Low DHEAS levels are associated with depressive symptoms in elderly Chinese men: results from a large study

Introduction

Depression is a prevalent disorder in the elderly and causes significant morbidity and mortality.¹ There is epidemiological evidence of an association between depression or clinically relevant depressive symptoms and low levels of free testosterone in elderly men.² There are theoretical grounds to support the hypothesis that low testosterone levels can lead to depression. First, oestrogens and androgens affect mood and behaviour, and studies have demonstrated relationships between the levels of sex steroid hormones and neurotransmitters, including serotonin, both in humans and other animals.³ Second, the well-established gender differences in the prevalence of depression suggest that testosterone may play a protective role in the pathogenesis of depression in men, resulting in a lower prevalence of depression in men than in women, although other social factors, such as differences in socialisation and help-seeking behaviours between the two sexes, may also play important roles.⁴ Furthermore, the high prevalence of depression in elderly men,⁵ which coincides with a physiological drop in testosterone levels, suggests that there may be a relationship between low testosterone levels and depression in men, although more research is needed to confirm whether this relationship exists.⁶

Several large studies have explored the relationship between depression and low levels of testosterone. The Rancho Bernardo study⁷ demonstrated the association between bioavailable testosterone and depressive symptoms using a cross-sectional design. Morsink et al.⁸ showed that dehydroepiandrosterone sulphate (DHEAS) levels are inversely associated with depressive symptoms in both elderly men and women. Moreover, men in the lowest testosterone quartile reported significantly more depressive symptoms than men in the other total testosterone quartiles. In the Health in Men study, Almeida et al.⁹ showed that low free testosterone concentrations are associated with depressive symptoms in older men.

In addition to cross-sectional studies, Shores et al.¹⁰ demonstrated an association between baseline testosterone levels and incident depression in elderly Americans. Recently, using data from the Longitudinal Aging Study in Amsterdam, Joshi et al.⁴ demonstrated an association between testosterone levels and depressive symptoms.

However, despite the findings from these studies, not all studies have shown statistically significant results.^{11, 12} Moreover, to our knowledge, no large epidemiological studies have been conducted outside the United States or Europe or in Asians. The present study investigated the association between sex steroid hormone levels and depressive symptoms cross-sectionally in a cohort of elderly Chinese men. We hypothesized that low testosterone levels are associated with depressive symptoms in elderly Chinese men.

Materials and methods

Results

The basic sociodemographic information of the sample of 1147 elderly men for which sex steroid hormone data are available is shown in Table 1. The median (interguartile range) of the total testosterone level in this sample was 5.18 ng ml⁻¹ (2.44 ng ml⁻¹). About 8.0% of this sample had clinically relevant depressive symptoms.

Table 2 shows the results from the linear regression analyses that assessed the association between the levels of sex steroid hormones and depressive symptoms before and after adjustments for age, education, BMI, smoking and chronic diseases. There was a significant inverse association between DHEA level and GDS score (unadjusted β=−0.134, P=0.003), which was attenuated after adjustment for the factors noted above (β=−0.110, P=0.015). Similarly, there was a significant inverse association between DHEAS level and GDS score (unadjusted β=−0.090, P=0.018), which was also attenuated after adjustment (β=−0.074, P=0.055). However, total testosterone, free testosterone, SHBG and oestradiol levels and the free androgen index were not significantly associated with GDS scores.

Similar to the method used by Morsink et al.,⁸ we divided the sex steroid hormone levels into quartiles to examine whether there was a ‘threshold effect' or a ‘gradient effect' in the association between sex steroid hormone levels and depressive symptoms. Depressive symptom scores among patients in the lowest quartile of any sex steroid hormone levels collected in this study were not significantly different from those for patients in higher quartiles of sex steroid hormone levels or the free androgen index before or after adjusting for potential confounding variables (P>0.05; Table 3). Figure 1 shows the relationships between various sex hormone levels and GDS scores. There were no significant threshold effects in the association between sex steroid hormone levels and depressive symptoms.

Discussion

This study explored the association between depressive symptoms and the total testosterone, free testosterone, DHEA, DHEAS, oestradiol, free testosterone and SHBG levels and the free androgen index in a sample of community-dwelling elderly Chinese men. To our knowledge, we are the first to explore these relationships in Asians using data from a large prospective cohort. In contrast to previous findings from studies of large cohorts,^{4, 8} we found a significant inverse association only between depressive symptoms (defined by GDS score) and DHEA or DHEAS, and we did not find an association between depressive symptoms and the serum levels of total testosterone, oestradiol, free testosterone or the free androgen index in these Chinese elderly men.

Our finding of an inverse association between DHEA or DHEAS level and depressive symptoms confirms the relationship that has previously been shown in Caucasians and African-Americans.^{8, 19, 20} We measured both DHEA and DHEAS levels, because DHEA is the precursor of DHEAS. DHEAS is the sulphated version of DHEA with a conversion that is reversibly catalysed by sulphotransferase (SULT2A1). Indeed, it has been suggested that among all of the sex steroid hormones that are associated with depressive symptoms in the elderly, findings on the association between DHEAS/DHEA levels and depressive symptoms are the most consistent.¹⁹

DHEA has antidepressant effects through the modulation of neurotransmitter systems in the brain in animal studies.²¹ Overall, DHEAS increases the excitability of the brain and has effects similar to antidepressants on the neurotransmitter system, resulting in an improvement in mood. Other possibilities for the association between low DHEAS level and depressive symptoms include the possibility that depressive symptoms could cause lower DHEAS levels or the possibility that the relationship is confounded by the presence of medical comorbidity or a lack of physical activity or malnutrition, which is common in the elderly and could have affected both the DHEAS levels and the depressive symptoms. Previous research has also shown that the presence of multiple chronic conditions lowers DHEAS levels and increases depressive symptoms.²²

In further support of the idea that DHEA/DHEAS level has a role in depression, a recent randomized controlled trial²³ in the United Kingdom showed that DHEA improved subjective mood and decreased the evening cortisol concentration, which is elevated in people with depression.²⁴

The reason for the lack of an association between the free androgen index and depressive symptoms is not known, but this result could result from the indirect measurement of the free androgen index in our study. We calculated the free androgen index using a formula¹⁶ and did not directly measure free testosterone or bioavailable testosterone. Moreover, no measurement of albumin was taken in our study, and this measurement is needed to calculate free testosterone using the method of Vermeulen et al.,²⁵ which was also used in a recent study by Joshi et al.⁴

We do not know why our study did not find an association between total testosterone levels and depressive symptoms. The value of the mean level of total testosterone in our study was comparable to or higher than the levels of total testosterone from previous studies (e.g., mean total testosterone=5.47 ng l⁻¹ or 18.96 mmol l⁻¹ in our study, 3.82 ng l⁻¹ in Morsink et al.'s study and 14.61 mmol l⁻¹ in Joshi et al.'s study) that demonstrated this association.^{4, 8} The presence of very few hypogonadal men in our study may explain the lack of association. However, the relationship between total testosterone level and depressive symptoms is inconsistent in the literature, with some studies reporting a significant association⁸ and others reporting no association.^{7, 26, 27}

Our present study has several limitations. First, we defined depression as scoring higher than the standard cutoff score with the use of a validated Chinese screening instrument for detecting depressive symptoms in elderly Chinese; there may have been an underestimation of depression if these Chinese elderly did not present with psychological symptoms or complaints. It should be noted that the prevalence of depression in our sample was lower than the prevalence that has been reported previously in Hong Kong in elderly Chinese,²⁸ probably as a result of recruiting physiologically well-functioning elderly men in our study. Second, this was simply a cross-sectional study, and a causal relationship between sex steroid hormone levels and depressive symptoms could not be derived. Third, we did not make direct measurements of free testosterone and albumin levels, which may have affected the accuracy of our calculations of the free or bioavailable testosterone levels. Fourth, we only crudely assessed the presence of chronic comorbidities, and medical comorbidities may have affected the levels of sex hormones. Fifth, our sample of men was not recruited by random sampling, and 511 participants were excluded because serum samples were not available for hormone measurement; therefore, it is possible that there is selection bias in this study.

We found a significant association between the DHEA level and depressive symptoms in elderly Chinese. Further longitudinal studies in various ethnic groups are required to further evaluate this relationship.

Author contributions

All authors contributed equally to this work. SYSW, TK, PCL, and JW designed and performed experiments; SYSW, JCL, TK, CO, and LV analyzed data; SYSW, JCL, TK, and PCL wrote the manuscript.

All co-authors have no financial conflicts or other conflicts of interest in this area.