



Figure 2 Behavioural performance by groups on (A) Tower of London (planning items) where difficulty is manipulated by the number of 
movements required; (B) Spatial Rotations Task (rotation items), where difficulty is manipulated by the complexity of the items to rotate; 
and (C) Encoding Memory Task, where difficulty is manipulated by the number of expositions in the memory task. A1 and B1 show 
response latency against the three level of difficulty for patients and controls. A2 and B2 show results in accuracy (the number of correct 
responses) against levels of difficulty for patients and controls. C1 shows the number of correct, incorrect and unseen responses during the 
post-scan test for patients and controls. C2 shows the number of correct responses for patients and controls, against exposure fold, (once 
versus twice). *Significant interaction between condition and difficulty (A1 and A2), significant interaction between disease and difficulty 
(B1 and B2), significant exposure effect (C1) and disease effect (C2), P < 0.05. PD = Parkinson's disease. 



MAPT or APOE genotype on activation for the Tower of London 
Task (F < 1). 

Spatial Rotations Task 

Across the two sites the number of patient participants 
(Cambridge/Newcastle) completing the study was n Patient = 1 34 
(46/88) and for healthy controls n Co nt ro i = 73 (49/24). 
Behavioural performance is illustrated in Fig. 2. 



Latency of response 

Task condition [control versus planning, F(1 ,207) = 312.534; 
P< 0.001] and difficulty [F(2,414) = 45.548; P< 0.001] along 
with the interaction between them [F(2,414) = 14.665; P < 0.001] 
were all significant, confirming that more difficult planning items 
required more time to be solved. Repeated-measures ANOVA re- 
vealed a significant effect of site [F(1 ,207) = 1 7.689; P < 0.001 ] but 
no effect of disease (F < 1) with no significant interaction (F < 1). 



2750 I Brain 2014: 137; 2743-2758 



C. Nombela et al. 




Figure 3 For the Tower of London Task (top left), the activation in regions of interest is presented separately by COMT genotype and 
LEDD in patients (bottom). The y-axis of each graph represents the mean activation in terms of average parameter estimates. The data are 
subdivided by a median split of LEDD (above versus below 275 mg/day) for each region of interest (top right). *P < 0.05. 



There was also a significant interaction between difficulty and dis- 
ease [F(2,414) = 2.988; P < 0.05], reflecting longer times to perform 
more difficult items by patients than controls. 

There was no significant effect of age (F < 1), MMSE (F < 1), 
years of education [F(1 ,206) = 2.425; not significant], MOCA 
(F<1), verbal and category fluency (F<1) or NART 
[F(1 ,206) = 1 .744; not significant] on latency of response. 

For the Parkinson's disease group, those with higher UPDRS-III 
scores [F(1, 1 34) = 5.637, P< 0.019] showed longer response 
latencies. There was a trend towards an effect of duration 
[F(1,134) = 3.457, P < 0.065] but no effect of LEDD (F<1), 
MAPI, COMT or APOE genotype on latency (F < 1). 

A stepwise multiple regression in the Parkinson's disease group 
revealed a minimal model [F(1 ,134) = 4.079, P < 0.045] including 
just category fluency [£(116) = -2.020; P< 0.045], which ex- 
plained only 2.9% of the variance. 

Accuracy 

Among within-subject factors, there was a significant effect of 
condition [F(1 .207) = 179.697; P< 0.001] and difficulty 
[F(2,414) = 21.691; P < 0.001] and a significant interaction 
[F(2,414) = 66.130; P < 0.001]. There was an effect of site on 
accuracy [F(1 ,207) = 42.611, P< 0.001, higher in Site 1] and a 
trend towards a disease effect [F(1,207) = 3.319, P < 0.07, lower 
score in patients] but there was no significant interaction. 

Accuracy was higher in younger volunteers [F(1,207) = 
3.715; P<0.05], and those with higher category fluency 



[F(1,207) = 7.264;P < 0.008] with weak trends for years of edu- 
cation [F(1 ,207) = 3.554; P< 0.061] and MOCA 
[F(1 ,207) = 3.385; P < 0.067], but no effects of MMSE, NART 
or verbal fluency [F(1,207) < 1.8; not significant]. 

The Parkinson's disease group with lower UPDRS-III score 
achieved higher accuracy [F(1 ,134) = 6.839; P< 0.001] and 
there was a weak trend for shorter duration of disease 
[F(1,134) = 6.839; P < 0.079] but no effect of LEDD [F(1,134) = 
2.702; not significant] or MAPT. A significant interaction between 
MAPT and difficulty [F(2, 134) = 39.135; P< 0.001] was found, 
confirming that H1 haplotype homozygotes achieved lower accur- 
acy in the more difficult items (Fig. 4). There was no significant 
effect of COMT or APOF genotype (F < 1). 

The stepwise multiple regression in the Parkinson's disease 
group revealed an explanatory model [F(2, 134) = 12.317, 
P< 0.001] that included years of education [£(116 = 4.115; 
P<0.001] and age [£(116) = -2.501; P< 0.014], which ex- 
plained 15% of the variance. 

Functional MRI regional activity 

To determine brain regions specifically activated by the rotational 
task, 'all rotation events minus baseline conditions' were analysed. 
Repeated-measures ANOVA showed no effect of site, disease or 
interaction effects between disease and site (all F<1). The re- 
gions differed in their activity as revealed by a main effect of 
region of interest [F(7,1428) = 85.004; adjusted P< 0.001] and 
there was a significant interaction between site and region of 



Early diagnosis of cognitive decline in Parkinson's disease 



Brain 2014: 137; 2743-2758 I 2751 



Rotation condition 



Match condition 



* 7 





I 7 



5 




HI homozygotcs 



H2 carriers 



Figure 4 Behavioural responses in the Spatial Rotations Task, showing the number of correct responses during experimental (left) and 
control (right) conditions, respectively. Repeated-measures ANOVA indicated a significant interaction between MAPT (H1/H1 versus H2 
carriers) and difficulty at rotation condition during the Spatial Rotations Task. *P < 0.05. Difficulty is manipulated by the complexity of the 
items to rotate in the Spatial Rotation Task. 



interest [F(7,1428) = 3.374; adjusted P < 0.001] and between dis- 
ease and region of interest [f (7,1 428) = 1.998; P < 0.05] such 
that controls achieved greater activation than patients in a 
subset of regions of interest. Post hoc t-tests analysis showed 
that significant effects were localized to the left parietal 
[f(207) = 1.917; P < 0.05] and precuneus [f(207 = 2.241 ; 
P < 0.026]. 

In the Parkinson's disease patient group, the addition of be- 
tween-subject variables (UPDRS-III, LEDD and duration) in separ- 
ate repeated-measures ANCOVAs indicated a significant effect of 
LEDD [F(1,134) = 1.696; P < 0.041] but no significant effect of 
UPDRS-III or duration (all F<1) on region of interest activity. 
Subsequent repeated-measures ANOVA including MAPT genotype 
and site as between-subject factors confirmed an effect of MAPT 
on beta activity within the regions of interest [F(1 ,134) = 6.600; 
P < 0.011, Fig. 5]. Post hoc f-test analysis indicated that H2 car- 
riers reached significantly higher values than H1 homozygotes in 
the right caudate [«134) = 4.045; P< 0.047], left caudate 
[«134) = 6.215; P< 0.014] and left parietal [«134) = 5.343; 
P < 0.023, Fig. 5]. There was no effect of COMT or APOF on 
region of interest activation during the Spatial Rotations Task 
(F< 1). 

Memory Encoding Task 

Across the two sites the number of patient participants 
(Cambridge/Newcastle) completing the encoding memory was 

^Patient = 

128 (41/87) and for healthy controls n Co ntroi = 80 (48/ 
32). Behavioural performance is illustrated in Fig. 2. 

Latency of response 

The within-subjects factor of exposure fold (once versus twice) 
was significant [F(1,208) = 62.401; P < 0.001]: in both groups la- 
tency of response was shorter for pictures exposed twice than for 



pictures exposed once. Repeated-measures ANOVA revealed sig- 
nificant effects of site [F(1 ,208) = 46.070; P < 0.001], but no dis- 
ease effect or interaction between site and disease on latency. 

There were no effects of age [F(1,208) = 1.203; not significant], 
MMSE [F(1 ,208) = 1.293; not significant] years of education 
(F < 1), letter fluency (F < 1), category fluency (F < 1), MOCA 
[F(1 , 208) = 1.501; not significant] or NART (F<1). In patients 
with Parkinson's disease, UPDRS-III (F<1), LEDD 
[F(1, 128) = 2.402; not significant], duration (F<1), COMT, 
MAPT or APOE (F < 1) had no significant effect on latency of 
response. A stepwise multiple regression model 
[F(1,128) = 14.245, P < 0.001] indicated that duration of disease 
[f(128) = -3.774; P < 0.001] explained 12.4% of the variance. 

Accuracy 

There was a main effect of site [F(1 ,208) = 22.476; P< 0.001, 
higher at Site 1] and disease [F(1, 208) = 4.165; P < 0.043] on 
accuracy, indicating more recognized pictures by controls than 
patients, but there was no interaction between disease and site 
(F < 1). The exposure fold (once versus twice) affected accuracy 
[F(1 ,208) = 170.973; P< 0.001], in both patient and control 
groups with no interaction between disease and site. See Fig. 2 
for details. Further analysis including d' scores per participant indi- 
cated higher scores for controls for both pictures seen once 
[«208) = 2.937; P < 0.004] and for those seen twice 
[f(208) = 3.524; P < 0.001]. 

There was no significant effect of age [F(1 ,208) = 1 .203; not 
significant], MMSE [F(1 , 208) = 1.293; not significant], years of 
education (F<1), MOCA [F(1 ,208) = 1 .501 ; not significant], 
letter fluency (F < 1), category fluency (F < 1) or NART (F < 1) 
on encoding memory task. 

In the Parkinson's disease group, there was no significant effect 
of LEDD [F(1,128) = 2.402; not significant], UPDRS-III (F<1), 



2752 I Brain 2014: 137; 2743-2758 



C. Nombela et at. 




Figure 5 For the Spatial Rotations Task (top left), the activation within each region of interest (top right) is plotted separately for H1 
patient homozygotes and H2 patient carriers. The y-axes represent the mean parameter estimate, in arbitrary scaled units. See text for 
details of the gene by region interaction. Post hoc f-test analysis indicated that region of interest and MAPT genotype interaction occurred 
at marked areas (bottom). *P < 0.05. 



duration (F < 1), COMT, MAPT or APOE genotype on accuracy. 
The stepwise multiple regression analysis in patients revealed no 
single significant explanatory variables for accuracy variance. 

Functional MRI regional activity 

The contrast between correctly encoded pictures 'seen once' 
minus baseline was used for repeated-measures ANOVA of re- 
gional activation. There were significant effects of site 
[F(1 ,208) = 226.369; P< 0.001] and effect of disease 
[F(1 ,208) = 6.050; P < 0.15] with higher beta values in controls 
and in Site 1 . There was an interaction between site and disease 
[F(1,208) = 22.878; P < 0.01]. The regions differed in the magni- 
tude of activation [main effect of region of interest, 
F(7, 1 260) = 11.920; P< 0.001] with an interaction between 
region of interest and site [F(7,1 260) = 68.392; P< 0.001, 
higher at Site 1] and interactions between region of interest and 
disease [F(7, 1260) = 9.729; P < 0.001]. Post hoc f-tests revealed 
significantly lower activations in patients within the left hippocam- 
pus K(207)= -1.792; P< 0.048], left inferior frontal gyrus 
[f(208) = -4.587, P < 0.001], right inferior frontal gyrus pars tri- 
angularis [f(208) = -4.896, P < 0.001], right inferior frontal gyrus 
pars opercularis [f(207) = -3.333, P< 0.001], left parietal 
[f(180)= -4.139; P< 0.001], left occipital [f(207) = -7.056; 
P< 0.001] and temporo-parieto-occipital areas 

[f(207) = -5.008; P < 0.001], 

There was a significant effect of AAOCA on accuracy 
[F(1 ,207) = 4.959; P < 0.028] but not age (F<1), years of 



education [F(1 ,207) = 2.262; not significant], AAMSE (F<1), 
letter fluency [F(1 ,207) = 2.187; not significant] or category flu- 
ency scores (F < 1) or NART (F < 1) . 

In the Parkinson's disease group, there was an effect of LEDD 
[F(1,107) = 7.992; P < 0.006] but no effect of UPDRS-III or dur- 
ation (all F<1) on regional activity. The addition of between 
subject variables (LEDD) in a repeated-measures ANCOVA re- 
vealed an interaction between region of interest and APOE geno- 
type [F(1 4,609) = 1.422; P < 0.05], with APOE4 carriers 
manifesting lower activation. Post hoc t-test analysis showed 
that the effect was focused on right hippocampus 
[«107) = 1.866, P< 0.048], left hippocampus [f(107 = 2.635, 
P<0.01], right inferior frontal gyri pars triangularis [f(107) = 
2.739, P< 0.007], left inferior frontal gyrus [f(107 = 2.623, 
P<0.01], left parietal [f(107 = 2.498, P<0.01], left occipital 
[f(1 07 = 2.784; P < 0.007] and temporo-parieto-occipital areas 
[f(107 = 2.702, P< 0.008] (Fig. 6). There was no significant 
effect of COMT or MAPT genotype in region of interest activity 
during the Encoding Memory Task (all F < 1). 

In summary, our data showed a longer latency of response 
(Spatial Rotations Task) and lower accuracy (Spatial Rotation and 
Encoding Memory Tasks) in patients with respect to controls. 
Score differences were stressed by demographical (age and years 
of education), neuropsychological (verbal fluency, MMSE and 
MOCA) and clinical (UPDRS-III, duration and LED) covariates. 
Patient impairments were reflected in brain functional measures: 
(i) working memory performance interacted with COMT poly- 
morphisms and LEDD; (ii) spatial abilities was particularly impaired 



Early diagnosis of cognitive decline in Parkinson's disease 



Brain 2014: 137; 2743-2758 I 2753 




Figure 6 Regional activation during encoding of items in the Encoding Memory Task (top left), illustrating the significant interaction 
between regional activation and APOE genotype in Parkinson's disease patients (see text for details). The y-axes represent the mean 
parameter estimate, in arbitrary scaled units. Post hoc t-test analysis indicated that region of interest and APOE genotype interaction 
occurred at marked areas (bottom). *P < 0.05. TPO = temporo-parieto-occipital. 



in H1 homozygotes (MAPI); and (iii) encoding abilities engaged 
lower beta values as a function of APOE polymorphisms. 

Discussion 

The principal results of this study, in line with our hypotheses, 
were that (i) soon after diagnosis, neurocognitive changes are evi- 
dent in fronto-striatal and parieto-temporal systems; and (ii) 
common polymorphisms in the COAAT, MAPT and APOE genes 
are associated with differences in regional brain activity associated 
with executive, visuospatial and memory functions, respectively. 
Our results demonstrate a significant impact of these genes on 
cortical activity associated with cognitive tasks, either alone or 
through an interaction with dopaminergic medication. This study 
goes beyond previous work, not only in the power afforded by the 
cohort size, but also in its emphasis on early disease, with patients 
being scanned within a median of 5 and 19 months from diagnosis 
at the two sites, respectively — namely within 2 years of their 
diagnosis. 

Our cohort was also representative of Parkinson's disease soon 
after diagnosis: our 168 patients did not differ in their demo- 
graphic variables from the larger ICICLE-PD cohort of 219 patients 
from which they were recruited (Yarnall et al., 2014). In ICICLE- 
PD, the patients' age, UPDRS-III, cognitive abilities and years of 
education were similar to previous large studies of community 
acquired cohorts in the UK undertaken in the last decade 



(Foltynie et al., 2004a, b; Williams-Cray et al., 2007a, 2009a; 
Elgh et al., 2009; Fallon et al., 2013). 

Tasks and cognition 

Although Parkinson's disease is associated with dysfunction of the 
fronto-striatal circuits supporting executive systems (Owen et al., 
1992; Kehagia et al., 2010), recent evidence indicates multiple 
affected domains (Janvin et al., 2006; Hely et al., 2008; Elgh 
et al., 2009; Aarsland and Kurz, 2010; Pedersen et al., 2013). 
The dynamic nature of neurodegeneration, neurotransmitter loss 
and progressive neuropathology led to the Dual Syndrome hy- 
pothesis of cognitive deficits in Parkinson's disease (Goris et al., 
2007; Kehagia et al., 2013; Winder-Rhodes et al., 2013): frontos- 
triatal dopaminergic dysfunction impairs planning, working 
memory, response inhibition and attention control, while posterior 
cortical pathology and cholinergic deficits impairs visuospatial, 
mnemonic and semantic functions. 

Our choice of functional AARI tasks succeeded in making differ- 
ential demands on fronto-striatal and temporoparietal systems for 
planning, spatial rotation and memory (Grant et al., 2013; 
Hampshire et al., 2013). The Tower of London Task is an execu- 
tive task that requires planning and working memory, which re- 
cruits a frontoparietal network that includes the prefrontal 
associative cortex (DLPFC) and posterior parietal cortex (Owen, 
1998; Owen et al., 1998; Rowe et al., 2000, 2001). At all 
stages of Parkinson's disease, impairments on this task have 



2754 I Brain 2014: 137; 2743-2758 



C. Nombela ef al. 



been reported with longer response times, reduced accuracy and 
poor neural efficiency with respect to age-matched controls 
(Owen ef al., 1992; Owen, 1998; Perfetti et al., 2010) and re- 
gional impairments identified by functional MRI and PET (Baker 
et al., 1996; Owen et al., 1996; Williams-Gray et al., 2007b). 
Lesion studies have confirmed that this task requires the integrity 
of the prefrontal cortex (Bor et al., 2006) whereas pharmaco- 
logical interventions and withdrawal indicate dopamine depend- 
ence (Cools et al., 2002). 

There was evidence of dopamine dependent Tower of London 
deficits in some patients, with a non-linear relationship between 
cortical dopamine tone and regional activation indicated by the 
significant LEDD by COMT interaction. Specifically, prefrontal 
cortex and caudate nuclei were more activated in Met/Met homo- 
zygotes on low-dose dopaminergic medication and Val/Val homo- 
zygotes on high-dose medication. This interaction is predicted by 
the inverted 'U-shaped function' relating dopaminergic tone and 
function, by which either too high or too low dopaminergic tone 
impairs working memory and executive performance (Goldberg 
and Weinberger, 2004; Williams-Gray et al., 2007b; Rowe 
et al., 2008; Cools and D'Esposito, 2011; Fallon et al., 2013). 

Our second task required mental spatial rotation, emphasizing 
visuospatial functions. Impairments in this domain are predictive of 
dementia in Parkinson's disease (Williams-Gray et al., 2009b). 
Neuroimaging of similar spatial rotations tasks in healthy adults 
indicates posterior parietal activation (Corballis, 1997) and pre- 
frontal activation (Selemon and Goldman-Rakic, 1988; Goldberg 
and Weinberger, 2004). Parkinson's disease increases response 
latencies and errors on this task (Lee et al., 1998; Amick et al., 
2006), and reduces posterior parietal activation (Crucian et al., 

2003) . We replicated both effects, more so in MAPT H1 
homozygotes. 

The final task involved required visual episodic memory encod- 
ing. This task evokes hippocampal and medial temporal lobe ac- 
tivity during encoding in healthy controls (Dove et al., 2006), 
which we replicated. We found that even in the early stages of 
Parkinson's disease, a reduction was seen in the neocortical acti- 
vation associated with this task, although the magnitude and dir- 
ection of hippocampal effects was similar (Fig. 6). Parkinson's 
disease- mild cognitive impairment and later stages of 
Parkinson's disease impair episodic memory (Weintraub et al., 

2004) although the relationship of early poor memory perform- 
ance to the development of Parkinson's disease dementia is un- 
clear (Williams-Gray et al., 2009b). Memory impairment is 
associated with reduced hippocampal volume in Parkinson's dis- 
ease (Davidson et al., 2013; Pereira et al., 2013) as well as in early 
Alzheimer's disease (Sahakian ef al., 1988), supported by objective 
measures of impaired memory encoding (Weintraub ef al., 2011; 
Beyer et al., 2013). 

Genetic influences on cognitive systems 
in Parkinson's disease 

We examined common polymorphisms that modulate the behav- 
ioural and neural consequences of Parkinson's disease. COMT 
regulates prefrontal cortical dopamine metabolism (Chen et al., 



2004) and influences macroscopic cortical structure (Rowe et al., 
2010). Both functional MRI (Rowe et al., 2008; Williams-Gray 
ef al., 2008; Fallon ef al., 2013) and F-DOPA PET (Wu ef al., 
2012) studies have shown significant functional consequences of 
the Val157Met polymorphism in Parkinson's disease. 

The COMT effect is complex, with modulation by both levo- 
dopa therapy and task demands (Williams-Gray et al., 2007b, 
2009a). Both the COMT genotype and dose of extrinsic dopamin- 
ergic medication follow a non-linear U-shape function for a given 
task, with either too-high or too-low frontal cortical dopamine 
levels adversely affecting cognitive performance and activation 
(Rowe et al., 2008). Consistent with the proposed dopaminergic 
modulation of frontostriatal circuits, the interaction between 
COMT genotype and LEDD was significant in dorsolateral and 
frontopolar prefrontal cortices and caudate nuclei. 

However, some studies do not find evidence for COMT modu- 
lation of frontal dopamine function. For example, no interaction 
between COMT genotype and Tower of London performance was 
reported by Hoogland et al. (2010) or between COMT and pre- 
frontal activation by Stokes ef al. (2011). In Hoogland ef al. 
(2010) a different Tower of London version was used (Foltynie 
et al., 2004b), and no functional MRI was conducted, perhaps 
limiting the sensitivity to an effect of COMT. Interestingly, there 
was an interaction between LEDD and COMT on verbal reasoning 
consistent with a genotype interaction with dopaminergic medica- 
tion to influence frontal cognitive ability in Parkinson's disease. 
Stokes ef al. (2011) applied a similar MRI Tower of London ver- 
sion to ours, but in fewer subjects and healthy middle-aged con- 
trols. Here, the ICICLE-PD data from a larger sample corroborate 
the COMT genotype modulation of frontostriatal function early in 
the course of Parkinson's disease. 

A second gene of interest was MAPT. The H1 haplotype in- 
creases the risk of developing Parkinson's disease, and the risk 
of early Parkinson's disease dementia (Goris ef al., 2007; 
Williams-Gray ef al., 2009a). Here we show that H1 carrier pa- 
tients were less accurate with difficult spatial rotations, and sus- 
tained less activity in the parietal cortex and caudate nuclei 
(Williams-Gray ef al., 2009a), essential areas for spatial rotations 
(Harris ef al., 2000). Others have argued that there is no relation- 
ship between MAPT haplotype and visuospatial performance 
(Goldberg and Weinberger, 2004; Ezquerra ef al., 2008; Rowe 
ef al., 2008; Morley ef al., 2012), which was the case here for 
easy items. Our hypothesis is that as Parkinson's disease pro- 
gresses, the difference between H1 and H2 haplotype will 
emerge but initially only for more difficult visuospatial tasks. Our 
data suggest that the posterior cortical functions underlying spatial 
rotations task performance are not significantly regulated by dopa- 
mine, in support of the dual syndrome hypothesis. 

The third gene of interest was APOE. During memory encoding, 
we found reduced brain activity within the temporo-parietal net- 
work and impaired performance in carriers of APOE4. Although 
the number of APOE4 carriers was small, this observation is con- 
sistent with the literature (Pulkes et al., 2011; Domenger ef al., 
2012; Federoff et al., 2012; Peplonska ef al., 2013; Multhammer 
ef al., 2014). It has been suggested that APOE4 Parkinson's dis- 
ease carriers present more severe cortical atrophy (Wakabayashi 
et al., 1998; Li et al., 2004) and more frequent cognitive decline 



Early diagnosis of cognitive decline in Parkinson's disease 



Brain 2014: 137; 2743-2758 I 2755 



than patients without an APOE4 allele (Irwin et al., 2012). Our 
data are the first to suggest that APOE4 also influences brain 
activity in the caudate nuclei, hippocampus and posterior cortical 
areas during a memory encoding task in recently diagnosed pa- 
tients with Parkinson's disease, a result that is in agreement with 
studies of Alzheimer's disease (Bookheimer and Burggren, 2009). 

The specificity of gene x task interactions suggests a contrast 
between COA/lT/dopamine effects on frontostriatal networks for 
working memory and executive function, versus MAPT/APOE 
modulation of temporo-parietal systems engaged in visuospatial 
and mnemonic functions. Other genetic factors are likely to con- 
tribute to cognitive function (Caccappolo et al., 2011; Chung 
et al., 2012), but our data clearly support a role for COMT, 
AAAPT and APOE in early disease expression, and possibly disease 
onset (Goris et al., 2007). The influence of these genetic variants 
is not necessarily specific to Parkinson's disease, and we saw in the 
introduction how they have been associated with risk, imaging 
and cognitive performance differences in several neurological 
and psychiatric disorders. However, the variation of these three 
genes appears to alter the neural substrates for major cognitive 
domains even soon after diagnosis of Parkinson's disease, which 
we suggest is directly relevant to their modification of the risk of 
cognitive impairment or dementia in the context of Parkinson's 
disease (APOE4, MAPT) and the potentially deleterious effects 
of high dose levodopa therapy on some aspects of cognition in 
a subset of patients (COMT). The mechanisms of these genetic 
influences may include pharmacological interactions at the synapse 
(especially for COMT in relation to cortical dopamine transmis- 
sion). However, they may also include neuroplasticity conse- 
quences of COMT, APOE and MAPT functional polymorphisms 
in the context of Parkinson's disease pathogenesis, or develop- 
mental effects even if these diminish with older age (e.g. for 
COMT) (de Frias et al., 2005; Starr ef al., 2007; Rowe et al., 
2010). 

Limitations 

The large size of ICICLE-PD and the systematic recruitment meth- 
ods have obvious advantages, but there remain methodological 
and inferential limitations with this study. Even with 168 partici- 
pants, the non-significant results of genetic variance or LEDD may 
result in type II error. Our statistical methods prioritize type I 
errors, especially with respect to the functional MRI studies. 
Moreover, we suggest that more subtle effects of genotype, medi- 
cation or other clinical-demographic factors may emerge with dis- 
ease progression. We also rely on clinical diagnostic criteria, 
Although we are relatively protected against potential misdiagnosis 
as ICICLE-PD relies on reapplying the clinicopathologically vali- 
dated diagnostic criteria after 18 months, and this is expected to 
be >90% accurate. 

Several performance and imaging results differed between sites, 
despite the same research protocol (Yarnall et al., 2014). Site dif- 
ferences are unlikely to reflect fundamental differences in the 
onset, risks or pathology of Parkinson's disease. The site differ- 
ences were not restricted to socioeconomic and cognitive meas- 
ures, but also included the interval from diagnosis to scanning, and 
the levodopa dose equivalent at the time of scanning. 



Interestingly, the difference in UDPRS-III motor signs severity 
was not significant suggesting that local treatment decisions 
were effectively managing what may have been differential pro- 
gression of the underlying disease between sites over time. 
Although there may be some genetic variation between northern 
and eastern England, we suggest that it is more likely that the 
differences between sites arise from different referral pathways 
and treatment practises. We fortunately obtained control partici- 
pant data from both sites, to reduce the potential impact of re- 
gional differences in culture, genetics, education, prior health and 
access to care services. Socioeconomic and educational norms may 
influence some cognitive score differences between sites, but the 
sites remain comparable on the most important demographic and 
cognitive tests metrics (age, gender, MMSE, AAOCA, NART). Most 
importantly for the interpretation of the regional activations, the 
behavioural data in the functional MRI tasks did not differ be- 
tween sites. It remains to be seen whether geographical factors 
continue to affect the cognitive and neural markers as disease 
progresses, or whether the sites converge over time as their dif- 
ferential delay to participation gradually becomes a smaller frac- 
tion of the total disease duration. 

We did not find many significant or large group effects in terms 
of behavioural measures. This may at first seem disappointing, 
given the behavioural deficits that emerge in studies of patients 
with more advanced disease. However, the lack of major effects in 
terms of behavioural data provides more relevance to the signifi- 
cant differences between patients and controls in the functional 
imaging: functional MRI may be more sensitive to the factors 
that modify the function of neural systems than the cognitive 
performance that depend on those systems at least at early 
stages of the disease; and the specificity of region by group inter- 
actions also raises the possibility that at early stages of the dis- 
ease, compensatory mechanisms can allow for a normal 
performance. It also reduces the ambiguity in interpreting func- 
tional MRI data that otherwise arises if there are marked behav- 
ioural differences such that activation differences could be the 
cause or consequence of altered behaviour (Price and Friston, 
1999; Poldrack, 2007). 

This study is focused on the early presentation of Parkinson's 
disease, with a median time from diagnosis to inclusion of 8 
months. The genetic and clinical factors that we identify might 
be used to study earlier or pre-manifest states in future studies 
which would also avoid issues of treatment effects. However, this 
was beyond the scope of the ICICLE-PD study. The potential 
interaction between genetic variants and the rate of cognitive 
decline following presentation of Parkinson's disease in the 
ICICLEPD cohort (without dementia at presentation) will require 
longitudinal investigation which will be the subject of future 
research papers. 

Conclusion 

This functional imaging study in ICICLE-PD revealed that soon 
after diagnosis, there are already changes in brain function and 
cognitive performance in patients with Parkinson's disease. The 
regional activations associated with three major cognitive domains 



2756 I Brain 2014: 137; 2743-2758 



C. Nombela et al. 



interact with genotype in the context of Parkinson's disease. Even 
recently diagnosed patients had impaired performance and altered 
regional brain activity in three tasks that spanned frontostriatal 
and parieto-temporal systems. The anatomical, functional, genetic 
and behavioural data support the dual syndrome hypothesis for 
Parkinson's disease cognition, with (i) an executive syndrome that 
is frontally mediated, dopamine-dependant and modulated by 
COMT genotype; versus (ii) a temporo-parietal system subject to 
MAPT and APOE, but not dopaminergic modulation, that is 
required for visuospatial and memory tasks. 

Acknowledgements 

We would like to thank all volunteers for their participation. 



Funding 

This study was supported by Parkinson's UK (C.N.), Lockhart 
Parkinson's Disease Research Fund (T.K.K.), Michael J. Fox 
Foundation (A.J.Y.), the National Institute for Health Research 
(NIHR, RG64473) Cambridge Biomedical Research Centre, the 
Wellcome Trust (JBR 088324); the Medical Research Couciil 
Cognition and Brain Sciences Unit, Cambridge (MC-A060- 
5PQ30); the NIHR Newcastle, Biomedical Research Unit based 
at Newcastle-upon-Tyne Hospitals, NHS Foundation Trust and 
Newcastle University; the NIHR Dementias and 
Neurodegenerative Diseases Research Network (J.T.O.) and 
Raymond and Beverly Sadder studentship (D.P.B.). The views ex- 
pressed are those of the authors and not necessarily those of the 
NHS, the NIHR or the Department of Health. 

References 

Aarsland D, Andersen K, Larsen JP, Lolk A, Kragh-Sorensen P. Prevalence 
and characteristics of dementia in Parkinson disease: an 8-year pro- 
spective study. Arch Neurol 2003; 60: 387-92. 

Aarsland D, Bronnick K, Fladby T. Mild cognitive impairment in 
Parkinson's disease. Curr Neurol Neurosci Rep 2011; 11: 371-8. 

Aarsland D, Kurz MW. The epidemiology of dementia associated with 
Parkinson's disease. Brain Pathol 2010; 20: 633-9. 

Amick MM, Schendan HE, Ganis G, Cronin-Golomb A. Frontostriatal 
circuits are necessary for visuomotor transformation: mental rotation 
in Parkinson's disease. Neuropsychologia 2006; 44: 339-49. 

Baker SC, Rogers RD, Owen AM, Frith CD, Dolan RJ, Frackowiak RS, 
et al. Neural systems engaged by planning: a PET study of the Tower 
of London task. Neuropsychologia 1996; 34: 515-26. 

Barone P, Aarsland D, Burn D, Emre M, Kulisevsky J, Weintraub D. 
Cognitive impairment in nondemented Parkinson's disease. Mov 
Disord 2011; 26: 2483-95. 

Benton AL. Differential behavioral effects in frontal lobe disease. 
Neuropsychologia 1968; 6: 53-60. 

Beyer MK, Bronnick KS, Hwang KS, Bergsland N, Tysnes OB, Larsen JP, 
et al. Verbal memory is associated with structural hippocampal 
changes in newly diagnosed Parkinson's disease. J Neurol Neurosurg 
Psychiatry 2013; 84: 23-8. 

Bookheimer S, Burggren A. APOE-4 genotype and neurophysiological 
vulnerability to Alzheimer's and cognitive aging. Annu Rev Clin 
Psychol 2009; 5: 343-62. 



Bor D, Duncan J, Lee AC, Parr A, Owen AM. Frontal lobe involvement in 
spatial span: converging studies of normal and impaired function. 
Neuropsychologia 2006; 44: 229-37. 

Bronnick K, Alves G, Aarsland D, Tysnes OB, Larsen JP. Verbal memory 
in drug-naive, newly diagnosed Parkinson's disease. The retrieval def- 
icit hypothesis revisited. Neuropsychology 2011; 25: 114-24. 

Caccappolo E, Alcalay RN, Mejia-Santana H, Tang MX, Rakitin B, 
Rosado L, et al. Neuropsychological profile of Parkin mutation carriers 
with and without Parkinson Disease: the CORE-PD Study. J Int 
Neuropsychol Soc 2011; 17: 91-100. 

Chen J, Lipska BK, Halim N, Ma QD, Matsumoto M, Melhem S, et al. 
Functional analysis of genetic variation in catechol- 
O-methyltransferase (COMT): effects on mRNA, protein, and 
enzyme activity in postmortem human brain. Am J Hum Genet 
2004; 75: 807-21. 

Chung SJ, Armasu SM, Biernacka JM, Anderson KJ, Lesnick TG, 
Rider DN, et al. Genomic determinants of motor and cognitive out- 
comes in Parkinson's disease. Parkinsonism Relat Disord 2012; 18: 
881-6. 

Cohen MS, Kosslyn SM, Breiter HC, DiGirolamo GJ, Thompson WL, 
Anderson AK, et al. Changes in cortical activity during mental rotation. 
A mapping study using functional MRI. Brain 1996; 119 (Pt 1): 
89-100. 

Cools R, D'Esposito M. Inverted-U-shaped dopamine actions on human 
working memory and cognitive control. Biol Psychiatry 2011; 69: 
e1 13-25. 

Cools R, Stefanova E, Barker RA, Robbins TW, Owen AM. 
Dopaminergic modulation of high-level cognition in Parkinson's dis- 
ease: the role of the prefrontal cortex revealed by PET. Brain 2002; 
125: 584-94. 

Corballis MC. Mental rotation and the right hemisphere. Brain Lang 

1997; 57: 100-21. 
Corder EH, Saunders AM, Strittmatter WJ, Schmechel DE, Gaskell PC, 

Small GW, et al. Gene dose of apolipoprotein E type 4 allele and the 

risk of Alzheimer's disease in late onset families. Science 1993; 261: 

921-3. 

Cronin-Golomb A, Braun AE. Visuospatial dysfunction and problem sol- 
ving in Parkinson's disease. Neuropsychology 1997; 11: 44-52. 

Crucian GP, Barrett AM, Burks DW, Riestra AR, Roth HL, Schwartz RL, 
et al. Mental object rotation in Parkinson's disease. J Int Neuropsychol 
Soc 2003; 9: 1078-87. 

Davidson PS, Cook SP, McGhan L, Bouchard T, Camicioli R. Source 
memory in normal aging and Parkinson's disease. J Neuropsychol 
2013; 7: 179-92. 

de Frias CM, Annerbrink K, Westberg L, Eriksson E, Adolfsson R, 
Nilsson LG. Catechol O-methyltransferase Val158Met polymorphism 
is associated with cognitive performance in nondemented adults. 
J Cogn Neurosci 2005; 17: 1018-25. 

Domenger D, Dea D, Theroux L, Moquin L, Gratton A, Poirier J. The 
MPTP neurotoxic lesion model of Parkinson's disease activates the 
apolipoprotein E cascade in the mouse brain. Exp Neurol 2012; 233: 
513-22. 

Dove A, Brett M, Cusack R, Owen AM. Dissociable contributions of the 
mid-ventrolateral frontal cortex and the medial temporal lobe system 
to human memory. Neuroimage 2006; 31: 1790-801. 

Ekman U, Eriksson J, Forsgren L, Mo SJ, Riklund K, Nyberg L. Functional 
brain activity and presynaptic dopamine uptake in patients with 
Parkinson's disease and mild cognitive impairment: a cross-sectional 
study. Lancet Neurol 2012; 11: 679-87. 

Elgh E, Domellof M, Linder J, Edstrom M, Stenlund H, Forsgren L. 
Cognitive function in early Parkinson's disease: a population-based 
study. Eur J Neurol 2009; 16: 1278-84. 

Ezquerra M, Campdelacreu J, Gaig C, Compta Y, Munoz E, Marti MJ, 
et al. Lack of association of APOE and tau polymorphisms with de- 
mentia in Parkinson's disease. Neurosci Lett 2008; 448: 20-3. 

Fallon SJ, Williams-Gray CH, Barker RA, Owen AM, Hampshire A. 
Prefrontal dopamine levels determine the balance between cognitive 
stability and flexibility. Cereb Cortex 2013; 23: 361-9. 



Early diagnosis of cognitive decline in Parkinson's disease 



Brain 2014: 137; 2743-2758 I 2757 



Federoff M, Jimenez-Rolando B, Nails MA, Singleton AB. A large study 
reveals no association between APOE and Parkinson's disease. 
Neurobiol Dis 2012; 46: 389-92. 

Folstein MF, Folstein SE, McHugh PR. "Mini-mental state". A practical 
method for grading the cognitive state of patients for the clinician. 
J Psychiatr Res 1975; 12: 189-98. 

Foltynie T, Brayne CE, Robbins TW, Barker RA. The cognitive ability of 
an incident cohort of Parkinson's patients in the UK. The CamPaIGN 
study. Brain 2004a; 127: 550-60. 

Foltynie T, Goldberg TE, Lewis SG, Blackwell AD, Kolachana BS, 
Weinberger DR, et al. Planning ability in Parkinson's disease is influ- 
enced by the COMT val158met polymorphism. Mov Disord 2004b; 
19: 885-91. 

Goetz CG, Nutt JG, Stebbins GT. The Unified Dyskinesia Rating Scale: 
presentation and clinimetric profile. Mov Disord 2008; 23: 2398-403. 

Goldberg TE, Weinberger DR. Genes and the parsing of cognitive pro- 
cesses. Trends Cogn Sci 2004; 8: 325-35. 

Gomperts SN, Locascio JJ, Marquie M, Santarlasci AL, Rentz DM, 
Maye J, et al. Brain amyloid and cognition in Lewy body diseases. 
Mov Disord 2012; 27: 965-73. 

Gomperts SN, Locascio JJ, Rentz D, Santarlasci A, Marquie M, 
Johnson KA, et al. Amyloid is linked to cognitive decline in patients 
with Parkinson disease without dementia. Neurology 2013; 80: 
85-91 . 

Goodglass H, Gleason JB, Bernholtz NA, Hyde MR. Some linguistic struc- 
tures in the speech of a Broca's aphasic. Cortex 1972; 8: 191-212. 

Goris A, Williams-Gray CH, Clark GR, Foltynie T, Lewis SJ, Brown J, et al. 
Tau and alpha-synuclein in susceptibility to, and dementia in, 
Parkinson's disease. Ann Neurol 2007; 62: 145-53. 

Gowers WR. A manual of diseases of the nervous system. Philadelphia: 
Blakiston; 1893. 

Hampshire A, Highfield RR, Parkin BL, Owen AM. Fractionating human 

intelligence. Neuron 2012; 76: 1225-37. 
Hampshire A, MacDonald A, Owen AM. Hypoconnectivity and 

hyperfrontality in retired American football players. Sci Rep 2013; 3: 

2972. 

Harris IM, Egan GF, Sonkkila C, Tochon-Danguy HJ, Paxinos G, 
Watson JD. Selective right parietal lobe activation during mental rota- 
tion: a parametric PET study. Brain 2000; 123 (Pt 1): 65-73. 

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. 

Hoogland J, de Bie RM, Williams-Gray CH, Muslimovic D, Schmand B, 
Post B. Catechol-O-methyltransferase val158met and cognitive func- 
tion in Parkinson's disease. Mov Disord 2010; 25: 2550-4. 

Huang X, Chen P, Kaufer Dl, Troster Al, Poole C. Apolipoprotein E and 
dementia in Parkinson disease: a meta-analysis. Arch Neurol 2006; 63: 
189-93. 

Hughes AJ, Daniel SE, Ben-Shlomo Y, Lees AJ. The accuracy of diagnosis 

of parkinsonian syndromes in a specialist movement disorder service. 

Brain 2002; 125: 861-70. 
Irwin DJ, White MT, Toledo JB, Xie SX, Robinson JL, Van Deerlin V, et al. 

Neuropathologic substrates of Parkinson disease dementia. Ann Neurol 

2012; 72: 587-98. 
Jacobs HI, Van Boxtel MP, Heinecke A, Gronenschild EH, Backes WH, 

Ramakers IH, et al. Functional integration of parietal lobe activity in 

early Alzheimer disease. Neurology 2012; 78: 352-60. 
Janvin CC, Larsen JP, Aarsland D, Hugdahl K. Subtypes of mild cognitive 

impairment in Parkinson's disease: progression to dementia. Mov 

Disord 2006; 21: 1343-9. 
Kehagia AA, Barker RA, Robbins TW. Neuropsychological and clinical 

heterogeneity of cognitive impairment and dementia in patients with 

Parkinson's disease. Lancet Neurol 2010; 9: 1200-13. 
Kehagia AA, Barker RA, Robbins TW. Cognitive impairment in 

Parkinson's disease: the dual syndrome hypothesis. Neurodegener 

Dis 2013; 11: 79-92. 
Lee AC, Harris JP, Calvert JE. Impairments of mental rotation in 

Parkinson's disease. Neuropsychologia 1998; 36: 109-14. 



Lewis SJ, Dove A, Robbins TW, Barker RA, Owen AM. Cognitive impair- 
ments in early Parkinson's disease are accompanied by reductions in 
activity in frontostriatal neural circuitry. J Neurosci 2003; 23: 6351-6. 

Li YJ, Hauser MA, Scott WK, Martin ER, Booze MW, Qin XJ, et al. 
Apolipoprotein E controls the risk and age at onset of Parkinson dis- 
ease. Neurology 2004; 62: 2005-9. 

Morley JF, Xie SX, Hurtig HI, Stern MB, Colcher A, Horn S, et al. Genetic 
influences on cognitive decline in Parkinson's disease. Mov Disord 
2012; 27: 512-8. 

Multhammer M, Michels A, Zintl M, Mendoza MC, Klunemann HH. A 
large ApoE epsilon4/epsilon4 homozygous cohort reveals no 
association with Parkinson's disease. Acta Neurol Belg 2014; 114: 
25-31. 

Muslimovic D, Post B, Speelman JD, Schmand B. Cognitive profile of 
patients with newly diagnosed Parkinson disease. Neurology 2005; 
65: 1239-45. 

Nagano-Saito A, Habak C, Mejia-Constain B, Degroot C, Monetta L, 
Jubault T, et al. Effect of mild cognitive impairment on the patterns 
of neural activity in early Parkinson's disease. Neurobiol Aging 2014; 
35: 223-31. 

Nasreddine ZS, Phillips NA, Bedirian V, Charbonneau S, Whitehead V, 
Collin I, et al. The Montreal cognitive assessment, MoCA: a brief 
screening tool for mild cognitive impairment. J Am Geriatr Soc 2005; 
53: 695-9. 

Nelson HE, O'Connell A. Dementia: the estimation of premorbid intelli- 
gence levels using the New Adult Reading Test. Cortex 1978; 14: 
234-44. 

Owen A. Working memory in dorsolateral frontal cortex. Trends Cogn 
Sci 1998; 2: 239. 

Owen AM, Doyon J, Petrides M, Evans AC. Planning and spatial working 
memory: a positron emission tomography study in humans. Eur J 
Neurosci 1996; 8: 353-64. 

Owen AM, James M, Leigh PN, Summers BA, Marsden CD, Quinn NP, 
et al. Fronto-striatal cognitive deficits at different stages of Parkinson's 
disease. Brain 1992; 115 (Pt 6): 1727-51. 

Owen AM, Sahakian BJ, Semple J, Polkey CE, Robbins TW. Visuo-spatial 
short-term recognition memory and learning after temporal lobe exci- 
sions, frontal lobe excisions or amygdalo-hippocampectomy in man. 
Neuropsychologia 1995; 33: 1-24. 

Owen AM, Stern CE, Look RB, Tracey I, Rosen BR, Petrides M. 
Functional organization of spatial and nonspatial working memory pro- 
cessing within the human lateral frontal cortex. Proc Natl Acad Sci 
USA 1998; 95: 7721-6. 

Pedersen KF, Larsen JP, Tysnes OB, Alves G. Prognosis of mild cognitive 
impairment in early Parkinson disease: the Norwegian ParkWest study. 
JAMA Neurol 2013; 70: 580-6. 

Peplonska B, Safranow K, Gaweda-Walerych K, Maruszak A, 
Czyzewski K, Rudzinska M, et al. TOMM40 and APOE common gen- 
etic variants are not Parkinson's disease risk factors. Neurobiol Aging 
2013; 34: 2078-2. 

Pereira JB, Junque C, Bartres-Faz D, Ramirez-Ruiz B, Marti MJ, Tolosa E. 
Regional vulnerability of hippocampal subfields and memory deficits in 
Parkinson's disease. Hippocampus 2013; 23: 720-8. 

Perfetti B, Varanese S, Mercuri P, Mancino E, Saggino A, Onofrj M. 
Behavioural assessment of dysexecutive syndrome in Parkinson's dis- 
ease without dementia: a comparison with other clinical executive 
tasks. Parkinsonism Relat Disord 2010; 16: 46-50. 

Poldrack RA. Region of interest analysis for fMRI. Soc Cogn Affect 
Neurosci 2007; 2: 67-70. 

Price CJ, Friston KJ. Scanning patients with tasks they can perform. Hum 
Brain Mapp 1999; 8: 102-8. 

Pulkes T, Papsing C, Mahasirimongkol S, Busabaratana M, 
Kulkantrakorn K, Tiamkao S. Association between apolipoprotein 
E genotypes and Parkinson's disease. J Clin Neurosci 2011; 18: 
1333-5. 

Rowe JB, Owen AM, Johnsrude IS, Passingham RE. Imaging the 
mental components of a planning task. Neuropsychologia 2001; 
39: 315-27. 



2758 I Brain 2014: 137; 2743-2758 



C. Nombela et al. 



Rowe JB, Hughes L, Ghosh BC, Eckstein D, Williams-Gray CH, Fallon S, 
et al. Parkinson's disease and dopaminergic therapy — differential ef- 
fects on movement, reward and cognition. Brain 2008; 131: 
2094-105. 

Rowe JB, Hughes L, Williams-Gray CH, Bishop S, Fallon S, Barker RA, 
et al. The val158met COMT polymorphism's effect on atrophy in 
healthy aging and Parkinson's disease. Neurobiol Aging 2010; 31: 
1064-8. 

Rowe JB, Toni I, Josephs O, Frackowiak RS, Passingham RE. The pre- 
frontal cortex: response selection or maintenance within working 
memory? Science 2000; 288: 1656-60. 

Sahakian BJ, Morris RG, Evenden JL, Heald A, Levy R, Philpot M, et al. 
A comparative study of visuospatial memory and learning in 
Alzheimer-type dementia and Parkinson's disease. Brain 1988; 111 
(Pt 3): 695-718. 

Selemon LD, Goldman-Rakic PS. Common cortical and subcortical 
targets of the dorsolateral prefrontal and posterior parietal cortices 
in the rhesus monkey: evidence for a distributed neural net- 
work subserving spatially guided behavior. J Neurosci 1988; 8: 
4049-68. 

Shallice T. Specific impairments of planning. Philos Trans R Soc Lond B 

Biol Sci 1982; 298: 199-209. 
Starr JAA, Fox H, Harris SE, Deary IJ, Whalley LJ. COMT genotype and 

cognitive ability: a longitudinal aging study. Neurosci Lett 2007; 21: 

57-61 . 

Stokes PR, Rhodes RA, Grasby PM, Mehta MA. The effects of the 
COMT Val108/158Met polymorphism on BOLD activation during 
working memory, planning, and response inhibition: a role for the 
posterior cingulate cortex? Neuropsychopharmacology 2011; 36: 
763-71 . 

Taylor JM, Main BS, Crack PJ. Neuroinflammation and oxidative stress: 

co-conspirators in the pathology of Parkinson's disease. Neurochem Int 

2013; 62: 803-19. 
Tomlinson CL, Stowe R, Patel S, Rick C, Gray R, Clarke CE. Systematic 

review of levodopa dose equivalency reporting in Parkinson's disease. 

Mov Disord 2010; 25: 2649-53. 
Wakabayashi K, Kakita A, Hayashi S, Okuizumi K, Onodera O, 

Tanaka H, et al. Apolipoprotein E epsilon4 allele and progression of 

cortical Lewy body pathology in Parkinson's disease. Acta Neuropathol 

1998; 95: 450-4. 



Weintraub D, Moberg PJ, Culbertson WC, Duda JE, Stern MB. Evidence 
for impaired encoding and retrieval memory profiles in Parkinson dis- 
ease. Cogn Behav Neurol 2004; 17: 195-200. 

Weintraub D, Doshi J, Koka D, Davatzikos C, Siderowf AD, Duda JE, 
et al. Neurodegeneration across stages of cognitive decline in 
Parkinson disease. Arch Neurol 2011; 68: 1562-8. 

Williams-Gray CH, Evans JR, Goris A, Foltynie T, Ban M, Robbins TW, 
et al. The distinct cognitive syndromes of Parkinson's disease: 5 
year follow-up of the CamPaIGN cohort. Brain 2009a; 132: 2958-69. 

Williams-Gray CH, Foltynie T, Brayne CE, Robbins TW, Barker RA. 
Evolution of cognitive dysfunction in an incident Parkinson's disease 
cohort. Brain 2007a; 130: 1787-98. 

Williams-Gray CH, Hampshire A, Robbins TW, Owen AM, Barker RA. 
Catechol O-methyltransferase Val158Met genotype influences fronto- 
parietal activity during planning in patients with Parkinson's disease. 
J Neurosci 2007b; 27: 4832-8. 

Williams-Gray CH, Hampshire A, Barker RA, Owen AM. Attentional con- 
trol in Parkinson's disease is dependent on COMT val 158 met geno- 
type. Brain 2008; 131: 397-408. 

Williams-Gray CH, Goris A, Saiki M, Foltynie T, Compston DA, 
Sawcer SJ, et al. Apolipoprotein E genotype as a risk factor for sus- 
ceptibility to and dementia in Parkinson's disease. J Neurol 2009b; 
256: 493-8. 

Williams-Gray CH, Mason SL, Evans JR, Foltynie T, Brayne C, 
Robbins TW, et al. The CamPaIGN study of Parkinson's disease: 10- 
year outlook in an incident population-based cohort. J Neurol 
Neurosurg Psychiatry 2013; 84: 1258-64. 

Winder-Rhodes SE, Evans JR, Ban M, Mason SL, Williams-Gray CH, 
Foltynie T, et al. Glucocerebrosidase mutations influence the natural 
history of Parkinson's disease in a community-based incident cohort. 
Brain 2013; 136: 392-9. 

Wu K, O'Keeffe D, Politis M, O'Keeffe GC, Robbins TW, Bose SK, et al. 
The catechol-O-methyltransferase Val(158)Met polymorphism modu- 
lates fronto-cortical dopamine turnover in early Parkinson's disease: a 
PET study. Brain 2012; 135: 2449-57. 

Yarnall Al, Breen DP, Duncan GW, Khoo TK, Coleman SY, Firbank MJ, 
et al. Characterizing mild cognitive impairment in incident Parkinson 
disease: the ICICLE-PD study. Neurology 2014; 82: 308-16. 

Zacks JM. Neuroimaging studies of mental rotation: a meta-analysis and 
review. J Cogn Neurosci 2008; 20: 1-19.