More than three years after 9/11/01, adults who had greater proximity to the terrorist attacks on the World Trade Center had significantly lower gray matter volume in amygdala, anterior hippocampus, insula, anterior cingulate, and medial prefrontal cortex. The 9/11-exposed group had no areas of significantly increased gray matter relative to the comparison group. All study subjects were free of mental and physical health disorder, so the regional differences in gray matter volumes observed here are not attributable to the presence of clinical disorder. These results provide evidence of structural change in gray matter in healthy adults following psychological trauma.
Examination of the amygdala as a region of interest confirmed the decrease in amygdala gray matter volume in the 9/11-exposed group. In general, these analyses also showed that decreases in amygdala gray matter volume were associated with increased BOLD signal change in the left amygdala in response to fearful versus calm faces. Leftward lateralization is consistent with reports that the left amygdala is more likely to be activated in response to tasks involving the neural processing of negative emotional stimuli (Baas, Aleman, & Kahn, 2004
; Wager, Phan, Liberzon Taylor, 2003
) that involve detailed evaluation of emotional intensity (Glascher & Adolphs, 2003
). Further analysis of the relationship between amygdala function and structure and lifetime trauma exposure revealed significant differences between those with fewer traumas (≤ 5) throughout their lifetime and those with many traumas (> 5). Overall, the relationship between lifetime trauma exposure and amygdala gray matter volume took the form of a first-order quadratic, such that the slope of this relationship reversed at higher levels of trauma exposure. This held for the comparison group alone, as well as for the sample as a whole.
Analyses of lower and higher lifetime trauma subgroups helped to clarify associations between behavioral measures of anxiety and both the structural and functional imaging data for the amygdala. In the group with five or fewer traumas in their lifetime (70% of the sample), state anxiety increased as amygdala gray matter volume decreased. Among those with more than five traumas, this relationship was reversed; anxiety increased as amygdala gray matter volume increased. This produced a significant interaction between level of lifetime trauma exposure and amygdala gray matter volume in predicting state anxiety. Consistent with the inverse relationship between gray matter volume and amygdala functional reactivity in this sample, we found an interaction between level of trauma exposure and amgydala BOLD signal (fearful versus calm faces) in predicting anxiety that was the inverse of the previous interaction. Among those with five or fewer traumas, anxiety increased with increasing BOLD signal to fearful versus calm faces – however, it decreased with increasing BOLD signal in those with five or more traumas. The relationships between symptoms of PTSD, level of trauma exposure, and amgydala gray matter and response to emotional faces took a similar, although nonsignificant, form. These results require replication with a larger sample of individuals who have more than five traumas in their lifetime. However, they do suggest that there may be qualitative differences among those with higher versus lower levels of trauma exposure in the relationships between behavioral anxiety and amygdala structure and reactivity. These data also provide supporting evidence for a nonlinear trajectory in amygdala stress-related neural plasticity with accumulating trauma exposure.
The use of VBM allowed us to perform a bias-free analysis of gray matter differences across the whole brain. Because VBM allows comparison of the local composition of tissue, it is useful for showing group differences in structure that may be more subtle than traditional volumetric measurement can resolve. VBM analyses are limited by an inability to interpret the nature of the specific microstructural changes that may be measured (e.g., changes in neuropil, neuronal size, dendritic, or axonal arborisation) (Ashburner and Friston, 2001
). This issue remains unresolved and will require future studies that utilize methods other than MRI. Thus, we are unable to know what specific types of gray matter changes were associated with trauma exposure in this study. VBM analysis also includes possible confounds related to the normalization and segmentation process, which can be especially problematic in comparisons of atypical populations (Bookstein, 2001
). To help address these concerns, we used an “optimized” VBM procedure, set significance levels arbitrarily high, and included only healthy subjects for whom no group differences in global brain shape would be expected. Because VBM analyses may be less sensitive to volume loss in the amygdala than traditional region-of-interest volumetric measurement (Good et al., 2002
), the results reported here may be conservative.
These findings suggest multiple avenues for future research. Most of the brain regions identified here as having lower gray matter volume in the 9/11-exposed group (amygdala, insula, anterior cingulate, and medial prefrontal cortex) have been identified as playing key roles in the evaluation and regulation of emotional stimuli in humans (Ochsner et al., 2004
; Phan et al., 2002
), suggesting that traumatic experiences may specifically affect the neural processing and regulation of emotion in nonclinical populations. This possibility clearly requires further study. Assessment of participants’ prior trauma exposure may be an important inclusion in studies of the neural correlates of emotion processing and regulation, and exploration of possible effects of stress-related neural change on emotion regulation may also help shed light on the emerging neuroscience of emotion-cognition interactions (Ochsner and Phelps, 2007
In addition, we report trauma-related decreases in gray matter in many of the same brain regions in which gerontologists report normal age-related gray matter atrophy (Good et al., 2001
; Resnick et al., 2003
). The established prevalence of trauma exposure in the general population (Breslau et al., 1998
; Kessler et al., 1995
) suggests the possibility that some of the gray matter atrophy attributed to aging may instead be due to accumulated lifetime trauma exposure. It would be of interest to delineate the relative contribution of age versus lifetime trauma exposure to the gray matter atrophy seen in older adults. Studies of aging samples have also identified increased BOLD signal change to emotional faces (Wright et al., in press
) and decreased amygdala volume (Shiino et al., 2006
) in those with mild Alzheimer’s disease. In the present study, we found decreased amygdala gray matter volume and increased amygdala functional reactivity to emotional stimuli in subjects with moderately high levels of lifetime trauma exposure (approximately 4 to 6 traumas). It would be of interest to examine whether stress-related changes in the amygdala represent a vulnerability factor for Alzheimer’s disease or conversely, if these differences may be attributable entirely or in part to increased stress that is associated with disease onset.
Amygdala gray matter atrophy, often accompanied by hippocampal volume atrophy, has also been reported in depression (Campbell et al., 2004
; Siegle, Konecky, Thase, Carter, 2003
; Sheline et al., 1999), borderline personality disorder (Schmahl et al., 2003
), and anxiety disorder (Milham et al., 2005
). There are also reports of increased amygdala reactivity to emotional stimuli in these populations (Siegle et al., 2003
; Schmahl et al., 2003
; Sheline et al., 2001
; Thomas et al., 2001
). For example, amygdala hyperactivity in depression (Drevets et al., 1992
; Sheline et al., 2001
) appears to be associated initially with amygdala hypertrophy (Frodl et al., 2002
), followed by amygdala atrophy after multiple episodes of major depression (Sheline et al., 1999). Drawing on animal models of chronic stressor exposure, it has been argued (McEwen, 2003
) that chronic hyperexcitability of the amygdala may produce these progressive alterations in amygdala volume over time, possibly through the influence of excitatory amino acids on cell survival and neuronal architecture.” In the present study, we find that decreases in amygdala and hippocampal gray matter coupled with increased amygdala reactivity are associated with moderately high levels of lifetime trauma exposure. Because of the relatively high incidence of trauma exposure in the general population (Kessler et al., 1995
), it may be that trauma exposure plays a common explanatory role in some portion of the observed neural differences in the medial temporal lobe in these disorders. Consideration of nonlinearity in these relationships may help to clarify some of the inconsistencies in this literature (Campbell et al., 2004
A negative correlation between amygdala gray matter volume and BOLD signal in the amygdala for fearful versus calm faces was identified in the 9/11-exposed group but not in the comparison group. This lack of significance is reported with caution because technical difficulties with the scanner (see Methods) caused BOLD data to be lost for a number of subjects who had good quality data for gray matter volume. This narrowed the range of values for amygdala gray matter volume in the comparison group (note the relative ranges of values for gray matter volume in and ). This loss of data affects only the direct comparison of gray matter volume and BOLD data illustrated in . All other analyses reported here employ the full range of gray matter volume data from this sample; within this expanded data set, the effect of anxiety, symptoms, and traumas plotted against amygdala gray matter volume is consistently the inverse of the effect of anxiety, symptoms, and traumas plotted against amygdala BOLD signal (F>C), suggesting an inverse relationship between these two sets of imaging data. Full articulation of this relationship requires replication within a larger sample.
The current analyses control for sex, which may mask real variation in the data. Examination of the relationships between gray matter volume, BOLD signal, and behavioral data within subgroups of males and females was not possible due to the small size of the sample, although we found no significant statistical contribution of sex to the results reported here.
Our findings suggest that trauma exposure plays a causal role in changes to both brain structure and function, even in nonclinical adult populations. These relationships may differ depending on the amount of lifetime trauma exposure a person has experienced. Greater understanding of possible stress-related neural change has the potential to inform a number of disciplines, including the understanding of the alteration in brain function and structure associated with normal aging and a range of mental health disorders, as well current understanding of the socioemotional consequences of stress and trauma exposure. Research is needed to examine whether there is evidence for a similar pattern of neural change with the accumulation of other types of stressor exposure (e.g., poverty, divorce, job loss) and whether these effects are the same or different in children and the elderly. This may, in turn, aid in designing interventions to help keep environmental risk from manifesting as decreased health and well-being across the lifespan.