Sex-specific associations of serum cortisol with brain biomarkers of Alzheimer's risk

Abstract

Emerging evidence implicates chronic psychological stress as a risk factor for Alzheimer's disease (AD). Herein, we examined the relationships between serum cortisol and multimodality brain AD biomarkers in 277 cognitively normal midlife individuals at risk for AD. Overall, higher cortisol was associated with lower total brain volume, lower glucose metabolism (CMRglc) in frontal cortex, and higher β-amyloid (Aβ) load in AD-vulnerable regions; and marginally associated with phosphocreatine to ATP ratios (PCr/ATP) in precuneus and parietal regions. Sex-specific modification effects were noted: in women, cortisol exhibited stronger associations with Aβ load and frontal CMRglc, the latter being more pronounced postmenopause. In men, cortisol exhibited stronger associations with gray matter volume and PCr/ATP measures. Higher cortisol was associated with poorer delayed memory in men but not in women. Results were adjusted for age, Apolipoprotein E (APOE) epsilon 4 status, midlife health factors, and hormone therapy use. These results suggest sex-specific neurophysiological responses to stress, and support a role for stress reduction in AD prevention.

Figure 1

Associations of serum cortisol with brain biomarkers. Scatterplots showing associations between cortisol and brain biomarkers: (A) total brain volume. (B) Phosphocreatine to ATP ratio (PCr/ATP) in precuneus and parietal regions (Factor 2 from principal component analysis of ³¹P-MRS data). (C) Cerebral metabolic rates of glucose (CMRglc) in middle frontal regions (Factor 3 from principal component analysis of ¹⁸F-FDG PET data). (D) Amyloid-beta (Aβ) load in AD-mask. Analyses are multivariable-adjusted by age, APOE-4 status, midlife health variables, and modality specific confounders. Cortisol measures underwent a standardized asinh(x) transformation prior to analysis. Standardized values are displayed in the graphs.

Figure 2

Associations of serum cortisol with brain biomarkers by sex. Scatterplots showing sex effects on the associations of cortisol with brain biomarkers: (A) Gray matter volume in precuneus, inferior and superior parietal regions (Factor 2 from principal component analysis of MRI data). (B) Phosphocreatine to ATP ratio (PCr/ATP) in frontal, temporal and posterior cingulate regions (Factor 1 from principal component analysis of ³¹P-MRS data). (C) Cerebral metabolic rates of glucose (CMRglc) in middle frontal regions (Factor 3 from principal component analysis of ¹⁸F-FDG data). (D) Amyloid-beta (Aβ) load in AD-mask. Analyses are multivariable-adjusted by age, APOE-4 status, midlife health variables, and modality specific confounders. Cortisol measures underwent a standardized asinh(x) transformation prior to analysis. Standardized values are displayed in the graphs. Men = blue; women = red. In (A–C), cortisol-biomarker associations demonstrate a differential effect of cortisol on these outcomes by sex, as evidence by significant interaction terms. Conversely, in (D), no sex-based modification effects were detected in the associations between cortisol and Aβ. Corresponding statistics are reported in Table 3.

Figure 3

Associations between serum cortisol and regional biomarker measures on a region-by-region basis. Heatmaps showing associations between cortisol levels and (A) regional brain volumes, (B) PCr/ATP, (C) glucose metabolism, and (D) Aβ load in individual regions of interest. Multivariable-adjusted regression coefficients are displayed on a red-to-blue color-coded scale, with red indicating negative associations and blue indicating positive associations. Corresponding statistics are found in Supplementary Table 2.