 |
 |
 |
|
|
The hippocampus has been implicated in the neurobiology of posttraumatic stress disorder (PTSD). Magnetic resonance imaging (MRI) studies report reduced hippocampal volumes in this condition (1–4), but see also (5). These volumetric changes have been interpreted as hippocampal atrophy (1), but the issue remains controversial (6). Proton magnetic resonance spectroscopy (1H-MRS) provides more information about neuronal viability than do volumetric studies alone. 1H-MRS allows the detection in vivo of N-acetyl-aspartate (NAA), choline-containing compounds (Cho), and creatine (Cr), among other neurometabolites (7). NAA is present primarily in neurons (8) and is considered a neuronal marker. Cr and Cho are involved in energy and membrane metabolism, respectively. 1H-MRS studies in PTSD report decreased hippocampal NAA (5,9,10), consistent with decreased neuronal density. A limitation of these studies, however, has been the use of neurometabolite ratios rather than absolute concentrations. Recently, highly reproducible methods have been developed to determine absolute neurometabolite concentrations from 1H-MRS (11,12). This study investigated the specificity of hippocampal NAA changes in PTSD by also investigating the occipital white matter (OWM) as a control region. Patients And Methods Patients We recruited participants at the University of New Mexico Health Sciences Center. All participants gave informed consent by signing an Institutional Review Board–approved form. Patients met criteria for PTSD, established with the Structured Clinical Interview for Axis I Diagnoses, patient version (SCID-P for DSM-IV) (13), and had a score of 60 or higher on the Clinician-Administered PTSD Scale (CAPS) (14). Healthy comparison subjects were currently free of any major Axis I diagnosis on the SCID-NP (nonpatient version). Exclusion criteria for both groups included major medical or psychiatric diagnoses, alcohol or substance dependence, alcohol or substance abuse in the previous year, a history of head trauma with loss of consciousness, seizures, or a neurological disorder. We matched comparison subjects by age, sex, race, years of education, and handedness. We estimated total weeks of lifetime alcohol use and alcohol intoxication . The Beck Depression Inventory (BDI) and Beck Anxiety Inventory (BAI) were also administered. MRI and 1H-MRS Subjects underwent quantitative MRI and 1H-MRS of the brain at the Clinical and Magnetic Resonance Research Center, University of New Mexico. The facility’s routine methodology was employed (15). Spectroscopic and imaging experiments were performed at 1.5 tesla using a standard clinical MR scanner, head coil, and software (Signa 5.4, GE Medical Systems, Waukesha [WI]). The imaging protocol included a T1-weighted volume coronal series oriented to the long axis of the hippocampus (fast-SPGR, TE = 6.9 ms, TR = 17 ms, flip = 25 , 256 x 192 matrix, 1.5 mm contiguous slices). 1H-MRS was used to examine 15.3 ´ 20.3 ´ 30 mm voxels in both hippocampi with point resolved spectroscopy (PRESS) (TE = 40 ms, TR = 2000 ms, 128 averages). The hippocampus was identified from the T1-weighted images in the coronal plane (see Figure 1). The anterior border of the voxel was located in the most anterior slice that showed hippocampus but not amygdala and was extended posteriorly. Two other voxels in the right and left OWM, were acquired using stimulated echo acquistion mode (STEAM) (20.3 ´ 20.3 ´ 21 mm, TE = 30 ms, TR = 2000 ms, 128 averages), consistent with previous studies (15). Occipital voxels were used as a control region to determine whether there were nonhippocampal brain changes. The peaks of spectra were identified from well-known resonance positions determined in previous studies. A rater blind to the subject’s diagnosis determined areas of peaks from NAA, Cr, and Cho, using Magnetic Resonance User Interface (MRUI, Katholieke Universiteit, Leuven, Belgium) (see Figure 2). The percentage of tissue within each spectroscopic voxel was obtained by segmenting the T1-weighted fast-SPGR images, using automated K-means segmentation described previously (16). Metabolite concentrations were corrected for the percentage of tissue in the voxel. Statistical Analysis We compared variables between groups with unpaired, 2-tailed t-tests. Effect sizes (ES) were also calculated with the following formula ES = t• p1 / n1 + 1 / n2 Figure 1 Coronal MRI scan showing the
typical location of the 15 x 20 x 30 mm voxel
Results We evaluated 8 patients with PTSD and 5 comparison subjects (see Table 1). One of the control subjects had been diagnosed with premenstrual dysphoric disorder and was receiving sertraline. Two control subjects (40%) had a history of major depression but were in remission at the time of evaluation. The groups did not differ in age (mean 43.35 years, SD 7.6 vs mean 44.2 years, DS 7.7; t = 0.19, P = 0.85), years of education (mean 15.8, SD 4.5 vs mean 15.4, SD 5; t = 0.15, P = 0.88) or lifetime weeks of alcohol intoxication (mean 12.7, SD 14.9 vs mean 11.6, SD 13.2; t = 0.15, P = 0.88). As expected, subjects with PTSD had higher scores on the BDI (mean 21.3, SD 12.7 vs mean 0.6, SD 1.3; t = 4.58, P < 0.01) and BAI (mean 21.7, SD 10 vs mean 0.6, SD 0.9; t = 5.83, P < 0.001). A trend toward reduced left hippocampal NAA was found in PTSD patients, compared with control subjects (P = 0.054), with a large ES of 1.49 (17). This finding indicates that more subjects in our sample would confirm the result. No difference was found for right hippocampal NAA or bilateral hippocampal Cho. There was a trend toward reduced left hippocampal Cre (P = 0.08) in the PTSD group. Occipital Cre was lower bilaterally in subjects with PTSD (P < 0.05). (See Table 2.) Discussion This preliminary report has some limitations that we would like to acknowledge. First, the small sample size has limited power to detect differences. Second, 55% of the subjects with PTSD were receiving psychotropic medications. Third, 50% of the subjects with PTSD had current major depression—a condition with hippocampal abnormalities (reviewed in [18]). We recruited 2 control subjects with a history of depression and 1 control subject with premenstrual dysphoric disorder, but we did not match control subjects for severity or chronicity of depressive symptoms or medication use. Our preliminary results show a trend (and a large ES) toward reduced hippocampal NAA concentrations in civilian patients with PTSD. This suggests that a larger sample size would confirm the difference. This finding is consistent with prior reports of decreased hippocampal NAA ratios in PTSD (5,9,10). However, in contrast to the present study, previous studies used neurometabolite ratios rather than unambiguous absolute neurometabolite concentrations. Neurometabolite ratios lead to ambiguity regarding which metabolite is changing—numerator or denominator. To our knowledge, this is the first 1H-MRS report studying civilians with PTSD, and also the first to use absolute neurometabolite concentrations in this population. Our finding of reduced hippocampal NAA replicates previous 1H-MRS studies in PTSD (5,9,10) and is consistent with decreased hippocampal neuronal integrity. The origin of hippocampal neuronal changes in PTSD remains controversial: it is not clear whether they predate trauma and PTSD (6). One hypothesis proposes the existence of hippocampal neurotoxicity secondary to trauma and PTSD (1). As expected, we found no differences in occipital NAA. This suggests that there are no generalized NAA changes in PTSD. The findings of decreased occipital Cre are puzzling, since Cre concentrations are thought to be very constant (7). One could speculate that Cre differences reflect changes in energy metabolism; however, a more plausible explanation is a type I error (detecting a difference when there is none). These findings need replication in a larger sample of PTSD subjects using multiple voxels. Figure 2 Typical hippocampal
spectra cho-choline, cre-creatine,
|
||||||
