Mediation of the association between disadvantaged neighborhoods and cortical microstructure by body mass index

Living in a disadvantaged neighborhood (area deprivation) is linked to worse health outcomes, including poor brain health, such as decreased brain volume. Mechanisms underlying how neighborhood conditions affect brain health are unclear, but obesity represents a plausible pathway, as disadvantaged neighborhoods are associated with poorer food quality, limited opportunities for physical activity, and higher intake of trans-fatty acids (TFAs) and sodium. Chronic neighborhood stressors may also elevate allostatic load and alter eating behaviors, increasing preference for highly palatable, unhealthy foods. Prior neuroimaging work indicates that stress can alter brain structure and function, promoting food cravings and obesity risk. Neighborhood disadvantage, high BMI, and chronic stress have been linked to cortical microstructure differences measured by the T1w/T2w ratio, a proxy sensitive to intracortical myelin (though not specific and may also reflect neurite/synaptic density). Given laminar differences in cortical inputs/outputs and function, evaluating cortical microstructure across layers can indicate how adverse environments impact specific information-processing routes. The present study investigated the relationship between the area deprivation index (ADI) and cortical microstructure (T1w/T2w ratio) across multiple cortical ribbon levels, and tested potential mediators BMI and perceived stress. In a subset, the relationship between TFA intake and cortical microstructure was examined. The authors hypothesized that worse ADI would be associated with higher BMI, obesogenic diet (high TFA intake), and higher stress, and that these factors would relate to adverse cortical microstructural patterns in reward-related, emotion regulation, and cognitive regions.

Prior research shows neighborhood disadvantage relates to worse brain outcomes, including reduced cerebral and hippocampal volumes. Disadvantaged neighborhoods are associated with greater obesity risk due to poor dietary environments and constraints on physical activity; neighborhood disadvantage correlates with higher intake of TFAs and sodium. TFAs, abundant in fried fast foods, contribute to obesity, especially abdominal adiposity. Chronic stressors tied to disadvantaged environments elevate allostatic load and can drive preferences for highly palatable, unhealthy foods as a coping response; neuroimaging studies show stress alters brain structure and function, enhances food cue reactivity, and impairs self-control, increasing obesity risk. BMI has been implicated as a mediator linking neighborhood disadvantage to reduced brain volume. Cortical microstructure assessed by the T1w/T2w ratio relates to intracortical myelination, which is crucial for neural synchrony and fine-tuning cortical circuits and cognition. The T1w/T2w ratio is sensitive to intracortical myelin (though limited in specificity and may index neurite/synaptic density) and has demonstrated utility regarding cortical maturation and cognition. Cortical layers differ in cell populations and connectivity; upper layers contain many myelinated inhibitory interneurons supporting gamma oscillations, while deeper layers contain myelinated excitatory neurons integrating numerous inputs. Prior work links socioeconomic disadvantage and stress to T1w/T2w alterations and myelin-sensitive measures (e.g., magnetization transfer), with reports of slower myelin growth under childhood neighborhood disadvantage and lower myelination associated with adult SES. BMI has been associated with intracortical myelin differences, and perceived stress has been linked to lower T1w/T2w ratio in supramarginal gyrus.

Design and participants: Cross-sectional analysis of 92 adults (27 men, 65 women) from Los Angeles who completed neuroimaging (T1w and T2w scans) and provided residential addresses between Oct 30, 2019 and Jul 14, 2022. Recruitment via campus/community flyers, emails, social media, and ClinicalTrials.gov (NCT05120908). Exclusion criteria: major neurological condition, current/past psychiatric illness, vascular disease, weight loss/abdominal surgery, substance use disorder, CNS-interfering medications, pregnancy/breastfeeding, strenuous exercise (>8 h/week continuous exercise), weight >400 lbs, metal implants, and poor-quality MRI (assessed with MRIQC using published thresholds). No participants had poor-quality images. IRB-approved protocols (UCLA IRB Nos. 16-000281, 20-000540, 20-000515, 20-002326). Written informed consent obtained. Assessments: Demographics, height/weight for BMI (kg/m²). Neighborhood disadvantage measured using the 2020 Neighborhood Atlas Area Deprivation Index (ADI) at the California State level (deciles 1–10; higher = greater deprivation), based on participant census block group; state ADI used due to non-normal distribution of national ADI in this California-only sample. Perceived stress measured with the 10-item Perceived Stress Scale (PSS; scores 0–40; higher = greater stress; reliability Cronbach’s alpha ~0.74–0.89). Diet assessed in a subset (N=81) using the VioScreen web-based graphical food frequency system (156 items, portion sizes, preparation methods; prior 3 months), yielding nutrient estimates including individual trans-fatty acids: trans-hexadecenoic acid, trans-octadecenoic (elaidic) acid, trans-octadecadienoic (linolelaidic) acid, and total TFA. Imaging acquisition: 3T Siemens Prisma scanner, HCP protocols v4.3. T1w (TE=1.81 ms, TR=2500 ms, slice thickness=0.8 mm, 208 slices, matrix 320×300, voxel 1.0×1.0×0.8 mm). T2w (TE=564 ms, TR=3200 ms, slice thickness=0.8 mm, 208 slices, matrix 320×300, voxel 1.0×1.0×0.8 mm). Spin-echo fieldmaps in AP/PA for distortion correction. Preprocessing and T1w/T2w computation: HCP minimal pipelines with FreeSurfer 6.0 for volume segmentation and cortical surface reconstruction. T1w/T2w ratio estimated at 5% increments across cortical thickness from gray–white to pial boundary; averaged into four cortical ribbon levels: deep (5–25%), lower-middle (30–50%), upper-middle (55–75%), superficial (80–100%). Resulting myelin maps parcellated using HCP MMP1.0 atlas into 360 cortical regions, each with four T1w/T2w estimates. Statistical analysis: Partial correlations (controlling for age and sex) between ADI and BMI, PSS, and TFA intake (for diet subset) using SPSS v28 with 5000 bootstrap samples; significance p<.05. Non-rotated partial least squares correlational (PLSC) analyses conducted separately by cortical level (n=92) to identify regions where T1w/T2w covaried with ADI (age and sex included in the demographic block; weights preset to identify ADI-sensitive regions); reliability assessed via bootstrap estimation (5000 samples). Regions with bootstrap ratio >3.3 (≈p<.001) considered significant. For significant parcels, T1w/T2w values were extracted and combined as weighted averages (by parcel size) to create indices for ADI-positive (positively related to ADI) and ADI-negative (negatively related) areas. Mediation modeling: Structural equation modeling (SEM) using lavaan in R to test mediation of ADI associations with average T1w/T2w in ADI-positive and ADI-negative areas by BMI and PSS, controlling for age and sex. Variables standardized prior to modeling; missing PSS values (n=4) handled via maximum likelihood. Model fit evaluated by chi-square p-value, comparative fit index (CFI), and standardized root mean square residual (SRMR). Paths with p<.05 deemed significant.

- Sample: N=92; mean age 28.0±10.3 years (range 18–58); 29% male. ADI (state decile) positively correlated with BMI (r=0.27, p=.01) and PSS (r=0.22, p=.04); BMI and PSS not correlated (r=0.04, p=.73). - PLSC results: Worse ADI associated with increased T1w/T2w ratio predominantly in middle/superficial cortical levels within medial prefrontal and cingulate regions (ADI-positive areas; p<.001). Worse ADI associated with decreased T1w/T2w ratio in middle/deep levels within supramarginal, middle temporal, and primary motor regions (ADI-negative areas; p<.001). - Mediation (SEM): Model fit good (χ²(2)=0.025, p=.99; CFI=1.0; SRMR=0.002). BMI showed significant positive direct effects on both brain indices (Z=4.26, p<.001; Z=2.14, p=.03). BMI partially mediated the relationship between ADI and average T1w/T2w in ADI-positive areas (indirect effect Z=2.39, p=.02), accounting for 29% of the total effect; mediation for ADI-negative areas was not significant (Z=1.71, p=.09). PSS had negative, non-significant direct effects on both brain indices (Z≈-1.10 and -1.02) and did not mediate ADI–brain relationships (all p>.30). - Diet correlations (n=81): Average T1w/T2w in ADI-positive areas positively correlated with trans-octadecenoic (elaidic) acid (r=.29, p=.01) and total TFA intake (r=.28, p=.01). No significant correlations for ADI-negative areas. BMI positively correlated with TFA measures including elaidic acid and total TFA (r=.27, p=.02 for both). - Overall: ADI relates to distinct laminar and regional cortical microstructural patterns; BMI partially explains ADI-associated increases in superficial/middle medial prefrontal and cingulate T1w/T2w; TFA intake aligns with these increases.

The study addressed whether neighborhood disadvantage (ADI) is associated with cortical microstructure across cortical layers and whether BMI and perceived stress mediate these relationships, with an additional examination of dietary TFA intake. Findings show that worse ADI relates to increased T1w/T2w in middle/superficial medial prefrontal and cingulate cortices and decreased T1w/T2w in middle/deep supramarginal, middle temporal, and primary motor regions. BMI partially mediated the ADI association with T1w/T2w increases in prefrontal/cingulate regions, while stress did not significantly mediate any associations. These results suggest that obesogenic features of disadvantaged neighborhoods, including higher BMI and poorer diet quality, are linked to cortical microstructural alterations in regions subserving reward processing, emotion regulation, and higher cognition. Alterations in superficial layers could impair top-down modulation and flexible state-dependent processing of inputs. The T1w/T2w increases might reflect changes in intracortical myelination, though the measure’s limited specificity means contributions from neurite/synaptic density or lipid-related changes cannot be excluded. Correlations with elaidic acid and total TFA intake support a potential role for diet-related mechanisms; in obesity, a fatty acid-rich neural milieu, lipid droplets, lipid-laden astrocytes, and altered BBB transport may influence cortical microstructure. Decreases in T1w/T2w in middle/deep layers of regions associated with the mirror neuron system suggest potential impacts on intercortical and subcortical-cortical communication underlying social interaction and motor functions. While perceived stress correlated with ADI, it did not correlate with the identified T1w/T2w indices here; lateralization (e.g., right vs. left supramarginal) and reliance on acute rather than chronic stress measures may explain discrepancies with prior studies. Overall, the findings support a pathway where neighborhood disadvantage influences cortical microstructure through obesity-related factors, with implications for cognitive, emotional, and social processing.

Worse neighborhood deprivation (ADI) is associated with cortical microstructural alterations characterized by decreased T1w/T2w ratio in middle/deep layers of supramarginal, middle temporal, and primary motor regions and increased T1w/T2w ratio in middle/superficial layers of medial prefrontal and cingulate regions. The latter association is partially mediated by higher BMI and is positively related to dietary trans-fat intake. These results indicate that obesogenic aspects of disadvantaged neighborhoods may disrupt flexible information processing in circuits for reward, emotion regulation, and cognition, with broader implications for brain health. Future research should employ complementary imaging modalities to clarify biological underpinnings (myelin vs. neurite/synaptic vs. lipid effects), incorporate longitudinal and life-course measures of neighborhood context (including childhood ADI), assess chronic stress biomarkers, and refine adiposity metrics beyond BMI to better delineate mediating mechanisms across development and adulthood.

- Biological specificity: The T1w/T2w ratio, while sensitive to intracortical myelin, has limited specificity and may reflect neurite or synaptic density; alternative mechanisms such as lipid-related changes in obesity are speculative. Additional myelin- and microstructure-specific modalities are needed. - Cross-sectional neighborhood measure: ADI was assessed at a single timepoint without residence duration, life-course exposure, or childhood neighborhood data; effects may differ based on timing and duration of exposure. - Stress assessment: PSS reflects perceived stress over the prior month; lack of chronic stress measures may underestimate associations with cortical microstructure and mediation effects. - BMI limitations: BMI imperfectly indexes adiposity and provides no information on fat distribution or composition; muscle mass can confound BMI. - Sample representativeness: Potential selection bias; race/ethnicity proportions differ from 2020 Census distributions, possibly limiting generalizability.