Testosterone increases fat-free mass (FFM) in men infected with human immunodeficiency virus (HIV), but its effects on muscle performance, physical function, mood, and quality of life are poorly understood. Sixty-one HIV-infected men with weight loss were randomized to receive weekly intramuscular injections of 300 mg of testosterone enanthate or placebo for 16 wk. The primary outcome of interest was physical function (walking speed, stair-climbing power, and load-carrying ability). Secondary outcome measures included body weight and composition, muscle performance, sexual function, mood, and quality of life. Serum nadir free and total testosterone levels increased (+188.0 ± 29.6 and +720 ± 86 ng/dl) in the testosterone, but not placebo, group. Testosterone administration was associated with increased FFM (2.8 ± 0.5 kg), which was significantly greater than in the placebo group (P < 0.0001). Leg press strength increased significantly in testosterone-treated (P = 0.027), but not placebo-treated, men; the difference between groups was not significant. Other measures of muscle performance and physical function did not change significantly in either group. Men receiving testosterone demonstrated significantly greater improvements in mental health and quality-of-life scores than those receiving placebo and improvements in fatigue/energy and mood scores that were not significantly different from those receiving placebo. Sexual function scores did not change in either group. In HIV-infected men with weight loss, a supraphysiological dose of testosterone significantly increased FFM but did not improve self-reported or performance-based measures of physical function. Improvements in mood, fatigue, and quality-of-life measures in the testosterone group, although clinically important, need further confirmation.
- body composition
- quality of life
- sexual function
- randomized controlled trial
even in an era of widespread use of highly active antiretroviral therapy, human immunodeficiency virus (HIV)-associated weight loss remains a common problem (40, 47) and is an independent predictor of mortality (12, 15, 24, 27, 28). HIV-infected patients with unintentional loss of ≥5% of body weight in a 6-mo period have a higher frequency of opportunistic infections and hospitalizations (22, 48) and a decreased survival compared with those without weight loss (12, 22). Weight loss in HIV infection is multifactorial and involves loss of fat and lean mass (27, 31). Loss of lean body mass in HIV-infected men is associated with diminished quality of life (49).
A number of clinical trials have assessed the impact of treatment with physiological and supraphysiological doses of androgens in HIV-infected men. Many of these trials have shown an increase in weight and lean body mass with androgen therapy (6, 7, 9, 18, 19, 21, 38, 42, 43). Meta-analyses of randomized trials in HIV-infected men with weight loss have shown a greater increase in body weight and lean body mass with androgen therapy than with placebo (4, 26). The effects of testosterone therapy on muscle strength have been inconsistent across trials; some studies have shown improvements in muscle strength, and others have reported no change (3, 7, 9, 18, 19, 43). It has not been clear whether gains in lean body mass during testosterone administration are associated with improvements in physical function. Similarly, data on the effects of testosterone on health-related quality of life, mood, and fatigue are inconsistent for non-HIV-infected (25, 33, 35, 36, 39, 45, 46) and HIV-infected (3, 6, 9, 14, 17, 20, 37, 38, 42) men.
Accordingly, the primary objective of this randomized clinical trial was to determine the effects of a supraphysiological dose of testosterone on physical function and muscle strength in HIV-infected men with mild-to-moderate weight loss. Inasmuch as the anabolic effects of testosterone on lean body mass and muscle strength are related to testosterone dose and concentrations (4, 10), we reasoned that a supraphysiological dose of testosterone would produce greater increments in muscle mass than a dose associated with testosterone replacement used in many previous studies and, therefore, would be more likely to improve muscle strength and physical function. We hypothesized that weekly treatment with 300 mg of testosterone enanthate by intramuscular injection would not only increase body weight and lean body mass but also improve performance-based measures of physical function. We also determined the effects of a supraphysiological dose of testosterone on fatigue, psychological distress, and self-reported physical function as secondary end points.
Study Design and Participants
This double-blind, placebo-controlled, randomized trial was conducted between May 2003 and June 2006 at Charles R. Drew University of Medicine and Science. It consisted of a 2-wk screening and control period, a 16-wk treatment period, and a 16-wk posttreatment observation period. The protocol was approved by the Institutional Review Boards at Charles R. Drew University of Medicine and Science and Boston University. Written informed consent was provided by each participant (ClinicalTrials.gov identifier NCT00260143).
The participants were 18- to 60-yr-old HIV-infected men with self-reported unintentional weight loss ≥5% over the preceding 6 mo or body mass index <20 kg/m2 and serum testosterone <400 ng/dl. Participants were required to be on stable highly active antiretroviral therapy or to be stable without retroviral therapy for the preceding 12 wk. The exclusion criteria included CD4+ T lymphocyte count <50/mm3; change in antiretroviral therapy within the preceding 12 wk; significant diarrhea; acquired immune deficiency syndrome-defining illness other than HIV wasting syndrome within the preceding 3 mo; severe lipodystrophy; any systemic cancer; diabetes mellitus; use of systemic corticosteroids, androgens, growth hormone, or other anabolic/orexigenic agent within 6 mo before screening; significant alcohol or drug abuse; severe or untreated obstructive sleep apnea; or American Urological Association lower urinary tract symptom score >14. Participants were excluded if they were participating in a vigorous resistance-training program or if they had neurological or musculoskeletal disorders precluding muscle strength measurements. Men with aspartate aminotransferase and alanine aminotransferase more than five times the upper limit of normal, severe dyslipidemia, hematocrit >48%, or prostate-specific antigen (PSA) ≥4 ng/ml were excluded.
Randomization and Drug Administration
Participants were randomized using a computer-generated randomization list with block size of 6 and received 300 mg of testosterone enanthate or placebo (sesame oil) by intramuscular injection once weekly for 16 wk. Injections were administered in the General Clinical Research Center by blinded research nurses.
Measures of body composition, muscle strength, and physical function were obtained at baseline and during week 16 of the treatment period. Questionnaires assessing mood and psychological distress, sexual function, and quality of life were administered at baseline and during treatment weeks 8 and 16. Blood counts and chemistries, PSA, and American Urological Association symptom scores were monitored on multiple occasions.
Anthropometrics and Body Composition
Body weight was measured weekly during the control and treatment periods and monthly during the observation period. Height was measured on the first two visits and averaged. The same dual-energy X-ray absorptiometery (DXA) scanner (model 4500A QDR; Hologic, Waltham, MA) was used for all body composition measurements. DXA data combined with a concurrent measure of actual weight were used to calculate fat mass and fat-free mass (FFM), a close correlate of lean body mass. The DXA scanner was calibrated weekly using a body composition analysis step phantom.
Maximal voluntary muscle strength.
Maximal voluntary muscle strength in the leg press exercise was measured by using the one-repetition maximum (1-RM) technique (1) on a seated leg press machine (Keiser Sport, Fresno, CA) with pneumatic resistance (9, 10, 41). Since maximal voluntary strength measurements are highly effort dependent, several strategies were used to ensure reliability and to minimize the learning effect. Tests were performed in duplicate, with careful standardization of starting knee flexion angle (∼90° by goniometry), foot placement, and completion of full range of motion. To avoid the familiarization effect, we instructed the participants in the proper execution of the exercise and allowed practice trials. After familiarization, participants completed a warm-up including 5 min of cycle ergometer or treadmill exercise. The participants performed progressively fewer lifts at heavier loads before attempts at 1 RM. Initial loads were set at 50% of the subject's estimated 1 RM using reference values established in our laboratory. Attempts at 1 RM were interspersed with 2-min rest intervals and continued until 1 RM was identified as the greatest amount of weight lifted through the complete range of motion. Strength tests were conducted in duplicate within 1 wk on nonconsecutive days, with scores required to be within 5%. All participants met these criteria, and the highest value in the trials was taken as the 1 RM.
Leg power was determined using the previously validated (2) Nottingham leg rig (University of Nottingham, Nottingham, UK). The movement pattern for this exercise is similar to that for the leg press exercise, except participants pushed as hard and as fast as possible using only the right leg with the foot precisely positioned on a foot pedal. Peak power (W and W/kg body wt) was calculated using the mass of the flywheel and its revolution frequency. The participants were familiarized with the procedures, asked to complete a warm-up, and positioned so that the knee was flexed to 90°. Trials were continued until a plateau was reached, typically within 15 trials; a 30-s rest period was provided between trials. Peak power was identified as the highest power output.
Local muscle endurance of the leg and hip extensors was determined with the leg press exercise using the same pneumatic resistance device used for the 1-RM tests. Using loads of 80% of their individual baseline 1-RM, participants performed as many leg press repetitions as possible.
We selected tests of physical function that reflect common, everyday activities, including stair climbing, walking speed, and load carrying. Recognizing the possibility of a ceiling effect in this relatively young and functional population, we selected physical function tests that had relatively high effort requirements.
A modification of the Margaria stair climb (30) was used to assess stair-climbing power. Participants were instructed to ascend a 12-step staircase as rapidly as possible, with ascent time recorded over the middle four steps (0.66 m) of the climb. Handrail holding was permitted only for balance. Power (W) was calculated as the product of body mass, total step rise, and the acceleration of gravity all divided by time to traverse the middle four steps.
Participants were instructed to walk a 160-m flat course as fast as possible without running. Photoelectric cells and timers were used to determine time to the nearest 0.01 s. The faster of two trials was used to calculate walking speed (m/s).
To simulate daily activities such as carrying groceries or suitcases, participants carried a load equal to 20% of their body weight while walking a distance of 20 m as fast as possible on a flat course. Two trials were given, with time to complete the course measured with photocells and timers. The better of these trials was used as the criterion score. A 2-min rest period separated the trials.
Quality of Life, Fatigue, and Psychological Distress
Questionnaires were administered at baseline and at weeks 8 and 16. Quality of life was measured using the Medical Outcomes Study-Short Form 30 (MOS-30), previously validated for use in HIV-infected individuals (50). It is self-administered and scored to generate 11 subscales using an established algorithm, such that higher values represent better health. These scales are then linearly transformed to range from 0 to 100. For each subscale, the score was considered missing if more than half of the questions were not answered for that subscale. A standard five-item self-reported sexual function instrument was administered concurrently and scored in the same manner as the MOS-30 subscales. The five items were as follows: 1) “I found it easy to achieve an erection when I wanted to,” 2) “I have lost interest in sex,” 3) “I found it difficult to sustain an erection when I wanted to,” 4) “I had problems achieving an orgasm,” and 5) “I am generally satisfied with the sex that I have.” Responses to each item were on a four-level Likert scale. Psychological distress was measured by the Depression, Anxiety, Stress Scale 21 (DASS-21) (29). This validated, self-administered instrument generates three subscale scores ranging from 0 to 42, with higher scores representing greater psychological distress.
Total testosterone was initially measured for screening by RIA in the Charles R. Drew hospital laboratory (normal range 300–1,200 ng/dl). To maintain blinding, all subsequent testosterone measures were made on frozen serum samples after the last participant had completed the study. These measurements were performed using an RIA (ICN Biomedical, Costa Mesa, CA) that has been validated against liquid chromatography-tandem mass spectrometry (8, 9). Serum was obtained for these measures in the morning on 2 days in the pretreatment period, and the values were numerically averaged. On-treatment testosterone levels were measured during weeks 12 and 16, 1 wk after the previous injection (nadir levels), and the values were averaged. Serum concentrations of luteinizing hormone (LH) and sex hormone-binding globulin (SHBG) were measured by time-resolved fluoroimmunoassays (DELFIA, Wallac). The detection limits of LH and SHBG assays were 0.05 IU/l and 1 nmol/l, respectively. Free testosterone was calculated from total testosterone, SHBG, and albumin concentration according to law of mass action equations as described by Vermeulen et al. (44).
Our primary end point was change in physical function, assessed by the modified Margaria stair climb, walking speed, and load-carrying ability. Body weight; lean body mass; muscle performance; and self-reported physical function, fatigue, and psychological distress were considered secondary end points. The primary analysis was by intention to treat, and all randomized participants were included in this analysis. For participants who dropped out in the treatment period, the last available observation was carried forward and included in the analysis. Participants were encouraged to return for follow-up outcome measurements, even if treatment had been discontinued early. If no follow-up data were available, baseline values were carried forward.
Statistical analyses were performed using SAS 9.1 (SAS Institute, Cary, NC). Continuous variables were assessed for normality of distribution, and the appropriate descriptive statistics were utilized. The baseline characteristics in Table 1 are expressed as means ± SD to characterize the study population. For comparisons between groups of normally distributed continuous variables, standard error is used to provide a measure of the precision of the sample mean. Where the assumption of normality was not met, the median and interquartile range are also given. For normally distributed continuous variables, change from baseline was assessed using two-tailed paired-sample t-tests, and comparison between groups was performed using two-tailed independent-sample t-tests. For nonnormal variables, statistical comparisons were made using nonparametric tests. Frequency counts between groups were compared using χ2 test unless individual cell counts were <5, in which case Fisher's exact test was used. For all statistical tests, P < 0.05 was considered significant.
Of 184 potential participants who were screened, 80 met eligibility criteria and 61 were randomized. Thirty-one subjects were assigned to the placebo group and 30 to the testosterone group. In the placebo group, 27 subjects completed the end-of-treatment measurements and four did not. Of those who did not complete the end-of-treatment measurements, one withdrew because of a death in the family, one because he was feeling ill with diarrhea and nasal congestion, and two for unknown reasons. In the testosterone group, 21 completed the end-of-treatment measurements and nine did not. Of those who did not complete the end-of-treatment measurements, one was withdrawn because of increased bilirubin, one was found to have multiple abscesses on day 4 of treatment, one withdrew after noting an aggressive mood, one found a job, one decided he was no longer interested, and five were lost to follow-up for unknown reasons. The drop-out rate was not significantly different between groups (P = 0.103). Baseline characteristics of the placebo- and testosterone-treated subjects were not significantly different (Table 1).
There were no significant differences in hormone levels between groups at baseline (Table 2). Total and free testosterone levels increased to a supraphysiological range during the 16-wk treatment period in the testosterone, but not placebo, group (P < 0.0001). As expected, serum LH concentrations decreased significantly in the testosterone, but not placebo, group (P < 0.0001). Serum SHBG levels also decreased significantly in the testosterone-treated, but not placebo-treated, men (P = 0.041).
Body Weight and Composition
There were no significant differences between groups at baseline in body weight, FFM, fat mass, or percent body fat (Table 1). There was a significant increase in FFM in the testosterone group (2.8 ± 0.5 kg, P < 0.0001) but no significant change in the placebo group. The change from baseline in FFM was significantly greater in the testosterone- than placebo-treated men (P < 0.0001; Fig. 1). Serum testosterone level during treatment was significantly correlated with the change in FFM (Pearson's correlation coefficient = 0.474, P = 0.0006). The testosterone group also experienced a significant increase in total body weight (1.8 ± 0.6 kg, P = 0.003) as well as significant decreases in whole body fat mass (1.0 ± 0.3 kg, P = 0.007) and percent body fat, which decreased from 17.5 ± 1.1% to 15.9 ± 1.0% (change = −1.5 ± 0.4%, P = 0.0003). In the placebo group, there were no significant changes in body weight or fat mass. The differences between groups in the changes in these variables were not statistically significant.
Muscle Performance and Physical Function Measures
Maximal voluntary strength in the leg press exercise increased significantly in the testosterone group but did not change significantly in the placebo group (Table 3). There was no significant difference in change from baseline between the two groups. There were no significant changes in leg press power in either group. There was a trend toward improved leg press fatigability in the testosterone group, with an increase in number of repetitions that was not significantly different from the placebo group but trended toward significance (P = 0.053).
The two groups did not differ significantly at baseline in their performance of physical tasks, such as walking, stair climbing, or load carrying (Table 3). There was no significant change in any of the three measures of physical function in either group during treatment. There were no significant differences between the two groups in the change from baseline in physical function measures.
Quality of Life and Psychological Distress
At baseline, there were no significant differences between groups in any of the 11 MOS-30 subscale scores or the sexual function score (Table 4). Testosterone administration was associated with significant improvements in the overall health (P = 0.020), health distress (P = 0.022), mental health (P = 0.0002), energy/fatigue (P = 0.004), and quality-of-life (P = 0.001) subscales. The improvements in the mental health and quality-of-life subscales were greater in the testosterone than placebo group (P = 0.036 and 0.012, respectively). There was no difference between groups in sexual function or other MOS-30 subscales.
Psychological distress was measured using the DASS-21 instrument, which is scored to generate three subscales, each ranging from 0 to 42, with higher scores corresponding to greater distress (Table 5). In the testosterone group, there were significant reductions from baseline in the stress, anxiety, and depression scales (within-group P = 0.0009, 0.0002, and 0.020, respectively); the changes in these scales were not significantly different from the placebo group.
Adverse Events and Safety Measures
Five subjects reported new or worse acne (4 in the placebo and 1 in the testosterone group) and two reported increased hair growth (1 in the testosterone and 1 in the placebo group). In the testosterone group, one subject reported gynecomastia as well as an increased bilirubin and another reported aggressive mood. Four patients received care in the emergency ward during the course of the study; of these, three were in the testosterone group. One was treated for epiglotitis, one for atypical chest pain, and one for multiple abscesses. In the placebo group, one subject was treated for a thumb laceration sustained while opening a can. All these events were considered unrelated to participation in the study. Hemoglobin increased significantly in the testosterone group (P < 0.0001) and to a greater extent than in the placebo group (P = 0.0005; Table 6). In two testosterone-treated men, hemoglobin increased to 17.3 and 17.8 g/dl, respectively, but did not reach the prespecified 18 g/dl threshold for treatment discontinuation. Serum PSA levels increased slightly from baseline in the testosterone group, with a significant difference compared with the placebo group (P = 0.033; Table 6), but an increase in PSA to >4 ng/ml was noted in only one subject, who was in the placebo group. In the testosterone group, there was a significant decrease in HDL cholesterol, but the between-group differences in change from baseline were not significant (Table 6). There were no significant changes from baseline in either group in total cholesterol, LDL cholesterol, or triglycerides (Table 6). There was a slight increase in CD4+ T lymphocyte count in the placebo group (P = 0.036) and no change in the testosterone group (Table 6). The proportion of participants with detectable HIV RNA on treatment was not significantly different between groups (P = 0.560).
Of the six prior placebo-controlled studies of androgen replacement in HIV-infected men with weight loss, four demonstrated significant increases in FFM (9, 19, 21, 42) and two did not (7, 18). Two meta-analyses of randomized trials in HIV-infected men with weight loss showed a difference in change in lean body mass between the testosterone and placebo group of 1.3 and 1.22 kg, respectively (4, 26). The dose of testosterone enanthate used in the present study was higher than that used in these previous trials; therefore, it is not surprising that the increment in FFM in our study was greater than that previously reported with replacement doses of testosterone. Testosterone administration has also been reported to increase lean body mass in healthy hypogonadal men, eugonadal men, and healthy older men with low or low-normal testosterone levels (4, 5). Studies in non-HIV-infected men have generally shown a significant decrease in fat mass with testosterone treatment; body weight often does not change because of reciprocal changes in lean and fat mass (8, 10, 11, 34). In contrast, testosterone trials in HIV-infected men with weight loss have shown no change in whole body fat mass or only modest reductions (4, 9). Androgen therapy has been shown to be comparable to other established anabolic therapies for HIV-associated weight loss, often with a more favorable side-effect profile (16, 42).
Our study comprehensively assessed testosterone's effects on measures of muscle performance (maximal voluntary strength, leg power, and muscle fatigability) and physical function (walking speed, stair-climbing power, and load-carrying ability). Despite significant gains in lean body mass, the improvements in maximal voluntary strength were modest in the testosterone group and not significantly different from the placebo group. Leg power, muscle fatigability, and the measures of physical function (stair-climbing power, walking speed, and load-carrying ability) showed no significant improvement. Self-reported physical function, as assessed by the MOS-30 physical function subscale, also did not change significantly. Our study had 80% power to detect a difference between groups in muscle performance and physical function measures as small as 6.2%. Therefore, small but significant differences could have been missed. Other studies in healthy older men also reported no detectable improvements in physical function measures, even when substantial improvements in lean body mass and maximal voluntary muscle strength were noted (9, 19, 21, 41, 42; unpublished observations). We controlled for the learning effect by familiarizing the subjects with the instrumentation and the technique. Additionally, the tests were repeated 2–7 days apart to ensure stability of baseline measurement. This approach was effective in minimizing the learning effect, as indicated by no significant change in muscle performance and physical function measures in the placebo-treated men. The performance-based measures of physical function used in the present study are susceptible to ceiling effects, and it is possible that the baseline function of these stable patients without functional limitations or disability exceeded the ceiling required for maximal performance in these tests of physical function.
Testosterone therapy was associated with significantly greater improvements in the mental health and quality-of-life MOS-30 subscales than placebo. There also were modest but significant improvements from baseline in the testosterone group in overall health, energy/fatigue, and health distress subscales. Several domains of mood (stress, anxiety, and depression) assessed by the DASS-21 instrument improved significantly with testosterone therapy. In healthy, young hypogonadal men, testosterone replacement improves positive aspects of mood and attenuates negative aspects of mood (45). Although some testosterone trials in men with depression and in HIV-infected men have reported improvements in depression measures, others have reported conflicting results. Mood disorders and fatigue are common among HIV-infected patients; therefore, improvements in mood, fatigue, mental health, and quality-of-life measures observed in our study are clinically relevant, but these findings need confirmation in adequately powered trials.
The supraphysiological dose of testosterone used in this trial was well tolerated, with very few adverse events. Consistent with a large body of clinical trial data (5), testosterone administration increased PSA and hemoglobin. There is concern about the potential effects of testosterone therapy on the growth of subclinical prostate cancer, especially in older men; this issue requires much larger and longer-term trials for adequate evaluation (5). The small decrease in HDL cholesterol in testosterone-treated men is consistent with earlier studies in HIV-infected (19, 21), as well as non-HIV-infected (5, 23), men. However, the small sample size and relatively short duration of the present study do not allow complete assessment of testosterone's safety in this population, particularly for >16 wk.
Our study had several limitations. The relatively small sample size did not allow power to confidently exclude smaller differences between groups. This was particularly true for adverse events, because the rates of these events were low. The relatively small sample size also may have limited the ability of randomization to achieve complete balance between the two groups on all baseline characteristics. The fact that there were no statistically significant differences between groups in the variables measured at baseline does not exclude the possibility of confounding factors. The study was limited to men; other testosterone studies in HIV-infected women using smaller doses of testosterone have shown no significant improvements in lean body mass, body weight, or muscle strength (13, 32). The drop-out rate of ∼20%, although high, is consistent with other intervention trials in HIV-infected patients. The study was designed to include men with total testosterone <400 ng/dl, and all participants qualified under this criterion at the time of screening. However, the mean starting testosterone, as reported in Table 1, was >400 ng/dl in both groups. We attribute this discrepancy to the phenomenon of regression to the mean and to interassay and biological variation in testosterone levels.
We adopted a strict intention-to-treat approach, such that baseline values were carried forward for all participants who did not complete follow-up measures. Although this approach provides a rigorous efficacy evaluation, it also may have reduced the observed change from baseline and difference between groups for true treatment effects. Our findings of improved mood and quality-of-life scores in the testosterone group should be interpreted cautiously, because these were secondary end points. Several secondary end points were evaluated on the basis of specific hypotheses; we did not perform specific statistical adjustments for multiple testing, but we provide actual P values for all comparisons to allow for accurate interpretation of our results. Highly significant differences were observed for several quality-of-life and psychological distress subscales, but others showed less significant change.
Larger studies of longer duration, specifically designed to confirm the effects of testosterone therapy on mood, fatigue, and quality of life, are required. Our results do not justify a general recommendation for testosterone therapy in HIV-infected men who do not have functional limitations.
This study was supported primarily by National Institutes of Health (NIH) Grant 1RO1 DK-49296 and additionally by NIH Grant 1RO1 DK-59627-01 and Research Centers in Minority Institutions Grants P20 RR-11145 and G12 RR-03026 and a grant from the Universitywide AIDS Research Program DrewCares HIV Center.
We thank the staff of the Charles R. Drew Clinical Research Center for help with these studies. The infrastructural support provided by the University of California AIDS Research Program-supported DrewCares HIV Center is gratefully acknowledged. Savient Pharmaceuticals provided testosterone enanthate.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2008 by American Physiological Society