Combining lifestyle risks to disentangle brain structure and functional connectivity differences in older adults

Figures

• 

  Fig. 1. Combined lifestyle risk measured with the developed combined lifestyle risk score. Lifestyle related data, that is, alcohol consumption (g/week), smoking (pack years), social integration (social integration index23), and physical activity (metabolic equivalent per week22) were z-transformed to obtain uniform measure units. Z-transformed data on the protective variables physical activity and social integration were first multiplied with (−1) and then summed up with z-transformed data on the risk variables of alcohol consumption and lifetime smoking. This resulted in an individual score of combined lifestyle risk. a The combined lifestyle risk score shows a nearly normal distribution. b Rank-sorted data of the combined lifestyle risk score: The participant with the lowest score gets rank 1 and the highest rank 549. Green colors indicate more protective lifestyle behavior and magenta colors indicate more risky lifestyle behavior

• 

  Fig. 2. Differences in cortical folding associated with lifestyle risk in different risk score models. a–i Small boxes represent different risk score models derived from the stepwise exclusion approach with excluded single lifestyle variables grayed out. Associations between cortical folding and the combined lifestyle risk score, or the different risk score models, respectively, are depicted on the inflated surfaces of the fsaverage brain. Arrows indicate the work flow of the stepwise exclusion. Negative associations are represented in blue, and positive associations in red. The recurrent negative associations between higher lifestyle risk and reduced cortical folding in left dorsal premotor cortex (dPMC) and right ventro-lateral prefrontal cortex (vlPFC) are colored in pink and orange, respectively. All results are corrected for age and gender and corrected for multiple comparisons using Monte Carlo Z simulations with α = 0.05. Some of the associated regions in specific risk score models marked with an asterisk did not pass an additional correction for multiple comparisons at α = 0.05/13 = 0.0033. For detailed information see Supplementary Table 2. n.s. = not significant, ACT = physical…

• 

  Fig. 3. Lifestyle risk-associated increases in resting-state functional connectivity (RSFC) to the seed in the left dorsal premotor cortex (dPMC). a–m Abbreviations in the small boxes refer to the same variables as in Fig. 2. Coronar sections show brain regions exhibiting increases in RSFC depicted in red. The recurrent pattern of higher RSFC between the dPMC and the somato-motor cortex is highlighted in pink. Arrows indicate the work flow of the stepwise exclusion. Transparent sections represent not significant (n.s.) results. This analysis included a subsample of the study population (n = 501). Results were significant at α = 0.05 (cluster level corrected, cluster-forming threshold α = 0.001). S1 = primary somatosensory cortex, MTG = middle temporal gyrus, M1 = primary motor cortex, DOC = dorsal occipital cortex, EC = entorhinal cortex, TPJ = temporo-parietal junction, L = left hemisphere, R = right hemisphere

• 

  Fig. 4. Lifestyle risk-associated increases in resting-state functional connectivity (RSFC) to the seed in right ventro-lateral prefrontal cortex (vlPFC). a–m Transversal sections show brain regions exhibiting increases in RSFC depicted in red. The recurrent pattern of higher RSFC between right vlPFC and superior frontal gyrus is outlined in orange. Arrows indicate the work flow of the stepwise exclusion. This analysis included a subsample of the study population (n = 501). For other conventions, see Fig. 2. Results were significant at α = 0.05 (cluster level corrected, cluster-forming threshold α = 0.001). SFG = superior frontal gyrus, MTG = middle temporal gyrus, THAL = thalamus, NC = caudate nucleus, n.s. = not significant, L = left hemisphere, R = right hemisphere

• 

  Fig. 5. Comparison of analyses without adjustment for polygenetic risk scores (GRS) and with adjustment for polygenetic risk score (PRS). a The lower row shows the results of the analyses of cortical folding and resting-state functional connectivity (RSFC) with adjustment for PRS. b The upper row shows the results of the analyses of cortical folding and RSFC with adjustment for PRS. Whereas the general pattern of the results did not change after adjustment for PRS, an additional decrease in cortical folding in right precuneus was observed. n.s. = not significant, L = left hemisphere, R = right hemisphere

• 

  Fig. 6. Analyses of cortical folding with additional adjustment for a set of non-lifestyle covariates. The illustrations show the associations between the combined lifestyle risk score and cortical folding when correcting for a set of non-lifestyle covariates. The additional adjustments did not change the general associations between the combined lifestyle risk and decreased cortical folding within the left dorsal premotor cortex (dPMC) (pink) and the right ventro-lateral prefrontal cortex (vlPFC) (orange). BDI-II = depressive symptomatology, as measured with Beck’s depression inventory-II36, ISCED = educational level as measured with international classification of education37, PRS = polygenic risk score