Search

We seek to understand the mechanisms by which early life adversity (negative life events, familial conflict, and neighborhood violence) are associated with anhedonic and depressive symptoms by examining functions of the reward and emotion centers of the brain. Previous work has shown the relationship between early life adversity and anhedonia as a risk factor for depression. Thus, the goal of this paper is to inform our understanding of the mechanisms by which ELA is associated with anhedonic and depressive symptoms.

There is a rich literature linking individual differences in reward-related behaviors and neurobiology to various forms of psychopathology, such as depression and substance use problems. Associations with various disorders leaves questions of specificity; differences in reward functioning may be related to various forms of psychopathology in distinct ways (e.g., decreased reward drive in depression, increased sensation seeking behaviors in substance use problems), or, be related to transdiagnostic clinical factors associated with multiple disorders (e.g.., negative emotionality or distress in both depressive and anxiety disorders, decreased reward sensitivity in both depression and substance use problems). Clarifying this distinction is critical for understanding etiologies of various types of psychopathology and improving the precision of clinical associations with reward functioning. Here, we will model a hierarchical structure of psychopathology in late adolescents and examine prospective associations at the disorder and higher-order dimensional level with various neural and behavioral measures of reward processing measured in early adolescence. Results will provide insights into how reward functioning in early adolescence is prognostically associated with specific disorders, transdiagnostic dimensions, or both. Though examining a broad range of psychopathology, we will particularly focus on associations with depression and substance use problems after accounting for variance from higher order dimensions as they are strongly associated with reward functioning and are at increased risk during late adolescence. We have explored the factor structure of clinical variables. This preregistration is focused on testing the prospective covariates.

Adolescence is characterized by rapid neurodevelopment of reward systems and increases in reward-seeking behavior. This developmental period is adaptive for learning behavioral approaches related to reward acquisition, but heightened sensation seeking is also associated with risky behaviors such as substance use. Moreover, adolescence is a key risk period for psychopathology such as depression and anhedonia, which are associated with decreased reward functioning. In contrast, cognitive inhibitory control shows a slower, but still increasing, developmental trend across adolescence. Adolescents with less inhibitory control have also been shown to engage in more risky behaviors such as substance use. Overall, adolescents tend to show a sharp increase in reward-related functioning, and relatively less steep increases in inhibitory control, and individual differences in these domains are associated with key outcomes such as substance use and depression. Thus, examining factors that predict individual differences in the trajectories of reward responsivity and inhibitory control may provide key insights into the etiologies of key outcomes as well as provide opportunities for early intervention. While the general trajectories of reward responsivity and inhibitory control across adolescence are relatively well understood (Shulman et al., 2016, DCN), little work has examined predictors of individual differences in these developmental trajectories. Here, we examine developmental trajectories of reward sensitivity and inhibitory control across adolescents ages 9-16. We also examine baseline neural and clinical predictors of trajectories of reward function and inhibitory control. Specifically, we will examine functional connectivity network strength of a subcortical reward hub and a cortical control hub during a reward task, parental diagnoses of depression and substance use problems, and child depression as clinical predictors.

Exploratory latent class analysis will be used to identify sub-groups of youth based on distinct patterns of brain volumes in subcortical and cortical regions that are empirically implicated in depression.

Reward network functioning is implicated in several key outcomes such as risk-taking behavior and psychopathology, which are particularly of interest in adolescence. Conventionally, a single a priori or data-driven aggregate-level network is estimated for all individuals in a sample and quantitative metrics about network functioning are potential factors related to behavioral or clinical outcomes. However, connectivity networks are increasingly recognized as having heterogenous structures across individuals, 1-2 such that multiple individuals have qualitatively different network structures than a single group aggregate network. Subgroup and individualized networks may better reflect individual differences in network functioning and improve the ability to detect associations between network functioning and external variables. Here, we parse network heterogeneity by using different group-, subgroup-, and individual-level information to assess differences in identifying brain-behavior associations in reward functioning. We use Group Iterative Multiple Model Estimation (GIMME), a structural equation modeling-based approach to modeling effective connectivity, to compare these approaches in a sample of early adolescents scanned during a monetary reward fMRI task that has previously been used in adolescents (DOI: 10.1176/appi.ajp.2008.07081336). First, GIMME will be used to create three sets of networks: (1) a group-level network of paths that represent a majority of the sample, (2) subgroup-level networks that represent smaller, more homogeneous groups of individuals, and (3) individual-level networks which add paths significant to the individual onto the group network for each person. We will then examine if identified subgroups are robust to incrementally increased noise (i.e., perturbations). If subgroups are validated, network paths and subgroup membership will be associated with multiple external reward outcomes. If valid subgroups are not identified, GIMME will be rerun without subgroups and then group- and individual-level networks will be re-estimated and associated with reward outcomes. Then, purely individual-level (idiographic) networks will be separately identified by beginning with a true null model and adding on individual level paths to each person. The presence or absence of individual paths in these networks will also be associated with reward outcomes. The pure individual-level networks will then be aggregated to create a single group level network that will be qualitatively compared to the GIMME-group level network.

See https://osf.io/6n5j2