Journal of Affective Disorders 253 (2019) 248–256

Contents lists available at ScienceDirect

Journal of Affective Disorders

journal homepage: www.elsevier.com/locate/jad

Research paper

Functional disconnectivity of the hippocampal network and neural T

correlates of memory impairment in treatment-resistant depression
Ruiyang Gea, Ivan Torresb, Jennifer J. Browna, Elizabeth Gregorya, Emily McLellana,
Jonathan H. Downarc,d, Daniel M. Blumbergerc,e, Zafiris J. Daskalakisc,e, Raymond W. Lamb,
⁎
Fidel Vila-Rodrigueza,
a
Non-Invasive Neurostimulation Therapies (NINET) Laboratory, Department of Psychiatry, University of British Columbia, 2255 Wesbrook Mall, Vancouver, BC V6T 2A1,
Canada
b
Department of Psychiatry, University of British Columbia, 2255 Wesbrook Mall, Vancouver, BC V6T 2A1, Canada
c
Department of Psychiatry, University of Toronto, Toronto, ON, Canada
d
MRI-Guided rTMS Clinic, Toronto Western Hospital, University Health Network, Toronto, ON, Canada
e
Temerty Centre for Therapeutic Brain Intervention, Campbell Family Research Institute, Centre for Addiction and Mental Health, Toronto, ON, Canada

A R T I C LE I N FO A B S T R A C T

Keywords: Background: Major depressive disorder (MDD) is a disabling neuropsychiatric condition associated with cogni-
Depression tive impairment. Neuroimaging studies have consistently linked memory deficits with hippocampal atrophy in
Hippocampus MDD patients. However, there has been a paucity of research examining how the hippocampus functionally
Resting state FMRI contributes to memory impairments in MDD. The present study examined whether hippocampal networks dis-
Subregion
tinguish treatment-resistant depression (TRD) patients from healthy controls (HCs), and whether these networks
Functional connectivity
RAVLT
underlie declarative memory deficits in TRD. We hypothesized that functional connectivity (FC) of the posterior
hippocampus would correlate preferentially with memory in patients, whereas FC pattern of the anterior and
intermediate hippocampus would correlate with emotion-mediated regions and show a significant correlation
with memory.
Methods: Resting-state functional magnetic resonance imaging (fMRI) scans were acquired in 56 patients and 42
age- and sex-matched HCs. We parcellated the hippocampus into three subregions based on a sparse re-
presentation-based method recently developed by our group. FC networks of hippocampal subregions were
compared between patients and HCs and correlated with clinical measures and cognitive performance.
Results: Decreased connectivity of the right intermediate hippocampus (RIH) with the limbic regions was a
distinguishing feature between TRD and HCs. These functional abnormalities were present in the absence of
structural volumetric differences. Furthermore, lower right amygdalar connectivity to the RIH related to a longer
current depressive episode. Declarative memory deficits in TRD were significantly associated with left posterior
and right intermediate hippocampal FC patterns.
Limitations: Our patient samples were treatment-resistant, the conclusions from this study cannot be generalized
to all MDD patients directly. Task-based imaging studies are needed to demonstrate hippocampal engagement in
the memory deficits of patients. Finally, our findings are strongly in need of replication in independent vali-
dation samples.
Conclusions: These findings demonstrate a transitional property of the intermediate hippocampal subregion
between its anterior and posterior counterparts in TRD patients, and provide new insights into the neural net-
work-level dysfunction of the hippocampus in TRD.

1. Introduction by the World Health Organization (World Health Organization, 2017).

The core psychopathological and diagnostic feature is depressed mood
Major depressive disorder (MDD) is a debilitating neuropsychiatric and its associated constellation of related symptoms (e.g. anhedonia or
disorder recently declared as the leading cause of disability worldwide suicidal ideation). While concentration difficulties are a diagnostic

⁎
Corresponding author.
E-mail address: fidelvil@mail.ubc.ca (F. Vila-Rodriguez).

https://doi.org/10.1016/j.jad.2019.04.096
Received 17 December 2018; Received in revised form 29 January 2019; Accepted 27 April 2019
Available online 29 April 2019
0165-0327/ © 2019 Elsevier B.V. All rights reserved.
R. Ge, et al. Journal of Affective Disorders 253 (2019) 248–256

criterion, broader cognitive impairment is a frequent and disabling i.e., the intermediate subregion may be involved in memory and cog-
feature in depression (Lee et al., 2012; Porter et al., 2003; Reppermund nitive processes, as well as in emotional functions.
et al., 2009), and it appears to be independent of depressed mood In the present study, we aimed to investigate whether there are
(Rock et al., 2014). Specifically, impairments in cognitive domains such differences in the hippocampal networks between patients with TRD
as learning and memory, attention, and executive functioning are often and healthy controls (HCs), and whether these networks underlie the
present in MDD and have been shown to contribute to poor functional declarative memory deficits in TRD. Specifically, the following hy-
outcomes (Lee et al., 2012; Porter et al., 2003). Dysfunction in learning potheses were tested: (1) Abnormal FC patterns between the anterior
and memory is of particular concern as it has been reported to occur and intermediate hippocampus and emotion-mediated regions (e.g.,
even in the absence of other neurocognitive deficits (Austin et al., 2001; anterior hippocampus, amygdala and orbitofrontal regions) will be
Sweeney et al., 2000), tends to persist despite remission of mood and present; (2) abnormal FC patterns of the intermediate hippocampus will
affective symptoms (Maeshima et al., 2016; Neu et al., 2005; be associated with both clinical variables as well as with declarative
Reppermund et al., 2009), and may be associated to the increase risk of memory deficits; (3) abnormal FC patterns of the posterior hippo-
dementia later in life (Potter et al., 2013). Furthermore, there are some campus will be associated to declarative memory deficits only.
data pointing to premorbid memory impairment as a predictive feature
for subsequently developing MDD (Mannie et al., 2009). Treatment-
2. Materials and methods
resistant depression (TRD) represents MDD patients who do not respond
to adequate trials of antidepressant medications (Sackeim, 2001). The
2.1. Participants
investigation of cognitive impairments in TRD has received little at-
tention, and a better understanding of its underlying neurobiology in
Sixty-two TRD patients and 42 HCs were recruited for clinical trials
TRD may provide important clues about the mechanism of action of
using repetitive transcranial magnetic stimulation (rTMS) (Blumberger
therapeutic interventions that modify the hippocampus (e.g. electro-
et al., 2018; Gennatas et al., 2017). All MRI scans used in this study
convulsive therapy).
were baseline scans acquired before any rTMS administration; however,
Neuroimaging studies investigating the neural substrates of learning
the inclusion and exclusion criteria reflect the requirements of the rTMS
and memory deficits in MDD have focused mostly on the structural
trials. The inclusion and exclusion criteria of patients and HCs are
abnormalities of the hippocampus. Specifically, research employing
outlined in supplementary material Table S1 and Table S2. Specifically,
neuroimaging techniques has shown that MDD patients’ memory dys-
patients were recruited if they had a moderate to severe depressive
function is associated with hippocampal atrophy (Hickie et al., 2005;
episode, had not responded to at least one adequate trial of anti-
Jayaweera et al., 2016). However, there has been a relative paucity of
depressant medication (but less than four trials) or were unable to
research examining how hippocampus functionally contributes to
tolerate at least two separate trials of antidepressants of inadequate
learning and memory impairments in MDD.
dose and duration. One patient was absent from the resting-state fMRI
The human hippocampal formation is a fundamental part of the
scanning. Participants who had excessive head movement during the
limbic system and is considered to play an important role in MDD due
scan (head motion > 3 mm or 3° (Dai et al., 2014; Tamm et al., 2006;
to its connection with other limbic and paralimbic structures (Catani
Yao et al., 2013), n = 4), or had severe artifacts in the functional data
et al., 2013; Lewis and Shute, 1967; Morgane et al., 2005). The hip-
induced by orthodontic hardware (n = 1) were excluded. In order to
pocampus is also involved in the default mode network (DMN), which is
rule out the potential confounding effect of head motion on resting-
dysfunctional in MDD (Whitfield-Gabrieli and Ford, 2012). Several
state connectivity, the mean frame-wise displacement (FD)
functional magnetic resonance imaging (fMRI) and positron emission
(Power et al., 2012) value was used as a nuisance regressor in the fol-
tomography (PET) studies have demonstrated dysfunction and treat-
lowing statistical analyses. The final sample included in the analyses
ment modulation of the hippocampus in MDD (Goldapple et al., 2004;
consisted of 56 patients and 42 HCs (Table 1). Note that all of the
Gong et al., 2018; Kennedy et al., 2001; Milne et al., 2012; Zhang et al.,
participants were native English speakers, except three in the control
2011). Although most of these studies have regarded the hippocampus
group.
as a singular and unitary region, the hippocampus is a functionally
The protocol was approved by the Clinical Research Ethics Board of
heterogeneous region along its anterior-posterior axis (Strange et al.,
the University of British Columbia, and the Vancouver Coastal Health
2014). In rodent models, gene expression patterns define at least three
Authority, and written informed consent was obtained from all parti-
longitudinal subregions: ventral, middle, and dorsal subregions, which
cipants. The studies were registered at clinicaltrials.gov (NCT01887782
are homologous to the human anterior, intermediate, and posterior
and NCT02800226).
subregions (Dong et al., 2009; Strange et al., 2014). The ventral/ante-
rior subregion is involved in processing emotional responses, and the
Table 1
dorsal/posterior subregion mediates memory and cognitive functions,
Sample demographics, clinical and cognitive characteristics for the two groups.
while the specific function of the middle/intermediate subregion is less
understood (Fanselow and Dong, 2010; Strange et al., 2014). The in- Patients Healthy volunteers p
(n = 56) (n = 42)
vestigation of these fine-grained subregions may improve our under-
standing about the role of the hippocampus in the pathophysiology of Gender (M/F) 24/32 17/25 0.81 a

b
MDD. Specifically, prominent deficits in memory, particularly in hip- Age (SD) 43.02 (11.87) 40.36 (13.20) 0.30
b
pocampal-mediated declarative memory, support the disruption of Educational level (SD) 15.04 (2.31) 16.00 (2.14) 0.04
Handedness (L/R/A) 8/47/1 3/37/2 –
hippocampal function in MDD (Jayaweera et al., 2016; Lee et al.,
AED [mg (SD)] 19.48 (18.48) –
2012). At a system level, therefore, a correlation between memory HDRS (SD) 22.09 (3.92) – –
performance and hippocampal functional connectivity (FC) patterns NAART (SD) 114.61 (5.49) 111.27 (9.66) 0.05 b
would be predicted in MDD patients, particularly in the posterior hip- RAVLT1-5 (SD) 48.59 (9.82) 54.40 (7.29) 0.002 b
pocampus (Fanselow and Dong, 2010; Strange et al., 2014). However, RAVLT-7 (SD) 10.07 (3.18) 11.21 (2.20) 0.04b

evidence points to a model whereby the patterns of connectivity be-

tween the hippocampus and cortical/subcortical regions along the
longitudinal axis transit in a gradual fashion rather than being orga-
nized in a discrete way (Amaral and Witter, 1989; Strange et al., 2014).
Therefore, it is plausible that the intermediate subregion exhibits
transitional properties between the anterior and posterior subregions,
Abbreviations: A, ambidextrous; AED, equivalent escitalopram dose; F, female;
HAMD, 17-item Hamilton Depression Rating Scale; L, left; M, male; NAART:
National Adult Reading Test; R, right; RAVLT, Rey Auditory Verbal Learning
Test; SD, standard deviation.
a
Chi-square test.
b
Two-sample t test for patients and healthy volunteer groups.

249
R. Ge, et al.
2.2. Clinical and cognitive outcome measures
Symptom severity was assessed with the 17-item Hamilton
Depression Rating Scale (HDRS) (Hamilton, 1960; Williams, 1988) at
baseline. Neuropsychological assessments were conducted by research
personnel, who were trained and supervised by a registered clinical
neuropsychologist (I.T.). In the present study, the hippocampal-medi-
ated cognitive outcome measures included declarative learning and
memory, assessed by the Rey Auditory Verbal Learning Test (RAVLT)
immediate recall (summed score over five trials, RAVLT-1-5) and de-
layed recall (trial 7, RAVLT-7) (Strauss et al., 2006). The premorbid IQ
was estimated using the North-American Adult Reading Test (NAART)
(Blair and Spreen, 1989).
2.3. Imaging data acquisition and preprocessing
Imaging was performed at the University of British Columbia's MRI
Research Centre on a Philips Achieva 3.0T scanner. During the scan-
ning, all participants were asked to keep still with their eyes open and
to try not to think of anything systematically. A total of 300 vol of echo
planar images were obtained with the following parameters:
TR = 2000 ms, TE = 30 ms, flip angle (FA) = 90°, field of view
(FOV) = 220 mm × 220 mm × 155 mm, acquisition matrix = 64 × 64,
in-plane resolution = 2.75 mm × 2.75 mm and slice thickness = 5 mm
with 1 mm slice gap. Twenty-six axial slices parallel to the AC-PC line
were obtained in an interleaved bottom-up order to effectively cover
the whole brain. High-resolution T1-weighted images were also ac-
quired with the following parameters: 165 axial slices, TR = 8.1 ms,
TE = 3.5 ms, FA = 8°, FOV = 256 mm × 256 mm × 165 mm, acquisi-
tion matrix = 256 × 250, in-plane resolution = 1 mm × 1 mm and
slice thickness = 1 mm.
Functional images were preprocessed using SPM12 software
(http://www.fil.ion.ucl.ac.uk/spm) and DPABI (Yan et al., 2016).
Briefly, the first ten functional volumes were discarded to allow for
stabilization of the initial signal and adaptation of the participants to
the circumstances. The remaining fMRI images were then slice-timing
corrected and realigned. Head displacement across the resting scan did
not differ significantly between the two groups (expressed as the mean
FD values, p = 0.370). The individual T1-weighted structural image
was co-registered to the mean functional image after motion correction
using a linear transformation. The transformed structural images were
then segmented into gray matter (GM), white matter (WM), and cere-
brospinal fluid (CSF) by using a unified segmentation algorithm
(Ashburner and Friston, 2005). The motion corrected (realigned)
functional images were spatially normalized to the Montreal Neurolo-
gical Institute (MNI) space using the normalization parameters esti-
mated during unified segmentation. The functional data were then re-
sampled to 3-mm isotropic voxels and smoothed with a 6-mm full width
at half maximum (FWHM) Gaussian kernel. Linear trends and nuisance
signals (Friston-24 head motion parameters, CSF, and WM signals with
the SPM's a priori tissue probability maps) (Friston et al., 1996) were
then removed, and temporal band-pass filtering (0.01–0.08 Hz) was
performed. The residuals were used for the following analyses. Note
that the global signal regression (GSR) step in fMRI preprocessing is
controversial: the implementation of GSR could introduce spurious
negative correlations in functional connectivity studies (Murphy et al.,
2009; Murphy and Fox, 2017), and there is growing evidence that the
global signal contains diagnostic information in clinical populations
(Hahamy et al., 2014; Yang et al., 2014). Thus we opted for not using
this step or its alternative (i.e., CompCor) (Behzadi et al., 2007;
Muschelli et al., 2014), because while CompCor does not explicitly
include GSR, the practical results of its application are quite similar
(Ciric et al., 2017).
250
Journal of Affective Disorders 253 (2019) 248–256
2.4. fMRI data analysis
2.4.1. Functional parcellation of the hippocampus
The bilateral hippocampi were extracted from the Harvard-Oxford
subcortical structural atlas (Caviness et al., 1996; Kennedy et al., 1998;
Smith et al., 2004) (http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/Atlases) with a
threshold of 25% minimum probability. The Harvard-Oxford atlas
provides probabilities that each voxel falls into a particular anatomical
region. Each voxel was assigned to the most likely region, as long as it
had a likelihood of 25% or greater (Craddock et al., 2012; Salomons
et al., 2016; Schiller et al., 2013; Smoski et al., 2013). We divided the
hippocampus, along its longitudinal axis, into three sections (Chen and
Etkin, 2013; Strange et al., 2014). Specifically, hippocampi were par-
cellated into three sub-divisions on each hemisphere, (i.e., left/right
anterior hippocampus (LAH/RAH), left/right intermediate hippo-
campus (LIH/RIH), and left/right posterior hippocampus (LPH/RPH)
using a sparse representation-based method (Ge et al., 2017). Specifi-
cally, we used a robust parcellation approach, which constructs a sparse
similarity graph based on the sparse representation coefficients be-
tween hippocampal voxels and then uses a spectral clustering algorithm
to identify distinct subregions (see supplementary material for details of
this parcellation approach).
2.4.2. Comparison of functional organizations of hippocampal subregions
between groups
One aim of the present study was to assess the potential differences
in the functional organizations of hippocampal subregions between
TRD patients and HCs. In the present study, functional organization was
measured as the parcel size (number of voxels) in each hippocampal
subregion (Nebel et al., 2014; Yamada et al., 2016). Therefore, we
examined between-group differences of the volumes of specific sub-
regions by performing permutation tests (5000 times permutations
using an in-house in-house MATLAB17a program which is available
upon request), as previously described (Ge et al., 2017). Specifically, we
randomly assigned participants to one of the two groups by permuting
the labels of all the subjects and applied our parcellation method
(Ge et al., 2017) to each of the permuted groups and repeated this
procedure 5000 times. Parcel size differences between corresponding
subregions from each permuted group were calculated and null dis-
tribution of between-group parcel size differences was generated. We
deemed between-group parcel size difference statistically significant if
it falls above the 95th percentile of the relevant empirical null dis-
tribution.
2.4.3. Functional connectivity of the hippocampal subregions
Regions of interest (ROIs, i.e., LAH, LIH, LPH, RAH, RIH and RPH
seeds) were constructed from group-level parcellation results, and the
whole-brain correlation maps were produced by extracting the mean
time series from each hippocampal subregion and computing the
Pearson's correlation coefficient between that time series and time
series from all other brain voxels. Correlation values were then con-
verted into z-scores using a Fisher transformation. The functional
connectivity analysis was confined to the SPM a priori probabilistic GM
mask, with a minimum threshold of 0.25. For the group comparisons,
two-sample t-tests were performed for each ROI-related correlation
map, respectively. The significance level for each comparison was set to
α< 0.05/6 = 0.008 (Bonferroni correction was applied here to correct
for multiple comparisons of the 6 subregions), with a cluster level fa-
mily-wise error (FWE) correction and height threshold of p< 0.005.
To investigate the correlation between the clinical outcome, cog-
nitive outcomes and the FC of the hippocampal subregions, regression
analyses for the HDRS, length of current episode, length of lifetime
episode, RAVLT1-5, RAVLT-7, and NAART scores were conducted. We
first confined our regression analysis using small-volume correction to
the regions that survived the between-group comparisons, we then
analyzed the whole brain. The significance level for each regression
R. Ge, et al.
map was set to α< 0.05/72 = 0.0007 (Bonferroni correction was ap-
plied to correct for multiple comparisons of the 6 subregions × 6 re-
gressors × 2 [one small-volume correction, and one whole brain cor-
rection]) with cluster level FWE correction and height threshold
p< 0.001. To summarize the extent to which the significant clusters
correlated to the neurocognitive profiles, the averaged z-score values
across voxels within the significant clusters obtained from the regres-
sion analyses were extracted for patients and HCs, respectively, and
correlated with the neurocognitive outcomes. All aforementioned sta-
tistical tests were conducted with age, sex, educational level, handed-
ness, mean FD values, and mean GM volume over the subregion as
nuisance regressors.
2.5. Supplementary analyses
Patients with MDD have previously been reported to have smaller
hippocampus (Bremner et al., 2000; McKinnon et al., 2009) (but see
(Vythilingam et al., 2004)), here we examined whether changes in gray
matter volume within the hippocampus, as measured by voxel-based
morphometry (VBM), may contribute to the functional connectivity
alterations observed in our TRD patients. VBM (Ashburner and
Friston, 2000) was performed using the VBM8 toolbox. The individual
T1-weighted images from all subjects were first segmented into GM,
WM, and CSF, using a unified segmentation routine. GM population
templates were generated from the entire image dataset using the Dif-
feomorphic Anatomical Registration Through Exponentiated Lie Al-
gebra (DARTEL) technique (Kurth et al., 2015). After an initial affine
registration of the GM DARTEL templates to the tissue probability maps
in MNI space, non-linear warping of GM images was performed to the
DARTEL GM template in MNI space. Images were then modulated and
smoothed with a 6-mm FWHM Gaussian kernel. The volumes of interest
corresponding to the modulated (Chen and Etkin, 2013; Gennatas et al.,
2017) images of the bilateral hippocampal subregions were extracted
by using the “spm_read_vols” function in SPM12 for the between-group
comparisons by two-sample t-tests. Specifically, we extract the GMV
(gray matter volume) values of voxels within each hippocampal sub-
region, and the average GMV value of these voxels was used for the
comparisons between groups.
As a supplementary analysis, we re-analyzed our data without any
nuisance variables added to the statistical models to mitigate the risk
that the present results were driven by these nuisance variables
(Gennatas et al., 2017).
3. Results
3.1. Demographic and cognitive outcomes
Patients and HCs did not differ significantly in age, sex (Table 1),
and head motion (expressed as FD, mean (SD), patients vs. HCs: 0.16
(0.09) vs. 0.14 (0.05), p = 0.34). There was a significant difference
between groups for highest level of education (Table 1). On average,
depressive symptom severity was in the moderate to severe range
(mean HDRS = 22.09, SD = 3.92, range 14–32) for depressed subjects.
There were no significant differences between the two groups with
regards to the predicted IQ using NAART. Patients scored lower on the
immediate (RAVLT1-5) and delayed recall (RAVLT-7) compared to HCs
(Table 1).
3.2. Functional parcellation of hippocampus
The unilateral hippocampus was parcellated, along its longitudinal
axis, into three sections (Fig. 1a), i.e., the left/right anterior hippo-
campus (LAH/RAH), left/right intermediate hippocampus (LIH/RIH),
and left/right posterior hippocampus (LPH/RPH) hereafter. The left
and right anterior subregions occupied 35% and 39% of the entire left
and right hippocampus; the left and right intermediate subregions
251
Journal of Affective Disorders 253 (2019) 248–256
Fig. 1. (a) Illustration of the hippocampal subregions using sparse representa-
tion-based parcellation method. (b) Brain regions whose functional con-
nectivity (FC) with the RIH showed significant difference between patients and
healthy controls. LAH/RAH, left/right anterior hippocampus; LIH/RIH, left/
right intermediate hippocampus; LPH/RPH, left/right posterior hippocampus.
occupied 38% and 38% of the entire left and right hippocampus; and
the left and right posterior subregions occupied 27% and 23% of the
entire left and right hippocampus, respectively. The correspondence
between the subregions and anatomical subfields of the hippocampus
can be found in the supplement (Supplement Fig. S2). Supplement Fig.
S3 summarizes the results of the permutation tests on the functional
organization of the hippocampal subregions. The results revealed no
significant differences in the organizational arrangement of each sub-
region between patients and HCs. In the supplementary material, we
presented the FC patterns of the subregions (supplement Fig. S5).
3.3. Differences of the hippocampal FC patterns between patients and HCs
The clusters show lower FC of the RIH seed in patients relative to
HCs with the bilateral anterior hippocampus, the parahippocampus, the
amygdala, and the orbitofrontal gyrus (Fig. 1b and Table 2).
3.4. Correlation between hippocampal FC with clinical and cognitive
outcomes
Within regions that exhibited significant between-group difference,
small-volume correction revealed that lower right amygdalar con-
nectivity to the RIH correlated with increased duration of current epi-
sode (p< 0.001, Fig. 2 and Table 2). No significant correlations were
found between the hippocampal FC and other clinical (including the
HDRS scores) and neurocognitive measurements within regions that
exhibited significant between-group differences.
The regression analysis in the whole brain level showed that better
delayed recall (RAVLT-7) scores were significantly associated with
lower FC patterns, using the LPH and RIH as the seeds (Table 2). For the
FC pattern of LPH, the significant clusters located at the bilateral dorsal
medial prefrontal cortex (DMPFC, primarily in the right hemisphere)
(r= −0.44, p = 1.03 × 10−5), the right dorsolateral prefrontal cortex
(DLPFC) (r= −0.44, p = 1.03 × 10−5) and the right posterior parietal
cortex (rPPC) (r= −0.43, p = 1.98 × 10−5; Fig. 3a) were significantly
negatively-correlated with RAVLT-7 scores. No FC patterns of the other
subregions showed significant association with the delayed recall or
immediate recall scores. However, if we lowered the multiple com-
parisons correction threshold to cluster level FWE correction p < 0.05
and height threshold p < 0.001, uncorrected for the 72 multiple
comparisons, the RIH FC pattern showed negative correlation with the
delayed recall scores. Specifically, the clusters located at the bilateral
DMPFC (primarily in the right hemisphere, r= −0.43,
p = 1.74 × 10−5) and right DLPFC (r= −0.40, p = 7.56 × 10−5;
R. Ge, et al. Journal of Affective Disorders 253 (2019) 248–256

Table 2
Significant clusters revealed in the two-sample t-tests and regression analysis.
Contrast Cluster Location MNI coordinates Cluster Size Peak
x y z K t

PTs < HCs 1 Left inferior frontal cortex −18 18 −24 100 4.53
Seed: RIH (Fig. 1) Left amygdala −24 −6 −21 3.52
2 Right hippocampus 27 −9 −18 120 5.39
Right amygdala 21 −6 −18 3.79
Association with length of current episode 1 Right amygdala 21 −3 −18 25 4.11
Seed: RIH (Fig. 2)
Association with RAVLT7 1 Bilateral DMPFC 3 45 45 190 4.77
Seed: LPH (Fig. 3) 2 Right DLPFC 39 15 42 183 4.84
3 Right PPC 51 −54 42 215 5.09
Association with RAVLT7 1 Bilateral DMPFC 3 54 36 57 4.06
Seed: RIH (Fig. 3) 2 Right DLPFC 39 18 42 55 4.03

Note. DMPFC, dorsomedial prefrontal cortex; DLPFC, dorsolateral prefrontal cortex; PPC, posterior parietal cortex.

3.5. Supplementary analyses

We examined the GMV in each hippocampal subregion, and found

Fig. 2. Right intermediate (RIH) hippocampal connectivity to the right amyg-
dala was negatively correlated with length of current episode. Residual hip-
pocampal connectivity and residual episode length (after correcting for nui-
sance variables) were used in the scatterplots. The r values represent partial
correlation coefficient. Note that there was an outlier point (marked with a
white arrow), however, the correlation remained significant after excluding the
outlier (p< 0.001, upper right panel).
Fig. 3b) showed significantly negative correlation with RAVLT-7 scores.
To determine the relationship between the hippocampal FC and RAVLT
scores for each group separately, we calculated the partial correlation
coefficients between the z-scores, averaged across the voxels within the
clusters and the corresponding RAVLT scores for each group, respec-
tively. Results revealed that the delayed recall scores were significantly
anti-correlated with the RIH hippocampal FC pattern in both patients
and HCs. Similarly, the delayed recall scores were negatively correlated
with LPH FC pattern in patients significantly (p< 0.001 for both frontal
clusters and parietal cluster), and the anti-correlation was marginally
significant in HCs (p = 0.07 for the right DLPFC cluster and p = 0.08
for the parietal cluster). To further explore whether the relationship
between hippocampal FC patterns differed by clinical status, post-hoc
exploratory 2-way continuous covariate interaction analyses were
conducted within the regions that showed significant correlation with
the RAVLT-7 scores (see supplementary material for details). The re-
sults (supplement Fig. S6) showed that for the LPH FC pattern, there
were two sub-clusters (within the right DLPFC and right PPC clusters)
that showed significant difference of the correlation to the RAVLT-7
scores between TRD patients and HCs (supplement Fig. S6a). For the
RIH FC pattern, there was a sub-cluster (within the right DLPFC cluster)
that showed significant difference of the correlation to the RAVLT-7
scores between TRD patients and HCs (supplement Fig. S6b).
There was no significant correlation/anti-correlation of the clinical
measures and NAART scores with hippocampal FC in any regions in the
entire brain. Sensitivity analyses excluding the three HC who were not
native English speakers did not change results.
no significant difference between groups (see supplement Fig. S4).
Additionally, the GMV in these subregions did not account for group
differences in the RIH connectivity, suggesting that the GM volume did
not likely account for the group differences in hippocampal FC patterns
observed in the present study. When the analyses were repeated
without inclusion of the nuisance variables, we found that the observed
group differences of the RIH FC pattern between patients and HCs re-
mained significant (Fig. S7), and the negative correlation between the
RIH FC pattern and the duration of current depressive episode remained
significant (Fig. S8). The correlations between the LPH FC pattern and
RIH FC pattern with the RAVLT-7 scores remained (Fig. S9). These
results suggested that the present results were not mainly driven by the
nuisance variables.
Furthermore, we have looked at the effects of antidepressant med-
ication on the present results. We calculated antidepressant equivalent
doses (AED) using previously defined ratios from the literature (Ge
et al., 2019; Hayasaka et al., 2015; Inada and Inagaki, 2015), and added
this normalized dose into the statistical analyses. The results showed
that including AED as a nuisance variable did not significantly change
the present results, suggesting that the present results were unlikely a
result of different antidepressant medication levels.
4. Discussion
The purpose of this study was to examine the differences in hip-
pocampal networks between TRD patients and HCs, and to investigate
whether these networks underlie declarative memory and clinical
profiles observed in TRD. We provide evidence for functional organi-
zations of the human hippocampal subregions in both TRD and HCs by
using a novel and robust brain parcellation method. As predicted, we
demonstrated a distinctive pattern of RIH connectivity within the
anterior hippocampus, parahippocampus, the amygdala, the putamen
and the orbitofrontal gyrus between the two groups, in the absence of
structural and functional organization changes of the hippocampal
subregions. Further, we found that lower RIH-amygdala functional
connectivity was related to the duration of current depressive episode.
Significant anti-correlations were found between FC pattern of LPH,
RIH and neurocognitive memory performance, suggesting a role for the
posterior and intermediate hippocampus in memory deficits of TRD
patients.
4.1. Hippocampal volume and functional organization are similar in TRD
and HCs
Although decreased hippocampal volume has been shown in TRD
patients (McKinnon et al., 2009), this finding has not been universally

252
R. Ge, et al.
Fig. 3. Hippocampal functional connectivity (FC) correlates with cognitive outcome measures. (a) Left posterior (LPH) hippocampal connectivity to the dorsal medial
prefrontal cortex (DMPFC), the right dorsolateral prefrontal cortex (DLPFC) and the right posterior parietal cortex (rPPC) was negatively correlated with the delayed
recall (RAVLT-7) scores. (b) Right intermediate (RIH) hippocampal connectivity to the DMPFC and right DLPFC was negatively correlated with the delayed recall
(RAVLT-7) scores. Residual hippocampal connectivity and residual recall performance (after correcting for nuisance variables) were used in the scatterplots to show
linear relationship between hippocampal connectivity and recall performance in each group. The r values represent partial correlation coefficient controlling for
nuisance variables.
replicated (Vythilingam et al., 2004). Our data are convergent with the
latter, as we did not observe significant difference in the GM volume
between TRD and HCs for all subregions. A plausible explanation of the
discrepancies is that the studies revealing differences were mostly based
on the measurement of hippocampal volume size (Kempton et al., 2011;
McKinnon et al., 2009; Schmaal et al., 2016; Tae et al., 2008), whereas
we compared the GM volume which is the “modulated” GM con-
centration (Gennatas et al., 2017) within an a priori-defined hippo-
campal mask. Specifically, GMV employed in the present study quan-
tifies the amount of gray matter existing in a voxel, and it is a voxel-
based measurement. In contrast, the volume of hippocampus used in
other studies is a measurement of the hippocampal size, rather than the
intensity of the hippocampal voxels. Compared to the large amount of
studies which found smaller hippocampal size in MDD patients
(Kempton et al., 2011; Schmaal et al., 2016), there is relatively a lack of
evidence that demonstrates the lower GMV of hippocampus in MDD
patients (Bora et al., 2012), and our results were convergent with the
latter. We further demonstrated normal functional organizations of all
hippocampal subregions in patients compared to HCs.
4.2. Decreased subregional hippocampal connectivity in TRD
Contrary to one of our hypotheses, we did not observe dysfunctional
connectivity of the bilateral anterior hippocampus subregions in pa-
tients relative to HCs. However, a major finding of our study was the
hypoconnectivity of the FC profile of the RIH subregion with the
anterior hippocampus, the parahippocampus, amygdala, and orbito-
frontal gyrus in TRD patients, suggesting that hippocampal FC dys-
function is a distinguishing feature of TRD in the absence of detectable
structural (Vythilingam et al., 2004) and functional (Ge et al., 2017)
organization changes. Our findings of the RIH FC patterns echo the
results that have demonstrated the key role of the right hippocampus in
MDD (Campbell et al., 2004; Videbech and Ravnkilde, 2004). The bi-
lateral anterior hippocampal subregions in the present study mainly
253
Journal of Affective Disorders 253 (2019) 248–256
located at the CA and subiculum subfields, rather than the dentate
gyrus. The dentate gyrus is believed to be the primary site of newly
developing neurons (Eriksson et al., 1998; Ho et al., 2013), and it has
been correlated with memory performance in healthy older adults
(Engvig et al., 2012). More importantly, the dentate gyrus has been
associated with the impairments in memory functioning in MDD pa-
tients (Bremner et al., 2004; MacQueen and Frodl, 2011). Considering
the key role of dentate gyrus in depression, the negative findings of the
anterior hippocampal regions in the present study might due to this
subregion did not overlap with the dentate gyrus.
The functional role of the hippocampus remains a topic of much
debate (Poppenk et al., 2013; Strange et al., 2014). Given the evidence
of gene expression, behavioral domains and anatomical connectivity
(Amaral and Witter, 1989; Chase et al., 2015; Dong et al., 2009), a
provisional model of anterior, intermediate and posterior thirds was
proposed (Strange et al., 2014). This model suggests that the anterior
portion of hippocampus is more engaged in limbic processes (i.e., “hot”
processes), and the posterior portion is preferentially enrolled in cog-
nitive functions such as spatial navigation and declarative memory (i.e.,
“cold” processes) (Fanselow and Dong, 2010; Robinson et al., 2016),
while the function of the intermediate portion remains poorly char-
acterized. Given the role of the anterior hippocampus in affective
processing, one might expect to discover dysfunction of the anterior
hippocampal FC patterns. However, we found no significant differences
in LAH and RAH-related FC patterns. Instead, our results provide evi-
dence that the intermediate portion may play a role in the emotional
deficits of TRD, indicated by a hypoconnectivity with the limbic regions
(i.e., the hippocampal formation and the amygdala) and the orbito-
frontal gyrus that are intensively implied in emotional processing
(Fanselow and Dong, 2010; Phan et al., 2002; Strange et al., 2014) and
may be impaired in MDD (Ballmaier et al., 2004; Drysdale et al., 2017;
Hamani et al., 2011; Lui et al., 2011; Matthews et al., 2008).
The network of decreased RIH-amygdala-frontal connectivity
emerging from the present study corresponds well to the limbic
R. Ge, et al.
compartment of the limbic-cortical model of MDD (Goldapple et al.,
2004; Mayberg, 2003). In support of this model, our findings further
demonstrate that the intermediate hippocampus showed disrupted FC
with other limbic regions in our samples. We also found that a lower
RIH-amygdala connectivity was related to a longer duration of current
depressive episode in TRD. This phenomenon was in line with the
functional hypoconnectivity between RIH and the amygdala in the
patient group relative to the healthy controls, suggesting that RIH-
amygdala connectivity is reflective of the clinical profile of the current
episode length. Note that though bilateral amygdala showed decreased
FC with the RIH, the decreased FC pattern was primarily located at the
right amygdala. Therefore, it is likely that the right lateralized amyg-
dala-RIH correlation with the clinical data was due to the pronounced
hypoconnectivity of the right amygdala. No significant correlation was
found between hippocampal networks and other clinical variables ex-
cept for the length of current episode, indicating that the abnormal FC
pattern in amygdala may be state-dependent associated to the current
depressive episode. In other words, the RIH-amygdalar connectivity
was related to the length of the current depressive episode might be
signaling that abnormal FC pattern might be a state marker which
fluctuates with current episode and presents only during the acute stage
of episode (Graham et al., 2013; Maalouf et al., 2011). The lack of
correlation of FC patterns with other measures such as degree of
treatment resistance, age of onset, or prior number of episodes would
increase the plausibility of such interpretation. This hypothesis, how-
ever, would warrant further research to test this interpretation of cur-
rent findings. Moreover, the missing link between abnormal FC to the
cognitive impairment may suggest that cognitive impairment being an
independent component from mood disability.
4.3. Declarative memory deficits are related to intrinsic functional
connectivity patterns
Indeed, we demonstrated a declarative memory deficit in TRD pa-
tients. Specifically, we found impairment in immediate declarative
memory, as well as in delayed memory function, as indexed by the
RAVLT. Somewhat opposed to this, others have demonstrated pre-
served verbal memory (Porter et al., 2003). Thus, factors that might
contribute to these disparate findings, such as depression subtype,
sample characteristic, severity and neurocognitive tests methodologic
issues, should be considered in future studies. In addition, our data
showed that these declarative memory impairments were associated
with patterns of functional connectivity between hippocampal sub-
regions and areas of the brain not overlapping with between-group FC
differences. Our finding of correlates between the posterior hippo-
campal FC pattern and declarative memory performance in the patients
as well as in the healthy controls is in keeping with cognitive, genetic
and clinical evidence relating learning and memory with this structure
(Fanselow and Dong, 2010; Jayaweera et al., 2016; Strange et al.,
2014). Importantly, we demonstrated that the FC pattern of the inter-
mediate hippocampus was also related to memory performance. Taken
together, the evidence from the intermediate hippocampus (vide supra)
demonstrates a dual property of this region (Fanselow and Dong, 2010).
The intermediate hippocampus, which has partially overlapping char-
acteristics with its anterior and posterior neighbors, may serve as a
transitional zone between the anterior and posterior portions, as de-
monstrated by showing hypoconnectivity with emotion-mediated re-
gions and significant correlates with memory performance in TRD pa-
tients.
For intermediate and posterior FC patterns, regions that showed
correlates with memory performance were located within the right
frontoparietal central executive network (CEN, included the DLPFC, the
DMPFC and the posterior parietal cortex) (Hamilton et al., 2013;
Mulders et al., 2015; Seeley et al., 2007). The CEN is a network of
structures that increases in activation during the performance of at-
tention-demanding tasks. The CEN and DMN are often seen as opposing
254
Journal of Affective Disorders 253 (2019) 248–256
networks; these two networks usually show anti-correlation, and the
disrupted interactions between them may contribute to the pathology of
a variety of mental disorders (Menon, 2011), including MDD
(Manoliu et al., 2014). Because the hippocampus is an important node
of the DMN (Buckner et al., 2008), it is plausible that the higher
functional connectivity between the hippocampus (representative of
the DMN) and the DLPFC, the DMPFC and the posterior parietal cortex
(representatives of the CEN) was detrimental to the cognitive proces-
sing, thereby producing the lower behavioral performance in RAVLT
tests that was discovered in our samples. The functional connectivity
patterns (to both LPH and RIH seeds) that related to the RAVLT-7 scores
were located at a set of right-lateralized regions, including right DLPFC,
PPC, and DMPFC. This right-lateralized pattern could be considered
consistent with the explanation that the right central executive network
(CEN) interacts with the default mode network (DMN, with hippo-
campus as a representative node) to support memory recall (Fornito
et al., 2012). A potential implication is that enhancing the hippo-
campus-related memory functions of the TRD patients targeting the
right lateralized CEN nodes may be an avenue to address hard-to-treat
cognitive impairment in TRD. Non-invasive neurostimulation treat-
ments such as repetitive transcranial magnetic stimulation (Wang et al.,
2014) or magnetic seizure therapy (Wang et al., 2018) have been re-
ported to accomplish such improvements and the mechanism may be
associated with targeting this particular network.
4.4. Limitations
Despite these encouraging results, limitations should be acknowl-
edged. First, growing evidence suggest that TRD is a sub-category of
MDD, and there are biological differences between TRD and treatment-
responsive patients (Wu et al., 2011). Because our patient samples were
treatment-resistant, the conclusions from this study cannot be gen-
eralized to MDD patients directly. Future research is needed to compare
TRD with treatment-responsive patients. Second, because our results
were obtained with the assessment of intrinsic connectivity patterns,
task-based imaging studies are needed to demonstrate hippocampal
engagement in the memory deficits of patients. Third, our findings are
strongly in need of replication in independent validation samples,
ideally across multi-site studies.
4.5. Concluding remarks
In summary, based on a recently introduced brain parcellation
method and functional connectivity technique with resting state fMRI,
we demonstrated functional hypoconnectivity in the RIH of MDD pa-
tients in the absence of detectable structural and functional organiza-
tion changes of the hippocampal subregions. Within these disconnec-
tions, the RIH-amygdala connectivity was correlated with the duration
of patients’ current depressive episodes. Neural correlates of declarative
memory deficits in MDD were found within the intermediate and pos-
terior hippocampal FC patterns. Taken together, this evidence suggests
a dual property of the intermediate hippocampus in processing cogni-
tive memory and affective information in MDD, and provides us with
new insights into the network-level neural correlates of memory deficits
in MDD.
4.6. Declarations of interest
The authors declare no financial interests relevant to this work.
D.M.B receives research support from the Canadian Institutes of Health
Research (CIHR), National Institutes of Health – US (NIH), Weston
Brain Institute, Brain Canada and the Temerty Family through the
CAMH Foundation and the Campbell Research Institute. He received
research support and in-kind equipment support for an investigator-
initiated study from Brainsway Ltd. and he is the site principal in-
vestigator for three sponsor-initiated studies for Brainsway Ltd. He
R. Ge, et al.
received in-kind equipment support from Magventure for this in-
vestigator-initiated study. He received medication supplies for an in-
vestigator-initiated trial from Indivior. He participated on advisory
board for Janssen. F.V.R received in-kind equipment support from
Magventure for an investigator-initiated study. He has participated on
an advisory board for Janssen. Other authors report no biomedical fi-
nancial interests or potential conflicts of interest. R.W.L. has received
honoraria for ad hoc speaking or advising/consulting, or received re-
search funds, from: Akili, Allergan, Asia-Pacific Economic Cooperation,
BC Leading Edge Foundation, Brain Canada, Canadian Institutes of
Health Research, Canadian Depression Research and Intervention
Network, Canadian Network for Mood and Anxiety Treatments,
Canadian Psychiatric Association, CME Institute, Hansoh, Janssen,
Lundbeck, Lundbeck Institute, Medscape, Mind Mental Health
Technologies, Otsuka, Pfizer, St. Jude Medical, University Health
Network Foundation, and VGH Foundation.
Contributors
RG and FVR conceived and designed the study. IT, JB, EG, EM, JD,
DMB, ZJD, and RWL provided input on the study design. RG and FVR
developed the plan for data analyses. RG analyzed the data. RG and
FVR drafted the manuscript. All authors made revisions to the manu-
script.
Role of the funding source
None
Acknowledgments
The NINET lab is immensely grateful for the philanthropic support
received for this study. The authors would like to thank all participants
and NINET lab members contributed to this work.
Supplementary materials
Supplementary material associated with this article can be found, in
the online version, at doi:10.1016/j.jad.2019.04.096.
References
Amaral, D., Witter, M., 1989. The three-dimensional organization of the hippocampal
formation: a review of anatomical data. Neuroscience 31, 571–591.
Ashburner, J., Friston, K.J., 2000. Voxel-based morphometry—the methods. Neuroimage
11, 805–821.
Ashburner, J., Friston, K.J., 2005. Unified segmentation. Neuroimage 26, 839–851.
Austin, M.-P., Mitchell, P., Goodwin, G.M., 2001. Cognitive deficits in depression. Br. J.
Psychiatry 178, 200–206.
Ballmaier, M., Toga, A.W., Blanton, R.E., Sowell, E.R., Lavretsky, H., Peterson, J., Pham,
D., Kumar, A., 2004. Anterior cingulate, gyrus rectus, and orbitofrontal abnormalities
in elderly depressed patients: an MRI-based parcellation of the prefrontal cortex. Am.
J. Psychiatry 161, 99–108.
Behzadi, Y., Restom, K., Liau, J., Liu, T.T., 2007. A component based noise correction
method (CompCor) for BOLD and perfusion based fMRI. Neuroimage 37, 90–101.
Blair, J.R., Spreen, O., 1989. Predicting premorbid IQ: a revision of the National Adult
Reading Test. Clin. Neuropsychol. 3, 129–136.
Blumberger, D.M., Vila-Rodriguez, F., Thorpe, K.E., Feffer, K., Noda, Y., Giacobbe, P.,
Knyahnytska, Y., Kennedy, S.H., Lam, R.W., Daskalakis, Z.J., 2018. Effectiveness of
theta burst versus high-frequency repetitive transcranial magnetic stimulation in
patients with depression (THREE-D): a randomised non-inferiority trial. Lancet 391,
1683–1692.
Bora, E., Fornito, A., Pantelis, C., Yücel, M., 2012. Gray matter abnormalities in major
depressive disorder: a meta-analysis of voxel based morphometry studies. J. Affect.
Disord. 138, 9–18.
Bremner, J.D., Narayan, M., Anderson, E.R., Staib, L.H., Miller, H.L., Charney, D.S., 2000.
Hippocampal volume reduction in major depression. Am. J. Psychiatry 157, 115–118.
Bremner, J.D., Vythilingam, M., Vermetten, E., Vaccarino, V., Charney, D.S., 2004.
Deficits in hippocampal and anterior cingulate functioning during verbal declarative
memory encoding in midlife major depression. Am. J. Psychiatry 161, 637–645.
Buckner, R.L., Andrews‐Hanna, J.R., Schacter, D.L., 2008. The brain's default network.
Ann. N. Y. Acad. Sci. 1124, 1–38.
Campbell, S., Marriott, M., Nahmias, C., MacQueen, G.M., 2004. Lower hippocampal
volume in patients suffering from depression: a meta-analysis. Am. J. Psychiatry 161,
255
Journal of Affective Disorders 253 (2019) 248–256
598–607.
Catani, M., Dell'Acqua, F., De Schotten, M.T., 2013. A revised limbic system model for
memory, emotion and behaviour. Neurosci. Biobehav. Rev. 37, 1724–1737.
Caviness Jr, V.S., Meyer, J., Makris, N., Kennedy, D.N., 1996. MRI-based topographic
parcellation of human neocortex: an anatomically specified method with estimate of
reliability. J. Cognit. Neurosci. 8, 566–587.
Chase, H.W., Clos, M., Dibble, S., Fox, P., Grace, A.A., Phillips, M.L., Eickhoff, S.B., 2015.
Evidence for an anterior–posterior differentiation in the human hippocampal for-
mation revealed by meta-analytic parcellation of fMRI coordinate maps: focus on the
subiculum. Neuroimage 113, 44–60.
Chen, A.C., Etkin, A., 2013. Hippocampal network connectivity and activation differ-
entiates post-traumatic stress disorder from generalized anxiety disorder.
Neuropsychopharmacology 38, 1889–1898.
Ciric, R., Wolf, D.H., Power, J.D., Roalf, D.R., Baum, G.L., Ruparel, K., Shinohara, R.T.,
Elliott, M.A., Eickhoff, S.B., Davatzikos, C., 2017. Benchmarking of participant-level
confound regression strategies for the control of motion artifact in studies of func-
tional connectivity. Neuroimage 154, 174–187.
Craddock, R.C., James, G.A., Holtzheimer III, P.E., Hu, X.P., Mayberg, H.S., 2012. A
whole brain fMRI atlas generated via spatially constrained spectral clustering. Hum.
Brain Mapp 33, 1914–1928.
Dai, Z., Yan, C., Li, K., Wang, Z., Wang, J., Cao, M., Lin, Q., Shu, N., Xia, M., Bi, Y., 2014.
Identifying and mapping connectivity patterns of brain network hubs in Alzheimer's
disease. Cereb. Cortex 25, 3723–3742.
Dong, H.-W., Swanson, L.W., Chen, L., Fanselow, M.S., Toga, A.W., 2009.
Genomic–anatomic evidence for distinct functional domains in hippocampal field
CA1. Proc. Natl. Acad. Sci. U.S.A. 106, 11794–11799.
Drysdale, A.T., Grosenick, L., Downar, J., Dunlop, K., Mansouri, F., Meng, Y., Fetcho,
R.N., Zebley, B., Oathes, D.J., Etkin, A., 2017. Resting-state connectivity biomarkers
define neurophysiological subtypes of depression. Nat. Med. 23, 28.
Engvig, A., Fjell, A.M., Westlye, L.T., Skaane, N.V., Sundseth, Ø., Walhovd, K.B., 2012.
Hippocampal subfield volumes correlate with memory training benefit in subjective
memory impairment. Neuroimage 61, 188–194.
Eriksson, P.S., Perfilieva, E., Björk-Eriksson, T., Alborn, A.-M., Nordborg, C., Peterson,
D.A., Gage, F.H., 1998. Neurogenesis in the adult human hippocampus. Nat. Med. 4,
1313–1317.
Fanselow, M.S., Dong, H.-W., 2010. Are the dorsal and ventral hippocampus functionally
distinct structures? Neuron 65, 7–19.
Fornito, A., Harrison, B.J., Zalesky, A., Simons, J.S., 2012. Competitive and cooperative
dynamics of large-scale brain functional networks supporting recollection. Proc. Natl.
Acad. Sci. U.S.A. 109, 12788–12793.
Friston, K.J., Williams, S., Howard, R., Frackowiak, R.S., Turner, R., 1996. Movement-
related effects in fMRI time-series. Magn. Reson. Med. 35, 346–355.
Ge, R., Blumberger, D.M., Downar, J., Daskalakis, Z.J., Tham, J., Lam, R., Vila-Rodriguez,
F., 2017. A sparse representation-based method for parcellation of the resting brain
and its application to treatment-resistant major depressive disorder. J. Neurosci.
Methods 290, 57–68.
Ge, R., Downar, J., Blumberger, D.M., Daskalakis, Z.J., Lam, R.W., Vila-Rodriguez, F.,
2019. Structural network integrity of the central executive network is associated to
the therapeutic effect of rTMS in treatment resistant depression. Prog. Neuro-
Psychopharmacol. Biol. Psychiatry 92, 217–225.
Gennatas, E.D., Avants, B.B., Wolf, D.H., Satterthwaite, T.D., Ruparel, K., Ciric, R.,
Hakonarson, H., Gur, R.E., Gur, R.C., 2017. Age-related effects and sex differences in
gray matter density, volume, mass, and cortical thickness from childhood to young
adulthood. J. Neurosci. 37, 5065–5073.
Goldapple, K., Segal, Z., Garson, C., Lau, M., Bieling, P., Kennedy, S., Mayberg, H., 2004.
Modulation of cortical-limbic pathways in major depression: treatment-specific ef-
fects of cognitive behavior therapy. Arch. Gen. Psychiatry 61, 34–41.
Gong, L., Hou, Z., Wang, Z., He, C., Yin, Y., Yuan, Y., Zhang, H., Lv, L., Zhang, H., Xie, C.,
2018. Disrupted topology of hippocampal connectivity is associated with short-term
antidepressant response in major depressive disorder. J. Affect. Disord. 225,
539–544.
Graham, J., Salimi-Khorshidi, G., Hagan, C., Walsh, N., Goodyer, I., Lennox, B., Suckling,
J., 2013. Meta-analytic evidence for neuroimaging models of depression: state or
trait? J. Affect. Disord. 151, 423–431.
Hahamy, A., Calhoun, V., Pearlson, G., Harel, M., Stern, N., Attar, F., Malach, R.,
Salomon, R., 2014. Save the global: global signal connectivity as a tool for studying
clinical populations with functional magnetic resonance imaging. Brain Connectivity
4, 395–403.
Hamani, C., Mayberg, H., Stone, S., Laxton, A., Haber, S., Lozano, A.M., 2011. The sub-
callosal cingulate gyrus in the context of major depression. Biol. Psychiatry 69,
301–308.
Hamilton, J.P., Chen, M.C., Gotlib, I.H., 2013. Neural systems approaches to under-
standing major depressive disorder: an intrinsic functional organization perspective.
Neurobiol. Dis. 52, 4–11.
Hamilton, M., 1960. A rating scale for depression. J. Neurol. Neurosurg. Psychiatry
23, 56.
Hayasaka, Y., Purgato, M., Magni, L.R., Ogawa, Y., Takeshima, N., Cipriani, A., Barbui, C.,
Leucht, S., Furukawa, T.A., 2015. Dose equivalents of antidepressants: evidence-
based recommendations from randomized controlled trials. J. Affect. Disord. 180,
179–184.
Hickie, I., Naismith, S., Ward, P.B., Turner, K., Scott, E., Mitchell, P., Wilhelm, K., Parker,
G., 2005. Reduced hippocampal volumes and memory loss in patients with early-and
late-onset depression. Br. J. Psychiatry 186, 197–202.
Ho, N., Hooker, J.M., Sahay, A., Holt, D.J., Roffman, J.L., 2013. In vivo imaging of adult
human hippocampal neurogenesis: progress, pitfalls and promise. Mol. Psychiatry 18,
404–416.
Inada, T., Inagaki, A., 2015. Psychotropic dose equivalence in Japan. Psychiatry Clin.
Neurosci. 69, 440–447.
Jayaweera, H., Hickie, I., Duffy, S., Mowszowski, L., Norrie, L., Lagopoulos, J., Naismith,
S., 2016. Episodic memory in depression: the unique contribution of the anterior
R. Ge, et al.
caudate and hippocampus. Psychol. Med. 46, 2189–2199.
Kempton, M.J., Salvador, Z., Munafò, M.R., Geddes, J.R., Simmons, A., Frangou, S.,
Williams, S.C., 2011. Structural neuroimaging studies in major depressive disorder:
meta-analysis and comparison with bipolar disorder. Arch. Gen. Psychiatry 68,
675–690.
Kennedy, D.N., Lange, N., Makris, N., Bates, J., Meyer, J., Caviness Jr, V.S., 1998. Gyri of
the human neocortex: an MRI-based analysis of volume and variance. Cerebral Cortex
(New York, NY: 1991) 8, 372–384.
Kennedy, S.H., Evans, K.R., Krüger, S., Mayberg, H.S., Meyer, J.H., McCann, S.,
Arifuzzman, A.I., Houle, S., Vaccarino, F.J., 2001. Changes in regional brain glucose
metabolism measured with positron emission tomography after paroxetine treatment
of major depression. Am. J. Psychiatry 158, 899–905.
Kurth, F., Gaser, C., Luders, E., 2015. A 12-step user guide for analyzing voxel-wise gray
matter asymmetries in statistical parametric mapping (SPM). Nat. Protoc. 10,
293–304.
Lee, R.S., Hermens, D.F., Porter, M.A., Redoblado-Hodge, M.A., 2012. A meta-analysis of
cognitive deficits in first-episode major depressive disorder. J. Affect. Disord. 140,
113–124.
Lewis, P.R., Shute, C., 1967. The cholinergic limbic system: projections to hippocampal
formation, medial cortex, nuclei of the ascending cholinergic reticular system, and
the subfornical organ and supra-optic crest. Brain 90, 521–540.
Lui, S., Wu, Q., Qiu, L., Yang, X., Kuang, W., Chan, R.C., Huang, X., Kemp, G.J., Mechelli,
A., Gong, Q., 2011. Resting-state functional connectivity in treatment-resistant de-
pression. Am. J. Psychiatry 168, 642–648.
Maalouf, F.T., Brent, D., Clark, L., Tavitian, L., McHugh, R.M., Sahakian, B.J., Phillips,
M.L., 2011. Neurocognitive impairment in adolescent major depressive disorder:
state vs. trait illness markers. J. Affect. Disord. 133, 625–632.
MacQueen, G., Frodl, T., 2011. The hippocampus in major depression: evidence for the
convergence of the bench and bedside in psychiatric research? Mol. Psychiatry 16,
252.
Maeshima, H., Baba, H., Satomura, E., Shimano, T., Inoue, M., Ishijima, S., Suzuki, T.,
Arai, H., 2016. Residual memory impairment in remitted depression may be a pre-
dictive factor for recurrence. J. Clin. Psychiatry 77, 247–251.
Mannie, Z., Barnes, J., Bristow, G., Harmer, C., Cowen, P., 2009. Memory impairment in
young women at increased risk of depression: influence of cortisol and 5-HTT gen-
otype. Psychol. Med. 39, 757–762.
Manoliu, A., Meng, C., Brandl, F., Doll, A., Tahmasian, M., Scherr, M., Schwerthöffer, D.,
Zimmer, C., Förstl, H., Bäuml, J., 2014. Insular dysfunction within the salience net-
work is associated with severity of symptoms and aberrant inter-network con-
nectivity in major depressive disorder. Front. Hum. Neurosci. 7, 930.
Matthews, S.C., Strigo, I.A., Simmons, A.N., Yang, T.T., Paulus, M.P., 2008. Decreased
functional coupling of the amygdala and supragenual cingulate is related to increased
depression in unmedicated individuals with current major depressive disorder. J.
Affect. Disord. 111, 13–20.
Mayberg, H.S., 2003. Modulating dysfunctional limbic-cortical circuits in depression:
towards development of brain-based algorithms for diagnosis and optimised treat-
ment. Br. Med. Bull. 65, 193–207.
McKinnon, M.C., Yucel, K., Nazarov, A., MacQueen, G.M., 2009. A meta-analysis ex-
amining clinical predictors of hippocampal volume in patients with major depressive
disorder. J. Psychiatry Neurosci. 34, 41.
Menon, V., 2011. Large-scale brain networks and psychopathology: a unifying triple
network model. Trends Cognit. Sci. 15, 483–506.
Milne, A.M., MacQueen, G.M., Hall, G.B., 2012. Abnormal hippocampal activation in
patients with extensive history of major depression: an fMRI study. J. Psychiatry
Neurosci. 37, 28.
Morgane, P.J., Galler, J.R., Mokler, D.J., 2005. A review of systems and networks of the
limbic forebrain/limbic midbrain. Prog. Neurobiol. 75, 143–160.
Mulders, P.C., van Eijndhoven, P.F., Schene, A.H., Beckmann, C.F., Tendolkar, I., 2015.
Resting-state functional connectivity in major depressive disorder: a review.
Neurosci. Biobehav. Rev. 56, 330–344.
Murphy, K., Birn, R.M., Handwerker, D.A., Jones, T.B., Bandettini, P.A., 2009. The impact
of global signal regression on resting state correlations: are anti-correlated networks
introduced? Neuroimage 44, 893–905.
Murphy, K., Fox, M.D., 2017. Towards a consensus regarding global signal regression for
resting state functional connectivity MRI. Neuroimage 154, 169–173.
Muschelli, J., Nebel, M.B., Caffo, B.S., Barber, A.D., Pekar, J.J., Mostofsky, S.H., 2014.
Reduction of motion-related artifacts in resting state fMRI using aCompCor.
Neuroimage 96, 22–35.
Nebel, M.B., Joel, S.E., Muschelli, J., Barber, A.D., Caffo, B.S., Pekar, J.J., Mostofsky, S.H.,
2014. Disruption of functional organization within the primary motor cortex in
children with autism. Hum. Brain Mapp. 35, 567–580.
Neu, P., Bajbouj, M., Schilling, A., Godemann, F., Berman, R.M., Schlattmann, P., 2005.
Cognitive function over the treatment course of depression in middle-aged patients:
correlation with brain MRI signal hyperintensities. J. Psychiatr. Res. 39, 129–135.
Phan, K.L., Wager, T., Taylor, S.F., Liberzon, I., 2002. Functional neuroanatomy of
emotion: a meta-analysis of emotion activation studies in PET and fMRI. Neuroimage
16, 331–348.
Poppenk, J., Evensmoen, H.R., Moscovitch, M., Nadel, L., 2013. Long-axis specialization
of the human hippocampus. Trends Cognit. Sci. 17, 230–240.
Porter, R.J., Gallagher, P., Thompson, J.M., Young, A.H., 2003. Neurocognitive impair-
ment in drug-free patients with major depressive disorder. Br. J. Psychiatry 182,
214–220.
Potter, G.G., Wagner, H.R., Burke, J.R., Plassman, B.L., Welsh-Bohmer, K.A., Steffens,
D.C., 2013. Neuropsychological predictors of dementia in late-life major depressive
256
Journal of Affective Disorders 253 (2019) 248–256
disorder. Am. J. Geriatr. Psychiatry 21, 297–306.
Power, J.D., Barnes, K.A., Snyder, A.Z., Schlaggar, B.L., Petersen, S.E., 2012. Spurious but
systematic correlations in functional connectivity MRI networks arise from subject
motion. Neuroimage 59, 2142–2154.
Reppermund, S., Ising, M., Lucae, S., Zihl, J., 2009. Cognitive impairment in unipolar
depression is persistent and non-specific: further evidence for the final common
pathway disorder hypothesis. Psychol. Med. 39, 603–614.
Robinson, J.L., Salibi, N., Deshpande, G., 2016. Functional connectivity of the left and
right hippocampi: evidence for functional lateralization along the long-axis using
meta-analytic approaches and ultra-high field functional neuroimaging. Neuroimage
135, 64–78.
Rock, P., Roiser, J., Riedel, W., Blackwell, A., 2014. Cognitive impairment in depression:
a systematic review and meta-analysis. Psychol. Med. 44, 2029–2040.
Sackeim, H.A., 2001. The definition and meaning of treatment-resistant depression. J.
Clin. Psychiatry 62, 10–17.
Salomons, T.V., Iannetti, G.D., Liang, M., Wood, J.N., 2016. The “pain matrix” in pain-
free individuals. JAMA Neurol. 73, 755–756.
Schiller, C.E., Minkel, J., Smoski, M.J., Dichter, G.S., 2013. Remitted major depression is
characterized by reduced prefrontal cortex reactivity to reward loss. J. Affect. Disord.
151, 756–762.
Schmaal, L., Veltman, D.J., van Erp, T.G., Sämann, P., Frodl, T., Jahanshad, N., Loehrer,
E., Tiemeier, H., Hofman, A., Niessen, W., 2016. Subcortical brain alterations in
major depressive disorder: findings from the ENIGMA Major Depressive Disorder
working group. Mol. Psychiatry 21, 806.
Seeley, W.W., Menon, V., Schatzberg, A.F., Keller, J., Glover, G.H., Kenna, H., Reiss, A.L.,
Greicius, M.D., 2007. Dissociable intrinsic connectivity networks for salience pro-
cessing and executive control. J. Neurosci. 27, 2349–2356.
Smith, S.M., Jenkinson, M., Woolrich, M.W., Beckmann, C.F., Behrens, T.E., Johansen-
Berg, H., Bannister, P.R., De Luca, M., Drobnjak, I., Flitney, D.E., 2004. Advances in
functional and structural MR image analysis and implementation as FSL. Neuroimage
23, S208–S219.
Smoski, M.J., Keng, S.-L., Schiller, C.E., Minkel, J., Dichter, G.S., 2013. Neural mechan-
isms of cognitive reappraisal in remitted major depressive disorder. J. Affect. Disord.
151, 171–177.
Strange, B.A., Witter, M.P., Lein, E.S., Moser, E.I., 2014. Functional organization of the
hippocampal longitudinal axis. Nat. Rev. Neurosci. 15, 655–669.
Strauss, E., Sherman, E.M., Spreen, O., 2006. A Compendium of Neuropsychological
Tests: Administration, Norms, and Commentary. American Chemical Society.
Sweeney, J.A., Kmiec, J.A., Kupfer, D.J., 2000. Neuropsychologic impairments in bipolar
and unipolar mood disorders on the CANTAB neurocognitive battery. Biol. Psychiatry
48, 674–684.
Tae, W.S., Kim, S.S., Lee, K.U., Nam, E.-C., Kim, K.W., 2008. Validation of hippocampal
volumes measured using a manual method and two automated methods (FreeSurfer
and IBASPM) in chronic major depressive disorder. Neuroradiology 50, 569.
Tamm, L., Menon, V., Reiss, A.L., 2006. Parietal attentional system aberrations during
target detection in adolescents with attention deficit hyperactivity disorder: event-
related fMRI evidence. Am. J. Psychiatry 163, 1033–1043.
Videbech, P., Ravnkilde, B., 2004. Hippocampal volume and depression: a meta-analysis
of MRI studies. Am. J. Psychiatry 161, 1957–1966.
Vythilingam, M., Vermetten, E., Anderson, G.M., Luckenbaugh, D., Anderson, E.R., Snow,
J., Staib, L.H., Charney, D.S., Bremner, J.D., 2004. Hippocampal volume, memory,
and cortisol status in major depressive disorder: effects of treatment. Biol. Psychiatry
56, 101–112.
Wang, J., Vila-Rodriguez, F., Jiang, W., Ren, Y.-P., Wang, C.-M., Ma, X., 2018.
Accelerated magnetic seizure therapy for treatment of major depressive disorder: a
report of 3 cases. J. ECT 00, 1.
Wang, J.X., Rogers, L.M., Gross, E.Z., Ryals, A.J., Dokucu, M.E., Brandstatt, K.L.,
Hermiller, M.S., Voss, J.L., 2014. Targeted enhancement of cortical-hippocampal
brain networks and associative memory. Science 345, 1054–1057.
Whitfield-Gabrieli, S., Ford, J.M., 2012. Default mode network activity and connectivity
in psychopathology. Annu. Rev. Clin. Psychol. 8, 49–76.
Williams, J.B., 1988. A structured interview guide for the Hamilton Depression Rating
Scale. Arch. Gen. Psychiatry 45, 742–747.
World Health Organization, 2017. Depression and other common mental disorders: global
health estimates. World Heal Organ., pp. 1–24.
Wu, Q.Z., Li, D.M., Kuang, W.H., Zhang, T.J., Lui, S., Huang, X.Q., Chan, R.C., Kemp, G.J.,
Gong, Q.Y., 2011. Abnormal regional spontaneous neural activity in treatment‐re-
fractory depression revealed by resting‐state fMRI. Hum. Brain Mapp. 32,
1290–1299.
Yamada, T., Itahashi, T., Nakamura, M., Watanabe, H., Kuroda, M., Ohta, H., Kanai, C.,
Kato, N., Hashimoto, R.-i., 2016. Altered functional organization within the insular
cortex in adult males with high-functioning autism spectrum disorder: evidence from
connectivity-based parcellation. Mol. Autism 7, 41.
Yan, C.-G., Wang, X.-D., Zuo, X.-N., Zang, Y.-F., 2016. DPABI: data processing & analysis
for (resting-state) brain imaging. Neuroinformatics 14, 339–351.
Yang, G.J., Murray, J.D., Repovs, G., Cole, M.W., Savic, A., Glasser, M.F., Pittenger, C.,
Krystal, J.H., Wang, X.-J., Pearlson, G.D., 2014. Altered global brain signal in schi-
zophrenia. Proc. Natl. Acad. Sci. U.S.A. 111, 7438–7443.
Yao, H., Liu, Y., Zhou, B., Zhang, Z., An, N., Wang, P., Wang, L., Zhang, X., Jiang, T.,
2013. Decreased functional connectivity of the amygdala in Alzheimer's disease re-
vealed by resting-state fMRI. Eur. J. Radiol. 82, 1531–1538.
Zhang, J., Wang, J., Wu, Q., Kuang, W., Huang, X., He, Y., Gong, Q., 2011. Disrupted
brain connectivity networks in drug-naive, first-episode major depressive disorder.
Biol. Psychiatry 70, 334–342.