Program Number 185.11
Poster Number FFF10
Sun. AM, Hall B-H
Individual differences in brain structure and functional connectivity related to
body mass index (BMI) and body fat percentage (BFP)
1,2,3,4
Chase R. Figley
Laboratory of Working Memory
and Cognition
1
1
1,2,3
, Erica Levenbaum , Judith S.A. Asem , and Susan M. Courtney
1
2
3
Department of Psychological & Brain Sciences; Department of Neuroscience; and F.M. Kirby Research Center for Functional Brain Imaging, Johns Hopkins University
4
Current Affiliations: Departments of Radiology; Physiology; and Biomedical Engineering, University of Manitoba
Functional Connectivity Analysis (rs-fMRI)
-24
A)
-20
28
0
32
4
36
8
40
-16
12
-12
16
44
20
-8
24
-4
28
0
32
4
8
40
36
12
-16
44
16
BMI was calculated using height and weight,
whereas BFP was measured using a Tanita BC-350
bioelectric impedance scale.
As shown in both Figure 1 and Figure 2: Age
ranged from 18 to 57 years (with a slight bias
toward younger subjects); BMI ranged from 18 to
37; and BFP ranged from 7.5 to 44.
3
3
3
2
2
2
2
1
1
1
1
0
20
5
40
60
0
20
30
40
0
10
20
30
40
50
0
64
68
72
48
76
Right
52
56
60
64
68
72
76
Left
Right
Negative Correlations:
BMI
Overlap
48
White Matter Volume
BFP
5
5
5
4
4
4
3
3
3
3
2
2
2
2
1
1
1
1
0
0
0
0
Male
4
20
5
40
60
20
30
40
10
20
30
40
50
5
5
5
4
4
4
3
3
3
3
2
2
2
2
1
1
1
1
Overall
4
0
20
40
60
Age
(years)
0
20
30
BMI
(kg/m2)
40
0
10
20
30
40
BFP
(%)
50
0
80
White Matter Microstructural Analyses (DTI)
100
A)
60
80
-48
16
80
60
2
(16 male; 16 female)
Normalized BMI
55
N = 32
3
50
45
C)
2
1
0
−1
40
−2
35
−3
−3
3
r = 0.23
p = 0.21
Normalized BFP
B)
1
0
−1
−2
0
−3
−3
3
Normalized Age
3
25
2
Normalized BFP
20
15
10
E)
r = 0.70
p < 0.0001
1
0
−1
−2
BMI
(kg/m2)
BFP
(%)
Age
(years)
−3
−3
3
Normalized BFP
D)
0
Normalized BMI
3
Normalized Age
(corrected for age and gender)
30
0
3
2
1
0
−1
−2
−3
−3
0
Normalized BMI
(corrected for age and gender)
BMI was significantly correlated with
BFP (Figure 2D), but even after
regressing out age and gender, there
were still small differences between
BMI and BFP (Figure 2E).
After body composition measurements
were acquired for each participant, 3T
MRI scans were performed.
r = 0.89
p < 0.0001
3
-40
-36
-32
-28
-24
-20
-48
-44
-40
-36
-32
-28
-24
20
24
28
32
36
40
−1
−2
0
-12
20
-8
24
-4
28
0
32
4
36
8
40
-16
12
-12
16
44
20
-8
24
-4
28
0
32
4
36
Figure 2: A) Body Mass Index (BMI), Body Fat Percentage (BFP) and
Age among the 32 healthy study participants, as well as correlations
between: B) normalized BMI and Age; C) normalized BFP and Age;
and D) normalized BFP and BMI. E) Partial correlations were also
performed between normalized BFP and normalized BMI to factor
out the effects of age and gender.
52
56
60
64
68
72
56
60
64
68
Left
72
48
76
Right
Negative Correlations:
Fractional Anisotropy (FA)
BMI
Overlap
BFP
52
Left
56
60
8
40
12
2
Overlap
68
BFP
−2
3
0
−1
−2
0
Executive Control Network
(ECN)
1
0
−1
−2
−3
−3
3
Normalized BMI
3
0
−1
−2
Normalized BFP
0
3
Normalized BMI
1
−3
−3
3
0
(corrected for age and gender)
rp = 0.11
p = 0.54
2
rp = 0.33
p = 0.06
2
1
0
−1
−2
−3
−3
3
Normalized BFP
(corrected for age and gender)
Default Mode Network
(DMN)
0
2
(corrected for age and gender)
1
−3
−3
Right
0
3
Normalized BFP
(corrected for age and gender)
(corrected for age and gender)
Salience Network
(SN)
Figure 5: Individual differences in resting state functional connectivity are correlated with body composition (i.e., BMI and/or BFP), corrected for age and gender. A) Previously defined regions of interest were taken from
an anatomical atlas (Shirer et al., 2012) of the default mode network (DMN; red), the executive control network (ECN; green), and the salience network (SN; blue). Partial correlations (corrected for age and gender) were
performed between body composition data (BMI or BFP) and: B) DMN connectivity, C) ECN connectivity, and D) SN connectivity.
Processing of resting state BOLD fMRI data (222 volumes; 7.4 minutes) was performed using the SPM8 Conn toolbox, and
included: realignment; slice-time correction; coregistration; spatial normalization; 6mm 3D smoothing; motion compensation
(CompCor); artifact detection and removal (ART); and band-pass filtering (0.01–0.08 Hz).
Each subject’s network connectivities were calculated by averaging bivariate correlations between all ROIs pairs (taken from a
previously published atlas; Figure 5A), and these were plotted against BMI and BFP, correcting for age and gender (Figure
5B-D). Of the three networks investigated (i.e., the DMN, ECN, and SN), we observed:
1) No significant relationships between DMN or ECN connectivity and either BMI or BFP (Figure 5B-C), and
Total brain size, as well as global gray and white matter
volumes were positively correlated with BMI and BFP in our
sample (Figure 6). However, many structures exhibited
significant volumetric decreases with higher BMI and BFP.
44
72
Several white matter regions also showed decreased FA
and/or increased MD (both markers of decreased white
matter integrity) among higher BMI and BFP individuals.
76
Right
Positive Correlations:
BMI
64
−1
−3
−3
3
rp = -0.18
p = 0.32
Gray Matter
Volume
Mean Diffusivity (MD)
Figure 4: Individual differences in regional white matter microstructure are correlated with body composition (i.e., BMI and/or BFP), corrected for age and gender. A) Fractional anisotropy (FA) values were negatively
correlated with BMI in three clusters, negatively correlated with BFP in two clusters, and positively correlated with both BMI and BFP in one cluster. B) Mean diffusivity (MD) values were positively correlated with BMI in
three clusters and negatively correlated with BFP in two clusters. No white matter structures demonstrated significant negative correlations with either BMI or BFP. *Note: all brain images are displayed in neurological
orientation, and are in normalized MNI space. Regions of the Salience Network (SN) are shown with a light brown outline.
Processing of diffusion-weighed images (30 directions; b = 700 s/mm2) was performed using CATNAP, MRIStudio, SPM8, and
included: coregistration; gradient orientation; skull stripping; linear (AIR) and non-linear (LDDMM) normalization to MNI space;
calculation of fractional anisotropy (FA) and mean diffusivity (MD) images; white matter masking; and 6 mm 3D smoothing.
By running separate cluster-level GLM analyses for FA and MD images, we found that:
1) The corona radiata, inferior fronto-occipital fasciculus, inferior longitudinal fasciculus, as well as orbitofrontal and temporal white matter
demonstrated significantly reduced FA with higher BMI and/or BFP. Only the left cerebellum showed increased FA (Figure 3A).
2) The cingulum, corpus callosum, cuneus and precuneus, as well as middle orbitofrontal and parahippocampal white matter demonstrated
significant increases in MD with higher BMI and/or BFP. No regions exhibited significantly decreased MD (Figure 3B).
Despite relatively under-developed gray and white matter
structures and reduced white matter connectivity proximal
to several nodes of the Salience Network (SN), functional
connectivity within this network was significantly increased
as a function of BMI and BFP.
In summary, we have shown that:
White Matter
Volume
A)
Cerebrospinal Fluid
Volume
B)
3
C)
3
rp = 0.55
p = 0.001
2
1
0
−1
−2
−3
52
0
(corrected for age and gender)
76
Left
1
Normalized BMI
44
Although the association between BMI and BFP was not
perfect (r=0.70), both measures were in good agreement
regarding the neural correlates of obesity.
-20
MMP
(%)
BFP, but not BMI, was significantly
correlated with age (Figure 2B-C).
r = 0.42
p = 0.018
-44
B)
100
48
A)
0
2
rp = 0.36
p = 0.04
Discussion and Conclusions
100
-16
60
12
3
rp = 0.30
p = 0.09
2) Significantly increased SN connectivity among higher BMI and BFP individuals (Figure 5D).
3) Neither gray nor white matter analyses (corrected for age, gender and intracranial volume) revealed local volumetric increases.
60
8
−2
0
1
0
−1
−2
−3
−2
0
−1
−2
−2
2
Normalized BFP
(corrected for age and gender)
0
2
1
0
−1
−2
−3
−2
0
1
0
−1
−2
−2
2
Normalized BFP
(corrected for age and gender)
0
0
−1
−2
−2
0
0
−1
−2
−2
2
Normalized BFP
(corrected for age and gender)
0
2
(corrected for age and gender)
3
rp = −0.18
p = 0.31
1
−3
1
Normalized BMI
(corrected for age and gender)
2
rp = 0.45
p = 0.01
2
−3
2
Normalized BMI
3
rp = 0.22
p = 0.23
3
rp = −0.13
p = 0.48
2
−3
2
(corrected for age and gender)
3
rp = 0.47
p = 0.007
2
0
Normalized BMI
(corrected for age and gender)
3
1
Total Intracranial
Volume
D)
3
rp = 0.36
p = 0.04
2
−3
2
Normalized BMI
(corrected for age and gender)
Participants were recruited to span a broad range of
age and weight, although sex was evenly balanced.
3
4
Normalized GM Volume
32 healthy adults were recruited from the
community at large and screened for prior
neurological or psychiatric trauma or illnesses,
alcohol or drug abuse, etc.
4
0
Normalized WM Volume
60
Processing of whole-brain T1-weighed images (1mm isotropic) was performed using SPM8 and the VBM8 toolbox, and included:
unified segmentation; non-linear (DARTEL) normalization to MNI space; image modulation; and 5 mm 3D smoothing.
5
4
-4
3
1
(corrected for age and gender)
56
2) Several white matter regions (i.e., the cingulum, corpus callosum, inferior fronto-occipital fasciculus, inferior and superior longitudinal
fasciculi, and regions proximal to the basal ganglia) also demonstrated volume reductions with higher BMI and/or BFP (Figure 3B).
4
-8
2
Normalized GM Volume
52
Figure 3: Individual differences in regional cortical volumes are correlated with body composition (i.e., BMI and/or BFP), corrected for age, gender and total intracranial volume. A) Gray matter volumes were negatively
correlated with BMI in three clusters and negatively correlated with BFP in two clusters. No gray matter structures demonstrated significant positive correlations with either BMI or BFP. B) White matter volumes were
negatively correlated with BMI in six clusters and negatively correlated with BFP in seven clusters. No white matter structures demonstrated significant positive correlations with either BMI or BFP. *Note: all brain images
are displayed in neurological orientation, and are in normalized MNI space. Regions of the Salience Network (SN) are shown with a light brown outline.
Body Composition (BMI and BFP)
4
-12
D)
rp = -0.17
p = 0.35
−3
−3
1) Several gray matter regions (i.e., the amygdala, basal ganglia, lingual gyri, precuneus, as well as the precentral and postcental gyri)
demonstrated significant volume reductions with higher BMI and/or BFP (Figure 3A).
5
C)
(corrected for age and gender)
24
-4
By running separate cluster-level GLM analyses for gray and white matter images, we found that:
5
B)
-20
Normalized Total IC Volume
20
-8
Gray Matter Volume
2) Comparing these to magnetic resonance imaging measures of local gray matter volume, white
matter volume, white matter microstructure, and functional connectivity using voxel-based
morphometry (VBM), diffusion tensor imaging (DTI), and resting state functional MRI (rs-fMRI).
Female
-24
(corrected for age and gender)
48
1) Acquiring two independent measures of body composition - i.e., body mass index (BMI) and body
fat percentage (BFP); and
5
-28
(corrected for age and gender)
-12
Left
D)
-32
3
This was achieved by:
C)
-36
(corrected for age and gender)
16
Therefore, the primary aim of this study was to identify differences in brain structure and
function related to normal individual variability in body composition.
B)
-40
Normalized DMN Connectivity
-16
Previous studies have investigated the neural correlates of these obesity-related cognitive
effects, and have included: studies of gray and white matter volumes; white matter
microstructure; or metabolic/functional activity. However, to the best of our knowledge, none
have investigated all of these factors in a single cohort.
A)
-44
3
There are also well-established associations between higher adiposity, elevated impulsivity
and decreased cognitive performance.
Figure 1: A) Age, B) Body Mass Index (BMI), C) Body Fat Percentage (BFP), and D) Muscle
Mass Percentage (MMP) for our study population (female; male; overall). *BMI less than
18.5 (dotted green line) is considered underweight; BMI between 18.5 and 25 (dotted
black line) is normal; BMI between 25 and 30 (dotted red line) is overweight; and BMI
greater than 30 is obese.
-48
Normalized Total IC Volume
-28
(corrected for age and gender)
-32
Normalized SN Connectivity
-36
(corrected for age and gender)
-40
Normalized SN Connectivity
-44
Normalized CSF Volume
-48
(corrected for age and gender)
B)
-20
Normalized CSF Volume
-24
(corrected for age and gender)
-28
(corrected for age and gender)
-32
Normalized ECN Connectivity
-36
(corrected for age and gender)
-40
(corrected for age and gender)
-44
Normalized WM Volume
Obesity affects more than 850 million adults worldwide and is one of the leading causes of
death in first-world countries.
-48
(corrected for age and gender)
A)
Normalized ECN Connectivity
Gray and White Matter Volumetric Analyses (VBM)
Background and Aims
Normalized DMN Connectivity
Lab Link
rp = 0.32
p = 0.07
2
1
0
−1
−2
−3
−2
0
2
Normalized BFP
(corrected for age and gender)
Figure 6: Individual differences in global brain volume related to body composition (i.e., BMI and BFP),
corrected for age and gender, and broken down by tissue type.
1) two independent measures of body composition (i.e., BMI and BFP) are correlated with aberrant
changes in brain structure and functional connectivity, and
2) many of the affected regions/networks offer a biologically plausible explanation for previously
reported deficits in reward processing and cognitive performance.
Thanks for stopping by. If you have any
questions or would like an electronic
reprint of this poster, please contact me
at [email protected]
This work was supported by The National
Institutes of Health (R01 MH082957 to
SMC) and the Canadian Institutes of Health
Research (postdoctoral fellowship to CRF).