Journal of Affective Disorders 109 (2008) 107–116www.elsevier.com/locate/jad

Research report

Gray matter reduction associated with psychopathology andcognitive dysfunction in unipolar depression:

A voxel-based morphometry study

Nenad Vasic a,⁎,1, Henrik Walter b,2, Annett Höse a,1, Robert Christian Wolf a,1

a Department of Psychiatry III, University of Ulm, Leimgrubenweg 12-14, 89075 Ulm, Germanyb Department of Psychiatry, Division of Medical Psychology, Friedrich Wilhelms University, Bonn, Germany

Received 15 July 2007; received in revised form 26 November 2007; accepted 26 November 2007Available online 9 January 2008

Abstract

Background: Functional neuroimaging studies on both cognitive processing and psychopathology in patients with major depressionhave reported several functionally aberrant brain areas within limbic-cortical circuits. However, less is known about the relation-ship between psychopathology, cognitive deficits and regional volume alterations in this patient population.Methods: By means of voxel-based morphometry (VBM) and a standardized neuropsychological test battery, we examined 15patients meeting DSM-IV criteria for major depression disorder and 14 healthy controls in order to investigate the relationshipbetween affective symptoms, cognitive deficits and structural abnormalities.Results: Patients with depression showed reduced gray matter concentration (GMC) in the left inferior temporal cortex (BA 20), theright orbitofrontal (BA 11) and the dorsolateral prefrontal cortex (BA 46). Reduced gray matter volume (GMV) was found in theleft hippocampal gyrus, the cingulate gyrus (BA 24/32) and the thalamus. Structure-cognition correlation analyses revealed thatdecreased GMC of the right medial and inferior frontal gyrus was associated with both depressive psychopathology and worseexecutive performance as measured by the Wisconsin Card Sorting Test (WCST). Furthermore, depressive psychopathology andworse performance during the WCST were associated with decreased GMVof the hippocampus. Decreased GMVof the cingulatecortex was associated with worse executive performance.Limitations: Moderate illness severity, medication effects, and the relatively small patient sample size should be taken into considerationwhen reviewing the implications of these results.Conclusions: The volumetric results indicate that regional abnormalities in gray matter volume and concentration may be associatedwith both psychopathological changes and cognitive deficits in depression.© 2007 Elsevier B.V. All rights reserved.

Keywords: Depression; Voxel-based morphometry; Gray matter; Executive dysfunction; Prefrontal cortex

⁎ Corresponding author. University of Ulm, Department ofPsychiatry III, Leimgrubenweg 12-14, 89075 Ulm, Germany.Tel.: +49 731 50061568; fax: +49 731 50061412.

E-mail address: nenad.vasic@uni-ulm.de (N. Vasic).1 Head: Prof. M. Spitzer, MD, PhD.2 Head: Prof. W. Maier, MD.

0165-0327/$ - see front matter © 2007 Elsevier B.V. All rights reserved.doi:10.1016/j.jad.2007.11.011

1. Introduction

Major depressive disorder (MDD) is associated withcerebral dysfunction of several distinct areas within alimbic-cortical network (Vasic et al., 2005). Functional

108 N. Vasic et al. / Journal of Affective Disorders 109 (2008) 107–116

neuroimaging techniques, including positron emissiontomography (PET) and functional magnetic resonanceimaging (fMRI), have identified a number of function-ally aberrant cortical and subcortical areas in patientswith depression during paradigms examining bothemotional processes and cognitive functions such asattention, working memory and executive processing(Mayberg, 2003). Working memory and executive defi-cits in MDD have been repeatedly associated with adysfunction of prefrontal cortical regions (Vasic et al.,2007). On the other hand, the affective component ofdepressive psychopathology may be primarily related tofunctional alterations in other brain regions, such as theanterior cingulate cortex (ACC) and the basal ganglia(Drevets, 2000; Mayberg et al., 2005).

Despite increasing evidence for a role of functionalbrain impairment in the etiology of MDD, the relation-ship between selective regional brain volume differ-ences and depressive symptoms remains unclear.Previous morphometric MRI studies using region-of-interest (ROI) analyses have revealed numerous struc-tural differences within subgroups of depressive patientscompared to healthy controls, e.g. in the orbitofrontalcortex (Ballmaier et al., 2004; Lacerda et al., 2004), theACC (Ballmaier et al., 2004; Caetano et al., 2006),(Ballmaier et al., 2004), the temporal cortex (Shah et al.,1998) and the hippocampus (Shah et al., 1998; Shelineet al., 2003). ROI analyses are inherently user-dependent (Wolkin et al., 1998), however, focusingonly on specific hypothesized brain regions, whilepotentially overlooking unpredicted or less specifiedareas (Goldstein et al., 1999). Consequently, voxel-based morphometry [VBM] (Ashburner and Friston,2000) is being increasingly used as a viable alternativemethodology for detecting structural abnormalities inpatients with neuropsychiatric disorders includingschizophrenia (Antonova et al., 2005; Yamada et al.,2007) and bipolar disorder (Nugent et al., 2006; Adleret al., 2007).

VBM is a user-independent, fully automated methodof analysis which allows for unbiased exploration ofbrain structures without a priori specification of ROIs,and can thus identify potentially unsuspected brainstructure abnormalities. At present, however, VBMstudies in patients with MDD are sparse (Shah et al.,1998; Campbell and MacQueen, 2006; Chen et al.,2007). Only one study has addressed the associationbetween structural alterations and psychopathology(Bell-McGinty et al., 2002), and no VBM study inpatients with MDD has investigated the relationshipbetween cortical volume alterations and performanceacross multiple cognitive domains.

In this study, we used VBM to examine the relation-ship between GM abnormalities, psychopathology andcognitive performance in patients with MDD aged 30to 45 years. We applied standard tasks that requiredivided attention, verbal and spatial working memory, aswell as executive processing and behavioral inhibition(Walter et al., 2007). We hypothesized that GMabnormalities in MDD detected by VBM would involveareas functionally subserving working memory andexecutive processing, cognitive domains which exhibitthe most prominent behavioral deficits. Additionally, wepredicted that the severity of depressive symptoms inpatients with MDD would correlate with GM abnorm-alities in areas of the frontal cortex known to be involvedin both affective and cognitive processing.

2. Materials and methods

2.1. Subjects

15 right-handed subjects with MDD (6 females) wererecruited from among the inpatients treated at theDepartment of Psychiatry III at the University of Ulm.All patients were diagnosed according to DSM-IVcriteria, excluding subjects with concurrent axis Idisorders. In addition to a detailed interview conductedby an experienced clinical psychiatrist (A.H.), case noteswere reviewed to corroborate a definitive diagnosis.Psychopathology was rated by means of the BriefPsychiatric Rating Scale (BPRS), the 21-item HamiltonDepression Scale (HAMD-21), Montgomery-AsbergDepression Rating Scale (MADRS) and the ClinicalGlobal Impression Scale (CGI) (see also Table 1). All ofthe patients were treated with antidepressants: 7 patientswere treated with citalopram alone (20–40 mg/d), 2 weretreatedwith venlafaxine alone (150 and 300mg/d), 2 weremedicated with citalopram (20–40mg/d) and mirtazapine(30 mg/d), and 4 were medicated with a monotherapy ofmirtazapine (30 mg/d), reboxetine (4 mg/d), fluoxetine(30 mg/d) and tranylcypromine (40 mg/d). None of thepatients were receiving a stable regime of benzodiaze-pines at the time of the neuropsychological testing. Allpatients were assessed within their first day/s of admis-sion, i.e. during a symptomatic phase before or shortlyafter changing a previous antidepressant drug regime.

The healthy control group consisted of 14 right-handedhealthy subjects (6 females)matched for age, handedness,education, and fluid intelligence, as measured by asubroutine from the ‘Leistungspruefsystem’ (Horn,1983). Subjects with a history of neurological orpsychiatric disorder or a family history ofmood disorders,substance abuse or dependence were excluded. The study

Table 1Demographic and clinical characteristics of patients with depression and control subjects; results of the neuropsychological assessment

MDD patients (n=15) Healthy controls (n=14) Analysis

Characteristic Mean SD Mean SD Statistic, p-value

Age (years) 37.4 8.5 31.4 9.6 F(1, 27)=3,2264, p=0.08Laterality scorea 90.7 17.7 84.8 18.1 F(1, 27)=0,80149, p=0.38Education (years) 11.9 3.1 11.5 1.6 F(1, 27)=0,15580, p=0.70Duration of illness (months) 43.4 37.3 n.a. n.a.HAMD score 16.9 5.6 n.a. n.a.MADRS score 23.3 4.0 n.a. n.a.BDI 20.8 8.6 0.6 1.6 F(1, 27)=75,346, p=0.0001CGI 5.2 1.8 n.a. n.a.LPS 3 (pts) 28.9 4.5 31.1 3.0 F(1, 27)=2,3603, p=0.14Tonic alertness (ms) 233.4 21.3 234.0 35.9 F(1, 27)= ,00349, p=0.95Phasic alertness (ms) 257.1 41.5 246.3 28.7 F(1, 27)= ,65417, p=0.43Divided attention (ms) 708.4 43.3 630.3 45.0 F(1, 27)=22,704, p=0.001Divided attention, omissions 2.6 1.5 1.1 1.4 F(1, 27)=7,5373, p=0.01Digit span, forward condition 8.1 1.2 10.1 2.1 F(1, 27)=10,092, p=0.004Digit span, backward condition 5.1 1.6 7.9 1.8 F(1, 27)=20,567, p=0.001Spatial span, forward condition 6.1 1.7 8.3 2.4 F(1, 27)=7,9045, p=0.009Spatial span, backward condition 5.0 1.2 7.6 2.0 F(1, 27)=18,156, p=0.0001WCST, perseverative errors 3.1 4.6 0.4 0.7 F(1, 27)=4,8775, p=0.04WCST, switch costs (sec) 2.9 2.4 1.1 0.8 F(1, 27)=7,5520, p=0.01Stroop test, reaction time (ms) 94.8 103.8 68.6 61.1 F(1, 27)= ,67275, p=0.42Stroop test, errors 12.1 6.2 2.2 1.9 F(1, 27)=32,649, p=0.001

MDD: major depressive disorder; HAMD: Hamilton Depression Rating Scale; MADRS: Montgomery-Asberg Depression Rating Scale; BDI: BeckDepression Inventory; CGI: Clinical Global Impression Scale; WCST: Wisconsin Card Sorting Test; LPS3: Leistungspruefsystem. Red: cognitivedomains in which patients with depression showed a significantly worse cognitive task performance compared with healthy controls (results of thebetween-group ANOVA, pb0.05). See text for a detailed description of the cognitive tasks, the statistical analysis and significance levels. n.a.indicates not applicable.aAs rated by the Edinburgh handedness questionnaire.

109N. Vasic et al. / Journal of Affective Disorders 109 (2008) 107–116

was approved by the local Institutional Ethics Committee.After complete description of the study to the subjects,written informed consent was obtained.

2.2. Neuropsychological tests

A neuropsychological test battery on alertness,divided attention, verbal and spatial working memory,executive function and behavioral inhibition wasadministered to each subject. Tonic and phasic alertness(tAL/pAL), as well as divided attention (DA) weremeasured using computerized tasks from a standardizedtest battery (Zimmermann and Fimm, 1993). During thetAL test, a gray cross was presented at random timeintervals on a black screen, prompting the subject topress a button as quickly as possible with the right indexfinger. During the pAL test, a short auditory signalpreceded the appearance of a visually presentedstimulus, and subjects were instructed to press a buttonimmediately after the presentation of the visual input (agray cross, identical to the stimulus presented during thetAL test). The DA test required the simultaneousprocessing of two concurrent tasks: subjects were

instructed to focus on a randomly presented 4×4 dotmatrix presented on a screen, as well as to alternatinghigh and low frequency pitches. Targets consisted eitherof 2×2 squares, or of two pitches identical in frequency.A button press with the right index finger was requiredonce a target had been identified. Verbal and spatialworking memory was assessed using the digit and spatialspan tests (Tewes, 1994; Schelling, 1997). During thedigit span test, a sequence of 2–12 digits was presented,and the subjects were instructed to repeat the sequence inthe same (forward condition) or in the reverse order(backward condition). Spatial span was measured usingthe Corsi block (Milner, 1971). This test comprises anarray of nine blocks; at the beginning of the test, twoblocks were tapped by the experimenter, and the subjectswere instructed to imitate the sequence in the same(forward condition) or in the reverse order (backwardcondition). The sequence length during the digit andspatial span tests was increased until performance brokedown. Executive function was measured using acomputerized version of the Wisconsin Card SortingTest [WCST] (Nelson, 1976). This WCST variantconsisted of 48 cards and a maximum of 5 category

Table 2Regions of grey matter reduction in patients with depression comparedto healthy controls

Anatomical region X y z Z

GMC Right inferior frontal gyrus (BA 44) 48 12 32 4.74Left inferior temporal gyrus (BA 20) −64 −24 −18 4.52Right medial frontal gyrus (BA 11) 24 42 −22 4.28Right medial/inferior frontal gyrus(BA 46)

50 46 4 4.04

Right transverse temporal gyrus(BA 41)

48 −18 14 3.99

Left angular gyrus (BA 39) −52 −74 30 3.75Left inferior frontal gyrus (BA 44) −54 10 30 3.53

GMV Thalamus 0 −24 10 4.47Left hippocampal gyrus (BA 28) −26 4 −22 4.29Cingulate gyrus (BA 24/32) 0 8 42 3.77

Results of the 2nd level ANCOVA, pb0.05 small volume corrected.GMC: grey matter concentration, GMV: grey matter volume, BA:Brodmann Area.

110 N. Vasic et al. / Journal of Affective Disorders 109 (2008) 107–116

switches. Inhibition was tested by a computerizedversion of the Stroop Word-Color Interference Test(Alvarez and Emory, 2006) based on randomized singletrials (20 trials per color and condition).

2.3. MRI data acquisition and preprocessing

MRI data were acquired using a 1.5 T MagnetomVISION whole body MRI system (Siemens, Erlangen,Germany) equipped with a standard head volumecoil. The MRI parameters of the three-dimensionalmagnetization-prepared rapid gradient-echo (3D-MPRAGE) sequences were as follows: TE=4.0 ms;TR=9.7 ms; TI=100 ms; FOV=256; slice plane=ax-ial; slice thickness=1 mm; resolution=1.0×1.0×1.0;number of slices=170. The data were processed usingStatistical Parametric Mapping 5 (SPM5) software(The Wellcome Department of Imaging Neuroscience,London, U.K.) with Matlab 7.3.0 (The Math Works,Natick, MA, U.S.A.).

The images were analyzed using the optimized VBMmethods, as previously described in detail by Ashburnerand Friston (2000) and Good et al., (2001). We used theVBM tools written by C. Gaser (http://dbm.neuro.uni-jena.de/vbm), an extension of the SPM5 algorithms. Inbrief, a study-specific whole brain template and GMprior images were created. Using the customizedtemplate and the priors, each participant's originalimage was spatially normalized and segmented intoGM, white matter, and cerebrospinal fluid (CSF), andeventually resliced with 1.0×1.0×1.0 mm voxels. Thisprocedure yielded ‘unmodulated’ and ‘modulated’ GMimages. All images were smoothed with a Gaussiankernel of 8 mm full width at half maximum. In thisstudy, we analyzed both modulated and unmodulateddata, since these parameters yield different but com-plementary information about the investigated braintissue. The ‘traditional’ VBM analysis compares theproportion of gray matter in each voxel, and does notaccount for the change in voxel size. The ‘optimized’VBM analysis includes an algorithm that modulateseach voxel with Jacobian determinants derived from thespatial normalization, thus allowing a comparison of theabsolute volume of each voxel (Good et al., 2001).Thus, unmodulated images were used for the groupcomparison of GM concentration (GMC), i.e. forcomparisons of a proportion of a tissue type relative toother tissue types within a region. Modulated imageswere used for the group comparison of GM volumedifferences (GMV), i.e. for comparisons of an absoluteamount of a tissue type within a region (Ashburner andFriston, 2000).

2.4. Data analysis

2.4.1. Behavioral data analysisPerformance measures were recorded as follows:

1. alertness: mean reaction times (RT in ms) during tALand pAL, as well as the number of omitted targets; 2. DA:mean reaction times (RT in ms) of correctly identifiedtargets and the number of omitted targets; 3. digit span,forward and backward condition: number of correctlyretrieved items; 4. spatial span, forward and backwardcondition: number of correctly retrieved items;5. WCST: number of perseverative errors and adjustedswitch costs (given in s) following (Spitzer et al., 2001);6. Stroop test: mean RT (in ms) and error differencesbetween the incongruent and congruent condition.A one-way between-group analysis of variance(ANOVA) was conducted between healthy controlsand patients with depression on each measure (pb0.05).

2.4.2. Between-group comparisons of regional graymatter reduction

To identify the brain regions of GMV/GMC reduc-tion in patients with depression relative to the healthycontrols, analyses of covariance (ANCOVA) wereperformed using SPM5. Age, gender and the globalGM volumes were included as nuisance covariates.These analyses yielded statistical parametric maps basedon a voxel-level height threshold of pb0.001 (uncor-rected). Small volume correction (SVC) was applied inorder to further protect against type I error. SVC wasperformed by centring a sphere of 9 mm as a volume-of-interest-(VOI) on all regions described in the Resultssection, i.e. on regions showing significant differencesbetween patients and healthy controls. Clusters found

Fig. 1. Upper panel: Gray matter concentration reductions in patients with depression. The adjusted gray matter concentration in the MOPFC andDLPFC is plotted against the degree of depression severity, as measured by the MADRS. Lower panel: Gray matter concentration reductions inpatients with depression. The adjusted gray matter concentration in the MOPFC and the ITC is plotted against the WCST task performance, asmeasured by the adjusted switch costs. Results of the 2nd level ANCOVA, pb0.05 small volume corrected. WCST: Wisconsin Card Sorting Test,MADRS: Montgomery-Asberg Depression Rating Scale, BA: Brodmann Area.

111N. Vasic et al. / Journal of Affective Disorders 109 (2008) 107–116

112 N. Vasic et al. / Journal of Affective Disorders 109 (2008) 107–116

within a volume of interest had to meet pb0.05 (clustercorrected) to be considered significant. All anatomicalregions and denominations are reported according to theatlases of Talairach and Tournoux (1988) and Duvernoy(1999). Coordinates are maxima in a given clusteraccording to the standard MNI template.

2.4.3. Relationship between gray matter structure,cognition and psychopathology

Each patient's GMV/GMC values were extracted foreach cluster of volume/concentration difference ob-tained by the procedures mentioned above. Using the‘volume of interest’ [VOI] function (spm_regions.m) inSPM5, this procedure yielded the predicted GMV/GMCvalues (y adjusted) for the percentage of total graymatter volume in regions of GMV/GMC reduction. Allmeasures of psychopathology and cognition were testedfor Gaussian distribution within the groups. Correlationmatrices were used to investigate the relationshipbetween the patients’ GMV/GMC in regions of reduc-tion and those psychopathological scores which showednormal distribution. Spearmans rank correlations testswere applied on the parameters which were not normallydistributed (neuropsychological scores), controlling forage. All analyses were performed with each psycho-pathological/neuropsychological parameter separatelyavoiding the effect of multiple testing. The correlativerelationship was considered to be significant at pb0.05.

Fig. 2. Gray matter volume reductions in patients with depression. The adjusWCST task performance, as measured by the adjusted switch costs, and theadjusted gray matter volume in the cingulate cortex is plotted against the WCSvolume corrected. WCST: Wisconsin Card Sorting Test, MADRS: Montgom

All correlations were performed using the Statisticasoftware (Version 6.0, StatSoft).

3. Results

3.1. Behavioral results

Patients performed significantly worse in tests ofdivided attention, verbal working memory (forward andbackward digit span) and spatial working memory(forward and backward spatial span). Worse perfor-mance during card sorting was characterized bysignificantly increased perseverative errors and higherswitch costs. No significant between-group differenceswere found for variables measuring tonic and phasicalertness and behavioral inhibition (Table 1).

3.2. Regional gray matter reductions in patientscompared to control subjects

Compared to healthy controls, the patients withdepression showed reduced GMC in the bilateralinferior frontal gyrus (Brodmann Area [BA] 44), theleft inferior temporal gyrus (BA 20), the rightorbitofrontal cortex (BA 11), the right medial/inferiorfrontal gyrus (BA 46), and the right transverse temporalgyrus (BA 41). Reduced GMVwas found in the bilateralthalamus, the left hippocampal gyrus (BA 28), and the

ted gray matter volume in the hippocampal gyrus is plotted against thedegree of depression severity, as measured by the MADRS score. TheT task performance. Results of the 2nd level ANCOVA, pb0.05 smallery-Asberg Depression Rating Scale, BA: Brodmann Area.

113N. Vasic et al. / Journal of Affective Disorders 109 (2008) 107–116

cingulate gyrus (BA 24/32). The inverse contrasts[patientsNcontrols] did not yield any differences inGMV and GMC at the chosen threshold. Post-hocShapiro–Wilks W-tests showed normal distribution ofthe structural data (pN0.05) (Table 2).

3.3. Relationship between brain structure, psychopa-thology and cognition

Significant correlations were found betweenMADRSscores and reduced GMC in the right orbitofrontal cortex(BA 11; r=−0.57) and the right dorsolateral prefrontalcortex (DLPFC, BA 46; r=−0.53); see also Fig. 1, upperpanel. Reduction of GMV in the left hippocampal gyrus(BA 28; r=−0.54) and in the left cerebellum (r=−0.59)was also correlated with the MADRS score; see alsoFig. 2.

An increasing number of omitted targets during theDA test was associated with decreased GMC in the rightinferior prefrontal cortex (BA 44; r=−0.63). ImpairedWCST performance, as assessed by the duration of theswitch costs during category switches, was associatedwith decreased GMC in the left inferior temporal cortex(BA 20; r=−0.60) and the right orbitofrontal cortex(BA 11; r=−0.63); see also Fig. 1, lower panel.Reduced GMC in the right dorsolateral prefrontal cortex(BA 46) was associated with the number of persevera-tive errors during WCST (r=−0.55). Furthermore,the duration of the switch costs during WCST taskperformance was negatively correlated with GMV

Table 3Correlations (adjusted for age) between grey matter abnormalities,cognitive function and symptoms of major depression

Clinical score/variable

Anatomical region Correlation(r)

MADRS Right medial frontal gyrus(BA 11)

−0.57

Right inferior/medial frontalgyrus (BA 46)

−0.53

Left hippocampal gyrus (BA 28) −0.54WCST, switch

costsLeft inferior temporal gyrus(BA 20)

−0.60

Right medial frontal gyrus (BA 11) −0.63Cingulate gyrus (BA 24/32) −0.59Left hippocampal gyrus (BA 28) −0.65

WCST,perseverations

Right inferior/medial frontal gyrus(BA 46)

−0.55

Divided attention,omissions

Right inferior frontal gyrus (BA 44) −0.63

Duration of illness Right inferior frontal gyrus (BA 44) −0.53

See the Results section for further details. MADRS: Montgomery-Asberg Depression Rating Scale, WCST: Wisconsin Card Sorting Test,BA: Brodmann Area.

reduction in the cingulate gyrus (BA 24/32; r=−0.59)and the hippocampal gyrus (BA 28; r=−0.65); see alsoFig. 2. No significant correlations between structuralabnormalities and measures of alertness, workingmemory, and inhibition were found (Table 3).

4. Discussion

In this study, we used VBM to investigate regionalgray matter (GM) alterations in medicated patients withmild to moderate MDD. We were primarily interested inthe relationship between GM abnormalities, psycho-pathology and cognitive impairment. In patients,decreased gray matter concentration (GMC) of theright medial and inferior frontal gyrus was associatedwith both psychopathology and worse executiveperformance as measured by the WCST. Decreasedgray matter volume (GMV) of the hippocampal gyrusand the cingulate gyrus predicted worse performanceduring the WCST. Decreased volume of the hippocam-pal gyrus was further associated with depressivepsychopathology.

In line with previous VBM findings in elder patientpopulations exhibiting symptoms of major depression(Bell-McGinty et al., 2002; Ballmaier et al., 2004;Lavretsky et al., 2004; Taki et al., 2005), our resultsargue that 3–7 years after the onset of the first episode ofmajor depression, patients with unipolar depressionpresent structural alterations in orbitofrontal, dorsolat-eral prefrontal, temporal and cingulate areas. Withregard to cognition, MDD-patients showed worse taskperformance across a wide range of cognitive tasksincluding divided attention, verbal and spatial workingmemory and executive function. Also in accordancewith previous neuropsychological and functional ima-ging findings, executive dysfunction seems to beparticularly prominent in patients with MDD (Zakzaniset al., 1998; Vasic et al., 2007). Furthermore, anassociation was found between structural differencesin patients with depression and variables measuringexecutive skills: GM abnormalities in the temporalcortex, the orbitofrontal cortex, the cingulate gyrus andthe hippocampal gyrus were associated with impairedperformance during the WCST. An overlap betweenexecutive deficits, psychopathology and GM abnorm-alities was found in the dorsolateral prefrontal cortex,the orbitofrontal cortex, and in hippocampal regions,suggesting that these areas might account for thepresence of both clinical symptoms and cognitiveimpairment. In contrast, GM alterations in the temporaland the cingulate cortex were not significantly related toclinical symptom measures, possibly indicating a neural

114 N. Vasic et al. / Journal of Affective Disorders 109 (2008) 107–116

dissociation between depressive psychopathology andcognitive dysfunction.

Our data showing a reduction of GM volume in thecingulate gyrus are in good accordance with the findingsof anatomical MRI studies which have focused on thisregion (Ballmaier et al., 2004; Caetano et al., 2006). Inaddition to volumetric MRI studies, post-mortem tissueexaminations indicate structural changes in the anteriorcingulate cortex showing a reduction in glial cell densityand neuronal size in patients with MDD. These findingsmay represent neuropathological changes specific toaffective symptoms as they do not appear in patientswith schizophrenia or bipolar disorder (Cotter et al.,2001). A recently published study combining structuraland functional MRI showed that faster rates of symptomimprovement in subjects with depression during drugtreatment were strongly associated with the extent ofGMV in the anterior cingulate cortex. The authorshypothesized that structural MRI measurements ofanterior cingulate cortex could potentially provide auseful predictor of antidepressant treatment response(Chen et al., 2007).

Also GM abnormalities in the orbitofrontal (BA 11)and the dorsolateral prefrontal (BA 46) cortex aresupported by structural MRI findings (Ballmaier et al.,2004; Lacerda et al., 2004) and post-mortem studies. Forinstance, VBM findings have shown significantlysmaller volume of the bilateral superior frontal gyrusin elderly subjects with symptoms of major depression(Taki et al., 2005). Post-mortem brain tissue examina-tion has also revealed distinct reductions in glial celldensity and neuronal size in the deeper cortical layers ofthe dorsolateral prefrontal cortex (BA 9) in patients withMDD compared to patients with schizophrenia, bipolardisorder, and healthy controls (Cotter et al., 2002).Moreover, orbitofrontal, lateral prefrontal and cingulateregions are known to exhibit metabolic and functionalchanges in patients with MDD (Vasic et al., 2005),suggesting that these frontal circuits might at least partlyaccount for depressive psychopathology and the man-ifestation of cognitive deficits in MDD-patients.

Similar to our findings, reduced hippocampal volumewas reported in both unmedicated (Saylam et al., 2006)and in drug-free remitted MDD-patients (Neumeisteret al., 2005). Moreover, our results confirm previousfindings showing that reduced hippocampal volumecorrelates with executive dysfunction inMDD as assessedby theWCST (Frodl et al., 2006). The lack of a correlationbetween hippocampal volume reduction and illnessduration in our data might be due to the relatively shortduration of illness in our patient sample (mean=3.5 years).Nevertheless, the significance of hippocampal volume

abnormalities in patients with MDD currently remainscontroversial. Smaller bilateral hippocampal volume, forinstance, has been attributed to occur in clinically de-pressed rather than in remitted patients, suggesting thatsmaller hippocampal volume might be a characteristic ofthe depressive state rather than of symptom remission(Caetano et al., 2004). Also structural alterations withinthe medial temporal cortex, both in terms of reducedconcentration (Shah et al., 1998) and reduced volume(Caetano et al., 2004), have been repeatedly associatedwith a more disadvantageous course of the illness or otheroutcome criteria, such as treatment resistance (Shah et al.,1998), a longer duration of untreated depressive episodes(Sheline et al., 2003), or the total duration of the illness(Sheline et al., 1999; Bell-McGinty et al., 2002; Caetanoet al., 2004).

Although our results are consistent with currentlyavailable functional neuroimaging and neuropsycholo-gical data (Zakzanis et al., 1998; Rose et al., 2006), thegeneralizability of the observed structural alterations inour study is limited by the patient sample size andpotential neuroprotective effects of antidepressantmedication. For instance, antidepressant treatment hasbeen demonstrated to increase brain-derived neuro-trophic factor (BDNF) levels in both animal models andhumans, possibly attenuating neuronal damage andneuronal death rate caused by corticosteroids (Hayneset al., 2004) or apoptosis (Kosten et al., 2007). Althoughour VBM findings in the hippocampus and the cingulatecortex are consistent with previous studies on medica-tion-naive patients (Neumeister et al., 2005; Saylamet al., 2006; Tang et al., 2007), we cannot rule out thepossibility of a beneficial ‘neuroprotective’ effect ofantidepressant therapy on cortical volume in our patientsample. Nevertheless, we sought to minimize additionalconfounds arising from aging processes, substanceabuse, and a longer total duration of drug treatment byincluding a carefully selected, relatively young patientsample without psychiatric comorbidity.

Another caveat with regard to our findings is that weobtained different volumetric results for GM concentra-tion and GM volume. Previous VBM studies in neu-ropsychiatric patient samples have reported differences inthe distribution of regional GM/WM concentration andvolume (Lochhead et al., 2004; Yamada et al., 2007), andit is still unclear why this occurs. Modulated and un-modulated data are considered to detect different, butcomplementary aspects of brain structure (Good et al.,2001). In particular, the optimized VBM approachincludes a step that modulates each voxel with Jacobiandeterminants derived from the spatial normalizationprocedure (Good et al., 2001), thus including information

115N. Vasic et al. / Journal of Affective Disorders 109 (2008) 107–116

about brain deformations. This algorithm could increasethe sensitivity for observing structural changes that mightbe lost during normalization without adjusting fordifferences in regional brain volume (Eckert et al.,2006). However, the reason why one analysis yielded asignificant difference in certain regions while the other didnot clearly necessitates further investigation. Furthermore,it remains unclear if the structural alterations observed inour study may represent state or trait characteristics of thispatient sample. At present, longitudinal data on brainvolume abnormalities in younger populations withdepression are lacking, but might shed further light onthe development of volume changes in MDD over time.This approach could help focus the research on a smallernumber of particularly significant structures, which couldallow formore sophisticated diagnostics of persons at risk,identification of prodromal features prior to first onset ofdepression and detailed monitoring of the effects oftherapy during the course of illness.

In conclusion, our results provide evidence forregionally-specific alterations of gray matter in MDD-patients. Gray matter abnormalities in frontal, cingulateand hippocampal regions predicted impaired cognitivefunction in the patients with depression, indicating thatselective gray matter alterations might at least partlyaccount for manifest cognitive deficits observed in thisdisorder. Moreover, we demonstrated an overlapbetween key regions involved in MDD, psychopathol-ogy and cognitive deficits, suggesting that both symp-tom domains are associated with volume alterations inmutual areas of the frontal cortex. However, the timecourse of these abnormalities has yet to be elucidated byfurther research.

Role of funding sourceFunding for this study was partly financially supported in a form of

grant support by Sanofi-Synthélabo (Grant-No.: D 1218); Sanofi-Synthélabo had no further role in study design; in the collection,analysis and interpretation of data; in the writing of the report; and inthe decision to submit the paper for publication.

Conflict of interestALL authors declare that they have no actual or potential conflict

of interest including any financial, personal or other relationships withother people or organizations within three (3) years of beginning thework submitted that could inappropriately influence, or be perceived toinfluence, their work.

Acknowledgements

This studywas partly supported by a grant fromSanofi-Synthélabo (Grant-No.: D 1218).We thankKatrinBrändlefor her technical assistance, Georg Grön for his critical

discussions and Timothy Laumann, Genes, Cognition andPsychosis Program, NIMH, Bethesda, for his insightfulcomments on a previous version of this manuscript.

References

Adler, C.M., Delbello, M.P., Jarvis, K., Levine, A., Adams, J.,Strakowski, S.M., 2007. Voxel-based study of structural changes infirst-episode patients with bipolar disorder. Biol. Psychiatry 61,776–781.

Alvarez, J.A., Emory, E., 2006. Executive function and the frontallobes: a meta-analytic review. Neuropsychol. Rev. 16, 17–42.

Antonova, E., Kumari, V., Morris, R., Halari, R., Anilkumar, A.,Mehrotra, R., Sharma, T., 2005. The relationship of structuralalterations to cognitive deficits in schizophrenia: a voxel-basedmorphometry study. Biol. Psychiatry 58, 457–467.

Ashburner, J., Friston, K.J., 2000. Voxel-based morphometry – themethods. Neuroimage 11, 805–821.

Ballmaier, M., Toga, A.W., Blanton, R.E., Sowell, E.R., Lavretsky, H.,Peterson, J., Pham, D., Kumar, A., 2004. Anterior cingulate, gyrusrectus, and orbitofrontal abnormalities in elderly depressedpatients: an MRI-based parcellation of the prefrontal cortex. Am.J. Psychiatry 161, 99–108.

Bell-McGinty, S., Butters, M.A., Meltzer, C.C., Greer, P.J., Reynolds3rd, C.F., Becker, J.T., 2002. Brain morphometric abnormalities ingeriatric depression: long-term neurobiological effects of illnessduration. Am. J. Psychiatry 159, 1424–1427.

Caetano, S.C., Hatch, J.P., Brambilla, P., Sassi, R.B., Nicoletti, M.,Mallinger, A.G., Frank, E., Kupfer, D.J., Keshavan, M.S., Soares,J.C., 2004. Anatomical MRI study of hippocampus and amygdalain patients with current and remitted major depression. PsychiatryRes. 132, 141–147.

Caetano, S.C., Kaur, S., Brambilla, P., Nicoletti, M., Hatch, J.P., Sassi,R.B., Mallinger, A.G., Keshavan, M.S., Kupfer, D.J., Frank, E.,Soares, J.C., 2006. Smaller cingulate volumes in unipolardepressed patients. Biol. Psychiatry 59, 702–706.

Campbell, S., MacQueen, G., 2006. An update on regional brainvolume differences associated with mood disorders. Curr. Opin.Psychiatry 19, 25–33.

Chen, C.H., Ridler, K., Suckling, J., Williams, S., Fu, C.H., Merlo-Pich, E., Bullmore, E., 2007. Brain imaging correlates ofdepressive symptom severity and predictors of symptom improve-ment after antidepressant treatment. Biol. Psychiatry 62, 407–414.

Cotter, D., Mackay, D., Landau, S., Kerwin, R., Everall, I., 2001.Reduced glial cell density and neuronal size in the anteriorcingulate cortex in major depressive disorder. Arch. Gen.Psychiatry 58, 545–553.

Cotter, D.,Mackay, D., Chana, G., Beasley, C., Landau, S., Everall, I.P.,2002. Reduced neuronal size and glial cell density in area 9 of thedorsolateral prefrontal cortex in subjects with major depressivedisorder. Cereb. Cortex 12, 386–394.

Drevets, W.C., 2000. Functional anatomical abnormalities in limbicand prefrontal cortical structures in major depression. Prog. Brain.Res. 126, 413–431.

Duvernoy, H.M. (1999). The human brain. Wien, New York, Springer.Eckert, M.A., Tenforde, A., Galaburda, A.M., Bellugi, U., Korenberg,

J.R., Mills, D., Reiss, A.L., 2006. To modulate or not to modulate:differing results in uniquely shaped Williams syndrome brains.Neuroimage 32, 1001–1007.

Frodl, T., Schaub, A., Banac, S., Charypar, M., Jager, M., Kummler, P.,Bottlender, R., Zetzsche, T., Born, C., Leinsinger, G., Reiser, M.,

116 N. Vasic et al. / Journal of Affective Disorders 109 (2008) 107–116

Moller, H.J., Meisenzahl, E.M., 2006. Reduced hippocampalvolume correlates with executive dysfunctioning in major depres-sion. J. Psychiatry Neurosci. 31, 316–323.

Goldstein, J.M., Goodman, J.M., Seidman, L.J., Kennedy, D.N.,Makris, N., Lee, H., Tourville, J., Caviness Jr., V.S., Faraone, S.V.,Tsuang, M.T., 1999. Cortical abnormalities in schizophreniaidentified by structural magnetic resonance imaging. Arch. Gen.Psychiatry 56, 537–547.

Good, C.D., Ashburner, J., Frackowiak, R.S., 2001. Computationalneuroanatomy: new perspectives for neuroradiology. Rev. Neurol.(Paris) 157, 797–806.

Haynes, L.E., Barber, D., Mitchell, I.J., 2004. Chronic antidepressantmedication attenuates dexamethasone-induced neuronal death andsublethal neuronal damage in the hippocampus and striatum.Brain. Res. 1026, 157–167.

Horn, W., 1983. Leistungspruefsystem (LPS). Handanweisung.Hogrefe, Göttingen.

Kosten, T.A., Galloway, M.P., Duman, R.S., Russell, D.S., D'Sa, C.,2007. Repeated unpredictable stress and antidepressants differen-tially regulate expression of the Bcl-2 family of apoptotic genesin rat cortical, hippocampal, and limbic brain structures. Neuro-psychopharmacology 33, 1545–1558.

Lacerda, A.L., Keshavan, M.S., Hardan, A.Y., Yorbik, O., Brambilla, P.,Sassi, R.B., Nicoletti, M., Mallinger, A.G., Frank, E., Kupfer, D.J.,Soares, J.C., 2004. Anatomic evaluation of the orbitofrontal cortex inmajor depressive disorder. Biol. Psychiatry 55, 353–358.

Lavretsky, H., Kurbanyan, K., Ballmaier, M., Mintz, J., Toga, A.,Kumar, A., 2004. Sex differences in brain structure in geriatricdepression. Am. J. Geriatr. Psychiatry 12, 653–657.

Lochhead, R.A., Parsey, R.V., Oquendo, M.A., Mann, J.J., 2004.Regional brain gray matter volume differences in patients withbipolar disorder as assessed by optimized voxel-based morpho-metry. Biol. Psychiatry 55, 1154–1162.

Mayberg, H.S., 2003. Modulating dysfunctional limbic-corticalcircuits in depression: towards development of brain-basedalgorithms for diagnosis and optimised treatment. Br. Med. Bull.65, 193–207.

Mayberg, H.S., Lozano, A.M., Voon, V., McNeely, H.E., Seminowicz,D., Hamani, C., Schwalb, J.M., Kennedy, S.H., 2005. Deep brainstimulation for treatment-resistant depression. Neuron 45, 651–660.

Milner, B., 1971. Interhemispheric differences in the localization ofpsychological processes in man. Br. Med. Bull. 27, 272–277.

Nelson, H., 1976. A modified card sorting test sensitive to frontal lobedeficits. Cortex 313–324.

Neumeister, A.,Wood, S., Bonne, O., Nugent, A.C., Luckenbaugh, D.A.,Young, T., Bain, E.E., Charney, D.S., Drevets, W.C., 2005. Reducedhippocampal volume in unmedicated, remitted patients with majordepression versus control subjects. Biol. Psychiatry 57, 935–937.

Nugent, A.C., Milham, M.P., Bain, E.E., Mah, L., Cannon, D.M.,Marrett, S., Zarate, C.A., Pine, D.S., Price, J.L., Drevets, W.C.,2006. Cortical abnormalities in bipolar disorder investigated withMRI and voxel-based morphometry. Neuroimage 30, 485–497.

Rose, E.J., Simonotto, E., Ebmeier, K.P., 2006. Limbic over-activity indepression during preserved performance on the n-back task.NeuroImage 29, 203–215.

Saylam, C., Ucerler, H., Kitis, O., Ozand, E., Gonul, A.S., 2006.Reduced hippocampal volume in drug-free depressed patients.Surg. Radiol. Anat. 28, 82–87.

Schelling, D., 1997. Block-Tapping-Test. Swets Test Services GmbH,Frankfurt.

Shah, P.J., Ebmeier, K.P., Glabus, M.F., Goodwin, G.M., 1998.Cortical grey matter reductions associated with treatment-resistantchronic unipolar depression. Controlled magnetic resonanceimaging study. Br. J. Psychiatry 172, 527–532.

Sheline, Y.I., Sanghavi, M., Mintun, M.A., Gado, M.H., 1999.Depression duration but not age predicts hippocampal volumeloss in medically healthy women with recurrent major depression.J. Neurosci. 19, 5034–5043.

Sheline, Y.I., Gado, M.H., Kraemer, H.C., 2003. Untreated depressionand hippocampal volume loss. Am. J. Psychiatry 160, 1516–1518.

Spitzer, M., Franke, B., Walter, H., Buechler, J., Wunderlich, A.P.,Schwab, M., Kovar, K.A., Hermle, L., Gron, G., 2001. Enantio-selective cognitive and brain activation effects of N-ethyl-3,4-methylenedioxyamphetamine in humans. Neuropharmacology 41,263–271.

Taki, Y., Kinomura, S., Awata, S., Inoue, K., Sato, K., Ito, H., Goto, R.,Uchida, S., Tsuji, I., Arai, H., Kawashima, R., Fukuda, H., 2005.Male elderly subthreshold depression patients have smaller volumeof medial part of prefrontal cortex and precentral gyrus comparedwith age-matched normal subjects: a voxel-based morphometry.J. Affect. Disord. 88, 313–320.

Talairach, J., Tournoux, P., 1988. Co-Planar Stereotaxic Atlas of theHuman Brain. Thieme, New York.

Tang, Y.,Wang, F., Xie, G., Liu, J., Li, L., Su, L., Liu, Y., Hu, X., He, Z.,Blumberg, H.P., 2007. Reduced ventral anterior cingulate andamygdala volumes in medication-naive females with majordepressive disorder: A voxel-based morphometric magneticresonance imaging study. Psychiatry Res. 156, 83–86.

Tewes, U., 1994. [Hamburg-Wechsler Intelligenztest für Erwachsene –Revision]. Hans Huber, Bern, Deutschland.

Vasic, N., Wolf, R.C., Walter, H., 2005. Neurofunctional mechanismsof unipolar depressive disorder. Nervenheilkunde 24, 603–610.

Vasic, N., Wolf, R.C., Walter, H., 2007. [Executive functions inpatients with depression. The role of prefrontal activation].Nervenarzt 78, 628–640.

Walter, H., Wolf, R.C., Spitzer, M., Vasic, N., 2007. Increased leftprefrontal activation in patients with unipolar depression: an event-related, parametric, performance-controlled fMRI study. J. Affect.Disord. 101, 175–185.

Wolkin, A., Rusinek, H., Vaid, G., Arena, L., Lafargue, T., Sanfilipo,M.,Loneragan, C., Lautin, A., Rotrosen, J., 1998. Structural magneticresonance image averaging in schizophrenia. Am. J. Psychiatry 155,1064–1073.

Yamada, M., Hirao, K., Namiki, C., Hanakawa, T., Fukuyama, H.,Hayashi, T., Murai, T., 2007. Social cognition and frontal lobepathology in schizophrenia: a voxel-based morphometric study.Neuroimage 35, 292–298.

Zakzanis, K.K., Leach, L., Kaplan, E., 1998. On the nature and patternof neurocognitive function in major depressive disorder. Neurop-sychiatry. Neuropsychol. Behav. Neurol. 11, 111–119.

Zimmermann, P., Fimm, B., 1993. [Testbatterie zur Aufmerksamkeit-sprüfung (TAP)]. Breisgau, Psytest, Freiburg.