Post-traumatic stress disorder develops after exposure to a traumatic event, and sufferers experience characteristic symptoms including nightmares or flashbacks of the trauma, hyperarousal and insomnia, poor concentration, and difficulty with recalling explicit memories of the traumatic event. A neurocircuitry model of PTSD postulates a hyperresponsivity of the amygdala to threat-related stimuli, inadequate control of the amygdala by the medial prefrontal cortex and anterior cingulate, and hippocampal deficits that lead to memory difficulties and problems identifying safe contexts (Bremner 2002; Pitman et al 2001).
Previous brain imaging studies have focused on identifying PTSD-related abnormalities within these brain structures at a single point in time. Many, but not all, studies have reported hippocampal volume decreases (Bremner et al., 1995
; Fennema-Notestine et al., 2002
; Gurvits et al., 1996
). A meta-analysis (Karl et al., 2006
) concluded that hippocampal volumes are reduced in PTSD, although the size of group differences is accounted for in part by MRI methodology, PTSD symptom severity, medication, age, and gender. Brain volume decreases have also been reported in the anterior cingulate, subcallosal cortex, amygdala, and prefrontal cortex (Rauch et al., 2003
; Schuff et al., 2008
; Weniger et al., 2008
; Yamasue et al., 2003
Studies of the effects of PTSD on brain anatomy have usually been hypothesis driven, associating specific region-of-interest (ROI) brain volumes to PTSD or PTSD symptoms. Fewer investigators have used voxel-wise analysis methods, which are useful for examining local tissue properties (e.g., gray matter volume or density) without a priori hypotheses, as they compare tissue properties at every image location. Using voxel-wise analysis, brain structure abnormalities associated with PTSD have been reported in limbic and paralimbic cortices, anterior cingulate and insular cortex (Chen et al., 2006
; Corbo et al., 2005
; De Nardo et al., 2009
Most prior studies of PTSD have also focused on a single timepoint, so less is understood regarding the progression of PTSD over time. Early longitudinal imaging studies did not find evidence for accelerated hippocampal volume changes associated with PTSD over 6 months following trauma (Bonne et al., 2001
) or in pediatric maltreatment-related PTSD during puberty (De Bellis et al., 2001
). In a longitudinal study of elderly Holocaust survivors with PTSD, Yehuda and colleagues showed that decades after trauma exposure, an accelerated decline in paired associates learning was observed (Yehuda et al., 2006
). Samuelson and colleagues followed the longitudinal trajectory of neuropsychological functioning in the present study's sample of primarily Vietnam veterans, showing accelerated decline in delayed facial recognition in veterans with PTSD (Samuelson et al., 2009
). These results shed some light on the effect of chronic PTSD on ongoing cognition, but it is unknown whether these cognitive changes are caused by abnormal neuronal function, progressive brain tissue loss, or both.
The primary goal of this study was to determine whether chronic PTSD is associated with brain tissue loss over time using deformation morphometry of brain MRI scans. Deformation morphometry is a voxel-wise technique that is sensitive to detection of disease effects on localized brain anatomy, such as the insula, that might be obscured when looking at tissue volume over large ROsIs that include those smaller structures.
Our specific aim was to compare a group of PTSD+ veterans to a group of PTSD- veterans to determine the longitudinal effects of PTSD. Baseline cross-sectional comparisons of cognition and brain imaging have been previously reported, and the present longitudinal sample comprises a subset of our earlier studies. In summary, PTSD+ showed reduced baseline N-acetyl-aspartate in the hippocampus and anterior cingulate cortex, and deficits in verbal memory, attention, and processing speed (Samuelson et al., 2006
; Schuff et al., 2008
; Schuff et al., 2001
), in the absence of significant lobar or hippocampal atrophy. We hypothesized that ongoing brain atrophy underlies these changes in baseline hippocampal and anterior cingulate NAA. Moreover, a two fold increase in the risk for dementia in veterans with PTSD has recently been reported (Yaffe et al.). This suggests that cumulative longitudinal tissue loss may be occurring in our PTSD+ veterans, who were in their sixth decade, when acceleration of brain atrophy normally begins (Raz et al., 2005
). Taken together, we expected that a follow-up assessment would detect longitudinal atrophy in all participants, with greater longitudinal atrophy in PTSD. We quantified brain volumes from T1-weighted MRIs and tested the a priori hypothesis of greater rates of atrophy in PTSD+ patients compared to PTSD- controls. An additional aim was to identify baseline predictors of ongoing brain atrophy. We hypothesized that greater longitudinal atrophy would be associated with lower education, greater baseline age, higher baseline CAPS, increasing PTSD symptoms (i.e., greater CAPS change), higher levels of alcohol use, and larger baseline grey matter (GM) volumes. A final aim was to determine whether ongoing tissue loss was related to longitudinal decline in cognitive functioning, as indexed by change in neuropsychological test scores in all participants.