Sleep duration and testosterone levels in community older men: results from the West China Health and Aging Trend study.

Abstract

To the Editor: Testosterone is a fundamental male sex hormone produced by the testicular Leydig cells. Testosterone levels are affected by age, peaking in the 20s and 30s and gradually declining thereafter.[1] Low testosterone level may predispose men, especially older men, to a poor prognosis or death in coronavirus disease 2019 (COVID-19).[2] Therefore, testosterone levels have a significant impact on the health status of older men. The identification of modifiable non-drug factors affecting testosterone levels in older men is important for improving their health. Chronic sleep changes are common in older adults. Testosterone production is dependent on sleep, and plasma testosterone levels vary with a circadian rhythm. However, there is a lack of large-scale population-based studies to explore the effect of sleep duration on testosterone levels in older men. The present study is a cross-sectional analysis based on the West China Health and Aging Trend (WeCHAT) study. WeCHAT is an ongoing prospective multicenter cohort study of community-dwelling Chinese older people and initiated in 2018, and it aims to understand the health status and influencing factors of the older adults in Western China. A total of 7538 people from 18 ethnic groups in four western provinces of China (Sichuan, Yunnan, Guizhou, and Xinjiang) participated in the study. The study was registered with a clinical trials registry (No. ChiCTR1800018895), the research protocol was approved by our institutional Ethics Committee (West China Hospital, China, No. 2017-445), and written informed consent was obtained from all participants. Detailed inclusion and exclusion criteria are shown in Supplementary Figure 1, https://links.lww.com/CM9/B325. We collected the required data through an electronic questionnaire and in-person interviews. Sleep duration was self-reported by participants in response to the following question: “During the past month, how many sleep time did you get at night (average hours for one night)?” Participants were also required to provide a blood sample for morning testing of the total testosterone (TT) level. Detailed processing and quality control steps are shown in Supplementary File, https://links.lww.com/CM9/B325. Baseline demographic information comprised body mass index (BMI), age, education level, and marital status. Smoking was evaluated by enquiring about smoking frequency, number of cigarettes, and whether the participant had attempted to quit. The frequency of drinking alcohol was used to assess drinking habit. This study used the modified Fried frailty phenotype to assess physical frailty. Nutritional status was measured using the Mini Nutritional Assessment Short Form, with scores of <8 points indicating normal nutritional status. The cognitive performances of the participants were assessed using the Chinese version of the 10-point Short Portable Mental Status Questionnaire. Scores of ≥5, suggesting moderate to severe cognitive function impairment, were categorized as cognitive impairment in the present study. Chronic diseases status, including diabetes, hypertension, chronic obstructive pulmonary disease, coronary heart disease, and stroke, was collected through face-to-face interview by medical staff. To address the non-linear association between sleep duration and TT and identify appropriate cut-off values of sleep duration, a generalized additive model and smooth curve fitting (penalized spline method) were used. The inflection point was calculated using a recursive algorithm, and then we constructed a two-phase piecewise linear regression model on either side of the inflection point to identify possible significant threshold effects. We determined the best fit model based on the P value of the log-likelihood ratio test. We identified the short- and long-sleep groups based on the inflection point. Differences in the patients’ baseline characteristics between the short-sleep and long-sleep groups were compared. Categorical variables were described as frequencies (percentage), and continuous variables were described as mean ± standard deviation or median (the 25th–75th percentile). Variables were compared using the chi-squared test, Fisher's exact test, Wilcoxon rank sum test, or Student's t-test as appropriate. All statistical tests were two-sided, and a P value of < 0.05 was considered as statistically significant. We used multivariable linear regression to analyze the possible association between sleep duration and TT level and constructed two models to test the stability of this relationship. Model I was adjusted for BMI, age, ethnic group, education level, marital status, smoking history, and alcohol use. Model II was adjusted for BMI, age, ethnic group, education level, marital status, smoking history, alcohol use, cognitive performance, nutritional status, physical frailty, and chronic diseases. Besides, subgroup analyses were performed by BMI. Using the likelihood ratio test, the P value for the interaction of sleep duration and BMI on TT level was also assessed. To verify the stability of our findings, we conducted a series of additional sensitivity analyses. The detailed methods and results are shown in Supplementary File, https://links.lww.com/CM9/B325. R statistical software (http://www.R-project.org, The R Foundation, Vienna, Austria) was used to perform the analyses above. We enrolled 1628 men aged ≥60 years in the current study. The mean age was 68.5 ± 5.9 years, mean BMI was 24.59 ± 3.72 kg/m2, and the median TT was 428.00 (314.00–554.25) ng/dL [Supplementary Table 1, https://links.lww.com/CM9/B325]. Results from smooth curve fitting and generalized additive models suggested that the association between sleep duration and testosterone levels in older men was non-linear when adjusting for the covariates (including age, ethnic group, education level, marital status, smoking history, alcohol use, nutrition status, and physical frailty) [Supplementary Figure 2, https://links.lww.com/CM9/B325]. Moreover, we found that the inflection point was 9.5 h by two-phage piecewise linear regression and recursive algorithm (P for log-likelihood ratio test: <0.001). There was a significant threshold effect, with insignificant effects on the left of the inflection point and a significant association of longer sleep duration with higher TT levels on the right of the inflection point (β = 640.92, 95% confidence interval [CI] = 380.42–901.42; P < 0.001). Based on the cut-off of 9.5 h, the enrolled patients were divided into a short-sleep group (<9.5 h, n = 1477) and a long-sleep group (≥9.5 h, n = 151). Supplementary Table 1, https://links.lww.com/CM9/B325 showed the detailed characteristics of the participants based on the different sleep durations. The results of the linear regression analyses revealed that long sleep duration was significantly associated with higher TT level (β = 436.48, 95% CI = 181.81–691.15; P < 0.001), after adjusting for BMI, age, ethnic group, education level, marital status, smoking history, and alcohol use status. After adding additional adjustment variables of cognitive performance, nutritional status, physical frailty, and the number of chronic diseases, the association between sleep duration and TT level remained stable, and the long sleep was significantly associated with obviously higher TT levels compared with short sleep group (β = 617.84, 95% CI = 227.45–1008.23; P = 0.002). Then, sleep duration was included in the analysis as a continuous variable, and the results of the linear regression analysis remained unchanged after adjusting for the covariates, showing that sleep duration was significantly associated with increased testosterone levels (β = 75.04, 95% CI = 8.21–141.86; P = 0.028). We then performed subgroup analyses by BMI, which also revealed an association between sleep duration and TT level [Supplementary Table 2, https://links.lww.com/CM9/B325]. There was no statistically significant association between sleep duration and TT level in the BMI ≥ 25 kg/m2 group. By contrast, longer sleep duration in the BMI < 25 kg/m2 group was significantly associated with increased TT level (β = 134.98, 95% CI = 7.90–262.06; P = 0.038). A significant interaction regarding BMI with the sleep duration and TT level was identified (P for the interaction = 0.010). To verify the stability of our findings, we conducted a series of additional sensitivity analyses. The detailed methods and results are shown in Supplementary Files and Supplementary Tables 3–8, https://links.lww.com/CM9/B325. The present study demonstrated that longer sleep duration was associated with a higher testosterone level among older men with BMI < 25 kg/m2. The physiological mechanisms behind the association between sleep and testosterone levels have not been determined. The hypothalamic-pituitary-gonadal axis is the primary mechanism regulating testosterone production and secretion. Basic research based on animal models has revealed that short sleep duration may not negatively affect gonadotropin-releasing hormone secretion in the hypothalamus, but negatively affect the pituitary gland, especially luteinizing hormone secretion, resulting in a decrease of testosterone production and secretion. On the other hand, sleep promotes inflammatory homeostasis by affecting multiple inflammatory mediators such as cytokines, which also affects the testosterone level. Chronic sleep deprivation (short sleep duration, sleep disturbance) contributes to chronic systemic low-grade inflammation and is associated with various diseases. Inflammatory factors, including IL-1, IL-6, IL-17, and tumor necrosis factor, have been reported to affect testosterone levels by activating inflammation. Besides, the interstitial macrophages located near the Leydig cells also affect testosterone levels by producing reactive oxygen species. This study has several strengths and limitations. Previously reported studies, which evaluated sleep duration and testosterone levels, were mainly limited to laboratory-controlled conditions with small cohorts. Given the background of aging society, and the changes of sleep and testosterone levels in the elderly, our study is one of the few studies focusing on sleep duration and testosterone levels in older men and using multicenter data from a large prospective cohort. The establishment of this cohort allowed comprehensive data collection and standardized laboratory measurements on a representative sample of older men in Western China. Of course, there are also some limitations. First, owing to the cross-sectional design, we could not draw a causal relationship between increased sleep duration and increased serum testosterone levels. Second, as in other population-based studies of sleep duration,[3,4] the sleep duration information was self-reported by participants, which may lead to recall bias. Third, owing to the limitations of the WeCHAT study database, the participants’ values of sex hormone-binding globulin and free testosterone were unavailable. Although low TT levels have been shown to be a predictor for a variety of men's health problems,[5] more precise indicators, such as free testosterone, can better reflect hormone activity. Therefore, we will conduct more comprehensive data collection in future research. Overall, our study revealed the association between sleep duration and testosterone levels, providing evidence for improving androgen levels through lifestyle modifications. We also provided population-based research evidence for future interventional clinical trials. In conclusion, for older men with normal BMI, achieving >9.5 h of sleep was associated with higher testosterone levels. Our findings can help aging management in older men. In men with low androgen concentrations, good sleep habits and ensuring >9.5 h of sleep may be an effective non-pharmacological intervention to improve androgen concentrations. Funding This work was supported by the National Key Research and Development Program of China (No. 2017YFC0908003), National Natural Science Foundation of China (Nos 81902578 and 81974098), China Post-doctoral Science Foundation (No. 2017M612971), Post-doctoral Science Research Foundation of Sichuan University (No. 2020SCU12041), Post-doctoral Research Project, West China Hospital, Sichuan University (No. 2018HXBH085), and National Clinical Research Center for Geriatrics, West China Hospital, Sichuan University (No. Z2018C01). Conflicts of interest None.