Meta-analysis result from 13 VBM studies: Thresholded Z map to match figure 2. Postive values are reduction in gray matter volume

Contributed by matthew.kempton on July 20, 2018

Collection: Meta-Analysis of 89 Structural MRI Studies in Posttraumatic Stress Disorder and Comparison With Major Depressive Disorder

Description: The following inclusion criteria were applied to the database of 113 studies: 1) gray matter VBM study comparing adult patients with PTSD to either non-traumatised-controls or traumatised-controls; 2) results presented in Talairach or MNI coordinates; 3) studies were only included if a whole brain analysis was performed rather than a small volume correction to ensure no bias in the regions reported. Thirteen studies met inclusion criteria and are listed in Table S1. We emailed all study authors who used SPM (Statistical Parametric Mapping) to process their data for a ‘T-map’ image comparing PTSD gray matter volume to the control group. ‘T-maps’ are three dimensional maps comprising statistical data of volume differences in thousands of voxels in the brain and provide far more detailed information than significant coordinates reported in studies. However, SDM allows both T-maps and coordinates to be combined in a single meta-analysis and the methodology reported in detail by Radua et al.8 We received 6 T-maps from 6 independent studies and these were included in the meta-analyses. In addition to the main meta-analysis comparing PTSD to all controls, three additional VBM analyses were conducted: 1) comparing the PTSD group with non-traumatised-controls only 2) comparing the PTSD group with traumatised-controls only, 3) comparing PTSD group with all controls and widening the criteria to include paediatric studies. T-maps and coordinates signifying gray matter volume changes from where we were unable to obtain T-maps were extracted from relevant studies and analysed using Seed-based d Mapping (SDM version 5.14, http://www.sdmproject.com). For studies where coordinate data was used, these were convolved with a Gaussian kernel (FWHM=20mm) in order to optimally compensate the sensitivity and specificity of the analysis. As is standard in SDM analyses, the number of randomizations were set to 100 and a threshold was set at p<0.005 as well as a cluster-level threshold of 10 voxels in order to increase sensitivity and correctly control false-positive rate.8 A jackknife sensitivity analysis was performed in order to assess the robustness of the results which was achieved by excluding one study in each of the analyses.

Tags: meta-analysis mri ptsd

Task View 3D View