Differences in resting-state network
functional connectivity between
treatment-resistant depression,
major depressive disorder
and healthy controls
Lisanne Fellinger
MSc in Brain and Cognitive Sciences Cognitive Neuroscience track
University of Amsterdam July 19, 2012 B. P. de Kwaasteniet (supervisor) H. G. Ruhe (co-supervisor) February 2012- Augustus 2012 28 ECTS 5804442
Introduction
Major depression disorder (MDD) is associated with strong feelings of sadness, guilt, worthlessness, hopelessness, fatigue and suicidal thoughts (DSM-IV, 1994). With therapy and medication treatments these symptoms can generally be relieved for most people, however, 15-30% of MDD patients do not respond to any form of regular treatment; they are diagnosed with treatment resistant depression (TRD) that can result in extreme hopelessness with
possible fatal consequences. About 60% of suicides are depression related (Wahlbeck & Mäkinen, 2008). Apart from these highly unwanted effects, the numerous treatments that are tried, though ineffective, can have negative side-effects and are very costly for society. Side-effects can range from significant weight gain to memory loss and in 2004 depression was the most costly brain disorder in Europe, accounting for 33% of the total costs.
If there was better knowledge about the neurobiological mechanism of TRD,
considerable suffering, ineffective medication intake and many costs can be prevented when these patients can be better diagnosed and treated. However, for better diagnosis and more effective treatments there needs to be a neural correlate for TRD that can be measured. At present the neural correlates of TRD are unknown. However, much research has been performed to determine the neural correlates of MDD.
Numerous functional brain differences between healthy controls and MDD were found and this study focused on differences in resting-state networks (RSNs). A RSN consists of different brain areas that are connected by similar activation patterns formed when a person is engaged in internally focused tasks (Mazoyer et al., 2001; Buckner et al., 2008). An important RSN is the default mode network (DMN), which comprises areas in dorsomedial,
dorsolateral- and ventromedial prefrontal cortices (dmPFC, dlPFC and vmPFC), medial and lateral inferior parietal lobe (IPL), medial and lateral temporal lobe, posterior cingulate cortex/precuneus (PCC) and anterior cingulate cortex (ACC) (Buckner et al., 2008; Raichle et al., 2001; Greicius et al., 2003; Meindl et al., 2010, Fransson & Marrelec, 2008). In the DMN the PCC acts as a hub, connecting itself with the medial temporal lobe and the inferior parietal lobe (Kobayashi & Amaral, 2007). The vmPFC acts as a second hub, encompassing prefrontal areas like subgenual ACC, ACC and dlPFC (Petrides & Pandya 1994; Öngür & Price, 2000). These hubs can be differentiated into functional subsystems within the DMN, which are also connected to the orbitofrontal cortex (OFC) and the limbic system, key players in emotion, affect regulation and memory (Frodl et al., 2010). The frontal subsystem, around the vmPFC, exerts a controlling function over the OFC and limbic system and the posterior/PCC
subsystem is said to be involved in retrieving associations from memory to be used in mental exploration (Wheeler & Buckner, 2004).
With functional connectivity (FC) analysis functional interactions within and between networks can be investigated by quantifying the temporal correlation of neural activity patterns of anatomically separated brain regions. Interestingly, disruptions in FC within the DMN and between DMN, OFC and limbic structures are present in various psychiatric diseases like Alzheimer’s disease, schizophrenia, autism and depression. This gives rise to the question of how the DMN with its subsystems contribute to cognitive function and
dysfunction. Understanding how the many cortical and limbic areas are functionally connected within the DMN and thus mapping their FC patterns and individual cognitive operations may reveal something about the function of the network as a whole.
Task-based studies have implicated the PCC in retrieval of episodic memories (Maddock et al., 2001; Fujii et al., 2002), the ACC is found to be involved in both cognitive and emotional attentional processes (Bush et al., 2000) and the mPFC has been linked to the integration of cognitive and emotional stimuli (Simpson et al., 2001). Furthermore, decreased FC between the OFC and cingulate cortex and precuneus was found in MDD relative to controls (Frodl et al., 2010) and decreased functional correlation was found between the ACC
and limbic areas when processing negative stimuli (Anand et al., 2005). Carballedo and colleagues (2011) also found functional disconnection between limbic and frontal brain regions including OFC. Based on this evidence, insufficient prefrontal and PCC control over the limbic system might thus be underlying the behavioral, cognitive and emotional
dysfunction seen in depression.
Results from resting state FC studies in MDD show differences in DMN FC between MDD compared to controls within the cortical-limbic circuit. Greicius and colleagues (2007) found increased network FC in the subgenual ACC. Veer and colleagues (2010) only found decreased FC within a limbic-cortical network (insula, dorsal ACC, vmPFC and amygdala). These findings are contradicting but both can explain the symptoms accompanying MDD. Increased subgenual ACC, mPFC and vACC resting-state FC is correlated to rumination, focusing on negative thoughts, characteristic of MDD (Berman et al. 2011; Zhu et al., 2012). Decreased cortical-limbic FC however, can explain impaired top-down control over amygdala that is causing increased negative affect, also characteristic of MDD. Reduced FC between posterior regions of the DMN (PCC and precuneus) was also found and is related to increased autobiographical memory (Zhu et al., 2012). Based on this evidence it has been suggested that the DMN contributes to maintaining (affect) information for interpretation and response to the environment (Raichle & Snyder, 2007) and that it has a function in retrieval of personal and general information (Greicius et al., 2003).
At present the neurobiological mechanism of TRD is unknown and few studies that investigated it cannot easily be compared because of methodology heterogeneity between these studies. However, it is clear that the DMN is disrupted in TRD. One study measuring the synchronisation of spontaneous fMRI signal oscillations between brain areas (regional homogeneity, ReHo) found decreased ReHo in TRD compared to controls in the left insula, superior temporal gyrus and inferior frontal gyrus (Guo et al., 2011). However, no MDD group was included, making it impossible to say anything about differences between treatment resistance and non-resistance. In a study by Greicius and colleagues (2007) a positive correlation was found between length of the current depressive episode and FC in the subgenual ACC. Lui and colleagues (2011) found decreased FC in MDD compared to TRD in the left amygdala, ACC, right insula and precuneus. The findings from Lui et al. and Greicius et al. do not contradict each other, but there have not yet been other FC studies to support any of the results. Lui et al. found that compared to healthy controls TRD patients showed
decreased FC in thalamo-frontal circuits and MDD patients mainly showed decreased FC in the limbic-striatal-pallidal-thalamic circuit. Similar results were found with ReHo analysis. TRD patients showed higher ReHo than the MDD group in a temporo-limbic network and lower ReHo than MDD in cortical regions (Wu et al., 2011). One explanation for these findings is that the thalamo-cortical circuit is less sensitive to antidepressants than the limbic system, which could underlie treatment refractoriness (Lopez et al., 1998).
FC analyses distinguished a thalamo-cortical and limbic-thalamic circuit that differentiates between MDD and TRD. The various methods of FC analysis across these studies make it difficult to quantitatively compare all findings; ReHo analysis showed a difference in a temporo-limbic network between the groups and ALFF detected a difference in anterior parts of the DMN. Nevertheless, it seems that the DMN can be divided into separate sub-networks. Buckner et al. (2008) suggested an anterior/posterior division of the DMN, with two central hubs (vmPFC and PCC). There have been more studies suggesting a functional division of the DMN; some researchers argue that there is a dorsal/ventral difference with decreased cognitive control from the dorsal over the emotional sensitive ventral network, resulting in MDD symptoms (Mayberg et al., 1994; Philips et al., 2003).
A method that can provide a clear picture of how the DMN could be divided into separate components based on functionality is anticorrelation analysis. Anticorrelation
between two networks indicates that they are temporally modulated in opposite directions. Networks that anticorrelate with the DMN are called task-positive networks (TPN) because they are active when a person is focusing on the outside world. It has been found that
anticorrelation analysis with different seed regions within the DMN result in different TPN’s, encompassing different brain areas. While there is a high positive correlation between all regions within the DMN, the finding of multiple anticorrelating networks suggests there is also a functional division within the DMN. Uddin and colleagues (2009) found two differing anticorrelation networks; a network in parietal visuospatial and temporal attention areas anticorrelating with a ventromedial vmPFC seed, and a prefrontal motor circuit anticorrelating with a seed in the PCC. Two important seeds within hubs of the DMN network that have very distinguishable anticorrelation networks suggest these hubs are part of separate sub-networks within the DMN. These seed regions may thus be functionally differentiated, whereas up to this point it was generally assumed the DMN is a homogenous network.
The current study will compare DMN FC between healthy control, MDD and TRD subjects to discover neural correlates that could explain treatment refractoriness. It is expected that MDD and TRD have overall less DMN FC compared to healthy controls. A further FC decrease is expected in TRD compared to MDD, mainly within a frontal-limbic network. Furthermore, we will try to distinguish two separate components within the DMN of healthy controls that have also been found by Uddin and colleagues (2009). This is a first step to separate the DMN in smaller sub regions that could make it easier to find the neural correlate of TRD when compared to MDD on these sub regions. We are expecting to find two networks that anticorrelate with a vmPFC and PCC seed in the DMN of healthy controls.
Methods Participants
All participants were tested and scanned in the Academic Medical Center (AMC) in Amsterdam. Three groups of participants were recruited; control subjects (n=20) were found by advertisements throughout the AMC and data from the MDD patients (n=17) was acquired from a direct colleague conducting related research. The TRD patients (n=19) in this
experiment were eligible for deep brain stimulation treatment (DBS) and therefore gave consent to participate in clinical research. About 50% of the patients were first seen in the St. Elisabeth Hospital in Tilburg and referred to the AMC by a psychiatrist.
All patients were experiencing a depressive episode at the time of scanning. The MDD patients stopped taking anti-depressants a month prior to scanning, but TRD patients
continued their medication. No subjects were excluded before data preprocessing. Procedure
Participants underwent a structural scan and a resting-state scan for which they were instructed to keep their eyes open. In total participants spent approximately 35 minutes inside the MRI.
Data acquisition
Data were obtained using a 3.0 Tesla Philips Intera full-body fMRI scanner (Philips Medical Systems BV, Eindhoven, the Netherlands) located at the Academic Medical Center, Amsterdam. Functional images for 13 of the controls and the TRD group were collected with a T2*-weighted echo planar imaging (EPI) sequence covering the entire brain except inferior parts of the cerebellum (with a 96x96 matrix, FOV of 220x105x220 mm, 35 slices, slice thickness of 3 mm, TR= 2.3 s, TE = 30 ms, flip angle of 80). Images from the other 7 control
subjects and the MDD group were collected with the same settings except for the FOV (220x120x220) and number of slices (40). Per subject 200 images were acquired. Data preprocessing
Before further processing with SPM8 the 3T EPI images were converted to Nifti files using a Matlab tool (http://r2agui.sourceforge.net/) and to image (.img) files using MRIcro (http://www.mccauslandcenter.sc.edu/mricro/mricro) which makes them readable for SPM (www.fil.ion.ucl.ac.uk/spm). Two dummy scans were made before the actual scans, therefore no slices were discarded. For each subject the scans were realigned first. Then a slice timing correction was applied and the resulting images were co-registered to the mean functional image per subject. Following was segmentation into white and gray matter and cerebrospinal fluid. Then the images were normalized to MNI space and smoothed (8 mm isotropic
Gaussian filter).
After these processing steps translation results were inspected to assure a maximum movement less than 2 mm for all subjects. Co-registration results were also inspected for distortions. This resulted in the exclusion of 1 subject with an artefact in the resting state image. With the functional connectivity option in REST, a resting-state MRI analysis toolbox (http://www.restfmri.net/), the linear trend was removed and the data were then filtered (between 0.01-0.08 Hz, with TR=2.3).
Statistical model
A full factorial experimental design was implemented with the MDD, TRD and control group as independent variables. Independence of the data and unequal variance between the groups were assumed.
Functional Connectivity
Voxel wise FC analysis using the REST toolbox was achieved with 2 spherical seeds centered in PCC and vmPFC, originally implemented by De Luca and colleagues (2006) and replicated by Uddin and colleagues (2009). See Figure 1 for the location of the seeds.
Anatomical locations were determined with the WFU pickatlas 3.0.3 (http://fmri.wfubmc.edu/software/PickAtlas).
To control for artifacts and to reduce false-positive activity across voxels, a mask was applied during FC analysis (Ridgway et al., 2009). Only voxels with a signal across an average of all subjects were included in the mask, resulting in the exclusion of false-positives.
The second level results were constructed with SPM8. Data were threshold at p=0.005 (not FWE corrected) and only FWE corrected whole-brain and cluster-level mean voxel changes were reported. If possible, small volume corrections with ROIs from the WFU pickatlas were applied. Otherwise 10 mm small volume corrections were applied to analyze the seed regions in PCC and vmPFC and regions that were not listed in WFU pickatlas.
Figure 1. Locations of seeds used for functional connectivity analysis. VmPFC
(red) coordinates = 2, 54, -3; PCC (blue) coordinates = -2, -51, 27.
Anticorrelation
Anticorrelation was analyzed in SPM8 in a similar way to regular functional
connectivity analysis. By using a different contrast (controls -1, MDD 0, TRD 0) during FC analysis the anticorrelation network of the DMN could be mapped for control subjects. Results
There were no sex differences between the groups. There was a significant difference between mean ages of the groups (F = 6,123, p = .004) (see Table 1). The MDD group was significantly younger than the control (p = .002) and TRD group (p = .005).
The mask to control for artefacts and false-positives was successfully applied and removed all extracranially active voxels from analysis, creating a more realistic comparison between the groups.
Because the control group consisted of data from two separate studies (n=13 and n=7), a two-sample t-test was used to test for significant FC differences between the subgroups. A significant FC difference between the subgroups with the PCC was found with the right Table 1. Mean age and standard deviation of the experimental groups.
N Men Women
Mean HDRS
Mean
Age Std. Deviation Minimum Maximum
controls 20 8 12 1.3 53,40 6,700 30 61
MDD 17 5 12 20.4 44,88 8,328 32 60
amygdala with a ROI mask correction (18, 0, -18, p < .05). A significant difference on the whole-brain level was found between vmPFC and an area including right precuneus (6, -80, 46, p < .05).
Functional Connectivity with PCC
MDD relative to controls. The MDD group showed decreased FC compared to controls between PCC and medial prefrontal areas (orbitofrontal cortex (p = .004), subgenual ACC (p = .000), dlPFC (p = .02), vmPFC (p = .001), along the midline to posterior dorsal cortical areas (inferior temporal gyrus (p = .000) and precuneus (p = .000)). Decreased FC was also seen between PCC and parts of the limbic system; putamen (p = .01), right
hippocampus (p = .009), bilateral caudate (p = .006), bilateral pallidum (p = .042) and right insula (p = .032).
TRD relative to controls. Decreased FC connectivity in the TRD group compared to controls was found between PCC and midline areas and ACC, in a medial orbitofrontal cortex-precuneus pathway. Furthermore, decreased FC was found between PCC and dlPFC (p = .0021), vmPFC (p = .002), subgenual ACC (p = .004) and bilateral medial frontal gyrus (p = .000). There was also decreased FC between PCC and the limbic system (bilateral amygdala (p = .032), bilateral insula (p = .000), bilateral putamen (p = .006), bilateral caudate (p = .046) and thalamus (p = .004).
TRD relative to MDD. Decreased FC between PCC and the left insular lobe (p = .033) was seen in the TRD compared to the MDD group. A significant decrease in FC difference was also found between PCC and right dlPFC (p = .035). Areas with increased FC with the PCC in the TRD compared to MDD group include right precuneus (p = .013), left caudate body (p = .004) and left inferior parietal lobe (p = .013). (See Table 2 for all significance levels).
In conclusion, significant differences in FC between the TRD and MDD group were found. Decreased FC in TRD compared to MDD was mainly found between PCC and parts of the limbic system and between PCC and right dlPFC. Increased FC in TRD compared to MDD was found between PCC and limbic structures. Furthermore, both the MDD and TRD group showed significantly reduced functional connectivity, compared to controls, between PCC and various regions of the DMN; medial frontal gyrus, anterior and middle cingulate cortex, inferior and superior temporal gyrus, orbitofrontal cortex, dlPFC, bilateral superior parietal lobules and precuneus.
Functional Connectivity with vmPFC
MDD relative to controls. Decreased FC in the MDD group compared to the control group was seen between vmPFC and posterior regions, limbic regions, precuneus (p = .000) and PCC (p = .003). A decrease in FC with posterior regions was seen between vmPFC and dorsal ACC (p = .002), subgenual ACC (p = .021) and dlPFC (p = .008). A decrease in FC with the limbic system was found between vmPFC and bilateral amygdala (p = .019), bilateral hippocampus (p = .040), bilateral insula (p = .026), left putamen (p = .016) and left caudate (p = .05).
TRD relative to controls. Significant FC decreases in TRD relative to controls between vmPFC and frontal regions include dorsal ACC (p = .039), medial cingulate cortex (p = .000), dlPFC (p = .042), medial (p = .006) and superior frontal gyrus (p = .006) and right inferior orbitofrontal cortex (p = .016). Significantly decreased FC in TRD was also seen between vmPFC and limbic areas (bilateral amygdala (p = .009), bilateral hippocampus (p = .032), bilateral putamen (p = .018), bilateral insula (p = .001) and bilateral thalamus (p = .003)).
Furthermore, significantly decreased FC in TRD was found between vmPFC and PCC (p = . 002) and precuneus (p = .000).
TRD relative to MDD. A significant decrease in FC in MDD compared to TRD was found between vmPFC and ACC (p = .03). A significant increase in FC in MDD compared to TRD was seen between vmPFC and right hippocampus (p = .036). (See Table 3 for all
significance levels.)
In conclusion, two significant FC differences between MDD and TRD were found. Compared to the MDD group, TRD showed significantly increased FC between vmPFC and ACC and significantly decreased FC between vmPFC and right hippocampus. Furthermore, both MDD and TRD groups showed significantly decreased FC with the vmPFC seed, compared to the control group. Decreases in FC were mainly seen between vmPFC and cortical areas; from left inferior orbitofrontal cortex, to inferior and medial temporal cortex and precuneus. Significant decreases were also found between vmPFC and dlPFC and the cingulate cortex. Furthermore, decreased FC was found between vmPFC and limbic areas (bilateral amygdala, bilateral hippocampus, bilateral insula, putamen and left caudate body). Anticorrelation
Brain regions of control subjects that anticorrelated with the PCC included bilateral insula (p = .000), middle temporal gyrus (p = .002), supplementary motor area (p = .000), supramarginal gyrus (p = .000), frontal gyrus (p = .000) and parts of the parietal and temporal cortex (p = .000) (see Table 4 for all significance levels).
Regions that anticorrelated with the vmPFC seed include parts of the parietal cortex (p = .000), middle temporal (p = .023) and occipital cortex (p = .01), superior frontal gyrus (p = . 027), supramargninal gyrus (p = .000), supplementary motor area (p = .027) and precuneus (p = .025). (See Table 5 for significance levels). See also Figure 2 and 3 for a visualization of correlation and anticorrelation analysis.
In conclusion, both networks largely overlapped one another. However, the PCC’s anticorrelation network was more widespread in anterior regions, like middle and superior frontal gyrus and inferior frontal operculum while the anticorrelation network of the vmPFC was more extensive in posterior areas, like bilateral middle occipital cortex and precuneus.
Analysis Regions of significant clusters
Side BA MNI coordinates Voxels in cluster
Statistics (p-values)
x y z
control > MDD Subgenual ACC Dorsal ACC vmPFC dlPFC Hippocampus Putamen Caudate Pallidum Insula Precuneus
Medial frontal gyrus Inf. orbitofrontal gyrus Inf. temporal gyrus Inf. parietal lobe
-L R R L R L R L R R L R L R R L R L R 32 11 46 46 7 6 37 37 -2 -6 6 54 -46 18 -18 22 -14 14 -18 22 26 -2 2 -26 34 31 -50 50 -54 42 36 28 52 40 40 -24 8 4 16 8 0 0 16 -60 -60 12 56 26 -56 -60 -52 -52 -10 22 -10 18 30 -10 10 10 6 14 6 6 -18 50 50 62 6 -21 -22 -18 38 54 72 176 63 11 14 2 38 27 49 34 5 4 11 315 215 -116 61 181 202 220 54 0.000 0.000 0.001 0.020** 0.015** 0.009** 0.010 0.019 0.006 0.013 0.042 0.045 0.032** 0.000 0.000 0.000* 0.000 0.004 0.000 0.000 0.000✝✝ 0.010✝✝ control > TRD Subgenual ACC
Dorsal ACC vmPFC dlPFC Amygdala Putamen Caudate Pallidum Insula Precuneus Thalamus
Med. orbitofrontal cortex Inf. temporal gyrus
-L R L R L R L R L R L R L R L R R R 32 32 46 46 34 34 7 5 10 10 6 -46 50 -30 26 -18 22 -14 14 -18 14 -30 34 -6 2 -10 14 22 58 28 24 48 40 40 4 4 12 4 16 24 4 0 28 20 -52 -40 -20 -24 44 -56 -5 26 -2 26 26 -18 -18 -10 6 -10 -6 -6 -2 -2 -18 66 58 2 2 -18 -14 42 -55 28 26 6 7 49 57 6 16 9 14 133 140 180 97 51 34 167 -0.004 0.000* 0.002 0.021 0.025 0.032 0.032 0.006 0.004 0.046** 0.042 0.028 0.018 0.000 0.000 0.000 0.002 0.004 0.011 0.000 0.000* MDD > TRD Insula dlPFC RL -3030 2836 38-2 59 0.033**0.035✝ TRD > MDD Caudate Precuneus Inf. parietal lobe
L R L 40 -18 26 -54 12 -52 -56 14 26 38 10 49 36 0.004** 0.013 0.013
Table 2. Regions with FC differences from PCC seed
ACC (anterior cingulate cortex), vmPFC (ventromedial prefrontal cortex), dlPFC (dorsolateral prefrontal cortex)
* whole-brain significance ✝10 mm sphere ** peak-level significance ✝✝ 20mm sphere
Table 3. Regions with FC differences from the vmPFC seed
Analysis Regions of significant
clusters Side BA MNI coordinates Voxels incluster Statistics (p-values)
x y z
control > MDD Subgenual ACC Dorsal ACC PCC dlPFC Amygdala Hippocampus Putamen Caudate Insula Precuneus
Medial frontal gyrus Inf. obitofrontal gyrus Inf. temporal gyrus Inf. parietal lobe
-L R L R L R L L L R L R L R L R L R 46 46 34 34 25 13 5 9 47 13 -10 10 2 -50 58 -30 30 -18 30 -30 -2 -30 34 -6 2 -34 42 -34 46 -34 38 32 48 -52 24 36 4 0 -8 -4 -8 8 16 -24 -48 -56 16 16 32 -40 -56 -52 -6 14 26 26 10 -18 -26 -22 -22 6 -6 -18 10 62 54 34 38 -6 18 38 38 16 71 53 9 31 12 15 7 27 31 14 33 42 342 279 10744 10744 10744 10744 203 181 0.021 0.002 0.003 0.008** 0.019 0.019 0.016 0.040** 0.023 0.016 0.050 0.026 0.016 0.000 0.000 0.000 0.000 0.000 0.000 0.000✝✝ 0.000✝✝ control > TRD Dorsal ACC
Medial cingulate cortex PCC dlPFC Amygdala Hippocampus Putamen Insula Precuneus Thalamus
Inf. orbitofrontal cortex Medial frontal gyrus Superior frontal gyrus Inf. parietal lobe
-L R L R L R L R L R L R L R R L R L L 46 46 34 5 5 10 10 2 -2 2 -54 54 -30 30 -18 30 -30 30 -30 38 -6 2 -18 22 50 -30 46 -22 -46 12 -12 -48 28 40 4 4 -28 -8 -16 -12 12 -28 -48 -40 -24 -24 48 60 28 64 -48 26 42 34 22 14 -18 -26 -6 -26 6 10 18 -18 62 58 -2 -2 -6 18 34 26 46 19 -54 18 -23 21 21 25 29 17 95 69 336 176 60 24 -250 0.039** 0.000* 0.002 0.042 0.016* 0.009 0.011 0.032 0.026 0.018 0.039 0.001 0.004 0.000 0.000 0.003 0.020 0.016* 0.006* 0.016* 0.006* 0.000✝✝ MDD > TRD Hippocampus R 42 -20 -18 9 0.036✝
TRD > MDD Anterior cingulate cortex R 10 18 48 10 11 0.030✝
ACC (anterior cingulate cortex), PCC (posterior cingulate cortex), dlPFC (dorsolateral prefrontal cortex)
* whole-brain significance
** peak-level significance
✝ 10 mm sphere
Table 4. Negatively correlated regions from the PCC seed
Regions of significant clusters
Side BA MNI coordinates Statistics (p-values)
x y z
Insula
Middle temporal gyrus Supramarginal gyrus Superior parietal cortex Middle frontal gyrus Inferior frontal operculum Supplementary motor area Superior temporal pole Superior frontal gyrus Inferior parietal cortex Superior occipital cortex Precentral gyrus L L R R R L R R R L R L R L R L L R R R L R 13 21 40 40 7 10 10 6 7 9 8 -38 -34 34 42 58 -62 62 54 22 -38 42 -46 54 -10 2 -50 -18 22 38 22 -58 54 12 16 20 -8 -52 -36 -28 -40 -56 48 40 4 12 4 4 8 -4 0 -44 -64 8 8 2 6 6 -10 -2 42 34 58 70 30 6 26 6 46 54 -6 74 54 42 46 34 42 0.000 0.000 0.000 0.000 0.002** 0.000* 0.000* 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
* whole brain level significance ** whole brain peak-level significance
Table 5. Negatively correlated regions from the vmPFC seed Regions of significant
clusters Side BA MNI coordinates Statistics (p-values)
x y z
Supramarginal gyrus Superior parietal cortex Middle temporal gyrus Inferior parietal cortex Supplementary motor area Superior frontal gyrus Middle occipital cortex Precuneus L R L R L R L R R L R R 2 7 7 6 -62 66 -22 22 -46 58 -30 6 22 -30 38 18 -28 -32 -56 -64 -52 -52 -48 8 0 -80 -76 -68 38 46 46 50 -2 -2 42 54 62 -26 22 46 0.000* 0.000* 0.000✝ 0.000* 0.023 0.004** 0.000 0.027 0.027 0.010✝ 0.015✝ 0.025
* Whole brain level significance
** peak-level significance
Discussion
Functional Connectivity
To find out more about the neural correlates of TRD we looked at the differences in resting state FC between MDD and TRD from two seed regions in PCC and vmPFC. Consistent with our expectations and previous studies, both the MDD and TRD group show reduced FC compared to controls between PCC and other parts of the DMN and the limbic system and between vmPFC and anterior to posterior cortical areas within the DMN and limbic areas.. Partly in accordance with our expectations our findings suggest increased FC between PCC and the posterior part of the DMN, part of the limbic system (left caudate body), inferior parietal lobe and between vmPFC and ACC in TRD compared to MDD. On the other hand, decreased FC between PCC and the left insular lobe and dorsolateral frontal regions of the DMN and between vmPFC and the hippocampus are seen in TRD compared to the MDD group.
Most of our expectations were confirmed; decreased prefrontal FC in depression is related to symptoms of heightened emotionality and increased negative rumination because of decreased control over the limbic system (Mayberg, 2003). Decreased FC between vmPFC and posterior cortical areas in depression is associated with decreased control over retrieval of episodic memories. This explains the negative bias seen in depressed patients in response to visual stimuli (Kumari et al., 2009; Gusnard & Raichle, 2001). However, some surprising FC differences between the MDD and TRD group need to be inspected in more detail.
Alterations of prefrontal cognitive control over emotional responses are consistently seen in depression disorder (Wang et al., 2008). Our study adds to this conclusion with the finding of decreased FC between dlPFC and PCC in TRD relative to MDD, which indicates less cognitive control over PCC in TRD. Interestingly, previous studies that compared FC between MDD and controls found increased FC in dlPFC and between dlPFC and PCC in MDD (Vasic et al., 2008; Zhou et al., 2010). Based on these findings it would therefore be expected that an even larger increase in FC with dlPFC is seen in TRD, however, our results show the opposite effect. This specific finding could be a biomarker of treatment resistance. However, our results show a decrease in FC between PCC and dlPFC in MDD compared to controls, but this result was only significant at the peak-level, so it seems the FC decrease between PCC and dlPFC is more specific to the TRD group. This is also confirmed by the finding of an inverse relationship between activity of the prefrontal cortex and severity of depression (Wang et al., 2008) and between prefrontal cortex metabolism and treatment resistance (Mayberg, 2003).
Inferior parietal cortex (IPC) activity is seen during working memory tasks (Jonides et al., 1998) and is functionally connected with the PCC in resting state (Greicius, 2003). In depression decreased FC within a prefrontal-parietal network including IPC is seen, more so during working memory related tasks (Vasic et al., 2009). Our results also show decreased FC between both seed regions and bilateral IPC in MDD compared to controls. Interestingly, a decrease in the TRD-controls comparison is only seen between vmPFC and left IPC.
Furthermore, there is actually a FC increase between PCC and left IPC in TRD compared to MDD. The difference in vmPFC-IPC FC between MDD and TRD is not significant.
Nonetheless, this overall FC increase is specific to the TRD group, which suggests that left IPC could also be related to treatment resistance, similar to dlPFC.
Decreased FC between PCC and the left caudate body can be related to symptoms of anhedonia, the inability to experience pleasure from activities, characteristic of depression. Intuitively a bigger decrease in FC between PCC and the caudate body would be expected for the TRD group, since this group experiences more severe and long lasting symptoms of depression. However, the caudate is involved in processing rewards (Delgado et al., 2000) and
the relative decrease for the MDD rather than TRD group can be explained because
dysfunction in this structure is one of the earliest manifestations of depression (Bluhm et al., 2009). Depression duration is much shorter for the MDD group compared to the TRD group, because before TRD diagnosis, multiple medication treatments and therapy have been completed. MDD patients in this study were scanned within a few months of diagnosis, decreased PCC-caudate FC as an early sign of depression might therefore still be visible in this group while medication use has altered FC in TRD over time.
Another limbic area that has increased FC with PCC in MDD relative to TRD is the left insula. The insula regulates emotional states by mediating the interpretation of sensory information from the body (Craig, 2002). In general there is a FC decrease visible between the insula and both seed regions in depression compared to the control group, underlying
depressive symptoms like a negative bias in interpreting interpersonal feedback (Cullen, 2009). A further decrease for TRD relative to MDD can be explained by greater severity and duration of depressive symptoms in TRD.
Our finding of increased FC between vmPFC and dorsal ACC in TRD relative to MDD is also in line with previous research. This suggests increased regulation of emotional and motivational responses given that the ACC is involved in assessing the salience of emotional and motivational information (Devinsky et al., 1995, Rogers et al., 2004) and the vmPFC is implicated in the generation of an abstract representation of reward value and integrating emotional and cognitive processes (Elliott et al., 2000; Gusnard & Raichle, 2001). This effect can be explained by the fact that people with depression disorder focus and think more about their feelings, particularly negative feelings, than healthy people. This is
supported by research about anhedonia. Anhedonia severity and the extent of vmPFC
response to happy stimuli are positively correlated; a larger vmPFC response is seen in people with more severe anhedonia symptoms. This effect shows that patients with severe anhedonia focus more attention towards happy stimuli than sad stimuli, possibly because in general they are already more focused on negative thoughts and extra vmPFC activation is not necessary to process these stimuli (Keedwell et al, 2005; Kumari et al 2003; Mitterschiffthaler et al., 2003). Increased FC from vmPFC to ACC in TRD compared to MDD therefore can also be explained by increased assessment of the salience of positive emotional and motivational information.
A decrease in FC between vmPFC and the hippocampus is related to memory retrieval and seen in TRD compared to MDD. This finding suggests a dysfunctional link between award valuation and memory retrieval in line with the finding of increased PCC FC and inhibited positive memory retrieval that is more severe in the TRD group. There is evidence that functional abnormalities in the hippocampus are specifically observed when a person is processing positive stimuli (Kumari et al., 2003). Negative rumination, very characteristic of depression disorder can be explained by these functional connectivity abnormalities in PCC FC and between vmPFC and the hippocampus. This could explain why people with
depression disorder do not attach meaning to positive events the same way as healthy people do.
In general, our findings are consistent with previous research and differences between MDD and TRD could all be explained by differences in depression severity but were not specific to either MDD or TRD, except for decreased PCC-dlPFC FC and increased PCC-left IPC in TRD compared to MDD. Based on results of studies investigating the effects of antidepressants it has already been suggested that an important biomarker of treatment resistance might be related to increased prefrontal dysfunction in TRD relative to MDD, because the thalamo-cortical circuit is less sensitive to antidepressants than the limbic system (Lopez et al., 1998). The consistent finding of prefrontal cortex normalisation in MDD after antidepressant treatment (Mayberg, 2009) provides further evidence for this conclusion.
Considering MDD patients are taking no antidepressants or much less and over a shorter period of time than TRD patients, it would be expected that FC in TRD is more similar to controls than MDD FC in the cortical circuit. However, this is not what we find; indicating that treatment resistance has an important relation to prefrontal dysfunction.
The finding of increased FC from both vmPFC and PCC to areas of the limbic system in MDD compared to TRD also suggests an underlying difference between treatment response and treatment resistance. Dysfunction within the limbic network can be normalized by
antidepressants, while they appear to have no effect on the cortical circuit in TRD (Naranjo et al., 2001).
All areas discussed above are directly and indirectly linked to each other within the DMN. Increased or decreased FC between different areas in depression does not only affect these specific areas, but has consequences for the functionality of the entire DMN. For example, the vmPFC is directly linked to many areas within the limbic system, to the PCC and to other prefrontal areas like dlFPC (Öngür & Price, 2000). We therefore suggest that insufficient prefrontal and PCC control over the limbic system might be underlying the behavioral, cognitive and emotional dysfunction seen in depression. This view of a top-down deficit in the regulation of attention for emotional control is also supported by Fales and colleagues (2008) and offers an explanation for the negativity bias seen in depressed patients. In further support of this statement Phan and colleagues (2005) found that inhibition of negative affect is associated with ACC and dlPFC activity. Thus, decreased FC between dlPFC and PCC in TRD relative to MDD may be causing the negative emotional state as a main characteristic of depression disorder.
Anticorrelation
Secondly we tried to distinguish an anterior and posterior component of the DMN of control subjects with anticorrelation analysis from a PCC and vmPFC seed. Anticorrelation between areas means that those areas are functioning antagonistically and therefore must also be coupled to each other. Two anticorrelation networks were mapped that largely overlapped each other, but nevertheless were distinguishable in some regions. The PCC’s anticorrelation network for example is more extensive in anterior regions like middle and superior frontal gyrus and inferior frontal operculum. The vmPFC anticorrelation network is more spread out in the posterior direction, including bilateral middle occipital cortex and precuneus areas. Even though the clearly distinct anticorrelation networks that were found by Uddin et al. (2009) were not as clearly separated in our study, these anterior and posterior differences in anticorrelation networks do suggest some amount of separation between an anterior and posterior component within the DMN. Moreover, the PCC is found to anticorrelate with bilateral insular lobe and the vmPFC seed anticorrelates with precuneus, while in a positive correlation analysis these regions are all highly correlated to each other. Buckner and
colleagues (2008) mapped correlation strengths between areas within the DMN and also show a distinction between anterior and posterior subsystems. These subsystems interact with each other through the vmPFC and PCC hubs.
Limitations
The differences in FC that are seen between MDD and TRD in this study are
consistent with previous research suggesting that the cortico-limbic circuit plays an important role in the neural correlates of treatment resistance. However, especially FC differences between vmPFC and PCC and the limbic system could also be explained by differences in medication between the MDD and TRD group. Some subjects in the MDD group had never taken antidepressants in their life and the ones that had, paused their intake from a month prior to scanning, while all of the TRD patients were on antidepressants throughout the
scanning procedure. Some were even on a combination of antidepressants. As mentioned before, antidepressants primarily act on the limbic system (Lopez et al., 1998) so any differences in FC that are found with this system between MDD and TRD could also be explained by the normalising effect of antidepressants in the TRD but not MDD group. On the other hand, FC normalisation in TRD due to medication intake could have eliminated
differences between MDD and TRD (Anand et al., 2005), resulting in the loss of significant differences that could otherwise have been found. If the TRD group had not taken any medication we would probably have found even bigger differences in FC between MDD and TRD.
In general, differences that were found between all groups could also be attributed to physical confounding factors that were not controlled for in this study (heart rate and
respiratory movement). For example, some studies found differences in respiratory rhythms between healthy people and people with panic disorder (Martinez et al., 2001). This could account for the differences that were found between TRD and the two comparison groups; none of the depressed patients in this study had panic disorder, but some TRD patients were diagnosed with other anxiety disorders that might have the same effects on respiratory rate. In contrast, almost none of the MDD patients and none of the controls were diagnosed with anxiety disorder. Even without differences between groups, respiratory rate influences FC analysis nonetheless and particularly in the DMN, because its signal changes occur at the same low frequencies as signal fluctuations related to FC (Birn et al., 2006). Therefore, our findings could possibly also reflect mean differences in respiratory rate.
Illness duration, which is much longer for the TRD group, could also account for differences in FC that were found between MDD and TRD. There is no conclusiveness yet about the exact causal relation between the DMN and depression. It could be that either a disruption of the DMN causes symptoms of depression, or depression disrupts normal functioning of the DMN. If the latter statement is found to be correct, longer illness duration will obviously have more disruptive effects on the DMN that are then more visible in the TRD compared to MDD group. Furthermore, despite careful age and sex matching of the subjects there was also a significant age difference between the groups, caused by a younger MDD group that could have influenced the results due to developmental/aging differences between the groups.
This was the first study trying to replicate the finding of two anticorrelational
networks of the DMN (Uddin et al., 2009). We did not find such a clear distinction between a vmPFC and PCC anticorrelation network that could suggest there is not a clear distinction between separate components within the DMN. Nevertheless, there is substantial evidence for the existence of subsystems within the DMN so our finding should not necessarily be
interpreted as evidence against the existence of separate components within the DMN (Uddin et al., 2009; Buckner et al., 2008; Damoiseaux et al., 2006). Future studies of DMN
anticorrelation networks should compare more seed regions within PCC and vmPFC because they are found to be important hubs within the DMN (Buckner et al., 2008), but the PCC specifically is an area that comprises numerous voxels within the brain. No specific argument for the chosen seed is made by Uddin and colleagues, except that it has been implemented in a previous study. Choosing a different seed region within the PCC, for example one that is located further anterior, could result in different anticorrelational networks. In the future, it should be investigated exactly which subsystems of the DMN cause depression symptoms and comparing them between MDD and TRD may also result in more knowledge about the neural correlates of TRD.
Comparing any resting-state FC study with another should not be done without taking some factors into consideration. There are differences in the resting-state paradigms applied in research. In our study subjects were instructed to stay awake with their eyes open while the
light in the scanning facility was still on and subjects could still look at the screen that was not turned off. In some other studies subjects were instructed to stay awake with their eyes closed or they had to fixate on a single point on a screen. A recent study however, found that FC strength is increased with eyes open or if one fixates, compared to eyes closed (van Dijk et al., 2010).
Conclusion
It is ahead of time to conclude we have found the neural correlates of TRD. Nevertheless, our study provides evidence for the involvement of some areas in non-responsiveness to therapy and antidepressants. Decreased FC between PCC and dlPFC and increased FC between PCC and left IPC are TRD specific and not related to depression severity.
Even though functional connectivity analysis does not reveal anything about
directionality, it can be successfully applied in the search of the neural correlates of TRD as seen in this study. Future research should however include directionality measurements to be able to draw conclusions about the specific relationship between the PCC and the two crucial areas mentioned above. Unraveling the specific relationship between the PCC and dlPFC and IPC in the depressed brain will tell us more about the cause and consequences of treatment resistance. Moreover, anticorrelation analysis provides a distinction within the DMN in healthy subjects. Future research should use anticorrelation analysis to investigate these frontal- and posterior components in TRD and compare their anticorrelation networks to those of MDD patients. This could provide more insight into the neural correlate of TRD.
References
Anand, A., Li, Y., Wang, Y., Wu, J., Gao, S., Bukhari, L., Mathews, V. P., Kalnin, A., & Lowe, M. J. (2005). Activity and connectivity of brain mood regulating circuit in depression: A functional magnetic resonance study. Biological Psychiatry, 57 (10), 1079-1088.
Berman, M. G., Nee, D. E., Casement, M., Kim, H. S., Deldin, P., Kross, E., & Jonides, J. (2011). Neural and behavioral effects of interference resolution in depression and rumination. Cognitive Affective, & Behavioral Neuroscience, 11 (1), 85-96.
Birn, R. M., Diamond, J. B., Smith, M. A., & Bandettini, P. A. (2006). Separating respiratory-variation-related fluctuations from neuronal-activity-related fluctuations in fMRI. Neuroimage, 31, 1536–1548.
Bluhm, R., Williamson, P., Lanius, R., Théberge, J., Densmore, M., Bartha, R., et al. (2009), Resting state default-mode network connectivity in early depression using a seed region-of-interest analysis: Decreased connectivity with caudate nucleus. Psychiatry and Clinical Neurosciences, 63, 754–761.
Buckner, R. L., Andrews-Hanna, J. R., & Schacter, D. L. (2008), The Brain's Default Network. Annals of the New York Academy of Sciences, 1124, 1–38.
Bush, G., Luu, P., & Posner, M. I. (2000). Cognitive and emotional influences in anterior cingulate cortex. Trends in Cognitive Science, 4, 215–222.
Carballedo, A., Scheuerecker, J., Meisenzahl, E., Schoepf, V., Bokde, A., Moller, H. J., et al. (2011). Functional connectivity of emotional processing in depression. Journal of Affective Disorders, 134 (1-3), 272–279.
Craig, A. D. (2002). How do you feel? Interoception: The sense of the physiological condition of the body. Nat. Rev. Neurosci., 3, 655–666.
Damoiseaux, J. S., Rombouts, S. A., Barkhof, F., Scheltens, P., & Stam, C. J., et al. (2006). Consistent resting-state networks across healthy subjects. Proc. Natl. Acad. Sci. U.S.A., 103, 13848–13853.
Delgado, M. R., Nystrom, L. E., Fissell, C., Noll, D. C., & Fiez, J. A. (2000). Tracking the hemodynamic responses to reward and punishment in the striatum. J. Neurophysiol, 84, 3072–3077.
Devinsky, O., Morrell, M. J., & Vogt, B. A. (1995). Contributions of anterior cingulate cortex to behaviour. Brain, 118, 279–306.
Van Dijk, K. R., Hedden, T., Venkataraman, A., Evans, K. C., Lazar, S. W., & Buckner, R. L. (2010). Intrinsic functional connectivity as a tool for human connectomics: Theory, properties, and optimization.and optimization. Journal of Neurophysiology, 103, 297– 321.
Elliott, R., Friston, K. J., & Dolan, R. J. (2000). Dissociable neural responses in human reward systems. Journal of Neuroscience, 20, 6159–6165.
Fitzgerald, P. B., Laird, A. R., Maller, J., & Daskalakis, Z. J. (2008). A meta-analytic study of changes in brain activation in depression. Human Brain Mapping, 29, 683–695.
Fox, M. D., & Raichle, M. E. (2007). Spontaneous fluctuations in brain activity observed with functional magnetic resonance imaging. Nat Rev Neurosci, 8(9), 700-711.
Fransson, P., & Marrelec, G. (2008). The precuneus/posterior cingulate cortex plays a pivotal role in the default mode network: Evidence from a partial correlation network analysis. NeuroImage, 42 (3), 1178-1184.
Frodl, T., Bokde, A. L., Scheuerecker, J., Lisiecka, D., Schoepf, V., Hampel, H., et al. (2010). Functional connectivity bias of the orbitofrontal cortex in drug-free patients with major depression. Biological Psychiatry, 67, 161–167.
Fujii, T., Okuda, J., Tsukiura, T., Ohtake, H., Miura, R., Fukatsu, R., et al. (2002). The role of the basal forebrain in episodic memory retrieval: A positron emission tomography study. NeuroImage, 15 (3), 501-508.
Greicius, M. D., Krasnow, B., Reiss, A. L., & Menon, V. (2003). Functional connectivity in the resting brain: A network analysis of the default mode hypothesis. Proc. Natl. Acad. Sci. 100, 253–258.
Greicius, M. D., Flores, B. H., Menon, V., Glover, G. H., Solvason, H. B., Kenna, H. et al. (2007). Resting-state functional connectivity in major depression: abnormally increased contributions from subgenual cingulate cortex and thalamus. Biological Psychiatry, 62, 429–437.
Grimm, S., Boesiger, P., Beck, J., Schuepbach, D., Bermpohl, F., Walter, M., et al. (2009). Altered negative BOLD responses in the default-mode network during emotion processing in depressed subjects. Neuropsychopharmacology, 34, 932–843. Guo, W., Sun, X., Liu, L., Xu, Q., Wu, R., Liu, Z., et al. (2011). Disrupted regional
homogeneity in treatment-resistant depression: A resting-state fMRI study. Progress in Neuro-Psychopharmacology and Biological Psychiatry, 35 (5), 1297-1302.
Gusnard, D. A., & Raichle, M. E. (2001). Searching for a baseline: Functional imaging and the resting human brain. Nat Rev Neurosci, 2, 685–692.
Van den Heuvel, M. P., & Hulshoff Pol, H. E. (2010). Exploring the brain network: a review on resting-state fMRI functional connectivity. Eur Neuropsychopharmacol, 20(8), 519-534.
Jonides, J., Schumacher, E. H., Smith, E. E., Koeppe, R. A., Awh, E., Reuter-Lorenz, P., et al. (1998). The role of parietal cortex in verbal working memory. (1998) J Neurosci, 18, 5026–5034.
Keedwell, P. A., Andrew, C., Williams, S. C., Brammer, M. J., Phillips, M. L. (2005). The neural correlates of anhedonia in major depressive disorder. Biological Psychiatry, 58, 843–853.
Kobayashi, Y., & Amaral, D. G. (2007). Macaque monkey retrosplenial cortex: III. Cortical efferents. J. Comp. Neurol., 502, 810–33.
Kumari, V., Mitterschiffthaler, M. T., Teasdale, J. D., Malhi, G. S., Brown, R. G.,
Giampietro, V., et al. (2003). Neural abnormalities during cognitive generation of affect in treatment-resistant depression. Biol Psychiatry, 54, 777–791.
Liu, Z., Xu, C., Xu, Y., Wang, Y., Zhao, B., Lv, Y., et al. (2010). Decreased regional
homogeneity in insula and cerebellum: a resting-state fMRI study in patients with major depression and subjects at high risk for major depression. Psychiatry Res, 182, 211–215. Lui, S., Wu, Q., Qiu, L., Yang, X., Kuang, W., Chan, R. C. K., et al. (2011). Resting-state
functional connectivity in treatment-resistant depression. Am. J. Psychiatry. 168, 642– 648.
Maddock, R. J., Garrett, A. S., & Buonocore, M. H. (2001). Remembering familiar people: the posterior cingulate cortex and autobiographical memory retrieval. Neuroscience, 104, (3), 667-676.
Martinez, J. M., Kent, J. M., Coplan, J. D., Browne, S. T., Papp, L. A., Sullivan, G. M., et al. (2001). Respiratory variability in panic disorder. Depress. Anxiety, 14: 232–237.
Mayberg, H. S. (1994). Frontal lobe dysfunction in secondary depression. J Neuropsychiatry Clin Neurosci, 6, 428–442.
Mayberg, H. S. (2003). Modulating dysfunctional limbic-cortical circuits in depression: Towards development of brain-based algorithms for diagnosis and optimised treatment. Br Med Bull, 65,193—207.
Mayberg H. S. (2009). Targeted electrode-based modulation of neural circuits for depression. J Clin Invest, 119, 717–725.
Mazoyer, B., Zago, L., Mellet, E., Bricogne, S., Etard, O., et al. (2001). Cortical networks for working memory and executive functions sustain the conscious resting state in man. Brain Res. Bull., 54, 287–98.
Meindl, T., Teipel, S., Elmouden, R., Mueller, S., Koch, W., Dietrich, O., et al. (2009). Test-retest reproducibility of the default-mode network in healthy individuals. Hum. Brain Mapp., 31, 237-46.
Mitterschiffthaler, M. T., Kumari, V., Malhi, G. S., Brown, R. G., Giampietro, V. P., Brammer, M. J., et al. (2003). Neural response to pleasant stimuli in anhedonia: an fMRI study. Neuroreport 14, 177–182.
Öngür, D., & Price, J. L. (2000). The organization of networks within the orbital and medial prefrontal cortex of rats, monkeys and humans. Cereb. Cortex, 10, 206–19.
Petrides, M., & Pandya, D. N. (1994). Comparative architectonic analysis of the human and the macaque frontal cortex. In F. Boller, & H. Spinnler (Eds.) Handbook of
Neuropsychology (pp. 17–58). Amsterdam: Elsevier Science B.V.
Phan, K. L., Fitzgerald, D. A., Nathan, P. J., Moore, G. L., Uhde, T. W., & Tancer, M. E. (2005). Neural substrates for voluntary suppression of negative affect: a functional magnetic resonance imaging study. Biol. Psychiatry, 57, 210–219.
Phillips, M. L., Drevets, W. C., Rauch, S. L., & Lane, R. (2003). Neurobiology of emotion perception I: the neural basis of normal emotion perception. Biological Psychiatry, 54, 504–514.
Raichle, M. E., MacLeod, A. M., Snyder, A. Z., Powers, W. J., Gusnard, D. A., et al. (2001). A default mode of brain function. Proc. Natl. Acad. Sci. U.S.A., 98, 676–82.
Ridgway, G. R., Omar, R., Ourselin, S., Hill, D. L., Warren, J. D., Fox, N. C. (2009). Issues with threshold masking in voxel-based morphometry of atrophied brains. Neuroimage, 44, 99–111.
Ritchey, M., Dolcos, F., Eddington, K. M., Strauman, T. J., & Cabeza, R. (2011). Neural correlates of emotional processing in depression: Changes with cognitive behavioral therapy and predictors of treatment response. Journal of Psychiatric Research, 45 (5), 577-587.
Rogers, R. D., Ramnani, N., Mackay, C., Wilson, J. L., Jezzard, P., Carter, C. S., et al. (2004). Distinct portions of anterior cingulate cortex and medial prefrontal cortex are activated by reward processing in separable phases of decision-making cognition. Biol.
Psychiatry, 55, 594–602.
Simpson, J. R. J., Drevets, W. C., Snyder, A. Z., Gusnard, D. A., & Raichle, M. E. (2001). Emotion-induced changes in human medial prefrontal cortex: II. During anticipatory anxiety. Proc. Natl Acad. Sci. USA 98, 688–691.
Simpson, J. R., Snyder, A. Z., Gusnard, D. A., & Raichle, M. E. (2001). Emotion-induced changes in human medial prefrontal cortex: I. During cognitive task performance. PNAS, 98 (2), 683-687.
Sobocki, P., Jonsson, B., Angst, J., & Rehnberg, C. (2006). Cost of depression in Europe. J. Ment. Health Policy Econ., 9 (2), 87–98.
Uddin, L. Q., Kelly, A. M. C., Biswal, B. B., Xavier Castellanos, F., Milham, M. P. (2009). Functional connectivity of default mode network components: correlation,
anticorrelation, and causality. Human Brain Mapping, 30, 625–637.
Vasic, N., Walter, H., Sambataro, F., Wolf, R. C. (2009). Aberrant functional connectivity of dorsolateral prefrontal and cingulate networks in patients with major depression during working memory processing. Psychol Med, 39, 977–987.
Vincent, J. L., Patel, G. H., Fox, M. D., Snyder, A. Z., Baker, J. T., Van Essen, D. C., et al. (2007). Intrinsic functional architecture in the anaesthetized monkey brain. Nature, 447, 83–86.
Wahlbeck, K., & Mäkinen, M. (2008). Prevention of depression and suicide. Consensus paper. Luxembourg: European Communities.
Wang, L., LaBar, K. S., Smoski, M., Rosenthal, M. Z., Dolcos, F., Lynch, T. R., et al. (2008). Prefrontal mechanisms for executive control over emotional distraction are altered in major depression. Psychiatry Res, 163(2), 143-55.
Wheeler, M. E., & Buckner, R. L. (2004). Functional-anatomic correlates of remembering and knowing. Neuroimage, 21, 1337–1349.
Zhou, Y., Yu, C., Zheng, H., Liu, Y., Song, M., Qin, W., et al. (2010). Increased neural resources recruitment in the intrinsic organization in major depression. J Affect Disord, 121, 220–230.
Zhu, X., Wang, X., Xiao, J., Liao, J., Zhong, M., Wang, W., et al. (2012). Evidence of a dissociation pattern in resting-state default mode network connectivity in first-episode, treatment-naive major depression patients. Biological Psychiatry, 71 (7), 611-617.
