10.7272/1TRB-WK91
Cardenas, Valerie
Samuelson, Kristin
Lenoci, Maryanne A.
Studholme, Colin
Neylan, Thomas C.
Marmar, Charles R.
Schuff, Norbert
Weiner, Michael W.
Changes in brain anatomy during the course of PTSD
University of California, San Francisco
2012
Adult
Male
Human
Magnetic Resonance Imaging
Brain
Deformation Morphometry
Longitudinal
Neuropsychological Testing
Stress Disorders, Post-Traumatic
Veterans
Vietnamese Conflict, 1961-1975
UCSF Center for Imaging of Neurodegenerative Diseases
1998/2005
2013-08-22T17:43:26+00:00
application/octet-stream
21683556
UCSF Datashare Data Use Agreement
Longitudinal structural T1-weighted images from middle-aged controls and
veterans with PTSD (post-traumatic stress disorder). Most patients were
Vietnam veterans. The goal of this study was to determine whether PTSD was
associated with an increase in time-related decline in macrostructural
brain volume and whether these changes were associated with accelerated
cognitive decline. To quantify brain structure, 3 dimensional T1-weighted
MRI scans were performed at baseline and again after a minimum of 24
months in 25 patients with PTSD and 22 controls. Longitudinal changes in
brain volume were measured using deformation morphometry. For the group as
a whole PTSD+ patients did not show significant ongoing brain atrophy
compared to PTSD-. PTSD+ patients were then subgrouped into those with
decreasing or increasing symptoms. We found little evidence for brain
markers of accelerated atrophy in PTSD+ veterans whose symptoms improved
over time, with only a small left parietal region showing greater ongoing
tissue loss than PTSD-. PTSD patients whose symptoms increased over time
showed accelerated atrophy throughout the brain, particularly brainstem
and frontal and temporal lobes. Lastly, for the sample as a whole greater
rates of brain atrophy were associated with greater rates of decline in
verbal memory and delayed facial recognition.
Participants: After complete description of the study to the subjects,
written informed consent was obtained to a protocol approved by the review
boards of both the University of California, San Francisco (UCSF) and the
San Francisco Veterans Affairs (SFVA) Medical Center. Participants had
previously participated in one of two earlier studies examining
neuroimaging and neuropsychological correlates of PTSD (Neylan et al.,
2004; Samuelson et al., 2006; Schuff et al., 2008; Schuff et al., 2001).
When initially studied, participants had given consent to be re-contacted
about future studies. Veterans were contacted a minimum of two years after
completion of the first assessment. Study procedures included completing a
neuropsychological test battery and MRI scanning. Participants met the
following basic inclusion criteria at baseline: veterans 25-65 years of
age; PTSD+ participants had current PTSD attributable to a traumatic life
event (e.g., Vietnam or Gulf War combat, experiencing or witnessing
serious accidents, illnesses, sudden death, physical and sexual assault),
PTSDparticipants had no current, subthreshold, or lifetime history of
PTSD. Exclusion criteria at baseline were: diagnosis of drug dependence or
abuse within the past 6 months, current or lifetime history of any
psychiatric disorder with psychotic features, current or lifetime history
of bipolar disorder, history of neurologic or systemic illness affecting
CNS function, history of head injury with loss of consciousness exceeding
10 minutes, and history of head injury with any persistent post-injury
symptoms. Alcohol abuse and dependence were allowable diagnoses for one of
the original studies. Patients and controls were studied twice, once after
enrollment (baseline) and again after a minimum of 24 months. Because we
were interested in the natural course of PTSD, we did not apply
exclusionary criteria at follow-up. At follow-up, three PTSD+ participants
met one of the exclusionary baseline diagnoses—two exhibiting psychotic
symptoms and one exhibiting symptoms of bipolar disorder not otherwise
specified. These participants did not represent outliers in terms of
neuropsychological functioning and were included in our previous study of
longitudinal neuropsychological functioning (Samuelson et al., 2009), and
were also included in this longitudinal imaging study. This analysis
included 25 PTSD+ male veterans and 22 PTSDmale veterans that had complete
longitudinal MRI and neuropsychological datasets. Clinical and cognitive
testing: Diagnoses of PTSD at baseline and follow-up were made by a
clinical psychologist using the Clinician Administered PTSD Scale, which
determines if DSM-IV diagnostic criteria were met (CAPS; (Blake et al.,
1995)). Individuals with no trauma exposure received a CAPS score of 0.
The Structured Clinical Interview for DSM-IV Diagnosis (SCID; (First et
al., 1996)) was used to diagnose comorbid and exclusionary conditions.
Lifetime alcohol use was obtained using the Lifetime Drinking History
questionnaire (LDH; (Skinner and Sheu, 1982)). All participants were
administered a test battery of neuropsychological measures at both
timepoints, including the California Verbal Learning Test (CVLT; (Delis et
al., 1987)), Faces I, Faces II, Family Pictures I, Family Pictures II,
Digit Span and Spatial Span subtests of the Wechsler Memory Scale-Third
Edition (WMS-III; (Wechsler, 1987)). Samuelson and colleagues previously
reported on the longitudinal neuropsychological changes observed in this
dataset (Samuelson et al., 2009), and found a subtle decline in delayed
facial recognition as indexed by performance on the Faces II subtest. In
light of this finding and the previous reports of longitudinal changes on
the CVLT due to PTSD (Yehuda et al., 2006), this study focused only on the
relationships of longitudinal changes in Faces II, CVLT total (short term
verbal memory) and CVLT long delay (long term verbal memory) with
longitudinal measures of brain atrophy. Longitudinal change on these
neuropsychological tests was defined as (scoretp1-scoretp2)/(test interval
in yrs). MRI acquisition and processing: T1-weighted images were acquired
on a clinical 1.5 Tesla MR scanner (Vision, Siemens Medical Systems,
Iselin NJ) using Magnetization Prepared Rapid Acquisition Gradient Echo
(TR/TI/TE = 9/300/4 ms, 1x1 mm2 in-plane resolution, 1.5 mm slabs); images
were acquired orthogonal to the long axis of the hippocampus. Deformation
based morphometry (DBM) analyses: Robust fluid registration was used to
nonlinearly register baseline and follow-up scans of each participant to
create maps of longitudinal atrophy. Each participant's baseline
image was then registered to an atlas; transformations were combined to
create maps of longitudinal atrophy in common space as described in
(Cardenas et al., 2007). The longitudinal atrophy maps were normalized by
interscan interval and these maps of annualized atrophy rate in common
space were used in statistical analysis using linear models. ANCOVA at
each voxel was used to test our first hypothesis that PTSD was associated
with greater tissue atrophy rates; the maps of longitudinal change were
the dependent variable, group status (i.e., PTSD+ or PTSD-) was the
categorical predictor, and age was a covariate. Linear regressions with
baseline clinical or imaging measures as the independent variable were fit
at each voxel in order to identify baseline predictors of ongoing atrophy.
Subsequent analyses compared PTSD+ participants with improving symptoms
vs. PTSD-, and PTSD+ with worsening symptoms vs. PTSD-, in order to
determine the relationship between improving mental health and ongoing
brain atrophy. Using all participants, linear regression with change in
neuropsychological test score as the independent variable was also fit at
each voxel, covarying for age, in order to determine the relationship
between rate of tissue atrophy and cognitive decline. Statistical maps
were corrected for multiple comparisons by thresholding at uncorrected
p=0.005, identifying suprathreshold clusters, and using nonstationary
random field theory (Worsley et al., 2002) to identify clusters with
corrected p<0.05. Within each cluster with corrected p<0.05, the
estimated effects (i.e., the voxel-wise beta coefficients from the linear
model) were averaged to determine the magnitude of the group effects (for
ANCOVA models) or the slope (for regression models).