Eur Arch Psychiatry Clin Neurosci (2015) 265:57–66 DOI 10.1007/s00406-014-0552-2
ORIGINAL PAPER
Dissociating pathomechanisms of depression with fMRI: bottomup or top-down dysfunctions of the reward system Roberto Goya-Maldonado • Kristina Weber Sarah Trost • Esther Diekhof • Maria Keil • Peter Dechent • Oliver Gruber
•
Received: 23 April 2014 / Accepted: 6 October 2014 / Published online: 21 October 2014 Ó Springer-Verlag Berlin Heidelberg 2014
Abstract Depression is a debilitating psychiatric disorder characterized among other aspects by the inability to properly experience or respond to reward. However, it remains unclear whether patients with depression present impaired reward system due to abnormal modulatory mechanisms. We investigated the activation of the nucleus accumbens (NAcc), a crucial region involved in reward processing, with functional magnetic resonance imaging using the desire-reason-dilemma paradigm. This task allows tracking the activity of the NAcc during the acceptance or the rejection of previously conditioned reward stimuli. Patients were assigned into subgroups of lower (LA) or higher (HA) NAcc activation according to beta weights. LA patients presented significant hypoactivation in the ventral tegmental area in addition to bilateral
Electronic supplementary material The online version of this article (doi:10.1007/s00406-014-0552-2) contains supplementary material, which is available to authorized users. R. Goya-Maldonado (&) K. Weber S. Trost E. Diekhof M. Keil O. Gruber Center for Translational Research in Systems Neuroscience and Psychiatry, Department of Psychiatry and Psychotherapy, University Medical Center, Georg August University, Go¨ttingen, Germany e-mail: [email protected] Present Address: E. Diekhof Neuroendocrinology Unit, Department of Human Biology, Biocenter Grindel and Zoological Museum, University of Hamburg, Hamburg, Germany P. Dechent Research Group ‘MR-Research in Neurology and Psychiatry’, Department of Cognitive Neurology, University Medical Center, Georg August University, Go¨ttingen, Germany
ventral striatum, confirming impairments in the bottom-up input to the NAcc. Conversely, HA patients presented significant hyperactivation in prefrontal areas such as the rostral anterior cingulate cortex and the anterior ventral prefrontal cortex in addition to bilateral ventral striatum, suggesting disturbances in the top-down regulation of the NAcc. Demographic and clinical differences explaining the abnormal co-activations of midbrain and prefrontal regions were not identified. Therefore, we provide evidence for dysfunctional bottom-up processing in one potential neurobiological subtype of depression (LA) and dysfunctional top-down modulation in another subtype (HA). We suggest that the midbrain and prefrontal regions are more specific pathophysiological substrates for each depression subtype. Above all, our results encourage the segregation of patients by similar dysfunctional mechanisms of the dopaminergic system, which would finally contribute to disentangle more specific pathogeneses and guide the development of more personalized targets for future therapies. Keywords Depression Nucleus accumbens Reward Pathophysiological subtypes Neuroimaging biomarkers Stratified therapy Abbreviations AD Antidepressant DC Desire Context HA High NAcc Activity LA Low NAcc Activity OFC Orbitofrontal Cortex NAcc Nucleus Accumbens rACC Rostral Anterior Cingulate Cortex RC Reason Context VTA Ventral Tegmental Area
123
58
Introduction Patients with depression demonstrate a genuine inability to experience reward, which strongly impairs motivation- and incentive-based learning, both necessary for daily adaptive behaviors [1]. At the center of the reward system is the nucleus accumbens (NAcc) [1–3] receiving dopaminergic bottom-up input from the ventral tegmental area (VTA) and substantia nigra (SN) and further extending it to the ventral striatum, to the ventral pallidum and back to the initial midbrain areas. The VTA sends bundles through thalamic nuclei [1, 2] to dorsal, medial, and orbital prefrontal regions so they can top-down modulate the activity of subcortical structures, closing a very refined loop. Imaging studies extended the knowledge about the dynamic involvement of the NAcc in different contexts of reward [4–9]. Although identifying the foundations for reward impairment in depression should be fundamental to comprehend pathogeneses and/or refine treatment options, this field has so far remained mostly unexplored. Some studies investigating reward processing in depression reported hypoactivity of the NAcc and ventral striatum [10–12], which was not always consistently replicated [13, 14]. Such heterogeneity in the NAcc could result from disturbed dopaminergic bottom-up input or impaired top-down regulation of the NAcc, as the model suggested by Savitz and Drevets [3]. Our research group has previously investigated the NAcc activity of healthy volunteers with functional magnetic resonance imaging (fMRI) using the desire-reason-dilemma (DRD) paradigm [15–17]. This task comprises two different contexts, one in which participants were allowed to freely accept previously conditioned reward stimuli and another, when these conditioned reward stimuli had to be rejected. In the first context, the NAcc is activated by dopaminergic bottom-up mechanisms, whereas in the second context, this activation is suppressed by top-down regulatory mechanisms. Here, we used this paradigm to investigate the levels NAcc functional activation in patients diagnosed with major depressive disorder. We expected patients displaying lower (LA) NAcc activation to exhibit dysfunctional coactivations in regions associated with bottom-up mechanisms, whereas those showing higher (HA) NAcc activation were anticipated to present concomitant dysfunctional regions related to top-down mechanisms.
Methods Participants Twenty-five inward and outward symptomatic patients diagnosed with major depressive disorder according to
123
Eur Arch Psychiatry Clin Neurosci (2015) 265:57–66
DSM-IV criteria and coming for treatment at the University Medical Center Goettingen were recruited. Inclusion criteria were 18- to 60-year-old right-handed patients. The exclusion criteria comprised past or actual presence of other axis I diagnoses, neurological, or other medical conditions that could be related to affective symptoms. As our study was majorly designed to evaluate the heterogeneity of the NAcc activation during reward, the ability of performing the reward desire-reason paradigm in patients had to be reassured. Thus, patients were expected to accept a minimum of a third (10/30) of the conditioned reward stimuli in the context named desire (DC) and reject more than a half (20/40) of the reward stimuli in the context named reason (RC). Otherwise, they would be considered not engaging in the task demands and therefore would be excluded from the final analysis. Also, in favor of not having the NAcc reactivity overly driven by demographic factors such as age and gender, each healthy control was selected to match age and gender characteristics of the patients enrolled, following identical study protocols. Controls reported no medical history, including neurological and psychiatric history, as well as no previous or actual use of psychotropic medication. Each participant received up to 30 euros according to task performance in our reward paradigm, which were added to the general reimbursement for participation. The project was approved by the local ethics committee. Participants provided written informed consent before investigation and after the study had been fully explained. Experimental procedure Outside the scanner, the experiment started with an operant conditioning task, in which participants associated specific stimuli colors to a monetary reward or to a neutral outcome. Squares of eight different colors were presented 20 times each in a randomized sequence. Two of them led to immediate reward, when collected with a button press, by a positive feedback. In the MRI environment, the task involved event-related blocks of two contexts: DC, in which participants were free to accept conditioned reward stimuli, and RC, in which conditioned reward stimuli had to be declined. Each trial had a duration of 1,900 ms, starting with 200 ms of blank screen, 900 ms of stimulus presentation, 700 ms of immediate feedback, and 100 ms of blank screen (for comprehensive task description, see [17]). In DC, NAcc activation would be elicited by the acceptance of previously conditioned stimuli. In RC, the activation of this region would be suppressed by prefrontal top-down mechanisms, as conditioned stimuli had to be declined. When a reward stimulus was collected in RC, the actual block terminated with the feedback ‘‘goal failure’’ and a new one started. In this way, the desire-reason-
Eur Arch Psychiatry Clin Neurosci (2015) 265:57–66
dilemma paradigm forces the participants to overcome the behavioral tendency to respond to conditioned reward stimuli in order to achieve a superordinate goal [15–17]. Data acquisition Clinical parameters Patients were systematically assessed with standardized psychometric scales administered by a trained investigator blinded to the research question. In addition, the Wechsler Adult Intelligence Scale (WAIS) vocabulary score, the Montgomery–Asberg Depression Rating Scale (MADRS), and the Beck Depression Inventory (BDI), clinical information such as age at illness onset, duration of illness, number of depressive episodes, and actual antidepressant (AD) treatment plan was collected for post hoc comparisons between subgroups. Functional neuroimaging Whole-brain fMRI was performed on 3 T system (Magnetom TRIO, Siemens Healthcare, Erlangen, Germany) using a 8-channel head coil, covering the whole brain with 31 slices, AC–PC aligned, ascending orientation, 64 9 64 matrix, 3-mm thickness, 0.6-mm spacing with a gradient-echo EPI sequence (TR 1.9 s, flip angle 70°, FOV 192 mm, TE 30 ms). Structural T1-weighted at 1-mm isotropic resolution was also acquired. The total time of acquisition for structural and functional protocols with the desire-reason-dilemma paradigm was approximately 12 min. Data analysis Behavioral and demographic data Behavioral data were acquired using Presentation software (Version 14.9, www.neurobs.com). Log files, accuracy (as number of completed blocks), and reaction times for correct responses were entered into a group-by-block analysis of variance (ANOVA two-tailed p \ 0.05) in SPSS (version 20, SPSS Inc., Chicago, Illinois, USA). Demographic and clinical data were compared between subgroups in SPSS in the same fashion. Imaging data Functional images were slice time corrected and unwarped after four initial volumes were discarded, realigned to the fifth volume using rigid body transformation, normalized to the EPI template in Montreal Neurological Institute (MNI) stereotactic space with 3 9 3 9 3 mm3 voxel resolution,
59
and spatially smoothed (FWHM 9 mm) using SPM8 (www.fil.ion.ucl.ac.uk/spm8) and MATLAB (The MathWorks, Inc., Natick, MA, USA). We carefully checked for motion artifacts and set a cutoff (over the entire experiment) of\2 mm for the three planes of translation and\3° for the three planes of rotation as a control of data quality. First-level contrasts for correct response trials were defined for DC and RC blocks. Marsbar (http://marsbar.source forge.net) was used to extract the beta weights from DC correct events with a 5-mm-radius sphere placed in the NAcc region centered at [x = 12, y = 12, z = -4] MNI coordinates. The levels of activation in the acceptance of conditioned reward stimuli allocated patients in either lower (LA) NAcc activity or higher (HA) NAcc activity. Second-level random effect analyses were performed for statistical inference on the group level for LA, HA, and respective matched controls, C\LA[ and C\HA[. To test for differences as a proof of concept, one-way ANOVA between groups was implemented in spm8 as suggested by Henson and Penny (http://www.fil.ion.ucl.ac.uk/*wpenny/ publications/rik_anova.pdf). A significant group effect at the right NAcc was centered at [x = 15, y = 6, z = -3] MNI coordinates (Fig. 1a, open arrow). The F contrast given was ½1 1=3 1=3 1=3; 1=3 1 1=3 1=3; 1=3 1=3 1=3 1; 1=3 1=3 1=3 1 . Notably, the main effect was driven by the groups LA and HA (Fig. 1b). However, considering that some variation in the NAcc activity was seen between in C\LA[ and C\HA[, we confirmed the coactivations with independent-sample t tests. A more conservative contrast regarding demographic factors between patient subgroups and paired controls for the results of each patient subgroup was taken. The t contrasts entered were ½1 1=2 1=2 0 (Fig. 2) and ½1=2 0 1 1=2 (Fig. 3). This allowed us to additionally disregard age and gender variations accounting for potential differences seen in the NAcc. Small volumes with 10 mm radius (Supplement 1) were a priori defined for multiple testing corrections (p \ 0.05) based on our hypotheses and on previous studies [15–17] using the DRD paradigm.
Results Behavioral and demographic data We recruited 25 patients and from those 20 (80 %) patients were able to perform the minimum requirements of our DRD paradigm as described in the exclusion criteria. Twenty healthy controls were demographically selected to match age and gender characteristics of the patients enrolled. Final analyses included 20 pairs of patients and healthy controls well matched for age and gender. Considering that dysfunctional bottom-up mechanisms should
123
60
Eur Arch Psychiatry Clin Neurosci (2015) 265:57–66
Fig. 1 Heterogeneous hemodynamic responses in the nucleus accumbens (NAcc). a Coronal sections showing small volumes covering the NAcc region bilaterally (in brown) and significant between-group differences in functional activation (in yellow) using the DRD paradigm (open arrow signalizes the right NAcc); b contrast estimates (mean, 90 % CI) at the right NAcc; c design matrix and F contrast
given as described in the ‘‘Methods’’ section; LA: patients with low hemodynamic responses in the NAcc (n = 10), C\LA[: age-/gendermatched controls for LA, HA: patients with high hemodynamic responses in the NAcc (n = 10), C\HA[: age-/gender-matched controls for HA
lead to low activity (LA) at the NAcc region and dysfunctional top-down mechanisms should lead to high activity (HA) at this region, patients were allocated into subgroups (LA or HA) according to their NAcc levels of activity as described in methods. In Table 1, demographic, behavioral, and clinical data are compared between patient subgroups and their controls.
Top-down dysfunction
Bottom-up dysfunction Bottom-up processes were evaluated when a reward stimulus was collected in DC. A robust hypoactivation for LA patients in comparison with C\LA[ and HA was seen bilaterally at ventral striatal region (Fig. 2a), indicating deficient bottom-up input in this patient group. Other significant hypoactive regions were seen at this contrast, among them the VTA as one of our regions of interest (Fig. 2b; Table 2). These findings were specific for LA, considering that neither C\LA[ nor HA presented differences in the VTA or other midbrain regions (Fig. 2b; Table 3).
123
Top-down processes were evaluated in RC, during which NAcc activation elicited by the presentation of reward stimuli should be physiologically suppressed (signalized as open arrows in controls, Fig. 2B ? C). In comparison with LA group and C\HA[, HA patients showed significantly increased bilateral ventral striatal activation in RC, revealing a lack of modulatory top-down regulation of the reward system (Fig. 2B ? C). When evaluated for potential differences in RC, LA patients did not present significant differences (see RC in Table 2), indicating that areas known to take part in top-down mechanisms were not deviant from controls in this group. Whole-brain analyses Whole-brain analyses were performed in order to identify further functional abnormalities in the two subgroups of depressive patients with bottom-up or top-down deficiencies of the reward system (Tables 2, 3). In general,
Eur Arch Psychiatry Clin Neurosci (2015) 265:57–66
61
Fig. 2 Dysfunctional bottom-up mechanism. a Left coronal slice (y = 6) displaying reduced hemodynamic responses in the nucleus accumbens (NAcc) in patients with ventral striatal LA (n = 10) in comparison with both age-/gender-matched controls (C\LA[, n = 10) and HA (n = 10) during desire context (DC) of the DRD paradigm; right contrast estimates (mean, 90 % CI) at the NAcc; b left axial
slice (z = -12) presenting reduced hemodynamic response in the ventral tegmental area (VTA) in patients with ventral striatal LA (n = 10) in comparison with age-/gender-matched controls (C\LA[, n = 10) and HA (n = 10) during desire context (DC) of the DRD paradigm; right contrast estimates (mean, 90 % CI) at the VTA; regions listed in Table 2
differential brain activations were seen in subcortical and cortical areas known to be involved in reward processing. The right orbitofrontal cortex (OFC) and the dorsal ACC showed reduced activity in LA patients in comparison with the other subgroups. Furthermore, the frontal eye fields were bilaterally hypoactive in LA patients in response to reward stimuli. In contrast, HA patients revealed hyperresponsivity to reward stimuli in the left OFC and other prefrontal regulatory regions such as the inferior and middle frontal gyrus as well as the avPFC. Taken together, LA patients showed downregulated regions involved in reward processing in contrast to HA, who presented upregulated regions in DC. No significant differences were
seen in RC for LA, whereas HA presented hyperactive posterior cingulate cortex and precuneus.
Discussion A hallmark of depression is the inability to properly experience reward. In the present study, we investigated reward processing in a group of 20 acutely depressed patients, focusing on the functional response of the NAcc to a reward task. Consistent with our hypothesis, NAcc activity and modulation in patients diverged abnormally from controls, when challenged by two reward-associated
123
62
Eur Arch Psychiatry Clin Neurosci (2015) 265:57–66
Fig. 3 Dysfunctional top-down mechanism. a Sagittal slice (x = 9) presenting increased hemodynamic responses in the nucleus accumbens (NAcc) in patients with ventral striatal HA (n = 10) in comparison with both age-/gender-matched controls (C\HA[, n = 10) and LA (n = 10) during desire context (DC) of the DRD paradigm; b contrast estimates (mean, 90 % CI) in the NAcc during
desire context (DC) of the DRD paradigm; c contrast estimates (mean, 90 % CI) at the NAcc during reason context (RC) of the DRD paradigm, open arrows signalize the top-down suppression from DC to RC (B ? C); d coronal slices (y = 6, y = 42) from a with crosshairs at the right NAcc and rostral ACC (rACC), respectively; regions listed in Table 3
contexts, DC and RC, of the DRD paradigm. More specifically, one group of patients displayed significantly lower NAcc activation (named here as LA patients) in comparison with matched controls, named as C\LA[. Other brain regions that were hypoactive in LA patients included the VTA, a region involved in bottom-up mechanisms of the dopaminergic system, confirming our a priori hypothesis of dysfunctional dopaminergic input. Conversely, the other group of patients showed significant hyperactivation and a lack of top-down regulation of the NAcc in comparison with matched controls, named here as HA and C\HA[, respectively. In these HA patients, hyperactivation was also found in prefrontal regions such as the avPFC, providing further support for our initial hypothesis of dysfunctional top-down modulation of the NAcc. Matched healthy controls showed NAcc activation during DC that was reduced during RC, replicating our previous findings
[15–17]. In contrast, patients did not present this adaptability in the NAcc activity across DC and RC, indicating dysfunctional modulatory mechanisms. Overall, our results provide specific neural substrates underlying dysfunctional bottom-up or top-down mechanisms as the model suggested for pathophysiological changes in depressive disorders [3]. This is also in agreement with the observation of heterogeneous NAcc hemodynamic responses during reward processing in patients with depression [10–14], which to the best of our knowledge is first explored here. The need to adjust to a frequently changing environment has possibly favored the development of complex neural regulatory substrates involved in reward processing. FMRI studies have extended the knowledge about the neural implementation [4–6] and adaptability [7–9] of the reward system in the healthy brain. In this study, we addressed the clinical inability to react to reward stimuli during the
123
Eur Arch Psychiatry Clin Neurosci (2015) 265:57–66 Table 1 Comparison of demographic, behavioral, psychometric, and general clinical parameters between subgroups of patients with depression exhibiting lower nucleus accumbens (NAcc) activity (LA), or higher NAcc activity (HA), and their age- and gender-matched controls in desire (DC) and reason contexts (RC) of the DRD paradigm
63
LA
HA
C\LA[
C\HA[
ANOVA
LA versus HA
N = 10 M (SD)
N = 10 M (SD)
N = 10 M (SD)
N = 10 M (SD)
F
t/v2
Age (year)
34.1 (9.6)
37.1 (11.3)
34.2 (9.5)
32.7 (10.3)
.32
.808
.64
.532
Education (year)
14.2 (2.5)
15.5 (4.2)
15.6 (3.5)
13.9 (3.1)
.61
.614
.83
.412
p, 2-tail
p, 2-tail
Gender (female/total)
7/10
7/10
6/10
8/10
.95
.813
.00
Truncated sequences (n)
12.5 (5.4)
17.4 (7.8)
10.6 (6.8)
13.8 (8.7)
1.54
.220
1.63
1 .120
Response time (ms) in DC
519.0 (28.8)
554.2 (69.4)
525.2 (56.9)
548.66 (51.6)
1.03
.390
1.48
.156
Response time (ms) in RC
522.3 (51.9)
570.4 (78.1)
521.91 (54.6)
559.92 (66.1)
1.57
.212
1.62
.122
WAIS vocabulary score
31.3 (3.5)
30.7 (3.7)
N/A
N/A
–
–
.38
.708
MADRS total score (10 items)
12.4 (6.3)
15.2 (5.5)
N/A
N/A
–
–
1.06
.302
BDI total score (21 items)
14.7 (10.8)
22.3 (8.9)
N/A
N/A
–
–
1.71
.103
Age first episode
23.4 (6.7)
27.4 (12.2)
N/A
N/A
–
–
.90
.378
Duration of disorder (year)
10.7 (9.5)
9.8 (10.5)
N/A
N/A
–
–
.20
.843
Depressive episodes (n)
6.5 (4.1)
5.1 (4.3)
N/A
N/A
–
–
1.12
.279
Actual AD plan
Antipressant (AD) monotherapy versus polytherapy between LA and HA v2 9.60 p = .008 Italic values: p \ 0.05
Escitalopram 10–30 mg/day
7/10
2/10
N/A
N/A
–
–
5.05
Sertraline 125 mg/day
1/10
1/10
N/A
N/A
–
–
.00
Mirtazapine 15–45 mg/day
1/10
4/10
N/A
N/A
–
–
2.40
.025 1 .121
Reboxetine 12 mg/day
–
1/10
N/A
N/A
–
–
1.05
.305
Venlafaxine 225–300 mg/ day
–
4/10
N/A
N/A
–
–
1.25
.264
Duloxetine 90 mg/day Pipamperon 40 mg/day
– –
1/10 1/10
N/A N/A
N/A N/A
– –
– –
1.05 1.05
.305 .305
Quetiapine 30 mg/day
–
1/10
N/A
N/A
–
–
1.05
.305
Olanzapine 5 mg/day
–
1/10
N/A
N/A
–
–
1.05
.305
Agomelatine 25 mg/day
–
1/10
N/A
N/A
–
–
1.05
.305
manifestation of depression by using an established paradigm to dissect mechanisms related to bottom-up and topdown influences on the NAcc [14–16]. In our sample of patients, heterogeneous NAcc hemodynamic reactivity to reward stimuli was observed, corroborating dysfunctional input or dysfunctional regulatory mechanisms. Dysfunctional bottom-up mechanisms in LA were evidenced by abnormal co-activation at the VTA, whereas in HA, dysfunctional top-down mechanisms were seen by abnormal avPFC and rACC co-activations. The innovation of this approach relies on the use of functional disturbances to define subgroups that may constitute different subtypes. This is exactly the reverse of defining patient subgroups by clinical parameters and evaluating their differential
correlates, which may comprise important limitations (for review, see [18]). Although the differential correlates between our patient subgroups in regard to NAcc heterogeneity could be consistently identified, they were not explained by demographic or clinical parameters in a post hoc evaluation. Clinical heterogeneity in depression is a theme of discussion per se in the literature [19, 20]. Previous studies [21, 22] have shown heterogeneous findings as a reflection of clinical subtypes of depression. In a SPECT study [23], patients with atypical depression presented increased right frontal perfusion in contrast to melancholic and undifferentiated depression. In addition, the atypical subtype has also been reported to perceive stress more intensively than other subtypes [24]. In our
123
64
Eur Arch Psychiatry Clin Neurosci (2015) 265:57–66
Table 2 Patients presenting low activity (LA, n = 10) in the ventral striatal region in comparison with both their age-/gender-matched controls (C\LA[, n = 10) and high activity (HA, n = 10) patients during desire (DC) and reason (RC) contexts Reward context
DC
Hemodynamic activity (contrast -2 1 1 0)
Decreased
Region
Peak voxel
k
T
p
MNI coordinates x
VST L
60
5.29
0.0002
VST R
84
5.73
0.00007
533 1,691
5.52 5.29
MTG L
37
dACC R LOFC R
TP L PCN L/R
RC
Cluster size
y
z
-15
3
-6
15
9
-3
0.000001 0.000003
-57 -9
-54 -48
18 45
4.39
0.00005
-45
-3
-18
22
4.16
0.00009
3
15
24
22
3.93
0.0002
30
39
-9 -12
VTA L/SN
6
3.68
0.014
-15
-18
MOG R
8
3.67
0.0004
33
-75
24
FEF R
10
3.65
0.0004
33
-6
57
FEF L
7
3.28
0.001
-27
12
51
No differences
VST ventral striatum, TP temporoparietal, PCN precuneus, MTG middle temporal gyrus, VTA ventral tegmental area, SN substantia nigra, dACC dorsal anterior cingulate cortex, LOFC lateral orbitofrontal cortex, MOG middle occipital gyrus, FEF frontal eye field Italic values: p \ 0.05 corrected, normal values: p \ 0.001 uncorrected
Table 3 Patients presenting high activity (HA, n = 10) in the ventral striatal region in comparison with both their age-/gender-matched controls (C\HA[, n = 10) and low activity (LA, n = 10) patients during desire (DC) and reason (RC) contexts Reward context
DC
RC
Hemodynamic activity (contrast -1 0 2 -1)
Increased
Increased
Region
Cluster size
Peak voxel
k
T
p
MNI coordinates x
y
z
VST R
46
5.77
0.00007
15
6
-3
VST L
17
4.72
0.001
-15
3
-3
PrC L
94
5.26
0.000003
-36
-6
48
TP L
492
5.11
0.000005
-54
-48
15
LOFC L
19
4.79
0.00001
-24
36
-12
PCC
34
4.30
0.00006
0
-27
30
MFG R IPS R
8 58
4.14 4.09
0.0001 0.0001
36 51
6 -48
45 48
TOc L
23
4.06
0.0001
-51
-60
-3
IFG R
11
4.01
0.0001
48
18
24
INS L
16
3.92
0.0002
-39
-6
3
SMA L
36
3.88
0.0002
-3
6
45
avPFC
13
3.68
0.0004
-39
42
6
rACC R
10
3.58
0.017
9
39
6
PCN L
10
3.45
0.0003
-6
-60
63
PCC L
8
3.43
0.0003
-3
-27
30
VST R
5
3.55
0.017
9
6
-3
VST ventral striatum, VP ventral pallidum, PrC precentral, TP temporoparietal, LOFC lateral orbitofrontal cortex, PCC posterior cingulate cortex, MFG middle frontal gyrus, IPS intraparietal sulcus, TOc temporooccipital, IFG inferior frontal gyrus, INS insula, SMA supplementary motor area, avPFC anterior ventral prefrontral cortex, rACC rostral anterior cingulate cortex, PCN precuneus Italic values: p \ 0.05 corrected, normal values: p \ 0.001 uncorrected
study, we could not identify relevant differences in clinical scores of depression between LA and HA (see Table 1), which is in line with the idea that different
123
pathophysiological processes in the brain may lead to similar depressive symptoms. Clinical parameters such as age at onset, duration of illness, and number of depressive
Eur Arch Psychiatry Clin Neurosci (2015) 265:57–66
episodes did not differ between our two subgroups of patients. Although these clinical differences were in general absent in our results, LA and HA presented important differences in medication profile. LA patients were mostly under serotonergic monotherapy, in contrast to HA patients among whom dual therapy with two or more AD agents was predominant. Considering that patients were grouped according to specific dysfunctional mechanisms and substrates, it was not surprising that the ADs prescribed in the clinic varied between subgroups. More striking was to observe that patient subgroups could not be characterized by any other of the clinical features evaluated. Pharmacological fMRI studies have been investigating AD interventions longitudinally. Studies [25, 26] evaluated the effect of serotonergic or dual agents modulating striatal reactivity to reward and positive affect, which correlated with clinical improvement. Based on our results, it would be very relevant to evaluate whether functional response to reward stimuli at the NAcc could predict beforehand the most effective AD agent for a certain patient. Medicationfree remitted patients have been shown to sustain impaired neural responses to reward stimuli [27], pointing toward a subclinical susceptibility to depression probably due to other factors such as genetic predisposition [28]. Further, associated genetic and imaging studies should investigate this question, especially in the scope of more individualized treatment responses and improvement in treatmentresistant depressions [29, 30]. Another important area showing differentially high hemodynamic responses to reward stimuli in HA in comparison with others was the rACC, also known as pregenual ACC. This region has been frequently reported in the literature of depression as a potential predictor of AD therapy response (for review, see [31, 32]). Impaired rACC deactivation during the performance of goal-oriented tasks has been associated with poorer AD response. Furthermore, in patients with AD refractory depression, deep brain stimulation at the NAcc induced deactivation of the rACC [33] and clinical improvement. Although the exact role of the rACC remains unclear, these findings suggest that the interaction between the rACC and the NAcc might comprise very relevant therapeutic value. Our results suggest that the hyperactivity of the rACC may be a potential biomarker of a subtype of patients with depression, showing dysfunctional top-down modulation of the NAcc in a reward task. This is in line with the results of a previous study with non-medicated patients presenting no NAcc hypoactivation during another reward task [13], where the hyperactivity of the rACC was also identifiable. Top-down deficiency has already been shown to play a role in the absence of amygdala suppression in depressed individuals [34]. Taken together, it seems plausible to us that specific pathomechanisms of depression might result
65
in abnormal involvement of the rACC in goal-directed tasks. Therefore, we speculate that the rACC, hyperactive in HAs during reward processing, could signalize this subgroup as eligible for particular AD agents. This question could be addressed by longitudinal trials using our DRD paradigm. Our findings should be carefully interpreted according to the study limitations. Comorbidities were present in some of the patients, but did not differ significantly between groups. In the LA group, they included nicotine dependency (3/10 patients), alcohol misuse (2/10 patients), cannabis misuse (1/10 patient), social phobia (1/10 patient), panic disorder (1/10 patient), and anankastic personality disorder (1/10 patient). In the HA group, they included nicotine dependency (5/10 patients), alcohol misuse (1/10 patient), cannabis misuse (1/10 patient), somatoform disorder (1/10 patient), dysthymia (1/10 patient), post-traumatic stress disorder (1/10 patient), and other specific personality disorders (F60.8, 2/10 patients). An association between dysfunctional bottom-up or top-down mechanisms and the choice of specific AD classes was seen, but with our study design, we cannot claim this to be a result of pharmacological interventions as the cross-sectional design clearly limits the evaluation of causative effects. Additionally, two non-medicated patients in each patient subgroup presented very low and two presented very high NAcc beta weights, which spoke in favor of differential underpinnings (pathomechanisms) existing before medication intake. Logically, this could only be proven by further longitudinal reward studies evaluating, e.g., ventral striatal, ventral tegmental, and rACC functional activity, along disease onset in highrisk samples. We suggest such studies to confirm and further characterize the association between the hyperactivation of rACC and deficient top-down mechanisms in the NAcc. Most of all, the potential benefit of revealing mechanism-related dysfunctional substrates could extend to more effective clinical decisions and substrate-specific therapeutic individualization with novel agents.
Conclusions Using a reward task, our results confirmed the existence of dysfunctional bottom-up mechanisms in patients with lower NAcc activation (LA subgroup) and dysfunctional top-down mechanisms in patients with higher NAcc activation (HA subgroup). This endorsed the functional heterogeneity in this region suggested by inconsistent findings in the literature of depression. Most importantly, we demonstrate that segregating patients by dysfunctional mechanisms may allow the identification of more specific neurobiological subtypes of depression with similar pathophysiological substrates.
123
66 Acknowledgments We acknowledge our colleague Bernd Kraemer for constant support and all MR technicians for competent imaging acquisition. We are very grateful to our participants Conflict of interest The authors report no biomedical relationships with financial or commercial interests and identify no potential conflicts of interest.
References 1. Haber SN, Knutson B (2010) The reward circuit: linking primate anatomy and human imaging. Neuropsychopharmacology 35(1):4–26 2. Sesack SR, Grace AA (2010) Cortico-Basal Ganglia reward network: microcircuitry. Neuropsychopharmacology 35(1):27–47 3. Savitz J, Drevets WC (2009) Bipolar and major depressive disorder: neuroimaging the developmental-degenerative divide. Neurosci Biobehav Rev 33(5):699–771 4. Elliott R, Friston KJ, Dolan RJ (2000) Dissociable neural responses in human reward systems. J Neurosci 20(16):6159–6165 5. Galvan A, Hare TA, Davidson M, Spicer J, Glover G, Casey BJ (2005) The role of ventral frontostriatal circuitry in reward-based learning in humans. J Neurosci 25(38):8650–8656 6. Wrase J, Kahnt T, Schlagenhauf F, Beck A, Cohen MX, Knutson B et al (2007) Different neural systems adjust motor behavior in response to reward and punishment. Neuroimage 36(4):1253–1262 7. Yacubian J, Glascher J, Schroeder K, Sommer T, Braus DF, Buchel C (2006) Dissociable systems for gain- and loss-related value predictions and errors of prediction in the human brain. J Neurosci 26(37):9530–9537 8. Akitsuki Y, Sugiura M, Watanabe J, Yamashita K, Sassa Y, Awata S et al (2003) Context-dependent cortical activation in response to financial reward and penalty: an event-related fMRI study. Neuroimage 19(4):1674–1685 9. Nieuwenhuis S, Heslenfeld DJ, von Geusau NJA, Mars RB, Holroyd CB, Yeung N (2005) Activity in human reward-sensitive brain areas is strongly context dependent. Neuroimage 25(4):1302–1309 10. Pizzagalli DA, Holmes AJ, Dillon DG, Goetz EL, Birk JL, Bogdan R et al (2009) Reduced caudate and nucleus accumbens response to rewards in unmedicated individuals with major depressive disorder. Am J Psychiatr 166(6):702–710 11. Robinson OJ, Cools R, Carlisi CO, Sahakian BJ, Drevets WC (2012) Ventral striatum response during reward and punishment reversal learning in unmedicated major depressive disorder. Am J Psychiatr 169(2):152–159 12. Steele JD, Kumar P, Ebmeier KP (2007) Blunted response to feedback information in depressive illness. Brain 130:2367–2374 13. Knutson B, Bhanji JP, Cooney RE, Atlas LY, Gotlib IH (2008) Neural responses to monetary incentives in major depression. Biol Psychiatry 63(7):686–692 14. Mitterschiffthaler MT, Kumari V, Malhi GS, Brown RG, Giampietro VP, Brammer MJ et al (2003) Neural response to pleasant stimuli in anhedonia: an fMRI study. NeuroReport 14(2):177–182 15. Diekhof EK, Gruber O (2010) When desire collides with reason: functional interactions between anteroventral prefrontal cortex and nucleus accumbens underlie the human ability to resist impulsive desires. J Neurosci 30(4):1488–1493 16. Diekhof EK, Nerenberg L, Falkai P, Dechent P, Baudewig J, Gruber O (2011) Impulsive personality and the ability to resist immediate reward: an fMRI study examining interindividual differences in the neural mechanisms underlying self-control. Hum Brain Mapp 33(12):2768–2784
123
Eur Arch Psychiatry Clin Neurosci (2015) 265:57–66 17. Diekhof EK, Keil M, Obst K, Henseler I, Dechent P, Falkai P, Gruber O (2012) A functional neuroimaging study assessing gender differences in the neural mechanisms underlying the ability to resist impulsive desires. Brain Res 1473:63–77 18. Cardoso de Almeida JR, Phillips ML (2013) Distinguishing between unipolar depression and bipolar depression: current and future clinical and neuroimaging perspectives. Biol Psychiatry 73(2):111–118 19. Ghaemi SN, Vohringer PA (2011) The heterogeneity of depression: an old debate renewed. Acta Psychiatr Scand 124(6):497 20. Østergaard SD, Jensen SOW, Bech P (2011) The heterogeneity of the depressive syndrome: when numbers get serious. Acta Psychiatr Scand 124(6):495–496 21. Sanacora G, Gueorguieva R, Epperson CN, Wu YT, Appel M, Rothman DL et al (2004) Subtype-specific alterations of gammaaminobutyric acid and glutamate in patients with major depression. Arch Gen Psychiatry 61(7):705–713 22. Drevets WC, Price JL, Bardgett ME, Reich T, Todd RD, Raichle ME (2002) Glucose metabolism in the amygdala in depression: relationship to diagnostic subtype and plasma cortisol levels. Pharmacol Biochem Behav 71(3):431–447 23. Fountoulakis KN, Iacovides A, Gerasimou G, Fotiou F, Ioannidou C, Bascialla F et al (2004) The relationship of regional cerebral blood flow with subtypes of major depression. Prog Neuro Psychopharmacol Biol Psychiatry 28(3):537–546 24. Farabaugh AH, Mischoulon D, Fava M, Green C, Guyker W, Alpert J (2004) The potential relationship between levels of perceived stress and subtypes of major depressive disorder (MDD). Acta Psychiatr Scand 110(6):465–470 25. Stoy M, Schlagenhauf F, Sterzer P, Bermpohl F, Hagele C, Suchotzki K et al (2012) Hyporeactivity of ventral striatum towards incentive stimuli in unmedicated depressed patients normalizes after treatment with escitalopram. J Psychopharmacol 26(5):677–688 26. Heller AS, Johnstone T, Light SN, Peterson MJ, Kolden GG, Kalin NH et al (2013) Relationships between changes in sustained fronto-striatal connectivity and positive affect in major depression resulting from antidepressant treatment. Am J Psychiatr 170(2):197–206 27. McCabe C, Cowen PJ, Harmer CJ (2009) Neural representation of reward in recovered depressed patients. Psychopharmacology 205(4):667–677 28. Gotlib IH, Hamilton JP, Cooney RE, Singh MK, Henry ML, Joormann J (2010) Neural processing of reward and loss in girls at risk for major depression. Arch Gen Psychiatry 67(4):380–387 29. Fagiolini A, Kupfer DJ (2003) Is treatment-resistant depression a unique subtype of depression? Biol Psychiatry 53(8):640–648 30. Gaynes BN (2009) Identifying difficult-to-treat depression: differential diagnosis, subtypes, and comorbidities. J Clin Psychiatry 70:10–15 31. Mayberg HS, Lozano AM, Voon V, McNeely HE, Seminowicz D, Hamani C, Schwalb JM, Kennedy S (2005) Deep brain stimulation for treatment-resistant depression. Neuron 3(45):651–660 32. Pizzagalli DA (2011) Frontocingulate dysfunction in depression: toward biomarkers of treatment response. Neuropsychopharmacology 36(1):183–206 33. Schlaepfer TE, Cohen MX, Frick C, Kosel M, Brodesser D, Axmacher N et al (2008) Deep brain stimulation to reward circuitry alleviates anhedonia in refractory major depression. Neuropsychopharmacology 33(2):368–377 34. Johnstone T, van Reekum CM, Urry HL, Kalin NH, Davidson RJ (2007) Failure to regulate: counterproductive recruitment of topdown prefrontal-subcortical circuitry in major depression. J Neurosci 27(33):8877–8884
ORIGINAL PAPER
Dissociating pathomechanisms of depression with fMRI: bottomup or top-down dysfunctions of the reward system Roberto Goya-Maldonado • Kristina Weber Sarah Trost • Esther Diekhof • Maria Keil • Peter Dechent • Oliver Gruber
•
Received: 23 April 2014 / Accepted: 6 October 2014 / Published online: 21 October 2014 Ó Springer-Verlag Berlin Heidelberg 2014
Abstract Depression is a debilitating psychiatric disorder characterized among other aspects by the inability to properly experience or respond to reward. However, it remains unclear whether patients with depression present impaired reward system due to abnormal modulatory mechanisms. We investigated the activation of the nucleus accumbens (NAcc), a crucial region involved in reward processing, with functional magnetic resonance imaging using the desire-reason-dilemma paradigm. This task allows tracking the activity of the NAcc during the acceptance or the rejection of previously conditioned reward stimuli. Patients were assigned into subgroups of lower (LA) or higher (HA) NAcc activation according to beta weights. LA patients presented significant hypoactivation in the ventral tegmental area in addition to bilateral
Electronic supplementary material The online version of this article (doi:10.1007/s00406-014-0552-2) contains supplementary material, which is available to authorized users. R. Goya-Maldonado (&) K. Weber S. Trost E. Diekhof M. Keil O. Gruber Center for Translational Research in Systems Neuroscience and Psychiatry, Department of Psychiatry and Psychotherapy, University Medical Center, Georg August University, Go¨ttingen, Germany e-mail: [email protected] Present Address: E. Diekhof Neuroendocrinology Unit, Department of Human Biology, Biocenter Grindel and Zoological Museum, University of Hamburg, Hamburg, Germany P. Dechent Research Group ‘MR-Research in Neurology and Psychiatry’, Department of Cognitive Neurology, University Medical Center, Georg August University, Go¨ttingen, Germany
ventral striatum, confirming impairments in the bottom-up input to the NAcc. Conversely, HA patients presented significant hyperactivation in prefrontal areas such as the rostral anterior cingulate cortex and the anterior ventral prefrontal cortex in addition to bilateral ventral striatum, suggesting disturbances in the top-down regulation of the NAcc. Demographic and clinical differences explaining the abnormal co-activations of midbrain and prefrontal regions were not identified. Therefore, we provide evidence for dysfunctional bottom-up processing in one potential neurobiological subtype of depression (LA) and dysfunctional top-down modulation in another subtype (HA). We suggest that the midbrain and prefrontal regions are more specific pathophysiological substrates for each depression subtype. Above all, our results encourage the segregation of patients by similar dysfunctional mechanisms of the dopaminergic system, which would finally contribute to disentangle more specific pathogeneses and guide the development of more personalized targets for future therapies. Keywords Depression Nucleus accumbens Reward Pathophysiological subtypes Neuroimaging biomarkers Stratified therapy Abbreviations AD Antidepressant DC Desire Context HA High NAcc Activity LA Low NAcc Activity OFC Orbitofrontal Cortex NAcc Nucleus Accumbens rACC Rostral Anterior Cingulate Cortex RC Reason Context VTA Ventral Tegmental Area
123
58
Introduction Patients with depression demonstrate a genuine inability to experience reward, which strongly impairs motivation- and incentive-based learning, both necessary for daily adaptive behaviors [1]. At the center of the reward system is the nucleus accumbens (NAcc) [1–3] receiving dopaminergic bottom-up input from the ventral tegmental area (VTA) and substantia nigra (SN) and further extending it to the ventral striatum, to the ventral pallidum and back to the initial midbrain areas. The VTA sends bundles through thalamic nuclei [1, 2] to dorsal, medial, and orbital prefrontal regions so they can top-down modulate the activity of subcortical structures, closing a very refined loop. Imaging studies extended the knowledge about the dynamic involvement of the NAcc in different contexts of reward [4–9]. Although identifying the foundations for reward impairment in depression should be fundamental to comprehend pathogeneses and/or refine treatment options, this field has so far remained mostly unexplored. Some studies investigating reward processing in depression reported hypoactivity of the NAcc and ventral striatum [10–12], which was not always consistently replicated [13, 14]. Such heterogeneity in the NAcc could result from disturbed dopaminergic bottom-up input or impaired top-down regulation of the NAcc, as the model suggested by Savitz and Drevets [3]. Our research group has previously investigated the NAcc activity of healthy volunteers with functional magnetic resonance imaging (fMRI) using the desire-reason-dilemma (DRD) paradigm [15–17]. This task comprises two different contexts, one in which participants were allowed to freely accept previously conditioned reward stimuli and another, when these conditioned reward stimuli had to be rejected. In the first context, the NAcc is activated by dopaminergic bottom-up mechanisms, whereas in the second context, this activation is suppressed by top-down regulatory mechanisms. Here, we used this paradigm to investigate the levels NAcc functional activation in patients diagnosed with major depressive disorder. We expected patients displaying lower (LA) NAcc activation to exhibit dysfunctional coactivations in regions associated with bottom-up mechanisms, whereas those showing higher (HA) NAcc activation were anticipated to present concomitant dysfunctional regions related to top-down mechanisms.
Methods Participants Twenty-five inward and outward symptomatic patients diagnosed with major depressive disorder according to
123
Eur Arch Psychiatry Clin Neurosci (2015) 265:57–66
DSM-IV criteria and coming for treatment at the University Medical Center Goettingen were recruited. Inclusion criteria were 18- to 60-year-old right-handed patients. The exclusion criteria comprised past or actual presence of other axis I diagnoses, neurological, or other medical conditions that could be related to affective symptoms. As our study was majorly designed to evaluate the heterogeneity of the NAcc activation during reward, the ability of performing the reward desire-reason paradigm in patients had to be reassured. Thus, patients were expected to accept a minimum of a third (10/30) of the conditioned reward stimuli in the context named desire (DC) and reject more than a half (20/40) of the reward stimuli in the context named reason (RC). Otherwise, they would be considered not engaging in the task demands and therefore would be excluded from the final analysis. Also, in favor of not having the NAcc reactivity overly driven by demographic factors such as age and gender, each healthy control was selected to match age and gender characteristics of the patients enrolled, following identical study protocols. Controls reported no medical history, including neurological and psychiatric history, as well as no previous or actual use of psychotropic medication. Each participant received up to 30 euros according to task performance in our reward paradigm, which were added to the general reimbursement for participation. The project was approved by the local ethics committee. Participants provided written informed consent before investigation and after the study had been fully explained. Experimental procedure Outside the scanner, the experiment started with an operant conditioning task, in which participants associated specific stimuli colors to a monetary reward or to a neutral outcome. Squares of eight different colors were presented 20 times each in a randomized sequence. Two of them led to immediate reward, when collected with a button press, by a positive feedback. In the MRI environment, the task involved event-related blocks of two contexts: DC, in which participants were free to accept conditioned reward stimuli, and RC, in which conditioned reward stimuli had to be declined. Each trial had a duration of 1,900 ms, starting with 200 ms of blank screen, 900 ms of stimulus presentation, 700 ms of immediate feedback, and 100 ms of blank screen (for comprehensive task description, see [17]). In DC, NAcc activation would be elicited by the acceptance of previously conditioned stimuli. In RC, the activation of this region would be suppressed by prefrontal top-down mechanisms, as conditioned stimuli had to be declined. When a reward stimulus was collected in RC, the actual block terminated with the feedback ‘‘goal failure’’ and a new one started. In this way, the desire-reason-
Eur Arch Psychiatry Clin Neurosci (2015) 265:57–66
dilemma paradigm forces the participants to overcome the behavioral tendency to respond to conditioned reward stimuli in order to achieve a superordinate goal [15–17]. Data acquisition Clinical parameters Patients were systematically assessed with standardized psychometric scales administered by a trained investigator blinded to the research question. In addition, the Wechsler Adult Intelligence Scale (WAIS) vocabulary score, the Montgomery–Asberg Depression Rating Scale (MADRS), and the Beck Depression Inventory (BDI), clinical information such as age at illness onset, duration of illness, number of depressive episodes, and actual antidepressant (AD) treatment plan was collected for post hoc comparisons between subgroups. Functional neuroimaging Whole-brain fMRI was performed on 3 T system (Magnetom TRIO, Siemens Healthcare, Erlangen, Germany) using a 8-channel head coil, covering the whole brain with 31 slices, AC–PC aligned, ascending orientation, 64 9 64 matrix, 3-mm thickness, 0.6-mm spacing with a gradient-echo EPI sequence (TR 1.9 s, flip angle 70°, FOV 192 mm, TE 30 ms). Structural T1-weighted at 1-mm isotropic resolution was also acquired. The total time of acquisition for structural and functional protocols with the desire-reason-dilemma paradigm was approximately 12 min. Data analysis Behavioral and demographic data Behavioral data were acquired using Presentation software (Version 14.9, www.neurobs.com). Log files, accuracy (as number of completed blocks), and reaction times for correct responses were entered into a group-by-block analysis of variance (ANOVA two-tailed p \ 0.05) in SPSS (version 20, SPSS Inc., Chicago, Illinois, USA). Demographic and clinical data were compared between subgroups in SPSS in the same fashion. Imaging data Functional images were slice time corrected and unwarped after four initial volumes were discarded, realigned to the fifth volume using rigid body transformation, normalized to the EPI template in Montreal Neurological Institute (MNI) stereotactic space with 3 9 3 9 3 mm3 voxel resolution,
59
and spatially smoothed (FWHM 9 mm) using SPM8 (www.fil.ion.ucl.ac.uk/spm8) and MATLAB (The MathWorks, Inc., Natick, MA, USA). We carefully checked for motion artifacts and set a cutoff (over the entire experiment) of\2 mm for the three planes of translation and\3° for the three planes of rotation as a control of data quality. First-level contrasts for correct response trials were defined for DC and RC blocks. Marsbar (http://marsbar.source forge.net) was used to extract the beta weights from DC correct events with a 5-mm-radius sphere placed in the NAcc region centered at [x = 12, y = 12, z = -4] MNI coordinates. The levels of activation in the acceptance of conditioned reward stimuli allocated patients in either lower (LA) NAcc activity or higher (HA) NAcc activity. Second-level random effect analyses were performed for statistical inference on the group level for LA, HA, and respective matched controls, C\LA[ and C\HA[. To test for differences as a proof of concept, one-way ANOVA between groups was implemented in spm8 as suggested by Henson and Penny (http://www.fil.ion.ucl.ac.uk/*wpenny/ publications/rik_anova.pdf). A significant group effect at the right NAcc was centered at [x = 15, y = 6, z = -3] MNI coordinates (Fig. 1a, open arrow). The F contrast given was ½1 1=3 1=3 1=3; 1=3 1 1=3 1=3; 1=3 1=3 1=3 1; 1=3 1=3 1=3 1 . Notably, the main effect was driven by the groups LA and HA (Fig. 1b). However, considering that some variation in the NAcc activity was seen between in C\LA[ and C\HA[, we confirmed the coactivations with independent-sample t tests. A more conservative contrast regarding demographic factors between patient subgroups and paired controls for the results of each patient subgroup was taken. The t contrasts entered were ½1 1=2 1=2 0 (Fig. 2) and ½1=2 0 1 1=2 (Fig. 3). This allowed us to additionally disregard age and gender variations accounting for potential differences seen in the NAcc. Small volumes with 10 mm radius (Supplement 1) were a priori defined for multiple testing corrections (p \ 0.05) based on our hypotheses and on previous studies [15–17] using the DRD paradigm.
Results Behavioral and demographic data We recruited 25 patients and from those 20 (80 %) patients were able to perform the minimum requirements of our DRD paradigm as described in the exclusion criteria. Twenty healthy controls were demographically selected to match age and gender characteristics of the patients enrolled. Final analyses included 20 pairs of patients and healthy controls well matched for age and gender. Considering that dysfunctional bottom-up mechanisms should
123
60
Eur Arch Psychiatry Clin Neurosci (2015) 265:57–66
Fig. 1 Heterogeneous hemodynamic responses in the nucleus accumbens (NAcc). a Coronal sections showing small volumes covering the NAcc region bilaterally (in brown) and significant between-group differences in functional activation (in yellow) using the DRD paradigm (open arrow signalizes the right NAcc); b contrast estimates (mean, 90 % CI) at the right NAcc; c design matrix and F contrast
given as described in the ‘‘Methods’’ section; LA: patients with low hemodynamic responses in the NAcc (n = 10), C\LA[: age-/gendermatched controls for LA, HA: patients with high hemodynamic responses in the NAcc (n = 10), C\HA[: age-/gender-matched controls for HA
lead to low activity (LA) at the NAcc region and dysfunctional top-down mechanisms should lead to high activity (HA) at this region, patients were allocated into subgroups (LA or HA) according to their NAcc levels of activity as described in methods. In Table 1, demographic, behavioral, and clinical data are compared between patient subgroups and their controls.
Top-down dysfunction
Bottom-up dysfunction Bottom-up processes were evaluated when a reward stimulus was collected in DC. A robust hypoactivation for LA patients in comparison with C\LA[ and HA was seen bilaterally at ventral striatal region (Fig. 2a), indicating deficient bottom-up input in this patient group. Other significant hypoactive regions were seen at this contrast, among them the VTA as one of our regions of interest (Fig. 2b; Table 2). These findings were specific for LA, considering that neither C\LA[ nor HA presented differences in the VTA or other midbrain regions (Fig. 2b; Table 3).
123
Top-down processes were evaluated in RC, during which NAcc activation elicited by the presentation of reward stimuli should be physiologically suppressed (signalized as open arrows in controls, Fig. 2B ? C). In comparison with LA group and C\HA[, HA patients showed significantly increased bilateral ventral striatal activation in RC, revealing a lack of modulatory top-down regulation of the reward system (Fig. 2B ? C). When evaluated for potential differences in RC, LA patients did not present significant differences (see RC in Table 2), indicating that areas known to take part in top-down mechanisms were not deviant from controls in this group. Whole-brain analyses Whole-brain analyses were performed in order to identify further functional abnormalities in the two subgroups of depressive patients with bottom-up or top-down deficiencies of the reward system (Tables 2, 3). In general,
Eur Arch Psychiatry Clin Neurosci (2015) 265:57–66
61
Fig. 2 Dysfunctional bottom-up mechanism. a Left coronal slice (y = 6) displaying reduced hemodynamic responses in the nucleus accumbens (NAcc) in patients with ventral striatal LA (n = 10) in comparison with both age-/gender-matched controls (C\LA[, n = 10) and HA (n = 10) during desire context (DC) of the DRD paradigm; right contrast estimates (mean, 90 % CI) at the NAcc; b left axial
slice (z = -12) presenting reduced hemodynamic response in the ventral tegmental area (VTA) in patients with ventral striatal LA (n = 10) in comparison with age-/gender-matched controls (C\LA[, n = 10) and HA (n = 10) during desire context (DC) of the DRD paradigm; right contrast estimates (mean, 90 % CI) at the VTA; regions listed in Table 2
differential brain activations were seen in subcortical and cortical areas known to be involved in reward processing. The right orbitofrontal cortex (OFC) and the dorsal ACC showed reduced activity in LA patients in comparison with the other subgroups. Furthermore, the frontal eye fields were bilaterally hypoactive in LA patients in response to reward stimuli. In contrast, HA patients revealed hyperresponsivity to reward stimuli in the left OFC and other prefrontal regulatory regions such as the inferior and middle frontal gyrus as well as the avPFC. Taken together, LA patients showed downregulated regions involved in reward processing in contrast to HA, who presented upregulated regions in DC. No significant differences were
seen in RC for LA, whereas HA presented hyperactive posterior cingulate cortex and precuneus.
Discussion A hallmark of depression is the inability to properly experience reward. In the present study, we investigated reward processing in a group of 20 acutely depressed patients, focusing on the functional response of the NAcc to a reward task. Consistent with our hypothesis, NAcc activity and modulation in patients diverged abnormally from controls, when challenged by two reward-associated
123
62
Eur Arch Psychiatry Clin Neurosci (2015) 265:57–66
Fig. 3 Dysfunctional top-down mechanism. a Sagittal slice (x = 9) presenting increased hemodynamic responses in the nucleus accumbens (NAcc) in patients with ventral striatal HA (n = 10) in comparison with both age-/gender-matched controls (C\HA[, n = 10) and LA (n = 10) during desire context (DC) of the DRD paradigm; b contrast estimates (mean, 90 % CI) in the NAcc during
desire context (DC) of the DRD paradigm; c contrast estimates (mean, 90 % CI) at the NAcc during reason context (RC) of the DRD paradigm, open arrows signalize the top-down suppression from DC to RC (B ? C); d coronal slices (y = 6, y = 42) from a with crosshairs at the right NAcc and rostral ACC (rACC), respectively; regions listed in Table 3
contexts, DC and RC, of the DRD paradigm. More specifically, one group of patients displayed significantly lower NAcc activation (named here as LA patients) in comparison with matched controls, named as C\LA[. Other brain regions that were hypoactive in LA patients included the VTA, a region involved in bottom-up mechanisms of the dopaminergic system, confirming our a priori hypothesis of dysfunctional dopaminergic input. Conversely, the other group of patients showed significant hyperactivation and a lack of top-down regulation of the NAcc in comparison with matched controls, named here as HA and C\HA[, respectively. In these HA patients, hyperactivation was also found in prefrontal regions such as the avPFC, providing further support for our initial hypothesis of dysfunctional top-down modulation of the NAcc. Matched healthy controls showed NAcc activation during DC that was reduced during RC, replicating our previous findings
[15–17]. In contrast, patients did not present this adaptability in the NAcc activity across DC and RC, indicating dysfunctional modulatory mechanisms. Overall, our results provide specific neural substrates underlying dysfunctional bottom-up or top-down mechanisms as the model suggested for pathophysiological changes in depressive disorders [3]. This is also in agreement with the observation of heterogeneous NAcc hemodynamic responses during reward processing in patients with depression [10–14], which to the best of our knowledge is first explored here. The need to adjust to a frequently changing environment has possibly favored the development of complex neural regulatory substrates involved in reward processing. FMRI studies have extended the knowledge about the neural implementation [4–6] and adaptability [7–9] of the reward system in the healthy brain. In this study, we addressed the clinical inability to react to reward stimuli during the
123
Eur Arch Psychiatry Clin Neurosci (2015) 265:57–66 Table 1 Comparison of demographic, behavioral, psychometric, and general clinical parameters between subgroups of patients with depression exhibiting lower nucleus accumbens (NAcc) activity (LA), or higher NAcc activity (HA), and their age- and gender-matched controls in desire (DC) and reason contexts (RC) of the DRD paradigm
63
LA
HA
C\LA[
C\HA[
ANOVA
LA versus HA
N = 10 M (SD)
N = 10 M (SD)
N = 10 M (SD)
N = 10 M (SD)
F
t/v2
Age (year)
34.1 (9.6)
37.1 (11.3)
34.2 (9.5)
32.7 (10.3)
.32
.808
.64
.532
Education (year)
14.2 (2.5)
15.5 (4.2)
15.6 (3.5)
13.9 (3.1)
.61
.614
.83
.412
p, 2-tail
p, 2-tail
Gender (female/total)
7/10
7/10
6/10
8/10
.95
.813
.00
Truncated sequences (n)
12.5 (5.4)
17.4 (7.8)
10.6 (6.8)
13.8 (8.7)
1.54
.220
1.63
1 .120
Response time (ms) in DC
519.0 (28.8)
554.2 (69.4)
525.2 (56.9)
548.66 (51.6)
1.03
.390
1.48
.156
Response time (ms) in RC
522.3 (51.9)
570.4 (78.1)
521.91 (54.6)
559.92 (66.1)
1.57
.212
1.62
.122
WAIS vocabulary score
31.3 (3.5)
30.7 (3.7)
N/A
N/A
–
–
.38
.708
MADRS total score (10 items)
12.4 (6.3)
15.2 (5.5)
N/A
N/A
–
–
1.06
.302
BDI total score (21 items)
14.7 (10.8)
22.3 (8.9)
N/A
N/A
–
–
1.71
.103
Age first episode
23.4 (6.7)
27.4 (12.2)
N/A
N/A
–
–
.90
.378
Duration of disorder (year)
10.7 (9.5)
9.8 (10.5)
N/A
N/A
–
–
.20
.843
Depressive episodes (n)
6.5 (4.1)
5.1 (4.3)
N/A
N/A
–
–
1.12
.279
Actual AD plan
Antipressant (AD) monotherapy versus polytherapy between LA and HA v2 9.60 p = .008 Italic values: p \ 0.05
Escitalopram 10–30 mg/day
7/10
2/10
N/A
N/A
–
–
5.05
Sertraline 125 mg/day
1/10
1/10
N/A
N/A
–
–
.00
Mirtazapine 15–45 mg/day
1/10
4/10
N/A
N/A
–
–
2.40
.025 1 .121
Reboxetine 12 mg/day
–
1/10
N/A
N/A
–
–
1.05
.305
Venlafaxine 225–300 mg/ day
–
4/10
N/A
N/A
–
–
1.25
.264
Duloxetine 90 mg/day Pipamperon 40 mg/day
– –
1/10 1/10
N/A N/A
N/A N/A
– –
– –
1.05 1.05
.305 .305
Quetiapine 30 mg/day
–
1/10
N/A
N/A
–
–
1.05
.305
Olanzapine 5 mg/day
–
1/10
N/A
N/A
–
–
1.05
.305
Agomelatine 25 mg/day
–
1/10
N/A
N/A
–
–
1.05
.305
manifestation of depression by using an established paradigm to dissect mechanisms related to bottom-up and topdown influences on the NAcc [14–16]. In our sample of patients, heterogeneous NAcc hemodynamic reactivity to reward stimuli was observed, corroborating dysfunctional input or dysfunctional regulatory mechanisms. Dysfunctional bottom-up mechanisms in LA were evidenced by abnormal co-activation at the VTA, whereas in HA, dysfunctional top-down mechanisms were seen by abnormal avPFC and rACC co-activations. The innovation of this approach relies on the use of functional disturbances to define subgroups that may constitute different subtypes. This is exactly the reverse of defining patient subgroups by clinical parameters and evaluating their differential
correlates, which may comprise important limitations (for review, see [18]). Although the differential correlates between our patient subgroups in regard to NAcc heterogeneity could be consistently identified, they were not explained by demographic or clinical parameters in a post hoc evaluation. Clinical heterogeneity in depression is a theme of discussion per se in the literature [19, 20]. Previous studies [21, 22] have shown heterogeneous findings as a reflection of clinical subtypes of depression. In a SPECT study [23], patients with atypical depression presented increased right frontal perfusion in contrast to melancholic and undifferentiated depression. In addition, the atypical subtype has also been reported to perceive stress more intensively than other subtypes [24]. In our
123
64
Eur Arch Psychiatry Clin Neurosci (2015) 265:57–66
Table 2 Patients presenting low activity (LA, n = 10) in the ventral striatal region in comparison with both their age-/gender-matched controls (C\LA[, n = 10) and high activity (HA, n = 10) patients during desire (DC) and reason (RC) contexts Reward context
DC
Hemodynamic activity (contrast -2 1 1 0)
Decreased
Region
Peak voxel
k
T
p
MNI coordinates x
VST L
60
5.29
0.0002
VST R
84
5.73
0.00007
533 1,691
5.52 5.29
MTG L
37
dACC R LOFC R
TP L PCN L/R
RC
Cluster size
y
z
-15
3
-6
15
9
-3
0.000001 0.000003
-57 -9
-54 -48
18 45
4.39
0.00005
-45
-3
-18
22
4.16
0.00009
3
15
24
22
3.93
0.0002
30
39
-9 -12
VTA L/SN
6
3.68
0.014
-15
-18
MOG R
8
3.67
0.0004
33
-75
24
FEF R
10
3.65
0.0004
33
-6
57
FEF L
7
3.28
0.001
-27
12
51
No differences
VST ventral striatum, TP temporoparietal, PCN precuneus, MTG middle temporal gyrus, VTA ventral tegmental area, SN substantia nigra, dACC dorsal anterior cingulate cortex, LOFC lateral orbitofrontal cortex, MOG middle occipital gyrus, FEF frontal eye field Italic values: p \ 0.05 corrected, normal values: p \ 0.001 uncorrected
Table 3 Patients presenting high activity (HA, n = 10) in the ventral striatal region in comparison with both their age-/gender-matched controls (C\HA[, n = 10) and low activity (LA, n = 10) patients during desire (DC) and reason (RC) contexts Reward context
DC
RC
Hemodynamic activity (contrast -1 0 2 -1)
Increased
Increased
Region
Cluster size
Peak voxel
k
T
p
MNI coordinates x
y
z
VST R
46
5.77
0.00007
15
6
-3
VST L
17
4.72
0.001
-15
3
-3
PrC L
94
5.26
0.000003
-36
-6
48
TP L
492
5.11
0.000005
-54
-48
15
LOFC L
19
4.79
0.00001
-24
36
-12
PCC
34
4.30
0.00006
0
-27
30
MFG R IPS R
8 58
4.14 4.09
0.0001 0.0001
36 51
6 -48
45 48
TOc L
23
4.06
0.0001
-51
-60
-3
IFG R
11
4.01
0.0001
48
18
24
INS L
16
3.92
0.0002
-39
-6
3
SMA L
36
3.88
0.0002
-3
6
45
avPFC
13
3.68
0.0004
-39
42
6
rACC R
10
3.58
0.017
9
39
6
PCN L
10
3.45
0.0003
-6
-60
63
PCC L
8
3.43
0.0003
-3
-27
30
VST R
5
3.55
0.017
9
6
-3
VST ventral striatum, VP ventral pallidum, PrC precentral, TP temporoparietal, LOFC lateral orbitofrontal cortex, PCC posterior cingulate cortex, MFG middle frontal gyrus, IPS intraparietal sulcus, TOc temporooccipital, IFG inferior frontal gyrus, INS insula, SMA supplementary motor area, avPFC anterior ventral prefrontral cortex, rACC rostral anterior cingulate cortex, PCN precuneus Italic values: p \ 0.05 corrected, normal values: p \ 0.001 uncorrected
study, we could not identify relevant differences in clinical scores of depression between LA and HA (see Table 1), which is in line with the idea that different
123
pathophysiological processes in the brain may lead to similar depressive symptoms. Clinical parameters such as age at onset, duration of illness, and number of depressive
Eur Arch Psychiatry Clin Neurosci (2015) 265:57–66
episodes did not differ between our two subgroups of patients. Although these clinical differences were in general absent in our results, LA and HA presented important differences in medication profile. LA patients were mostly under serotonergic monotherapy, in contrast to HA patients among whom dual therapy with two or more AD agents was predominant. Considering that patients were grouped according to specific dysfunctional mechanisms and substrates, it was not surprising that the ADs prescribed in the clinic varied between subgroups. More striking was to observe that patient subgroups could not be characterized by any other of the clinical features evaluated. Pharmacological fMRI studies have been investigating AD interventions longitudinally. Studies [25, 26] evaluated the effect of serotonergic or dual agents modulating striatal reactivity to reward and positive affect, which correlated with clinical improvement. Based on our results, it would be very relevant to evaluate whether functional response to reward stimuli at the NAcc could predict beforehand the most effective AD agent for a certain patient. Medicationfree remitted patients have been shown to sustain impaired neural responses to reward stimuli [27], pointing toward a subclinical susceptibility to depression probably due to other factors such as genetic predisposition [28]. Further, associated genetic and imaging studies should investigate this question, especially in the scope of more individualized treatment responses and improvement in treatmentresistant depressions [29, 30]. Another important area showing differentially high hemodynamic responses to reward stimuli in HA in comparison with others was the rACC, also known as pregenual ACC. This region has been frequently reported in the literature of depression as a potential predictor of AD therapy response (for review, see [31, 32]). Impaired rACC deactivation during the performance of goal-oriented tasks has been associated with poorer AD response. Furthermore, in patients with AD refractory depression, deep brain stimulation at the NAcc induced deactivation of the rACC [33] and clinical improvement. Although the exact role of the rACC remains unclear, these findings suggest that the interaction between the rACC and the NAcc might comprise very relevant therapeutic value. Our results suggest that the hyperactivity of the rACC may be a potential biomarker of a subtype of patients with depression, showing dysfunctional top-down modulation of the NAcc in a reward task. This is in line with the results of a previous study with non-medicated patients presenting no NAcc hypoactivation during another reward task [13], where the hyperactivity of the rACC was also identifiable. Top-down deficiency has already been shown to play a role in the absence of amygdala suppression in depressed individuals [34]. Taken together, it seems plausible to us that specific pathomechanisms of depression might result
65
in abnormal involvement of the rACC in goal-directed tasks. Therefore, we speculate that the rACC, hyperactive in HAs during reward processing, could signalize this subgroup as eligible for particular AD agents. This question could be addressed by longitudinal trials using our DRD paradigm. Our findings should be carefully interpreted according to the study limitations. Comorbidities were present in some of the patients, but did not differ significantly between groups. In the LA group, they included nicotine dependency (3/10 patients), alcohol misuse (2/10 patients), cannabis misuse (1/10 patient), social phobia (1/10 patient), panic disorder (1/10 patient), and anankastic personality disorder (1/10 patient). In the HA group, they included nicotine dependency (5/10 patients), alcohol misuse (1/10 patient), cannabis misuse (1/10 patient), somatoform disorder (1/10 patient), dysthymia (1/10 patient), post-traumatic stress disorder (1/10 patient), and other specific personality disorders (F60.8, 2/10 patients). An association between dysfunctional bottom-up or top-down mechanisms and the choice of specific AD classes was seen, but with our study design, we cannot claim this to be a result of pharmacological interventions as the cross-sectional design clearly limits the evaluation of causative effects. Additionally, two non-medicated patients in each patient subgroup presented very low and two presented very high NAcc beta weights, which spoke in favor of differential underpinnings (pathomechanisms) existing before medication intake. Logically, this could only be proven by further longitudinal reward studies evaluating, e.g., ventral striatal, ventral tegmental, and rACC functional activity, along disease onset in highrisk samples. We suggest such studies to confirm and further characterize the association between the hyperactivation of rACC and deficient top-down mechanisms in the NAcc. Most of all, the potential benefit of revealing mechanism-related dysfunctional substrates could extend to more effective clinical decisions and substrate-specific therapeutic individualization with novel agents.
Conclusions Using a reward task, our results confirmed the existence of dysfunctional bottom-up mechanisms in patients with lower NAcc activation (LA subgroup) and dysfunctional top-down mechanisms in patients with higher NAcc activation (HA subgroup). This endorsed the functional heterogeneity in this region suggested by inconsistent findings in the literature of depression. Most importantly, we demonstrate that segregating patients by dysfunctional mechanisms may allow the identification of more specific neurobiological subtypes of depression with similar pathophysiological substrates.
123
66 Acknowledgments We acknowledge our colleague Bernd Kraemer for constant support and all MR technicians for competent imaging acquisition. We are very grateful to our participants Conflict of interest The authors report no biomedical relationships with financial or commercial interests and identify no potential conflicts of interest.
References 1. Haber SN, Knutson B (2010) The reward circuit: linking primate anatomy and human imaging. Neuropsychopharmacology 35(1):4–26 2. Sesack SR, Grace AA (2010) Cortico-Basal Ganglia reward network: microcircuitry. Neuropsychopharmacology 35(1):27–47 3. Savitz J, Drevets WC (2009) Bipolar and major depressive disorder: neuroimaging the developmental-degenerative divide. Neurosci Biobehav Rev 33(5):699–771 4. Elliott R, Friston KJ, Dolan RJ (2000) Dissociable neural responses in human reward systems. J Neurosci 20(16):6159–6165 5. Galvan A, Hare TA, Davidson M, Spicer J, Glover G, Casey BJ (2005) The role of ventral frontostriatal circuitry in reward-based learning in humans. J Neurosci 25(38):8650–8656 6. Wrase J, Kahnt T, Schlagenhauf F, Beck A, Cohen MX, Knutson B et al (2007) Different neural systems adjust motor behavior in response to reward and punishment. Neuroimage 36(4):1253–1262 7. Yacubian J, Glascher J, Schroeder K, Sommer T, Braus DF, Buchel C (2006) Dissociable systems for gain- and loss-related value predictions and errors of prediction in the human brain. J Neurosci 26(37):9530–9537 8. Akitsuki Y, Sugiura M, Watanabe J, Yamashita K, Sassa Y, Awata S et al (2003) Context-dependent cortical activation in response to financial reward and penalty: an event-related fMRI study. Neuroimage 19(4):1674–1685 9. Nieuwenhuis S, Heslenfeld DJ, von Geusau NJA, Mars RB, Holroyd CB, Yeung N (2005) Activity in human reward-sensitive brain areas is strongly context dependent. Neuroimage 25(4):1302–1309 10. Pizzagalli DA, Holmes AJ, Dillon DG, Goetz EL, Birk JL, Bogdan R et al (2009) Reduced caudate and nucleus accumbens response to rewards in unmedicated individuals with major depressive disorder. Am J Psychiatr 166(6):702–710 11. Robinson OJ, Cools R, Carlisi CO, Sahakian BJ, Drevets WC (2012) Ventral striatum response during reward and punishment reversal learning in unmedicated major depressive disorder. Am J Psychiatr 169(2):152–159 12. Steele JD, Kumar P, Ebmeier KP (2007) Blunted response to feedback information in depressive illness. Brain 130:2367–2374 13. Knutson B, Bhanji JP, Cooney RE, Atlas LY, Gotlib IH (2008) Neural responses to monetary incentives in major depression. Biol Psychiatry 63(7):686–692 14. Mitterschiffthaler MT, Kumari V, Malhi GS, Brown RG, Giampietro VP, Brammer MJ et al (2003) Neural response to pleasant stimuli in anhedonia: an fMRI study. NeuroReport 14(2):177–182 15. Diekhof EK, Gruber O (2010) When desire collides with reason: functional interactions between anteroventral prefrontal cortex and nucleus accumbens underlie the human ability to resist impulsive desires. J Neurosci 30(4):1488–1493 16. Diekhof EK, Nerenberg L, Falkai P, Dechent P, Baudewig J, Gruber O (2011) Impulsive personality and the ability to resist immediate reward: an fMRI study examining interindividual differences in the neural mechanisms underlying self-control. Hum Brain Mapp 33(12):2768–2784
123
Eur Arch Psychiatry Clin Neurosci (2015) 265:57–66 17. Diekhof EK, Keil M, Obst K, Henseler I, Dechent P, Falkai P, Gruber O (2012) A functional neuroimaging study assessing gender differences in the neural mechanisms underlying the ability to resist impulsive desires. Brain Res 1473:63–77 18. Cardoso de Almeida JR, Phillips ML (2013) Distinguishing between unipolar depression and bipolar depression: current and future clinical and neuroimaging perspectives. Biol Psychiatry 73(2):111–118 19. Ghaemi SN, Vohringer PA (2011) The heterogeneity of depression: an old debate renewed. Acta Psychiatr Scand 124(6):497 20. Østergaard SD, Jensen SOW, Bech P (2011) The heterogeneity of the depressive syndrome: when numbers get serious. Acta Psychiatr Scand 124(6):495–496 21. Sanacora G, Gueorguieva R, Epperson CN, Wu YT, Appel M, Rothman DL et al (2004) Subtype-specific alterations of gammaaminobutyric acid and glutamate in patients with major depression. Arch Gen Psychiatry 61(7):705–713 22. Drevets WC, Price JL, Bardgett ME, Reich T, Todd RD, Raichle ME (2002) Glucose metabolism in the amygdala in depression: relationship to diagnostic subtype and plasma cortisol levels. Pharmacol Biochem Behav 71(3):431–447 23. Fountoulakis KN, Iacovides A, Gerasimou G, Fotiou F, Ioannidou C, Bascialla F et al (2004) The relationship of regional cerebral blood flow with subtypes of major depression. Prog Neuro Psychopharmacol Biol Psychiatry 28(3):537–546 24. Farabaugh AH, Mischoulon D, Fava M, Green C, Guyker W, Alpert J (2004) The potential relationship between levels of perceived stress and subtypes of major depressive disorder (MDD). Acta Psychiatr Scand 110(6):465–470 25. Stoy M, Schlagenhauf F, Sterzer P, Bermpohl F, Hagele C, Suchotzki K et al (2012) Hyporeactivity of ventral striatum towards incentive stimuli in unmedicated depressed patients normalizes after treatment with escitalopram. J Psychopharmacol 26(5):677–688 26. Heller AS, Johnstone T, Light SN, Peterson MJ, Kolden GG, Kalin NH et al (2013) Relationships between changes in sustained fronto-striatal connectivity and positive affect in major depression resulting from antidepressant treatment. Am J Psychiatr 170(2):197–206 27. McCabe C, Cowen PJ, Harmer CJ (2009) Neural representation of reward in recovered depressed patients. Psychopharmacology 205(4):667–677 28. Gotlib IH, Hamilton JP, Cooney RE, Singh MK, Henry ML, Joormann J (2010) Neural processing of reward and loss in girls at risk for major depression. Arch Gen Psychiatry 67(4):380–387 29. Fagiolini A, Kupfer DJ (2003) Is treatment-resistant depression a unique subtype of depression? Biol Psychiatry 53(8):640–648 30. Gaynes BN (2009) Identifying difficult-to-treat depression: differential diagnosis, subtypes, and comorbidities. J Clin Psychiatry 70:10–15 31. Mayberg HS, Lozano AM, Voon V, McNeely HE, Seminowicz D, Hamani C, Schwalb JM, Kennedy S (2005) Deep brain stimulation for treatment-resistant depression. Neuron 3(45):651–660 32. Pizzagalli DA (2011) Frontocingulate dysfunction in depression: toward biomarkers of treatment response. Neuropsychopharmacology 36(1):183–206 33. Schlaepfer TE, Cohen MX, Frick C, Kosel M, Brodesser D, Axmacher N et al (2008) Deep brain stimulation to reward circuitry alleviates anhedonia in refractory major depression. Neuropsychopharmacology 33(2):368–377 34. Johnstone T, van Reekum CM, Urry HL, Kalin NH, Davidson RJ (2007) Failure to regulate: counterproductive recruitment of topdown prefrontal-subcortical circuitry in major depression. J Neurosci 27(33):8877–8884
