The aftermath of 9/11: effect of intensity and recency of trauma on outcome.

Abstract

Does trauma exposure have a long-term impact on the brain and behavior of healthy individuals? The authors used functional magnetic resonance imaging to assess the impact of proximity to the disaster of September 11, 2001, on amygdala function in 22 healthy adults. More than three years after the terrorist attacks, bilateral amygdala activity in response to viewing fearful faces compared to calm ones was higher in people who were within 1.5 miles of the World Trade Center on 9/11, relative to those who were living more than 200 miles away (all were living in the New York metropolitan area at time of scan). This activity mediated the relationship between group status and current symptoms of posttraumatic stress disorder. In turn, the effect of group status on both amygdala activation (fearful vs. calm faces) and current symptoms was statistically explained by time since worst trauma in lifetime and intensity of worst trauma, as indicated by reported symptoms at time of the trauma. These data are consistent with a model of heightened amygdala reactivity following high-intensity trauma exposure, with relatively slow recovery.

Full text

Page 1

The Aftermath of 9/11: Effect of Intensity and Recency
of Trauma on Outcome
Barbara Ganzel and B. J. Casey
Cornell University
Gary Glover
Stanford University
Henning U. Voss and Elise Temple
Cornell University
Does trauma exposure have a long-term impact on the brain and behavior of healthy individuals? The
authors used functional magnetic resonance imaging to assess the impact of proximity to the disaster of
September 11, 2001, on amygdala function in 22 healthy adults. More than three years after the terrorist
attacks, bilateral amygdala activity in response to viewing fearful faces compared to calm ones was
higher in people who were within 1.5 miles of the World Trade Center on 9/11, relative to those who were
living more than 200 miles away (all were living in the New York metropolitan area at time of scan). This
activity mediated the relationship between group status and current symptoms of posttraumatic stress
disorder. In turn, the effect of group status on both amygdala activation (fearful vs. calm faces) and
current symptoms was statistically explained by time since worst trauma in lifetime and intensity of worst
trauma, as indicated by reported symptoms at time of the trauma. These data are consistent with a model
of heightened amygdala reactivity following high-intensity trauma exposure, with relatively slow
recovery.
Keywords: amygdala, trauma, stress, neuroplasticity, 9/11
Psychological traumas have been defined as events that threaten
death or injury to self or others and that engender intense feelings
of fear, helplessness, or horror (e.g., rape, combat, witnessing
violence or disaster, or the sudden death of a loved one; American
Psychiatric Association, 2000). Trauma exposure is a potent envi-
ronmental risk factor that predicts immediate and lifetime in-
creases in a diverse array of mental health disorders (Breslau et al.,
1998; Kessler, Sonnega, Bromet, Hughes, & Nelson, 1995), in-
cluding anxiety and depression (Brown, 1993; Kendler, Hettema,
Butera, Gardner, & Prescott, 2003; McCauley, Kern, Kolodner,
Dill, & Schroeder, 1997). Trauma exposure is also a definitional
prerequisite for posttraumatic stress disorder (PTSD) and a predis-
posing factor for further incidence of PTSD if there is a subsequent
trauma(Bremner,Southwick,Johnson,Yehuda,&Charney,1993).In
cases where trauma exposure is severe and protracted, rates of mor-
tality and chronic illness far exceed the norm later in life, and mean
life expectancy is sharply decreased (McFarlane, 1997).
Trauma exposure is not uncommon. Results from an epidemi-
ological study of 5,877 people within the United States found that
more than 50% of women and 60% of men experienced at least
one trauma in their lifetime, and more than a quarter of the sample
experienced two or more traumas (Kessler et al., 1995). Although
only a small percentage of the trauma-exposed population devel-
ops PTSD (6% to 9%, Breslau et al., 1998; 8% to 20%, Kessler et
al., 1995), this disorder has been the focus of studies of the effects
of trauma on humans. The response to trauma in people without a
specific clinical disorder has been less well characterized (Bunce,
Larsen, & Peterson, 1995; McFarlane, 1997). Thus, although
trauma exposure is an established environmental risk factor for a
wide range of disorders, the neural mechanisms underlying this
overall increase in risk are unclear.
A small number of epidemiological and community-based studies
have examined the effects of trauma exposure in individuals without
PTSD (Otte et al., 2005: Yehuda, Golier, & Kaufman, 2005; Young
& Breslau, 2004a, 2004b; Young, Tolman, Witkowski, & Kaplan,
2004) by using peripheral biological indicators of the stress response
(e.g., urinary cortisol, dopamine, epinephrine). From these efforts, the
physiological correlates of trauma exposure in healthy (nonclinical)
populations are beginning to emerge. For example, trauma-exposed
young adults without PTSD were found to have significantly lower
urinary catecholamine levels (dopamine and epinephrine) than
nontrauma-exposed individuals, whereas a group with lifetime PTSD
Barbara Ganzel, Sackler Institute for Developmental Psychobiology,
Weill Medical College of Cornell University, and Department of Human
Development, Cornell University, Ithaca, New York; B. J. Casey, Sackler
Institute for Developmental Psychobiology, Weill Medical College of
Cornell University, Ithaca, New York; Gary Glover, Lucas Magnetic
Resonance Image Center, Stanford University, Palo Alto, California;
Henning U. Voss, Citigroup Biomedical Imaging Center, Weill Medical
College of Cornell University, Ithaca, New York; Elise Temple, Depart-
ment of Human Development, Cornell University, Ithaca, New York.
We thank Jason Zevin, Chiyoko Kobayashi, Todd Hare, John Ecken-
rode, and Jason Buhle for helpful discussions throughout the course of this
study. Special thanks to Clayton Eccard for valuable assistance with this
project and thanks also to Elizabeth Aronstam, Gregor Carrigan, Molly
Henderson, Eric Horowitz, Judith Katz, and Michoel Snow. This research
was supported by NIH Kirchstein NRSA MH68139-01A2 to BG.
Correspondence concerning this article should be addressed to Barbara
Ganzel, Department of Human Development, MVR Hall, Cornell Univer-
sity, Ithaca, NY 14853. E-mail: blg4@cornell.edu
Emotion
2007, Vol. 7, No. 2, 227–238
Copyright 2007 by the American Psychological Association
1528-3542/07/$12.00 DOI: 10.1037/1528-3542.7.2.227
227

Page 2

had higher levels (Young & Breslau, 2004b). Another study (Otte et
al., 2005) found significantly higher basal levels of MHPG (3-
methoxy-4-hydroxy-phenylglycol, a major metabolite of nonrepi-
nephrine) and greater MHPG reactivity in individuals with childhood
trauma as compared to those with no childhood trauma. Although
these findings are not conclusive, they are suggestive of long-term
effects of trauma exposure on the central nervous system that may be
more directly investigated with noninvasive neuroimaging tech-
niques.
There have been a number of neuroimaging studies examining
brain function and structure in individuals with PTSD. This re-
search has provided substantial evidence for long-term central
nervous system effects of trauma exposure that is accompanied by
PTSD. Some of these studies have found enhanced amygdala
activation in response to a variety of negatively valenced stimuli
(e.g., Rauch et al., 2000; Shin et al., 2004; although, see Lanius et
al., 2003; Phan, Britton, Taylor, Fig, & Liberzon, 2006; Sakamoto
et al., 2005). The majority of this work has involved reexperienc-
ing paradigms (e.g., Britton, Phan, Fig, Taylor, & Liberzon, 2005;
Lanius et al., 2005), which include reinstatement of memories of
the traumatic event. The format of these paradigms makes it
difficult to assess whether reported results are specific to memory
reinstatement or are a reflection of the disorder itself. It has also
been argued (Williams et al., 2006) that the use of standardized
probes of limbic activity allows for generalization of findings
to other populations. As a consequence, a growing number
of neuroimaging studies of PTSD have used standardized probes
of amygdala activity (e.g., emotional faces, Armony, Corbo,
Clement, & Brunet, 2005; Rauch et al., 2000; Shin et al., 2005;
Williams et al., 2006; emotional scenes, Hendler at al., 2003; Phan
et al., 2006; emotional Stroop task, Shin et al., 2001; emotionally
valenced words, Bremner et al., 2003; Protopopescu et al., 2005).
The most frequently used standardized probe of amygdala ac-
tivity in samples with PTSD has been presentation of emotional
faces (Armony et al., 2005; Rauch et al., 2000; Shin et al., 2005;
Williams et al., 2006). In general, these studies have reported
increased amygdala activity to negative emotional expressions in
those with PTSD. Shin and colleagues (2005) found increased
activity in the right amygdala and decreased activity in medial
prefrontal regions in men with chronic PTSD of long duration.
Rauch et al. (2000) and Armony et al. (2005) found increased
amygdala activation in samples with PTSD, but did not report
significant differences in prefrontal activation. Williams et al.
(2006) reported increased activity in ventral amygdala and de-
creased activity in dorsal amygdala in a mixed sample of men and
women with PTSD of relatively short duration. Thus, increased
activity in the amygdala is the most consistently observed neural
correlate of PTSD as probed with emotional faces, with some
evidence that this activation is localized to subregions of the
amygdala and is associated with alterations in medial prefrontal
activity.
In comparison to the rich body of neuroimaging research on
PTSD, there has been very little work examining the neural cor-
relates of trauma exposure in people without a clinical disorder;
this is surprising, considering the known impact of negative life
events on psychological distress and disorder in the overall pop-
ulation (e.g., Dohrenwend, 2006). One recent structural MRI paper
has addressed this question from a developmental perspective
(Cohen et al., 2006), showing that retrospective report of accumu-
lated adverse childhood events predicted smaller anterior cingulate
and caudate volumes in adulthood. Adverse childhood events with
the most impact on adult brain morphology were those more likely
to meet criteria as traumas (e.g., death of a parent, witnessing
domestic violence, sexual assault). Also, a recent functional neu-
roimaging study (Sharot, Martorella, Delgado, & Phelps, in press)
used a flashbulb memory paradigm that focused on events of
September 11th. These results showed significant increases in left
amygdala activation in response to evoked memories of the events
of September 11th in adults who were in downtown Manhattan
that day, relative to those who were in midtown Manhattan, five
miles away.
Animal models of stress have highlighted the effects of severe
stressors on the amygdala and related structures. Exposure to acute
uncontrollable stressors produces extended hyperexcitability of the
amygdala in laboratory animals (Adamec, Blundell, & Burton,
2005; Maier & Watkins, 1998), which renders the amygdala and
related structures more readily activated independently of the
triggering stimulus (Rosen & Schulkin, 1998). This effect has been
associated with increased vigilance and fearful responses (e.g.,
freezing) to ambiguous or mild standardized stressors (Adamec et
al., 2005; Rosen & Schulkin, 1998). Exposure to acute stressors
increases spine synapse formation in the basolateral amygdala
(BLA), which may underlie the associated increases in anxiety-like
behavior (Mitra, Jadhav, McEwen, & Chattarji, 2005; Vyas, Mitra,
Shankaranarayana Rao, & Chattarji, 2002). Exposure to repeated/
chronic stressors produces anxiety-like behavior in response to stan-
dardized stressors (e.g., an open field), along with dendritic growth in
the BLA, greater increases spine density than seen with acute stres-
sors, and dendritic retraction in the hippocampus (McEwen, 2005;
Mitra et al., 2005).
Taken together, these data suggest the hypothesis that the amyg-
dala and closely related structures are persistently more reactive
after trauma exposure in healthy adults (i.e., in individuals without
a clinical disorder) and that these effects will be observable using
mild, standardized stressors (i.e., that they do not require a reex-
periencing paradigm). Epidemiological studies indicate that it is
the nature of the worst trauma (i.e., the event that precipitates
diagnosis and/or the most symptoms) that best predicts long-term
negative psychological consequences of trauma exposure. Worst
traumas that are more recent (Kessler et al., 1995) and more severe
(e.g., violent assault: Breslau et al., 1998; Kessler et al., 1995) are
more likely to predict distress and disorder. Selection of a sample
with relatively recent, high-intensity, worst-trauma exposure
would maximize the possibility of observing the hypothesized
increase in amygdala reactivity.
The events of September 11, 2001, and the experiences of the men
and women who were in close proximity to that disaster, provide a
unique window into the neural correlates of environmental stressor
exposure. Notably for this study, proximity to this disaster have been
shown to predict psychological morbidity (Blanchard et al., 2004;
Galea et al., 2002), making it possible to prospectively titer the effects
of the disaster across study groups by studying participants who were
near the World Trade Center (WTC) in New York City on September
11, 2001, relative to those who lived far away at that time. We
hypothesize that relative increases in amygdala reactivity would be
more apparent in the group with closer proximity to the WTC on
September 11th. Building on previous assays of amygdala respon-
siveness (Brieter et al., 1996; Thomas et al., 2001), participants
228
GANZEL ET AL.

Page 3

passively viewed blocks of fearful and calm faces while undergoing
functional MRI (fMRI).
Methods
Participants
Twenty-two right-handed adults participated in this study; all
were living in the New York metropolitan area at time of scan.
Recruitment material requested participants who were at different
distances from the WTC on September 11th. Eleven participants
were within 1.5 miles of the WTC on September 11, 2001 (5
women, aged 30.3 ? 2.5 years [mean ? SE]; range ? 19 to 41
years). Eleven lived at least 200 miles from the New York City
area on September 11, 2001 (5 women, aged 29.3 ? 1.4 years;
range ? 23 to 37 years). People who lived in Washington, DC, on
September 11th and those who had friends or relatives on aircraft
involved in the disaster were excluded. Data were collected be-
tween 41 and 48 months after September 11th.
Before imaging, all participants were screened for current or
past psychiatric, neurological, or medical illness and trauma ex-
posure, as well as for any contraindications for fMRI. Approxi-
mately 30 minutes elapsed between time of interview and time of
scan. This investigation was conducted within institutional guide-
lines established for protection of human subjects. All participants
provided informed written consent.
Behavioral Measures
Three different clinical and standardized assessments were used to
assess psychiatric symptoms and diagnosis on the day of the MRI
visit: (1) the PTSD module of the University of Michigan Composite
International Diagnostic Interview (UM-CIDI: Kessler et al., 1994)
was used in conjunction with the Life History Calendar methodology
developedbyCaspietal.(1996)toassesslifetimeincidenceoftrauma
exposure, current PTSD, and PTSD symptoms in lifetime. The UM-
CIDI is a fully structured diagnostic interview that allows the assess-
ment of current and lifetime mental disorders in the form of the third
revised Diagnostic and Statistical Manual for Mental Disorders
(American Psychiatric Association, 1986); (2) the Impact of Events
Scale (Horowitz, Wilner, & Alvarez, 1979) is a 15-item scale that was
used to assess PTSD stress reactions in the seven days before scan-
ning; (3) the Speilberger State Trait Anxiety Inventory (STAI-S:
Spielberger, 1973) is a 20-item scale that was used to assess current
state anxiety.
Stimuli and Procedure
Participants viewed gray-scaled pictures of faces of eight dif-
ferent actors (four female) demonstrating fearful and calm facial
expressions (see Figure 1). Images were selected from a standard-
ized picture set (Tottenham et al., in press). Face stimuli were
presented on an overhead liquid crystal panel in a pseudorandom
sequence of nine blocks of fixation, fearful faces, or calm faces.
Participants were instructed, “Please look at the faces and the ?’s.
You don’t have to press any buttons.” There was no reference to
the events of September 11th during imaging. Order of presenta-
tion was counterbalanced across subjects and across runs using the
two following sequences: ?FC ? FC ? FC or ? CF ? CF ? CF
(where F indicates a block of fearful faces, C indicates a block of
calm faces, and ? indicates fixation). In each block of faces, 10
images were presented for 4 seconds each. Fixation blocks were 30
seconds. Total block duration was 330 seconds.
Imaging Protocols
Subjects were scanned with a General Electric Signa 3-Tesla
fMRI scanner (General Electric Medical Systems, Milwaukee, WI)
using a quadrature head coil. After a three-plane localizer and a
whole-head coronal localizer, a T2-weighted two-dimensional an-
atomical image with a fast spin-echo (FSE) sequence was ac-
quired: time for repetition (TR) ? 4000, time for echo (TE) ? 68
ms, flip ? 90°, field of view (FOV) ? 20, 29 slices, 5-mm slice
thickness, 0-mm gap, matrix ? 256 ? 192, axial-oblique. A
three-dimensional spoiled gradient recalled (SPGR) T1-weighted
anatomic scan was also acquired (124 axial slices, TR ? 25 ms,
TE ? 5 ms, flip ? 20°, FOV ? 24 cm, 1.5-mm thickness, 0 mm
gap, matrix ? 256 ? 256 x 160). Functional data was acquired
using a spiral in-out sequence (Preston, Thomason, Ochsner,
Cooper, & Glover, 2004) and the same spatial prescription as the
FSE: TR ? 2000 ms, TE ? 30 ms, matrix ? 64 ? 64 mm, 29
slices per volume.
Data Analysis
Preprocessing and statistical analysis of fMRI data was per-
formed by using Statistical Parametric Mapping (SPM2: Well-
come Department of Neurology, London, United Kingdom) im-
plemented on MatLab 7.0. During preprocessing, the first four
acquisitions were discarded and functional scans were realigned to
the initial image, generating a set of realignment parameters for
each run and a mean functional image. The mean functional image
was used to coregister functional scans to the FSE anatomical
images, which were then coregistered to the SPGR. The resulting
parameters were used to realign the functional scans. The SPGR
was then transformed to Montreal Neurological Institute (MNI)
space, and these parameters were applied to the functional scans.
The normalized functional data was smoothed by using a 6-mm
full-width/half-maximum kernel.
Individual level analysis was performed by using the general
linear model (Friston, Holmes, Price, Buchel, & Worsley, 1999)
Figure 1.
a passive-viewing paradigm that was counterbalanced ? FC ? FC ? FC
or ? CF ? CF ? CF.
Fearful and calm faces and a fixation cross were presented in
229
THE AFTERMATH OF 9/11

Page 4

with a fixed effects model. Contrast images comparing each of the
block types were generated for each individual. These individual-
level contrasts served as the basis for group-level random effects
analyses (Friston et al., 1999). Data were first analyzed by using
whole-brain analysis to identify significant areas activated to fear-
ful versus calm faces in all participants. Then, a one-way analysis
of variance was performed to compare brain activation across
groups (9/11-exposed and control) for fearful versus calm faces.
For all analyses, results were reported as significant if they met a
voxel-wise threshold of p ? .001, with clusters of 20 or more
contiguous voxels (Forman, Cohen, Fitzgerald, Eddy, Mintun, &
Noll, 1995). Given our strong, directional hypotheses regarding the
amygdala, results were reported as significant if p ? .01, with
clusters of 3 or more voxels (this provides a conservative estimate
of statistical significance: Foreman et al., 1995). In addition, small
volume corrections from a 5-mm sphere around a priori locations
of activation in the amygdala are reported (Worsley, 1996). All
activations are reported using MNI coordinates.
In order to test a priori hypotheses regarding activation in the
9/11-exposed group relative to the comparison group, a region-of-
interest (ROI) analysis was conducted. ROIs were defined func-
tionally as the voxels found to be reliably activated in the whole-
brain analysis of the fearful versus calm contrast at thresholds of
p ? .01, with cluster extents of three or more contiguous voxels.
Because there were no medial prefrontal areas reliably activated in
the whole brain analysis of variance, post hoc ROI analyses were
limited to the amygdala. The association between the behavioral
measures and blood oxygen-level dependent (BOLD) signal
change in the amygdala for the fearful-calm contrast in each ROI
was examined by using multiple regression analyses.
Results
Behavioral Results
Demographics and trauma.
trauma variables across groups revealed no significant differences in
Comparison of demographic and
age, gender, age at first trauma, number of traumas in lifetime,
number of traumas in lifetime with associated shock and/or horror, or
history of PTSD (see Table 1). The control group had a significantly
longer mean time since their worst trauma in lifetime than the 9/11-
exposed group. The 9/11-exposed group also retrospectively reported
more symptoms of avoidance and arousal at time of worst trauma in
response to probes from the PTSD module of the UM-CIDI. Most,
but not all, 9/11-exposed individuals rated the 9/11 disaster as the
worst trauma that they had experienced in their lifetime.
Current symptoms and anxiety.
group did not meet diagnostic criteria for any disorder, they had a
higher mean level of current symptoms, as assessed by both
subscales of the IES (see Table 1). There were no significant
differences between groups on the state measure of the STAI-S.
Although the 9/11-exposed
Imaging Results
BOLD signal in left and right amygdala was elevated in the
group that was closer to the WTC on September 11th in the
contrast of fearful versus calm faces (see Figure 2). Voxel-wise
analysis of variance results indicated that the 9/11-exposed group
had significantly greater activity than controls for fearful versus
calm faces in the left amygdala (?25, ?9, ?20; z ? 3.13, p ?
.001, psvc? .01, 19 contiguous voxels) and in the right amygdala
(22, ?9, ?20; z ? 2.97, p ? .001, psvc? .01, 5 contiguous
voxels).
The ROI analysis also showed that the mean signal change in
the left, t (22) ? ?3.3, p ? .003 and right amygdala, t (22) ?
?2.7, p ? .01 was higher in the 9/11-exposed group than in the
comparison group for the fearful versus calm contrast. Signal
change in these regions was not correlated with gender, age,
number of traumas in lifetime, age at first trauma, or years since
most recent trauma.
Table 2 shows correlations between BOLD signal change in
right and left amygdala (fearful-calm contrast) with behavioral
variables across the whole group. Signal change in the left amyg-
Table 1
Comparison of Demographic, Trauma, and Behavioral Variables Across Study Groups
Comparison mean (SE)9/11-exposed mean (SE)
Age at scan (y)
Gender
Age at first trauma
Years since most recent trauma
Years since worst trauma
Number of traumas in lifetime
Number of traumas in lifetime with shock/horror
Total number of symptoms at worst trauma
Number intrusion symptoms at worst trauma
Number of avoidance symptoms at worst trauma
Number of arousal symptoms at worst trauma
IES
IES subscale (intrusion)
IES subscale (avoidance)
State-Trait Anxiety Inventory
History of posttraumatic stress disorder in lifetime
29.3 (1.4)
5F, 6M
12.5 (1.6)
6.0 (1.7)
11.7 (2.6)
6.6 (2.7)
3.4 (0.9)
3.2 (1.1)
1.2 (0.5)
0.8 (0.4)
1.2 (0.5)
15.7 (0.5)
7.8 (0.4)
8.1 (0.1)
30.1 (1.1)
0.2 (0.1)
30.5 (2.5)
5F, 6M
17.3 (2.6)
3.3 (.41)†
4.9 (1.3)*
4.4 (0.7)
3.4 (0.6)
7.4 (1.2)*
2.2 (0.4)
2.3 (0.5)*
2.8 (0.5)*
22.3 (2.6)*
10.8 (1.2)*
11.4 (1.6)*
28.5 (2.0)
0.3 (0.1)
Note.
(where n[comparison] ? 10).
*p ? .05.
F ? female; M ? male; IES ? Impact of Events Scale. n ? 11 in each group, except for trauma variables
†p ? .10.
230
GANZEL ET AL.

Page 5

Figure 2.
elicited greater amygdala activity than calm faces in the whole group. Voxel-wise analysis of variance for this
contrast indicated that the 9/11-exposed group showed significantly increased bilateral amygdala activation
relative to the comparison group. Region of interest analysis showed mean signal change in the (B) left amygdala
(p ? .01) and (C) right amygdala (p ? .05) (fearful vs. calm contrast) was higher in the group that was exposed
to the disaster on September 11, 2001 than in the comparison group.
Amygdala activity and proximity to the WTC on September 11th, 2001. (A) Fearful emotional faces
231
THE AFTERMATH OF 9/11

Page 6

dala was correlated robustly with current symptoms as reported on
the IES, as well as with retrospectively reported symptoms of
avoidance on the UM-CIDI. Signal change in the right amygdala
was less strongly correlated with overall current symptoms on the
IES and with retrospective report of avoidance symptoms at time
of worst trauma. One outlier was removed for these analyses
(Mahalanobis distance ? 12.37: Darlington, 1990). These relation-
ships were not driven by data from individuals with a past history
of PTSD; control for history of PTSD strengthened the signifi-
cance of the relationship between BOLD signal and IES score (left
amygdala: ?R2? .17, p ? .007; right amygdala: ?R2? .14, p ?
.05) and had no effect on the association between signal and
symptoms at worst trauma (?R2? .009, not significant). With
control of 9/11 group status, there was an association between
STAI-S and percent signal change in the left amygdala that was
significant (? ? .35; p ? .05) in this sample, which is in keeping
with previous findings (Bishop, Duncan, & Lawrence, 2004).
Mediation by amygdala activation.
activity drives the association between trauma exposure and in-
creased vulnerability, then it should statistically mediate the rela-
tionship between the two (Baron & Kenny, 1986; Holland, 1986).
To determine whether amygdala activation mediated the relation-
ship between 9/11 status and current symptoms as reported on the
IES, we performed a standard test of mediation (Baron & Kenny,
1986: please see appendix). Subjects’ 9/11 status predicted higher
levels of current symptoms and of amygdala activation, and in-
creased amygdala activation predicted increased current symp-
toms. When mean signal change in the left amygdala (fear vs. calm
contrast) was entered into a linear regression model predicting
current symptoms from 9/11 status, the relationship between 9/11
status and symptoms became nonsignficant, indicating mediation
of the relationship between 9/11 status and current symptoms by
left amygdala activity. Furthermore, this model explained nearly
half of the variance in current symptoms in this sample; R2
0.46. Appropriate testing (MacKinnon & Dwyer, 1993; Sobel,
1986) found this mediated effect to be significant, z ? 2.80, p ?
.01. Thus, amygdala reactivity is a reasonable candidate for a
potentially causal mechanism associating 9/11 exposure and symp-
tom presentation (Baron & Kenny, 1986). The next analysis iden-
tifies the key parameters of exposure to the 9/11 disaster that were
associated with persistent amygdala reactivity and increased symp-
toms in this sample.
If increased amygdala re-
adj?
Years since worst trauma.
this sample, which provided the opportunity to examine temporal
effects in the psychobiological correlates of trauma exposure,
regardless of type of trauma. We previously noted that 9/11-
exposed participants had significantly fewer years since worst
trauma exposure than those in the comparison group (see Table 1).
Of the 21 participants with trauma exposure, 11 identified the 9/11
disaster as their worst trauma in lifetime (eight in the group that
was closest to the WTC on 9/11/01 and three in the group that was
farther away). The remaining 10 identified other lifetime traumas
as their worst trauma (three in the group that was closest to the
WTC, seven in the group that was farther away). Overall, years
since worst trauma ranged from less than one year to nearly 20
years, with the exception of one person who experienced very
early abuse (nearly 40 years before scanning). In the whole group,
years since worst trauma predicted reductions in current symptoms
(? ? ?0.51, p ? .02), as well as signal in the left (? ? ?0.46, p ?
.05) and right (? ? ?0.45, p ? .06) amygdala to fearful versus
calm faces (outlier excluded; see Figure 3). Slopes of these rela-
tionships are similar for the whole group and for the group that
identified non-9/11-related traumas as their worst traumas in life-
time (left amygdala: ? ? ?0.49; right amygdala: ? ? ?0.36),
suggesting that this finding may be generalizable to non-9/11
traumas. These analyses included control for age at scan, which
was found to serve as a classical suppressor variable in these
models (age was not significantly correlated to years since worst
trauma or to amygdala activation but the presence of age increased
the multiple correlation of the above regression models, suggesting
that age was suppressing some of what would otherwise be error
variance in the relationship between years since worst trauma and
amygdala activation: Darlington, 1990).
The associations between years since worst trauma and all
elements of the mediation model discussed above suggest that 9/11
status may be secondary to this factor in the prediction of current
symptoms and amygdala activation (fear vs. calm). In support of
this, statistical control of years since worst trauma in the relation-
ship between 9/11 group status and current symptoms rendered the
contribution of 9/11 status nonsignificant, suggesting that the
contribution of 9/11 status in this model was accounted for by
variation in years since worst trauma. Similarly, control of this
factor in the relationship between 9/11 status and percent signal
Trauma exposure was prevalent in
Table 2
Correlation of Percent Signal Change in Right and Left Amygdala (Fearful Versus Calm
Contrast) With Behavioral Variables (Pearson r), in the Whole Group
Signal change in left
amygdala (r)
Signal change in right
amygdala (r)
Total number of symptoms at worst trauma
Number of intrusion symptoms at worst trauma
Number of avoidance symptoms at worst trauma
Number of arousal symptoms at worst trauma
Impact of Events Scale (IES)
IES subscale (intrusion)
IES subscale (avoidance)
State-Trait Anxiety Inventory
0.46*
0.35
0.58**
0.27
0.72**
0.67**
0.60**
0.25
0.32
0.32
0.42†
0.13
0.56*
0.41†
0.54*
0.20
*p ? .05.
**p ? .01.
†p ? .10.
232
GANZEL ET AL.

Page 7

change in left amygdala (fear vs. calm) rendered the contribution
of 9/11 group status less significant (? ? 0.64, p ? .003 to ? ?
0.57, p ? .02), indicating that years since worst trauma accounts
for part of the explained variance in this relationship.
Trauma intensity.
Trauma intensity is difficult to define and
quantify, except for its immediate impact on the individual.
This relative level of arousal and distress at the time of trauma
is an important predictor of long-term vulnerability to mental
health disorders (Ozer, Best, Lipsey, & Weiss, 2003). For the
purposes of this analysis, we operationally defined intensity by
the participant’s own report of symptoms at worst trauma. In
this sample, the 9/11-exposed participants reported more symp-
toms of avoidance and arousal at time of worst trauma (see
Table 1). Trauma intensity was highly related to current symp-
toms (? ? 0.60; p ? .005). Most of the variance in this
relationship was explained by years since worst trauma (its
inclusion with trauma intensity in a model predicting current
symptoms reduced the contribution of trauma intensity to a
trend: ? ? 0.47; p ? .07). Together, years since worst trauma
and trauma intensity explained a substantial portion of the
variance in the relationship between 9/11 group status and
current symptoms (R2? 0.40).
Figure 3.
at scan of (A) mean signal change in left amygdala (fearful vs. calm) versus years since worst trauma in lifetime:
? ? ?0.46, p ? .05. (B) mean signal change in right amygdala (fearful vs. calm) versus years since worst trauma
in lifetime: ? ? ?0.45, p ? .06. Slopes of these relationships for the group with non-9/11 traumas as worst
traumas are shown for comparison (see p. 232).
Amygdala activity and time since worst trauma. Partial plots for the whole group controlling for age
233
THE AFTERMATH OF 9/11

Page 8

Increased symptoms of avoidance at time of worst trauma also
predicted increases in percent signal change in the left and right
amygdala to fearful versus calm faces (Table 2; Figure 4). While
trauma intensity was correlated with years since worst trauma (r ?
?0.48; p ? .05), it also contributed significantly to the explained
variance of percent signal in the left amygdala (fear vs. calm)
beyond that already accounted for by years since worst trauma
(? ? 0.60, p ? .02; ?R2? 0.23, p ? .02). When intensity of worst
trauma was entered with years since worst trauma into a model
predicting signal change in the left amygdala from 9/11 status,
these two variables fully controlled the contribution of 9/11 group
status. This suggests that the amount of time since an individual’s
worst trauma and the intensity of that worst trauma accounted for
the observed increases in amygdala reactivity (fearful vs. calm) in
participants who were close to the WTC on September 11th,
relative to those who were farther away.
Figure 4.
reported symptoms of avoidance at worst trauma and signal change in left amygdala (fearful vs. calm contrast):
r ? 0.58, p ? .01. (B) Correlation between the number of retrospectively reported symptoms of avoidance at
worst trauma and signal change in right amygdala (fearful vs. calm contrast): r ? 0.42, p ? .06.
Amygdala activity and trauma intensity. (A) Correlation between the number of retrospectively
234
GANZEL ET AL.

Page 9

Discussion
This fMRI study examined the neural correlates of trauma
exposure in a sample of healthy adults. We found that participants
who were within 1.5 miles of the WTC on September 11th, 2001,
had significantly higher mean levels of activation in bilateral
amygdala to fearful versus calm faces relative to those who were
living more than 200 miles away, despite having no current diag-
noses of PTSD, depression, or anxiety disorder. The 9/11-exposed
group also had higher levels of current symptoms and reported
more symptoms at time of worst trauma. Signal change in the
amygdala (fearful vs. calm contrast) fully mediated the relation-
ship between 9/11-exposure and number of current symptoms. In
turn, the effect of group status on both amygdala activation (fearful
vs. calm faces) and current symptoms was statistically explained
by time since worst trauma and intensity of worst trauma, as
indicated by reported symptoms at worst trauma. These data are
consistent with a model of heightened amygdala reactivity follow-
ing high-intensity trauma exposure, with relatively slow recovery.
Notably, similar results have recently been identified in a similar
population (Sharot et al., in press) using a paradigm that evoked
specific memories of the disaster of September 11th. The present
study supports and extends this research by suggesting that long-
term trauma-related modulation of the amygdala is observable
using mild, standardized emotional stimuli (fearful vs. calm faces),
indicating that these effects may extend further into everyday life
than previously thought.
Animal models of exposure to stress may provide insight into
the processes of brain plasticity that underlie these results. Trau-
matic stressors encountered in the natural environment are likely to
involve a mixture of conditioned and unconditioned response
(Adamec, Blundell, & Burton, 2006). The most analogous animal
models are likely to be those involving relatively severe uncon-
trollable stressors (for reviews, see Adamec et al., 2006 [unpro-
tected predator stress]; Maier & Watkins, 1998 [inescapable
shock]; McEwen, 2005 [acute and chronic restraint stress]). Con-
trollable stressors do not activate the amygdala enough to produce
even temporarily increased reactivity of the amygdala; uncontrol-
lable stressors have, however, produced moderately persistent hy-
perexcitability of the amygdala and related structures in laboratory
animals (Adamec et al., 2005; Maier & Watkins, 1998). This
hyperexcitability has been associated with increased fearful re-
sponse to ambiguous or mild stressors (Adamec et al., 2006; Maier
& Watkins, 1998).
At the neuronal level, one type of uncontrollable stressor (chronic
restraint stress or CRS) has been shown to produce hypertrophy of the
dendritic arborization in the BLA and extended amygdala, accompa-
nied by dendritic atrophy and decreases in spine density in medial
prefrontal areas and the hippocampus (Mitra et al., 2005; Vyas et al.,
2002). In rats, these changes are associated with standard behavioral
indicators of anxiety in rodents, including decreased time spent in the
open (exposed) arms of an elevated plus maze and reduction in open
armentries(Mitraetal.,2005;Vyasetal.,2002).Acute(single-event)
restraint stress produces similar increases in spine density in the BLA
and behavioral indicators of anxiety, with no associated changes in
prefrontal areas (Mitra et al., 2005). Consistent with these findings,
single-event unprotected predator stress produces persistent increases
in evoked potentials from neurons in the BLA, which are also asso-
ciated with increased open arm entry and decreased in exploration in
the elevated plus maze (Ademac, Blundell, & Burton, 2005). Thus, a
single noxious event of various types can drive alterations in the
amygdala that are associated with persistent increases in behavioral
reactivity in situations of uncertainty and potential threat. Notably,
although CRS-induced changes to prefrontal areas and hippocampus
are reversible with an extended stress-free period, hypertrophy of the
amygdala and increases in anxiety-related behaviors are more persis-
tent (Vyas, Pillai, & Chattarji, 2004). The effects of predator stress are
also reported to be slow to return to baseline (Ademac et al., 2005).
This highlights the role of the amygdala in maintaining stress-induced
anxiety-like behaviors, as well as the slow recovery of this system
over time after exposure to intensive stressors.
The findings of the present study point to the potential durability of
human stress-related neural plasticity, even in adulthood, and suggest
that such changes may be driven by the recency and intensity of worst
adverse event (worst trauma in lifetime). Although accumulated
trauma exposure did not add to the statistical prediction any of the
outcomes in this analysis, it is a point for future research whether a
more general measure of cumulative environment risk would do so.
Research has established the relationship between accumulated envi-
ronmental risk and negative socioemotional outcomes (e.g., Rutter,
1979; Sameroff, Seifer, Baldwin, & Baldwin, 1993). However, in-
creases in cumulative environmental risk (e.g., low socioeconomic
status, minority status, compromised social support, mental illness in
family) may be confounded with increased likelihood for high-
intensity trauma exposure in most community-based samples, and
cumulative risk may also exacerbate trauma outcomes (Galea et al.,
2002). More basic research is needed on the neural correlates of
stressful life events of all types in healthy individuals; this is an
exciting new area for investigation in cognitive neuroscience—one
that in the past year has begun to show forward movement (e.g.,
Cohen et al., 2006; Sharot et al., in press; this study: also see recent
work on the effects of in-scanner stressor exposure, Dedovic et al.,
2005; Wang et al., 2005).
Finally, amygdala hyperexcitability after trauma may provide a
tool for examining the neural processes that modulate stress-
related neural plasticity. Chief among these potential modulatory
processes is emotion regulation. Regulation of emotions is impor-
tant for social adaptation, and neuroimaging evidence suggests that
regulation of negative emotions, in particular, depends on modu-
lation of amygdala response using prefrontal and anterior cingulate
control systems (Ochsner & Gross, 2005). For example, amygdala
activation is attenuated in response to fearful or angry emotional
faces during cognitive evaluation of the type of emotion (Hariri,
Bookheimer, & Mazziotta, 2000). The same occurs during cogni-
tive evaluation of threatening emotional scenes and is accompa-
nied by decline in skin conductance (Hariri, Mattay, Tessitore,
Fera, & Weinberger, 2003). Different strategies of emotion regu-
lation (e.g., reappraisal vs. suppression: Gross & John, 2003; John
& Gross, 2004) are likely to modulate amygdala hyperexcitability
and associated behavioral outcomes following trauma exposure.
The latter would be a point of interest for intervention.
Limitations
This study uses a retrospective report of trauma exposure. The
difficulties with retrospective reporting include revisionist recall,
bias because an ensuing disorder is known to have occurred, bias
by respondent’s depression or cognitive impairment, and normal
235
THE AFTERMATH OF 9/11

Page 10

forgetting (Hardt & Rutter, 2004; Moffitt et al., 2006). Prospective
studies of the effects of trauma exposure on the brain would be
preferable in many ways. However, this is made difficult because
trauma exposure typically occurs unexpectedly. In the absence of
a prospective study, it has been argued (Moffitt, Caspi, & Rutter,
2006) that the use of the life history calendar (Caspi et al., 1998)
is a highly valid and reliable method for collecting retrospective
data on adverse life events. We used this method in the present
study to reduce retrospective reporting bias to a minimum.
An additional limitation of this study is that the comparison
group in the current sample exhibited a high rate of trauma
exposure. This may be due in part to participant self-selection,
although high rates of trauma exposure have also been observed in
large community-based studies (e.g., Breslau et al., 1998; Young
& Breslau, 2004a and 2004b; Young et al., 2004) that used the
UM-CIDI measure (Kessler et al., 1994) that was employed in the
present study. Because there were few subjects without trauma
exposure in this sample, we were unable to test differences be-
tween participants with and without trauma exposure. We note,
too, that because most of the sample consisted of trauma-exposed
individuals without PTSD, this may have inadvertently become a
study of resilience; if so, these findings suggest that resilience
under stress does not imply lack of physiological consequences
following stressor exposure. In addition, specific examination of
gender differences was not possible due to the small size of the
study, although we found no significant statistical contributions of
gender to the results reported here.
A further limitation to this study is that we used a passive
viewing paradigm in this “first look” for neural correlates of
trauma exposure in adult humans, because passive viewing of
emotional faces is one of the most powerful standardized methods
for evoking amygdala response (e.g., Whalen et al., 2001) and
because tasks that require an active response to emotional stimuli
can lead to top-down suppression of the amygdala response (Lange
et al., 2003). The limitation of this method is the lack of accom-
panying behavioral data to determine level of attention to the
stimuli. Having established with Sharot et al. that there are robust
neural correlates of trauma exposure, the next set of research
questions would address the source of the observed differences
between groups using more sophisticated paradigms, for example,
examination of differences in attention or emotion regulation using
more active paradigms.
Finally, in the present research, participants were aware that
the study was related to the effects of the 9/11 disaster, and they
were interviewed for trauma exposure before scanning. This
knowledge, and the screening process, may have had a priming
effect on amygdala response. This is an unexplored area and
one that requires more research. Because there was trauma
exposure in both groups (with the exception of one individual in
the comparison group), the trauma inventory might have been
expected to have a similar impact across all participants. How-
ever, it may be that specific variables related to prescreening
interviews (e.g., length and timing of the trauma interview) had
significant effects on the BOLD signal response, even years
after the trauma. If so, it would suggest even more powerful and
subtle long-term effects of previous trauma exposure on the
brain than are indicated in the current study.
Conclusions
Close proximity to the WTC on September 11th, 2001, was
associated with increased amygdala activation and higher levels of
symptoms in healthy adults. Amygdala activation mediated the
association between trauma exposure and current symptoms, sug-
gesting a role for amygdala hyperexcitability in the association
between trauma exposure and subsequent vulnerability to mental
health disorder. Overall, amygdala activation was found to de-
crease with time since worst trauma, and to increase with the
intensity of each individual’s worst-ever trauma, as measured by
symptoms at time of trauma. Together, these two variables statis-
tically explained the effect of proximity to the 9/11 disaster on
both amygdala activation and current symptoms. This finding
suggests that there is heightened amygdala activity from high
intensity traumas, and that recovery occurs over many years, even
in those without a current clinical disorder.
References
Adamec, R. E., Blundell, J., & Burton, P. (2005). Neural circuit changes
mediating lasting brain and behavioral response to predator stress. Neu-
roscience and Biobehavioral Reviews, 29, 1225–1241.
Adamec, R. E., Blundell, J., & Burton, P. (2006). Relationship of predatory
attack experience to neural plasticity, pCREB expression and neuroen-
docrine response. Neuroscience and Biobehavioral Reviews, 30, 356–
375.
American Psychiatric Association. (2000). Diagnostic and statistical man-
ual of mental disorders (4th ed.). Washington, DC: Author.
Armony, J. L., Corbo, V., Clement, M. H., & Brunet, A. (2005). Amygdala
responseinpatientswithacutePTSDtomaskedandunmaskedemotionalfacial
expressions. American Journal of Psychiatry, 162, 1961–1963.
Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable
distinction in social psychological research: Conceptual, strategic, and
statistical considerations. Journal of Personality and Social Psychology,
51, 1173–1182.
Bishop, S., Duncan, J., & Lawrence, A. (2004). State anxiety modulation
of the amygdala response to unattended threat-related stimuli. Journal of
Neuroscience, 24, 10364–10368.
Blanchard, E. B., Kuhn, E., Rowell, D., Hickling, E. J., Wittrock, D.,
Rogers, R., et al. (2004). Studies of the vicarious traumatization of
college students by the September 11th attacks: Effects of proximity,
exposure and connectedness. Behaviour Research and Therapy, 42,
191–295.
Bremner, J. D., Southwick, S., Johnson, D., Yehuda, R., & Charney, D.
(1993). Childhood physical abuse and combat-related posttraumatic
stress disorder in Vietnam veterans. American Journal of Psychiatry,
150, 235–239.
Bremner, J. D., Vythilingam, M., Vermetten, E., Southwick, S. M.,
McGlashan, T., Nazeer, A., et al. (2003). MRI and PET study of
deficits in hippocampal structure and function in women with child-
hood sexual abuse and posttraumatic stress disorder. American Jour-
nal of Psychiatry,160, 924–932.
Breslau, N., Kessler, R. C., Chilcoat, H. D., Schultz, L. R., Davis, G. C.,
& Andreski, P. (1998). Trauma and posttraumatic stress disorder in the
community: The 1996 Detroit Area Survey of Trauma. Archives of
General Psychiatry, 57, 626–632.
Brieter,H.C.,Etcoff,N.L.,Whalen,P.J.,Kennedy,W.A.,Rauch,S.L.,Buckner,
R. L., et al. (1996). Response and habituation of the human amygdala during
visual processing of facial expression. Neuron, 17, 875–887.
Britton, J. C., Phan, K. L., Taylor, S. F., Fig, L. M., & Liberzon. I. (2005).
Corticolimbic blood flow in posttraumatic stress disorder during script-
driven imagery. Biological Psychiatry 57, 832–40.
236
GANZEL ET AL.

Page 11

Brown, G. (1993). Life events and affective disorder: Replications and
limitations. Psychosomatic Medicine, 55, 248–259.
Bunce, S., Larsen, R., & Peterson, C. (1995). Life after trauma: Personality
and daily life experiences of traumatized people. Journal of Personality,
63, 165–188.
Caspi, A., Moffitt, T. E., Thornton, A., Freedman, D., Amell, J. W.,
Harrington, H. L., et al. (1998). The Life History Calendar: A research
and clinical assessment method for collecting retrospective event-history
data. International Journal of Methods in Psychiatric Research, 6,
101–114.
Cohen, R., Grieve, S., Hoth, K., Paul, R., Sweet, L., Tate, D., et al. (2006).
Early life stress and morphometry of the adult anterior cingulate cortex
and caudate nuclei. Biological Psychiatry, 59, 975–982.
Darlington, R. (1990): Regression and linear models. New York: McGraw-
Hill.
Dedovic, K., Renwick, R., Mahani, N. K., Engert, V., Lupien, S. J.,
Pruessner, J. C. (2005). The Montreal Imaging Stress Task: Using
functional imaging to investigate the effects of perceiving and process-
ing psychosocial stress in the human brain. Journal of Psychiatry and
Neuroscience, 30, 319–325.
Dohrenwend, B. P. (2006). Inventorying stressful life events as risk factors
for psychopathology: Toward resolution of the problem of intracategory
variability. Psychological Bulletin, 132, 477–495.
Forman, S., Cohen, J., Fitzgerald, M., Eddy, W., Mintun, M., & Noll, D.
(1995). Improved assessment of significant activation in functional
magnetic resonance imaging (fMRI): Use of a cluster-size threshold.
Magnetic Resonance Medicine, 33, 636–647.
Friston, K. J., Holmes, A. P., Price, C., Buchel, C., & Worsley, K. J.
(1999). Multisubject fMRI analyses and conjunction analyses. Neuro-
image, 10, 385–396.
Galea, S., Ahern, J., Resnick, H., Kilpatrick, D., Bucuvalas, M., Gold, J.,
& Wlahov, D. (2002). Psychological sequelae of the September 11
terrorist attacks in New York City. New England Journal of Medicine,
346, 982–987.
Gross, J. J., & John, O. (2003). Individual differences in two emotion
regulation processes: Implications for affect, relationships, and well-
being. Journal of Personality and Social Psychology, 85, 348–362.
Hardt, J., & Rutter, M. (2004). Validity of adult retrospective reports of
adverse childhood experiences: Review of the evidence. Journal of
Child Psychology, Psychiatry, and Allied Disciplines, 45, 260–273.
Hariri, A. R., Bookheimer, S. Y., & Mazziotta, J. C. (2000). Modulating
emotional responses: Effects of a neocortical network on the limbic
system (2000). Neuroreport, 11, 43–48.
Hariri, A. R., Mattay, V. S., Tessitore, A., Fera, F., & Weinberger, D.
(2003). Neocortical modulation of the amygdala response to fearful
stimuli. Biological Psychiatry, 53, 494–501.
Hendler, T., Rotshtein, P., Yeshurun, Y., Weizmann, T., Kahn, I., Ben-Bashat,
D., et al. (2003). Sensing the invisible: Differential sensitivity of visual
cortex and amygdala to traumatic context. Neuroimage, 19, 587–600.
Holland, P. W. (1986). Statistics and causal inference. Journal of the
American Statistical Association, 81, 945–970.
Horowitz, M., Wilner, N., & Alvarez, W. (1979). Impact of Event Scale:
A measure of subjective stress. Psychosomatic Medicine, 41, 209–218.
John, O., & Gross, J. J. (2004). Health and unhealthy emotion regulation:
Personality processes, individual differences, and life span development.
Journal of Personality, 72, 1301–1334.
Kendler, K. S., Hettema, J. M., Butera, F., Gardner, C. O., & Prescott,
C. A. (2003). Life event dimensions of loss, humiliation, entrapment,
and danger in the prediction of onsets of major depression and gener-
alized anxiety. Archives of General Psychiatry, 60, 789–796.
Kessler, R., Sonnega, A., Bromet, E., Hughes, M., & Nelson, C. (1995).
Posttraumatic stress disorder in the National Comorbidity Study. Ar-
chives of General Psychiatry, 52, 1048–1059.
Kessler, R. C., McGonagle, K., Zhao, S., Nelson, C. B., Hughes, M.,
Eshelman, S., et al. (1994). Lifetime and 12-month prevalence of DSM-
IIIR psychiatric disorders in the United States: Results from the National
Comorbidity Survey. Archives of General Psychiatry, 51, 8–19.
Lange, K., Williams, L. M., Young, A. W., Bullmore, E. T., Brammer, M. J.,
Williams, S. C., et al. (2003). Task instructions modulate neural responses
to fearful facial expressions. Biological Psychiatry, 53, 226–232.
Lanius, R. A., Williamson, P. C., Bluhm, R. L., Densmore, M., Boksman,
K., Neufeld, R. W., et al. (2005). Functional connectivity of dissociative
responses in posttraumatic stress disorder: A functional magnetic reso-
nance imaging investigation. Biological Psychiatry, 57, 873–884.
Lanius, R. A., Williamson, P. C., Hopper, J., Densmore, M., Boksman, K.,
Gupta,M.A.,etal.(2003).Recallofemotionalstatesinposttraumaticstress
disorder: An fMRI investigation. Biological Psychiatry, 53, 204–210.
MacKinnon, D. P., & Dwyer, J. H. (1993). Estimating mediated effects in
prevention studies. Evaluation Review, 17, 144–158.
Maier, S., Watkins, L. (1998). Stressor controllability, anxiety, and sero-
tonin. Cognitive Therapy and Research, 22, 595–613.
McCauley, J., Kern, D., Kolodner, K., Dill, L., & Schroeder, A. (1997).
Clinical characteristics of women with a history of childhood abuse:
Unhealed wounds. Journal of the American Medical Association, 277,
1362–1368.
McEwen, B. (2005). Glucocorticoids, depression and mood disorders:
Structural remodeling in the brain. Metabolism: Clinical and Experi-
mental, 54, 20–23.
McFarlane A (1997): The prevalence and longitudinal course of PTSD:
Implications for neurobiological models of PTSD. In: R. Yehuda & A.
McFarlane (Eds.), Psychobiology of posttraumatic stress disorder (Vol.
821, pp. 10–24). New York: New York Academy of Sciences.
Mitra, R., Jadhav, S., McEwen, B. S., & Chattarji, S. (2005). Stress
duration modulates the spatioteporal patterns of spine formation in the
basolateral amygdala. Proceedings of the National Academy of Sciences,
USA, 102, 9371–9376.
Moffitt, T. E., Caspi, A., & Rutter, M. (2006). Measured gene-environment
interactions in psychopathology. Concepts, research strategies, and im-
plications for research, intervention, and public understanding of genet-
ics. Perspectives in Psychological Science 1, 5–27.
Ochsner, K., & Gross, J. (2005). The cognitive control of emotion. Trends
in Cognitive Neuroscience, 9, 242–249.
Otte, C., Neylan, T. C., Pole, N., Metzler, T., Best, S., Henn-Haase, C., et
al. (2005). Association between childhood trauma and catecholamine
response to psychological stress in police academy recruits. Biological
Psychiatry, 57, 27–32.
Ozer, E., Best, S., Lipsey, T., & Weiss, D. (2003). Predictors of posttrau-
matic stress disorder and symptoms in adults: A meta-analysis. Psycho-
logical Bulletin, 129, 52–73.
Phan, K. L., Britton, J. C., Taylor, S. F., Fig, L. M., & Liberzon, I. (2006).
Corticolimbic blood flow during nontraumatic emotional processing in
posttraumatic stress disorder. Archives of General Psychiatry, 63, 184–192.
Preston, A. R., Thomason, M. E., Ochsner, K. N., Cooper, J. C., & Glover,
G. H. (2004). Comparison of Spiral-In/Out and Spiral-out BOLD fMRI
at 1.5T and 3T. Neuroimage, 21, 291–301.
Protopopescu, X., Pan, H., Tuescher, O., Cloitre, M., Goldstein, M.,
Engelien, W., et al. (2005). Differential time courses and specificity of
amygdala activity in posttraumatic stress disorder subjects and normal
control subjects. Biological Psychiatry, 57, 464–473.
Rauch, S., Whalen, P., Shin, L., McInerney, S., Macklin, M., & Lasko, N.
(2000). Exaggerated amygdala response to masked facial stimuli in
posttraumatic stress disorder: A functional MRI study. Biological Psy-
chiatry, 47, 769–776.
Rosen, J., & Schulkin, J. (1998). From normal fear to pathological anxiety.
Psychological Review, 105, 325–350.
Rutter, M. (1979). Protective factors in children’s responses to stress and
disadvantage. In M. W. Kent & J. E. Rolf (Eds.), Primary Prevention of
237
THE AFTERMATH OF 9/11

Page 12

Psychopathology (Vol. 3, pp. 49–74). Hanover, NH: University Press of
New England.
Sakamoto, H., Fukuda, R., Okuaki, T., Rogers, M., Kasai, K., Machida, T.,
et al. (2005). Parahippocampal activation evoked by masked traumatic
images in posttraumatic stress disorder: A functional MRI study. Neu-
roimage, 26, 813–821.
Sameroff, A. J., Seifer, R., Baldwin, A., & Baldwin, C. (1993). Stability of
intelligence from preschool to adolescence: The influence of social and
family risk factors. Child Development, 64, 80–97.
Sharot, T., Martorella, E. A., Delgado, M. R., & Phelps, E. (in press). Remem-
bering 9/11: How personal experience modulates the neural circuitry of recol-
lection. Proceedings of the National Academy of Sciences, USA.
Shin, L., Orr, S., Carson, M., Rauch, S., Macklin, M., Lasko, N., et al.
(2004). Regional cerebral blood flow in the amygdala and medial pre-
frontal cortex during traumatic imagery in male and female Vietnam
veterans with PTSD. Archives of General Psychiatry, 61, 168–176.
Shin, L. M, Whalen, P. J., Pitman, R. K., Bush, G., Macklin, M. L., Lasko,
N. B., et al. (2001). An fMRI study of anterior cingulate function in
posttraumatic stress disorder. Biological Psychiatry, 50, 932–942.
Shin, L. M., Wright, C. I., Cannistraro, P. A., Wedig, M. M., McMullin, K.,
Martis, B., et al. (2005). A functional magnetic resonance imaging study
of amygdala and medial prefrontal cortex responses to overtly presented
fearful faces in posttraumatic stress disorder. Archives of General Psy-
chiatry, 62, 273–281.
Sobel, M. E. (1986). Some new results on indirect effects and their
standard errors in covariance structure models. In N. Tuma (Ed.), So-
ciological methodology (pp. 159–186). Washington, DC: American
Sociological Association.
Spielberger, C. D. (1973). Manual for the state-trait anxiety inventory for
children. Palo Alto, CA: Consulting Psychologists Press.
Thomas, K., Drevets, W., Whalen, P., Eccard, C., Dahl, R., Ryan, N., et al.
(2001). Amygdala response to facial expressions in children and adults.
Biological Psychiatry, 49, 309–316.
Tottenham, N., Tanaka, J., Leon, A. C., McCarry, T., Nurse, M., Hare,
T. A., et al. (in press). Validity and reliability of the nimStim set of facial
expressions.
Vyas, A., Mitra, R., Shankaranarayana Rao, B. S., & Chattarji, S. (2002).
Chronic stress induces contrasting patterns of dendritic remodeling in
hippocampal and amygdalaloid neurons. Journal of Neuroscience, 22,
6810–6818.
Vyas, A., Pillai, A. G., & Chattarji, S. (2004). Recovery after chronic stress
fails to reverse amygdaloid neuronal hypertrophy and enhanced anxiety-
like behavior. Neuroscience, 128, 667–673.
Wang, J., Rao, H., Wetmore, G. S., Furlan, P. M., Korczykowski, M.,
Dinges, D. F., et al. (2005). Perfusion functional MRI reveals cerebral
blood flow pattern under psychological stress. Proceedings of the Na-
tional Academy of Sciences USA, 102, 17804–17809.
Whalen, P. J., Shin, L. M., McInerney, S. C., Fischer, H., Wright, C. I., &
Rauch, S. L. (2001). A functional MRI study of human amygdala
responses to facial expressions of fear versus anger. Emotion, 1, 70–83.
Williams, L. M., Kemp, A. H., Felmingham, K., Barton, M., Olivieri, G.,
Peduto,A.,etal.(2006).Traumamodulatesamygdalaandmedialprefrontal
responses to consciously attended fear. NeuroImage, 29, 347–357.
Worsley, K. J. (1996). A unified statistical approach for determining
significant signals in images of cerebral activation. Human Brain Map-
ping, 4, 58–73.
Yehuda, R., Golier, J. A., & Kaufman, S. (2005). Circadian rhythm of
salivary cortisol in Holocaust survivors with and without PTSD. Amer-
ican Journal of Psychiatry, 152, 998–1000.
Young, E., & Breslau, N. (2004a). Cortisol and catecholamines in post-
traumatic stress disorder: An epidemiologic community study. Archives
of General Psychiatry, 61, 394–401.
Young, E., & Breslau, N. (2004b). Saliva cortisol in posttraumatic stress
disorder: A community epidemiologic study. Biological Psychiatry, 56,
205–209.
Young, E., Tolman, R., Witkowski, K., & Kaplan, G. (2004). Salivary
cortisol and Posttraumatic Stress Disorder in a low-income community
sample of women. Biological Psychiatry, 55, 621–626.
Appendix
Statistical mediation (Baron & Kenny, 1986, p. 1176)
A variable functions as a mediator of the significant relationship
between an independent and dependent variable when (a) variation
in the levels of the independent variable significantly accounts for
variations in the presumed mediator, (b) variations in the presumed
mediator significantly account for variations in the dependent
variable, and (c) both the independent variable and the presumed
mediator are entered simultaneously into a regression model pre-
dicting the dependent variable, then the previously significant
relationship between the independent variable and the dependent
variable is reduced in significance or (in the strongest condition) is
no longer significant. MacKinnon & Dwyer (1993) provide a
statistical test for significance of this overall process, which was
used here.
Received September 5, 2006
Revision received January 11, 2007
Accepted January 12, 2007 ?
238
GANZEL ET AL.