Introduction
Functional imaging studies have implicated a wide range of brain regions in major depressive disorder (MDD), including but not limited to the lateral and medial frontal cortex, several other frontal regions, parts of the temporal lobe cortex, the insula and the amygdala (Drevets, Reference Drevets1998; Rogers et al. Reference Rogers, Kasai, Koji, Fukuda, Iwanami, Nakagome, Fukuda and Kato2004; Drevets et al. Reference Drevets, Price and Furey2008a ; Fitzgerald et al. Reference Fitzgerald, Laird, Maller and Daskalakis2008a ). One such region, the anterior cingulate cortex, has been a focus of particular interest over a sustained period. Originally, Drevets et al. (Reference Drevets, Videen, Price, Preskorn, Carmichael and Raichle1992) documented increased resting blood flow in depressed patients in an area which included parts of the ventrolateral prefrontal cortex and extended inwards to include part of the left medial prefrontal cortex. This abnormality was not seen in a separate group of patients who were recovered, suggesting that it was a state rather than a trait characteristic. Subsequently, Mayberg et al. (Reference Mayberg, Lozano, Voon, McNeely, Seminowicz, Hamani, Schwalb and Kennedy2005) argued that a similar but smaller region, the subgenual anterior cingulate cortex (sgACC), was pathologically overactive in MDD. This was based on a series of studies which found that increased resting activity in this cortical region characterized both normal sadness and clinical depression, and that different modalities of antidepressant treatment tended to reduce activation in the same region. The current use of the sgACC as a site for treatment of MDD with deep brain stimulation stems directly from these findings (Mayberg et al. Reference Mayberg, Lozano, Voon, McNeely, Seminowicz, Hamani, Schwalb and Kennedy2005; Lozano et al. Reference Lozano, Mayberg, Giacobbe, Hamani, Craddock and Kennedy2008, Reference Lozano, Giacobbe, Hamani, Rizvi, Kennedy, Kolivakis, Debonnel, Sadikot, Lam, Howard, Ilcewicz-Klimek, Honey and Mayberg2012).
Nevertheless, increased resting sgACC activation has not been a consistent finding in patients with MDD, with some studies instead finding reduced activation (see Rogers et al. Reference Rogers, Kasai, Koji, Fukuda, Iwanami, Nakagome, Fukuda and Kato2004; Drevets et al. Reference Drevets, Savitz and Trimble2008b ). Drevets et al. (Reference Drevets, Price, Simpson, Todd, Reich, Vannier and Raichle1997) argued that this contradiction might be explained by the presence of a corresponding reduction in volume in this area that they found to be present in groups of both bipolar and unipolar depressed patients. Further structural imaging studies, however, have once again proved inconclusive on this point. On the one hand, meta-analyses of studies using region-of-interest measurements have not supported the presence of volume reductions in the sgACC in MDD (Kempton et al. Reference Kempton, Salvador, Munafo, Geddes, Simmons, Frangou and Williams2011; Arnone et al. Reference Arnone, McIntosh, Ebmeier, Munafo and Anderson2012). On the other, two meta-analyses of studies using voxel-based morphometry (VBM), which avoids the necessity of preselecting regions of interest, identified a cluster of volume reduction in the anterior cingulate cortex. In one, this was adjacent to the genu of the corpus callosum, extending superiorly and to a minor extent inferiorly into the sgACC (Bora et al. Reference Bora, Fornito, Pantelis and Yucel2012), and in the other it was centred on the sgACC itself (Du et al. Reference Du, Wu, Yue, Li, Liao, Kuang, Huang, Chan, Mechelli and Gong2012).
The sgACC and adjacent parts of the orbitofrontal cortex have been argued to form part of a ‘medial prefrontal network’ which is closely connected to the amygdala, and which is implicated in the modulation of visceral function in response to sensory or emotional stimuli (Price & Drevets, Reference Price and Drevets2012). The medial frontal cortex is also an important component of the so-called default mode network (Gusnard & Raichle, Reference Gusnard and Raichle2001; Raichle et al. Reference Raichle, MacLeod, Snyder, Powers, Gusnard and Shulman2001), a series of brain regions that are highly active at rest but which deactivate during performance of attention-demanding tasks. As well as having putative functions related to introspective or self-directed thought, there is evidence that the default mode network is involved in emotion processing (Maddock, Reference Maddock1999). Default mode network dysfunction has been implicated in a number of psychiatric and neuropsychiatric disease states (Broyd et al. Reference Broyd, Demanuele, Debener, Helps, James and Sonuga-Barke2009), including MDD (Greicius et al. Reference Greicius, Flores, Menon, Glover, Solvason, Kenna, Reiss and Schatzberg2007; Frodl et al. Reference Frodl, Scheuerecker, Albrecht, Kleemann, Muller-Schunk, Koutsouleris, Moller, Bruckmann, Wiesmann and Meisenzahl2009; Grimm et al. Reference Grimm, Boesiger, Beck, Schuepbach, Bermpohl, Walter, Ernst, Hell, Boeker and Northoff2009; Sheline et al. Reference Sheline, Barch, Price, Rundle, Vaishnavi, Snyder, Mintun, Wang, Coalson and Raichle2009); see also Hamilton et al. (Reference Hamilton, Chen and Gotlib2013).
The aim in this study was to examine brain structural and functional change in the same group of patients with MDD, using whole-brain voxel-based techniques. We used VBM to measure structural change and investigated brain activations by means of functional magnetic resonance imaging (fMRI) during performance of a widely used working memory task, the n-back task. We also examined deactivations, since, as an attention-demanding cognitive task, n-back performance would also be expected to produce deactivation in the default mode network.
Method
Participants
Right-handed patients were recruited from in-patient and out-patient services of a psychiatric hospital and the private practice of one of the authors. They all met Diagnostic and Statistical Manual of Mental Disorders, fourth edition (DSM-IV) criteria for MDD, based on a detailed clinical interview and review of case-notes. They were excluded if: (a) they were younger than age 18 years or older than 70 years; (b) they had a history of brain trauma or neurological disease; (c) they had a history of mania or hypomania; (d) they showed psychotic features in the current episode; (e) they had shown alcohol or substance dependence within 12 months prior to participation; or (f) if they had undergone electroconvulsive therapy in the previous 24 months. The patients were also required to have an intelligence quotient (IQ) in the normal range (i.e. ⩾70) as estimated using the Word Accentuation Test (Test de Acentuación de Palabras, TAP; Del Ser et al. Reference Del Ser, Gonzalez-Montalvo, Martinez-Espinosa, Delgado-Villapalos and Bermejo1997; Gomar et al. Reference Gomar, Ortiz-Gil, McKenna, Salvador, Sans-Sansa, Sarro, Guerrero and Pomarol-Clotet 2011 ). This is a word reading test conceptually similar to the British National Adult Reading Test (Nelson & Willis, Reference Nelson and Willis1991) and the American Wide Range of Achievement Test (Wilkinson, Reference Wilkinson1993); it requires pronunciation of low-frequency words whose accents have been removed.
The patients were all in a depressive episode and had a 21-item Hamilton Depression Rating Scale score ⩾18 within 12 days prior to scanning. A total of 28 patients were on treatment with antidepressants [five on tricyclics, one on monoamine oxidase inhibitors (MAOIs); 11 on selective serotonin reuptake inhibitors (SSRIs)/other newer antidepressants; 11 on combinations of antidepressant]. One of the patients was taking a mood stabilizer (lamotrigine), six were on treatment with antipsychotics, two were not taking any treatment and one was on only a very small and subclinical dose of antidepressant. Medication data were missing for one patient. All treated patients were also on treatment with benzodiazepines for anxiety or insomnia.
Controls were drawn from a pool of subjects recruited via poster and web-based advertisement in the hospital and local community, plus word-of-mouth requests from staff in the research unit. All were questioned and excluded if they reported a history of mental illness and/or treatment with psychotropic medication. The controls were selected to be similar to the patients in terms of age, sex and TAP-estimated IQ. Because we anticipated that differences would be small in the structural imaging part of the study – two recent meta-analyses (Kempton et al. Reference Kempton, Salvador, Munafo, Geddes, Simmons, Frangou and Williams2011; Arnone et al. Reference Arnone, McIntosh, Ebmeier, Munafo and Anderson2012) found effect sizes of −0.06 and −0.07 for whole-brain volume and −0.10 and −0.04 for grey matter volume – the patients were matched 1:2 with healthy controls.
All participants gave written informed consent. The study was approved by the local research ethics committee.
Procedure
Subjects were scanned in the same 1.5 Tesla GE Signa scanner (General Electric Medical Systems, USA) located at the Sant Joan de Déu Hospital in Barcelona (Spain).
Structural imaging
High-resolution structural T1 MRI data were acquired with the following acquisition parameters: matrix size = 512 × 512; 180 contiguous axial slices; voxel resolution 0.47 × 0.47 × 1 mm3; echo time (TE) = 3.93 ms; repetition time (TR) = 2000 ms; inversion time = 710 ms; flip angle 15°. Analysis was carried out using FSL-VBM tools, an optimized VBM-style analysis (Good et al. Reference Good, Johnsrude, Ashburner, Henson, Friston and Frackowiak2001) included in the FSL software package (Smith et al. Reference Smith, Jenkinson, Woolrich, Beckmann, Behrens, Johansen-Berg, Bannister, De Luca, Drobnjak, Flitney, Niazy, Saunders, Vickers, Zhang, De Stefano, Brady and Matthews2004; http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/). In a first step, structural images were brain-extracted using BET (Brain Extraction Tool; Smith, Reference Smith2002). Next, tissue-type segmentation was carried out and the resulting grey matter partial volume images were aligned to the Montreal Neurological Institute (MNI) 152 standard space using the FSL linear (FLIRT) and non-linear (FNIRT) tools. The resulting images were averaged to create a study-specific template, to which the native grey matter images were non-linearly re-registered. These images were modulated (to correct for local expansion or contraction) by dividing by the Jacobian of the warp field, and smoothed with an isotropic Gaussian kernel with a sigma of 4 mm.
Group comparisons were carried out with permutation-based non-parametric tests. These were made with the randomise function implemented in FSL, using the recently developed threshold-free cluster enhancement method, for proper statistical inference of spatially distributed patterns (Smith & Nichols, Reference Smith and Nichols2009). All statistical tests in the analysis were corrected for multiple comparisons by means of family-wise error, p < 0.05.
Total intracranial volume was estimated using sienax (Smith et al. Reference Smith, De Stefano, Jenkinson and Matthews2001, Reference Smith, Zhang, Jenkinson, Chen, Matthews, Federico and De Stefano2002), part of FSL. Brain and skull images were extracted from the single whole-head input data (Smith, Reference Smith2002). The brain images were then affine-registered to MNI152 space (Jenkinson & Smith, Reference Jenkinson and Smith2001; Jenkinson et al. Reference Jenkinson, Bannister, Brady and Smith2002) in order to obtain the volumetric scaling factor. This factor was used to obtain a measure of the volume, scaling this ratio by the MNI total intracranial volume.
fMRI
The participants performed a sequential-letter version of the n-back task (Gevins & Cutillo, Reference Gevins and Cutillo1993) in the scanner (summarized in Fig. 1). Two levels of memory load (1-back and 2-back) were presented in a blocked-design manner. Each block consisted of 24 letters that were shown every 2 s (1 s on, 1 s off) and all blocks contained five repetitions (1-back or 2-back depending on the block) located randomly within the blocks. Individuals had to indicate repetitions by pressing a button. Four 1-back and four 2-back blocks were presented in an interleaved way, and between them a baseline stimulus (an asterisk flashing with the same frequency as the letters) was presented for 16 s. To identify which task had to be performed, characters were shown in green in 1-back blocks and in red in the 2-back blocks. All participants first went through a training session outside the scanner.
Fig. 1. Schematic of the n-back task.
The behavioural measure used was the signal detection theory index of sensitivity, d’ (Green & Swets, Reference Green and Swets1966). Higher values of d’ indicate better ability to discriminate between targets and distractors. Subjects who had negative d’ values in either or both of the 1-back or 2-back versions of the task, which suggests that they were not performing it, were excluded from the study.
In each individual scanning session 266 volumes were acquired. A gradient echo echo-planar imaging (EPI) sequence depicting the blood oxygen level-dependent (BOLD) contrast was used. Each volume contained 16 axial planes acquired with the following parameters: TR = 2000 ms; TE = 40 ms; flip angle = 70°; section thickness = 7 mm; section skip = 0.7 mm; in-plane resolution = 3 × 3 mm. The first 10 volumes were discarded to avoid T1 saturation effects.
Image analysis was performed with the fMRI Expert Analysis Tool (FEAT) module, included in FSL software (Smith et al. Reference Smith, Jenkinson, Woolrich, Beckmann, Behrens, Johansen-Berg, Bannister, De Luca, Drobnjak, Flitney, Niazy, Saunders, Vickers, Zhang, De Stefano, Brady and Matthews2004). At a first level, images were corrected for movement and then co-registered to a common stereotaxic space (MNI template). To minimize unwanted movement-related effects, individuals with an estimated maximum absolute movement >0.3 mm or an average absolute movement >0.3 mm were excluded from the study. General linear models were fitted to generate individual activation maps for the baseline versus 2-back contrast, the contrast delivering the most meaningful and interpretable results, in terms of activations and deactivations, in previous studies by our group (e.g. Pomarol-Clotet et al. Reference Pomarol-Clotet, Salvador, Sarro, Gomar, Vila, Martinez, Guerrero, Ortiz-Gil, Sans-Sansa, Capdevila, Cebamanos and McKenna2008, Reference Pomarol-Clotet, Moro, Sarro, Goikolea, Vieta, Amann, Fernandez-Corcuera, Sans-Sansa, Monte, Capdevila, McKenna and Salvador2012; Fernandez-Corcuera et al. Reference Fernandez-Corcuera, Salvador, Monte, Salvador Sarro, Goikolea, Amann, Moro, Sans-Sansa, Ortiz-Gil, Vieta, Maristany, McKenna and Pomarol-Clotet2013).
Group comparisons between patients and controls were performed within the FEAT software tool, with mixed-effects GLM models (Beckmann et al. Reference Beckmann, Jenkinson, Woolrich, Behrens, Flitney, Devlin and Smith2006). Z (Gaussianized T) statistical images were thresholded using clusters determined by Z >2.3 and a cluster significance threshold of p < 0.05, corrected for multiple comparisons across space (Worsley, Reference Worsley, Jezzard, Matthews and Smith2001).
Results
Demographic, clinical and neuropsychological data for the two groups are shown in Table 1. The groups were matched for sex and estimated pre-morbid IQ (TAP score).
Table 1. Demographic and clinical features of the patients and controls
IQ, Intelligence quotient; TAP, Word Accentuation Test (Test de Acentuación de Palabras); GAF, Global Assessment of Functioning; HAMD-21, Hamilton Depression Rating Scale, 21-item version; MADRS, Montgomery–Åsberg Depression Scale.
Data are given as mean (standard deviation).
Co-morbid Axis I disorders were present in seven of the patients prior to the onset of the current episode. These included: dysthymia (n = 1); panic disorder without agoraphobia (n = 2); anxiety disorder NOS (not otherwise specified; n = 1); adjustment disorder (n = 1); alcohol/sedative drug abuse (more than 2 years before scanning; n = 2).
Structural imaging findings
All 32 patients and 64 healthy controls participated in this part of the study. Total intracranial volume was 1313.90 (s.d. = 118.90) ml in the controls and 1338.26 (s.d. = 142.50) ml in the patients (p = 0.38).
VBM revealed that the patients showed a single large cluster of reduced volume compared with the controls. This affected the orbitofrontal cortex bilaterally, extending superiorly to the subcallosal medial frontal cortex, and marginally posteriorly to both insulas. It also involved the temporal poles bilaterally, extending to the inferior and middle temporal cortex on the right. On the left, the cluster reached the parahippocampal gyrus and hippocampus and marginally the amygdala and left fusiform gyrus (3943 voxels; MNI 52, −14, −44; Z score = 1.00; p = 0.004). The findings are shown in Fig. 2.
Fig. 2. Brain regions showing significant differences between the depressed patients and the controls in the voxel-based morphometry analysis. All clusters represent volume reduction in the patients. The right side of the images represents the right side of the brain.
There were no regions where the depressed patients showed increased volume compared with the controls.
Functional imaging findings
A total of 26 patients participated in this part of the study, including one patient whose structural scan could not be included because of movement. Reasons for exclusion for the remaining patients were excessive head movement (n = 4) and negative d’ values, suggesting that they were not performing the task (n = 2). These 26 patients were matched 1:2 with 52 of the healthy subjects from the original sample. The two groups remained matched for age [patients: 46.50 (s.d. = 13.29) years; controls: 46.25 (s.d. = 10.21) years; p = 0.92], sex (patients: 10 males and 16 females; controls: 20 males and 32 females; p = 1.0) and TAP-estimated IQ [patients: 99.48 (s.d. = 7.28); controls: 99.36 (s.d. = 10.27); p = 0.96].
Task performance
Using the d’ measure of performance (see Method), the depressed patients were significantly impaired compared with the controls on the 1-back version of the task (mean d’ = 3.76 (s.d. = 0.96) v. 4.23 (s.d. = 0.86), p = 0.03) and the 2-back version (mean d’ = 2.26 (s.d. = 0.88) v. 3.03 (s.d. = 1.01), p = 0.001).
Activations and deactivations in controls
Clusters of activation and deactivation for the 2-back versus baseline are shown in Fig. 3 and Table 2. There were clusters of significant activation in a network of regions including the anterior insula bilaterally, neighbouring regions of the frontal operculum extending dorsally to the precentral gyrus, and reaching both the left and right prefrontal cortex (dorsolateral prefrontal cortex; DLPFC) and the supplementary motor area. Other areas of activation were seen in the cerebellum, extending bilaterally to temporal and occipital regions, precuneus and parietal areas; as well as in the basal ganglia and the thalamus.
Fig. 3. Activations and deactivations in the 2-back versus baseline contrast in the controls. Yellow indicates a positive association (activation) with the task. Blue indicates areas where the task led to a decrease in the blood oxygen level-dependent (BOLD) response (deactivation). Numbers refer to Montreal Neurological Institute (MNI) z coordinates of the slice shown. The right side of each image represents the right side of the brain. Colour bars indicate Z scores from the group-level analysis.
Table 2. Significant activations and deactivations in the controls during n-back task performance (2-back versus baseline contrast)
DLPFC, Dorsolateral prefrontal cortex.
Clusters of significant deactivation were seen in an extensive area in the medial frontal cortex, and in the posterior cingulate cortex/precuneus and the angular gyrus, and also in the temporal poles bilaterally, the amygdala, the parahippocampus and hippocampus, the middle and superior temporal cortex and left posterior insula, and the right postcentral cortex.
Activations and deactivations in depressed patients
The pattern of activations and deactivations in the depressed patients was broadly similar to but less marked than that in the controls (see online Supplementary Fig. S1 and Supplementary Table S1 reporting peak activations).
Differences between patients and controls
Findings for the 2-back versus baseline contrast are shown in Fig. 4 a. The depressed patients showed significantly reduced activation in four clusters. One cluster was in the left DLPFC, extending to the precentral gyrus and the frontal operculum. This cluster also included the left and right thalamus and posterior part of the left caudate (2406 voxels; peak activation in MNI −12, −10, 12; Z score = 4.29; p < 0.001). Another cluster was in the precuneus, reaching the cuneus and bilateral superior parietal cortex (1731 voxels; peak activation in Brodmann area 7; MNI −4, −78, 60; Z score = 3.85; p < 0.001). Finally, there were two clusters situated in the cerebellum which also reached the left inferior occipital cortex (3614 voxels; peak activation in cerebellar vermis; MNI 4, −44, −20; Z score = 4.68; p < 0.001; and 1212 voxels; peak activation in left cerebellum; MNI −54, −66, −28; Z score = 3.84; p < 0.005).
Fig. 4. (a) Activation and deactivation differences between the depressed patients and the controls in the 2-back versus baseline contrast. Areas where the patients showed reduced activation compared with the controls are shown in yellow; areas where they showed failure of deactivation are shown in blue. The right side of each image represents the right side of the brain. (b) Scatterplot of mean activation values for the patients and controls in a region of interest based on the cluster of deactivation differences. 1, Patient taking antipsychotic; 2, patient taking mood stabilizer; 3, drug-free/subtherapeutic dose antidepressant-treated patient.
Fig. 4 a also shows that the depressed patients showed significantly reduced deactivation compared with the controls in a single cluster in the subcallosal medial frontal cortex and perigenual anterior cingulate cortex, extending laterally to include the gyrus rectus (1216 voxels; peak activation in Brodmann area 10; MNI −10, 46, −2; Z score = 3.86; p = 0.0048). Fig. 4 b shows the mean activations in the patients and controls in a region of interest based on this cluster. It confirms that the group differences reflected failure of deactivation (rather than being interpretable as differential hyperactivation in the two groups). Mean activation values in the patients who were taking antipsychotics or mood stabilizers are also indicated, as are those who were on no or subtherapeutic doses of antidepressants.
The relationship between the structural and functional imaging findings in the medial frontal cortex is shown in Fig. 5. It can be seen that both clusters were in close proximity, although they did not actually overlap; the cluster of failure of deactivation was more anteriorly placed than the cluster of volume reduction identified in the VBM analysis.
Fig. 5. Overlap of structural and functional imaging changes in the medial frontal cortex. Red represents volume reduction in voxel-based morphometry analysis; blue represents failure of deactivation during n-back performance.
Discussion
This study applying whole-brain, voxel-based structural and functional imaging to a group of patients with MDD found abnormality in both modalities affecting the sgACC, a region that has been argued to be crucial to the underlying biology of the disorder (Drevets et al. Reference Drevets, Price, Simpson, Todd, Reich, Vannier and Raichle1997, Reference Drevets, Savitz and Trimble2008b ; Mayberg et al. Reference Mayberg, Lozano, Voon, McNeely, Seminowicz, Hamani, Schwalb and Kennedy2005; Hamani et al. Reference Hamani, Mayberg, Stone, Laxton, Haber and Lozano2011). The fact that the functional imaging change took the form of failure of deactivation during cognitive task performance further suggests that MDD pathology in this area involves default mode network dysfunction.
Our structural imaging findings are broadly in keeping with those of the two recent meta-analyses of studies using conventional region-of-interest measurements: both of these identified the orbitofrontal cortex as an area of volume reduction (Kempton et al. Reference Kempton, Salvador, Munafo, Geddes, Simmons, Frangou and Williams2011; Arnone et al. Reference Arnone, McIntosh, Ebmeier, Munafo and Anderson2012), as in our study, and one of them also found reduced volume in the right anterior cingulate cortex (Arnone et al. Reference Arnone, McIntosh, Ebmeier, Munafo and Anderson2012). Although neither of these meta-analyses found significant volume reductions in the sgACC specifically, Du et al.'s (Reference Du, Wu, Yue, Li, Liao, Kuang, Huang, Chan, Mechelli and Gong2012) meta-analyses of voxel-based studies specifically implicated this area, and that by Bora et al. (Reference Bora, Fornito, Pantelis and Yucel2012) identified a single cluster close to it (which expanded to include it when a lower significance threshold was used).
In our study VBM also revealed clusters of volume reduction in the hippocampus and parahippocampal gyrus on the left in MDD patients. Hippocampal volume reduction in MDD patients has been a regular finding in studies using conventional volumetric measurement, and was supported in both the meta-analyses of Kempton et al. (Reference Kempton, Salvador, Munafo, Geddes, Simmons, Frangou and Williams2011) and Arnone et al. (Reference Arnone, McIntosh, Ebmeier, Munafo and Anderson2012). The right hippocampus was also identified as an area of volume reduction in Du et al.'s (2012) meta-analysis of VBM studies, although findings were negative in that of Bora et al. (Reference Bora, Fornito, Pantelis and Yucel2012); it should be noted that these two meta-analyses did not pool data from precisely the same set of studies. Also possibly relevant to the discrepant findings here is a recent study by Arnone et al. (Reference Arnone, McKie, Elliott, Juhasz, Thomas, Downey, Williams, Deakin and Anderson2013) which found that hippocampal volume reduction in MDD was mood-state dependent, being present in the depressive phase but normalizing after recovery.
Functional imaging during working memory performance revealed reduced activation in an area of the left lateral frontal cortex that included the DLPFC. This finding is in line with several other studies using cognitive paradigms in patients with MDD (Elliott et al. Reference Elliott, Baker, Rogers, O'Leary, Paykel, Frith, Dolan and Sahakian1997; Audenaert et al. Reference Audenaert, Goethals, Van Laere, Lahorte, Brans, Versijpt, Vervaet, Beelaert, Van Heeringen and Dierckx2002; Okada et al. Reference Okada, Okamoto, Morinobu, Yamawaki and Yokota2003; Siegle et al. Reference Siegle, Thompson, Carter, Steinhauer and Thase2007). However, some studies have failed to find activation differences (Berman et al. Reference Berman, Doran, Pickar and Weinberger1993; Barch et al. Reference Barch, Sheline, Csernansky and Snyder2003; Videbech et al. Reference Videbech, Ravnkilde, Kristensen, Egander, Clemmensen, Rasmussen, Gjedde and Rosenberg2003; Holmes et al. Reference Holmes, MacDonald, Carter, Barch, Andrew Stenger and Cohen2005; Rose et al. Reference Rose, Simonotto and Ebmeier2006), or have reported increased activation (Harvey et al. Reference Harvey, Fossati, Pochon, Levy, Lebastard, Lehericy, Allilaire and Dubois2005; Walter et al. Reference Walter, Wolf, Spitzer and Vasic2007; Fitzgerald et al. Reference Fitzgerald, Srithiran, Benitez, Daskalakis, Oxley, Kulkarni and Egan2008b ). These inconsistencies do not appear to be attributable to differences among the tasks used, and a number of studies with negative findings used the n-back task (Barch et al. Reference Barch, Sheline, Csernansky and Snyder2003; Harvey et al. Reference Harvey, Fossati, Pochon, Levy, Lebastard, Lehericy, Allilaire and Dubois2005; Rose et al. Reference Rose, Simonotto and Ebmeier2006; Fitzgerald et al. Reference Fitzgerald, Srithiran, Benitez, Daskalakis, Oxley, Kulkarni and Egan2008b ).
We found that reduced frontal activation in the depressed patients was accompanied by failure of deactivation in the medial frontal cortex. This finding strongly implies dysfunction in the default mode network: this network normally deactivates during performance of a wide range of attention-demanding tasks and the medial frontal cortex is one of its two midline ‘nodes’ or ‘hubs’ (e.g. Buckner et al. Reference Buckner, Andrews-Hanna and Schacter2008). Three previous studies have documented reduced task-related deactivation in the territory of the default mode network in MDD using cognitive–emotional tasks (Frodl et al. Reference Frodl, Scheuerecker, Albrecht, Kleemann, Muller-Schunk, Koutsouleris, Moller, Bruckmann, Wiesmann and Meisenzahl2009; Grimm et al. Reference Grimm, Boesiger, Beck, Schuepbach, Bermpohl, Walter, Ernst, Hell, Boeker and Northoff2009; Sheline et al. Reference Sheline, Barch, Price, Rundle, Vaishnavi, Snyder, Mintun, Wang, Coalson and Raichle2009), and this predominated in the medial frontal cortex in two of them (Frodl et al. Reference Frodl, Scheuerecker, Albrecht, Kleemann, Muller-Schunk, Koutsouleris, Moller, Bruckmann, Wiesmann and Meisenzahl2009; Sheline et al. Reference Sheline, Barch, Price, Rundle, Vaishnavi, Snyder, Mintun, Wang, Coalson and Raichle2009). Rose et al. (Reference Rose, Simonotto and Ebmeier2006) also had what were in all probability similar findings in a study using the n-back task – they found that a region involving the medial orbitofrontal and the subgenual/rostral anterior cingulate cortex was relatively more active in depressed patients than in the controls, but also noted that both groups showed task-related decreases in activation in this region. On the other hand, Davey et al. (Reference Davey, Yucel, Allen and Harrison2012) failed to find deactivation differences between patients with major depression aged 15–24 years using the multi-source interference task; however, this study also failed to find activation differences between the groups.
Another way of examining default mode network function is by means of resting state connectivity, although few studies have so far applied this technique to patients with MDD. Greicius et al. (Reference Greicius, Flores, Menon, Glover, Solvason, Kenna, Reiss and Schatzberg2007) used independent component analysis (ICA) to isolate the default mode network and found that depressed patients showed increased connectivity between parts of this network and the whole of the rest of the brain. These parts included the sgACC, and the precuneus, as well as the thalamus and the orbitofrontal cortex (which in their study also formed part of the ‘best-fit’ ICA component representing the default mode network). They argued on various grounds that the changes were particularly important in the sgACC. Sheline et al. (Reference Sheline, Price, Yan and Mintun2010) used a seed region placed in the precuneus, the posterior midline node of the default mode network, and examined connectivity within the network itself. They found increased resting state connectivity in several medial frontal regions in depressed patients compared with controls.
In conclusion, this study finds evidence that the sgACC is a site of both brain volume reduction in MDD and also functional brain abnormality, specifically failure of deactivation during task performance. This latter finding raises the intriguing question of whether such a task-related change might be related to the increased resting state activity in this area documented by Drevets et al. (Reference Drevets, Videen, Price, Preskorn, Carmichael and Raichle1992), Mayberg et al. (Reference Mayberg, Lozano, Voon, McNeely, Seminowicz, Hamani, Schwalb and Kennedy2005) and other authors (see Sacher et al. Reference Sacher, Neumann, Funfstuck, Soliman, Villringer and Schroeter2012). While it seems intuitive that the two findings might be different manifestations of the same underlying pathology, it is important to note that there is no a priori reason why one should automatically entail the other. Two important limitations of the study also need to be acknowledged. First, the patients in the study were on drug treatment, which has the potential to affect functional imaging findings. Although Phillips et al. (Reference Phillips, Travis, Fagiolini and Kupfer2008) have argued against marked effects due to this factor in bipolar disorder, this applied only to activation changes; the relationship with failure of deactivation has not yet been examined in drug-free patients with either bipolar disorder or major depression. Second, failure of deactivation is only an incomplete way to document default mode network dysfunction, and this finding needs to be complemented by further studies of resting state connectivity in MDD.
Supplementary material
For supplementary material accompanying this paper visit http://dx.doi.org/10.1017/S0033291714000841.
Acknowledgements
This work was supported by: (i) the Centro de Investigación Biomédica en Red de Salud Mental (CIBERSAM); (ii) several grants from the Instituto de Salud Carlos III (Miguel Servet Research Contract to R.S. (no. CP07/00048) and to E.P.-C. (no. CP10/00596); Intensification grant to S.S. (no. 10/231); and (iii) the Comissionat per a Universitats i Recerca del DIUE from the Catalonian Government (grant no. 2009SGR211 and no. 2009SGR1022).
Declaration of Interest
None.
