Stressor-evoked brain activity, cardiovascular reactivity, and subclinical atherosclerosis in midlife adults

Figure 1.. Correlogram. A correlogram involving the main variables in our mediation framework and several cardiovascular disease risk factors. Colors and annotations display the strength of their associations as measured by Pearson’s correlation coefficients. A cross represents those associations whose p-values after false discovery rate correction were above the significance level α=0.05. SBP, systolic blood pressure; HDL, high-density lipoproteins; Trigs, triglycerides; CA-IMT, carotid artery intima-media thickness; mm, millimeter, SBP_AUCg, Area under the curve with respect to the ground; SBP_AUCi, Area under the curve with respect to increase; ascvd_10y, 10-year atherosclerosis cardiovascular disease risk score.

Figure 2.. Main effects. For each cohort (PIP and NOAH) the stressor-evoked brain activation patterns at the group level and in terms of the Cohen’s d effect size measure in both fMRI tasks (Stroop and MSIT). Only effects whose p-value is below 0.05 after false discovery rate correction and with a cluster size threshold of k > 50 voxels are displayed. Orange colors represent greater average activity in the incongruent condition compared to the congruent condition, and blue colors the opposite. In arrows, the spatial similarity between activation patterns as measured by Pearson’s correlations.

Figure 3.. Out-of-sample performances. For each possible mediator variable (M), the out-of-sample performance using a L2-penalized principal component regression and applied to the different paths in the mediation analysis framework. Here, the input variable (X) is voxelwise responses and the output variable (Y) is CA-IMT. Each dot represents the coefficient of determination calculated from the observed .vs predicted values generated from a particular run of the nested cross-validation procedure. In addition, bars and error bars display the average and the 95% Confidence intervals across these values.

Figure 4.. Encoding weight maps of indirect effects. For the scenario with SBP_AUCg as mediator, the encoding weight maps obtained from transforming back to voxel space those principal components whose a·b products were significant based on bootstrapping (5000 resamples) and after correcting for a false discovery rate of 0.05. Warm and cool colors represent a positive and negative mediated association between stressor-evoked brain activity and CA-IMT respectively. For visualization purposes, only weights with |z| > 1 after spatial standardization are displayed.