Posttraumatic Stress Disorder and Depression Symptom Severities Are Differentially Associated With Hippocampal Subfield Volume Loss in Combat Veterans

Background Two decades of human neuroimaging research have associated volume reductions in the hippocampus with posttraumatic stress disorder. However, little is known about the distribution of volume loss across hippocampal subfields. Recent advances in neuroimaging methods have made it possible to accurately delineate 10 gray matter hippocampal subfields. Here, we apply a volumetric analysis of hippocampal subfields to data from a group of combat-exposed Veterans. Method Veterans (total, n = 68, posttraumatic stress disorder, n = 36; combat control, n = 32) completed high-resolution structural magnetic resonance imaging. Based on previously validated methods, hippocampal subfield volume measurements were conducted using FreeSurfer 6.0. The Clinician-Administered PTSD Scale assessed posttraumatic stress disorder symptom severity; Beck Depression Inventory assessed depressive symptom severity. Controlling for age and intracranial volume, partial correlation analysis examined the relationship between hippocampal subfields and symptom severity. Correction for multiple comparisons was performed using false discovery rate. Gender, intelligence, combat severity, comorbid anxiety, alcohol/substance use disorder, and medication status were investigated as potential confounds. Results In the whole sample, total hippocampal volume negatively correlated with Clinician-Administered PTSD Scale and Beck Depression Inventory scores. Of the 10 hippocampal subfields, Clinician-Administered PTSD Scale symptom severity negatively correlated with the hippocampus–amygdala transition area (HATA). Beck Depression Inventory scores negatively correlated with dentate gyrus, cornu ammonis 4 (CA4), HATA, CA2/3, molecular layer, and CA1. Follow-up analysis limited to the posttraumatic stress disorder group showed a negative correlation between Clinician-Administered PTSD Scale symptom severity and each of HATA, CA2/3, molecular layer, and CA4. Conclusion This study provides the first evidence relating posttraumatic stress disorder and depression symptoms to abnormalities in the HATA, an anterior hippocampal region highly connected to prefrontal-amygdala circuitry. Notably, dentate gyrus abnormalities were associated with depression severity but not posttraumatic stress disorder symptoms. Future confirmatory studies should determine the extent to which dentate gyrus volume can differentiate between posttraumatic stress disorder- and depression-related pathophysiology.

FreeSurfer, which is available for download at http://surfer.nmr.mgh.harvard.edu, was used (standard recon-all pipeline) to process three structural MRI for each subject, reconstructing the cortical surface and segmenting cortical and subcortical volumes. The FreeSurfer automated pipeline is well validated and has been described in the literature. Briefly, recon-all automatically performed normalization of image intensity, S1 motion correction, and averaging S2 of two T1-weighted MPRAGE and skull stripping. S3 Following Talairach transformation, FreeSurfer segments the gray matter (GM), subcortical white matter (WM), and deep gray matter volumetric structures. S4, S5 The GM/WM boundary is then tessellated and overall topology is corrected. S6, S7 To complete individual cortical modeling, the software optimizes the placement of GM/WM and GM/CSF boundaries based on shifts in intensity gradients. S8-S10 FreeSurfer measurements have been validated against histology and manual measurements. S10-S13 These methods use information from the entire brain volume, such as spatial intensity gradients across tissue classes. As mentioned in the manuscript, 10 GM subfields were evaluated based on enhanced subfield segmentation in FreeSurfer 6.0. S14, S15

Effects of Age and PTSD Symptom Severity on Subfield Volumes
Given that previous studies have reported evidence of age effect on hippocampal volumes, S16 age was included as covariate in all primary analyses. However, to assess whether an interaction between age and PTSD severity existed, we constructed a general linear model (GLM) examining the age x PTSD symptom severity interaction. We found significant CAPS-by-age interaction in the subiculum (F=5.50, p=0.02) and the parasubiculum (F=4.71. p=0.03). No other interactive effects were found. In addition, age did not correlate with any hippocampal subfield, controlling for intracranial volume. The results of these analyses are reported below in columns 1 & 2 of Table S4.

PTSD vs Non-PTSD Group Differences
This study was optimized to conduct an analysis of PTSD symptom severity on a continuum, not to create a contrast between PTSD and non-PTSD groups. However, to inform future studies, we conducted a between group multivariate analysis removing the effect of age and estimated total intracranial volume (eTIV), and using log-transformed volumes as required for normal distribution, consistent with the methods reported in the primary manuscript. No significant differences were found between PTSD and Non-PTSD groups, but the results and estimated marginal means (by group) are reported in columns 3-5 of Table  S4 below.
For completeness, we also repeated the post-hoc analyses in the Non-PTSD group, following the exploratory methods described in the manuscript for the PTSD-only group. As described in the manuscript, this post-hoc data was not corrected for multiple comparisons. No significant correlations were identified. The results are reported below in Table S5.     Abbreviations: r, partial correlation controlling for age and total intracranial volume; CA, cornu ammonis; ML, molecular layer; DG, granule cell and molecular layers of the dentate gyrus; HATA, hippocampal-amygdala transition area; CAPS, Clinician Administered PTSD Scale for DSM-IV. Correlations were conducting controlling for the effects of age and total intracranial volume. * indicates significance if p is less than or equal to 0.05 (p ≤ 0.05). Exploratory analysis-not corrected for multiple comparisons. For Full Group and PTSD-Only Group results, see Table 2 in the primary manuscript.

Video Legends:
Video S1. 10 Gray Matter Subfields of the Hippocampus -Animated 3D Model. This video was created in Blender using meshes from one of the combat control (non-PTSD) participants described in the manuscript. It demonstrates the gray matter subfields of the hippocampal formation, and steps through the anatomy to better visualize the more hidden subfields.