Voxelwise meta-analysis of gray matter reduction in major depressive disorder

Abstract

Background

Voxel-based morphometry (VBM) has been widely used in studies of major depressive disorder (MDD) and has provided cumulative evidence of gray matter abnormalities in patients relative to controls. Thus we performed a meta-analysis to integrate the reported studies to determine the consistent gray matter alterations in MDD.

Methods

A systematic search was conducted to identify VBM studies which contrasted MDD patients against a comparison group. The coordinates of gray matter change across studies were meta-analyzed using the activation likelihood estimation (ALE) method hybridized with the rank-based Genome Scan Meta-Analysis (GSMA) to quantitatively estimate regional gray matter reductions in MDD.

Results

A total of 20 VBM studies comparing 543 major depressive patients with 750 healthy control subjects were included. Consistent gray matter reductions in all MDD patients relative to healthy controls were identified in the bilateral anterior cingulate cortex (ACC), right middle and inferior frontal gyrus, right hippocampus and left thalamus.

Conclusions

Meta-analysis of all primary VBM studies indicates that significant gray matter reductions in MDD are localized in a distributed neural network which includes frontal, limbic and thalamic regions. Future studies will benefit from the use of a longitudinal approach to examine anatomical and functional abnormalities within this network and their relationship to clinical profile, particularly in first-episode and drug-naive MDD patients.

Introduction

Structural neuroimaging techniques have been widely used to study brain alterations in major depressive disorder (MDD). The large majority of brain morphology studies in MDD have used a region of interest (ROI) approach in which investigators manually drew the brain regions of interest and then calculated and compared their volume in patients to healthy controls. Meta-analyses of these ROI studies have identified gray matter volume reductions in the amygdala (Hamilton et al., 2008), hippocampal gyrus (Campbell et al., 2004, McKinnon et al., 2009, Videbech and Ravnkilde, 2004), subgenual cingulate gyrus (Hajek et al., 2008) and frontal regions, particularly the anterior cingulate cortex (ACC) and the orbitofrontal cortex (Koolschijn et al., 2009). However, most of the studies applying ROI methods to examine gray matter changes in MDD focused on a relatively small number of specific brain regions (Goldstein et al., 1999), such as the hippocampus and the amygdala. This may have biased the conclusions of meta-analyses of ROI-based studies in favor of those regions which have been examined in most individual studies and, furthermore, may account for some of the inconsistent conclusions between meta-analyses.

VBM is a whole-brain and automatic technique which can be used to perform a voxel-wise comparison of the local concentration of gray matter between groups (Ashburner and Friston, 2000). It does not require the investigators to manually draw a priori regions of interest and is not biased towards one particular brain region; therefore, this technique overcomes the drawbacks of standard ROI-based methods. VBM has been widely used in a large number of studies of MDD to identify the neuroanatomical abnormalities associated with this disorder. Gray matter alterations have been reported in the orbitofrontal cortex (Egger et al., 2008, Scheuerecker et al., 2010, Wagner et al., 2008), temporal gyrus (Mak et al., 2009, Scheuerecker et al., 2010), anterior cingulate gyrus (Leung et al., 2009, Tang et al., 2007, Treadway et al., 2009), thalamus (Kim et al., 2008, Vasic et al., 2008) and hippocampus (Vasic et al., 2008, Wagner et al., 2010, Zou et al., 2010). However, there have been a number of non-replications and contrasting results across studies. For example, gray matter reduction in the cingulate cortex in patients with MDD relative to healthy controls have been reported in a number of studies (Tang et al., 2007, Treadway et al., 2009, Vasic et al., 2008), but opposite effect was also reported (Yuan et al., 2008). Similarly, two studies reported decreased gray matter in the hippocampus (Wagner et al., 2010, Zou et al., 2010), but another two studies reported no group difference in this region (Tang et al., 2007, Treadway et al., 2009). As is known to us, there was considerable variation in terms of the demographic characteristics of patients, the used imaging protocols and clinical factors in these published studies. For instance, several studies only included female MDD participants, which could limit the generalizability of their findings (Kim et al., 2008, Leung et al., 2009, Mak et al., 2009, Wagner et al., 2008). Additionally, some published MDD studies had used relatively small sample sizes which may result into false positive or false negative findings (Scheuerecker et al., 2010, Tang et al., 2007, Vasic et al., 2008). These variations may account for the inconsistent findings.

Given the apparent inconsistencies among previous reports of gray matter abnormalities in MDD, in the present study, we conducted a meta-analysis of the published VBM studies of MDD to identify the consistent regional gray matter abnormalities in MDD.