Nucleus Accumbens Functional Connectivity with the Default Mode Network: Black Children’s Diminished Returns of Household Income

Introduction: Although research has established a link between socioeconomic status (SES) and neuroimaging measures, weaker SES effects are shown for Blacks than Whites. This is, in part, due to processes such as stratification, racism, mineralization, and othering of Black people in the US. Purpose: This study had two aims: First to test the association between household income and the nucleus accumbens (NAcc) resting-state functional connectivity with the Default Mode Network (DMN) in children, and second, to investigate racial heterogeneity in this association. Methods: This cross-sectional study used data from the Adolescent Brain Cognitive Development (ABCD) study. We analyzed the resting-state functional connectivity data using Magnetic Resonance Imaging (rsfMRI) of 7903 US pre-adolescents who were between ages 9 and 10 years old. The main outcome was the NAcc resting-state functional connectivity with DMN. The independent variable was household income. Age, sex, and family structure were the study covariates. Race was the moderator. Mixed-effects regression models were used for data analysis with and without interaction terms between household income and race. Results: Higher household income was associated with higher NAcc resting-state functional connectivity with DMN. Race showed a statistically significant interaction with household income suggesting that the NAcc resting-state functional connectivity with DMN was significantly weaker for Black compared to White pre-adolescents. Conclusions: In line with Minorities’ Diminished Returns (MDRs), the association between household income and pre-adolescents NAcc resting-state functional connectivity with DMN is weaker in Black than in White children. This result is of interest because DMN’s functional connectivity with NAcc may have a role in cognitive flexibility and reward processing. The weaker links between SES indicators and neuroimaging findings for Blacks than for Whites may reflect the racialization of Black people in the US. Social stratification, racism, and discrimination may minimize the returns of SES for Black families, who have been oppressed for centuries.

Among various functional connectivity measures evaluated by rsfMRI that are linked to SES, is connectivity between NAcc and DMN (Rakesh et al., 2021). Analysis of connectivity between DMN and brain regions such as NAcc is important, because they reflect cross-system-level measures and are involved in higher-order brain function. As such, NAcc-DMN functional connectivity may reflect some of the coordinated multi-system cognitive, emotional, and behavioral tasks across multiple regions and networks. NAcc resting-state functional connectivity with DMN may correlate with altered cognition, emotions, and psychiatric disorders (Karcher, O'Brien, Kandala, & Barch, 2019;Smallwood et al., 2021).

Aims
Built on the MDRs, this study used data of a national sample of 9-to-10-year-old pre-adolescents from the Adolescent Brain Cognitive Development research (ABCD) data (B. Casey et al., 2018;Karcher et al., 2019;Lisdahl et al., 2018;Luciana et al., 2018;Research & Staff, 2018) for two aims: First, to test the association between household income and NAcc resting-state functional connectivity with DMN; and second, to explore racial heterogeneity in this association. We expected a positive association between household income and NAcc resting-state functional connectivity with DMN. Additionally, in line with the MDRs theory (Shervin Assari, Shanika Boyce, Mohsen Bazargan, & Cleopatra H Caldwell, 2020; Boyce et al., 2020), we expected weaker effects of household income on NAcc resting-state functional connectivity with DMN for Black pre-adolescents compared to White pre-adolescents.

Design and Settings
This secondary cross-sectional analysis was based on the Adolescent Brain Cognitive Development (ABCD) study (B. Casey et al., 2018;Karcher et al., 2019;Lisdahl et al., 2020;Luciana et al., 2018;Research & Staff, 2018), the largest child brain development study ever, with a diverse sample in terms of SES, sex, race and ethnicity (Auchter et al., 2018;Research & Staff, 2018). Detailed information regarding ABCD methods is available here (Auchter et al., 2018).

Participants and Sampling
The ABCD study participants were 9 to 10 years old and were selected from 21 sites across 15 states, encompassing over 20 % of the total United States population of 9/10-year-old children (Auchter et al., 2018;Garavan et al., 2018). For sampling and selection, school selection was informed by sex, race, ethnicity, SES, and urbanicity. These recruitment processes were precisely designed, implemented, and evaluated across the 21 study sites (Ewing, Bjork, & Luciana, 2018). Although the ABCD sample is not representative, the sample is a near approximation of U.S. children (Garavan et al., 2018). Participants, 7903 children aged 9-to-10-year-olds, could be included regardless of race and ethnicity (Garavan et al., 2018). They needed to have complete data on our variables and meet satisfactory criteria for rsfMRI (according to the DEAP baseline data).

Brain Imaging:
Functional MRI (rsfMRI) was used to estimate pre-adolescents' NAcc resting-state (task-negative) functional connectivity with DMN. Brain imaging in the ABCD study was based on three 3 tesla (T) scanner platforms: Philips Healthcare, GE Healthcare, and Siemens Healthcare .
T1-weighted and T2-weighted brain images, carefully harmonized, were drawn from the MRI devices (B. Casey et al., 2018). In order to reduce bias due to variation in imaging sites, images were corrected for gradient non-linearity distortions (Jovicich et al., 2006). Pre-processed structural data are available from the ABCD study, and are calculated based on T1-and T2-weighted images that maximize mutual information's relative position and orientation across images (Wells III, Viola, Atsumi, Nakajima, & Kikinis, 1996). By using tissue segmentation and sparse spatial smoothing, the ABCD study performed intensity non-uniformity correction. Moreover, the images have been resampled with 1-mm isotropic voxels into rigid alignment within the brain atlas. Furthermore, using FreeSurfer software, version 5.3.0 (Harvard University), these volumetric measures were constructed. The images have also undergone surface optimization (Fischl & Dale, 2000;Fischl, Sereno, & Dale, 1999), and nonlinear registration to a spherical surface-based atlas (Fischl et al., 1999).

Study Variables
The study variables included household income (independent variable), race (moderator), age, sex, and family structure (confounders), and NAcc resting-state functional connectivity with DMN (dependent variable).

Independent Variable:
Household income: Household income was a three-level nominal variable: less than 50k, 50-100k, and 100k+ per year. Household income less than 50k was the reference group.

Dependent Variable:
NAcc resting-state functional connectivity with DMN: This variable was a continuous measure and reflected the Pearson correlation test between the BOLD measures of the two networks over time.
DMN was defined according to the Gordon parcellation scheme that divides brain networks into 12 predefined resting state networks (RSN) (Gordon et al., 2016). In this study, we only used data of DMN, not the other 11 brain networks. To calculate this information, the ABCD completed 4-5 five-minute resting state scans (eyes open). This was used to ensure at least eight minutes of relatively low-motion data. More details are explained here . Preprocessing was carried out by the ABCD Data Analysis and Informatics Core using the standardized ABCD pipeline . Next, www.scholink.org/ojs/index.php/rhs Research in Health Science Vol. 6, No. 3, 2021 fMRI time courses were projected onto FreeSurfer's cortical surface. Using these time courses, withinand between-network connectivity (Pearson correlation) were calculated on the basis of standard protocols based on the Gordon scheme (Gordon et al., 2016). For more information regarding these processes, please see here (B. J. Casey et al., 2018;. Family SES is shown to be correlated with rsfMRI of the DMN (Karcher et al., 2019). Our outcome was associated with a wide range of cognitive measures as an indicator of validity of our measure (Supplementary Figure).

Moderator:
Race. Race was reported by the parent and was treated as a nominal variable: Black, Asian, Other/Mixed, and White (reference group).

Confounders:
Age. Age was a continuous variable. Parents reported the child's age as months.
Sex. Sex was a categorical variable with 1 for boys and 0 for girls.
Family Structure. Family Structure was also a dichotomous variable, self-reported by the parent interviewed, and coded 1 vs. 0 for married and unmarried (any other condition).

Data Analysis
We used the Data Exploration and Analysis Portal (DEAP), which is a user-friendly online platform for multivariable analysis of the ABCD data, for data analysis. For multivariable analyses, two mixed-effects regression models were estimated (Supplementary Table). Model 1 tested the additive effects of household income, race, and covariates. Model 2 also included interaction terms between household income and race. In all models, the NAcc resting-state functional connectivity with DMN was the outcome. Regression coefficient (b), standard error (SE), and p-value were reported from our regression models.

Ethical Aspect
While the original ABCD research protocol went through an Institutional Review Board (IRB) in several institutions, including the University of California, San Diego (UCSD), our analysis was found to be exempt from further IRB review by the Charles R Drew University of Medicine and Science (CDU). Moreover, several institutional IRBs approved the study protocol. All children provided assent.

Results
This study was performed on 7903 children aged between 9 to 10 years old.   weaker positive association between household income and the DMN's resting-state functional connectivity in Black than White children.

Discussion
First, we found a positive association between household income and the NAcc resting-state functional connectivity with DMN. Second, there were racial differences in the associations between household income and the resting-state DMN's functional connectivity with the striatum. In line with the MDRs phenomenon, the correlation between household income and the NAcc resting-state functional connectivity with DMN was larger for White than Black children.
Our first finding aligns with the well-described effects of SES indicators such as household income on brain structure and function in adolescents and young people. However, most of this work has been conducted on individual brain structures (Hanson, Chandra, Wolfe, & Pollak, 2011;Jednoróg et al., 2012;Lawson, Duda, Avants, Wu, & Farah, 2013;Noble, Houston, Kan, & Sowell, 2012). High SES, for example, is linked to the activity and size of brain structures such as the NAcc, amygdala, hippocampus, and cerebral cortex (Hanson et al., 2011;Muscatell et al., 2012;Noble et al., 2012;Noble et al., 2015).
One study using the ABCD study showed that SES is linked to the DMN functional connectivity at rest (Rakesh et al., 2021). These neurodevelopmental correlates of SES may mediate why SES influences language, reading, social cognition, executive functions, and spatial skills (Noble et al., 2015). However, very few studies to date have explored the relationship between household income and the NAcc resting-state functional connectivity with DMN.
The association between household income and NAcc resting-state functional connectivity with DMN may be because household income is a proxy of low-risk (Spann et al., 2014), high-support (Anton, Jones, & Youngstrom, 2015;Woods-Jaeger, Cho, Sexton, Slagel, & Goggin, 2018), low-stress (Parkes, Sweeting, & Wight, 2015), and positive social environment, which predicts healthy brain development  Household income and race have multiplicative rather than additive effects on NAcc resting-state functional connectivity with DMN. Under racism, Black pre-adolescents remain at high risk, regardless of their SES. This is in contrast to White pre-adolescents for whom high SES considerably reduces the risk. These unequal associations may be due to racism.
In this study, race was a social rather than biological determinant. Race is a social factor in all the MDRs literature, including studies on adolescents' brain development. Subsequently, the racial differences reported here have resulted from the differential treatment by society, which is preventable, not differences due to genes that are innate. Race is a proxy of racism, including labor market discrimination, low school quality, segregation, and differential policing, leading to reduced household income, even for high SES and successful people who have secured economic and human resources.
This view contrasts with a biologic deterministic approach that attributes racial differences to genetics (Herrnstein & Murray, 2010).

Conclusions
While high household income is associated with higher NAcc resting-state functional connectivity with DMN, this link may be weaker for Black than White American pre-adolescents. This racial variation might be in part due to racism, social stratification, and segregation, all reducing the effects of SES Random: ~(1|abcd_site/rel_family_id) a) Correlation between the resting-state functional connectivity between NAcc and default mode network and positive urgency (upps_ss_positive_urgency = y axis) b) Correlation between the resting-state functional connectivity between NAcc and default mode network and reward responsiveness (bisbas_ss_basm_rr= y axis)