Salivary cortisol and prefrontal cortical thickness in middle-aged men: A twin study.

Abstract

Although glucocorticoid receptors are highly expressed in the prefrontal cortex, the hippocampus remains the predominant focus in the literature examining relationships between cortisol and brain. We examined phenotypic and genetic associations of cortisol levels with the thickness of prefrontal and anterior cingulate cortex regions, and with hippocampal volume in a sample of 388 middle-aged male twins who were 51-59 years old. Small but significant negative phenotypic associations were found between cortisol levels and the thickness of left dorsolateral (superior frontal gyrus, left rostral middle frontal gyrus) and ventrolateral (pars opercularis, pars triangularis, pars orbitalis) prefrontal regions, and right dorsolateral (superior frontal gyrus) and medial orbital frontal cortex. Most of the associations remained significant after adjusting for general cognitive ability, cardiovascular risk factors, and depression. Bivariate genetic analyses suggested that some of the associations were primarily accounted for by shared genetic influences; that is, some of the genes that tend to result in increased cortisol levels also tend to result in reduced prefrontal cortical thickness. Aging has been associated with reduced efficiency of hypothalamic-pituitary-adrenal function, frontal lobe shrinkage, and increases in health problems, but our present data do not allow us to determine the direction of effects. Moreover, the degree or the direction of the observed associations and the extent of their shared genetic underpinnings may well change as these individuals age. Longitudinal assessments are underway to elucidate the direction of the associations and the genetic underpinnings of longitudinal phenotypes for changes in cortisol and brain morphology.

Full text

Page 1

Salivary cortisol and prefrontal cortical thickness in middle-aged men: A twin study
William S. Kremena,b,c,⁎, Robert C. O'Briena, Matthew S. Panizzona, Elizabeth Prom-Wormleyd,
Lindon J. Eavesd, Seth A. Eisene, Lisa T. Eylera,c, Richard L. Haugera,c, Christine Fennema-Notestinea,f,
Bruce Fischlg, Michael D. Granth, Dirk H. Hellhammeri, Amy J. Jaka,c, Kristen C. Jacobsonj, Terry L. Jernigana,
Sonia J. Lupienk, Michael J. Lyonsh, Sally P. Mendozal, Michael C. Nealed, Larry J. Seidmanm,
Heidi W. Thermenosm, Ming T. Tsuanga,b,c, Anders M. Daleg,n, Carol E. Franza
aDepartment of Psychiatry, University of California, San Diego, La Jolla, CA, USA
bCenter for Behavioral Genomics, University of California, San Diego, La Jolla, CA, USA
cVA San Diego Healthcare System, La Jolla, CA, USA
dDepartments of Psychiatry and Human Genetics, Virginia Commonwealth University, Richmond, VA, USA
eDepartment of Veterans Affairs, Washington, DC and Departments of Medicine and Psychiatry, Washington University, St. Louis, MO, USA
fDepartment of Radiology, University of California, San Diego, La Jolla, CA, USA
gDepartment of Radiology, Harvard Medical School and Massachusetts General Hospital, Boston, MA, USA
hDepartment of Psychology, Boston University, Boston, MA, USA
iDepartment of Clinical and Physiological Psychology, University of Trier, Trier, Germany
jDepartment of Psychiatry and Behavioral Neuroscience, The University of Chicago, Chicago, IL, USA
kMental Health Research Centre Fernand Seguin, Hôpital Louis-H Lafontaine, Université de Montréal, Canada
lUniversity of California, Davis Primate Center, Davis, CA, USA
mDepartment of Psychiatry, Harvard Medical School, Boston, MA, USA
nDepartment of Neurosciences, University of California, San Diego, La Jolla, CA, USA
a b s t r a c ta r t i c l ei n f o
Article history:
Received 1 September 2009
Revised 12 January 2010
Accepted 10 February 2010
Available online 13 February 2010
Keywords:
Heritability
Magnetic resonance imaging (MRI)
Hippocampus
HPA axis structure
Genetic correlation
Although glucocorticoid receptors are highly expressed in the prefrontal cortex, the hippocampus remains
the predominant focus in the literature examining relationships between cortisol and brain. We examined
phenotypic and genetic associations of cortisol levels with the thickness of prefrontal and anterior cingulate
cortex regions, and with hippocampal volume in a sample of 388 middle-aged male twins who were 51–
59 years old. Small but significant negative phenotypic associations were found between cortisol levels and
the thickness of left dorsolateral (superior frontal gyrus, left rostral middle frontal gyrus) and ventrolateral
(pars opercularis, pars triangularis, pars orbitalis) prefrontal regions, and right dorsolateral (superior frontal
gyrus) and medial orbital frontal cortex. Most of the associations remained significant after adjusting for
general cognitive ability, cardiovascular risk factors, and depression. Bivariate genetic analyses suggested
that some of the associations were primarily accounted for by shared genetic influences; that is, some of the
genes that tend to result in increased cortisol levels also tend to result in reduced prefrontal cortical
thickness. Aging has been associated with reduced efficiency of hypothalamic–pituitary–adrenal function,
frontal lobe shrinkage, and increases in health problems, but our present data do not allow us to determine
the direction of effects. Moreover, the degree or the direction of the observed associations and the extent of
their shared genetic underpinnings may well change as these individuals age. Longitudinal assessments are
underway to elucidate the direction of the associations and the genetic underpinnings of longitudinal
phenotypes for changes in cortisol and brain morphology.
© 2010 Published by Elsevier Inc.
The human neuroendocrine stress response is initiated and
coordinated by the activation of brain corticotropin releasing factor
(CRF) neurotransmission and the release of hypothalamic paraven-
tricular CRF which, in turn, binds to anterior pituitary CRF1receptors
to trigger secretion of adrenocorticotropic hormone (ACTH); ACTH, in
turn, binds to its receptor in the adrenal cortex to stimulate secretion
of cortisol (Hauger et al., 2006). Although activation of the CRF-ACTH-
cortisol cascade is critical for survival in the context of internal or
external threats to homeostasis, rapid counter-regulation of this
process is equally important for reestablishing physiological homeo-
stasis upon threat termination. As illustrated in Fig. 1, this regulation
takes place, in part, via feedback from target brain regions. The most
well known of these is the hippocampus. Indeed, since the seminal
study of McEwen and colleagues (1968) showed that the highest
NeuroImage 53 (2010) 1093–1102
⁎ Corresponding author. Department of Psychiatry, University of California, San
Diego, 9500 Gilman Drive (MC 0738), La Jolla, CA 92093, USA. Fax: +1 858 822 5856.
E-mail address: wkremen@ucsd.edu (W.S. Kremen).
1053-8119/$ – see front matter © 2010 Published by Elsevier Inc.
doi:10.1016/j.neuroimage.2010.02.026
Contents lists available at ScienceDirect
NeuroImage
journal homepage: www.elsevier.com/locate/ynimg

Page 2

concentration of receptors for corticosteroids in the rodent brain was
in the hippocampus, cognitive and brain research on cortisol regu-
lation has focused overwhelmingly on hippocampal dysfunction and
hippocampal-dependent episodic memory. Later cellular and molec-
ular research identified two corticosteroid receptor subtypes: the
glucocorticoid receptor (GR) and the mineralocorticoid receptor (MR)
(Lupien and Lepage, 2001). Interestingly, the prefrontal cortex in
primates was found to be a major site of low affinity glucocorticoid
receptor expression, thereby indicating that prefrontal cortical
neurons are also an important target of cortisol (Lupien and Lepage,
2001; Mizoguchi et al., 2003; Sánchez et al., 2000; Sarrieau et al.,
1986). In his evolving work, McEwen (2007) too has noted the
importance of other brain regions, including prefrontal cortex, in the
adaptation to stress.
The high expression of GRs in the prefrontal cortex may also be an
important factor in brain aging. Although evidence for increased
cortisol dysregulation with age is mixed, the predominant view is that
aging is associated with reduced efficiency of the hypothalamic–
pituitary–adrenal (HPA) axis (Bergendahl et al., 2000; Deuschle et al.,
1997;Ferrarietal.,2001;Inglisetal.,1999;Lupienetal.,1994;Seeman
etal.,2001;VanCauteretal.,1996,2000).Multiplestudiesalsosuggest
that the frontal lobes undergo greater age-related shrinkage than
other brain tissue regions (Raz, 2000; Raz and Rodrigue, 2006), and
thereisevidencethatboththefrontalcortexandtheunderlyingwhite
matter of the frontal lobes are disproportionately affected by aging
(Jernigan et al., 2001). Aging is also associated with reductions in
prefrontal dopamine (Wenk et al., 1989). Moreover, one of the pre-
dominant models of normative aging in humans is that it consists
largely of frontal–subcortical changes in contrast to the predomi-
nant medial temporal changes associated with Alzheimer's disease
(Buckner, 2004; Hedden and Gabrieli, 2004; Pugh and Lipsitz, 2002).
Other evidence suggests that chronic cortisol exposure—perhaps
duetostressorallostaticload—could haveanimpactonthestructure of
prefrontal regions. Chronic corticosteroid exposure in animal studies
has been associated with both dopaminergic and serotonergic changes
(Piazzaetal.,1996;Stenforsetal.,2004),dendriticremodeling(Shansky
etal.,2009;Wellman,2001),andasymmetriesofbrainactivity(Kalinet
al., 1998). Because age-related brain changes in humans appear to
preferentially affect prefrontal–subcortical circuit components, these
effects of corticosteroid exposure might also increase with age.
Overall, human research provides a mixed picture regarding
associations between chronic cortisol exposure and brain structure.
Mixed results may be due, in part, to the fact that most of the studies
had small sample sizes. When significant results were observed, they
were often in comparisons of extreme groups such as normal controls
versus patients with Cushing's disease, Alzheimer's disease, or
depression, or low- versus hyper-secreters (O’Brien et al., 1996;
Sheline et al., 1996; Lupien, et al., 1998; Sapolsky et al., 2000). Small
correlations have been reported between cortisol levels and prefron-
tal or anterior cingulate regions, but often within the context of mixed
results in which even the direction of association was not always
consistent (Gold et al., 2005; MacLullich et al., 2005, 2006). Results
based on extreme groups or people with specific disorders may not
extrapolate to the general population. Several findings were also
based on challenges such as dexamethasone suppression tests.
Although these constitute valid and useful ways to assess HPA axis
dysregulation, acute response to challenge may not necessarily
parallel the effects of chronic hyperactivity of the HPA axis.
Individual differences in both human cortisol output (Bartels et al.,
2003; Kirschbaum et al., 1992; Levene et al., 1972; Meikle et al., 1988)
and brain structure (Schmitt et al., 2007; Kremen et al., 2010) are
influenced by genetic factors to varying degrees, but we are unaware
of any examination of the genetic relationship between cortisol and
brain structure. In the present study, we sought to examine the
relationship of cortisol levels and brain structure in a large sample
that is representative of U.S. men in the age range under study. In
addition, our twin data allowed us to address the key issue of the
extent to which any significant phenotypic correlations between
cortisol levels and brain structure are accounted for by underlying
genetic or environmental factors.
Based on knowledge of GR concentrations and the relatively small
amount of existing MRI research, our primary focus was on prefrontal
cortex. However, previous reports also led us to examine the anterior
cingulatecortex and hippocampus as well. We tested whethercortisol
levels would be associated with reductions in the size of these brain
regions. Because prefrontal cortex is large and heterogeneous, we
examined specific prefrontal cortical parcellation units rather than
simply measuring the frontal lobes as a whole.
All of the previous reports used volume measures of brain regions.
We also used volume measures for the hippocampus, but for regions
across the cortical ribbon we examined measures of average thickness
and total surface area. The volume of a cortical regionis essentially the
product of its thickness and surface area. These two measures reflect
relatively independent processes in brain development as articulated
in Rakic's (1988, 1995, 2007) radial unit hypothesis. Elsewhere we
have shown that cortical thickness and surface area are each
influenced by different, independent sets of genes (Panizzon et al.,
2009). When volume is measured, these two different processes are
combined and it is not possible to determine whether one or the other
or both is accounting for size differences. Furthermore, thickness and
surface area might counteract one anotherto yield no correlation with
an external variable, if both are associated independently but in
opposite directions. Based on the radial unit hypothesis, we would
expect that cortisol would be related primarily to cortical thickness
rather than surface area.
Finally, the cortisol awakening response (CAR)—the pronounced
cortisol increase within 30–60 min after awakening—is also of
interest, in part, because it is significantly associated with self-
reported stress (Pruessner et al., 1999). The CAR has been associated
with hippocampal volume in young adult men, although the direction
of association may differ in younger and older individuals (Pruessner
et al., 2007). To our knowledge, the association between the CAR and
the thickness of prefrontal cortical regions has not been examined.
Methods
Participants
A total of 1237 twins participated in wave 1 of the longitudinal
Vietnam Era Twin Study of Aging (VETSA) project, an overview of
Fig. 1. Schematic representation of the hypothalamic–pituitary–adrenal (HPA) axis.
Beginning with the perception of a stressor, the hypothalamus releases CRF which
ultimately leads to the secretion of cortisol by the adrenal gland and its delivery to
target systems in the brain and body. Minus signs indicate negative (inhibitory)
feedback; plus sign indicates positive (excitatory) feedback.
1094
W.S. Kremen et al. / NeuroImage 53 (2010) 1093–1102

Page 3

which can be found elsewhere (Kremen et al., 2006). Participants
were randomly selected from an earlier study of twins in the Vietnam
Era Twin Registry study (Tsuang et al., 2001). Registry members are
male-male twin pairs who both served in the United States military
between 1965 and 1975. The Registry is neither a VA nor a patient
sample, and the large majority was not exposed to combat. The
VETSA MRI Study and the VETSA Cortisol Study were begun after the
parent VETSA project was underway. We began the VETSA MRI study
in the third year of the primary VETSA study. At the time of this
report there were 474 individual VETSA participants with analyzable
MRI data; 241 were scanned in San Diego and 233 were scanned in
Boston. The VETSA Cortisol Study began sometime after the start of
the MRI study. The present analyses are based on 388 individuals (86
monozygotic [MZ] pairs, 78 dizygotic [DZ] pairs, and 60 unpaired
twins) who participated in both the VETSA MRI and Cortisol studies,
and had data on all of the variables that were included in the
statistical models. Zygosity was initially classified according to
questionnaire and blood group information. These classifications
are being updated on the basis of 25 microsatellite markers. To date,
56% of the MRI study participants have DNA-determined zygosity.
Consistent with the overall VETSA project, 95% of the questionnaire-
based classifications were in agreement with the DNA-based
classifications; when differences occurred, we used the DNA-based
classifications.
Participants live throughout the United States and were given the
option of traveling to San Diego or Boston for a day-long series of
assessments. The MRI session was typically the day after the in-lab
evaluation. Only 6% of VETSA participants who were eligible to
undergo MRI declined to participate; 59% were included. The
remaining participants were excluded for reasons such as possible
metal in the body (7%), claustrophobia (3%), testing being conducted
in the twins' hometown (5%), scanner problems (8%), co-twin being
excluded (9%), and other reasons (3%).
Basic demographic and health characteristics of the VETSA
sample are comparable to U.S. census data for similarly aged men
(Centers for Disease Control and Prevention, 2003; National Center
for Disease Statistics, 2003). Mean age of the MRI participants was
55.8 (SD=2.6) years (range: 51–59), mean years of education was
13.9 (SD=2.1), and 85.2% were right-handed. Most participants
were employed full-time (74.9%), 4.2% were employed part-time,
and 11.2% were retired. There were 88.3% non-Hispanic white, 5.3%
African-American, 3.4% Hispanic, and 3.0% “other” participants. Self-
reported overall health status was as follows: excellent (14.8%); very
good (36.5%); good (37.4%); fair (10.4%); and poor (0.9%). These
demographic characteristics did not differ from the entire VETSA
sample, nor were there significant differences between MZ and DZ
twins.
Salivary cortisol
Saliva collection
On the day of testing, saliva samples were obtained at awakening,
30 minutes post-awakening, 10:00 am, 3:00 pm, and 9:00 pm or
bedtime. The first two morning samples and the evening sample
were provided by participants at the hotel where they stayed for the
study. Twins brought the morning samples with them to the
laboratory and dropped off the evening sample the next morning at
the hotel front desk. If necessary, participants chewed original Trident
gum to stimulate saliva production and removed the gum prior to
providing the sample. Previous testing by one of the investigators
(SPM) showed that this particular gum does not alter cortisol
concentrations.
Cortisol assays
Samples were sent to the laboratory of Dr. Mendoza at the
University of California, Davis where all assays were performed. Prior
to conducting the assays, samples were centrifuged at 3000 rpm for
20 minutes to separate the aqueous component from mucins and
other suspended particles. Salivary concentrations of cortisol were
estimated in duplicate using commercial radioimmunoassay kits
(Siemens Medical Solutions Diagnostics, Los Angeles, CA). Assay pro-
cedures were modified to accommodate overall lower levels of
cortisol in human saliva relative to plasma as follows: 1) standards
were diluted to concentrations ranging from 2.76 to 345 nmol/L; 2)
sample volume was increased to 200 μl, and 3) incubation times were
extended to 3 hours. Serial dilution of samples indicates that the
modified assay displays a linearity of 0.98 and an assay sensitivity
(least detectable dose) of 1.385 nmol/L. Intra- and inter-assay coeffi-
cients of variation are 3.96% and 5.66%. Samples from each participant
were analyzed in the same assay; one to three individuals were
includedin thesameassaybatch.Whenthree people wereincludedin
a batch, it typically included two co-twins and one unrelated twin.
Comparison of related and unrelated twins revealed significant batch
effects. Therefore, cortisol levels are adjusted for batch in all analyses.
Cortisol assays were always performed without knowledge of the
zygosity of the participant.
Cortisol measures
We used the mean of the five samples as the primary cortisol
measure for three reasons. The primary reason was that the mean
output on the day of testing was the most heritable cortisol measure,
and we were interested in determining the genetic overlap between
cortisol and brain structure (see Statistical analysis). Second, the
mean output on the day of testing provides a good index of overall
output. Third, if there is an association between cortisol levels and
brain structure, it is likely to be a function of chronic rather than
acute effects. Although the mean will, in part, reflect overall
responsivity to the sustained testing, it is more likely to reflect
chronic cortisol exposure as compared with specific responsivity
measures such as the cortisol awakening response (CAR) or cortisol
slope. We also examined area under the curve (AUC) for the five
samples as another index of overall output. The AUC is often
considered the best overall output measure, but it was not as highly
heritable as the mean. Moreover, it was very highly correlated with
the mean (r=0.87, pb0.001) and yielded similar results. Finally, we
examined the AUC for the cortisol awakening response (CAR),
defined as the AUC for the awakening and 30 minutes post-
awakening time points.
Cortisol values above 50 nmol were converted to missing. This
value corresponds approximately to average maximum values for the
participants plus 3 standard deviations. The cortisol measures had
skewed distributions and were log transformed to normalize the
distributions. If data were missing for one of the five samples, we
imputed the missing values based, in part, on individual profiles. For
example, if time point 5 was missing, we calculated the average
decrease for all participants from time point 4 to 5. We then decreased
an individual's owntime point4 by thatamountin order toimpute his
time point 5 value. If two or more time points were missing, the mean
and AUC values were treated as missing because it was felt that there
was an insufficient percentage of data present.
MRI acquisition
Images were acquired on Siemens 1.5 Tesla scanners (241 at
University of California, San Diego; 233 at Massachusetts General
Hospital [MGH]). Sagittal T1-weighted MPRAGE sequences were
employed with a TI=1000 ms, TE=3.31 ms, TR=2730 ms, flip
angle=7degrees, slice thickness=1.33 mm, voxel size 1.3×
1.0×1.3 mm. Raw DICOM MRI scans (including two T1-weighted
volumes per case) were downloaded to the MGH site. These data
were reviewed for quality, registered, and averaged to improve
signal-to-noise.
1095
W.S. Kremen et al. / NeuroImage 53 (2010) 1093–1102

Page 4

MRI processing
Volume measures
Volumetric measures were created for hippocampus using volumet-
ric segmentation (Fischl et al., 2002, 2004a) and cortical surface
reconstruction (Dale et al., 1999; Dale and Sereno, 1993; Fischl et al.,
1999,2002,2004a,b)methodsbasedonthepubliclyavailableFreeSurfer
software package. The automated, fully 3D whole-brain segmentation
procedure uses a probabilistic atlas and applies a Bayesian classification
rule to assign a neuroanatomical label to each voxel (Fischl et al., 2002,
2004a). A widely used training atlas has been shown to be comparable
to that of expert manual labeling and is sensitive to subtle brain changes
in Alzheimer's disease and normal aging (Fischl et al., 2002, 2004a).
However, we created a new manually-derived training set from 20
unrelated, randomly selected VETSA participants. This VETSA-specific
atlas was created by the same laboratory at the MGH Center for
Morphometric analysis using the same neuroanatomic criteria and the
samereliabilitystandardsastheoriginalatlas.Ourrationalewasthatthe
VETSA-specificatlaswouldbemorerepresentativeoftheVETSAsample,
thereby yielding more accurate measurements. Comparison of the
atlases did show that the VETSA-specific measures were closest to the
“gold standard” manually-derived measures (Kremen et al., 2010).
Estimated total intracranial volume, derived according to the method
of Buckner et al. (2004), is also provided in FreeSurfer and was used to
control for differences in head size. Hippocampal volumes were,
therefore, adjusted for intracranial volume.
Cortical thickness and surface area measures
Using semi-automated cortical surface reconstruction methods
(Dale et al., 1999; Dale and Sereno, 1993; Fischl and Dale, 2000; Fischl
et al., 1999, 2004b) based on the publicly available FreeSurfer
software package, we measured thickness at each surface location,
or vertex. Intensity variations due to magnetic field inhomogeneities
are corrected, a normalized intensity image is created, and the skull
(non-brain) is removed from the normalized image. The preliminary
segmentation is partitioned using a connected components algorithm,
with connectivity not allowed across the established cutting planes
that separate the cerebral hemispheres and disconnect brainstem
and cerebellum. Any interior holes in the components representing
white matter are filled, resulting in a single filled volume for each
cortical hemisphere. The resulting surface is covered with a triangular
tessellation and smoothed to reduce metric distortions. After the
initial surface model has been constructed, a refinement procedure is
applied to obtain a representation of the gray/white boundary. This
surface is subsequently deformed outwards to obtain an explicit
representation of the pial surface.
The surface was then divided into distinct cortical regions of
interest (ROIs) (Fischl et al., 2004b). Each surface location, or vertex,
was assigned a neuroanatomical label based on (1) the probability of
each label at each location in a surface-based atlas space, based on a
manually parcellated training set; (2) local curvature information;
and (3) contextual information, encoding spatial neighborhood
relationships between labels (conditional probability distributions
derived from the manual training set). The parcellation scheme labels
cortical sulci and gyri (Desikan et al., 2006), which form the basis for
defining ROIs. In the present analyses, we examined the thickness and
surface area of the nine prefrontal and two anterior cingulate ROIs in
each hemisphere (Fig. 2). Cortical thickness was defined as the
average distance between the gray-white boundary and the pial
surface within each ROI. Surface area was defined as the sum of the
areas of each triangular tesselation falling within a given ROI in each
individual's native space.
Fig. 2. Cortical Parcellation. The cortical parcellation units are generated by Freesurfer and are based on the parcellation system of Desikan et al. (2006). All of the parcellation units
are shown. The 9 prefrontal and 2 anterior cingulate regions examined in the present study are labeled. G.=gyrus. ctx.=cortex. Numbers in yellow are the phenotypic correlations
between mean cortisol level and thickness of the region. Bracketed numbers in white repeat the correlations with the superior frontal gyrus, but are shown on the medial surface.
1096
W.S. Kremen et al. / NeuroImage 53 (2010) 1093–1102

Page 5

In addition to the hippocampal, prefrontal, and anterior cingulate
measures that were the focus of the present study, we also examined
66 other cortical and subcortical ROIs (see Kremen et al., 2010) in
order to gauge the specificity of cortisol-brain structure associations.
MRI quality control
Of the 493 scans available at the time of these analyses, quality
control measures excluded 0.6% (3 cases) due to scanner artifact and
3% (16 cases) due to inadequate image processing results (e.g., poor
contrast caused removal of non-brain to fail). Scans were visually
inspected—blind to participant characteristics—and manually edited
by trained technicians. In conjunction with the Morphometry
Biomedical Informatics Research Network (BIRN; http://www.
nbirn.net/research/morphometry/index.shtm), which is sponsored
by the National Institutes of Health and the National Center for
Research Resources, the reliability and validity of these image acqui-
sition and processing methods across sites and scanners has been
demonstrated (Dickersonet al., 2008;Fennema-Notestine et al., 2007;
Han et al., 2006; Jovicich et al., 2009, 2006). Studies have also
demonstrated a high correlation of automatic and manual measures
in vivo and ex vivo (Fischl and Dale, 2000; Walhovd et al., 2005).
Statistical analysis
Non-twin analyses
These analyses examined the phenotypic relationship between
cortisol levels and brain structure by means of generalized linear
mixed models using SAS PROC GLIMMIX (SAS Institute, 2000).
Clustering of twins pairs was included in the models as a random
effect; this clustering serves to adjust the degrees of freedom to
account for the fact that twins within pairs (i.e., within the same
family) are not independent observations. All models were adjusted
for batch effects of the cortisol assays; batchwas included as a random
effect. Site effects were included as fixed effects in order to adjust for
possible scanner differences at the two sites. Analyses of ROIs other
than the hippocampus and those in prefrontal and anterior cingulate
cortices were considered exploratory because we did not have a priori
expectations about their association with cortisol levels.
In a second set of analyses, the effects of several demographic and
health characteristics that might be associated with the size of brain
structures or with cortisol levels were included as covariates and
treated as fixed effects in addition to the adjustments for batch and
site effects. These included age, young adult general cognitive ability,
hypertension, cardiovascular illness, depression, diabetes, smoking,
and alcohol consumption.
Descriptive statistics for these variables are shown in Table 1. The
index of young adult cognitive ability was the score on the Armed
Forces Qualification Test (AFQT). The AFQT was given to VETSA
participants just prior to induction into the military at average age 20.
It is highly correlated (r=0.81–0.84) with Wechsler IQ and provides a
good g measure (Grafman et al., 1988; McGrevy et al., 1974). AFQT
scores are percentiles, and these were log transformed to reflect a
normal distribution. AFQT provides a measure of general cognitive
ability that is more precise than the proxy variable of education, and it
is also not confounded by later aging effects. Hypertension was a
dichotomous variable based on whether the participant had in-
laboratory systolic blood pressure over 140, diastolic blood pressure
over 90, or was taking antihypertensive medications. Cardiovascular
illness was a dichotomous variable thatwas rated “yes” if a participant
had a history of heart surgery, heart catheterization, stroke, heart
attack, heart failure, or peripheral vascular disease. Depression was
based on the Center for Epidemiological Studies Depression Scale
(CES-D). CES-D scores were log transformed in order to obtain a more
normal distribution. Smoking was defined based on whether or not a
participant was a current smoker. Alcohol consumption was based on
a 4-point scale for drinking during the past two weeks: 0=non-
drinker; 1=up to 1 drink per day on average; 2=more than 1 and up
to 2 drinks per day on average; 3=more than 2 drinks per day on
average. A similar scale has been used in large epidemiological studies
(Paul et al., 2008).
Twin analyses
The standard twin (“ACE”) model estimates the proportion of
phenotypic variance due to additive genetic effects (A), shared or
common environmental effects (C), and individual-specific environ-
mental effects(E) (Eaves et al., 1978;Neale andCardon, 1992). Shared
environmental influences are those that make twins similar;
individual-specific environmental influences are those that make
twins different. Because measurement error is assumed to be random,
it is uncorrelated within twin pairs; consequently, it forms part of the
individual-specific environmental variance.
In the basic univariate ACE model : (1) additive genetic factors
correlate 1.0 for MZ twins and 0.5 for DZ twins; (2) shared envi-
ronmentalfactorscorrelate1.0acrosstwinsregardlessof zygosity; (3)
individual-specific environmental factors are uncorrelated across
twins; and (4) the variance of the underlying latent genetic and
environmental factors is fixed at 1.0. If MZ twin pairs are more highly
correlated than DZ pairs, it suggests that genetic factors account for
some of the individual differences in a phenotype.
Table 1
Sample characteristics.
Measure MeanSD
Cortisol (nmol/L)
Mean
Area under the curve
Cortisol awakening response
Prefrontal cortical thickness (mm)
L superior frontal gyrus
R superior frontal gyrus
L rostral middle frontal gyrus
R rostral middle frontal gyrus
L caudal middle frontal gyrus
R caudal middle frontal gyrus
L pars orbitalis
R pars orbitalis
L pars triangularis
R pars triangularis
L pars opercularis
R pars opercularis
L frontal pole
R frontal pole
L lateral orbital frontal cortex
R lateral orbital frontal cortex
L medial orbital frontal cortex
R medial orbital frontal cortex
L rostral anterior cingulate cortex
R rostral anterior cingulate cortex
L caudal anterior cingulate cortex
R caudal anterior cingulate cortex
Hippocampal volume (mm3)
L hippocampus
R hippocampus
Covariates
General cognitive ability in early
adulthood (AFQT)
Depression (CES-D)
1.68
23.81
1.19
0.41
7.00
0.52
2.19
2.20
1.85
1.81
2.03
2.04
2.19
2.19
1.91
1.92
2.04
2.05
2.38
2.34
2.11
2.07
1.85
1.84
2.01
2.00
2.07
2.20
0.11
0.11
0.10
0.10
0.13
0.14
0.17
0.16
0.13
0.13
0.14
0.15
0.26
0.25
0.14
0.13
0.16
0.15
0.25
0.27
0.33
0.27
3991.75
4225.29
390.98
431.40
61.11 22.83
7.95
% Yes
54.6%
15.2%
7.6%
23.6%
% L0
40.9%
7.81
% No
45.4%
84.8%
92.4%
76.4%
% L2
6.3%
Hypertension
Cardiovascular illness
Diabetes
Current smoking
% L1
44.1%
% L3
8.7%
Alcohol use (level [L])a
L=Left. R=Right. AFQT=Armed Forces Qualification Test. CES-D=Center for
Epidemiological Studies Depression Scale.
a0=non-drinker; 1=N 0 and ≤1 drink per day; 2=N 1 and ≤2 drinks per day; 3=N
2 drinks per day.
1097
W.S. Kremen et al. / NeuroImage 53 (2010) 1093–1102

Page 6

We have previously reported the heritabilities of the individual
measures, but our interest in this article was in the relationships
between measures. The ACE model is easily extended to a bivariate
model that includes the genetic, shared and individual-specific
environmental covariances between two measures. Fig. 3 shows a
bivariate correlated factors model, but for ease of presentation, only
one twin is shown. Bivariate twin models were examined when there
were statistically significant phenotypic correlations between mean
cortisol level and brain structure measures. Genetic correlations (rg),
shared environmental correlations (rc), and individual-specific
environmental correlations (re) are derived from the bivariate
models. A genetic correlation is like a standard phenotypic correla-
tion except that it is based on genetic covariances only, i.e., the ge-
netic covariance between two variables divided by the square root of
the product of their genetic variances (Neale and Cardon, 1992). For
the most part, genetic correlations indicate the amount of genetic
overlap between measures. Similarly, shared environmental correla-
tions reflect only the overlap between shared environmental
influences on each measure, and individual-specific environmental
correlations reflect only the overlap between individual-specific
environmental correlations. All three sets of influences contribute to
observed (phenotypic) correlations, but it is not possible to
determine which of these factors accounts for the observed
correlations in non-twin studies. The twin models were tested
formally with Mx, a maximum-likelihood-based structural equation
modeling program (Neale et al., 2003).
Results
Phenotypic associations
Table 2 shows the results for mean cortisol level and Fig. 2
illustrates the regions. Higher mean cortisol level was significantly
associated with thinner cortex in seven prefrontal regions: left and
right superior frontal gyrus; left rostral middle frontal gyrus; left pars
opercularis; left pars triangularis; left pars orbitalis; and right medial
orbital frontal cortex. After adjusting for the covariates, the associa-
tions between mean cortisol level and cortical thickness remained
significant in five of these seven prefrontal regions: left and right
superior frontal gyrus; left pars opercularis, left pars triangularis; and
right medial orbital frontal cortex. Significance was reduced to trend
levels for left rostral middle frontal gyrus and left pars orbitalis. The
significant correlations ranged from −0.10 to −0.15 (ps=0.05 to
0.004) when adjusted for batch and site effects only, and from −0.11
to −0.13 (ps=0.04 to 0.01) after adjusting for all of the covariates.
The results were similar for cortisol AUC (Table 3). Thinner cortex
in eight prefrontal regions was associated with larger AUC values: left
and right superior frontal gyrus; left rostral middle frontal gyrus; left
pars opercularis; left pars triangularis; right lateral orbital frontal
cortex; and left and right medial orbital frontal cortex (rs=−0.10 to
−0.15; ps=−0.05 to 0.004). After adjusting for covariates, the
associations remained significant between four of these prefrontal
regains and cortisol AUC: left pars opercularis, right lateral orbital
frontal cortex; and left and right medial orbital frontal cortex (rs=
−0.10 to −0.13; ps=0.05 to 0.01). The area under the curve for the
CAR was not significantly associated with any of these brain regions.
There were no significant associations between cortisol level and the
surface area of the prefrontal and anterior cingulate ROIs, and there
were no significant associations between any of the cortisol measures
and hippocampal volumes.
Fig. 3. Bivariate Correlated Factors ModelA=Additive genetic influences. C=Shared or
common environmental influences. E=Nonshared or individual-specific environmen-
tal influences. rgor ra=Genetic correlation. rc=Shared environment correlation. re=
Unique environment correlation. Arrows from A1, C1, and E1to mean cortisol level
represent parameter estimates for the contribution of those components to that
variable. The same is true for arrows from A2, C2, and E2to prefrontal cortical thickness.
Squaring these parameter estimates provides the proportion of variance accounted for
by each component. For ease of presentation, only one twin is represented.
Table 2
Association of cortisol level and prefrontal cortical thickness: Mixed model results and
phenotypic correlations for mean cortisol level.
Region of interestBatch/site only (n=388)a
All covariates (n=383)a
dftrb
pdftrp
Left superior frontal
gyrus
Right superior frontal
gyrus
Left rostral middle
frontal gyrus
Left pars opercularis
Left pars triangularis
Left pars orbitalis
Right medial orbital
frontal cortex
114
−2.94
−0.15
0.004
102
−2.40
−0.12
0.02
114
−2.94
−0.15
0.004
102
−2.49
−0.13
0.01
114
−2.45
−0.12
0.02
102
−1.79
−0.09
0.08
114
114
114
114
−1.99
−2.61
−2.03
−2.44
−0.10
−0.13
−0.10
−0.12
0.05
0.01
0.04
0.02
102
102
102
102
−2.15
−2.38
−1.91
−2.16
−0.11
−0.12
−0.10
−0.11
0.03
0.02
0.06
0.03
Batch/Site Only=Analyses adjusted for batch and site effects only. All Covariates=
Analyses adjusted for batch and site effects plus demographic and health covariates.
Associations that were not significant at the 0.05 level in either analysis are not shown.
Associations that were significant at the 0.05 level are shown in bold; associations that
were significant at trend levels (pb0.10) are shown in italics.
aAll analyses were based on mixed models that account for non-independence of
observations within twin pairs.
Table 3
Association of cortisol and prefrontal cortical thickness: mixed model results and
phenotypic correlations for cortisol area under the curve.
Region of interest Batch/site only (n=387)a
All covariates (n=382)a
dftrb
p dftrp
Left superior frontal
gyrus
Right superior frontal
gyrus
Left rostral middle
frontal gyrus
Left pars opercularis
Left pars triangularis
Right lateral orbital
frontal cortex
Left medial orbital
frontal cortex
Right medial orbital
frontal cortex
113
−2.12
−0.11
0.04
101
−1.55
−0.08 0.13
113
−2.07
−0.10
0.04
101
−1.60
−0.080.11
113
−2.20
−0.11
0.03
101
−1.55
−0.08 0.12
113
113
113
−1.95
−2.16
−2.04
−0.10
−0.11
−0.10
0.05
0.03
0.04
101
101
101
−1.95
−1.90
−1.95
−0.10
−0.10
−0.10
0.05
0.06
0.05
113
−2.07
−0.10
0.04
101
−2.00
−0.10
0.05
113
−2.91
−0.15
0.004
101
−2.63
−0.13
0.01
Batch/Site Only=Analyses adjusted for batch and site effects only. All covariates=
analyses adjusted for batch and site effects plus demographic and health covariates.
Associations that were not significant at the 0.05 level in either analysis are not shown.
Associations that were significant at the 0.05 level are shown in bold; associations that
were significant at trend levels (pb0.10) are shown in italics.
aAll analyses were based on mixed models that account for non-independence of
observations within twin pairs.
1098
W.S. Kremen et al. / NeuroImage 53 (2010) 1093–1102

Page 7

Only six of the 66 other ROIs were significantly associated with
mean cortisol level at the 0.05 level with adjustment for all of the
covariates. These were left inferior parietal cortex, left supramarginal
gyrus, right cuneus, right amygdala, and the left and right inferior
lateral ventricles.
Bivariate twin analyses
We performed bivariate twin analyses for the ROIs that were
significantly correlated with mean cortisol levels at the phenotypic
level. These analyses were limited to mean cortisol levels because, as
noted in the introduction, mean cortisol had the highest heritability of
the cortisol measures, and the brain regions with significant
associations in the other analyses were largely consistent with those
for mean cortisol levels. In previous univariate analyses, the
heritability of mean cortisol level during the study was 0.43, but it
was only 0.18 for the AUC (Franz et al., 2010). Heritabilities for the
relevant prefrontal regions were: left superior frontal gyrus (0.75);
right superior frontal gyrus (0.68); left rostral middle frontal gyrus
(0.45); left pars opercularis (0.62); left pars triangularis (0.44); left
pars orbitalis (0.37) and right medial orbital frontal cortex (0.39)
(Kremen et al., 2010).
Table 4 shows the results of the model fitting for the mean cortisol
level and left superior frontal gyrus. In the full model, genetic and
environmental factors are allowed to covary between the two
measures. The reduced models were compared to the full model to
determine whether dropping particular parameters would result in a
significant reduction in the fit of the model. This procedure is used to
establish the most parsimonious explanation of the covariance. As
shown in Table 4, Model 6—in which both the shared environmental
(rc) and the unique environmental (re) correlations are dropped—was
the most parsimonious model. In other words, the covariance
between mean cortisol level and left superior frontal gyrus thickness
can be explained by the genetic correlation alone without a significant
reduction in model fit. In this model, rg=−0.61 (95% CI=−1.00 to
−0.17). Similarly, for the right superior frontal gyrus, rcand recould
be dropped from the model. Here again, the genetic correlation (rg)
alone provided an adequate explanation of the data (rg=−0.70; (95%
CI=−1.00 to −0.14).
In the full models for left pars opercularis, left pars triangularis,
and right medial orbital frontal cortex, the genetic correlations ranged
from −0.23 to −0.68. These genetic correlations were not significant
in the full model, but rcand recould still be dropped from the models
without a significant reduction in fit. In the full models for left rostral
middle frontal gyrus and left pars orbitalis, rgand rccould be dropped
without a significant loss of fit. In the full models, neither shared
environmental (rc) nor unique environmental (re) correlations
between an ROI and a given cortisol measure were statistically
significant.
Discussion
We found that higher salivary cortisol levels were associated with
thinner cortex in prefrontal regions in a large-scale study of middle-
aged men. Significant associations were observed for mean cortisol
level and AUC, but not CAR. Neither cortisol index was significantly
associated with cortical surface measures. The significant phenotypic
correlations were of small magnitude, but they can be considered
reliable given our large sample size. In our effort to have a
representative sample, we did not exclude participants on the basis
of health measures. However, adjusting for the effects of age, general
cognitive ability, depression, cardiovascular risk factors, diabetes,
smoking, and alcohol use had relatively little impact on the overall
results. There were no significant associations of hippocampal
volumes with any cortisol measures. As seen in Fig. 2, the significant
correlations between cortisol level and prefrontal ROIs were in
partially contiguous regions. This outcome suggests a meaningful
pattern rather than a handful of chance results. Our data showing that
higher salivary cortisol levels were significantly correlated with
prefrontal cortical thinning, and that this association may be
genetically regulated may provide insight into the susceptibility of
cognitive function to the dysregulation of cortisol secretion during
aging.
Regarding the exploratory analysis of 66 other ROIs, it is tempting
to speculate about the inferior parietal and supramarginal regions
because elsewhere we have shown that there are substantial genetic
correlations between cortical thickness in these regions and several of
the prefrontal regions listed in Table 2 of the present study (Rimol
et al., 2010). Although one might think about a similar relationship
between the inferior lateral ventricles and the hippocampus, we
previously found that ventricular and hippocampal size were
primarily determined by independent sets of genes (Eyler et al., in
press). On the other hand, we did not make predictions about these
other ROIs, and 3.3 of these six correlations would be expected to be
significant by chance. Consequently, the most reasonable conclusion
is probably that the inverse cortisol-brain associations are relatively
specific to prefrontal cortex. As noted in the Introduction, previous
studies have frequently focused on selected participants, and many of
the associations between cortisol levels and hippocampal volume
were found in relatively extreme groups. In contrast, the VETSA
sample is representative of men in this age range. It may be that
associations are of greater magnitude at the extremes (e.g., disease
states such as Alzheimer's or Cushing's; high vs. low dexamethasone
suppression response; hypertensive vs. normotensive), but are of
lesser magnitude in our non-patient sample.
To our knowledge, this was also the first study to examine the
underlying genetic and environmental influences on the existing
relationships between cortisol levels and brain structure. There were
significant genetic correlations in two of the seven bivariate twin
models. Those genetic correlations of −0.61 for left and −0.70 for
right superior frontal gyrus, respectively, indicate that most of the
correlations were accounted for by genetic factors that are shared
between cortisol level and superior frontal gyrus thickness. Some of
the other genetic correlations were moderate but they were not
significant. None of the shared or unique environmental correlations
was significant. For three prefrontal regions—pars opercularis, left
pars triangularis, and right medial orbital frontal cortex—the
environmental correlations could be dropped from the models even
though the genetic correlations in the full model were not significant.
This result suggests that common genes were still important
determinants of the cortisol-prefrontal correlations, but even with
our relatively large sample size, the study was underpowered to
demonstrate that conclusively. Like the phenotypic correlations, the
Table 4
Model fitting results for bivariate genetic analysis of mean cortisol level and superior
frontal gyrus thickness.
−2LL
df
LRT
Δdfp
AIC
1
2
3
4
5
6
7
8
Full model
Drop rg
Drop rc
Drop re
Drop rgand rc
Drop rcand re
Drop rgand re
Drop all covariance
parameters
2285.938
2301.032
2286.869
2287.988
2301.044
2288.491
2301.337
2326.287
851
852
852
852
853
853
853
854
–
15.095
0.931
2.051
15.107
2.554
15.399
40.349
–
1
1
1
2
2
2
3
–
b0.001
0.335
0.152
0.001
0.279
b0.001
b0.001
–
13.095
−1.069
0.051
11.107
−1.446
11.399
34.349
−2LL=−2 log-likelihood; LRT=likelihood ratio chi-square test; Δdf=change in df;
AIC=Akaike information criterion; rg=genetic correlation; rc=shared (common)
environmental correlation; re=individual-specific environmental correlation.
Models 2–8 are tested against the full (Cholesky) model. The most parsimonious (best-
fitting) model is shown in boldface type; it indicates that the phenotypic correlation
between the two measures can be accounted for almost entirely by genetic factors.
1099
W.S. Kremen et al. / NeuroImage 53 (2010) 1093–1102

Page 8

genetic correlations were negative, indicating that genes that result in
increased cortisol levels also tend to result in thinner prefrontal
cortex.
Several previous studies have examined prefrontal cortex as a
single region, but significant phenotypic associations in the present
study were found in selected prefrontal sub-regions. In the left
hemisphere these regions showed partial contiguity from superior to
inferior frontal gyri that included dorsolateral (Brodmann areas [BA] 9
and 46) and ventrolateral (BA 47, 45, and 44; inferior frontal gyrus)
regions. In the right hemisphere the regions included dorsolateral (BA
9 and 46) and orbital-medial (BA 10, 11 and 12) prefrontal cortex.
Decreased glucose metabolism in BA 9 and 10 has been associated
with cortisol increases during a 25-minute stress test (Kern et al.,
2008). This regional overlap with some of our structural findings
supports the link between these regions and cortisol output.
The original impetus for examining relationships between cortisol
levels and prefrontal brain regions was the fact that glucocorticoid
receptors are highly expressed in prefrontal cortex but the links have
less well studied compared with the hippocampus. We considered
average cortisol level and AUC for the day of testing as indices
reflecting long-term corticosteroid exposure. It seems mostly likely
that the associationof cortisol with brain structure would reflect long-
term exposure. At least two factors can account for high levels of
cortisol: (1) responsivity that occurs in response to stressors or
challenges in which the system is activated and rapidly returns to
basal levels; and (2) dysregulation of the system that leads to chronic
elevationsevenin theabsenceof challenge.It is importantto notethat
responsivity does not imply a pathological process as it occurs
naturally daily in response to awakening (CAR). Persistently high
levels, however, may be pathological and may preclude the unfolding
of processes that are inhibited by cortisol and may even cause
retraction of dendritic arborization and loss of neurons (e.g., Shansky
et al., 2009). Change in the CAR in response to an acute challenge may
reflect responsivity, but without a clear before-and-after comparison,
the CAR may also reflect long-term response to stress in the present
study. It is also difficult to explain how the response to stressors on a
given day could measurably affect brain structure. Given that the
other factors we examined did not appear to mediate the association
between cortisol levels and brain structure, it seems reasonable to
conclude that the thinning of some prefrontal cortical regions may be
related to HPA dysregulation associated with chronic secretion of
glucocorticoids.
Our results cannot shed light on the direction of this relationship.
Chronically elevated glucocorticoids have been related to brain
volume reductions and cognitive impairment in humans (Lupien et
al., 1999), and chronic stress and exogenous glucocorticoids result in
brain atrophy in rats (Cerqueira et al., 2005; Cook and Wellman,
2004). On the other hand, factors causing prefrontal cortical thinning
might result in disruptions of HPA axis functioning considering that
studies indicate the prefrontal cortex exerts an inhibitory action on
stress-sensitive neurons in the extended amygdala. Prefrontal lesions
in humans can result in elevated morning cortisol levels (Tchiteya
et al., 2003), and animal studies have shown that medial prefrontal
lesions result in HPA axis dysregulation (Gerrits et al., 2003). Genetic
factors have received little or no attention when researchers have
described these prefrontal-cortisol links. As the VETSA participants
age, these different factors may also combine to have a significant
mediating or moderating effect on the association between cortisol
and brain structure. Additional evidence suggests that the direction of
associationsbetweencognitionandhippocampal volumeare opposite
in younger and older individuals (Pruessner et al., 2007). Thus, it is
possible that even the direction of the relationship between cortisol
and prefrontal cortical structure might change withage. Given the fact
that aging has been associated with both HPA axis dysfunction and
frontal lobe shrinkage, it is entirely possible that there are bidirec-
tional effectssuchthatboth processes increasinglyaffect eachotheras
a person ages. A goal of the VETSA follow-up assessments, which are
now underway, is to shed light on the genetic influences underlying
changes in both brain structure and HPA axis functioning. One way to
assess directionality would be to see if genes that influence HPA axis
functioning (e.g., CRF gene) are associated with prefrontal cortical
thinning over time.
Strengths of the study include the large sample size, use of cortisol
measurements that were based on multiple (5) time points, and the
ability to examine the genetic underpinnings of the associations
between cortisol and brain structure. Some limitations should be
noted as well. Like each of the studies cited that examined cortisol and
prefrontal or cingulate cortex structure, our sample included only
men. Therefore, these relationships need to be examined in women.
We do not know whether these associations are the same at different
ages, and it is not possible to determine the direction of effect in these
cross-sectional analyses. It is a goal of the VETSA to follow participants
as they age, which would enable us to replicate the findings and
address the issue of changes in mid-to-later life.
In sum, consistent with theory and previous evidence, cortical
thickness in prefrontal regions was associated with cortisol level. We
observed small, but significant negative correlations of cortisol level
with thickness in left dorsolateral and ventrolateral prefrontal cortex,
and right dorsolateral and orbital-medial frontal cortex. There were
no significant associations with anterior cingulate cortex thickness or
hippocampal volume, but given the small size of the significant
associations with prefrontal regions, it may be best to think of these
associations as being different in degree rather than present versus
absent. For the most part, the associations were not due to general
cognitive ability, cardiovascular risk factors, or depression. The results
suggest that at least some of the associations were primarily
accounted for by shared genetic influences, but the study was
underpowered to draw a definitive conclusion in this regard. Stress
in animal models has been shown to decrease dendritic arborization
and spine density of prefrontal cortex neurons (Shansky et al., 2009).
Furthermore, attentional control and prefrontal cortical processing
were disrupted in young individuals exposed to experimental
conditions with increased psychosocial stress (Liston et al., 2009). It
is likely that alteration in prefrontal cortex functioning in both
paradigms resulted from increased cortisol secretion. We hypothesize
that stress–cortisol–prefrontal cortex interactions may be even more
pronounced in humans as they age. Moreover, genetic factors may
increase an individual's sensitivity to the detrimental impact of
glucorticoids on prefrontal cortex during aging. Alternatively, genetic
factors may confer “resilience” to cortisol–prefrontal cortex interac-
tions. Longitudinal analysis could shed light on the direction or on
changes in the degree of the associations with age. The demographic
and health factors that we assessed may also begin to have greater
impacts on the association of cortisol levels and brain structure as
these study participants age.
Acknowledgments
Funded by National Institute on Aging (AG022982, AG022381,
AG018384, AG018386); National Center for Research Resources (P41-
RR14075; NCRR BIRN Morphometric Project BIRN002); National
Institute for Biomedical Imaging and Bioengineering
(R01EB006758); National Institute for Neurological Disorders and
Stroke (R01 NS052585-01); Mental Illness and Neuroscience Discov-
ery (MIND) Institute, part of the National Alliance for Medical Image
Computing (NAMIC), funded by the National Institutes of Health
through the NIH Roadmap for MedicalResearch, Grant U54 EB005149.
Additional support was provided by The Autism & Dyslexia Project
funded by the Ellison Medical Foundation. The U.S. Department of
Veterans Affairs has provided financial support for the development
and maintenance of the Vietnam Era Twin (VET) Registry. Numerous
organizations have provided invaluable assistance in the conduct of
1100
W.S. Kremen et al. / NeuroImage 53 (2010) 1093–1102

Page 9

this study, including: Department of Defense; National Personnel
Records Center, National Archives and Records Administration;
Internal Revenue Service; National Opinion Research Center; National
Research Council, National Academy of Sciences; the Institute for
Survey Research, Temple University. Most importantly, the authors
gratefully acknowledge the continued cooperation and participation
of the members of the VET Registry and their families. Without their
contribution this research would not have been possible.
References
Bartels, M., Van den Berg, M., Sluyter, F., Boomsma, D.I., de Geus, E.J., 2003.
Heritability of cortisol levels: review and simultaneous analysis of twin studies.
Psychoneuroendocrinology 28, 121–137.
Bergendahl, M., Iranmanesh, A., Mulligan, T., Veldhuis, J.D., 2000. Impact of age on
cortisol secretory dynamics basally and as driven by nutrient-withdrawal stress.
J. Clin. Endocrinol. Metab. 85, 2203–2214.
Buckner, R.L., 2004. Memory and executive function in aging and AD: multiple factors
that cause decline and reserve factors that compensate. Neuron 44, 195–208.
Buckner, R.L., Head, D., Parker, J., Fotenos, A.F., Marcus, D., Morris,J.C., Snyder, A.Z., 2004.
A unified approach for morphometric and functional data analysis in young, old,
and demented adults using automated atlas-based head size normalization:
reliability and validation against manual measurement of total intracranial volume.
Neuroimage 23, 724–738.
Centers for Disease Control and Prevention, 2003. Public health and aging: trends in
aging—United States and worldwide. MMWR CDC Surveill. Summ. 52, 101–106.
Cerqueira, J.J., Catania, C., Sotiropoulos, I., Schubert, M., Kalisch, R., Almeida, O.F., Auer,
D.P., Sousa, N., 2005. Corticosteroid status influences the volume of the rat cingulate
cortex: a magnetic resonance imaging study. J. Psychiatr. Res. 39, 451–460.
Cook, S.C., Wellman, C.L., 2004. Chronic stress alters dendritic morphology in rat medial
prefrontal cortex. J. Neurobiol. 60, 236–248.
Dale, A.M., Sereno, M.I., 1993. Improved localization of cortical activity by combining
EEG and MEG with MRI cortical surface reconstruction: a linear approach. J. Cogn.
Neurosci. 5, 162–176.
Dale, A.M., Fischl, B., Sereno, M.I., 1999. Cortical surface-based analysis. I: segmentation
and surface reconstruction. Neuroimage 9, 179–194.
Desikan, R.S., Segonne, F., Fischl, B., Quinn, B.T., Dickerson, B.C., Blacker, D., Buckner, R.L.,
Dale, A.M., Maguire, R.P., Hyman, B.T., Albert, M.S., Killiany, R.J., 2006. An automated
labeling system for subdividing the human cerebral cortex on MRI scans into gyral
based regions of interest. Neuroimage 31, 968–980.
Deuschle, M., Gotthardt, U., Schweiger, U., Weber, B., Korner, A., Schmider, J.,
Standhardt, H., Lammers, C.H., Heuser, I., 1997. With aging in humans the activity
of the hypothalamus–pituitary–adrenal system increases and its diurnal amplitude
flattens. Life Sci. 61, 2239–2246.
Dickerson, B.C., Fenstermacher, E., Salat, D.H., Wolk, D.A., Maguire, R.P., Desikan, R.,
Pacheco, J., Quinn, B.T., Van der Kouwe, A., Greve, D.N., Blacker, D., Albert, M.S.,
Killiany, R.J., Fischl, B., 2008. Detection of cortical thickness correlates of cognitive
performance: reliability across MRI scan sessions, scanners, and field strengths.
Neuroimage 39, 10–18.
Eaves, L.J., Last, K.A., Young, P.A., Martin, N.G., 1978. Model-fitting approaches to the
analysis of human behavior. Heredity 41, 249–320.
Eyler,L.T.,Prom-Wormley,E.,Fennema-Notestine,C.,Panizzon,M.S.,Neale,M.C.,Jernigan,
T.L., Fischl, B., Franz, C.E., Lyons, M.J., Stevens, A., Pacheco, J., Perry, M.E., Schmitt, J.E.,
Spitzer,N.C.,Seidman,L.J.,Thermenos,H.W.,Tsuang,M.T.,Dale,A.M.,Kremen,W.S.,in
press. Genetic patterns of correlation among subcortical volumes in humans: Results
from a magnetic resonance imaging twin study. Hum. Brain Mapp.
Fennema-Notestine, C., Gamst, A.C., Quinn, B.T., Pacheco, J., Jernigan, T.L., Thal, L.,
Buckner, R., Killiany, R., Blacker, D., Dale, A.M., Fischl, B., Dickerson, B., Gollub, R.L.,
2007. Feasibility of multi-site clinical structural neuroimaging studies of aging
using legacy data. Neuroinformatics 5, 235–245.
Ferrari, E., Cravello, L., Muzzoni, B., Casarotti, D., Paltro, M., Solerte, S.B., Fioravanti, M.,
Cuzzoni, G., Ponitggia, B., Magri, F., 2001. Age-related changes of the hypothalamic–
pituitary–adrenal axis: pathophysiological correlates. Eur. J. Endocrinol. 144,
319–329.
Fischl, B., Dale, A.M., 2000. Measuring the thickness of the human cerebral cortex from
magnetic resonance images. Proc. Natl. Acad. Sci. 97, 11050–11055.
Fischl, B., Sereno, M.I., Dale, A.M., 1999. Cortical surface-based analysis. II: Inflation,
flattening, and a surface-based coordinate system. Neuroimage 9, 195–207.
Fischl, B., Salat, D.H., Busa, E., Albert, M., Dieterich, M., Haselgrove, C., van der Kouwe, A.,
Killiany, R., Kennedy, D., Klaveness, S., Montillo, A., Makris, N., Rosen, B., Dale, A.M.,
2002. Whole brain segmentation: automated labeling of neuroanatomical
structures in the human brain. Neuron 33, 341–355.
Fischl, B., Salat, D.H., van der Kouwe, A.J., Makris, N., Segonne, F., Quinn, B.T., Dale, A.M.,
2004a. Sequence-independent segmentation of magnetic resonance images.
Neuroimage 23 (Suppl. 1), S69–S84.
Fischl, B., van der Kouwe, A., Destrieux, C., Halgren, E., Segonne, F., Salat, D.H., Busa, E.,
Seidman, L.J., Goldstein,J., Kennedy, D., Caviness, V., Makris, N., Rosen, B., Dale, A.M.,
2004b. Automatically parcellating the human cerebral cortex. Cereb. Cortex 14,
11–22.
Franz, C.E., York, T.P.., Eaves, L.J., Mendoza, S.P., Hauger, R.L., Hellhammer, D.H.,
Jacobson, K.C., Levine, S., Lupien, S.J., Lyons, M.J., Prom-Wormley, E., Xian, H.,
Kremen, W.S., 2010. Genetic and environmental influences on cortisol regulation
across days and contexts in middle-aged men. Behav. Genet. 40, 467–479.
Gerrits, M., Westenbroek, C., Fokkema, D.S., Jongsma, M.E., Den Boer, J.A., Ter Horst, G.J.,
2003. Increased stress vulnerability after a prefrontal cortex lesion in female rats.
Brain Res. Bull. 61, 627–635.
Gold, S.M., Dziobek, I., Rogers, K., Bayoumy, A., McHugh, P.F., Convit, A., 2005.
Hypertension and hypothalamo–pituitary–adrenal axis hyperactivity affect frontal
lobe integrity. J. Clin. Endocrinol. Metab. 90, 3262–3267.
Grafman, J., Jonas, B.S., Martin, A., Salazar, A.M., Weingartner, H., Ludlow, C., Smutok,
M.A., Vance, S.C., 1988. Intellectual function following penetrating head injury in
Vietnam veterans. Brain 111, 169–184.
Han, X., Jovicich, J., Salat, D., van der Kouwe, A., Quinn, B., Czanner, S., Busa, E., Pacheco,
J., Albert, M., Killiany, R., Maguire, P., Rosas, D., Makris, N., Dale, A., Dickerson, B.,
Fischl, B., 2006. Reliability of MRI-derived measurements of human cerebral
cortical thickness: the effects of field strength, scanner upgrade and manufacturer.
Neuroimage 32, 180–194.
Hauger, R.L., Risbrough, V., Brauns, O., Dautzenberg, F.M., 2006. Corticotropin releasing
factor (CRF) receptor signaling in the central nervous system: new molecular
targets. CNS Neurol. Disord. Drug Targets 5, 453–749.
Hedden, T., Gabrieli, J.D., 2004. Insights into the ageing mind: a view from cognitive
neuroscience. Nat. Rev. Neurosci. 5, 87–96.
Inglis, G.C., Ingram, M.C., Holloway, C.D., Swan, L., Birnie, D., Hillis, W.S., Davies, E.,
Fraser, R., Connell, J.M.C., 1999. Familial pattern of corticosteroids and their
metabolism in adult human subjects—The Scottish Adult Twin Study. J. Clin.
Endocrinol. Metab. 84, 4132–4137.
Jernigan, T.L., Archibald, S.L., Fennema-Notestine, C., Gamst, A.C., Stout, J.C., Bonner, J.,
Hesselink, J.R., 2001. Effects of age on tissues and regions of the cerebrum and
cerebellum. Neurobiol. Aging 22, 581–594.
Jovicich, J., Czanner, S., Greve, D., Haley, E., van der Kouwe, A., Gollub, R., Kennedy, D.,
Schmitt, F., Brown, G., Macfall, J., Fischl, B., Dale, A.M., 2006. Reliability in multi-site
structural MRI studies: effects of gradient non-linearity correction on phantom and
human data. Neuroimage 30, 436–443.
Jovicich, J., Czanner, S., Han, X., Salat, D., van der Kouwe, A., Quinn, B., Pacheco, J., Albert,
M., Killiany, R., Blacker, D., Maguire, P., Rosas, D., Makris, N., Gollub, R., Dale, A.,
Dickerson, B.C., Fischl, B., 2009. MRI-derived measurements of human subcortical,
ventricular and intracranial brain volumes: reliability effects of scan sessions,
acquisition sequences, data analyses, scanner upgrade, scanner vendors and field
strengths. Neuroimage 46, 177–192.
Kalin, N.H., Larson, C., Shelton, S.E., Davidson, R.J., 1998. Asymetric frontal brain activity,
cortisol, and behavior associated with fearful temperament in rhesus monkeys.
Behav. Neurosci. 112, 286–292.
Kern, S., Oakes, T.R., Stone, C.K., McAuliff, E.M., Kirschbaum, C., Davidson, R.J., 2008.
Glucose metabolic changes in the prefrontal cortex are associated with HPA axis
response to a psychosocial stressor. Psychoneuroendocrinology 33, 517–529.
Kirschbaum, C., Wust, S., Faig, H.-G., Hellhammer, D.H., 1992. Heritability of cortisol
responses to human corticotropin-releasing hormone, ergometry, and psycholog-
ical stress in humans. J. Clin. Endocrinol. Metab. 75, 1526–1530.
Kremen, W.S., Thompson-Brenner, H., Leung, Y.J., Grant, M.D., Franz, C.E., Eisen, S.A.,
Jacobson, K.C., Boake, C., Lyons, M.J., 2006. Genes, environment, and time: The
Vietnam Era Twin Study of Aging (VETSA). Twin Res. Hum. Genet. 9, 1009–1022.
Kremen, W.S., Prom-Wormley, E., Panizzon, M.S., Eyler, L.T., Fischl, B., Neale, M.C., Franz,
C.E., Lyons, M.J., Pacheco, J., Perry, M.E., Stevens, A., Schmitt, J.E., Grant, M.D.,
Seidman, L.J., Thermenos, H.W., Tsuang, M.T., Eisen, S.A., Dale, A.M., Fennema-
Notestine, C., 2010. Genetic and environmental influences on the size of specific
brain regions in midlife: the VETSA MRI study. Neuroimage 49, 1213–1223.
Levene, R.Z., Schwartz, B., Workman, P.L., 1972. Heritability of plasma cortisol. Arch.
Ophthalmol. 87, 389–391.
Liston, C., McEwen, B.S., Casey, B.J., 2009. Psychosocial stress reversibly disrupts pre-
frontal processing and attentional control. Proc. Natl. Acad. Sci. U. S. A. 106, 912–917.
Lupien, S.J., Lepage, M., 2001. Stress, memory, and the hippocampus: can't live with it,
can't live without it. Behav. Brain Res. 127, 137–158.
Lupien, S., Lecours, A.R., Lussier, I., Schwartz, G., Nair, N.P.V., Meaney, M.J., 1994. Basal
cortisol levels and cognitive deficits in human aging. J. Neurosci. 14, 2893–2903.
Lupien, S.J., Leon, M.D., Santi, S.D., Convit, A., Tarshish, C., Nair, N.P.V., Thakur, M.,
McEwen, B.S., Hauger, R.L., Meaney, M.J., 1998. Cortisol levels during human aging
predict hippocampal atrophy and memory deficits. Nat. Neurosci. 1, 69–73.
Lupien, S.J., Nair, N.P.V., Briere, S., Maheu, F., Tu, M.T., Lemay, M., McEwen, B.S.,
Meaney, M.J., 1999. Increased cortisol levels and impaired cognition in human
aging: implication for depression and dementia in later life. Rev. Neurosci. 10,
117–139.
MacLullich, A.M., Deary, I.J., Starr, J.M., Ferguson, K.J., Wardlaw, J.M., Seckl, J.R., 2005.
Plasma cortisol levels, brain volumes and cognition in healthy elderly men.
Psychoneuroendocrinology 30, 505–515.
MacLullich, A.M., Ferguson, K.J., Wardlaw, J.M., Starr, J.M., Deary, I.J., Seckl, J.R., 2006.
Smaller left anterior cingulate cortex volumes are associated with impaired
hypothalamic–pituitary–adrenal axis regulation in healthy elderly men. J. Clin.
Endocrinol. Metab. 91, 1591–1594.
McEwen, B.S., 2007. Physiology and neurobiology of stress and adaptation: central role
of the brain. Physiol. Rev. 87, 873–904.
McEwen, B.S., Weis, J.M., Schwartz, L.S., 1968. Selective retention of corticosterone by
limbic structure in rat brain. Nature 220, 911–912.
McGrevy, D.F., Knouse, S.B., Thompson, R.A., 1974. Relationships among an individual
intelligence test and two air force screening and selection tests. Personnel Research
Division, Air Force Human Resources Laboratory Technical Report, AFHRL-TR-74-
25. Brooks Air Force Base, TX.
Meikle, A.W., Stringham, J.D., Woodward, M.G., Bishop, D.T., 1988. Heritability of
variation of plasma cortisol levels. Metabolism: Clinical and Experimental 37,
514–517.
1101
W.S. Kremen et al. / NeuroImage 53 (2010) 1093–1102

Page 10

Mizoguchi, K., Ishige, A., Aburada, M., Tabira, T., 2003. Chronic stress attenuates
glucocorticoid negative feedback: involvement of the prefrontal cortex and
hippocampus. Neuroscience 119, 887–897.
National Center for Disease Statistics, 2003. Health, United States, 2003. U.S.
Department of Health and Human Services, Centers for Disease Control and
Prevention, Hyattsville, MD.
Neale, M.C., Cardon, L.R., 1992. Methodology for genetic studies of twins and families.
Kluwer Academic Publishers, Dordrecht, The Netherlands.
Neale, M.C., Boker, S.M., Xie, G., Maes, H.H., 2003. Mx: Statistical Modeling, 6th ed.
Department of Psychiatry, Medical College of Virginia, Richmond, VA.
O'Brien, J.T., Ames, D., Schweitzer, I., Colman, P., Desmond, P., Tress, B., 1996. Clinical
and magnetic resonance imaging correlates of hypothalamic–pituitary–adrenal
axis function in depression and Alzheimer's disease. Br. J. Psychiatry 168, 679–687.
Panizzon, M.S., Fennema-Notestine, C., Eyler, L.T., Jernigan, T.L., Prom-Wormley, E.,
Neale, M.C., Jacobson, K.C., Lyons, M.J., Grant, M.D., Franz, C.E., Xian, H., Tsuang, M.T.,
Fischl, B., Seidman, L.J., Dale, A.M., Kremen, W.S., 2009. Distinct genetic influences
on cortical surface area and cortical thickness. Cereb. Cortex 19, 2728–2735.
Paul, C.A., Au, R., Fredman, L., Massaro, J.M., Seshadri, S., Decarli, C., Wolf, P.A., 2008.
Association of alcohol consumption with brain volume in the Framingham study.
Arch. Neurol. 65, 1363–1367.
Piazza, P.V., Rouge-Pont, F., Deroche, V., Maccari, S., Simon, H., Le Moal, M., 1996.
Glucocorticoids have state-dependent stimulant effects on the mesencephalic
dopaminergic transmission. Fraga, M.F. 93, 8716–8720.
Pruessner, J.C., Hellhammer, D.H., Kirschbaum, C., 1999. Burnout, perceived stress, and
cortisol responses to awakening. Psychosom. Med. 61, 197–204.
Pruessner, M., Pruessner, J.C., Hellhammer, D.H., Bruce Pike, G., Lupien, S.J., 2007. The
associations among hippocampal volume, cortisol reactivity, and memory
performance in healthy young men. Van Snellenberg, J.X. 155, 1–10.
Pugh, K.G., Lipsitz, L.A., 2002. The microvascular frontal–subcortical syndrome of aging.
Neurobiol. Aging 23, 421–431.
Rakic, P., 1988. Specification of cerebral cortical areas. Science 241, 170–176.
Rakic, P., 1995. A small step for the cell, a giant leap for mankind: a hypothesis of
neocortical expansion during evolution. Trends Neurosci. 18, 383–388.
Rakic, P., 2007. The radial edifice of cortical architecture: from neuronal silhouettes to
genetic engineering. Brain Res. Rev. 55, 204–219.
Raz, N., 2000. Aging of the brain and its impact on cognitive performance: integration of
structural and functional findings. In: Craik, F.I.M., Salthouse, T.A. (Eds.), The
Handbook of Aging and Cognition. Erlbaum, Hillsdale, NJ, pp. 1–90.
Raz, N., Rodrigue, K.M., 2006. Differential aging of the brain: patterns, cognitive
correlates and modifiers. Neurosci. Biobehav. Rev. 30.
Rimol, L.M., Panizzon, M.S., Fennema-Notestine, C., Eyler, L.T., Fischl, B., Franz, C.E.,
Hagler, D.J., Lyons, M.J., Neale, M.C., Pacheco, J., Perry, M.E., Schmitt, J.E., Grant, M.D.,
Seidman, L.J., Thermenos, H.W., Tsuang, M.T., Eisen, S.A., Kremen, W.S., Dale, A.M.,
2010. Cortical thickness is influenced by regionally-specific genetic factors. Biol.
Psychiatry 67, 493–499.
Sánchez, M.M., Young, L.J., Plotsky, P.M., Insel, T.R., 2000. Distribution of corticosteroid
receptors in the rhesus brain: relative absence of glucocorticoid receptors in the
hippocampal formation. J. Neurosci. 20, 4657–4668.
Sapolsky, R.M., Romero, L.M., Munck, A.U., 2000. How do glucocorticoids influence
stress responses? Integrating permissive, suppressive, stimulatory, and preparative
actions. Endocrinol. Rev. 21, 55–89.
Sarrieau, A.S., Dussaillant, M., Agid, F., Moguilewsky, M., Philibert, D., Agid, Y., Rostene,
W.H., 1986. Radioautographic localization of glucocorticosteroid and progesterone
binding sites in the human post-mortem brain. J. Steroid Biochem. 25, 717–721.
SAS Institute, 2000. SAS/STAT User's Guide, Volume 8. SAS Institute, Carey, NC.
Schmitt, J.E., Eyler, L.T., Giedd, J.N., Kremen, W.S., Kendler, K.S., Neale, M.C., 2007.
Review of twin and family studies on neuroanatomic phenotypes and typical
neurodevelopment. Twin Res. Hum. Genet. 10, 683–694.
Seeman, T.E., Singer, B., Wilkinson, C.W., McEwen, B., 2001. Gender differences in age-
related changes in HPA axis reactivity. Psychoneuroendocrinology 26, 225–240.
Shansky, R.M., Hamo, C., Hof, P.R., McEwen, B.S., Morrison, J.H., 2009. Stress-induced
dendritic remodeling in the prefrontal cortex is circuit specific. Cereb. Cortex 19,
2479–2484.
Sheline, Y.I., Wang, P.W., Gado, M.H., Csernansky, J.G., Vannier, M.W., 1996. Hip-
pocampal atrophy in recurrent major depression. Proc. Nat. Acad. Sci. U S A 93,
3908–3913.
Stenfors, C., Hallerback, T., Larsson, L.G., Wallsten, C., Ross, S.B., 2004. Pharmacology of a
novel selective 5-hydroxytryptamine1B receptor antagonist, AR-A000002. Naunyn
Schmiedebergs Arch. Pharmacol. 369, 330–337.
Tchiteya, B.M., Lecours, A.R., Elie, R., Lupien, S.J., 2003. Impact of a unilateral brain lesion
on cortisol secretion and emotional state: anterior/posterior dissociation in
humans. Psychoneuroendocrinology 28, 674–686.
Tsuang, M.T., Bar, J.L., Harley, R.M., Lyons, M.J., 2001. The Harvard Twin Study of
Substance Abuse: what we have learned. Harv. Rev. Psychiatry 9, 267–279.
Van Cauter, E., Leproult, R., Kupfer, D.J., 1996. Effects of gender and age on the levels and
circadian rhythmicity of plasma cortisol. J. Clin. Endocrinol. Metab. 81, 2468–2473.
Van Cauter, E., Leproult, R., Plat, L., 2000. Age-related changes in slow wave sleep and
REM sleep and relationship with growth hormone and cortisol levels in healthy
men. J. Am. Med. Assoc. 284, 861–868.
Walhovd, K.B., Fjell, A.M., Reinvang, I., Lundervold, A., Fischl, B., Salat, D., Quinn, B.T.,
Makris, N., Dale, A.M., 2005. Cortical volume and speed-of-processing are comple-
mentary in prediction of performance intelligence. Neuropsychologia 43, 704–713.
Wellman,C.L.,2001.Dendriticreorganizationinpyramidalneuronsinmedialprefrontal
cortex after chronic corticosterone administration. J. Neurobiol. 49, 245–253.
Wenk, G.L., Pierce, D.J., Struble, R.G., Price, D.L., Cork, L.C., 1989. Age-related changes in
multiple neurotransmitter systems in the monkey brain. Neurobiol. Aging 10, 11–19.
1102
W.S. Kremen et al. / NeuroImage 53 (2010) 1093–1102