Fluctuating Asymmetry and Individual Variation in Regional Gray and White Matter Volumes: A Voxel-Based Morphometry Study

Evolutionary Psychology
www.epjournal.net – 2008. 6(4): 613-627
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Original Article
Fluctuating Asymmetry and Individual Variation in Regional Gray and White
Matter Volumes: A Voxel-Based Morphometry Study

Matthew Euler, Department of Psychology, University of New Mexico, Albuquerque, New Mexico, USA

Robert J. Thoma, MIND Research Network and Department of Psychiatry, University of New Mexico,
Albuquerque, New Mexico, USA

Lauren Parks, Department of Psychology, University of New Mexico, Albuquerque, New Mexico, USA

Steven W. Gangestad, Department of Psychology, University of New Mexico, Albuquerque, New Mexico,
USA

Ronald A. Yeo, Department of Psychology, University of New Mexico, Albuquerque, New Mexico, USA.
Email: ryeo@unm.edu (corresponding author)

Abstract: Composite measures of fluctuating asymmetry (FA) of skeletal features are
commonly used to estimate developmental instability (DI), the imprecise expression of
developmental design due to perturbations during an individual’s growth and maturation.
Though many studies have detailed important behavioral correlates of FA, very little is
known about its possible neuroanatomical correlates. In this study we obtained structural
brain MRI scans from 20 adults and utilized voxel-based morphometry (VBM) to identify
specific regions linked to FA. Greater FA predicted greater whole brain white matter
volume, and a trend in the same direction was noted for whole brain gray matter volume.
Greater FA was associated with significantly greater gray and white matter volumes in
discrete brain regions, most prominently in the frontal lobes and in the right cerebral
hemisphere. Developmental studies are needed to identify when FA-related brain
differences emerge and to elucidate the specific neurobiological mechanisms leading to
these differences.
Keywords: fluctuating asymmetry, developmental instability, voxel-based morphometry
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

---

Fluctuating asymmetry
Introduction
Developmental instability (DI) refers to an organism’s ability to develop the
appropriate species-specific phenotype despite genetic and environmental perturbations that
tend to disrupt development, such as mutations, interbreeding, toxins, parasites, injuries,
and starvation (Gangestad and Thornhill, 1999; Møller and Swaddle, 1997). DI is often
operationalized as fluctuating asymmetry (FA), a composite measure of an individual’s
deviations from symmetry in traits that are symmetrical at the population level, without
regard to side (Gangestad, Bennett, and Thornhill, 2001). Interest in DI stems in part from
the wide range of studies demonstrating that humans with greater FA may show reduced
fecundity, health, social dominance, and mating success (Gangestad and Simpson, 2000;
Johnstone, 1995; Leung and Forbes, 1997; Thornhill and Møller, 1997; Zebrowitz, Hall,
Murphy, and Rhodes, 2002). However, there is substantial variability in the strengths of
associations across studies, and reasons for inconsistencies are poorly understood.

To date, greater FA has been reported in a variety of neurodevelopmental disorders
(e.g., Edgar et al., 2006; Yeo, Gangestad, and Thoma, 2007) and has also been linked in
humans with traits such as intelligence, jealousy, and physical violence (Kowner, 2001).
More specifically, individuals with greater FA have been reported to show relatively lower
general intellectual functioning (Bates, 2007; Furlow, Armijo-Prewitt, Gangestad, and
Thornhill, 1997; Luxen and Buunk, 2006; Prokosch, Yeo, and Miller, 2005; Thoma, Yeo,
Gangestad, Halgren, Sanchez, and Lewine, 2005; though see also Johnson, Segal, and
Bouchard, 2008) and to express more jealousy in mating situations (Brown and Moore,
2003). Furthermore, greater FA predicts fewer self-reported lifetime sexual partners
(Gangestad, Bennett, and Thornhill, 2001) and fewer episodes of physical violence
(Furlow, Gangestad, and Armijo-Prewitt, 1998).

Given these behavioral correlates, one would expect FA to vary systematically with
aspects of brain anatomy and/or physiology. Identifying specific neural correlates of FA
would facilitate our understanding of the mechanisms linking the genetic and
environmental determinants of FA with individual differences in behavior. To date, studies
linking neural variation with FA in humans are sparse. In general, FA appears to be
associated with atypical functional lateralization of the human brain (Thoma, Yeo,
Gangestad, Lewine, and Davis, 2000; Yeo, Gangestad, Thoma, Shaw, and Repa, 1997). In
addition, we previously reported a positive relationship between skeletal FA and a
composite measure of asymmetries of gray matter, white matter, and planum temporale
volumes (Thoma, Yeo, Gangestad, Lewine, and Davis, 2002; Thoma et al., 2005).
However, we did not find any significant relationships between measures of brain volumes
(total brain, gray matter, or white matter) and FA in either of these studies. Consistent with
these results, Furlow et al. (1997) found no relationship between head circumference and
FA. Further, although FA, gray matter, and total brain volumes individually predict
intellectual ability, we have found that these volumes as assessed from MRI scans do not
mediate the FA-intelligence relationship (Thoma et al., 2005). However, Bates (2007)
recently reported that skull size, which is correlated with brain volume, did mediate the
relationship between intelligence and FA.

In the current study, we examined the neuroanatomic correlates of FA using voxel-
based morphometry (VBM). In contrast to traditional region of interest (ROI) magnetic
resonance imaging (MRI) analysis methods, which are spatially limited and labor-intensive,
Evolutionary Psychology – ISSN 1474-7049 – Volume 6(4). 2008. -614-

---

Fluctuating asymmetry
VBM allows for a fully automated and objective assessment of variations in regional
cortical volume, on a whole-brain basis. Studies comparing VBM to manual ROI methods
have attested to the validity of the former technique, and have even highlighted the
complementarity of the two approaches (Good et al., 2002; Giuliani, Calhoun, Pearlson,
Francis, and Buchanan, 2005). As well as quantifying between-group differences in brain
volume, VBM has also been used to highlight the relationship between local changes in
cortical tissue and measures of cognitive performance (Haier, Jung, Yeo, Head, and Alkire,
2004), disease severity and heterogeneity (Koutsouleris, Gaser, Jager, et al., 2008), and
demographics (Smith, Chebrolu, Weckstein, Schmitt, and Markesbery, 2007). Here, we
employed VBM to explore the relationship between variations in regional cortical gray and
white matter volumes and measured of, in a sample of neurologically intact adults. VBM is
especially well suited for exploring the neural correlates of FA. The wide range of
behavioral associations of FA does not suggest the central importance of any single brain
region. Rather, numerous brain regions could possibly be related to FA, rendering the ROI
approach inadequate. In this study, we used VBM to investigate whether specific regions of
cortical gray matter and underlying white matter are correlated with FA.
Materials and Methods
Participants
Twenty normal participants (13 male, 7 female) were recruited through ads in the
local media. Participants were adults ranging in age from 26 to 62 (see Table 1); one
female and two male participants were left-handed. None had a history of head injury,
neurological disorder, or unstable medical illness. All participants were screened with the
Structured Clinical Interview for DSM-IV Axis-I Disorders, Clinician Version (SCID-CV;
First, Spitzer, Gibbon, and Williams, 1996). Individuals with a history of alcohol or other
substance abuse in the three months preceding the study were excluded, as were individuals
who demonstrated any history of either marijuana or cocaine dependence.

FA Measurement
FA measures reflect deviation from perfect symmetry in bilateral features that,
across the relevant population, are typically symmetrical. Alternatively, directional
asymmetries (DA) occur in features for which a population of organisms shows a
consistent structural or functional bias for a particular side of the body. Examples of
directional asymmetry in humans include the placement of the heart and number of lobes in
the right versus the left lung. Skeletal FA was assessed in 20 participants using calipers
across the feet, ankles, elbows, wrists, and hands. The length and width of each ear and the
length of the first four fingers were also measured. Each of these measures was conducted
twice to increase reliability, and the mean of the two measurements was calculated and
used as the feature size for calculation of FA. Of the seven body parts measured, only hand
width was determined to be directionally asymmetric; it was excluded from the present
analyses. FA for individual traits was calculated by taking the absolute value of the
difference between left and right sides, divided by one-half times the sum of left plus right
sides {individual FA = |R-L| / [.5 x (R+L)]}. FA scores for the ten remaining measurements
were summed to derive a total FA score for each subject.
Image Acquisition
Evolutionary Psychology – ISSN 1474-7049 – Volume 6(4). 2008. -615-

---

Fluctuating asymmetry
For each subject, high-resolution 3D T1-weighted MR images were acquired with a
1.5 T Picker Edge Imager at the VA Functional Neuroimaging Center, using a Field Echo
3D Sagittal sequence (Picker) (TR = 15 ms, TE = 4.4 ms, FOV = 256 mm, flip angle = 25
degrees, matrix 192 x 256, slice thickness = 1.5 mm).

Voxel Based Morphometry

All image analysis was performed using and SPM2
(http://www.fil.ion.ucl.ac.uk/spm/) running in MATLAB version 6.5. The fundamental
principles behind VBM have previously been introduced by Ashburner and Friston (2000).
First, MR images were manually aligned to a common space in SPM based on the locations
of the anterior and posterior commissures. VBM procedures followed the “optimized
VBM” protocol as described by Good et al. (2001), including creation of study-specific
gray and white matter templates. This involved spatially normalizing each study image to
the generic T1 template within SPM using an affine transformation and non-linear basis
functions, and segmenting the resulting images into gray matter, white matter, and csf
based on voxel intensity values and a priori information about brain tissue distribution. The
obtained normalization parameters were then recursively applied to the original images in
native space to achieve optimal normalization and segmentation. Optimally segmented gray
and white matter images were then averaged to obtain study-specific templates which were
smoothed via convolution with a Gaussian kernel of 8mm full width half-maximum
(FWHM).

To achieve optimal segmentation of the individual study images, the above
recursive procedure was performed where each image was initially segmented and
normalized to their respective gray or white matter template. Again, the normalization
parameters obtained from this segmentation and warping were applied to the original study-
images in native space, and the resulting optimally normalized images underwent final
segmentation into gray matter, white matter, and CSF. Next, the “modulation” step was
performed to correct for erroneous volume changes introduced during spatial
normalization. The resulting gray and white matter probability maps were then smoothed
with a kernel of 12 mm FWHM, giving the analysis a resolution of approximately 12mm
spatial scale for regions significantly associated with FA based on the Matched Filter
Theorem (Ashburner and Friston, 2001).

Statistical analyses were conducted using the multiple regression model in SPM2.
Individual FA values were the variable of interest, whereby FA scores were tested on a
voxel-by-voxel basis for significant positive or negative association with regional gray and
white matter volumes. Additionally, age, sex, and raw gray/white matter volumes were
treated as covariates of no interest (i.e., nuisance variables) in the analyses. For all
analyses, contrasts were defined to test for clusters of voxels of brain tissue that were
positively or negatively associated with FA. Clusters of voxels were considered significant
at p < 0.001 (uncorrected for multiple comparisons), and additionally thresholded for
display at using a minimum of 100 contiguous voxels. For each significant cluster, spatial
coordinates were entered into the xjView image viewing program
(http://people.hnl.bcm.tmc.edu/cuixu/xjView/) program to determine its approximate
anatomical locus (Cui, Jeter, Yang, Montague, and Eagleman, 2007). Importantly, as the
current study utilized a study-specific template and is hence in unique study-specific space,
Evolutionary Psychology – ISSN 1474-7049 – Volume 6(4). 2008. -616-

---

Fluctuating asymmetry
the reported x, y, z coordinates and associated regions only approximate their
corresponding regions in Talairach space.
Results
Table 1 provides descriptive statistics on age, FA, and tissue compartment volumes.
Partial correlations, controlling for age and sex, were obtained for relationships between
segmented overall tissue volumes and FA. We found a significant correlation between total
FA and white matter volume (r = .53, p = .025) and a trend toward a significant
relationship between FA and total gray matter volume (r = .43, p = .079). A total of eight
VBM analyses were performed. In the first four, VBM analyses examined positive and
negative relationships of FA with gray and white matter volumes, and age and sex were
treated as covariates. The second four analyses added another covariate, either total gray
matter volume for the gray matter analyses or total white matter volume for the white
matter analyses. As our results below suggest, treating total tissue volume as a covariate
reduces statistical power for detecting significantly increased or decreased regions related
to FA, yet serves to highlight those regions which are most strongly linked to FA.

Table 1. Descriptive statistics on age, fluctuating asymmetry (FA), and tissue compartment
volumes (in cubic centimeters).



Mean
SD
Range
Age
42.75
11.54
26-62
FA
.31
.08
.18-.47
Gray Matter
590.28-
737.91
59.40
866.70
White Matter
399.88-
465.08
55.76
584.45

1. Positive gray matter relationships with FA (sex and age as covariates).
Significant clusters of increased grey matter volume in relation to skeletal FA
were found prominently in the right middle frontal gyrus and left medial frontal
gyrus. Additionally, significant gray matter clusters positively associated with
skeletal FA were found in the right angular gyrus, right medial frontal gyrus,
right precentral gyrus, left inferior frontal gyrus, left uncus, left anterior
cingulate gyrus, right and left superior frontal gyri, and right and left inferior
temporal gyri. The size, location, and statistical value for each significant region
is summarized in Table 2 and displayed in Figure 1.



Evolutionary Psychology – ISSN 1474-7049 – Volume 6(4). 2008. -617-

---

Fluctuating asymmetry
Table 2. VBM analysis of gray matter clusters demonstrating significant positive
covariation with total FA, with age, and sex as covariates.
Positive
X Y Z
Cluster
Size
t
(p < .0001)
Right Middle
40 36 48 501
5.82
Frontal gyrus
Left Medial
-10 29 41 2154
5.48
Frontal gyrus
Right Inferior
52 -21 -36
1152
4.94
Temporal
gyrus
Left Inferior
-69 -18 -28 792
4.92
Temporal
gyrus
Right Medial
17
31 31 497
4.89
Frontal gyrus

Right Angular
43 -59 20 2928
4.88
gyrus
Right Middle
44 31 29
1664
4.79
Frontal gyrus
Right
51 0 7
2811
4.70
Precentral
gyrus
Left Inferior
-51
26 21
1897
4.65
Frontal gyrus

Right Inferior
69
-17 -26 618
4.45
Temporal

gyrus
Left Anterior
-8
14 26 188
4.34
Cingulate

gyrus
Left Uncus
-17
10 -45 184
4.34

Right Superior
15
27 60 471
4.30
Frontal gyrus

Left Superior
-9
28 58 134
4.10
Frontal gyrus


2. Negative gray matter relationships with FA (sex and age as covariates). No
significant clusters were identified.

3. Positive white matter relationships with FA (sex and age as covariates). Significant
clusters of increased white matter volume in relation to skeletal FA were found in
regions underlying the left middle temporal gyrus, right medial frontal gyrus, right
middle frontal gyrus, left middle frontal gyrus, right middle temporal gyrus, right
Evolutionary Psychology – ISSN 1474-7049 – Volume 6(4). 2008. -618-

---

Fluctuating asymmetry
angular gyrus, and right insula. All of these regions contained greater than 1000
contiguous voxels. Other significant white matter regions were found underlying
the left middle temporal gyrus, left inferior temporal gyrus, left postcentral gyrus,
left insula, right midbrain, left inferior parietal lobule, right supramarginal gyrus,
and left and right angular gyri. Table 3 lists the size, location, and statistical value
for each significant region and Figure 2 shows the anatomic location of some of
these clusters.

Figure 1. Anatomic representation of gray matter clusters demonstrating significant (p <
0.001) positive covariation with total FA (with age, and sex as covariates) in VBM. The
crosshairs are located in the right hemisphere. Only those clusters located in the planes
defined by the crosshairs are shown. The color bar denotes value of t-
statistics.


4. Negative white matter relationships with FA (sex and as covariates). No significant
clusters were identified.

5. Positive gray matter relationships with FA (sex, age, and total gray matter volume
as covariates). Significant clusters of increased gray matter volume in relation to
Evolutionary Psychology – ISSN 1474-7049 – Volume 6(4). 2008. -619-

---

Fluctuating asymmetry
skeletal FA were found in the right middle frontal gyrus (Figure 1B), left medial
frontal gyrus, right angular gyrus, and right precentral gyrus. Table 4 lists the size,
location, and statistical value for each significant region.

Table 3. VBM analysis of white matter clusters demonstrating significant positive
covariation with total FA, with age, and sex as covariates.
Positive
X Y Z
Cluster
Size
t
(p < .001)
Left Middle
-60 -32 -16 1685

8.27
Temporal gyrus
Right Medial
14
25
45
1701
7.68
Frontal gyrus
Right Middle
40 25 18 4352
6.65
Frontal gyrus
Left Middle
-37 20 10 6081
5.76
Frontal gyrus
Right Sub-lobar
41 -12 0 1370
5.50
Insula
Right Middle
55 -31 -14 3318
5.37
Temporal gyrus
Left Middle
-49 3 -37 300
5.27
Temporal gyrus
Right Angular
50 -61 36 2489
4.81
gyrus
Left Inferior
-45 -65 -6 313 4.80
Temporal gyrus
Left Postcentral
-9 -44 64 354
4.57
gyrus
Left Sub-lobar
-43 -11 -3 145 4.44
Insula
Right Midbrain
4 -9 -10 149
4.36

Left Inferior
-60 -45 21 118 4.28
Parietal Lobule
Right Angular
46 -70 22 199
4.03
gyrus
Left Angular
-42 -60 32 202 3.94
gyrus

Right
51 -44 30 147
3.94
Supramarginal
gyrus


Evolutionary Psychology – ISSN 1474-7049 – Volume 6(4). 2008. -620-

---

Fluctuating asymmetry
Figure 2. Anatomic representation of white matter clusters demonstrating significant (p <
0.001) positive covariation with total FA (with age, and sex as covariates) in Voxel Based
Morphometry. The crosshairs are located in the right hemisphere. Only those clusters
located in these planes are revealed. The color bar denotes value of t-statistics.

6. Negative gray matter relationships with FA (sex, age, and total gray matter volume
as covariates). No significant clusters were identified.

7. Positive white matter relationships with FA (sex, age, and total white matter volume
as covariates).





Evolutionary Psychology – ISSN 1474-7049 – Volume 6(4). 2008. -621-

---

Fluctuating asymmetry
Table 4. Voxel Based Morphometry (VBM) analysis of gray matter clusters demonstrating
significant positive covariation with total FA, with age, sex, and total gray matter volume
as covariates.
Positive
X Y Z
Cluster
Size
t
(p < .0001)
Right Middle
40 36 48 169
5.16
Frontal gyrus
Left Medial
-10 29 41 299
4.53
Frontal gyrus
Right Angular
44 -59 21 169
4.06
gyrus
Right
51 0 7 121
3.96
Precentral
gyrus

Significant clusters of increased white matter in relation to skeletal FA were
detected in regions underlying the left middle temporal gyrus, right medial frontal gyrus,
left and right middle frontal gyri, and right insula. Notably, all five of these regions were
among the most significant regions detected in the prior white matter analysis that did not
control total white matter volume. The size, location, and statistical value for each
significant region are summarized in Table 5.

Table 5. VBM analysis of white matter clusters demonstrating significant positive
covariation with total FA, with age, sex, and total white matter volume as covariates.
Positive
X Y Z
Cluster
Size
t
(p < .001)
Left Middle
-60 -33 -16 482 6.90
Temporal gyrus
Right Middle
43 26 19 1864
6.27
Frontal gyrus
Right Medial
14 25 45 465
5.98
Frontal gyrus
Right Sub-lobar
43 -14 -1 394
4.61
Insula
Left Middle
-35 20 12 150
4.38
Frontal gyrus

8. Negative white matter relationships with FA (sex, age, and total white matter
volume as covariates). Decreased white matter concentration was associated
with greater skeletal FA in several regions underlying the right superior parietal
lobule, left parahippocampal gyrus, right inferior temporal gyrus, and right
cingulate gyrus (see Table 6).



Evolutionary Psychology – ISSN 1474-7049 – Volume 6(4). 2008. -622-

---