Brain activation during the anticipation phase

In the current meta-analysis, the most important finding is that studies consistently reported that people with MDD exhibit hyper-responsivity to monetary reward in the mPFC, MFG, median and anterior cingulate gyri, and postcentral areas during anticipation. The medial walls of the frontal lobes, comprising the cingulate cortex and mPFC, are thought to be engaged in reward anticipation, and anhedonia, core features of MDD (Heilbronner & Hayden, Reference Heilbronner and Hayden2016). Specifically, as the cognitive part of the cingulate cortex, the ACC encodes reward evaluation and comparison and subserves monitoring throughout the decision-making process, such as the representation of the value of stimuli or actions and the monitoring of somatic states (Holroyd & Yeung, Reference Holroyd and Yeung2012). The frontal medial regions of the prefrontal cortex are key components of neural circuits involved in detecting the motivational significance of external stimuli. Altered functioning of the ACC in particular, which plays a central role in detecting the salience of external stimuli and in reward monitoring (Seeley et al., Reference Seeley, Menon, Schatzberg, Keller, Glover, Kenna and Greicius2007; Whitton et al., Reference Whitton, Kakani, Foti, Van't Veer, Haile, Crowley and Pizzagalli2016), has been observed in patients with MDD during reward processing tasks. Increased ACC activation has been observed in healthy individuals (Fan et al., Reference Fan, Kolster, Ghajar, Suh, Knight, Sarkar and McCandliss2007) and depressed individuals (Dichter et al., Reference Dichter, Kozink, McClernon and Smoski2012; Knutson et al., Reference Knutson, Bhanji, Cooney, Atlas and Gotlib2008) under conditions involving uncertainty and conflict, when errors are likely. Stronger activations are also observed in the MCC given that this region is a hub linking incoming affective information (Shackman et al., Reference Shackman, Salomons, Slagter, Fox, Winter and Davidson2011) and engages in reward processing (Vogt, Reference Vogt2016). The MCC plays an important role in decision-making, especially in relation to reward (Bush, Luu, & Posner, Reference Bush, Luu and Posner2000; van Heukelum et al., Reference van Heukelum, Mars, Guthrie, Buitelaar, Beckmann, Tiesinga and Havenith2020). The anticipation phase is characterised by heightened arousal and increased attention. These results support the notion that increased activation of the mPFC and cingulate cortex is due to reward expectation, salience processing, and cognitive response to the task. It should be noted that hyperactivation during reward anticipation is located in the dorsal ACC, which is also termed the anterior MCC (Holroyd & Yeung, Reference Holroyd and Yeung2012), rather than the pregenual aspect of the ACC in individuals with MDD. Some brain imaging evidence pointed to decreased pregenual ACC activation in depressed individuals (Ito et al., Reference Ito, Yokokawa, Yahata, Isato, Suhara and Yamada2017; Philippi et al., Reference Philippi, Cornejo, Frost, Walsh, Hoks, Birn and Abercrombie2018), but these findings might reflect activation in response to emotional stimulus rather than anticipatory reward processing. Such a finding of greater responses in the mPFC and anterior MCC in the MDD group may suggest greater monitoring of responses, increased motivational salience processing, and enhanced reward evaluation of the forthcoming reward.

During monetary anticipation, blunted brain activation was most prominent in the mesolimbic circuit, including the striatum, insula, and amygdala, when compared to healthy controls. In line with previous literature, a blunted striatal response to reward is indeed a hallmark of pathological reward anticipation in MDD. This region of the VS receives afferent projections from several cortical and limbic regions, including the prefrontal cortex, amygdala, and hippocampus, and, in turn, sends projections to the ventral pallidum, ventral tegmental area, and substantia nigra (Haber & Knutson, Reference Haber and Knutson2010). Many of the regions linked to the striatum, particularly prefrontal regions, have been associated with the computation and representation of reward value as well as the regulation of affect and reward-related behaviour in animals and healthy individuals (Grabenhorst & Rolls, Reference Grabenhorst and Rolls2011; Hiser & Koenigs, Reference Hiser and Koenigs2018). The assessment of reward value, or the motivation to obtain a reward, triggers emotions and leads to an understanding of motivation or ‘wanting’ (Grabenhorst & Rolls, Reference Grabenhorst and Rolls2011). The mesolimbic circuit is one of the most important anatomical substrates for natural rewards (e.g. food, sex) as well as for secondary rewards (e.g. money, social interactions). An accumulating body of evidence indicates that hypo-responsivity in the striatal regions plays an important role in the onset and course of MDD. For example, reduced striatal activation is present among currently depressed individuals (Rappaport, Kandala, Luby, & Barch, Reference Rappaport, Kandala, Luby and Barch2020) and in MDD during periods of remission (Geugies et al., Reference Geugies, Mocking, Figueroa, Groot, Marsman, Servaas and Ruhe2019). This reduction in reward responsiveness has been found to predict later depressive episode MDD and the emergence of depressive symptoms despite antidepressant treatment (Morgan, Olino, McMakin, Ryan, & Forbes, Reference Morgan, Olino, McMakin, Ryan and Forbes2013; Stringaris et al., Reference Stringaris, Vidal-Ribas Belil, Artiges, Lemaitre, Gollier-Briant, Wolke and Paillère-Martinot2015). The strength of these blunt responses is correlated with the severity of depression (Geugies et al., Reference Geugies, Mocking, Figueroa, Groot, Marsman, Servaas and Ruhe2019; Rappaport et al., Reference Rappaport, Kandala, Luby and Barch2020). In addition to VS, a reduced response in the putamen was found. In previous studies, the putamen has been demonstrated to be associated with reward sensitivity (Meyer et al., Reference Meyer, Kruger, Wilson, Christensen, Goulding, Schaffer and Kennedy2001; Russo & Nestler, Reference Russo and Nestler2013). Deficits in reward sensitivity are particularly evident in the melancholic subtype of depression (Foti, Carlson, Sauder, & Proudfit, Reference Foti, Carlson, Sauder and Proudfit2014). Some studies with depressed patients have found an increase in putamen volume after treatment, suggesting that patients can be treated by modulating their reward/motivation circuits (Wang et al., Reference Wang, Wang, Liu, Chen, Liu, Nie and Kong2017). Common findings of hypo-responsivity in the mesolimbic circuit highlight the role of dopamine signalling in the pursuit of reward, contributing to motivational impairment in depression.

Among regions outside the striatum, a notable finding was the reduced activity in the OFC, amygdala and insula. The OFC is thought to play a major role in many complex functions including recognition of reward stimuli and stimulus-reward association, which is considered a therapeutic target for depression (Lacerda et al., Reference Lacerda, Keshavan, Hardan, Yorbik, Brambilla, Sassi and Soares2004). Reduced activity in the OFC during reward anticipation may reflect diminished tagging of normally rewarding stimuli with positive affective value (Fettes, Schulze, & Downar, Reference Fettes, Schulze and Downar2017). The amygdala shares interconnections with the orbitofrontal cortex and supplies information to generate and use expectancies of reinforcers in the guidance of goal-directed behaviour (Holland & Gallagher, Reference Holland and Gallagher2004). Recent evidence has indicated the important roles of the amygdala in the representation of incentive value and the evaluation of salient stimuli (Holland & Gallagher, Reference Holland and Gallagher2004). The insula, together with the mPFC and posterior cingulate cortex, is an intricate part of the salience/reward network. The insula is involved in risk prediction and incentive motivational processes, possibly due to its role in detecting salient information and processing interoceptive information (Naqvi & Bechara, Reference Naqvi and Bechara2009). Our findings of blunted reward activity in limbic regions confirm previous fMRI studies in MDD. The hypo-reactivity of a broader set of regions within the mesolimbic circuit indicates that the amygdala and insula may be key brain regions implicated in the pathophysiology of depression (Price & Drevets, Reference Price and Drevets2010).

Brain activation during the outcome phase

With regard to receiving reward outcomes, MDD showed decreased activity within the cortices (medial/dorsolateral prefrontal cortex, precentral gyrus, and cingulate cortex) and subcortical structures (ventral and dorsal striatum, caudate nucleus, thalamus) and increased brain activity in the ITG; notably, these mesocortical structures were previously implicated as deficient in patients with MDD. Previous studies of reduced activity in the VS associated with an abnormal response to feedback were frequently reported in currently depressed (Forbes et al., Reference Forbes, Hariri, Martin, Silk, Moyles, Fisher and Dahl2009; Pizzagalli et al., Reference Pizzagalli, Holmes, Dillon, Goetz, Birk, Bogdan and Fava2009; Spati et al., Reference Spati, Chumbley, Doerig, Brakowski, Grosse Holtforth, Seifritz and Spinelli2015) and high-risk individuals (Sharp et al., Reference Sharp, Kim, Herman, Pane, Reuter and Strathearn2014). Decreased responses in the mesocortical pathway could indicate depression-related difficulties with several aspects of outcome-related reward processing. MCC hypoactivation presumably contributes to choice processes when receiving the outcome, as the MCC is an integral part of feedback-mediated decision-making and modification of rewarded behaviours to accommodate current and predicted contexts (Vogt, Reference Vogt2016). The thalamus is an interface for brain reward circuits and with input signals arising from structures such as the prefrontal cortex and hypothalamus that are broadcast to downstream limbic targets (Otis et al., Reference Otis, Zhu, Namboodiri, Cook, Kosyk, Matan and Stuber2019). It mediates appetitive associative learning, motivational drive, planning and cognition for the development of goal-directed behaviour (Krebs, Boehler, Roberts, Song, & Woldorff, Reference Krebs, Boehler, Roberts, Song and Woldorff2012; Zhu et al., Reference Zhu, Nachtrab, Keyes, Allen, Luo and Chen2018). Disruption of the response in the thalamus, could suggest difficulties with the execution of action in response to rewarding cues. Notably, the striatal response in depressed patients seems to be more involved in the dorsal part of the striatum during the outcome than in the ventral part of the striatum during anticipation. In reinforcement learning, the VS signals reward prediction error and incentive motivational salience (Berridge, Reference Berridge2012; Bromberg-Martin, Matsumoto, & Hikosaka, Reference Bromberg-Martin, Matsumoto and Hikosaka2010), while the DS supports habit, formation, actions, and stimulus-driven behaviour (Hilario & Costa, Reference Hilario and Costa2008). One possible interpretation is that the response of the DS is blunted rewards during the processing of reward delivery, which seems to reflect a ‘habit’ of reduced reward response, leading to more automatic hedonic responses. Compared to reward anticipation, blunted reactivity to reward after exposure to visual stimuli (Kaposvari, Kumar, & Vogels, Reference Kaposvari, Kumar and Vogels2018; Liu et al., Reference Liu, Admon, Mellem, Belleau, Kaiser, Clegg and Pizzagalli2020). The increased ITG activation evoked by reward cues is consistent with these studies and implies improved sensory processing for feedback information.

Clinical implications

Core depressive symptom dimensions, including persistent low mood and anhedonia, reflect predominant features of dysfunctional reward processing. From the translational potential of the insights, mapping the different domains of reward processing onto depressive symptoms could have important implications for informing illness diagnostics. Impaired reward learning has emerged as a key characteristic of MDD and may be a vulnerability marker in individuals with MDD (Coccurello, Reference Coccurello2019; Hasler et al., Reference Hasler, Drevets, Manji and Charney2004). MDD is characterised by decreased responsiveness of the frontostriatal brain regions to rewarding stimuli, including decreased anticipation of forthcoming rewards; reduced hedonic experience derived from reward presentation may contribute to the core symptom of anhedonia in MDD (Der-Avakian & Markou, Reference Der-Avakian and Markou2012; Zhang et al., Reference Zhang, Chang, Guo, Zhang and Wang2013, Reference Zhang, Lin, Shi, Ongur, Auerbach, Wang and Wang2016). These abnormalities in reward circuitry may be associated with behaviours linked to constructs in the positive valence systems domain, such as reward expectancy and anhedonia (Der-Avakian, Barnes, Markou, & Pizzagalli, Reference Der-Avakian, Barnes, Markou and Pizzagalli2016). Our meta-analysis found that MDD may be characterised by hyper-responsivity in the cortical prefrontal regions during the anticipation phase, and hypo-responsivity of the mesocortico-limbic circuit across the two phases of the reward response. Consistent with the NIMH Research Domain Criteria initiative (Insel et al., Reference Insel, Cuthbert, Garvey, Heinssen, Pine, Quinn and Wang2010), identifying brain responses to reward stimuli may be particularly useful for identifying brain-behaviour relationships and clinically and pathophysiologically meaningful MDD endophenotypes. Therefore, identification of the different neurobiological pathways underlying the different facets of reward processing may serve as potential neural markers for MDD diagnosis based on the framework of the appetitive reward system.

Of particular relevance here, targeting reward dysfunction also allows us to identify treatment outcomes and relapse propensity. A growing body of literature addresses attenuations of striatal activation, and frontostriatal connectivity can serve as a potential predictor of treatment response in depression (Phillips et al., Reference Phillips, Chase, Sheline, Etkin, Almeida, Deckersbach and Trivedi2015; Walsh et al., Reference Walsh, Carl, Eisenlohr-Moul, Minkel, Crowther, Moore and Dichter2017). Aberrant brain connectivity in the striatum is reported to be associated with the course of depressive episodes, suggesting that the reorganisation of striatal connectivity may interact with the course of episodes in depression, thereby contributing to depressive relapse risk (Meng et al., Reference Meng, Brandl, Tahmasian, Shao, Manoliu, Scherr and Sorg2014). Recently, a novel psychological treatment (called Positive Affect Treatment) for anxiety and depression was designed to specifically target deficits in reward sensitivity and appetitive responding (Craske, Meuret, Ritz, Treanor, & Dour, Reference Craske, Meuret, Ritz, Treanor and Dour2016). After receiving this new treatment, patients with unipolar depression showed a significant reduction in depression symptoms and a significant improvement in positive affect, although the effectiveness of treating anhedonia awaits careful testing in further studies (Craske et al., Reference Craske, Meuret, Ritz, Treanor and Dour2016). This may help to explain the success of behavioural activation treatments for MDD, which specifically target motivational symptoms (Dimidjian et al., Reference Dimidjian, Hollon, Dobson, Schmaling, Kohlenberg, Addis and Jacobson2006). Therefore, investigation of the precise reward processes that are affected in MDD will facilitate the diagnosis and treatment of these disorders that include reward and motivation deficits as key symptoms.

Limitations and future directions

Several limitations of the current study should be addressed. First, our voxel-wise meta-analysis was based on the coordinates from published studies instead of raw statistical brain maps, which limited the accuracy of our results. Second, we could not investigate the effect of some clinical characteristics, such as subtypes of depression, severity of anhedonia, comorbidity with anxiety, and mediation status in the depressed individuals on the MID-related brain response, due to limited data. We cannot directly investigate the neural correlates of the construct of anhedonia given the insufficient data for measuring self-reports of anhedonia (see online supplement for the in-depth discussion). We are also underpowered to examine the differences in monetary processing between depression with anxiety and without anxiety during the anticipation and outcome phases. Similarly, only a few studies reported the percentage of participants who had received antidepressant treatments in the past, preventing further analysis. Additionally, the complicated effects of treatment, such as drug types, clinical response, and side effect profiles, could not be examined in our present study due to insufficient data. These results should be interpreted with caution, and very careful clinical observation is required to examine treatment-related effects on reward response. Third, all fMRI studies have used monetary rewards, but the processing of primary and secondary rewards may differ in important respects. The results should be interpreted with caution when applying the findings to primary reward enforcers. With regard to methodological heterogeneity, diverse imaging acquisition techniques (e.g. different MRI field strengths, MRI scanners, and imaging parameters) might bias the results. Furthermore, the included studies applied different templates, smoothing kernel size, slice thickness and statistical threshold. These variations in data analysis parameters make it difficult to directly compare findings among studies. Thus, imaging acquisition and fMRI analysis parameters varied throughout the literature, potentially limiting our ability to detect robust group differences.

In the present study, we examined different stages of reward processing in cMDD patients using a voxel-wise meta-analysis. Because previous research suggests that neural activation during reward processing might vary with depression symptom profiles (Rappaport et al., Reference Rappaport, Kandala, Luby and Barch2020), our findings of functional alterations in MDD may provide as state markers of MDD that reflect the pathophysiological processes of the illness. Future studies investigating cMDD individuals and rMDD individuals may be able to identify the state or trait characteristic of the illness. Future longitudinal studies are also needed to further characterise the persistence of altered neural responses to monetary stimulus when patients enter or recover from depressive episodes. Regarding the task paradigm, the MID task is not designed to capture temporal learning or any reward prediction error signal. However, it would be interesting to examine aspects of reward coding, including learning rate, reward sensitivity, and prediction error, in the future using theories and models of reinforcement learning, such as goal-directed v. habitual decision-making.