Neural unresponsiveness to fairness in MDD patients — state or trait?

Aug 20, 2020

Background and Motivations

Depression is one the most common mental illnesses and can cause significant impairment in the lives of its sufferers. Some common symptoms include sleep disturbance, anhedonia, feelings of hopelessness, suicidal thoughts, fatigue, difficulty with decision-making, and a loss of pleasure and interest in activities that were once enjoyable (NIMH). With the symptoms and experiences of those with depression being so varied, identifying behavioral and neural differences in the brains of depressed individuals is an important step in the development of better treatments. The Research Domain Criteria (RDoC) framework emphasizes studying symptoms using a variety of neuroscientific and behavioral approaches, which is achieved by the proposed study. One fruitful lens consists in the identification of neural patterns during strategic gameplay. To this end, previous work has demonstrated neural differences during strategic decision-making games between those with psychiatric conditions and healthy control groups (Robson et al., 2020). However, there is a lack of longitudinal studies of neuroeconomic gameplay in patient populations, which has made it difficult to discern if differences in gameplay and neural activation are static factors or if these changes dynamically and directly result from conditions that can be changed through treatment.

In the context of depression specifically, various functional magnetic resonance imaging (fMRI) studies have identified differences in brain activation between depressed patients and healthy controls when playing the Ultimatum Game (UG). The UG is a well-established neuroeconomic paradigm in which a responder decides whether to accept or reject a proposer’s offer of a share of the proposer’s initial endowment. Should the responder accept, the split is implemented. However, if the responder rejects the offer, neither party receives any of the endowment. We consider this paradigm useful for studying neural signature correlations because it involves social cognition and a gradient of fairness, which is apt for regression analysis. As the fairness of the offers increased, the control group showed increased activation in the nucleus accumbens (NAcc) and the dorsal caudate (DC) while the depressed group did not (Gradin et al., 2015).

These areas are thought to respond to reward and social information (Gradin et al., 2015). Importantly, there were no differences in rejection rates between the depressed and control groups, a finding replicated by other groups (Mukherjee et al., 2020). There was also a diminished response in the medial occipital lobe in the depressed group relative to controls as unfairness increased (Gradin et al., 2015). Figures 1 and 2 summarize these findings.

Previous work has also shown occipital lobe functioning differences in depressed patients in emotion and social related tasks, as well as in other disorders (e.g. social anxiety and schizophrenia) that include social dysfunction symptoms (Li et al., 2013; Zhao et al., 2020; Goldin et al., 2009; Bjorkquist and Herbener, 2013). In addition, the nucleus accumbens has been used as a target for deep brain stimulation in treatment-resistant depression with promising results (Bewernick et al., 2010; Bewernick et al., 2012; Taghva et al., 2013), suggesting that decreases in NAcc responses may play a causal role in the etiology of depression. Given previous findings that established clear differences between depressed and control groups in UG decision-making, social cognition, and reward-related processing, we are choosing the nucleus accumbens, the dorsal caudate, and the medial occipital lobe as the regions of interest in our neuroimaging study.

Current treatments for depression include medications (typically selective serotonin reuptake inhibitors or SSRIs), psychotherapy, and deep brain stimulation or electroconvulsive therapy in the case of treatment-resistant depression. SSRIs are known to have mixed effects. Indeed, while some patients do exhibit a significant response, up to 70% will not exhibit symptom remission following their first course of SSRI treatment (Gaynes et al. 2009; Trivedi et al., 2006). Additionally, up to 40% of patients will be classified as treatment-resistant after exhibiting minimal response to sequential SSRI treatment attempts (Kulikov 2018; Gaynes et al., 2009). With most SSRIs taking roughly 4-6 weeks to take effect and a typical treatment period of several months to a year followed by a weaning off period, a longitudinal design is useful to capture any potential behavioral and neural changes as a result of treatment. Previous neuroimaging data supports the idea that SSRI treatment has a normalizing effect on the depressed brain and may also offer insights into why different individuals have different responses to treatment (Wessa and Lois, 2015). However, such studies have had shorter timelines than what we propose and did not investigate behavior in neuroeconomic gameplay as well.

We seek to address the lack of longitudinal studies of neuroeconomic gameplay in those with depression through a longitudinal fMRI study of both depressed patients and healthy controls. We hypothesize patients with depression will initially exhibit decreased fairness-related responses to ultimatums in the nucleus accumbens and dorsal caudate, and decreased unfairness-related responses in the medial occipital lobe, compared to control patients. Additionally, we hypothesize that medication-assisted improvement in depression burden will be associated with an attenuation of the diminished neural response to fairness in UG responders. Regardless of the results of this study, the findings will have important implications in both the fields of neuroeconomics and computational psychiatry.

Aims and Innovation

Ample data suggest that a number of event-related hemodynamic signals are altered in MDD patients compared to healthy controls. However, few neuroimaging studies follow patients beyond a few weeks following the initiation of treatment (Wessa and Lois, 2015). Almost none have established whether these signatures are static or dynamic. If these patterns are static, we would expect them to exist throughout an individual’s lifetime as a preexisting psychiatric risk factor. If they are dynamic, they should normalize with successful treatment and, therefore, reflect only the diagnostic state. Similarly, very few studies have examined the stability of neural findings related to neuroeconomic gameplay (Robson et al., 2020). Thus, the proposed longitudinal study would be among the first of its kind. Our findings will therefore offer unprecedented insight into the stability of event-related findings in neuroeconomic gameplay, alongside an improved understanding of the dynamic or static status of neural signatures of MDD.

Significant heterogeneity characterizes the remission response of MDD patients to SSRIs. Using a longitudinal study design combined with regression analyses, we hope to determine whether this heterogeneity is also represented in neural responses to unfairness and inequity that have previously been shown in MDD patients. We also hope that by approaching heterogeneity using regression analyses rather than by using a questionnaire cut-off score for “responders” vs. “nonresponders”, as has often been done in previous studies, we will more sensitively capture the extent of this relationship. We anticipate that over the course of treatment, initial neural differences between the depressed and control subjects will decrease most robustly in patients that show strong clinical response to SSRIs, showing that the SSRIs are able to aid in any neural dysfunction that may lead to clinical depression. By employing a longitudinal study that exceeds the length of the average treatment course for SSRI therapy, we also hope to capture whether SSRIs can correct brain states only while pharmacologically active (i.e. during the treatment course) or whether the longer-term application of SSRIs specifically result in corrective plastic changes and thus allow for long-term remission of behavior.

Methods

The proposed experimental timeline is summarized in Figure 3. During the Recruitment phase, we recruit two groups of participants: one group of MDD patients and one group of healthy controls. The MDD group will be recruited through psychiatrists, who will refer interested candidates upon diagnosis. The inclusion criteria for the MDD group are a diagnosis from a referring psychiatrist, expected SSRI treatment onset in responding patients within six weeks of the first scan and treatment-naive prior to the experiment. This final specification is so that we can clearly state that the findings at the time of the initial scan are the result of untreated depression. The controls will be recruited via local advertising. During an initial screening session, a clinician will administer the 17-item Hamilton Depression Rating Scale (HAMD) to both recruitment populations, and we exclude from the control group applicants with a HAMD score greater than 7, indicating the possibility of depression or dysphoria (Tolentina and Schmidt, 2018; Carrozzino et al., 2020) or any history of psychiatric disorders, including depression. We use HAMD to screen for controls for two reasons: it has been shown to reliably isolate individuals with low depression symptomatology, and it differs from the Beck Depression Inventory-II, our regressor, in that the former is rated by an interviewer rather than self-rated. We do not employ a cutoff HAMD score for depression group members, as we consider a psychiatric diagnosis sufficient for inclusion. Nonetheless, we administer the evaluation to both groups for the sake of maintaining a uniform protocol. Subjects will then take a BDI questionnaire and undergo a Baseline Scan while playing the Ultimatum Game (described in detail below.)

During the Longitudinal Phase, every six months for three years, participants complete a HAMD interview, a Beck Depression Inventory-II (BDI-II) survey, and an Ultimatum Game task, the latter in an fMRI scanner. If control group members at any stage score above 7 on the HAMD evaluation, they are excluded from our analysis. The first session for depression group members must take place less than six weeks prior, but not following the start of, start of SSRI treatment. We choose a period of three years because, though some regimens may last indefinitely, SSRIs are typically administered for less than two years, and most take less than six months to wean off of a successful treatment regimen (Jung et al., 2016). Additionally, by three years post-diagnosis, treatment-resistant patients will have been identified if they are present in our sample. We want to know whether fairness/inequity sensitivity regresses or maintains after patients wean off, which is why we make the sample period long enough to capture post-treatment changes. A longer timeframe may complicate the subject groupings; some controls may become depressed or develop other psychiatric abnormalities over the course of the experiment. Finally, longitudinal studies typically exhibit high rates of attrition due to the challenges of re-testing the same subjects, as myriad uncontrollable factors may reduce subjects’ ability or desire to participate (e.g. cross-country moves, medical complications). As such, we believe three years achieves a satisfactory balance between assessing persistence of treatment-related effects, and ensuring the stability of psychiatric groupings while minimizing group attrition.

The Ultimatum Game task is as follows. There are two roles: proposers (pre-programmed) and responders (played by participants). So that the game is experienced as socially interactive, we use deception to tell responders that proposals are from real people, either playing simultaneously via a connected computer network or proposed previously. The proposer has an initial endowment, of which he offers a share to the responder. Should the responder accept, the endowment is divided accordingly. If the responder rejects the offer, neither player receives any of the endowment.

Participants are given fair and unfair offers and told that they will be paid a percentage of their earnings at the end of the experiment. Fair offers range from 38% to 50% of the initial endowment, and unfair offers from 8% to 33%. Twenty-eight trials of each kind are randomly interspersed, and the offers are matched for amount; for every fair offer of a given amount, there is an unfair offer of the same amount but smaller relative to the total endowment. This is to ensure that the effect of fairness rather than material payoff is being assessed.

Regarding neural activity, we are interested in the hemodynamic response in the regions of interest (ROIs) upon presentation of the offers—an event-related design. We focus on three key regions: the dorsal caudate, nucleus accumbens, and medial occipital lobe.

From a behavioral standpoint, we examine whether the acceptance and rejection rates differ between the two groups for both kinds of offers. In addition, after the scan, we present participants with eight fair/unfair offers (four of each type, matched for material payoff) and ask them to rate feelings on 9-point Likert scales along these four dimensions: happy, angry, sad, and betrayed. The average of four ratings for each individual is represented by $L_{g,j,k,t}$ where $g$ = {MDD, Controls}; $j$ = {happy, angry, sad, betrayed}; $k$ = {fair, unfair}; and $t$ is the time period.

Upon collection of this data, we first intend to replicate the analyses of Gradin et al. (2015). Primarily, this involves comparing the correlations between regional BOLD signal and fairness/inequity for the two groups in the first session and confirming that the mOCC is less sensitive to inequity and DC and NAcc are less sensitive to fairness among depressed people prior to treatment than healthy controls. Note that these correlations for each individual are represented by $r_{g,h,k,t}$ where $g$ = {MDD, Controls}; $h$ = {DC, NAcc, mOCC}; $k$ = {fairness, unfairness}; and $t$ is the time period. For evaluating effects of fairness, r is defined as the correlation between BOLD signal and the percent of the endowment for fair offers, ranging from 38% to 50%. For offer unfairness, r is the correlation between BOLD signal and offer inequality for unfair offers, ranging from 33% (minimally unequal) to 8% (maximally unequal). To ensure that the correlation between mOCC BOLD signal and unfairness will be positive like in Gradin et al. (2015), we encode unfair offer percentages as the proposer’s share (67% to 92%). The overall pattern of the expected findings are similar to those seen in Figures 1 and 2 above.

Like in Gradin et al. (2015), we do not expect to find differences in sadness, anger, or betrayal emotionality scores, either during the baseline scan or during subsequent scans, between control subjects and MDD patients. However, we predict that there will be a statistically significant difference between initial happiness ratings for fair offers and that the two groups’ ratings will converge as SSRI treatment progresses. We expect that the degree of convergence in each period will be correlated with the extent of symptom reduction (ΔBDI) for each MDD subject but that these two factors will exhibit minimal correlation for control subjects. Figure 5 shows an example of the expected findings for these experiments.

Using data from subsequent sessions, we wish to examine whether MDD group patterns—both neural and behavioral—converge with the control group patterns as depression symptoms abate. For each individual in the depression group, we will compute inter-period changes relative to controls in ROI responsiveness to fairness/inequity.

We also compute changes in self-reported emotionality scores relative to controls for fair/unfair offers.

We regress these values on changes in BDI-II score across periods. For ROI regression models, a negative coefficient implies that a decrease in BDI-II predicts an increase in regional responsiveness.

Similar patterns as those exhibited in Gradin et al. (2015) are expected with regards to ROI sensitivity to fairness and inequity. In the first set of trials, we expect that, relative to controls, the MDD group will exhibit less sensitivity to increased fairness in the NAcc and DC ROIs. Moreover, the depressed group is initially expected to show less mOCC sensitivity to inequity than controls. Insofar as treatment is effective and BDI-II scores show reduction, we expect to observe neural convergence between MDD patients and controls. Example results of the expected relationships for NAcc and mOCC responses are shown in Figure

1. We expect the pattern related to the DC to look very similar to that for the NAcc, with the same regressors.

The lack of response to fair and unfair responses could also be seen as a blunted response to a social reward and thus could represent social anhedonia, a common symptom of depression in which appreciation for social interactions is diminished (Gradin et al., 2015). We therefore hypothesize that anhedonia-related symptoms of depression may be driving any observed correlations between BDI change and neural recovery in our ROIs. To address this possibility, we regress $\Delta r$ against the BDI anhedonia subscore ($BDI_a$), calculated by summing the scores on questions related to low motivation and reward responsiveness (questions 4, 12, 15, and 21) to determine to what extent anhedonia symptoms drive any neural findings (Pizzagalli et al., 2005). We therefore expect that the correlation between both the NAcc response to fairness and the mOCC response to inequality will be a better fit for the data than than that for the overall BDI score. Shown in Figure 7 is expected data for these regressions.

We also regress inter-temporal changes in ROI responsiveness (Δr) on the two BDI-II subscale scores (noncognitive and cognitive dimensions) to identify which subscale is most strongly linked to UG neural/behavioral differences between the MDD group and controls (Kumar et al., 2002).

Limitations

While we believe the proposed experimental design is robust, several important confounds may limit the interpretation of our data. It is possible that, as an effect of administering the same procedure to participants every six months for three years, task familiarity will result in adaptation. That is, longitudinal changes in the hemodynamic response to inequity or unfairness observed in our ROIs may simply be a result of increasing familiarity with the task. We hope that the six-month intervals between scans will reduce this risk and will allow us to minimize such familiarity effects. Additionally, rather than assessing increased neural/behavioral sensitivities per se, we examine these sensitivities relative to controls. If convergence is observed despite longitudinal changes in both groups due to task familiarity, then one possible interpretation is that SSRI treatment induced normalization in fairness/inequity sensitivity. However, it may be that task familiarity affects control subjects and MDD patients differently. With a large enough sample size, we may be able to assess this possibility; if MDD patients showing minimal SSRI response exhibit similar changes to those of controls over time, but patients exhibiting robust SSRI response do not, then convergence may be more readily attributed to symptom reduction. If all groups exhibit a similar pattern, however, we must qualify our conclusions accordingly.

Another limitation is the heterogeneity we expect to see both in MDD etiology/symptomatology and in SSRI treatment response. Many symptoms can occur separately or together: anhedonia, weight changes, disordered sleep, lack of energy, attention deficit, etc. As noted previously, depression can have somatic, affective, and cognitive dimensions. Even within the class, SSRIs comprise a group of medications that are chemically diverse and, while all block reuptake mechanisms, are differently metabolized, exhibit differential off-target neurochemical effects, and, as a result, have different side effect profiles (Coleman and Goeaux, 2018). As such, longitudinal changes may differ amongst subjects according to the specific medication and the precise nature of their MDD symptomatology. While we hope to leverage the heterogeneity that results using regression analyses, should there be differences in neural signatures or behaviors that correlate with individual symptoms or specific SSRIs they may affect the statistical robustness of the study. Further studies may also inform this question by examining these neural signatures in patient groups displaying risk factors for SSRI non-response, such as the serotonin transporter polymorphism SLC6A4 (Luddington et al., 2009; El-Mallakh et al., 2019).

Relatedly, in the present studies we have limited our sample to patients undergoing treatment with SSRIs. There are, however, a variety of other treatments in use for MDD, including norepinephrine reuptake inhibition, monoamine oxidase inhibition, NMDA receptor antagonism, brain stimulation, and cognitive therapy (Tran et al., 2019). We chose to limit our findings to SSRIs because they are current first-line antidepressants; because there is known heterogeneity in response that is itself of interest in these studies; and because we are interested in controlling, as much as is reasonable, for the specific effects of pharmacological treatment on our results. Nonetheless, any findings we uncover may therefore not be generalizable to all patient populations. Further studies undertaken to examine the effects under other treatment conditions would be necessary to determine the generalizability of these effects. Nonetheless, these results will address whether medication can normalize brain states that are associated with depression symptomatology.

It is additionally possible that we will fail to replicate the NAcc, DC, and mOCC neural activity decreases in response to fairness/unfairness in our MDD subjects. We do not anticipate that we will fail to detect all three of these neural signatures described in Gradin et al., as each have been reported to show disruption in MDD studies across multiple experimental conditions (Kerestes et al., 2014; Li et al., 2013; Zhou et al., 2020; Bewernick et al., 2012; Bewernick et al., 2010; Robson et al., 2020). However, should we fail to detect alterations in any of these pre-identified areas, we will perform a voxelwise whole brain analysis from the data collected during the baseline scan to identify novel ROIs that show event-related changes during UG performance and, instead, track these throughout the longitudinal phase of the study.

Another limitation concerns the social gameplay inherent to the UG. It is possible that, because our subjects will be playing against a computer without a physical partner in the room, that we will fail to detect differences in brain regions due to fairness or unfairness of the offer based on the simple fact that the lack of a human partner will dilute the social value of the game for the scanned participant. Should this occur we expect that we will be able to detect this effect in our controls as well, in the form of a blunted NAcc/DC response to fair offers and a blunted mOCC response to unfair offers, similar to the effects predicted in MDD patients. If this arises with initial subjects, we will include an experimenter acting as the proposer in the room, who will offer through a headset worn into the fMRI scanner. If this effect is truly due to a depersonalization and thus reduction in the reward value of the outcome due to the computerized gameplay, we would expect to see a recovery of the blunting in control subjects but not in MDD subjects.

A final important consideration concerns the relationship between depression symptom severity and neural patterns. We hypothesize that neural patterns will change as depression severity improves following SSRI treatments. However, as these experiments address outstanding questions in the field, it is very possible that neural responses to fairness and inequity in our ROIs will not track with BDI-II score changes. We would interpret these results to indicate that responses to fairness may be not a dynamic result of psychiatric condition but instead a static factor indicating the possible lifetime presence of MDD.

Most of the individual components of this experiment are well-established; performing tasks in the context of fMRI, the task itself, the proposed data analyses, and the means of subject recruitment have all been performed by numerous studies in the past. The experimental procedures therefore do not represent high-risk aspects of the proposal. However, as with any longitudinal study, there is a major risk of subject attrition that could impact the statistical robustness of the final data. We hope that limiting the scanning interval to three years will allow us to strike a balance between the ability to capture longitudinal effects and remission following SSRI treatment in our MDD sample, while also limiting subject attrition due to uncontrollable factors. To encourage completion, all participants will be compensated for each session at a flat rate plus a percentage of total earnings from the UG. Ideally, this second component of subjects’ earnings will also encourage them to play the UG as they would if all offers were paid out.

In sum, the studies proposed here would substantially advance our understanding of the stability of neural signatures of neuroeconomic gameplay in MDD. The longitudinal study design will robustly address the outstanding question of whether signatures of depression represent a static state or a dynamic trait that resolves with successful treatment. Data collected here will also further our understanding—currently lacking—of the stability of neural signatures of economic gameplay. The design of this study could subsequently be extended to examine other disorders that disrupt social reward-processing, including schizophrenia and social anxiety disorders (Goldin et al., 2009; Bjorkquist and Herbener, 2013). We therefore believe the proposed studies will be impactful and trans-diagnostically relevant.

References

Bewernick, B., Kayser, S., Sturm, V. et al. Long-Term Effects of Nucleus Accumbens Deep Brain Stimulation in Treatment-Resistant Depression: Evidence for Sustained Efficacy. Neuropsychopharmacol 37, 1975–1985 (2012). https://doi.org/10.1038/npp.2012.44

Bewernick BH, Hurlemann R, Matusch A, Kayser S, Grubert C, Hadrysiewicz B, Axmacher N, Lemke M, Cooper-Mahkorn D, Cohen MX, Brockmann H, Lenartz D, Sturm V, Schlaepfer TE. Nucleus accumbens deep brain stimulation decreases ratings of depression and anxiety in treatment-resistant depression. Biol Psychiatry. 2010 Jan 15;67(2):110-6. doi: 10.1016/j.biopsych.2009.09.013. PMID: 19914605.

Bjorkquist OA, Herbener ES. Social perception in schizophrenia: evidence of temporo-occipital and prefrontal dysfunction. Psychiatry Res. 2013;212(3):175-182. <doi:10.1016/j.pscychresns.2012.12.002>

Carrozzino D, Patierno C, Fava G, A, Guidi J: The Hamilton Rating Scales for Depression: A Critical Review of Clinimetric Properties of Different Versions. Psychother Psychosom 2020;89:133-150. doi: 10.1159/000506879

Coleman JA, Gouaux E. Structural basis for recognition of diverse antidepressants by the human serotonin transporter. Nat Struct Mol Biol. 2018;25(2):170-175. <doi:10.1038/s41594-018-0026-8>

El-Mallakh RS, Ali Z. Therapeutic implications of the serotonin transporter gene in depression. Biomarkers in Neuropsychiatry. 2019; 1:100004. Doi: 10.1016/j.bionps.2019.100004

Gaynes BN, Warden D, Trivedi MH, Wisniewski SR, Fava M, Rush AJ. What did STAR*D teach us? Results from a large-scale, practical, clinical trial for patients with depression. Psychiatr Serv. 2009;60(11):1439-1445. <doi:10.1176/ps.2009.60.11.1439>

Gradin VB, Pérez A, MacFarlane JA, et al. Abnormal brain responses to social fairness in depression: an fMRI study using the Ultimatum Game. Psychol Med. 2015;45(6):1241-1251. <doi:10.1017/S0033291714002347>

Goldin PR, Manber T, Hakimi S, Canli T, Gross JJ. Neural bases of social anxiety disorder: emotional reactivity and cognitive regulation during social and physical threat. Arch Gen Psychiatry. 2009;66(2):170-180. <doi:10.1001/archgenpsychiatry.2008.525>

Jung, W. Y., Jang, S. H., Kim, S. G., Jae, Y. M., Kong, B. G., Kim, H. C., Choe, B. M., Kim, J. G., & Kim, C. R. (2016). Times to Discontinue Antidepressants Over 6 Months in Patients with Major Depressive Disorder. Psychiatry investigation, 13(4), 440–446. https://doi.org/10.4306/pi.2016.13.4.440

Kerestes R, Harrison BJ, Dandash O, et al. Specific functional connectivity alterations of the dorsal striatum in young people with depression. Neuroimage Clin. 2014;7:266-272. Published 2014 Dec 27. doi: 10.1016/j.nicl.2014.12.017

Kulikov AV, Gainetdinov RR, Ponimaskin E, Kalueff AV, Naumenko VS, Popova NK. Interplay between the key proteins of serotonin system in SSRI antidepressants efficacy. Expert Opin Ther Targets. 2018;22(4):319-330. <doi:10.1080/14728222.2018.1452912>

Kumar G, Steer RA, Teitelman KB, Villacis L. Effectiveness of Beck Depression Inventory-II subscales in screening for major depressive disorders in adolescent psychiatric inpatients. Assessment. 2002 Jun;9(2):164-70. doi: 10.1177/10791102009002007. PMID: 12066831.

Li J, Xu C, Cao X, Gao Q, Wang Y, Wang Y, Peng J, Zhang K. Abnormal activation of the occipital lobes during emotion picture processing in major depressive disorder patients. Neural Regen Res. 2013 Jun 25;8(18):1693-701. doi: 10.3969/j.issn.1673-5374.2013.18.007. PMID: 25206466; PMCID: PMC4145913.

Luddington NS, Mandadapu A, Husk M, El-Mallakh RS. Clinical implications of genetic variation in the serotonin transporter promoter region: a review. Prim Care Companion J Clin Psychiatry. 2009;11(3):93-102. <doi:10.4088/pcc.08r00656>

Mukherjee, D., Lee, S., Kazinka, R. et al. Multiple Facets of Value-Based Decision Making in Major Depressive Disorder. Sci Rep 10, 3415 (2020). https://doi.org/10.1038/s41598-020-60230-z

NIMH.gov. Depression Fact Sheet. (2020).

Pizzagalli DA, Jahn AL, O’Shea JP. Toward an objective characterization of an anhedonic phenotype: a signal-detection approach. Biol Psychiatry. 2005;57(4):319-327. <doi:10.1016/j.biopsych.2004.11.026>

Robson, S.E., Repetto, L., Gountouna, VE. et al. A review of neuroeconomic gameplay in psychiatric disorders. Mol Psychiatry 25, 67–81 (2020). https://doi.org/10.1038/s41380-019-0405-5

Taghva AS, Malone DA, Rezai AR. Deep brain stimulation for treatment-resistant depression. World Neurosurg. 2013 Sep-Oct;80(3-4):S27.e17-24. doi: 10.1016/j.wneu.2012.10.068. Epub 2012 Oct 27. PMID: 23111230.

Tolentino JC, Schmidt SL. DSM-5 Criteria and Depression Severity: Implications for Clinical Practice. Front Psychiatry. 2018;9:450. Published 2018 Oct 2. <doi:10.3389/fpsyt.2018.00450>

Tran BX, Ha GH, Vu GT, et al. Indices of Change, Expectations, and Popularity of Biological Treatments for Major Depressive Disorder between 1988 and 2017: A Scientometric Analysis. Int J Environ Res Public Health. 2019;16(13):2255. Published 2019 Jun 26. <doi:10.3390/ijerph16132255>

Trivedi MH, Rush AJ, Wisniewski SR, et al. Evaluation of outcomes with citalopram for depression using measurement-based care in STAR*D: implications for clinical practice. Am J Psychiatry. 2006;163(1):28-40. doi: 10.1176/appi.ajp.163.1.28

Wessa M, Lois G. Brain Functional Effects of Psychopharmacological Treatment in Major Depression: a Focus on Neural Circuitry of Affective Processing. Curr Neuropharmacol. 2015;13(4):466-79. doi: 10.2174/1570159x13666150416224801. PMID: 26412066; PMCID: PMC4790403.

Zhao Y, Niu R, Lei D, Shah C, Xiao Y, Zhang W, Chen Z, Lui S, Gong Q. Aberrant Gray Matter Networks in Non-comorbid Medication-Naive Patients With Major Depressive Disorder and Those With Social Anxiety Disorder. Front Hum Neurosci. 2020 Jun 10;14:172. doi: 10.3389/fnhum.2020.00172. PMID: 32587507; PMCID: PMC7298146.

← Back to all articles