Association Between Lifestyle at Different Life Periods and Brain Integrity in Older Adults

Abstract

Background and objectives: Lifestyle behaviors, including engagement in complex mental activities, have been associated with dementia risk and neuroimaging markers of aging and Alzheimer disease. However, the life period(s) at which lifestyle factors have the greatest influence on brain health remains unclear. Our objective was to determine the relative influence of lifestyle (i.e., engagement in complex mental activities) at different life periods on older adults' brain health.

Methods: This observational study included community-dwelling cognitively unimpaired seniors (older than 65 years) from the Age-Well randomized controlled trial (Caen, France). All participants completed at baseline the Lifetime of Experiences Questionnaire, assessing engagement in complex mental activities during young adulthood (13-30 years: LEQ-young), midlife (30-65 years: LEQ-midlife), and late-life (older than 65 years: LEQ-late). LEQ scores were divided into specific and non-specific activities. Multiple regressions were conducted including LEQ scores at the 3 life periods (same model) to predict gray matter volume (GMv; structural-MRI), glucose metabolism (fluorodeoxyglucose-PET), perfusion (early-Florbetapir-PET), or amyloid burden (late-Florbetapir-PET), both in AD-signature regions and voxel-wise (significance for voxel-wise analyses: p < 0.005_uncorrected, k > 100). Correlations between LEQ and neuroimaging outcomes were then compared between (1) life periods and (2) specific and non-specific activities. Analyses were controlled for age and sex.

Results: In 135 older adults (mean age = 69.3 years; women = 61.5%), no associations were found within AD-signature regions (all p > 0.25). Voxel-wise analyses revealed no association between LEQ-young and neuroimaging. LEQ-midlife showed stronger voxel-wise associations than the other periods with GMv, notably in the anterior cingulate cortex, and with amyloid burden in the precuneus. These correlations were stronger for the LEQ-midlife specific (i.e., occupation) than the non-specific subscore (GMv: z = 3.25, p < 0.001, 95% CI [0.1292-0.5135]; amyloid: z = -1.88, p < 0.05, 95% CI [-0.3810 to -0.0113]). LEQ-late showed stronger voxel-wise associations than the other periods with perfusion and glucose metabolism in medial frontal regions. The correlation of perfusion with LEQ-late was stronger for non-specific than specific subscore (z = 2.88, p < 0.01, 95% CI [0.0894-0.4606]).

Discussion: Lifestyle at different life periods may have complementary benefits on brain health in regions related to reserve/resilience in aging. While past (midlife) engagement could promote resistance against structural/pathologic alterations, current (late-life) engagement could enhance cognitive reserve. Future larger longitudinal studies should explore mechanisms by which lifestyle promotes reserve.

Figure 1. Design Overview

(A) In this cross-sectional design, 135 cognitively unimpaired older adults from the Age-Well cohort completed the Lifetime of Experiences Questionnaire (LEQ), allowing to retrospectively evaluate their lifestyle during young adulthood, midlife, and late-life. A score was obtained for each life period and reflected both activities that were specific and non-specific to this period. Multimodal neuroimaging was also obtained including structural MRI (gray matter volume), FDG-PET (cerebral glucose metabolism), and Florbetapir-PET scans (early frames/cerebral perfusion and late frames/amyloid burden). (B) Multiple linear regressions were performed including LEQ scores for the 3 life periods as predictors of AD-signature regions (1 model per imaging modality; hippocampal volume, glucose metabolism, or cerebral perfusion in temporoparietal areas, neocortical amyloid burden), controlling for age and sex. (C) Whole brain analyses: (1) Voxel-wise multiple linear regressions were conducted to assess the association of LEQ score at each life period (in the same model) with neuroimaging (1 model per neuroimaging modality). Results were considered significant at p < 0.005 (uncorrected), k > 100 (Figure 1C1, green regions). (2) When an association was found between lifestyle at a given life period and imaging, we conducted additional analyses within an inclusive mask including the regions highlighted to further understand the specificity of this association and the type of activity that could drive this effect. More specifically, within the regions significant at p < 0.05 (used as an inclusive mask), we assessed voxel-wise whether the association with this life period was stronger than the association with the 2 other life periods. The average volume/SUVR of the regions found to be more related to this life period than to the 2 others (Figure 1C2, orange regions) were extracted to (3) conduct a 2-by-2 correlation coefficient comparisons between the 3 life periods and (4) assess the relative influence of specific and non-specific activities conducting comparisons of the correlation coefficient for specific vs non-specific subscores. LEQ = Lifetime of Experiences Questionnaire; SUVR = Standardized Uptake Value Ratio.

Figure 2. Associations Between Midlife and Late-Life LEQ Scores and Brain Integrity

Left panel: (1) Voxel-wise associations between LEQ-midlife and gray matter volume (A) and amyloid burden (B), controlling for age, sex, and LEQ-young and LEQ-late scores. (2) Regions showing greater association with LEQ-midlife compared with the other 2 periods for gray matter volume (E) and amyloid burden (F) and associated graphic representations. (3) Associations between LEQ-midlife specific (dark blue) or non-specific (light blue) subscores with gray matter volume (I) and amyloid burden (J) in regions previously highlighted (panels E and F). Statistical values were obtained from multiple regression models including specific and non-specific subscores to predict the average volume/SUVR within the regions previously evidenced, controlling for age and sex. Right panel: (1) voxel-wise associations between LEQ-late and cerebral perfusion (G) and glucose metabolism (D), controlling for age, sex, and LEQ-young and LEQ-midlife scores. (2) Regions showing greater association with LEQ-late compared with the other 2 periods for cerebral perfusion (G) and glucose metabolism (H) and associated graphic representations. (3) Associations between LEQ-late-life specific (light pink) or non-specific (dark pink) subscores with glucose metabolism (K) and cerebral perfusion (L) in regions previously highlighted (panels G and H). Statistical values were obtained from multiple regression models including specific and non-specific subscores to predict the average volume/SUVR within the regions previously evidenced, controlling for age and sex. Voxel-wise results were considered significant at p < 0.005 uncorrected, k > 100. Linear regression lines and standard error (SE) bands corresponding to the 95% CI are represented. p < 0.05*, p < 0.01**, p < 0.001***. LEQ = Lifetime of Experiences Questionnaire; SUVR = Standardized Uptake Value Ratio.