We report independent associations of steep decline and extreme variability in DHEAS levels with mortality, whereas the absolute DHEAS level was not associated with mortality in a population-based cohort of men and women aged 65 years or older. This finding suggests that the absolute level of DHEAS, which is highly genetically influenced (14
), is less clinically relevant than an individual’s ability to maintain that set point. Furthermore, it provides evidence to support the premise that the optimal way to age is both slowly (no or slow rate of decline) and uniformly (small standard deviation in the rate of decline) (16
Our study confirms the previous cross-sectional and longitudinal data showing an overall decline in DHEAS levels with increasing age and shows a similar magnitude of decline (3
). However, our data demonstrate considerable heterogeneity among individuals in rates of change of DHEAS and, in a subset, substantial instability over time. This has implications for DHEA supplementation because our data suggest that there are significant endogenous fluctuations in DHEAS levels. It is also of particular concern because DHEA has not been subjected to the same rigor of clinical testing as other hormonal preparations due to its status as an over-the-counter supplement. The largest clinical trials in older people, the longest of which was 2 years in duration, do not support health benefits to DHEA supplementation (17
). The long-term health risks or benefits of DHEA supplementation are unknown and particularly not on a background of varying endogenous levels.
Studies in humans, rather than in animal models, are required to understand the physiological roles of DHEAS because DHEAS is not produced in mice or rats but only in higher order primates and humans (20
). Observational studies conducted in older populations have generally reported lower mortality in men with higher DHEAS levels and no association or a U-shaped relationship in women (13
), whereas we found no association of the absolute level of DHEAS with mortality in either sex. Roth and coworkers (22
) have shown that calorically restricted monkeys have higher DHEAS levels and live longer than monkeys on an unrestricted diet, supporting a role for DHEAS as a potential biomarker of aging. However, DHEAS levels were not affected by a 6-month study of caloric restriction in younger people (23
Although the adverse DHEAS trajectory patterns precede mortality, we cannot determine whether the changes in DHEAS levels contribute to mortality or are markers of the underlying health status of the individual. DHEAS may have biologic importance as a precursor to estrogens and androgens and as neurotransmitter in the brain (24
). Nevertheless, we favor the interpretation that the DHEAS trajectory reflects the underlying health status of the individual. Steep decline or extreme variability in other health parameters, such as blood pressure (25
), has also been associated with an increase in mortality rate. To our knowledge, this type of trajectory analysis has not been previously extended to examination of a putative biomarker of aging.
We also cannot determine whether the physiological genesis of steep decline and extreme variability in DHEAS levels is the same. Large variability in DHEAS levels suggests a complete inability of the individual to make adjustments to cope with internal changes and external stressors. Whereas minor changes in DHEAS levels could represent appropriate restoration of equilibrium, constantly fluctuating DHEAS levels are likely to be measuring a lack of adaptive capacity characteristic of frailty (26
The sex difference in the strength of association between trajectory pattern and mortality is intriguing. If DHEAS played a causal role, it would be expected to be more influential in women than in men. This is due to lower levels of endogenous androgens and estrogens and increased conversion of DHEAS to DHEA in postmenopausal women (27
), which should result in a decreased ability to buffer changes in DHEAS level. Fried and associates (29
) have previously reported a mortality difference between men and women enrolled in CHS. Adjustment for a large battery of traditionally measured predictors of mortality significantly diminished the association between age and mortality, demonstrating that these factors explained much of the effect of age, but they did not attenuate the twofold higher mortality of men compared with women. These findings suggest that there must be other factors, not traditionally measured in epidemiological studies of older people, which contribute to the greater longevity of women. Our data suggest a relative impairment of homeostasis and increased susceptibility in these older men, or a physiological resilience in women, in the face of the factors that lead to adverse DHEAS trajectory patterns.
A major strength of our study is the use of a large, multicenter, population-based cohort comprised older men and women who had serial examinations and phlebotomy, with up to 17 years of follow-up. However, our results may not be generalizable to younger populations, as we have no data on individuals who are younger than 65 years. Also, because we are using observational data, we cannot determine whether the hormonal trajectory is a sensitive marker of physiological dysregulation or whether alterations in the dynamics of DHEAS mediate physiological effects. Our data support thresholds of slope and variability in relation to mortality, and therefore, the DHEAS measures were modeled using cutpoints. However, the specific threshold values identified in this investigation represent estimates from our data and should be confirmed in other cohorts.
In summary, our data suggest that trajectories of DHEAS provide more biologic information about an older individual than the DHEAS level itself. This research is a departure from disease-based models of aging and instead focuses on age-related physiological susceptibility and resilience, which currently lack objective measures. In fact, the point estimates of mortality risk for steep decline and extreme variability were larger than those of any of the five baseline diseases we included as covariates, suggesting that the trajectory parameters were stronger predictors of mortality than established predictors such as CVD or diabetes mellitus. Although we are the first, to our knowledge, to examine the relationship between the pattern of a hormonal measure, irrespective of its absolute level, and mortality, it is likely that the dysregulation in DHEAS levels is also paralleled in other physiological systems. Trajectory analysis provides the opportunity to study the biology of aging at the population level and potentially offers insight into the dynamic process of maintaining homeostasis.