If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Department of Innovative Research and Education for Clinicians and Trainees (DiRECT), Fukushima Medical University Hospital, Fukushima, JapanDepartment of Healthcare Epidemiology, School of Public Health in the Graduate School of Medicine, Kyoto University, Kyoto, Japan
Address correspondence to Shunichi Fukuhara, MD, DMSc, MACP, Department of Healthcare Epidemiology, School of Public Health in the Graduate School of Medicine, Kyoto University, Yoshida Konoe-cho, Sakyo-ku, Kyoto 606-8501, Japan.
Department of Healthcare Epidemiology, School of Public Health in the Graduate School of Medicine, Kyoto University, Kyoto, JapanCenter for Innovative Research for Communities and Clinical Excellence (CIRC2LE), Fukushima Medical University, Fukushima, Japan
Findings from several experimental studies in animals have suggested a protective action of testosterone on kidney function, but hard evidence for such an association in humans is scarce. We examined the association between testosterone levels and kidney function among adult men living in super-aged communities.
Design, setting, participants, and measurements
We conducted cross-sectional study involving residents aged 40–80 years who participated in annual health check-ups in 2 communities. A total of 1031 men were recruited in 2010. Main exposure was salivary testosterone (sT) levels measured using an enzyme-linked immunosorbent assay. Main outcome was estimated glomerular filtration rate (eGFR) determined by age, gender, and serum creatinine levels.
Results
For the 848 participants analyzed, median age and eGFR were 69 years and 69.1 mL/min/1.73 m2, respectively. On comparison of 90th-percentile sT levels with lower levels, our general linear model with restricted cubic splines showed that lower sT levels were associated with decreased eGFR after adjustment for sociodemographic characteristics, comorbidities, and blood pressure. For example, fifth percentile sT was associated with decreased eGFR, with a difference in eGFR [−3.43 mL/min/1.73 m2 (95% confidence interval, CI −6.02 to −0.84)] comparable in magnitude to the reduction in eGFR observed for a 6-year increase in age in our population. The association between low testosterone levels and decreased eGFR remained similar even when analyses were restricted to participants aged over 60 years (734 participants, median age 71 years).
Conclusions
Results from our study indicated that having low testosterone levels was independently associated with reduced eGFR in adult men. Our finding of this association between low testosterone levels and reduced kidney function needs to be corroborated among persons with chronic kidney disease or in a longitudinal study.
Reduced kidney function is a growing public health problem, and increasing evidence suggests that chronic kidney disease is equivalent to diabetes as a coronary risk factor.
While a number of causes have been implicated in reduced kidney function, age-related decline in function via cellular senescence is recognized as a major contributor.
Separately, age-related decline in testosterone levels in men is relatively common, with approximately 30% of men aged 40–79 years showing some degree of testosterone deficiency.
Mounting evidence suggests that low testosterone levels are associated with increased cardiovascular morbidity and mortality in men, both with and without chronic kidney disease.
Endogenous testosterone and mortality due to all causes, cardiovascular disease, and cancer in men: European Prospective Investigation into Cancer in Norfolk (EPIC-Norfolk) Prospective Population Study.
High serum testosterone is associated with reduced risk of cardiovascular events in elderly men: The MrOS (Osteoporotic Fractures in Men) Study in Sweden.
However, although several experimental and clinical studies have suggested that testosterone has a protective role on kidney function, findings regarding the influence of testosterone on kidney function remain conflicting, particularly in adult men.
Experimental studies suggest that testosterone may protect against ischemia-reperfusion-induced acute kidney injury, possibly via attenuation of inflammation and mediators of ischemia.
Indeed, in line with these findings, androgen deprivation therapy has been associated with increased risk of acute kidney injury in patents with prostate cancer,
However, observations in general population studies regarding the connection between testosterone deficiency and chronic kidney dysfunction conflict with these previous findings. For example, a population-based study in the United States showed that testosterone deficiency was not associated with prevalence of chronic kidney disease [estimated glomerular filtration rate (eGFR) <60 mL/min/1.73 m2] among adult men.
However, given that the mean age of that study population was 41.9 years, the finding may not be applicable to aging populations, which are generally characterized by a relatively high prevalence and degree of testosterone deficiency. Identifying a link between magnitude of reduction in kidney function related to low testosterone levels and an equivalent magnitude of reduction in kidney function related to other factors such as aging may be of marked clinical significance.
Here, to examine whether or not testosterone deficiency is independently associated with reduced kidney function among adult men living in a super-aged community, we analyzed data from the Locomotive Syndrome and Health Outcome in Aizu Cohort Study (LOHAS).
Methods
Our cross-sectional study was approved by the Research Ethics Committee of Fukushima Medical University School of Medicine. The LOHAS is a population-based study initiated in 2008 involving residents aged 40–80 years who participated in annual health check-ups in 2 communities (Tadami and Minamiaizu towns) in Fukushima Prefecture, Japan. Tadami and Minamiaizu towns are considered “super-aged” societies based on the World Health Organization criterion of “≥ 20% of residents are aged ≥65 years” (2010 values were 41.3% and 35.7% in Tadami and Minamiaizu, respectively).
Statistics Bureau Summary of Iwate prefecture, Miyagi prefecture, and Fukushima prefecture (Results of basic complete tabulation on population and households: The 2010 Population Census).
Eligibility criteria for our present study were “aged ≥ 40 years” and “participated in the health check-up conducted in 2010.” No exclusion criteria were set. All participants provided written informed consent. Details of the design of the LOHAS have been reported previously.
Locomotor dysfunction and risk of cardiovascular disease, quality of life, and medical costs: Design of the Locomotive Syndrome and Health Outcome in Aizu Cohort Study (LOHAS) and baseline characteristics of the study population.
was the main exposure, measured via enzyme-linked immunosorbent assay (ELISA) using the RE52631 system (IBL International GmbH, Hamburg, Germany) and expressed in pg/mL (to convert to picomoles per liter, multiply by 3.47). Saliva was collected in the morning, at least 30 minutes after breakfast. Participants were asked to avoid brushing their teeth and instead gently clean their mouth with water. Whole saliva was collected by having participants spit directly into polypropylene tubes through a polypropylene straw tube. If a participant had little saliva, he was instructed to chew gum prepared specially for saliva collection. The supernatants of saliva obtained after centrifugation (3000 × g, 10 minutes) were kept at −80°C for further analysis.
In the present study, sT was measured in a laboratory at Teikyo University (Tokyo, Japan). As an external reference, measured in the same laboratory, median sT for men aged 20–40 years was found to be 139.4 pg/mL (10th–90th percentile, 43.8–288.0 pg/mL), with intra- and interassay coefficients of variance of 3.9%–8.8% and 6.7%–8.0%, respectively.
Slight cross-reactions with other natural steroids in the human body and their profiles were as follows: dihydrotestosterone, 2.5%; androstenedione, 0.85%; and others, <0.1%.
where eGFR represents the estimated GFR in mL/min/1.73 m2. This equation was developed in a population of 413 individuals with and without chronic kidney disease and validated on an additional 350 individuals. This Japanese equation has been found to predict measured GFR computed from inulin clearance more accurately than the Japanese version of the Modification of Diet in Renal Disease formula.
Serum creatinine was measured in the present study via enzymatic assay (N-assay L Creatinine Kit; Nittobo Medical Co, Ltd, Tokyo, Japan).
Measurement of Potential Confounding Variables
Potential confounding variables examined in the present study were sociodemographic characteristics, such as age and smoking history (defined as present if individuals answered “smoke currently” or “quit smoking” to the question concerning smoking habit), as well as the presence of cerebrovascular disease and the presence of heart disease, all obtained via self-reported questionnaire; body mass index and systolic and diastolic blood pressure, as measured by local nurse practitioners; diabetes, defined as having glycosylated hemoglobin values ≥6.1%, as described by the Japanese Diabetes Society [equivalent to ≥6.5% in National Glycohemoglobin Standardization Program (NGSP) values
The Committee of the Japan Diabetes Society on the Diagnostic Criteria of Diabetes Report of the Committee on the Classification and Diagnostic Criteria of Diabetes Mellitus.
] or by individuals reporting attending a physician for treatment; dyslipidemia, defined as having low-density lipoprotein cholesterol values ≥140 mg/dL, as described by the Japan Atherosclerosis Society
Diagnostic criteria for dyslipidemia. Executive summary of Japan Atherosclerosis Society (JAS) guideline for diagnosis and prevention of atherosclerotic cardiovascular diseases for Japanese.
or by individuals reporting attending a physician for treatment.
Statistical Analysis
Participants with complete data were entered into primary analyses, which were conducted using Stata v 12.0 (Stata Corp, College Station, TX). sT, sociodemographic characteristics, and comorbidities were described in overall values as well as by quintiles of sT. Differences in these characteristics by quintiles of sT were evaluated using the Kruskal-Wallis test for continuous variables and the χ2 test for categorical variables. Box plots for sT stratified by age categories were created. Effect measures in the present study were differences in mean eGFR by different sT levels, calculated using a general linear model. To estimate adjusted difference in mean eGFR, the potential confounding variables described above were simultaneously forced into the models along with sT. Although age is a negative factor in the eGFR formula, we included age as a covariate in our general linear models because age is a confounder from a biological point of view: aging is associated with declining kidney function and decreasing testosterone levels.
Given that the association of sT (as continuous variables) and eGFR might have been nonlinear, as nonlinear relationships are well-established for several hormonal systems in the endocrinology literature,
separate models were constructed to assess the shape of the association between sT and eGFR, where sT was included as (1) a linear variable, (2) a log-transformed variable, and (3) a transformation using restricted cubic splines with three knots. To assess the fitness of these models, the adjusted R2 values were reported.
Model superiority in order of increasing adjusted R2 value was as follows: linear (0.1534), logarithmic (0.1558), and restricted cubic spline (0.1572). We, therefore, chose the restricted cubic spline model for primary analysis and further testing, as this model provided a good fit and showed a continuous relationship between sT and eGFR. In this model, the 90th percentile of sT was chosen as a reference, as this value corresponds to the median of the highest quintile.
Sensitivity Analysis
In addition to the above analyses, we also conducted several sensitivity analyses. First, the association between sT and eGFR was reported, with log-transformed sT included. Second, the association between sT and eGFR was analyzed among participants aged over 60 years. Finally, the association between sT and eGFR was analyzed without adjustment for age. As described in the Results section, only 1 participant had missing values for any covariates. We, therefore, used only complete data and did not impute the missing value. P < .05 was considered statistically significant.
Results
Of the 1031 men who underwent the health check examination (Figure 1), 69 and 100 participants were missing data for sT and eGFR, respectively. After excluding 37 participants with poor-quality saliva specimens (because of inadequate amount obtained or suspected blood contamination), 849 (82.3%) remained with both sT and outcome variables. After excluding another participant with values for at least 1 confounding variable missing (cerebrovascular disease), the remaining 848 participants were ultimately entered into the primary analyses.
Fig. 1Flow chart of study participants. Sum of number of participants with missing values for salivary testosterone (sT) and outcome variables (eGFR) exceeds 145 because 25 participants had missing values for 2 variables.
Baseline characteristics are presented in Table 1. Median age in the present study was 69 years (interquartile range, 63–75 years), the same value as the mean age (69 years). The prevalence of diabetes, dyslipidemia, cerebrovascular disease, and heart disease were 13.4%, 32.6%, 5.5%, and 11.9%, respectively. Median sT was 61.9 pg/mL (interquartile range, 41.2–93.3 pg/mL). Box plots for sT showed a trend of decreasing median sT with increased age, with the lowest median value observed in those aged ≥80 years (Figure 2). Median eGFR was 69.1 mL/min/1.73 m2 (interquartile range, 60.3–76.1 mL/min/1.73 m2). When stratified by sT quintiles, median age was higher and median eGFR lower in lower sT quintiles, and the prevalence of diabetes and heart disease were highest in the lowest sT quintile. Differences in age, eGFR, and the prevalence of heart disease among sT quintiles were statistically significant. As a group, baseline characteristics among participants in the primary analysis set were similar to those in participants with missing variables or poor quality saliva samples, except for sT values, age, and diabetes (Supplementary Material).
Table 1Baseline Characteristics of the Analysis Population
For continuous variables, values are presented as median and interquartile range. Differences in baseline characteristics by quintiles of salivary testosterone were assessed using the Kruskal-Wallis test for continuous variables and the χ2 test for categorical variables.
Fig. 2Box plots for salivary testosterone (sT) levels stratified by age categories. The white line in the boxes indicates median sT in each category. The top and bottom of the boxes indicate 75th and 25th percentiles of sT in each category, respectively.
In the covariate-adjusted restricted cubic spline model, the splines demonstrated a nonlinear relationship between sT and difference in eGFR (P = .010 for nonlinearity). The estimated shape of the sT curve is shown in Figure 3, suggesting that the lower the sT level, the greater the difference in eGFR, with the 90th percentile of sT as reference. For example, the adjusted difference in eGFR in the fifth percentile of sT was −3.43 [95% confidence interval (CI) −6.02 to −0.84] (Table 2), a difference comparable in magnitude to the reduction in eGFR observed for a 6-year increase in age in our cohort (−3.12 mL/min/1.73 m2, 95% CI −3.75 to −2.49) and the reduction in eGFR observed for presence of heart disease, (−4.06 mL/min/1.73 m2, 95% CI −6.76 to −1.36).
Fig. 3Adjusted difference in eGFR rate in men with aged over 40 years. Linear regression model with restricted cubic spline adjusted for covariates (age, body mass index, smoking history, SBP, DBP, diabetes, dyslipidemia, cerebrovascular disease, and heart disease). The 90th percentile of salivary testosterone (sT) level (ie, median of the fifth quintile) was cited as reference. The left vertical axis shows difference in eGFR, and the solid line indicates point estimates of eGFR while dotted lines indicate CIs. Gray bars indicate frequency of the sT values. The right vertical axis shows frequency of each gray bar. CI, Confidence interval.
Linear regression model with restricted cubic spline adjusted for covariates (age, body mass index, smoking history, SBP, DBP, diabetes, dyslipidemia, cerebrovascular disease, and heart disease). The 90th percentile of sT level (ie, median of the fifth quintile) was cited as reference.
∗ Linear regression model with restricted cubic spline adjusted for covariates (age, body mass index, smoking history, SBP, DBP, diabetes, dyslipidemia, cerebrovascular disease, and heart disease). The 90th percentile of sT level (ie, median of the fifth quintile) was cited as reference.
In sensitivity analysis using log-transformed sT, lower log-transformed sTs were associated with lower eGFR [− 0.93 (95% CI −1.80 to −0.07) mL/min/1.73 m2 per 1 standard deviation decline] (Supplementary Material). Associations between sT and eGFR as assessed in another sensitivity analysis among participants aged over 60 years (mean, median, and interquartile range of age were 71, 71, and 66–76 years, respectively) were similar to those among participants aged over 40 years (Supplementary Material). In addition, sensitivity analysis with adjustment of covariates except for age showed stronger associations than those presented in Table 2 and Figure 3 (Supplementary Material, respectively).
Discussion
In this cross-sectional study of adult men living in super-aged communities, low testosterone levels were independently associated with reduced eGFR, with particularly steep reduction of eGFR observed when sT levels dropped below 100 pg/mL. Further, the strength of the association between reduced kidney function and testosterone deficiency was comparable to that of a 6-year increase in age or presence of heart disease. This association should be confirmed in a prospective study.
Our finding regarding the relationship between low testosterone levels and decreased eGFR is consistent with findings of a protective role of testosterone on kidney from several clinical and basic studies.
However, a conflicting result has also been reported regarding the association between testosterone levels and kidney disease from a study in the United States among community-dwelling men.
Such conflicting findings may be explained in part by differences between measured fractions of circulating testosterone and those in commercial assays. In addition, the population in that study in the United States was younger (mean age, 41.9 years) than the present one (mean age, 69 years) and may, therefore, have involved fewer participants with low testosterone levels, leading to a null association between testosterone levels and chronic kidney disease.
We believe that our findings here will influence the activities of physicians and basic researchers for several reasons. First, low testosterone level may be a modifiable risk factor for reduced kidney function. Given that endogenous production of testosterone can be induced by exercise, testosterone deficiency can be managed with exercise instruction.
Second, the biological connection between low testosterone levels and reduced kidney function might be independent of metabolic effects potentially induced by testosterone deficiency, such as obesity, diabetes, or dyslipidemia, as we observed that the relationship between testosterone and reduced kidney function persisted even after adjustment for such covariates. Testosterone deficiency may lead to reduced kidney function by a number of pathways other than obesity, diabetes, and dyslipidemia. For example, ischemia via reduced hemoglobin concentration may be involved, as low testosterone levels may be associated with anemia.
In addition, ischemia might be induced by endothelial dysfunction related to testosterone deficiency, as testosterone induces vasodilation in the renal vessels via the production of nitric oxide.
Inflammation may be also a mediator of kidney injury induced by testosterone deficiency, as levels of inflammatory cytokines or inflammatory markers have been found to be reduced by testosterone administration.
Further study is warranted to clarify the mechanisms by which testosterone deficiency chronically reduces kidney function.
Several strengths to the present study warrant mention. First, we demonstrated the relationship between testosterone and reduced kidney function among adult men living in a super-aged community, adjusting for potential confounding variables such as blood pressure, body mass index, and comorbidities such as diabetes and dyslipidemia, which may be related to both testosterone and kidney function. Given that societies in developed countries now are becoming super-aged, such as the communities in our present study, we believe that our findings will be generalizable to such societies worldwide. Second, we demonstrated that the strength of the association between low testosterone level and reduced kidney function was comparable to that of a 6-year increase in age or presence of heart disease, in terms of difference in eGFR. This finding highlights the potential importance of testosterone deficiency on kidney health as well as aging and vascular damage.
However, several limitations to the present study also warrant mention. First, although sT is a reliable and suitable metric for evaluating testosterone levels in population-based studies,
the clinical guideline for the diagnosis of hypogonadism does not recommend its clinical use, as the methodology has not been standardized and adult male ranges are not yet available in most hospital or reference laboratories.
Further, some endocrine specialists question the validity of sT when measured by ELISA. However, previous studies have shown that sT levels obtained via ELISA share good correlation with those obtained via liquid chromatography/mass spectrometry,
Diagnostic significance of salivary testosterone measurement revisited: Using liquid chromatography/mass spectrometry and enzyme-linked immunosorbent assay.
Despite these drawbacks, however, utilization of sT still allows for simple, minimally invasive screening to evaluate testosterone levels on a population-level basis.
Second, cases of nonprescriptional use of testosterone or methyl-testosterone could not be recorded. However, use of such medication in our population is unlikely, as these compounds require a physician's prescription and are not available as over-the-counter drugs in Japan. As such, any influence on our findings—if present at all—would have been negligible. Third, the cross-sectional design of the present study means that we cannot attribute causality from the associations between testosterone and reduced kidney function. Although we were unable to determine a potential mechanism for reverse causality in this community-dwelling population, men with advanced chronic kidney disease are more likely to suffer hypogonadism than community-dwelling men, as luteinizing hormone signaling may be inhibited by uremia at the level of Leydig cells.
Fourth, hemoglobin levels could not be included in our study because 766 (90.4%) of our participants did not receive hemoglobin measurement. However, we assume that any influence of anemia would be intermediary between testosterone deficiency and reduced kidney function (ie, not a confounding effect), as testosterone replacement leads to an increase in hemoglobin levels.
Fifth, although serum creatinine needs to be measured twice with an interval of at least 3 months between measurements to diagnose chronic kidney disease,
serum creatinine measurement was performed only once in our study, and, thus, we cannot exclude the presence of participants with acute kidney injury. However, given that all of our participants received health-checkup in community settings, we believe the influence of such injury to be negligible.
Conclusions
In conclusion, lower testosterone was associated with reduced kidney function among Japanese adult men living in a super-aged community. Further, the strength of this association was comparable to that of a 6-year increase in age or presence of heart disease. Our finding of this association between low testosterone levels and reduced kidney function needs to be corroborated among participants with chronic kidney disease or in a longitudinal study.
Acknowledgments
We thank Yan Lu (Department of Urology, Teikyo University School of Medicine) for technical assistance.
Endogenous testosterone and mortality due to all causes, cardiovascular disease, and cancer in men: European Prospective Investigation into Cancer in Norfolk (EPIC-Norfolk) Prospective Population Study.
High serum testosterone is associated with reduced risk of cardiovascular events in elderly men: The MrOS (Osteoporotic Fractures in Men) Study in Sweden.
Summary of Iwate prefecture, Miyagi prefecture, and Fukushima prefecture (Results of basic complete tabulation on population and households: The 2010 Population Census).
Locomotor dysfunction and risk of cardiovascular disease, quality of life, and medical costs: Design of the Locomotive Syndrome and Health Outcome in Aizu Cohort Study (LOHAS) and baseline characteristics of the study population.
Diagnostic criteria for dyslipidemia. Executive summary of Japan Atherosclerosis Society (JAS) guideline for diagnosis and prevention of atherosclerotic cardiovascular diseases for Japanese.
Diagnostic significance of salivary testosterone measurement revisited: Using liquid chromatography/mass spectrometry and enzyme-linked immunosorbent assay.
00049-9&id=gr1.jpg){kind=link}
00049-9&id=gr2.jpg){kind=link}
00049-9&id=gr3.jpg){kind=link}