Schematic overview of the study design and summary of enrollment.
Dolan S, Wilkie S, Aliabadi N, Sullivan MP, Basgoz N, Davis B, Grinspoon S. Effects of Testosterone Administration in Human Immunodeficiency Virus–Infected Women With Low WeightA Randomized Placebo-Controlled Study. Arch Intern Med. 2004;164(8):897-904. doi:10.1001/archinte.164.8.897
Copyright 2004 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2004
The prevalence of human immunodeficiency virus (HIV) disease is increasing among women, many of whom remain symptomatic with low weight and poor functional status. Although androgen levels may often be reduced in such patients, the safety, tolerability, and efficacy of testosterone administration in this population remains unknown.
A total of 57 HIV-infected women with free testosterone levels less than the median of the reference range and weight less than 90% of ideal body weight or weight loss greater than 10% were randomly assigned to receive transdermal testosterone (4 mg/patch) twice weekly or placebo for 6 months. Muscle mass was assessed by urinary creatinine excretion. Muscle function was assessed by the Tufts Quantitative Muscle Function Test. Treatment effect at 6 months was determined by analysis of covariance. Results are mean ± SEM unless otherwise specified.
At baseline, subjects were low weight (body mass index [calculated as weight in kilograms divided by the square of height in meters] 20.6 ± 0.4), with significant weight loss from preillness maximum weight (18.7% ± 1.2%), and demonstrated reduced muscle function (upper and lower extremity muscle strength, 83% and 67%, respectively, of predicted range). Testosterone treatment resulted in significant increases in testosterone levels vs placebo (total testosterone: 37 ± 5 vs –2 ± 2 ng/dL [1.3 ± 0.2 vs −0.1 ± 0.1 nmol/L] [P<.001]; free testosterone: 3.7 ± 0.5 vs –0.4 ± 0.3 pg/mL [12.8 ± 1.7 vs −1.4 vs 1.0 pmol/L] [P<.001]) and was well tolerated, without adverse effects on immune function, lipid and glucose levels, liver function, or body composition or the adverse effect of hirsutism. Muscle mass tended to increase (1.4 ± 0.6 vs 0.3 ± 0.8 kg; P = .08), and shoulder flexion (0.4 ± 0.3 vs −0.5 ± 0.3 kg; P = .02), elbow flexion (0.3 ± 0.4 vs –0.7 ± 0.4 kg; P = .04), knee extension (0.2 ± 1.0 vs –1.7 ± 1.3 kg; P = .02), and knee flexion (0.7 ± 0.5 vs 0.3 ± 0.7 kg; P = .04) increased in the testosterone-treated compared with the placebo-treated subjects.
Testosterone administration is well-tolerated and increases muscle strength in low-weight HIV-infected women. Testosterone administration may be a useful adjunctive therapy to maintain muscle function in symptomatic HIV-infected women.
Although the percentage of women among newly diagnosed AIDS cases has increased from 6.7% in 1986 to 32% in 2001,1,2 relatively little research has been done to investigate unique gender-specific disease factors and therapeutic interventions in this growing population of patients. Human immunodeficiency virus (HIV)–infected women may be less likely to begin or adhere to highly active antiretroviral therapy and less likely to use health services appropriately because of inadequate resources or other socioeconomic obstacles, which may contribute to increased disease severity in this population.1,3,4 Prior studies have demonstrated reduced androgen levels among HIV-infected women, particularly among those with weight loss.5,6 In preliminary, short-term, dose-ranging studies, we have shown that testosterone administration was safe in HIV-infected women with weight loss.7 However, the longer-term tolerability of testosterone and effects on muscle function in this population remain unknown.
Patients were recruited between 1998 and 2001 from the Massachusetts General Hospital Multidisciplinary HIV Clinic, Boston, and community-based practices and through newspaper advertisements. Patients were eligible based on the following criteria: female between the ages of 18 and 45 years; documented HIV infection; free testosterone level by equilibrium dialysis less than 3.0 pg/mL (<10.4 pmol/L) (median of the Endocrine Sciences [Calabasas Hills, Calif] reference range); weight less than 90% of ideal body weight or weight loss greater than 10% of weight before illness; or weight less than 100% of ideal body weight with weight loss greater than 5% of ideal body weight. Patients were excluded if pregnant or actively seeking pregnancy; requiring total parenteral nutrition; receiving androgen, estrogen, progestational derivatives, or megestrol acetate within 3 months of the study; or initiating new antiretroviral therapy within 6 weeks of the study. Of the 57 enrolled subjects, 7 had been in a prior study of testosterone administration, but none had received testosterone for over 18 months before study enrollment. Patients with an aspartate aminotransferase level higher than 125 U/L and/or clinically significant liver disease, creatinine level higher than 2.0 mg/dL (176.8 µmol/L), or hemoglobin level lower than 8.0 g/dL were excluded, as were patients with active substance or alcohol abuse.
All patients gave written informed consent, and the study was approved by the Human Research Committee of the Massachusetts General Hospital and the Committee on the Use of Humans as Experimental Subjects at the Massachusetts Institute of Technology, Cambridge. Eligibility was determined at a screening visit, at which free testosterone level, aspartate aminotransferase level, human chorionic gonadotropin level, weight, and weight history were determined.
Eligible patients were randomly assigned to receive either active testosterone transdermal delivery systems (Testosterone TTS [4.1 mg/patch, estimated delivery rate 150 µg/d]; Watson Pharmaceuticals Inc, Salt Lake City, Utah) twice weekly or identical placebo patches provided by the manufacturer. Randomization was stratified by weight (<90% or ≥90% of ideal body weight) in a permuted block design, with a block size of 4 based on randomly generated numbers. Randomization was performed by the Massachusetts General Hospital Pharmacy. All investigators and patients were blinded to drug assignment. Each patient underwent a directed history screening for testosterone-dependent effects, including symptoms of virilization and libido, and physical examination and evaluation for signs of virilization and hirsutism using the Lorenzo rating scale,8 a modification of the Ferriman Gallwey hirsutism score.
For women with regular menses, the baseline evaluation occurred within the early follicular phase of the menstrual cycle (days 1-7). Subjects were placed on an isocaloric, meat-free diet 3 days prior to the study and were admitted for a 24-hour overnight admission. A 24-hour urine collection was performed for determining creatinine levels. Fasting blood samples were obtained at approximately 8:30 AM for determining levels of testosterone, free testosterone, sex hormone–binding globulin, estradiol, luteinizing hormone, follicle-stimulating hormone, dehydroepiandrosterone sulfate, insulinlike growth factor I (IGF-I), lipids, glucose, aspartate aminotransferase, and alkaline phosphatase and for determining CD4 count and HIV viral load. Weight was measured and body composition assessed. Muscle strength of the upper and lower extremities was determined from the Tufts Quantitative Muscle Function Test. Muscle function was assessed by a 6-minute walk test. Caloric intake was determined by food record. Subjects returned for monthly visits, including physical examination, urine pregnancy test, and measurement of 8:30 AM fasting blood total and free testosterone levels and weight. Time of last testosterone application was recorded. Subjects recorded menses in a monthly menstrual diary. A new month's supply of study drug was given out only after documentation of a negative pregnancy test result, and used patches were collected at each visit. At the 6-month visit, subjects were evaluated and baseline testing was repeated.
All samples from the same patient were run in duplicate in the same assay. The free testosterone concentration was determined as the product of the percentage of free testosterone (measured by equilibrium dialysis) and the total testosterone concentration (Endocrine Sciences). The intra-assay coefficient of variation (CV) of free testosterone is 6.9%, and the intra-assay CV for total testosterone is less than 8.1%. The intra-assay CVs were developed using pooled serum samples covering the range of the assay. The normal range for total testosterone is 10 to 55 ng/dL (0.3-1.9 nmol/L) and for free testosterone is 1.1 to 6.3 pg/mL (3.8-21.8 pmol/L) in adult women. The interassay CV for testosterone is 8% to 15% and for free testosterone, 8.9% to 11.9%. The sensitivity of the total testosterone assay is 3 ng/dL (0.1 nmol/L). The sensitivity of the determination of the percentage of free testosterone by this method is 0.1%. Radioimmunoassays were performed for measuring levels of dehydroepiandrosterone sulfate (intra-assay CV, 3.8%-5.3%; interassay CV, 6.3%-11.0%) (Diagnostic Products Corporation, Los Angeles, Calif), luteinizing hormone (intra-assay CV, 2.6%; interassay CV, 4.5%-5.4%) (Nichols Institute, San Juan Capistrano, Calif), follicle-stimulating hormone (intra-assay CV, 1.6%-2.3%, interassay CV, 3.2%-3.8%) (Nichols Institute), and estradiol (intra-assay CV, 6.5%-8.9%; interassay CV, 7.5%-12.2%) (Diagnostic Systems Laboratories, Webster, Tex), and IGF-I (intra-assay CV, 3.9%-4.1%; interassay CV, 2.8%-4.8%) (Diagnostics Systems Lab). Sex hormone–binding globulin level was measured by immunoradiometric assay, with an intra-assay CV of less than 4% and an interassay CV of 7.8% to 10.6% (Endocrine Sciences). Cholesterol, low-density lipoprotein, high-density lipoprotein, triglyceride, hemoglobin, creatinine, glucose, and liver function tests were performed as previously described.9
CD4 counts were measured by flow cytometry using FACS Lysing Solution and a FACScan Analyzer (Becton-Dickinson Immunocytometry Systems, San Jose, Calif; normal range, 348-1456 CD4 cells/µL). Human immunodeficiency virus 1 RNA was quantified using a sandwich nucleic acid hybridization procedure (the Quantiplex HIV-RNA Assay [Chiron Corporation, Emeryville, Calif]), with a sensitivity of 50 copies/µL.
On the first day of each visit after an overnight fast, weight was determined on a calibrated scale and height was determined by stadiometer. Percentage of ideal body weight was calculated based on standard height and weight tables.10 Fat mass was determined by dual energy x-ray absorptiometry using a Hologic-4500 densitometer (Hologic Inc, Waltham, Mass), with a precision of 3%.11 Muscle mass was determined by urinary creatinine excretion on a meat-free diet as previously reported.12 Cross-sectional abdominal computed tomographic scanning was performed to assess subcutaneous and visceral abdominal fat. A lateral scout image was obtained to identify the level of the L4 pedicle, which served as a landmark for the single-slice image.13 Subjects were instructed on completion of a 4-day food record, which was analyzed for total calorie, protein, fat, and carbohydrate content (Minnesota Nutrition Data Systems [version 8A/2.6], Minneapolis). Resting energy expenditure was determined after an overnight fast by indirect calorimetry (Vmax 29; SensorMedics Inc, Yorba Linda, Calif). Physical activity was assessed by a questionnaire adapted from Kohl et al.14
Menstrual status was assessed by self-report (recall over 3 months prior to initiation of study) and by menstrual diary during the study. Subjects who had not had a period within the last 3 months prior to study initiation were categorized as amenorrheic, subjects with 1 or 2 periods in the last 3 months were characterized as oligomenorrheic, and subjects with 3 periods in the 3 months prior to the initiation of the study were categorized as eumenorrheic. The number of menstrual periods occurring during the 6 months of treatment were tabulated from the menstrual diary.
Subjects completed a Beck depression inventory at baseline and end of study to determine the effects of testosterone on depression indexes.15
Upper and lower extremity muscle strength were determined using the Tufts Quantitative Muscle Function Test.16,17 Peak isometric force of (1) shoulder flexion, (2) shoulder extension, (3) elbow flexion, (4) elbow extension, (5) knee flexion, (6) knee extension, (7) dorsiflexion, and (8) grip were measured on the best of 2 repetitions, during which subjects held a maximum contraction for 5 seconds. The predicted normal range for each subject was determined based on regression equations from an established database of 273 healthy female subjects previously tested using the Tufts Quantitative Muscle Function Test protocol.18 These equations factored in age, height, and weight to determine predicted strength for each muscle group tested. Percentages of predicted values were calculated by dividing the raw scores by the predicted scores derived from the regression equations. Upper and lower percent predicted scores were determined by averaging percent predicted scores for shoulder flexion, shoulder extension, elbow flexion, elbow extension, and grip strength (upper extremity) and knee flexion and extension and ankle dorsiflexion (lower extremity) as previously described.19 Distance walked over 6 minutes was assessed in the 6-minute walk test.20
Investigational active and placebo patches were supplied by Watson Pharmaceuticals Inc and were identical in size, shape, and consistency. The testosterone transdermal delivery system is a proprietary alcohol-free matrix patch (18 cm2) containing testosterone, sorbitan monooleate as a permeation enhancer, and a hypoallergenic acrylic adhesive. Each active patch contains 4.1 mg of testosterone and was expected to deliver testosterone at a nominal delivery rate of 150 µg/d over a 3- to 4-day application period (Watson Pharmaceuticals, unpublished data, July 1996).
Baseline comparisons were made by t tests between the groups. Categorical variables were compared by χ2 analysis. Treatment effect at 6 months was estimated using analysis of covariance. The measurement obtained at 6 months was the outcome variable, treatment assignment was the main effect, and baseline measurement of the variable was used as a covariate. The analysis was intent to treat using available data and last observation carried forward for subjects without end-of-study data. When interim data were not available, data from the baseline visit were carried forward and used in the analysis. The primary end points were muscle mass and muscle function. An estimated sample size of 50 patients was determined to be necessary to detect an increase in muscle mass of 2.0 kg with 80% power and a 2-tailed α level of .05.
Multivariate analysis of covariance was used to assess the overall treatment effect on strength testing in a combined analysis of individual strength tests. The measurement obtained at 6 months in the shoulder flexion, shoulder extension, elbow flexion, elbow extension, knee flexion, knee extension, dorsiflexion, and grip tests was included in the model; treatment assignment was the main effect; and baseline measurement was used as a covariate. An overall effect on strength was demonstrated (P = .02) by multivariate analysis of covariance, and determination of treatment effect for individual tests was made by analysis of covariance, controlling for baseline as a covariate.
Outlier analysis was performed using the Dixon criteria,21 and extreme outliers were excluded from the analysis. Significance is defined as P <.05. Statistical analyses were performed using SAS JMP (SAS Institute, Cary, NC). A data safety monitoring board met every 3 months to review any adverse events occurring in the study. Results are mean ± SEM, unless otherwise reported.
Seventy-nine patients were screened for the study, 66 met the inclusion criteria, and 57 patients were randomized and received treatment with testosterone (n = 29) or placebo (n = 28). Of the 29 patients randomized to testosterone and the 28 randomized to placebo, 27 and 25, respectively, completed the study (Figure 1), for an overall dropout rate of 8.7%. Compliance with study medication based on count of used patches was 95% in the placebo group and 96% in the testosterone group (P = .73). No differences were observed between eligible subjects enrolling (n = 57) and not enrolling (n = 9) in screening testosterone, free testosterone, body mass index (BMI), weight, weight loss, or safety parameters (data not shown). Dropouts (n = 5) were not different from completers (n = 52) in testosterone, free testosterone, weight, BMI, muscle mass, or strength, but demonstrated higher viral load (median [interquartile range], 74 509 [24 410-93 750] vs 1234 [50-12 824] copies/mL; P = .04 by median rank test).
At baseline (Table 1 and Table 2) subjects were low weight (BMI [calculated as weight in kilograms divided by the square of height in meters], 20.6 ± 0.4), with a CD4 count of 317 ± 32 cells/µL and an HIV viral load of 15 765 ± 3553 copies/mL. Baseline variables did not differ between testosterone and placebo groups (age, 38 ± 1 vs 38 ± 1 years [P = .56]; percentage of white subjects, 62% vs 50% [P = .12]; BMI, 21.0 ± 0.5 vs 20.2 ± 0.6 [P = .36]; and weight loss, 16.9% ± 1.8% vs 20.6% ± 1.6% [P = .14]). The percentage of subjects (testosterone vs placebo treated) receiving a protease inhibitor (68% vs 81%; P = .28), nucleoside reverse transcriptase inhibitor (96% vs 89%; P = .30), and nonnucleoside reverse transcriptase inhibitor (50% vs 56%; P = .68) did not differ between the treatment groups. Subjects demonstrated reduced strength at baseline for all muscle groups tested as well as for the upper and lower extremity composite scores (upper [83% of predicted range] and lower [67% of predicted range]; Table 3). Of the subjects, 13 (23%) were amenorrheic, 3 (5%) were oligomenorrheic, 36 (63%) were eumenorrheic, and 5 (9%) had previously undergone hysterectomy at baseline. Menstrual status did not differ by treatment and placebo groups (eumenorrheic, 19 [66%] vs 17 [61%]; oligomenorrheic, 8 [28%] vs 8 [29%]; and hysterectomy, 2 [7%] vs 3 [11%]; P = .26).
Total testosterone and free testosterone levels increased significantly in the testosterone- vs placebo-treated subjects at 6 months (Table 2). Free testosterone levels achieved at 6 months were within the normal range in most subjects receiving testosterone. Among the testosterone-treated subjects, end-of-study free testosterone levels were distributed as follows: 0% were below the normal range, 12% were in the lower half of the normal range, 64% were in the upper half of the normal range, and 24% were above the normal range (maximum level, 11.0 pg/mL [38.1 pmol/L]). In contrast, end-of-study free testosterone levels among the placebo-treated subjects were distributed as follows: 29% were below the normal range, 58% were in the lower half of the normal range, 13% were in the upper half of the normal range, and 0% were above the normal range. The mean ± SEM time of testosterone application prior to blood drawing for testosterone- vs placebo-treated was 139 ± 11 vs 132 ± 13 minutes, respectively (P = .68) and did not correlate with the testosterone level (r = 0.024; P = .89).
Insulinlike growth factor I level increased significantly in response to testosterone compared with placebo (46 ± 16 vs –13 ± 20 ng/mL [6.0 ± 2.1 vs 1.7 ± 2.6 nmol/L]; P = .045, testosterone vs placebo). The changes in estradiol, gonadotropin, dehydroepiandrosterone sulfate and sex hormone–binding globulin levels did not differ significantly between treatment groups at 6 months (Table 2). No significant differences in lipid or glucose levels were observed between the groups.
Weight did not change significantly between the treatment groups (Table 2). There was a trend toward increased muscle mass in the testosterone compared with the placebo-treated group (Table 2). Total body fat, abdominal subcutaneous fat, and abdominal visceral fat did not change significantly between the treatment groups. Although caloric intake and resting energy expenditure did not change significantly between the groups (Table 2), testosterone-treated patients tended to increase caloric intake and resting energy expenditure relative to placebo-treated patients. Spontaneous physical activity did not change between groups (−0.1 ± 0.1 vs –0.1 ± 0.3 h/wk; P = .31).
A significant overall effect of testosterone on strength was seen by multivariate analysis of covariance (P = .02). Left shoulder flexion (P = .02), elbow flexion (P = .04), knee extension (P = .02), knee flexion (P = .04), and overall lower extremity (P = .02) scores increased significantly in the testosterone- compared with the placebo-treated subjects (Table 3). A positive trend was also observed for increased shoulder extension (P = .07) and overall upper extremity score (P = .07) in the testosterone- vs placebo-treated subjects. Distance on the 6-minute walk test decreased 27.0 ± 27.6 m in the placebo-treated patients and increased 1.8 ± 15.6 m in the testosterone-treated group (P = .17). Strength (r = 0.30; P = .04 [knee extension vs testosterone]), but not muscle mass, correlated with the testosterone level at the end of the study. Strength also correlated with muscle mass (r = 0.36; P = .008 [knee extension vs muscle mass]) and 6-minute walking distance (r = 0.50; P<.001 [knee extension vs distance walked]) at the end of the study. In contrast, strength did not correlate with weight (r = 0.11; P = .42 [weight vs knee extension]).
CD4 count and HIV viral load did not change significantly between the treatment groups (Table 2). The percentages of patients (testosterone vs placebo treated) using protease inhibitor (12 [44%] vs 13 [52%]; P = .59), nucleoside reverse transcriptase inhibitor (21 [78%] vs 15 [60%]; P = .17), and nonnucleoside reverse transcriptase inhibitor (9 [33%] vs 8 [32%]; P = .92) was similar at the end of the study between treatment groups. Seventeen (31%) of the subjects switched classes of antiretroviral agents during the study and were evenly divided between the treatment groups (8 were receiving testosterone and 9 were receiving placebo; P = .78). Four patients discontinued antiretroviral therapy (2 in each group); 7 initiated protease inhibitor therapy (3 receiving testosterone and 4 receiving placebo); and 8 initiated nucleoside reverse transcriptase inhibitor therapy (5 receiving testosterone and 3 receiving placebo).
Hirsutism scores, liver function tests, hemoglobin level, and menstrual status did not change significantly between treatment groups (Table 2). No subjects demonstrated deepening of the voice, temporal balding, or significant acne. Libido increased in 4 of the testosterone-treated subjects vs 2 of the placebo-treated subjects (P = .67). The total number of periods in each treatment group was similar during the course of the study (3.9 ± 0.4 vs 4.3 ± 0.5 cycles per 6 months; P = .52). The completion rate was similar in the 2 treatment groups. Three patients in the placebo-treated group and 1 patient in the testosterone-treated group were unwilling to complete the protocol and withdrew from the study voluntarily without associated adverse events. One patient in the testosterone-treated group with a history of depression committed suicide during the study, but the Beck depression score did not change significantly between the groups. No other dropouts were due to adverse events, and the 4 remaining dropouts resulted from inability and willingness to return for scheduled study visits.
The prevalence of HIV disease is increasing among women in the United States and worldwide.1 Wanke et al22 demonstrated that wasting, low weight, and weight loss remain relatively common even in the era of highly active antiretroviral therapy. In contrast to men, relatively little is known about gender-specific therapies to improve functioning and health for symptomatic HIV-infected women. Decreased androgen levels have previously been shown among HIV-infected women with low weight,5,6 and short-term administration of relatively low-dose testosterone administration was previously shown to be safe in HIV-infected women.7 However, the effects of testosterone administration over a longer duration of treatment in terms of tolerability, body composition, and strength are not known in this population.
Subjects were selected for this study based on low weight and weight loss. Subjects were ambulatory. Of the subjects, 52 (91%) were receiving antiretroviral therapy at the start of the study, but viral load was undetectable in only 14 (25%). In this study, we report the novel finding of reduced strength and muscle function in this population of female patients. For example, baseline composite upper and lower extremity scores were 83% and 67%, respectively, of the predicted normal ranges based on gender, weight, and age (Table 3). We used a standardized strength assessment series that we have used in prior studies and that has been shown to predict functional status in other populations of patients. The series tests physical functioning in critical muscle groups and can be used to assess overall strength in comparison with a well-established normal range in healthy female subjects and also to determine change over time. Prior studies using survey or questionnaire data have suggested reduced physical functioning in patients with HIV disease,23,24 but to our knowledge, this is the first report exclusively among women to document reductions in actual strength and physical functioning.
In prior studies, we have demonstrated reduced androgen levels in HIV-infected women, particularly those with weight loss and low weight. Indeed, almost half of the subjects screened for this study demonstrated androgen levels below the normal range. For consistency, baseline blood tests were timed to the early follicular phase for all eumenorrheic subjects, and it is possible that higher baseline testosterone levels would be seen if levels had been assessed in the late follicular or luteal phases of the menstrual cycle. The present study was not limited to patients with absolute reductions in androgen levels. Rather, the study was designed, based on our prior study, to enroll subjects with a free testosterone level lower than the median of the normal range to allow safe administration of testosterone to symptomatic HIV-infected women. In this regard, our prior short-term dose-ranging study suggested that we could successfully identify a relatively androgen-deficient, symptomatic population of HIV-infected women using a cutoff of the median of the reference range, in whom low-dose testosterone treatment achieved generally physiologic testosterone levels and was well tolerated in the short-term. We therefore chose a similar entry criteria for our subsequent longer-term study. We chose this design to make the results of the study more generalizable to a broader population of HIV-infected women with low weight and weight loss and simultaneously to assure patient safety. Different results might be seen in studies with enrollment limited to women with absolute reduction in androgen levels below the normal range, and future studies of testosterone administration to this population will also be important.
In this study, we used a natural testosterone preparation, as opposed to an oral anabolic androgen preparation. Testosterone levels were raised, on average, to the upper end of the normal range and remained within the normal range for most patients. In this regard, neither gonadotropins nor sex hormone–binding globulin levels decreased significantly, suggesting that although testosterone levels were significantly increased, such changes generally within the physiologic range were not of a sufficient magnitude to cause a change in these parameters. Estradiol levels were not increased, which is consistent with the findings of prior studies using this preparation in oophorectomized women,25 suggesting that little of the testosterone was aromatized to estrogen. Thus, use of the testosterone as administered in the present study would therefore not be expected to be associated with estrogen-related adverse effects. Dosing was well tolerated, without significant dropout or adverse effects. One patient with a history of depression receiving testosterone committed suicide during the study, but the Data Safety Monitoring Board believed that it was unlikely that testosterone contributed to this adverse event. Furthermore, adverse effects of testosterone on depression indexes were not seen. Neither liver function test values nor hirsutism scores increased, and menstrual function was not adversely affected. Testosterone did not result in worsened glucose homeostasis or adverse effects on lipid levels. Furthermore, no adverse effects of testosterone on total body, visceral, or subcutaneous fat were observed at the dose used in this study. No effect on immune parameters was seen. Taken together, data from the present study and our prior pilot study7 suggest that administration of natural testosterone at a nominal delivery rate of 150 µg/d is well tolerated and safe in this population of patients for periods of up to 6 months.
Testosterone administration had a significant effect on strength at multiple sites in this symptomatic population. Strength increased for most muscle groups in patients receiving testosterone, whereas strength decreased at many sites in the upper and lower extremities in the placebo-treated subjects over the course of the 6-month study, resulting in significant differences between the testosterone and placebo-treated subjects in this randomized, double-blind study. Similarly, distance walked in the 6-minute walk test, a measure of functional status, decreased more in the placebo-treated patients and was correlated with strength. Reductions in physical function over time have been reported in HIV-infected patients.26 Data from our study suggest that testosterone may help to prevent a decline in muscle function in low-weight HIV-infected women. The effects of testosterone on muscle strength observed in this study may be direct or indirect and related to improved energy or ability to concentrate on the task at hand. Given the poor baseline scores on muscle function testing, substantial improvement in muscle function in the testosterone-treated subjects relative to the placebo-treated subjects is clinically important in this understudied population of patients. Further studies are needed to determine the long-term effects and mechanisms of testosterone effects on physical functioning in this population of patients.
Although the changes did not reach statistical significance, muscle mass increased by 1.1 kg more in the testosterone-treated compared with placebo-treated patients and correlated significantly with strength testing. The cause of improved strength in the absence of statistically significant increases in muscle mass is not clear but may relate to a partial improvement in muscle mass as well as direct or indirect (eg, through IGF-I) effects of testosterone. Testosterone levels correlated significantly with strength testing. Alternatively, subjects receiving the testosterone treatment may have had more energy and tried harder on the testing.
The change in weight was not significant between the groups. Although testosterone-treated patients tended to increase caloric intake, a relative imbalance in energy expenditure associated with improved strength and functionality might reduce overall weight gain associated with testosterone. Relatively few studies of natural testosterone administration have been performed in women. Results from the present study suggest that low-dose testosterone treatment improves strength and is well tolerated during prolonged administration in HIV-infected women. Larger doses of testosterone might result in a greater anabolic effect but may not be tolerated in the long-term in women. Furthermore, the study was powered to detect a greater change in muscle mass than what was observed between the groups, and larger studies may be necessary to detect significant changes in muscle mass in response to physiologic testosterone administration.
Insulinlike growth factor I levels increased significantly in response to testosterone. Insulinlike growth factor I levels are known to decrease with weight loss and undernutrition,27 and testosterone levels are known to increase growth hormone and IGF-I secretion in men.28- 30 Both dehydroepinadrosterone31 and oxandrolone32 have been shown to increase the IGF-I level in women, but prior studies have not investigated the effects of natural testosterone on IGF-I levels in women. Insulinlike growth factor I levels were relatively reduced at baseline in our subject population with low weight (eg, IGF-I level was decreased below the normal range in 19 subjects [34%]). Insulinlike growth factor I is an anabolic hormone,33 and an increased IGF-I level may contribute to increased muscle mass and strength in response to testosterone.
Gender-specific treatment options to improve functional status and strength for symptomatic HIV-infected women are limited. Prior studies assessing the effects of anabolic therapies have primarily recruited men and were not tailored therapies for women. Therefore, although anabolic steroids,34 megestrol acetate,35 and growth hormone36 have been shown to increase weight34- 36 and lean body mass36 in men, such therapies have not been tested in women. Such therapies may be associated with adverse effects of fluid retention, insulin resistance, fat accumulation, and hepatotoxic effects. Furthermore, these therapies are not physiologic at the recommended doses and are without clearly proven benefits on functional status in women. In contrast, testosterone at the dose administered in the present study was well tolerated and improved muscle strength relative to placebo treatment. In this regard, this is the first study, to our knowledge, to demonstrate significant effects on strength in women with HIV-related weight loss and decreased functional status.
Data from this study demonstrate that testosterone administration is well tolerated and results in significant beneficial effects on strength in HIV-infected women. As such, natural testosterone administered at low doses is a novel therapy of potential benefit for HIV-infected women with low weight and weight loss. These results may relate to a not heretofore demonstrated effect of testosterone to prevent or partially attenuate an ongoing decline in muscle function in this population. Further long-term studies are necessary to determine the optimal testosterone dosing strategy in HIV-infected women. These data also suggest the need for further studies of testosterone administration in other groups of symptomatic women with low weight and decreased muscle function.
Corresponding author: Steven Grinspoon, MD, Program in Nutritional Metabolism, Massachusetts General Hospital, LON207, Boston, MA 02114 (e-mail: firstname.lastname@example.org).
Accepted for publication May 20, 2003.
This study was supported by grant NIH DK-54167 from the National Institutes of Health, Bethesda, Md, and the Mary Fisher Clinical AIDS Research and Education (CARE) Fund, Birmingham, Ala.
We wish to thank the nursing and bionutrition staff of the Massachusetts General Hospital General Clinical Research Center and Colleen Corcoran, ANP, for their dedicated patient care and Jeff Breu, BS, of the MIT General Clinical Center Core Laboratory, Cambridge, Mass, for his help in the performance of radioimmunoassays.