How does allostatic load relate to cognitive ability, depressive symptoms and other measures of physiological weathering in older age?

Abstract

Allostasis is the process through which the human body aims to achieve homeostasis - long-term physiological stability in each of its systems - through short-term changes. If someone experiences chronic stress or repeated bouts of stress, their allostatic systems can experience wear and tear and enter a state of allostatic load. The thesis first considered how allostatic load is operationalised. It outlined problems with modelling allostatic load as a latent factor, including concerns based on psychometric and substantive studies. This chapter also posited desirable qualities of allostatic load operationalisations. The subsequent substantive chapters were empirical studies that explored the role of allostatic load in successful ageing. In the first three of four studies, we fitted latent growth curve models to data from Waves 1-4 of the Lothian Birth Cohort 1936 (n [M age] at Waves 1-4 = 1,091 [69.5 years], 866 [72.5 years], 697 [76.3 years], 550 [79.3 years]). Additional models tested the role of non-random dropout. The first of these studies investigated how allostatic load relates to four other measures of physiological weathering: [Klemera-Doubal] biological age; extrinsic epigenetic age; intrinsic epigenetic age; and telomere length. Allostatic load was most strongly related to biological age, which is an age-linked measure calculated using some of the same biomarkers as allostatic load. At Wave 1, higher allostatic load was related to older biological age, with a moderate effect size (r [SE] = .28 [.05], p < .001). Greater increase over time in allostatic load was strongly related to accelerated biological ageing (r [SE] = .53 [.15], p < .001). However, most correlations between physiological weathering measures were small, which suggested that they index different aspects of physiological weathering. The next study found that at Wave 1, higher allostatic load was related to lower general cognitive ability, with a small effect size (r [SE] = -.13 [.04], p = .004). However, after controlling for the variance related to participant dropout, the relationship was attenuated by around a third (r [SE] = -.08 [.04], p = .08). The third Lothian Birth Cohort 1936 study found that at Wave 1, higher allostatic load was related to having more depressive symptoms, with a small-to-moderate effect size (r [SE] = .19 [.05], p < .001). These studies also went beyond previous research by testing relationships of associated changes over time. These tests suggested changes in allostatic load are not strongly related to changes in cognitive ability and depressive symptoms. The last empirical study in this thesis used data from the Scotland-based 36-Day-Sample (n range for models = 280-332; M age at cognitive testing = 78.2 years). In this study, producing more cortisol during the first 45 minutes after waking was related to higher general cognitive ability (β [SE] = 0.14 [0.07], Bayes factor10 [BF10] = 5.23), but the total cortisol produced during the waking day was not related to general cognitive ability (BF10 < 1/3). We also found that in carriers of the e4 allele of the apolipoprotein E gene, some measures of cortisol dynamics were more negatively associated with residualised immediate recall scores and some measures of total cortisol production were more negatively associated with residualised Raven’s matrices scores. Additionally, this study found that depressive symptoms were related to lower general cognitive ability (β [SE] = -0.15 [0.08], BF10 = 3.47), but the relationship was not more negative in e4 allele carriers (β [SE] = 0.08 [0.07], BF10 = 0.21). The final chapter considered the results in a wider context, as well as possible future directions. We discussed the limitations of our allostatic load measure, which did not include any hypothalamic–pituitary–adrenal axis biomarkers. We also considered the constraints on generalisability of all allostatic load findings given the heterogeneity of the operationalisations used. We questioned the clinical utility of allostatic load measurement for cognitive decline and depressive illness, given the small effect sizes we obtained for the relevant relationships. Finally, we concluded that future research will benefit from using larger samples and teasing apart how these relationships vary based on the allostatic load biomarkers and operationalisations used.