Major depressive disorder (MDD) and anxiety disorders are among the most prevalent and debilitating psychiatric disorders in the U.S. (Kessler et al., 2005; World Health Organization, 2017). MDD and anxiety disorders are highly comorbid (Fava & Kendler, 2000; Lamers et al., 2011). However, most research measures clinical constructs using the Diagnostic Statistical Manual of Mental Disorders (DSM), which uses a categorical model and assumes each disorder is distinct and separate. This classification method may limit insights into the nature of mental illness. As a result, initiatives such as the Research Domain Criteria (RDoC) framework and Hierarchical Taxonomy Of Psychopathology (HiTOP), have focused their efforts on data driven constructs and dimensional measures of psychopathology. The present study extends previous structural neuroimaging literature by examining gray mater volume of emotion generation and regulation regions using dimensional symptoms of depression and anxiety from the Tri-level Model (Prenoveau et al., 2010). These regions have been previously implicated in the depression and anxiety literature.

The amygdala and nucleus accumbens (NAcc) are two brain regions involved in emotion generation, which is the facilitation of emotional processes and accompanying physiological processes. The amygdala is implicated in threat processing and mediates defensive emotional, behavioral, and physiological states (Hur et al., 2019). Although depression and anxiety are associated with elevated amygdala activation, the structural neuroimaging literature is more inconsistent (Hamilton et al., 2008). Many studies have shown an association between depression and smaller amygdala gray matter volume (Bora et al., 2012; Sacher et al., 2012), though others have shown enlarged (Anand & Shekhar, 2006; Frodl et al., 2002; Lange & Irle, 2004) or unchanged (Campbell et al., 2004; MacMaster et al., 2008) volumetric alterations. Anxiety has been associated with larger amygdala gray matter volumes (De Bellis et al., 2000; Hur et al., 2019; Schienle et al., 2011; Suor et al., 2020), though not always (e.g., Blackmon et al., 2011; Hayano et al., 2009; Warnell et al., 2018). How anxiety is defined and measured may contribute to these inconsistencies, in that larger amygdala volume is associated with generalized anxiety disorder (De Bellis et al., 2000; Schienle et al., 2011) and social anxiety (Suor et al., 2020), contrasted with smaller amygdala volume associated with panic disorder (Hayano et al., 2009), obsessive-compulsive disorder (Anand & Shekhar, 2006), and trait anxiety (Warnell et al., 2018). Inconsistencies in the neural structure literature may also reflect variability in the distribution of depression and anxiety symptoms not captured in categorical DSM diagnoses. For example, differences in amygdala structure may be driven in part by the common dysphoric mood and negative emotionality seen in depression and anxiety (Hur et al., 2019; Shackman et al., 2016).

The nucleus accumbens (NAcc) is a sub-component of the ventral striatum that mediates motivation, reward, emotion processing, and reward-related behaviors (Harvey et al., 2007; O'Doherty, 2004). Reward functioning in the NAcc is blunted in depression (Forbes et al, 2009) and implicated in anxiety (Guyer et al., 2012). Anatomically, smaller volume of the NAcc is associated with depression symptoms (Auerbach et al., 2017; Phillips, 2003; Wacker et al., 2009) and anxiety disorders (Anand & Shekhar, 2006; Hilbert et al., 2015), though a large portion of research did not support an association between depression or anxiety and alterations in NAcc structure (e.g., Besteher et al., 2020; Kempton et al., 2011). Some research suggests that smaller NAcc volumes may reflect reward-related dysfunction, impacting brain function and symptom presentation across internalizing disorders.

Emotion regulation is the process of implementing conscious or non-conscious processes to modulate the trajectory of an emotion (Phillips et al., 2008). The ventrolateral prefrontal cortex (VLPFC) and dorsolateral prefrontal cortex (DLPFC) have distinct roles from the orbitofrontal cortex (OFC) in regulating behaviors and emotion processing via structural connections to subcortical emotion generation regions (Phillips et al., 2008). Specifically, the VLPFC and DLPFC are involved in voluntary, purposeful emotion regulation processes, while the OFC is involved in automatic emotion regulation processes (Phillips et al., 2008). Some suggest that the OFC may mediate connections between the VLPFC and DLPFC regions and subcortical regions (Phillips et al., 2008).

Depression and anxiety are characterized by disordered emotion regulation (Amstadter, 2008; Cisler et al., 2010; Joormann & Stanton, 2016). In particular, depression and anxiety are associated with the use of maladaptive strategies and reduced ability to use effective strategies for emotion regulation (Cisler et al., 2010; Joormann & Stanton, 2016). This can include rumination, emotion suppression, experiential avoidance, emotional non-acceptance, negative reactivity to emotions and less use of reappraisal (Amstadter, 2008; Cisler et al., 2010; Joormann & Stanton, 2016). These regulatory strategies can maintain or increase depression and anxiety symptoms (Amstadter, 2008; Joormann & Stanton, 2016).

Anatomical differences in emotion regulation regions are associated with depression and anxiety. Depression is commonly characterized by reduced gray matter volume in the prefrontal cortex (Anand & Shekhar, 2006; Phillips, 2003), including in the OFC (Anand & Shekhar, 2006; Bremner et al., 2002; Koolschijn et al., 2009; Webb et al., 2014), DLPFC (Chang et al., 2011; Li et al., 2010), and VLPFC (Lener et al., 2016; Salvadore et al., 2011). Anxiety is also associated with reduced prefrontal cortex gray matter volume (Syal et al., 2012), particularly in the OFC (Anand & Shekhar, 2006; Blackmon et al., 2011; Zhao et al., 2017). Anxiety disorders are sometimes associated with smaller DLPFC (Fonzo et al., 2016; Hilbert et al., 2015) and VLPFC (Auday & Pérez‐Edgar, 2019) gray matter volume, though often show non-significant results (Mohlman et al., 2009; Schienle et al., 2011; Zhao et al., 2017). In sum, prefrontal alterations seem to be a common feature of both depression and anxiety and may be a mechanism that underlies the transdiagnostic emotion regulation dysfunction seen in depression and anxiety.

Most research on the pathophysiology of depression and anxiety focuses on DSM classification, however there are some limitations to the DSM that may play a role in the inconsistent results presented thus far (see Nikolaidis et al., 2022). First, categorical classification uses arbitrary cut off points for assignment of a psychiatric disorder, which does not reflect the continuous nature of psychological functioning and many forms of psychopathology (Haslam et al., 2012). Second, continuous measures of psychopathology outperform discrete measures in terms of reliability and validity (Markon et al., 2011). Third, psychiatric disorders are highly comorbid, particularly depression and anxiety (Kessler et al., 2005; Krueger & Markon, 2006; Rush et al., 2005), such that separation of discrete disorders may limit scientific research and applicability of interventions. Fourth, there is heterogeneity within and across DSM diagnoses, in that the clinical presentation for a disorder can vary widely, but also overlap with symptoms of different disorders (Allsopp et al., 2019). Further, phenomenology as defined by behavioral systems does not discretely map onto biology. Examples of this include common brain abnormalities across multiple illnesses (Krueger, 1999), overlapping genetic influences (Kendler, 1992), multiple etiological pathways leading to similar clinical manifestations (Kendler, 2019), and the effectiveness of the same treatments for multiple diagnoses (e.g. SSRIs; Barlow et al., 2017; Vaswani et al., 2003).

As a solution to these limitations, there has been a large push in the field to use dimensional analyses as additional models of psychopathology. These models could lead to a shift in how psychopathology is studied, classified, and treated (Michelini et al., 2021). Importantly, preliminary research shows that dimensional models have greater sensitivity and stronger associations to neural variables, compared to diagnoses (Kircanski et al., 2017; Michelini et al., 2021; Nikolaidis et al., 2022; Reininghaus et al., 2019).

The dimensional model we employ in the present study is the Tri-level Model, an empirically-derived, hierarchical structure underlying symptoms of unipolar mood and anxiety disorders (see Figure 1 for model structure; Prenoveau et al., 2010). A broad, transdiagnostic factor, termed General Distress, is characterized by shared features of depression and anxiety, including distress, negative emotionality, and dysphoric mood. Then there are two intermediate level factors: (a) Fears is loaded on by anxiety specific symptoms, such as interoceptive-agoraphobic fears, social fears, and fears of specific stimuli, and (b) Anhedonia-Apprehension (previously termed Anxious-Misery) is loaded on by relatively specific depression symptoms of anhedonia and hopelessness. The Fears factor is significantly related to social phobia, specific phobia, and obsessive-compulsive disorder (Prenoveau et al., 2010). The Anhedonia-Apprehension factor is significantly related to major depression, social phobia, and generalized-anxiety disorder (Prenoveau et al., 2010). Though the model also defines narrower factors, these were not used in the current analyses due to our specific interest in transdiagnostic and disorder differences, rather than specific disorder subtypes. Additionally, we were concerned about having enough power to detect associations at the narrow factor level. This model has been independently replicated and published on by our group (Naragon-Gainey et al., 2016; Prenoveau et al., 2010; Williams et al., 2021; Young et al., 2021; Zinbarg et al., 2022). The Tri-level Model is empirically driven and has never been examined in the context of structural brain data. We chose to use the Tri-level Model because it gives us the unique opportunity to study transdiagnostic features as well as those that are more disorder specific, which is not possible using DSM diagnosis.

For context, the Tri-level Model is similar to the Internalizing spectrum of the HiTOP model (Kotov et al., 2017). Both models have a hierarchical structure and utilize dimensional symptoms of psychopathology, though the Internalizing spectrum of HiTOP is embedded in a larger higher order model. Specific dimensions, namely, the HiTOP Internalizing spectrum, Fear subfactor, and Distress subfactor largely overlap with General Distress, Fears, and Anhedonia-Apprehension, respectively (Kotov et al., 2017; Naragon-Gainey et al., 2016; Prenoveau et al., 2010). There are a few minor differences between the two models. First, the Tri-level Model methodologically relies on self-report indicators to this point, which is not yet a component of the HiTOP model which relies on diagnostic data to this point. Second, the HiTOP Internalizing spectrum includes additional forms of internalizing symptoms (e.g., sexual problems, eating problems) not found in the Tri-level Model. However, despite these differences, they are highly similar constructs, and we expect the Tri-level Model to largely generalize to the HiTOP Internalizing spectrum dimensions.

In the present study, we investigate structural metrics of emotion generation- and regulation-related neural regions and Tri-level Model symptom dimensions in young adults. We focused on young adulthood as it is an important time in regards to (1) neurodevelopmental change, particularly in prefrontal cortical regions, which are the last to develop (Powers & Casey, 2015; Shaw et al., 2011), and (2) the emergence of psychopathology, in that the peak of onset for depression and anxiety disorders is during adolescence and young adulthood (Paus et al., 2008).

Though other work has explored the neural correlates of transdiagnostic and comorbid depression and anxiety (see review Sindermann et al., 2021), this is the first study to explore the association between brain structure and the Tri-level Model. This work has the potential to provide insight into the association between the structure of specific brain regions and specific dimensional symptoms of depression and anxiety. We expected alterations in gray matter volumes of emotion generation regions, the amygdala and NAcc, to be negatively associated with the General Distress symptom dimension based on the literature linking these regions to negative emotionality and dysphoric mood. We predicted that volumetric alterations in the amygdala and NAcc would also be associated with Fears and Anhedonia-Apprehension symptom dimensions, respectively. This is based on literature linking the amygdala to anxiety and NAcc to depression both functionally and structurally. Lastly, we expected that emotion regulation regions (OFC, VLPFC, DLPFC) would be negatively associated with the General Distress, Anhedonia-Apprehension, and Fears symptom dimensions, due to this association being mostly consistent across the depression and anxiety literature. We tested these predictions using linear regression analyses and significant results were followed up with specificity analyses by adjusting for the other Tri-level Model symptom dimensions.