Inflammatory pathways link socioeconomic inequalities to white matter architecture

Abstract

Socioeconomic disadvantage confers risk for aspects of ill health that may be mediated by systemic inflammatory influences on the integrity of distributed brain networks. Following this hypothesis, we tested whether socioeconomic disadvantage related to the structural integrity of white matter tracts connecting brain regions of distributed networks, and whether such a relationship would be mediated by anthropometric, behavioral, and molecular risk factors associated with systemic inflammation. Otherwise healthy adults (N= 155, aged 30-50 years, 78 men) completed protocols assessing multilevel indicators of socioeconomic position (SEP), anthropometric and behavioral measures of adiposity and cigarette smoking, circulating levels of C-reactive protein (CRP), and white matter integrity by diffusion tensor imaging. Mediation modeling was used to test associations between SEP indicators and measures of white matter tract integrity, as well as indirect mediating paths. Measures of tract integrity followed a socioeconomic gradient: individuals completing more schooling, earning higher incomes, and residing in advantaged neighborhoods exhibited increases in white matter fractional anisotropy and decreases in radial diffusivity, relative to disadvantaged individuals. Moreover, analysis of indirect paths showed that adiposity, cigarette smoking, and CRP partially mediated these effects. Socioeconomic inequalities may relate to diverse health disparities via inflammatory pathways impacting the structural integrity of brain networks.

Keywords: diffusion tensor imaging; inflammation; social health disparities; socioeconomic inequality; stress; white matter fractional anisotropy.

Figure 1.

Shown are associations between years of schooling and fractional anisotropy (FA). (A) Clusters of voxels where increased years of schooling exhibited associations with FA. Warm-colored voxels reflect positive associations, and cool-colored voxels show negative associations. Side panels show slice placements. All voxels were adjusted for multiple testing at the cluster level, with a minimal cluster size of 20 voxels. (B) Probability density functions of regression coefficients for the total effects of schooling on FA, after controlling for age and sex. Yellow shows the observed distribution, and the blue shows the average distribution from the bootstrap permutation tests. (C) Cumulative distribution functions for the same data as in B. Same color terms used for observed and bootstrapped data. (D) Q–Q plot comparing the observed distribution in B to a Gaussian distribution with the same mean and variance.

Figure 2.

Aassociations of occupant-adjusted household income and FA are shown, with the same plotting conventions as in Figure 1A–D.

Figure 3.

Associations of community-level SEP and FA are shown, with the same plotting conventions as in Figures 1 and 2A–D.

Figure 4.

Correlations between FA and axial diffusivity (AD; open bars) and radial diffusivity (RD, filled bars). Data were analyzed separately for voxels with significant positive or negative total path effects for years of schooling (A), household income (B), and community-level SEP (C). Error bars show standard deviation of the distributions.

Figure 5.

Results of mediation analysis. (A) Diagram of indirect path effects of smoking and adiposity (waist circumference) for each of the three SEP indicators. Line thickness denotes the percentage of white matter voxels with significant effects. These percentages are shown on the left along with the average path coefficient across all white matter voxels in the brain. (B) Direct and indirect path effects for smoking status and adiposity factors through the CRP term, with the same plotting conventions as in A. (C) Distribution of voxels with significant indirect paths for the results shown in A. These maps were adjusted for multiple-testing at the cluster level with a minimum cluster size of 20 voxels.

Figure 6.

Change in the percent of significant voxels for each indirect path shown in Figure 5A when the CRP term is included as a covariate in the model. Dashed lines show the upper and lower 95% CIs from the control analysis, wherein a random noise term was included instead of CRP. Solid line shows the mean value of these control analyses. Percentages are from all white matter voxels. Change from within the subset of originally significant voxels is reported in the text.