|
|
Background Dehydroepiandrosterone (DHEA) and testosterone are widely promoted as antiaging supplements, but the long-term benefits, as compared with potential harm, are unknown.
Methods We performed a 2-year, placebo-controlled, randomized, double-blind study involving 87 elderly men with low levels of the sulfated form of DHEA and bioavailable testosterone and 57 elderly women with low levels of sulfated DHEA. Among the men, 29 received DHEA, 27 received testosterone, and 31 received placebo. Among the women, 27 received DHEA and 30 received placebo. Outcome measures included physical performance, body composition, bone mineral density (BMD), glucose tolerance, and quality of life.
Results As compared with the change from baseline to 24 months in the placebo group, subjects who received DHEA for 2 years had an increase in plasma levels of sulfated DHEA by a median of 3.4 µg per milliliter (9.2 µmol per liter) in men and by 3.8 µg per milliliter (10.3 µmol per liter) in women. Among men who received testosterone, the level of bioavailable testosterone increased by a median of 30.4 ng per deciliter (1.1 nmol per liter), as compared with the change in the placebo group. A separate analysis of men and women showed no significant effect of DHEA on body-composition measurements. Neither hormone altered the peak volume of oxygen consumed per minute, muscle strength, or insulin sensitivity. Men who received testosterone had a slight increase in fat-free mass, and men in both treatment groups had an increase in BMD at the femoral neck. Women who received DHEA had an increase in BMD at the ultradistal radius. Neither treatment improved the quality of life or had major adverse effects.
Conclusions Neither DHEA nor low-dose testosterone replacement in elderly people has physiologically relevant beneficial effects on body composition, physical performance, insulin sensitivity, or quality of life. (ClinicalTrials.gov number, NCT00254371
[ClinicalTrials.gov]
.)
DHEA modestly increases testosterone levels in women, although in the absence of overt hypogonadism, DHEA replacement has a minimal effect on testosterone levels in men.9 It is unclear whether testosterone supplementation has a benefit in elderly men with a modest reduction in the level of "bioavailable" testosterone (i.e., the fraction of circulating testosterone that is not bound to sex hormone–binding globulin). Therefore, there is a growing debate about whether to treat the substantial proportion of elderly men (up to 90% in some reports)10 whose level of bioavailable testosterone is below that of young men. Moreover, testosterone replacement in elderly men may be associated with risks, especially of prostate cancer and progression of benign prostatic hypertrophy.11 This uncertainty recently prompted the Institute of Medicine to conclude that additional well-controlled studies examining the risks and benefits of testosterone replacement in elderly men should be performed before large-scale, long-term clinical trials are undertaken.12 In postmenopausal women, studies show that conventional estrogen replacement has substantial adverse effects; trials of ultralow estrogen replacement have demonstrated beneficial effects.13
We conducted a 2-year, randomized, placebo-controlled, double-blind study to determine the effects of full DHEA replacement and low-dose testosterone replacement on body composition, physical performance, bone mineral density (BMD), and glucose tolerance in elderly people with low androgen levels. We also determined whether receiving this hormone-replacement therapy had adverse effects related to the prostate.
Methods
Subjects
Subjects were eligible to participate in the study if they were at least 60 years of age. Eligibility criteria included, for men, a level of bioavailable testosterone that was less than 103 ng per deciliter (3.6 nmol per liter) and a sulfated DHEA level that was less than 1.57 µg per milliliter (4.3 µmol per liter), and for women, a sulfated DHEA level that was less than 0.95 µg per milliliter (2.6 µmol per liter). These cutoff values, which represented the 15th percentile of levels for normal young men and women,2 were chosen to ensure that a sufficient number of healthy elderly people could participate in the study. All volunteers underwent a medical history taking and physical examination and were excluded if there was evidence of clinically important coexisting illnesses or conditions that could have an effect on outcome measures.
In addition, we evaluated 38 healthy young women and 37 healthy young men between the ages of 18 and 31 years once in order to obtain a baseline for a comparison of outcome measures. Elderly men underwent a digital rectal examination and ultrasonography to quantify the size of the prostate and to detect any nodules, and all elderly men with a level of prostate-specific antigen (PSA) above the age-adjusted normal level were excluded.
Study Design
The study was approved by the institutional review board of the Mayo Foundation, and all subjects gave written informed consent. The study was designed and conducted entirely by the study team without industry support. Randomization schedules were prepared by study statisticians. Study groups included elderly men receiving a DHEA tablet (75 mg per day) and a transdermal placebo patch, a placebo tablet and a transdermal testosterone patch (5 mg per day; D-TRANS, Alza), or a placebo tablet plus a placebo transdermal patch. Elderly women received either a DHEA tablet (50 mg per day) or a placebo tablet. Identical-appearing blue capsules contained either DHEA (which was 95.5% pure on analysis) or placebo with lactose as filler. Only the statisticians and pharmacists had access to the coded treatment assignments. We randomly assigned 92 men and 60 women to the study groups (Figure 1).
|
If we observed an increase in the PSA level of 0.75 ng per milliliter or more, we repeated the PSA measurement in 3 months. If the PSA level remained elevated, the subject was examined by a urologist who was unaware of the subject's treatment assignment. A digital examination, ultrasonography, and a biopsy of the prostate were performed if such procedures were clinically warranted.
Outcome Measures
Primary outcome measures were physical performance, the peak aerobic capacity, body composition, BMD, and levels of plasma insulin and glucose after an overnight fast. Additional measurements included body weight, the proportion of body fat, the insulin-sensitivity index, quality of life, levels of various hormones, and levels of alkaline phosphatase, alanine aminotransferase, aspartate transferase, and hemoglobin. Adverse effects, including increases in the PSA, were assessed.
Measures of physical performance included muscle strength and the peak aerobic capacity, as reflected by the maximum volume of oxygen consumed per minute (VO2). The peak VO2 was measured during a graded-intensity treadmill-walking test, with expired gas exchange assessed as previously described.14 Isometric torque of the knee extensors was measured (the best of five maximal voluntary contractions) on the dominant leg while the subject was seated and the knee angle was fixed at 60 degrees of flexion. The one-repetition maximum (the highest weight that can be lifted one time) for the double leg press and chest press was determined from a progressive series of attempts on adjustable weight-stack machines. All subjects were familiarized with the equipment and procedures on a separate visit preceding the data collection, and all tests were supervised by exercise specialists.
Body composition and BMD, including the proportion of abdominal visceral fat and the fat-free mass, were measured with the use of dual-energy x-ray absorptiometry (DPX-IQ, Lunar),15 and the thigh-muscle area was measured with computed tomography.16 Abdominal visceral fat was measured as previously described.15,17 BMD was obtained at the anteroposterior mass of the lumbar spine (L2 to L4), femoral neck, total hip, distal radius, and ultradistal radius.
After an overnight fast, subjects ingested a mixed meal consisting of 45% carbohydrate, 40% fat, and 15% protein, totaling 10 kcal per kilogram of body weight.18 Arterialized venous blood was sampled at regular intervals for 30 minutes before and 6 hours after the meal to measure levels of glucose, insulin, and C peptide. The oral glucose minimal model19 was used to calculate the insulin-sensitivity index.
Ultrasonography of the prostate was performed with the use of a probe and biplanar imaging.
Levels of sulfated DHEA, total and bioavailable testosterone, follicle-stimulating hormone, and estradiol were measured by competitive chemiluminescence immunoassay (with a high-sensitivity competitive chemiluminescence immunoassay for subjects with low levels of estradiol); sex hormone–binding globulin was measured by solid-phase, two-site chemiluminescence immunometric assay (Immulite, Diagnostic Products). In subjects with low testosterone levels, values were obtained with the use of high-sensitivity competitive chemiluminescence immunoassay (ACS-180, Bayer Diagnostics); bioavailable testosterone and bioavailable estradiol were measured on the basis of differential precipitation of sex hormone–binding globulin by ammonium sulfate after the equilibration of serum samples with tracer amounts of tritium-labeled testosterone and estradiol.
We used the Health Status Questionnaire (HSQ) to evaluate subjects' quality of life.20 The HSQ adds three questions to the Medical Outcomes Study 36-item Short General Form Health Survey (SF-36)21 to provide a further assessment of emotional function. Although the HSQ gives rise to eight dimensions, or scales, of health status, it can also be scored to generate the two factor-derived scores from the SF-36 questionnaire (the physical component and the mental component). Each summary score is assigned a mean (±SD) score of 50±10 on the basis of an assessment of a general U.S. population without chronic conditions; individual scores were then compared with the normalized scores for the general population.21
Statistical Analysis
On the basis of preliminary estimates of the corresponding standard deviations, we determined that 30 subjects would be required in each group for the study to have a statistical power of 90% to detect clinically meaningful differences between groups. We therefore planned to enroll 150 elderly subjects (90 men and 60 women) during a 5-year period, with all subjects followed for 24 months. Difficulties in recruiting volunteers in a timely manner resulted in an extension of the study to 6 years. Recruitment was terminated on June 30, 2002; follow-up was terminated on December 31, 2003. Thus, not all subjects had their last follow-up visit at a full 24 months. For men receiving DHEA, the median duration of treatment was 23.2 months (interquartile range, 22.6 to 23.5); for men receiving testosterone, the median was also 23.2 months (interquartile range, 22.2 to 23.8); and for men receiving placebo, the median was 23.1 months (interquartile range, 22.7 to 24.0). For women receiving DHEA, the median duration of treatment was 23.0 months (interquartile range, 22.7 to 23.7), and for women receiving placebo, the median was 23.3 months (interquartile range, 22.4 to 23.5).
Changes in the end points of interest were calculated by comparing the value at baseline with that at the last measurement. The value at the last follow-up visit was used for subjects who were followed for less than 24 months but not less than 12 months. Univariate and multivariate associations of these changes for each end point were calculated for each treatment group with the use of multiple regression analysis. Independent variables included the group, the length of follow-up, the age of the subject, and the baseline value of the end point of interest. The dependent variable was the change (from lowest to highest) from baseline to 24 months. Four subjects for whom no data were available at or after 12 months were not included in the analysis. Two-sided tests were used, and P values of less than 0.05 were considered to indicate statistical significance.
Randomization and maintenance of the project database, data editing, and data analysis were carried out in the Division of Biostatistics at the Mayo Clinic.
Results
Baseline Characteristics
The characteristics of the elderly men and women did not differ significantly among the groups at baseline (Table 1). (Additional details are listed in Table 1 of the Supplementary Appendix, available with the full text of this article at www.nejm.org.) Baseline characteristics of young subjects are listed in Table 2 of the Supplementary Appendix.
|
Subjects in the DHEA groups (but not the placebo groups) had a significant increase from baseline to 24 months in levels of sulfated DHEA and estradiol (both total and bioavailable forms), and women had an increase in levels of total testosterone (Figure 2; additional details are shown in Table 3 of the Supplementary Appendix). Men in the testosterone group had a significant increase in levels of bioavailable testosterone and total testosterone, as compared with men in the placebo group. Neither men nor women in the DHEA group had significant changes in the levels of follicle-stimulating hormone or luteinizing hormone; men in the testosterone group had significantly lower levels of both hormones (data not shown). Subjects in both the DHEA group and the testosterone group had no significant changes in fasting plasma glucose or in the insulin-sensitivity index. Significant changes in fasting insulin levels were noted in the testosterone group but not in the placebo group. Taken together, men and women in the DHEA group had significant reductions in the levels of high-density lipoprotein cholesterol, but no other measures of lipids were affected by treatments.
|
Among the primary outcome measures, only fat-free mass differed significantly between the treatment groups and placebo groups. When men and women in the DHEA group were considered separately, no significant changes were seen in body-composition measurements. When men and women were combined, the DHEA group had a slight but significant increase in fat-free mass (less than 0.5 kg) and a decrease in the proportion of body fat (less than 1.5%). Men in the testosterone group had a significant increase in fat-free mass. The changes in the peak VO2 and the measures of muscle strength were similar in the combined DHEA group and the placebo group, as well as in the testosterone group and the placebo group.
BMD
In the DHEA group, women had a slight, but significant increase in the BMD of the ultradistal radius, and men had a slight, but significant increase in the BMD of the femoral neck. Men in the testosterone group had a significant increase in BMD only in the femoral neck. Subjects in neither the DHEA group nor the testosterone group had a significant increase in BMD at other sites.
Quality of Life
Subjects in the DHEA and testosterone groups had no significant change in scores on the Physical Component Scale and the Mental Component Scale of the HSQ (Table 2, and Figure 1, 2, and 3 of the Supplementary Appendix).
|
Measures of prostate volume, PSA levels, liver function, electrolyte levels, and hemoglobin levels were not significantly altered by treatment with either DHEA or testosterone (Table 3, and Tables 4, 5, and 6 of the Supplementary Appendix).
|
The administration of DHEA for about 23 months in elderly people with low androgen levels increased the levels of sulfated DHEA to values that would be considered in the high-normal range for young people. This therapy slightly increased levels of testosterone and estradiol in women and levels of estradiol in men. The administration of low-dose testosterone for just over 23 months significantly increased the levels of both total and bioavailable testosterone in men. However, treatment with neither DHEA nor testosterone had any detectable effect on physical performance, insulin sensitivity, or the physical and mental components of the quality of life. Testosterone replacement resulted in a small but significant increase in fat-free mass but no significant change in thigh-muscle area or muscle strength. DHEA had no effect on fat-free mass in men or women when the groups were analyzed separately according to sex. The lack of a significant effect on thigh-muscle area, strength, or fitness largely discounts the relevance of the change in fat-free mass. Among the five sites measured for BMD, the women in the DHEA group had a small but significant increase in BMD of the ultradistal radius; men in both the DHEA and testosterone groups had an increase in BMD in the femoral neck. On the other hand, treatment with DHEA or testosterone caused no detectable harm, since side effects (including PSA levels and prostate volume) did not differ significantly between men in the treatment groups and those in the placebo group.
On the basis of the 95% CIs shown in Table 2, it is not likely that our negative findings can be attributed to the small number of subjects in the study. Specifically, these CIs indicate that the number of subjects was sufficient to establish that any treatment effects that might exist were not clinically meaningful. For example, the 95% CI in the DHEA group for the measure of fat-free mass in elderly men ranged from 0 to 1.78 kg, thus ruling out an effect greater than 1.8 kg. If the study had enrolled more subjects, the CI would have been even narrower. The upper limit of the 95% CI for the effect of DHEA on fat-free mass in elderly women was 1.35 kg, and for testosterone in men it was 2.15 kg.
Similarly, the number of subjects in the study appears to have been adequate to rule out clinically meaningful effect sizes for muscle strength, as measured by the upper limits of the 95% CI for the chest press (2.27 kg for both men and women in the DHEA group and 4.54 kg in the testosterone group), the fasting glucose level (1.37 mg per deciliter [0.08 mmol per liter] for men and 2.45 mg per deciliter [0.14 mmol per liter] for women in the DHEA group and 2.99 mg per deciliter [0.17 mmol per liter] in the testosterone group), and BMD at the femoral neck (0.04 g per square centimeter of body-surface area for men and 0.02 g per square centimeter for women in the DHEA group and 0.05 g per square centimeter in the testosterone group). Even though women in the DHEA group had a significant increase in BMD at the ultradistal radius (P=0.005), the number of subjects was adequate to establish that the magnitude of the effect was less than that reported with current treatments, such as bisphosphonates.22 In contrast with our findings, 3 months of resistance exercise training increased chest-press strength by an average of 15 kg in older people.23
We administered 75 and 50 mg of DHEA per day to elderly men and women, respectively, to achieve circulating levels present in young people of the same sex.24,25 Previous studies have examined the effects of DHEA doses ranging from 50 mg26,27,28 to 1600 mg29,30 given as a single dose or as repeated daily doses for up to 1 year.25,31,32 These studies have had inconsistent results, with some reporting positive effects on body composition, muscle strength, BMD, and glucose tolerance, and others showing no effect.8,9,21,25,27,29,33,34,35,36,37,38,39,40,41 The lack of a detectable effect of DHEA replacement on strength, peak VO2, quality of life, or measures of insulin sensitivity does not preclude the possibility that DHEA had short-term effects on these measures or that pharmacologic doses could have a biologic effect. However, the current data indicate that if short-term restoration of DHEA levels in elderly subjects has favorable biologic effects, they are not sustained.
The effect of DHEA on BMD is less clear. Elderly women in the DHEA group had a small but significant increase in BMD of the ultradistal radius, and men in both the DHEA group and the testosterone group had an increase in BMD of the femoral neck. However, there was no significant increase in BMD at other sites in either group. Although this may have been a chance finding, an actual effect might have been influenced by the associated plasma estrogen levels. Since other well-tolerated and tested pharmacologic agents result in a far greater increase in BMD, the value of DHEA for either preventing or treating osteoporosis in elderly men or women is probably limited.
The administration of low-dose testosterone was associated with a small increase in fat-free mass but was not associated with a significant increase in the thigh-muscle area or with an improvement in any of the performance measures. In contrast to these results, a study of intramuscular testosterone therapy (at a dose of 200 mg every 2 weeks for 36 months)42 demonstrated a larger increase in fat-free mass (6.7%) and a larger decrease in the proportion of body fat (17.3%). A previous study in elderly people in which testosterone levels were increased to values that were in the middle of the normal range for young people also did not show any effect on muscle strength.43 Thus, although a pharmacologic dose of testosterone may have a marked effect on body composition and function, low-dose replacement has no demonstrable effect. Higher doses of testosterone that increase levels in older men to levels seen in young men may increase adverse effects (prostate enlargement and prostate cancer11), and any beneficial effects remain to be established.
It is reassuring that men in the DHEA, testosterone, and placebo groups had similar changes in both the PSA level and prostate size. However, the 2 years of observation may not be sufficient to detect subtle adverse effects of DHEA or testosterone on prostate growth and differentiation. Our study also does not preclude the possibility of adverse effects from higher doses of either of these agents. We therefore believe that the long-term safety of DHEA or testosterone therapy in elderly people remains uncertain.
In all, we found little if any beneficial effect of the restoration of DHEA levels in elderly men and women to those in healthy young people of the same sex. Although DHEA replacement has no detectable effect on body composition, physical performance, insulin action, or quality of life, it resulted in a minimal and inconsistent effect on BMD, the magnitude of which was far smaller than that of established therapies for osteoporosis. Additional long-term studies of testosterone are warranted to determine the risk–benefit ratio of higher doses. Taken together, our data provide no evidence that either DHEA or low-dose testosterone is an effective antiaging hormone supplement and argue strongly against the use of these agents for this purpose.
Supported by grants (PO1 AG14283 and MO1 RR00585) from the National Institutes of Health; by the Department of Medicine, Mayo Clinic; and by the Mayo Foundation. Dr. Nair is supported by the David Murdock Professorship.
Dr. Melton reports having received lecture fees from Amgen, Merck, Novartis, and Procter & Gamble. Dr. Nehra reports having served as a consultant to Pfizer. No other potential conflict of interest relevant to this article was reported.
We are indebted to Traci Hammer for coordinating the study; to Barbara Norby, Jean Feehan, and Laurie Wahlstrom for their nursing support; to Peggy Helwig and the Chemistry Core Laboratory of the General Clinical Research Center (GCRC) for their technical assistance; to Drs. Ann O'Berg, Terry M. Therneau, Scott Przybelski, and Claudia Powell for their statistical support; to Melissa Aakre for her secretarial support; and to members of the GCRC nursing and dietetic staff.
Source Information
From the Division of Endocrinology (K.S.N., R.A.R., K.D., K.R.S., A.B., R.B., S.K., M.D.J.) and the Departments of Health Sciences Research (P.O., L.J.M.), Urology (A.N., D.J.T.), Medicine (J.L.V.), Laboratory Medicine and Pathology (G.G.K.), and Psychology (G.E.S.), Mayo Clinic, Rochester, MN; and the Department of Information Engineering, University of Padua, Padua, Italy (C.C., G.T., C.D.M.).
Drs. Khosla and Jensen contributed equally to this article.
Address reprint requests to Dr. Nair at the Division of Endocrinology, Mayo Clinic, 200 First St. SW, 5-194 Joseph, Rochester, MN 55905, or at nair.sree{at}mayo.edu.
References
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||
Related Letters:
DHEA and Testosterone in the Elderly
Page S. T., Matsumoto A. M., Bremner W. J., Yasuda M., Horie S., Perls T. T., Gleicher N., Barad D., Nair K. S., Smith G.
Extract |
Full Text |
PDF
N Engl J Med 2007;
356:635-637, Feb 8, 2007.
Correspondence
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2008 Massachusetts Medical Society. All rights reserved. |
Previous
