R2* mapping for brain iron: associations with cognition in normal aging

Accepted Manuscript R2* mapping for brain iron: Associations with cognition in normal aging Christine Ghadery, MD Lukas Pirpamer, MSc Edith Hofer, PhD Christian Langkammer, PhD Katja Petrovic, PhD Marisa Loitfelder, PhD Petra Schwingenschuh, MD Stephan Seiler, MD Marco Duering, MD Eric Jouvent, MD PhD Helena Schmidt, MD, PhD Franz Fazekas, MD Jean-Francois Mangin, MD Hugues Chabriat, MD Martin Dichgans, MD Stefan Ropele, PhD Reinhold Schmidt, MD

PII:

S0197-4580(14)00618-6

DOI:

10.1016/j.neurobiolaging.2014.09.013

Reference:

NBA 9054

To appear in:

Neurobiology of Aging

Received Date: 27 February 2014 Revised Date:

11 September 2014

Accepted Date: 15 September 2014

Please cite this article as: Ghadery, C., Pirpamer, L., Hofer, E., Langkammer, C., Petrovic, K., Loitfelder, M., Schwingenschuh, P., Seiler, S., Duering, M., Jouvent, E., Schmidt, H., Fazekas, F., Mangin, J.F., Chabriat, H., Dichgans, M., Ropele, S., Schmidt, R., R2* mapping for brain iron: Associations with cognition in normal aging, Neurobiology of Aging (2014), doi: 10.1016/j.neurobiolaging.2014.09.013. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

1

ACCEPTED MANUSCRIPT

R2* mapping for brain iron: Associations with cognition in

2

normal aging

3

Christine Ghadery, MDa, Lukas Pirpamer, MSca, Edith Hofer, PhDa,b, Christian

4

Langkammer, PhDa, Katja Petrovic, PhDa, Marisa Loitfelder, PhDa, Petra

5

Schwingenschuh, MDa, Stephan Seiler, MDa, Marco Duering, MDc, Eric Jouvent, MD

6

PhDd, Helena Schmidt, MD, PhDe, Franz Fazekas, MD , Jean-Francois Mangin, MDf,

7

Hugues Chabriat, MDd, Martin Dichgans, MDc,g, Stefan Ropele, PhDa, Reinhold

8

Schmidt, MDa,*

a

10 11

b

M AN U

9

Department of Neurology, Medical University of Graz, Austria

Institute for Medical Informatics, Statistics and Documentation, Medical University of Graz, Austria

12 c

Institute for Stroke and Dementia Research and Department of Neurology, LudwigMaximilians-University, Munich, Germany

14 d

15 e

Institute of Molecular Biology and Medical Biochemistry, Medical University Graz,

19 20 21

AC C

17 18

INSERM UMR-740, Dept. of Neurology, CHU Lariboisière, France

EP

16

TE D

13

SC

a

RI PT

1

g

f

Austria

Neurospin, CEA Saclay, France

Munich Cluster for Systems Neurology (SyNergy), Munich, Germany

* Corresponding author:

Reinhold Schmidt, MD

22

Department of Neurology

23

Clinical Division of Neurogeriatrics

24

Medical University Graz

25

Auenbruggerplatz 22, A-8036 Graz, Austria

2

ACCEPTED MANUSCRIPT 1

Phone: +43-316-385-83397

2

Fax: +43-316-385-14178

3

E-mail: [email protected]

Number of Characters in Title: 71

6

Number of Words in Abstract: 174 words

7

Number of Figures: 3

8

Number of Colour Figures: 0

9

Number of Tables: 3

SC

5

RI PT

4

Number of Supplementary Tables: 1

11

Number of Supplementary Figures: 3

12

Supplemental data file: Figure_e_1, Figure_e_2, Figure_e_3, Table_e_1

13

TE D

14 15

19 20 21 22

AC C

18

EP

16 17

M AN U

10

3

ACCEPTED MANUSCRIPT 1

ABSTRACT

2

Brain iron accumulates during aging and has been associated with

4

neurodegenerative disorders including Alzheimer´s disease. MR-based R2* mapping

5

enables the in vivo detection of iron content in brain tissue. We investigated if during

6

normal brain aging iron load relates to cognitive impairment in region-specific

7

patterns in a community-dwelling cohort of 336 healthy, middle aged and older adults

8

from the Austrian Stroke Prevention Family Study (ASPS-Fam). MR imaging and R2*

9

mapping in the basal ganglia and neocortex was done at 3T. Comprehensive

M AN U

SC

RI PT

3

neuropsychological testing assessed memory, executive function and psychomotor

11

speed.

12

We found the highest iron concentration in the globus pallidus, and pallidal as well as

13

putaminal iron was significantly and inversely associated with cognitive performance

14

in all cognitive domains, except memory. These associations were iron load-

15

dependent. Vascular brain lesions and brain volume did not mediate the relationship

16

between iron and cognitive performance.

17

We conclude that higher R2*-determined iron in the basal ganglia correlate with

18

cognitive impairment during brain aging independent of concomitant brain

19

abnormalities. The prognostic significance of this finding needs to be determined.

AC C

EP

TE D

10

20 21

Keywords:

22 23

All Cognitive Disorders/Dementia, MRI, R2* brain iron mapping, cognition, cognitive aging

24

4

ACCEPTED MANUSCRIPT 1

1. Introduction

2 Iron plays a central role in many biological processes such as cellular aerobic

4

metabolism, neurotransmitter and myelin synthesis. It is also involved in the

5

generation of reactive oxygen species (ROS) (Crichton, 2001, Schipper, 2004, Zecca

6

et al., 2004) and aggregation of α-synuclein (Kostka et al., 2008) in neurons.

7

Iron progressively accumulates in the central nervous system (CNS) as a function of

8

aging and it is preferentially located in the basal ganglia, hippocampus, cerebellar

9

nuclei, and subcortical brain regions (Hallgren and Sourander, 1958). Abnormal

SC

RI PT

3

accumulation of iron has been reported in numerous neurodegenerative and

11

inflammatory CNS disorders (for a review, see Ref. (Brass et al., 2006, Zecca et al.,

12

2004) ). However, it is unclear if increased iron deposition contributes to the

13

pathogenesis of these diseases or only represents an epiphenomenon.

14

Quantitative MRI techniques, such as mapping of the relaxation rate R2* (1/T2*),

15

represent an indirect measure of iron, which has recently been confirmed in a post-

16

mortem study (Langkammer et al., 2010). These techniques also enable in vivo

17

studies and offer the opportunity to correlate regional iron concentrations in brain

18

tissue to clinical phenotypes.

19

If iron-related oxidative stress precedes the hallmark of neuropathological lesions of

20

neurodegeneration, it is likely that regional iron concentrations in brain tissue

21

correlate to clinical findings at the earliest stages of neurodegenerative disorders.

22

So far MRI has been mainly used to evaluate brain iron accumulation in multiple

23

sclerosis (Khalil et al., 2011) and neurodegenerative diseases including Alzheimer’s

24

disease and Parkinson’s disease (Bartzokis et al., 2000, Brass et al., 2006, Thomas

25

et al., 1993). Only few authors (Bartzokis et al., 2007, Pujol et al., 1992, Sullivan et

26

al., 2009) have investigated the association of brain iron and cognition in non-

AC C

EP

TE D

M AN U

10

5

ACCEPTED MANUSCRIPT demented individuals. A recent report has suggested an inverse relationship between

2

iron deposits in the basal ganglia and cognition in the general population (Penke et

3

al., 2012).

4

We extend previous work by reporting the correlations between R2*-based iron

5

mapping and cognitive functioning in the Austrian Stroke Prevention Family Study, a

6

large community-based sample of participants free of history or signs of stroke and

7

dementia. We hypothesized (1) that increased iron concentrations are inversely

8

associated with cognitive performance following region-specific patterns, and (2) that

9

such associations, if any, are mediated by the magnitude of brain atrophy or vascular

SC

brain lesions.

M AN U

10

RI PT

1

11 12

2. Methods

13 2.1. Participants

15

The Austrian Stroke Prevention Family Study (ASPS-Fam) is a prospective single-

16

center, community-based study on the cerebral effects of vascular risk factors in the

17

normal aged population of the city of Graz, Austria. ASPS-Fam represents an

18

extension of the Austrian Stroke Prevention Study (ASPS), which was established in

19

1991 (Schmidt et al., 1994, Schmidt et al., 1999). Between 2006 and 2013, study

20

participants of the ASPS and their first grade relatives were invited to enter ASPS-

21

Fam. Inclusion criteria were no history of previous stroke or dementia and a normal

22

neurologic examination. A total of 381 individuals from 169 families were included

23

into the study. The number of members per family ranged from 2 to 6. The entire

24

cohort underwent a thorough diagnostic work-up including clinical history, laboratory

25

evaluation, cognitive testing, and an extended vascular risk factor assessment. All

26

individuals underwent MRI, except for 26 who had contraindications. Thus, voxel-

AC C

EP

TE D

14

6

ACCEPTED MANUSCRIPT based R2* maps were available in a total of 355 participants. Nineteen participants

2

were excluded because of brain infarcts; therefore the sample for the current study

3

comprises 336 individuals. Dementia was an exclusion criterion and none of the

4

study participants fulfilled the Petersen criteria for amnestic mild cognitive impairment

5

(MCI) (Petersen, 2004). However, 7 individuals whose age was between 61 and 77

6

years had a Mini-Mental State Examination (MMSE) score below 25 with a minimum

7

of 22. One person had a score below 1.5 standard deviations of the mean in the

8

memory domain, but he did not report memory problems. His clinical dementia rating

9

(CDR) score was zero. All individuals with MMSE scores below 25 or memory

SC

RI PT

1

performance below 1.5 standard deviations of the mean had a low educational level

11

with a maximum of 8 years of schooling.

12

The study protocol was approved by the ethics committee of the Medical University of

13

Graz, Austria, and written informed consent was obtained from all participants.

M AN U

10

TE D

14 2.2. Vascular risk factors

16

Assessment of vascular risk factors included arterial hypertension, diabetes mellitus,

17

and cardiac disease and was determined based on history and measurements at the

18

examination as previously described (Schmidt et al., 1999).

AC C

19

EP

15

20

2.3. Neuropsychological testing

21

A test battery assessing memory, executive function as well as psychomotor speed

22

was administered as described previously (Schmidt et al., 1999). The tests employed

23

have been widely used in the German-speaking area and were always applied in the

24

same order and under same laboratory conditions. The tests chosen represent

25

examples of the cognitive domains memory and learning, executive functions and

26

psychomotor speed. Yet, it is important to note that the three groups of tests cannot

7

ACCEPTED MANUSCRIPT be considered statistically independent, and there exists potential overlap of tests

2

across domains.

3

Intermediate memory recall and learning ability was assessed by the “Bäumler’s

4

Lern- und Gedächtnistest” (LGT-3) (Bäumler, 1974) , a highly demanding, paper-

5

pencil procedure consisting of six subtests. Three subtests (word and digit

6

association tasks, and story recall) screen for verbal memory, and two subtests (trail

7

and design recall) screen for figural memory. The sum of scores from these subtests

8

and of an image recognition paradigm result in the total learning and memory

9

performance score. The stimulus sets of the word association task (German-Turkish

10

word pairs), the story (facts about construction of a hospital), and design recall (core

11

symbol and frame), and the recognition paradigm (objects) consist of 20 items each.

12

A trail in an abstracted city map serves as the trail recall test. These sets of stimuli

13

were presented to the person being tested for 1 minute. Two minutes were given for

14

learning the 13 items of the digit association task (three-digit telephone numbers and

15

names of extension holders). During a learning phase the six sets of stimuli are

16

subsequently presented to the person being tested. The recall phase starts

17

immediately thereafter and follows the same order. The delay between presentation

18

and recall for a given subtest ranges between 7 and 11 minutes. Executive functions

19

were tested by the Wisconsin Card Sorting Test (Heaton, 1981), part B of the Trail

20

Making Test (United States War Department., 1944) and Digit Span Backwards,

21

which is part of the Wechsler Adult Intelligence Scale, revised (Tewes, 1991).

22

Adhering to Milner's criteria (Milner B., 1963), the measures computed for the

23

Wisconsin Card Sorting test were categories completed, perseverative errors, and

24

total errors. Psychomotor speed was evaluated by the Purdue Pegboard Test (Tiffin

25

and Asher, 1948). To reduce floor and ceiling artifacts and other sources of

26

measurement error, we used summary measures of cognitive function in the

AC C

EP

TE D

M AN U

SC

RI PT

1

8

ACCEPTED MANUSCRIPT analyses rather than the results of individual tests. We formed composite measures

2

of the specific domains of cognitive function. Each summary measure was calculated

3

by converting individual test scores to z-scores within the group and by computing

4

the average of the scores in each cognitive domain.

5

Global cognitive function was calculated as the mean of all three cognition variables

6

including memory, executive function and psychomotor speed. Supplementary

7

figure_e_1 displays how the neuropsychological tests were combined into domain

8

scores.

SC

RI PT

1

9 2.4. Magnetic Resonance Imaging

11

The study participants underwent conventional MR imaging and R2* mapping on a

12

3T whole-body MR system (TimTrio; Siemens Healthcare, Erlangen, Germany). The

13

conventional protocol included an axial FLAIR sequence (TR=10000ms, TE=70ms,

14

TI=2500ms, number of slices=44, slice thickness=3mm, in-plane

15

resolution=0.86x0.86mm², FOV of 165 x 220 x 132mm 3 ) and a high resolution T1

16

weighted 3D sequence with magnetization preparation and rapid gradient echo

17

(MPRAGE) with whole brain coverage (TR=1900ms, TE=2.19ms, TI=900ms, flip

18

angle=9°, isotropic resolution of 1 mm) for assessi ng brain volume and for

19

segmentation. For R2* mapping, a spoiled 3D multi-echo gradient echo sequence

20

(fast low angle shot (FLASH)) was used (FOV of 256 x 256 x 128mm3, matrix = 256 x

21

256, BW= 190Hz/px, TE of first echo = 4.92ms, echo spacing = 4.92ms, TR=35ms,

22

slice thickness = 2mm, number of slices = 64, number of echoes = 6).

AC C

EP

TE D

M AN U

10

23 24

2.5. Visual and volumetric assessment

25

White matter hyperintensities (WMH) and lacunar infarcts were recorded on FLAIR

26

images as previously described (Schmidt et al., 1999). Lacunes were focal lesions

9

ACCEPTED MANUSCRIPT involving the basal ganglia, the internal capsule, the thalamus, the brainstem or the

2

white matter not exceeding a maximum diameter of 20 mm. WMH were outlined

3

using a home written IDL program (Exelis Visual Information Solutions, USA). They

4

were semiautomatically segmented by combined region growing and local

5

thresholding following manual selection. Total WMH volume (mm³) was calculated

6

from the lesion masks using the program FSLMATHS (FSL, Oxford,

7

www.fmrib.ox.ac.uk). Total brain tissue volume was calculated from the T1weighted

8

MPRAGE images using the fully automated structural image evaluation of

9

FREESURFER (Freesurfer, http://surfer.nmr.mgh.harvard.edu). WMH volume and

SC

RI PT

1

brain tissue volume were normalized for total intracranial volume (TIV)

11

[nVol= ⁄  × 100].

M AN U

10

12 2.6. Regional R2* mapping

14

To determine regional brain iron load, we used R2* mapping technique, which has

15

previously been validated by inductively coupled plasma mass spectrometry in post-

16

mortem studies by our own group (Langkammer et al., 2010). The results of these

17

validation studies indicated a correlation coefficient of 0.95 between R2* values and

18

iron concentration throughout the brain.

19

For regional R2* mapping, the gradient-echo (GRE) sequence with 6 echoes and

20

equidistant echo-spacing of 4.92ms was used. R2* values were calculated for deep

21

grey matter including pallidum, putamen, nucleus caudatus and thalamus as well as

22

for total cortex and hippocampus. We used a voxel based monoexponential fit

23

implemented in MATLAB, which is based on an efficient and stable Levenberg-

24

Marquard-Fletcher method (Balda, 15 Nov 2007 (Updated 18 Feb 2012)), and

25

considers the noise level measured for each echo.

26

In order to prevent fitting errors influencing the regional analysis, highly noisy voxels,

AC C

EP

TE D

13

10

ACCEPTED MANUSCRIPT in which the standard deviation of the fit was less than 0.8, were not taken into

2

account. Regional analysis was performed with FSLSTATS, in which R2* maps were

3

evaluated for regions obtained by FREESURFER segmentation. Region masks,

4

originally segmented in the T1-space, were registered to the GRE-space using FSL-

5

FLIRT (FSL, Oxford, www.fmrib.ox.ac.uk). To avoid partial volume effects introduced

6

by a normal “nearest-neighbor” registration of the masks, we used the “trilinear

7

interpolation” option and a threshold of 90%, which kept significant voxels in the

8

borderline.

SC

RI PT

1

9 2.7. Statistical analysis

11

Assumptions of normal distribution were tested with the Kolmogorov-Smirnov test.

12

Normally distributed variables are reported as mean +/- standard deviation and non-

13

normally distributed variables as median and interquartile range. Due to the bimodal

14

distribution of age we used tertiles of age (38 years to 60 years; 61 years to 70 years;

15

71 years to 86 years) in all statistical models. WMH volume had a skewed distribution

16

and therefore the volumes were natural log transformed.

17

Significant differences in means of normally distributed continuous variables were

18

tested by ANOVA and significant differences of non-normally distributed variables by

19

the Kruskal-Wallis test. Difference in proportions was assessed by Χ2 statistics.

20

Regression analyses were used to assess three major sets of hypotheses. First, we

21

tested the association between MRI findings and R2* based iron measurements

22

(hypothesis set 1). Second, we tested the association between iron load and

23

cognition (hypothesis set 2), and third, we determined the association between iron

24

load and cognition mediated by MRI findings (hypothesis set 3).

25

The association between MRI findings and iron load was assessed in six mixed

26

models, with the MRI variables brain volume, WMH volume and presence of lacunes

AC C

EP

TE D

M AN U

10

11

ACCEPTED MANUSCRIPT being the predictor variables, and iron content in each brain region including total

2

cortex, hippocampus, pallidum, putamen, caudate nucleus and thalamus being the

3

dependent variable. In total, 24 comparisons were made.

4

The association between R2* based iron measurements and cognition was tested in

5

24 mixed model analyses, with iron load in each of the six brain regions as the

6

predictor variable and each cognitive variable (memory, executive function,

7

psychomotor speed and global cognitive function) as dependent variable.

8

This association was also tested for verbal and figural memory separately in 12

9

additional analyses, as well as in each of the three age groups which resulted in 72

SC

RI PT

1

comparisons, in addition.

11

To determine the shape of iron load-related effects in regions significantly associated

12

with cognition, participants were categorized into quartiles according to iron value

13

distribution, and 8 mixed models adjusted for possible confounders with iron quartiles

14

being the explanatory variables and cognitive performance being the dependent

15

variable were calculated. The lowest quartile of iron load served as the reference. A

16

p-value for the linear trend of the iron load-related effect was calculated by including

17

the quartiles as a continuous variable in the mixed models.

18

To account for the sample relationships in the current family-based study, a random

19

effect was added to each model using a kinship matrix describing the family structure

20

(Newton-Cheh et al., 2009, Suchindran et al., 2009).

21

To test the hypothesis that R2*-based iron effects on cognition were mediated by

22

brain atrophy or vascular brain lesions we used simple mediation models for

23

estimating indirect effect size (Preacher and Hayes, 2004). Mediation was evaluated

24

in regions significantly associated with cognition for each of the structural brain

25

lesions separately and for each significant cognitive function, resulting in 36 models

26

(two models for each combination of predictor, mediator and independent variable).

AC C

EP

TE D

M AN U

10

12

ACCEPTED MANUSCRIPT We determined whether the influence of R2*-based iron load on cognition is

2

explained by brain volume, WMH volume and presence of lacunar infarcts. Mediator

3

effect sizes and 95% confidence intervals were estimated using a bootstrap-based

4

method developed by Preacher and Hayes (Preacher and Hayes, 2008). If the 95%

5

CI of the indirect effect does not contain zero, a significant mediating effect is

6

probable, whereas no mediation is present if zero is included in the 95% CI.

7

To determine the proportion of the age-related variance in cognition that is accounted

8

for by the R2*-based iron in regions significantly associated with cognition, we used a

9

method developed by Lindenberger and Pötter (Lindenberger and Pötter, 1998). The

SC

RI PT

1

proportion of the variance in cognition that is shared by age and R2* based iron is

11

defined as the shared effect of age and R2* based iron divided by the simple effect of

12

age. The simple age effect is the coefficient of determination (R2) of a model with age

13

as the predictor variable and cognition as the dependent variable. In order to obtain

14

the shared effect of age and R2* based iron, the unique age effect is subtracted from

15

the simple age effect. The unique age effect is the difference in the coefficients of

16

determination of two models with R2* based iron, and R2* based iron and age

17

respectively as independent variables and cognition as dependent variable.

18

All models (except the models for the determination of the age-related variance) were

19

adjusted for potential confounders to evaluate the independent effect of iron on

20

structural brain abnormalities and cognition. We considered age, sex, education,

21

hypertension, diabetes and cardiac disease as possible confounders. Multicollinearity

22

was assessed between the independent variables of the models using the variance

23

inflation factor (VIF). A VIF value > 10 was considered to be an indicator of

24

multicollinearity. There was no indication for multicollinearity in the models. The mean

25

VIF for predictor variables in our study was 1.53 (range 1.06 – 2.57).

AC C

EP

TE D

M AN U

10

13

ACCEPTED MANUSCRIPT For each regression coefficient ß, the 95% confidence interval and the p-value were

2

determined. A two-sided p-value <0.05 was considered to be statistically significant.

3

Multiple testing correction was performed using the Benjamini-Hochberg false

4

discovery rate (FDR) method (Benjamini and Hochberg, 1995). We corrected the p-

5

values of all analyses in hypothesis sets 1 and 2, which are 56 tests in total. We

6

report corrected p-values in the result section, and corrected as well as uncorrected

7

p-values in the tables. Mixed models and kinship matrices were calculated using the

8

coxme (Therneau, 2012) and kinship2 (Therneau, 2013) packages in R (R: A

9

language and Environment for Statistical Computing, Vienna, Austria. www.R-

SC

RI PT

1

project.org). R was also used to correct the p-values for multiple testing with the p-

11

adjust function.

12

Multicollinearity check and mediation analyses were performed with SPSS (IBM

13

Statistics for Windows, Version 20, Armonk, New York).

M AN U

10

15

TE D

14 3. Results

16

Demographics, frequency of vascular risk factors, neuropsychological test results and

18

MRI findings of the total cohort and in tertiles of age are displayed in table 1. Higher

19

age was associated with lower educational level, higher frequency of risk factors,

20

worse cognitive performance, greater extent of focal brain lesions and lower brain

21

volume. Figure 1 gives R2* values as an indirect measure of iron content in various

22

brain regions. As can be seen from this figure, iron load is highly variable depending

23

on brain region. The highest concentrations were observed in the pallidum, followed

24

by the putamen and caudate nucleus. Increased iron content correlated with higher

25

age in putamen (β=2.55; 95%CI 1.12, 3.97; p=0.01) only. No association was

26

detected between iron content in any brain region and sex (data not shown). Also,

AC C

EP

17

14

ACCEPTED MANUSCRIPT there were no significant associations between R2* values in any of the brain regions

2

and structural MRI changes including WMH volume and brain volume. Lacunes

3

related to higher R2* values in total cortex (β=0.11; 95%CI 0.03, 0.19; p=0.047), but

4

not in other tissue compartments. Table 2 presents the relationship between regional

5

R2* values and cognition. As can be seen from this table, higher iron load in the

6

pallidum related inversely with all cognitive measures, except memory.

7

Other significant associations existed between R2* in putamen and global cognitive

8

function and psychomotor speed. There were no significant relationships between

9

R2* values in neocortex and hippocampus and cognition.

SC

RI PT

1

Figure 2 and 3 are scatterplots displaying the inverse relationship between R2*-

11

based iron in the pallidum and in the putamen and domain-specific

12

neuropsychological test scores.

13

Neither verbal nor figural memory as determined by “Bäumler´s Lern und

14

Gedächtnis-Test” were significantly related to R2* values in the various brain regions

15

including the hippocampus (data not shown).

16

When assessing the associations between regional iron content and cognitive

17

functioning within tertiles of age, the associations were strongest in the highest age

18

group ranging from 71 to 86 years (see supplementary table_e_1). There existed an

19

almost linear relationship between quartiles of R2* iron load in the pallidum and

20

impairment in global cognitive function, executive function and psychomotor speed

21

(supplementary figure e_2). Similar results were seen between quartiles of R2* iron

22

load in the putamen and psychomotor speed (supplementary figure_e_3).

23

None of the structural MRI changes mediated the relationship between R2*

24

measured iron content in pallidum and putamen and cognition (table 3).

25

As determined by the method of Lindenberger and Pötter (Lindenberger and Pötter,

26

1998) the R2* based iron content in the pallidum accounted for 9% of the age-

AC C

EP

TE D

M AN U

10

15

ACCEPTED MANUSCRIPT 1

related variance in executive function, 7% in global cognitive function and 8% in

2

psychomotor speed results of study participants. R2* based iron load in the putamen

3

accounted for 24% of the age-related variance in executive function, 18% in global

4

cognitive function and 21% in psychomotor speed results.

RI PT

5 4. Discussion

7

In the present study, the largest investigation of brain iron accumulation in

8

community-dwelling individuals performed so far, we observed, as previously

9

reported, that the highest concentration of iron is in the pallidum, followed by

SC

6

putamen and caudate nucleus (Aoki et al., 1989, Drayer et al., 1986, Hallgren and

11

Sourander, 1958, Pfefferbaum et al., 2009). We confirmed Bartzokis´ (Bartzokis et

12

al., 1997) as well as Hallgren´s and Sourander´s (Hallgren and Sourander, 1958)

13

suggestion for the thalamus being an exception for iron accumulation by showing a

14

slow decrease in iron levels in old age.

15

Our results demonstrate that iron load in specific brain regions is actually related to

16

cognitive functioning. This was particularly true for iron in the pallidum and putamen,

17

while associations between iron load in other brain regions and cognition were

18

inconsistent. R2* based iron content in the pallidum accounted for 7% to 9% of the

19

age-related variance in domain specific cognitive functions. The effect of putaminal

20

iron load was considerably larger, it accounted for 18% of the age-related variance in

21

global cognitive function and 24% in executive function. The association between

22

iron in structures of the basal ganglia and cognition was strongest in the highest

23

tertile of age ranging from 71 to 86 years. The mechanisms underlying the

24

association between basal ganglial iron deposition and cognitive abilities are unclear.

25

Our findings argue against the view that iron accumulation represents only an

26

epiphenomenon (Zecca et al., 2004) of brain atrophy and vascular cerebral damage

AC C

EP

TE D

M AN U

10

16

ACCEPTED MANUSCRIPT which by itself might be the cause for both, increased iron deposition and cognitive

2

impairment. In such instance iron load in the pallidum as well as in the putamen

3

should have been related to these brain abnormalities and, moreover the association

4

between pallidal and putaminal iron and cognition should have been mediated by

5

either focal or diffuse brain abnormalities or both. We failed to observe such

6

relationships and thus contrast a recent study in patients with CADASIL, in which

7

cerebral small vessel disease-related brain lesions were associated to basal

8

ganglionic iron accumulation (Liem et al., 2012). A reason for these discrepant results

9

may be that the amount of brain lesions in CADASIL patients by far exceeds the

SC

RI PT

1

magnitude of cerebral damage that occurs with advancing age. The lack of an

11

association between iron load, white matter damage and brain atrophy in our study is

12

also against the view that age-related demyelination of white matter tracts or cortical

13

atrophy may lead to increased iron release and finally contribute to chronic

14

accumulation of iron in the basal ganglia (Bartzokis et al., 2007).

15

A causal relationship between iron load and cognition cannot be inferred by our

16

data. It is of note however, that increasing iron content in the basal ganglia related to

17

increasing executive dysfunction and psychomotor speed dysfunction. No such

18

association was seen for memory. This pattern of cognitive dysfunction is consistent

19

with involvement of frontal-subcortical circuits (Bonelli and Cummings, 2007, Duering

20

et al., 2011, Krause et al., 2012) which originate in the prefrontal cortex, project to the

21

striatum, connect to the pallidum and substantia nigra and from there to the thalamus

22

(Alexander et al., 1986, Purves et al., 2001). Moreover, we found an almost linear

23

association between the amount of R2*-determined iron in pallidum as well as in

24

putamen and cognitive performance, which also suggests a close relationship

25

between the two measures.

AC C

EP

TE D

M AN U

10

17

ACCEPTED MANUSCRIPT Multiple mechanisms may be responsible for direct damaging effects of increased

2

iron load. Iron- derived radical oxygen species (ROS) may lead to membrane

3

damage, abnormal cell signalling, mitochondrial and proteosomal dysfunction,

4

pathologic protein accumulation and inclusion body formation (Schipper, 2004).

5

Electrophysiological imbalances and synaptolysis, as well as apoptosis and death of

6

neurons and glial cells have also been reported (Schipper, 2004). Moreover, there is

7

evidence that metallochemical reactions resulting in formation of ROS might be the

8

common denominator underlying a spectrum of neurodegenerative diseases

9

including Alzheimer’s disease, amyotrophic lateral sclerosis, prion disease, cataracts,

SC

RI PT

1

mitochondrial disorders (Friedreich’s ataxia), and Parkinson’s disease (Bush, 2000,

11

Connor, 1997, Hirsch and Faucheux, 1998, Jellinger, 1999).

12

A limitation of our investigation lies in its cross sectional design which does not permit

13

to conclusively infer on a cause-effect relationship. The unspecific measurement of

14

iron, in which primary ferritin which stores iron in a non–toxic but bioavailable form, is

15

measured, is another shortcoming. However, at this point methods for specific

16

measurement of iron in its toxic form are not available.

17

We realize that although statistically significant, the magnitude of cognitive

18

impairment associated with increasing iron load seen in our study was small. The

19

clinical importance of iron accumulation during normal aging is currently unknown.

20

Longitudinal studies are now needed to explore if increasing iron content in the basal

21

ganglia indeed reflects age-related neurodegenerative processes which may

22

ultimately result in clinically relevant cognitive impairment or even dementia. Data on

23

the rate of iron accumulation during normal aging are yet not available.

24

Acknowledgements

25

Study funding:

AC C

EP

TE D

M AN U

10

18

ACCEPTED MANUSCRIPT 1

This work was supported by FP6 ERA-NET NEURON grant (01 EW1207) and by the

2

Austrian Science Fund (FWF) project I904

3 Financial disclosure

5

Christine Ghadery, Lukas Pirpamer, Edith Hofer, Christian Langkammer, Katja

6

Petrovic and Marisa Loitfelder report no disclosures. Petra Schwingenschuh received

7

funding to attend conferences from Boehringer Ingelheim, GlaxoSmithKline, Ipsen,

8

Novartis, UCB, and Merz pharma companies; received speaker honoraria from UCB;

9

and served on national advisory boards of UCB and Novartis. Stephan Seiler reports

10

no disclosures. Marco Duering received travel expenses for lectures from the United

11

Leukodystrophy Foundation, the European Stroke Conference and Kenes

12

International. Eric Jouvent reports no disclosures. Helena Schmidt received research

13

support by the Austrian Science Fund (Project P20545). Franz Fazekas serves on

14

scientific advisory boards for Bayer-Schering, Biogen Idec, Genzyme, Merck Serono,

15

Pfizer, Novartis, Perceptive Informatics, and Teva Pharmaceutical Industries Ltd.;

16

serves on the editorial boards of Cerebrovascular Diseases, Multiple Sclerosis, the

17

Polish Journal of Neurology and Neurosurgery, Stroke, and the Swiss Archives of

18

Neurology and Psychiatry; and has received speaker honoraria and support from

19

Biogen Idec, Bayer-Schering, Merck Serono, Novartis, Sanofi-Aventis, and Teva

20

Pharmaceutical Industries Ltd.. Jean-Francois Mangin reports no disclosures.

21

Hugues Chabriat served on scientific advisory boards of trials for Lundbeck, Servier

22

and Janssen and serves as an editorial board member of Cerebrovascular Diseases.

23

Martin Dichgans receives research support from the German Research Foundation

24

(DFG), German Federal Ministry of Education and Research (BMBF), the European

25

Union (EU FP6 and FP7), Vascular Dementia Research Foundation, Corona

26

Foundation and Jackstädt Foundation. He received travel expenses and honoraria

AC C

EP

TE D

M AN U

SC

RI PT

4

19

ACCEPTED MANUSCRIPT for lectures from Bayer Vital, Boehringer Ingelheim Pharma, Heel, Bristol-Meyer

2

Squibb, Lundeck, Sanofi-Aventis, Ever Pharma, and Shire as well as for educational

3

activities not funded by industry, received honoraria for articles for Thieme,

4

UpToDate and Kohlhammer, serves as an editor for Stroke, the International Journal

5

of Stroke, the Journal of Neurochemistry, as an editorial board member of

6

Cerebrovascular Diseases, and serves as a consultant for Bayer Vital, Boehringer

7

Ingelheim, Bristol-Meyer Squibb, Ever Pharma, and Heel. Stefan Ropele reports no

8

disclosures. Reinhold Schmidt receives research support from the Austrian Science

9

Fund and travel expenses and honoraria for lectures from Pfizer, Novartis, Janssen,

SC

RI PT

1

Merz Austria, Lundbeck, and Ever Pharma.

11

Disclosure statement

12

All authors have no actual or potential conflicts of interest to report. The study

13

protocol was approved by the ethics committee of the Medical University of Graz,

14

Austria, and written informed consent was obtained from all participants.

15

References

16 17 18

Alexander, G.E., DeLong, M.R., Strick, P.L., 1986. Parallel organization of functionally segregated circuits linking basal ganglia and cortex. Annu.Rev.Neurosci. 9, 357-381.

19 20 21

Aoki, S., Okada, Y., Nishimura, K., Barkovich, A.J., Kjos, B.O., Brasch, R.C., Norman, D., 1989. Normal deposition of brain iron in childhood and adolescence: MR imaging at 1.5 T. Radiology. 172, 381-385.

22 23

Balda, M., 15 Nov 2007 (Updated 18 Feb 2012). LMFnlsq, efficient and stable Levenberg-Marquard-Fletcher method for solving of nonlinear equations.

24 25 26

Bartzokis, G., Beckson, M., Hance, D.B., Marx, P., Foster, J.A., Marder, S.R., 1997. MR evaluation of age-related increase of brain iron in young adult and older normal males. Magn.Reson.Imaging. 15, 29-35.

27 28 29

Bartzokis, G., Sultzer, D., Cummings, J., Holt, L.E., Hance, D.B., Henderson, V.W., Mintz, J., 2000. In vivo evaluation of brain iron in Alzheimer disease using magnetic resonance imaging. Arch.Gen.Psychiatry. 57, 47-53.

AC C

EP

TE D

M AN U

10

20

ACCEPTED MANUSCRIPT Bartzokis, G., Tishler, T.A., Lu, P.H., Villablanca, P., Altshuler, L.L., Carter, M., Huang, D., Edwards, N., Mintz, J., 2007. Brain ferritin iron may influence age- and gender-related risks of neurodegeneration. Neurobiol.Aging. 28, 414-423.

4 5

Bäumler, G., 1974. Lern und Gedächtnistest (LGT 3). Verlag Psychologie, Göttingen, Germany.

6 7

Benjamini, Y., Hochberg, Y., 1995. Controlling the false discovery rate: a practical and powerful approach to multiple testing., 289-300.

8 9

Bonelli, R.M., Cummings, J.L., 2007. Frontal-subcortical circuitry and behavior. Dialogues Clin.Neurosci. 9, 141-151.

RI PT

1 2 3

Brass, S.D., Chen, N.K., Mulkern, R.V., Bakshi, R., 2006. Magnetic resonance imaging of iron deposition in neurological disorders. Top.Magn.Reson.Imaging. 17, 31-40.

13

Bush, A.I., 2000. Metals and neuroscience. Curr.Opin.Chem.Biol. 4, 184-191.

14 15

Connor, J.R., 1997. Metals and oxidative damage in neurological disorders. Plenum Press, New York.

16 17

Crichton, R.R., 2001. Inorganic Biochemistry of Iron Metabolism: From Molecular Mechanism to Clinical Consequences. John Wiley& Sons, Chichester.

18 19

Drayer, B., Burger, P., Darwin, R., Riederer, S., Herfkens, R., Johnson, G.A., 1986. MRI of brain iron. AJR Am.J.Roentgenol. 147, 103-110.

20 21 22 23

Duering, M., Zieren, N., Herve, D., Jouvent, E., Reyes, S., Peters, N., Pachai, C., Opherk, C., Chabriat, H., Dichgans, M., 2011. Strategic role of frontal white matter tracts in vascular cognitive impairment: a voxel-based lesion-symptom mapping study in CADASIL. Brain. 134, 2366-2375.

24 25

Hallgren, B., Sourander, P., 1958. The effect of age on the non-haemin iron in the human brain. J.Neurochem. 3, 41-51.

26 27

Heaton, R., 1981. Wisconsin card sorting test manual. Psychological Assessment Resources, Odessa, FL.

28 29

Hirsch, E.C., Faucheux, B.A., 1998. Iron metabolism and Parkinson's disease. Mov.Disord. 13 Suppl 1, 39-45.

30 31

Jellinger, K.A., 1999. The role of iron in neurodegeneration: prospects for pharmacotherapy of Parkinson's disease. Drugs Aging. 14, 115-140.

32 33 34 35

Khalil, M., Langkammer, C., Ropele, S., Petrovic, K., Wallner-Blazek, M., Loitfelder, M., Jehna, M., Bachmaier, G., Schmidt, R., Enzinger, C., Fuchs, S., Fazekas, F., 2011. Determinants of brain iron in multiple sclerosis: a quantitative 3T MRI study. Neurology. 77, 1691-1697.

AC C

EP

TE D

M AN U

SC

10 11 12

21

ACCEPTED MANUSCRIPT Kostka, M., Hogen, T., Danzer, K.M., Levin, J., Habeck, M., Wirth, A., Wagner, R., Glabe, C.G., Finger, S., Heinzelmann, U., Garidel, P., Duan, W., Ross, C.A., Kretzschmar, H., Giese, A., 2008. Single particle characterization of iron-induced pore-forming alpha-synuclein oligomers. J.Biol.Chem. 283, 10992-11003.

5 6 7

Krause, M., Mahant, N., Kotschet, K., Fung, V.S., Vagg, D., Wong, C.H., Morris, J.G., 2012. Dysexecutive behaviour following deep brain lesions--a different type of disconnection syndrome?. Cortex. 48, 97-119.

8 9 10

Langkammer, C., Krebs, N., Goessler, W., Scheurer, E., Ebner, F., Yen, K., Fazekas, F., Ropele, S., 2010. Quantitative MR imaging of brain iron: a postmortem validation study. Radiology. 257, 455-462.

11 12 13 14

Liem, M.K., Lesnik Oberstein, S.A., Versluis, M.J., Maat-Schieman, M.L., Haan, J., Webb, A.G., Ferrari, M.D., van Buchem, M.A., van der Grond, J., 2012. 7 T MRI reveals diffuse iron deposition in putamen and caudate nucleus in CADASIL. J.Neurol.Neurosurg.Psychiatry. 83, 1180-1185.

15 16 17

Lindenberger, U., Pötter, U., 1998. The complex nature of unique and shared effects in hierarchical linear regression: Implications for developmental psychology., 218230.

18 19

Milner B., 1963. Effects of different brain lesion on card sorting. The role of the frontal lobes., 100-110.

20 21 22 23 24 25

Newton-Cheh, C., Eijgelsheim, M., Rice, K.M., de Bakker, P.I., Yin, X., Estrada, K., Bis, J.C., Marciante, K., Rivadeneira, F., Noseworthy, P.A., Sotoodehnia, N., Smith, N.L., Rotter, J.I., Kors, J.A., Witteman, J.C., Hofman, A., Heckbert, S.R., O'Donnell, C.J., Uitterlinden, A.G., Psaty, B.M., Lumley, T., Larson, M.G., Stricker, B.H., 2009. Common variants at ten loci influence QT interval duration in the QTGEN Study. Nat.Genet. 41, 399-406.

26 27 28 29

Penke, L., Valdes Hernandez, M.C., Maniega, S.M., Gow, A.J., Murray, C., Starr, J.M., Bastin, M.E., Deary, I.J., Wardlaw, J.M., 2012. Brain iron deposits are associated with general cognitive ability and cognitive aging. Neurobiol.Aging. 33, 510-517.e2.

30 31

Petersen, R.C., 2004. Mild cognitive impairment as a diagnostic entity. J.Intern.Med. 256, 183-194.

32 33 34

Pfefferbaum, A., Adalsteinsson, E., Rohlfing, T., Sullivan, E.V., 2009. MRI estimates of brain iron concentration in normal aging: comparison of field-dependent (FDRI) and phase (SWI) methods. Neuroimage. 47, 493-500.

35 36 37

Preacher, K.J., Hayes, A.F., 2008. Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behav.Res.Methods. 40, 879-891.

38 39 40

Preacher, K.J., Hayes, A.F., 2004. SPSS and SAS procedures for estimating indirect effects in simple mediation models. Behav.Res.Methods Instrum.Comput. 36, 717731.

AC C

EP

TE D

M AN U

SC

RI PT

1 2 3 4

22

ACCEPTED MANUSCRIPT Pujol, J., Junque, C., Vendrell, P., Grau, J.M., Marti-Vilalta, J.L., Olive, C., Gili, J., 1992. Biological significance of iron-related magnetic resonance imaging changes in the brain. Arch.Neurol. 49, 711-717.

4 5

Purves, D., Augustine, G., Fitzpatrick, D., et al., 2001. Circuits within the basal ganglia., in: Anonymous Neuroscience. Sinauer Associates, Sunderland (MA).

6 7

Schipper, H.M., 2004. Brain iron deposition and the free radical-mitochondrial theory of ageing. Ageing Res.Rev. 3, 265-301.

8 9 10

Schmidt, R., Fazekas, F., Kapeller, P., Schmidt, H., Hartung, H.P., 1999. MRI white matter hyperintensities: three-year follow-up of the Austrian Stroke Prevention Study. Neurology. 53, 132-139.

11 12 13 14

Schmidt, R., Lechner, H., Fazekas, F., Niederkorn, K., Reinhart, B., Grieshofer, P., Horner, S., Offenbacher, H., Koch, M., Eber, B., 1994. Assessment of cerebrovascular risk profiles in healthy persons: definition of research goals and the Austrian Stroke Prevention Study (ASPS). Neuroepidemiology. 13, 308-313.

15 16 17 18

Suchindran, S., Vana, A.M., Shaffer, R.A., Alcaraz, J.E., McCarthy, J.J., 2009. Racial differences in the interaction between family history and risk factors associated with diabetes in the National Health and Nutritional Examination Survey, 1999-2004. Genet.Med. 11, 542-547.

19 20 21 22

Sullivan, E.V., Adalsteinsson, E., Rohlfing, T., Pfefferbaum, A., 2009. Relevance of Iron Deposition in Deep Gray Matter Brain Structures to Cognitive and Motor Performance in Healthy Elderly Men and Women: Exploratory Findings. Brain Imaging Behav. 3, 167-175.

23 24

Tewes, U., 1991. Hamburg-wechsler Intelligenztest für Erwachsene, Revision 1991. Manual. Bern, Switzerland.

25 26

Thomas, L.O., Boyko, O.B., Anthony, D.C., Burger, P.C., 1993. MR detection of brain iron. AJNR Am.J.Neuroradiol. 14, 1043-1048.

27 28

Tiffin, J., Asher, E.J., 1948. The Purdue pegboard; norms and studies of reliability and validity. J.Appl.Psychol. 32, 234-247.

29 30 31

United States War Department., 1944. Army individual test battery manual of directions and scoring. United States War Department, Adjutant General's Office, Washington, DC.

32 33

Zecca, L., Youdim, M.B., Riederer, P., Connor, J.R., Crichton, R.R., 2004. Iron, brain ageing and neurodegenerative disorders. Nat.Rev.Neurosci. 5, 863-873.

34 35

AC C

EP

TE D

M AN U

SC

RI PT

1 2 3

ACCEPTED MANUSCRIPT

23 Table 1. Demographics, risk factors, neuropsychological test performance and MRI findings of study participants. Age category 2 (61-70 years)

Age category 3 (71-86 years)

p

336 (100) 204 (60.7)

97 (28.9) 53 (54.6)

123 (36.6) 82 (66.7)

116 (34.5) 69 (59.5)

0.183

68 (20.2) 142 (42.3) 70 (20.8) 56 (16.7)

1 (1.0) 36 (37.1) 30 (30.9) 30 (30.9)

38 (30.9) 62 (50.4) 9 (7.3) 14 (11.4)

29 (25.0) 44 (37.9) 31 (26.7) 12 (10.3)

<0.001 a 0.071 a <0.001 a <0.001

212 (63.1) 33 (9.8) 79 (23.5)

41 (42.3) 3 (3.1) 10 (10.3)

84 (68.3) 15 (12.2) 29 (23.6)

87 (75.0) 15 (12.9) 40 (34.5)

<0.001 a 0.030 a <0.001

-1.75 – 2.21 -3.17 – 3.50 -2.78 – 1.36 -2.39 – 3.14 28.13 (1.47) 22 - 30

-1.16 – 2.21 -0.99 – 3.50 -1.17 – 1.36 -2.01 – 3.14 28.75 (1.01) 25 - 30

-1.13 – 1.13 -1.70 – 2.66 -1.82 – 0.95 -2.01 – 2.08 27.99 (1.56) 23 - 30

-1.75 - 1.22 -3.17 – 1.74 -2.78 – 1.13 -2.39 – 1.73 27.77 (1.55) 22 - 30

<0.001 * <0.001 * <0.001 * <0.001 * <0.001

2 (2.1)

11 (8.9)

11 (9.5)

0.069

70.89 (3.73)

73.02 (4.37)

70.54 (3.01)

69.46 (2.99)

<0.001

0.27 (0.15 – 0.52)

0.16 (0.08 – 0.29)

0.31 (0.17 – 0.53)

0.44 (0.23 – 0.80)

<0.001

SC

M AN U

TE D

EP

N (%) Women N (%) Education N (%) Primary school N (%) Apprenticeship N (%) High school diploma N (%) University degree N (%) B. Risk factors Hypertension N (%) Diabetes N (%) Cardiac disease N (%) C. Neuropsychological testing 1 Global cognitive function (range) 1 Memory (range) 1 Executive function (range) 1 Psychomotor speed (range) MMSE (mean, SD) MMSE (range) D. MRI variables Lacunes N (%) Brain volume normalized for TIV in % mean(SD) WMH volume normalized for TIV in % (median, IQR)

67.00 (55-72)

24 (7.1)

AC C

Age, years (median, IQR)

RI PT

Age category 1 (38-60 years)

Age (total)

a

a

a

*

a

*

b

Abbreviations: MRI= Magnetic Resonance Imaging; WMH= White Matter Hyperintensities; SD= Standard Deviation; IQR= interquartile range; MMSE= mini a

2

mental state examination; normalized in % of TIV= nVol=  ⁄  × 100; nVol= normalized Volume; TIV= total intracranial volume. Χ test, *ANOVA test, 1

Kruskal-Wallis test, Z-values.

b

ACCEPTED MANUSCRIPT

24

Global cognitive function (G-factor)

RI PT

Table 2. Multivariate linear regression analysis¹: R2* -based iron in different brain regions and cognitive functioning

Memory

95% CI

p†

corrected p*

Cortex Hippocampus Pallidum Putamen Caudatus Thalamus

0,0193 0,0046 -0,0192 -0,0165 -0,0156 0,0129

[-0.037, 0.075] [-0.015, 0.025] [-0.029, -0.010] [-0.029, -0.004] [-0.033, 0.002] [-0.025, 0.050]

0,50000 0,65000 0,00008 0,01000 0,07900 0,50000

0,66182 0,71373 0,00218 0,04667 0,26024 0,66182

95% CI

p†

Cortex Hippocampus Pallidum Putamen Caudatus Thalamus

0,0588 0,0193 -0,0158 -0,0189 -0,0145 0,0204

[-0.008, 0.126] [-0.004, 0.043] [-0.028, -0.004] [-0.038, -0.004] [-0.035, 0.006] [-0.024, 0.065]

0,08600 0,11000 0,00710 0,01300 0,17000 0,37000

AC C

0,0713 0,0095 -0,0134 -0,0075 -0,0085 0,0352

95% CI

p†

corrected p*

[-0.020, 0.163] [-0.023, 0.042] [-0.029, 0.002] [-0.028, 0.013] [-0.037, 0.020] [-0.026, 0.096]

0,13000 0,56000 0,09800 0,48000 0,56000 0,26000

0,331 0,665 0,289 0,662 0,665 0,502

Psychomotor speed

corrected p*

β

95% CI

p†

corrected p*

0,26756 0,30800 0,04667 0,05600 0,38080 0,57556

-0,0657 -0,0099 -0,0307 -0,0275 -0,0275 -0,0282

[-0.149, 0.018] [-0.040, 0.020] [-0.045, -0.017] [-0.046, -0.009] [-0.053, -0.002] [-0.084, 0.028]

0,12000 0,52000 0,00002 0,00330 0,03400 0,32000

0,32000 0,66182 0,00118 0,03080 0,11900 0,54303

EP

β

TE D

Executive function

Brain Region

β

SC

β

M AN U

Brain Region

¹Adjusted for age, sex, education, hypertension, diabetes, cardiac diseases and family structure † uncorrected p-value *corrected p-value (Benjamini Hochberg FDR)

ACCEPTED MANUSCRIPT

25 Table 3. Mediation models1 assessing the effect of brain volume, WMH volume and number of lacunes on the relationship between

Brain volume Total effect

Direct effect

Indirect effect

Bootstrapped CI

-0.0183

-0.0178

-0.0005

-0.0034; 0.0005

-0.0147

-0.0139

-0.0008

-0.0034; 0.0008

-0.031

-0.0304

-0.0006

-0.0045; 0.0006

WMH volume -0.0183

-0.0177

-0.0025

-0.0028; 0.0004

-0.0157

-0.0159

0.0004

-0.0012; 0.002

-0.0326

-0.0316

-0.0044

-0.0044; 0.0009

TE D

Global cognitive function(n=297 Executive function(n=301) Psychomotor speed (n=302)

Number of lacunes -0.0185

0.0003

-0.0001; 0.0013

-0.0147

-0.0147

0

-0.0007; 0.0002

-0.0305

-0.031

0.0005

EP

-0.0182

-0.0001; 0.0017

AC C

Global cognitive function(n=330) Executive function(n=335) Psychomotor speed (n=335)

Global cognitive function (n=328) Executive function (n=333) Psychomotor speed (n=333)

Brain volume

Total effect

Direct effect

Indirect effect

Bootstrapped CI

-0.0168

-0.0159

-0.0009

-0.0041; 0.0004

-0.0188

-0.0175

-0.0013

-0.0052; 0.0008

-0.0271

-0.026

-0.0011

-0.0054; 0.0006

SC

Global cognitive function (n=328) Executive function (n=333) Psychomotor speed (n=333)

R2* in Putamen

M AN U

R2* in Pallidum

RI PT

R2*-based iron in pallidum and putamen and cognitive performance.

Global cognitive function(n=297 Executive function(n=301) Psychomotor speed (n=302)

WMH volume -0.0173

-0.0165

-0.0008

-0.0037; 0.0002

-0.0206

-0.0208

0.0002

-0.0005; 0.0022

-0.0265

-0.0253

-0.0012

-0.0058; 0.0012

Number of lacunes Global cognitive function(n=330) Executive function(n=335) Psychomotor speed (n=335)

-0.017

-0.0173

0.0003

-0.0003; 0.0022

-0.0186

-0.0186

0

-0.0011; 0.0003

-0.0277

-0.0283

0.0006

-0.0005; 0.0032

Abbreviation: WMH volume= White matter hyperintensities volume 1

Dependent variable: global cognitive function, executive function and psychomotor speed. Independent Variable: R2* in pallidum and in putamen; Mediators are brain volume,

WMH volume and number of lacunes. The models are adjusted for age, sex, education, hypertension, diabetes and cardiac diseases. Note zero was always within the bootstrapped CI of the indirect effect, indicating that brain abnormalities had no mediating effects on the association between pallidal and putaminal R2* and cognitive functioning. Due to missing data the effective sample size varied between 297 and 335.

ACCEPTED MANUSCRIPT

26 Figure 1. R2*-based iron in different brain regions. The median of regional R2* values is displayed by the central line within the box.

RI PT

Box edges represent the 25th and 75th percentile values. Whiskers extend to the most extreme data points not considering outliers. Outliers are displayed by circles and most extreme outliers by asterisks.

SC

Figure 2. Scatterplots showing the correlation between R2*-based iron in pallidum and domain-specific neuropsychological test

M AN U

scores. X-axis displays z-values of neuropsychological test scores, y-axis gives R2*-based iron in pallidum. Pearson´s correlation showed significant associations of R2*-based iron with scores of global cognitive function (r=-0.231, p<0.001), memory (r=-0.129, p=0.019), executive function (r=-0.196, p<0.001), and psychomotor speed (r=-0.261, p<0.001).

TE D

Figure 3. Scatterplots showing the correlation between R2*-based iron in putamen and domain-specific neuropsychological test scores. X-axis displays z-values of neuropsychological test scores, y-axis gives R2*-based iron in putamen. Pearson´s correlation

EP

showed significant associations of R2*-based iron with scores of global cognitive function (r=-0.295, p<0.001), memory (r=-0.192,

AC C

p<0.001), executive function (r=-0.252, p<0.001), and psychomotor speed (r=-0.302, p<0.001).

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT Highlights

We examine the effect of R2*-based brain iron and cognition in healthy individuals.

•

Higher R2*-based iron in the basal ganglia correlate with cognitive impairment.

•

These associations are iron load-dependent.

•

Concomitant brain abnormalities do not mediate these findings.

AC C

EP

TE D

M AN U

SC

RI PT

•