Effects of nandrolone decanoate compared with placebo
or testosterone on HIV-associated wasting
 
 
 
Effects of nandrolone decanoate compared with placebo
or testosterone on HIV-associated wasting.  
 
  HIV Medicine
Volume 7 Page 146 - April 2006
 
J Gold1, MJ Batterham2, H Rekers3, MK Harms3, TBP Geurts3, PME Helmyr3, J Silva de Mendonca4, LH Falleiros Carvalho5, G Panos6, A Pinchera7, F Aiuti8, C Lee9, A Horban10, J Gatell11, P Phanuphak12, W Prasithsirikul13, B Gazzard14, M Bloch15 and SA Danner16 for the E-1696 Study Investigators
 
"...This study is the first large multicentre randomized placebo-controlled study to demonstrate the effectiveness of nandrolone decanoate over placebo in increasing fat free mass, weight and body mass index in HIV-positive males with wasting. In addition, fortnightly treatment with 150 mg nandrolone decanoate was superior to fortnightly treatment with 250 mg testosterone for increasing weight, increasing body mass index and (insignificantly) increasing fat-free mass. Subjectively, nandrolone was superior to both placebo and testosterone for improving perception of treatment benefit and recovery from symptoms..."
 
Table 3 Summary statistics of changes from baseline to endpoint during the initial treatment period in anorexia/cachexia results (intention-to-treat analysis)
"The use of this instrument in the present study clearly demonstrated a significant benefit of nandrolone over both placebo and testosterone for subjective improvements in recovery of symptoms from anorexia/cachexia and patient perception of benefit. This questionnaire was designed to target the aspects of quality of life that are specific to a nutritional/appetite intervention, such as body image, enjoyment of food, appetite per se and the actual intervention. These areas are not covered in traditional quality of life questionnaires and therefore a treatment effect may be missed."
 

parameter-1.gif

BACRI, Bristol-Myers Anorexia Cachexia Recovery Instrument; CI, confidence interval; SD, standard deviation.
 
ABSTRACT
Objectives

Current research is unclear about the most effective pharmacological agents for managing the loss of weight and fat-free mass common in HIV/AIDS. The aim of this study was to compare nandrolone decanoate with placebo and testosterone.
 
Methods
The study was a multicentre randomized double-blind placebo-controlled trial. Three hundred and three adult HIV-positive male patients with a weight loss of 5-15% in the last 12 months, or a body mass index of 17-19 kg/m2, or a body cell mass/height ratio lower than 13.5 kg/m, were randomly assigned to receive nandrolone decanoate (150 mg), testosterone (250 mg) or placebo intramuscularly every 2 weeks for 12 weeks. Fat-free mass, weight, immune markers and perception of treatment were the main outcome measures.
 
Results
Treatment with nandrolone resulted in significantly greater increases in fat-free mass [mean increase 1.34 kg; 95% confidence interval (CI) 0.60; 2.08 kg] and in weight (mean increase 1.48 kg; 95% CI 0.82; 2.14 kg) compared with placebo. The mean increase in weight with nandrolone of 1.00 kg (95% CI 0.27; 1.74 kg) when compared with testosterone was significant, although the difference in fat free mass did not reach significance (mean increase 0.69 kg; 95% CI-0.13; 1.51 kg). Patient perception of benefit was significantly greater in the nandrolone group when compared with both the placebo and the testosterone groups.
 
Conclusions
Treatment with nandrolone decanoate increased body weight when compared with placebo and testosterone. Nandrolone decanoate treatment resulted in greater increases in fat-free mass than placebo and demonstrated a trend for a significant increase when compared with testosterone.
 
LIMITATIONS
The results reported in this study are relatively short-term reporting changes over a 12-week period. Grinspoon et al [34] have conducted a long-term study of the use of testosterone showing sustained significant increases in fat-free mass and smaller changes in weight. However, the long-term use of nandrolone requires investigation, as clinically patients are using these treatments long-term where the efficacy has not yet been demonstrated.
 
Clinically, this study demonstrates unequivocally the superiority of nandrolone decanoate over placebo for increasing fat-free mass, weight, body mass index, and perception of treatment benefit when compared with placebo. In addition, nandrolone was superior to testosterone for increasing weight and perception of treatment benefit and tended to increase fat-free mass. This study provides important information for decision making in clinical practice. Further research should investigate the longer term effectiveness of these agents.
 
The use of bioelectrical impedance is a limitation to this study. The study was conducted in many centres where the use of other body composition methodology such as dual energy X-ray absorptiometry was not possible. The prediction equations used in this study were developed and validated in HIV-positive subjects for both cross-sectional and longitudinal measures [14], and bioelectrical impedance analysis (BIA) is now commonly used in the assessment of body composition changes in studies of this nature [16,33].
 
...No significant differences were detected between the placebo and nandrolone groups for changes in serum cholesterol (total, LDL or HDL), triglycerides, glucose and insulin in the present study, although a significant difference did exist between the nandrolone and testosterone groups for LDL and total cholesterol. These tests should be viewed in the context of multiple post-hoc comparisons. Nevertheless, caution is needed in using these agents in HIV-positive patients with pretreatment elevations in lipids and poor glycaemic control, as use in these populations remains untested..."
 
INTRODUCTION
Despite the introduction of highly active antiretroviral therapy (HAART), weight loss continues to be prevalent in HIV disease [1,2]. More importantly, the introduction of HAART has not been associated with increases in the fat-free mass [3], the body composition component associated with survival in HIV disease progression [4]. Reduced energy intake has been shown to be the primary determinant of weight loss in HIV disease [5]; however, other metabolic, psychological, endocrinological and gastrointestinal factors may contribute to negative energy balance [6]. Lowered testosterone levels have also been associated with loss of fat-free mass in HIV disease [7].
 
There are a number of different pharmacological agents available to assist in management of HIV-associated weight loss [8]. These agents include appetite stimulants such as corticosteroids, megestrol acetate, thalidomide and anabolic and androgenic agents (e.g. growth hormone, nandrolone decanoate and testosterone). The anabolic and androgenic agents are the only ones to have shown consistent improvements in both weight and fat-free mass [9-12].
 
Currently, in the clinical management of weight loss in HIV-infected patients, there remains uncertainty about the best approach. After investigation of any underlying cause of weight loss, such as inadequate dietary intake, diarrhoea, opportunistic infections or malignancy, there are no clear guidelines as to the most appropriate agent to use to assist patients to increase fat free mass.
 
There were two primary aims of this study. First, we intended to compare the effectiveness of 12 weeks of therapy with nandrolone decanoate with placebo in HIV-positive men with mild to moderate wasting on fat-free mass. Secondly, we intended to compare nandrolone decanoate and testosterone in order to provide data for clinicians for decision making on the most effective agent for increasing fat-free mass. Secondary analyses compared changes in weight, body fat, intracellular and extracellular water, body cell mass, perception of treatment, HIV disease markers and safety between nandrolone decanoate and placebo or testosterone.
 
RESULTS (tables immediately follow the text below, followed by author Discussion & Methods)
From January 2000 to October 2001, 303 subjects from 21 sites in 11 countries were recruited and randomized. All subjects commenced treatment. Of the 157 patients randomized to nandrolone, 152 completed the 12-week treatment (97%), of the 66 randomized to testosterone, 59 completed the treatment (89%), and of the 80 randomized to placebo, 77 completed (96%) the 12-week study period. Flow through the study is outlined in Fig. 1. Seven subjects discontinued because of adverse events, five from the testestorone arm and one each from the nandrolone and placebo arms. In two cases of adverse events related to study withdrawal from the testosterone arm (one subject withdrew because of buttock pain on injection and one because of ongoing weight increase), the cause was likely to be related to the study drug. Depression as an adverse event resulted in study termination in two subjects (one in the nandrolone arm and one in the placebo arm) and was thought to be unrelated to the study medication. Three other subjects in the testosterone arm experienced adverse events resulting in study termination which were thought to be unrelated to the study medication (rash in one patient, pain not otherwise specified, dehydration, diarrhoea, pyrexia and mouth ulceration in one patient, and respiratory insufficiency in the other patient).
 
Baseline demographics for each of the groups and the whole study population are shown in Table 1. Changes over time for the body composition parameters for nandrolone versus placebo are presented in Table 2; the analysis presented in this table was performed on an intention-to-treat basis. When analysed on a per protocol basis, the changes of fat-free mass in the nandrolone group (mean increase 1.71 kg; n=145) were significantly greater than in the placebo group (mean increase 0.32 kg; n=72; P=0.0007); the estimated mean contrast was 1.38 kg [95% confidence interval (CI) 0.59; 2.17]. Table 2 also shows the contrasts and confidence intervals for the differences in the changes in body composition variables between the nandrolone and testosterone groups.
 
Table 3 shows the results for the two BACRI domains. Table 4 presents summary statistics of changes from baseline to endpoint during the initial treatment period in CD4 and CD8 count and viral load for the intention-to-treat group. Considering the large variability in these immunological parameters, no relevant differences were noted. Changes in total (P=0.024) and LDL (P=0.038) cholesterol were significantly different among the three groups. Post hoc analysis showed that the change in cholesterol was significantly different between the nandrolone decanoate group (change of 0.346±0.879 mmol/L from baseline) and the testosterone group (change of -0.105±0.959 mmol/L from baseline, P=0.005); the change in the placebo group (0.348±2.027 mmol/L) was not significantly different from that in the other two groups. The post hoc analysis for LDL cholesterol similarly showed a significant difference between the nandrolone (change of 0.241±0.827 mmol/L) and testosterone groups (-0.171±0.899, P=0.010), with the change in the placebo group (0.058±0.958 mmol/L) being not significantly different from that in the other two groups. Changes in HDL cholesterol (P=0.318), triglycerides (P=0.266), glucose (P=0.109), insulin (P=0.828), alanine aminotransferase (ALT) (P=0.208), aspartate aminotransferase (AST) (P=0.289), bilirubin (P=0.862) and PSA (P=0.599) were not significantly different among the three groups. The interaction term between baseline testosterone quartile and treatment was not significant for any of the testosterone categories (P>0.214). Therefore there is no testosterone level for which nandrolone treatment would be superior to testosterone or vice versa.
 
Table 2 Summary statistics of the change from baseline to endpoint of the initial treatment period of other body composition parameters, body weight and body mass index (intention-to-treat analysis)
 

parameter-2.gif

treatment-3.gif

tableSum-4.gif

parameter-5.gif

Discussion
 
This study is the first large multicentre randomized placebo-controlled study to demonstrate the effectiveness of nandrolone decanoate over placebo in increasing fat free mass, weight and body mass index in HIV-positive males with wasting. In addition, fortnightly treatment with 150 mg nandrolone decanoate was superior to fortnightly treatment with 250 mg testosterone for increasing weight, increasing body mass index and (insignificantly) increasing fat-free mass. Subjectively, nandrolone was superior to both placebo and testosterone for improving perception of treatment benefit and recovery from symptoms.
 
Testosterone and nandrolone were chosen as the two agents to compare in this study as they were the treatments most commonly prescribed at the participating research centres for treatment of weight loss. Growth hormone has also been shown, in two large randomized placebo-controlled studies [9,12] and in a single-arm study [16], to increase weight and fat free mass in people with HIV-associated wasting [9]. However, the option of using growth hormone is not available in many countries where high cost, difficulty with daily injections, issues of injection equipment safety with home injection and side effects all limit its current and potential usefulness. In view of the escalating nature of health care costs, it is critical that an effective anabolic agent that will improve fat free mass and perception of treatment benefit is of low cost and has minimal side effects so that it may be of use in resource-poor settings.
 
The initial randomized controlled trial of growth hormone for HIV-associated wasting reported a change in weight of 1.1±4.0 kg (P<0.0001 compared with placebo; intention-to-treat) and a change in fat-free mass of 2.5±3.4 kg (P<0.001 compared with placebo; intention-to-treat) [9]. The more recent trial showed weight increases of 1.5 or 2.2 kg (dose dependent) (P<0.0001) with increases in fat free mass of 3.3 and 5.3 kg (P<0.0001) versus placebo [12]. The changes in weight in the present study are similar to those found in this previous research, although the fat-free mass change was greater with growth hormone. Differences in entry criteria and patient demographics preclude a direct comparison of the results of this study and the growth hormone research; however, the economic costs of the treatment cannot be overlooked in the current health care environment. A significantly greater proportion of patients receiving growth hormone reported swelling/puffiness, arthralgia/myalgia and diarrhoea compared with the placebo group. In the most recent study, 49.8-51.8% of those receiving growth hormone reported adverse events including arthralgia, myalgia, peripheral oedema, hypothesia and paraesthesia compared with 28.7% of the placebo group. These side effects must be considered when making treatment decisions.
 
The results of the present study support previous research prior to [10,17] and since [18,19] the introduction of highly active antiretroviral therapy which demonstrated significant within-group changes in weight and fat free mass in HIV-positive male subjects receiving nandrolone in open-label studies. These previous studies reported sample sizes of between eight and 16 subjects, with doses varying from 100 mg/fortnight to an average of 550 mg/week for periods between 10 and 16 weeks. Three studies investigated the effects of nandrolone in weight-losing subjects [10,17,18] and one in weight-stable subjects [19]. Strawford et al [10] specifically recruited hypogonadal subjects. Testosterone levels were within the normal range in one study [19] and not measured in two studies [17,18]. Virological control and CD4 levels and methods of assessing body composition differed in these previous studies. Within-group weight increases in these studies were between 2 and 5 kg and fat free mass increases were between 3 and 4 kg.
 
It is of particular interest that the subjects in the highest dose study [19] did not show greater weight and fat free mass increases than those found in the other published studies, despite using a substantially higher dose of nandrolone. Subjects in this high-dose study were weight stable and eugonadal, and this suggests that nandrolone may be more effective in hypogonadal and/or weight-losing subjects. This contradicts the idea of the dose-response curve proposed by Forbes [20], which suggests that responses to pharmacological doses of anabolic agents are greater than those to replacement doses.
 
The within-group weight change in the present study was 2.1±2.2 kg (P<0.0001 versus placebo) and that in fat free mass was 1.6±3.0 kg (P=0.0006 versus placebo). It is likely that the smaller changes in the present study are a direct result of the large randomized sample size including a diverse range of subjects in terms of pre-entry weight loss, testosterone status, disease stage and ethnic background.
 
A recent meta-analysis of the effects of testosterone versus placebo on fat-free mass showed a significant mean increase in the testosterone group of 0.51 kg (95% CI 0.09; 0.93) for the fixed effects model and 1.22 kg (95% CI 0.23; 2.22) for the random effects model [21]. The same analysis showed positive weight increases of 0.63 kg (95% CI-0.01; 1.28) for the fixed effects model and 1.04 kg (95% CI-0.01; 2.10) for the random effects model. The results of the present study are consistent with those of this meta-analysis, although heterogeneity in dose, routes of administration, and patient characteristics, such as hypogonal versus eugonadal and weight-losing versus weight-stable, makes a direct comparison of studies difficult.
 
Rietschel et al [22] reported that 21% of their HIV-positive male subjects using HAART had lower than normal free testosterone levels, suggesting that hypogonadism remains common in HIV-positive male subjects taking HAART. Prior to the introduction of HAART, this same research group reported lowered free testosterone levels in 49% of a sample of HIV-positive male patients with weight loss [7]. In the present study, 17% of subjects (n=50) had low free testosterone (<42 pmol/L) and 5% (n=15) had low total testosterone (<12.1 nmol/L). We would perhaps have expected this prevalence to be higher based on the previous literature and the bias in our subject group. As the previous literature reports a relationship between loss of fat free mass and lower testosterone levels, we would have expected a high prevalence of hypotestosteronaemia in this sample.
 
It has been reported that the lower than normal testosterone level in HIV-positive male patients is associated with decreased sexual functioning, depression and loss of fat free mass [23,24]. The aetiology of this physiological observation is unclear. HIV may have an impact on the hypothalamic pituitary axis or a direct impact on the testis, or both. In addition to involuntary weight loss, lowered testosterone levels are associated with a poor prognosis [25]. Because of these documented associations, the relationship between baseline androgen levels and treatment with androgenic anabolic agents warrants investigation. Bhasin et al [26] did not find a correlation between change in fat free mass and baseline testosterone levels in their study investigating the effects of testosterone replacement in hypogonadal HIV-positive men with weight loss. In the present study, we aimed to determine whether one of the treatments would be superior in increasing fat free mass at different testosterone levels. The testosterone-by-treatment interaction term was not significant, revealing important clinical information that neither treatment is superior to the other in increasing fat free mass at lower or higher testosterone levels.
 
This study also addresses the concern that anabolic/androgenic agents may increase serum lipids and liver function test results, and have a negative effect on glycaemic control [27,28]. Current evidence does not support these concerns. Replacement doses of testosterone in middle-aged men without HIV infection have been associated with reduced visceral adipose tissue, lower glucose levels and improved insulin sensitivity [29]. Epidemiological studies have shown inverse relationships between total and free serum testosterone and visceral fat, cardiovascular disease and type 2 diabetes [29]. Sattler et al [30] demonstrated no detrimental effects of nandrolone on triglycerides, or total or LDL cholesterol. HDL cholesterol did decrease transiently during nandrolone treatment, but returned to near-baseline levels when assessed 12 weeks after the treatment was finished. Previously we have also investigated changes in serum lipids, fasting insulin and glucose over an 8-week period in 10 HIV-positive male patients with body composition changes consistent with lipodystrophy and found no significant changes in these parameters [31]. No significant differences were detected between the placebo and nandrolone groups for changes in serum cholesterol (total, LDL or HDL), triglycerides, glucose and insulin in the present study, although a significant difference did exist between the nandrolone and testosterone groups for LDL and total cholesterol. These tests should be viewed in the context of multiple post-hoc comparisons. Nevertheless, caution is needed in using these agents in HIV-positive patients with pretreatment elevations in lipids and poor glycaemic control, as use in these populations remains untested.
 
27 Glazer G. Artherogenic effects of anabolic steroids on serum lipid levels. A literature review. Arch Intern Med 1991; 151: 1925-1933.
 
28 Kopera H. (1993) Side effects and contraindications of anabolic steroids, In: Kopera H. ed. Anabolic-Androgenic Steroids Towards the Year 2000, pp. 262-266. Vienna, Blackwell MZV.
 
29 Bhasin S. Effects of testosterone administration on fat distribution, insulin sensitivity, and atherosclerosis progression. Clin Infect Dis 2003; 37 (Suppl. 2): S142-S149.
 
30 Sattler FR, Schroeder ET, Dube MP et al. Metabolic effects of nandrolone decanoate and resistance training in men with HIV. Am J Physiol Endocrinol Metab 2002; 283: E1214-E1222.
 
31 Gold J, Batterham M Nandrolone decanoate; use in HIV-associated lipodystrophy syndrome: a pilot study. Int J STD AIDS 1999; 10: 558.
 
Adverse events thought to be related to the study medication were rare and occurred primarily in the testosterone group, although the differences between groups were not statistically significant.
 
The BACRI was specifically designed and validated to show subjective recovery from the symptoms of anorexia/cachexia and patient perception of benefit with nutritional interventions [15]. The instrument was initially tested in a dose-ranging randomized placebo-controlled trial of megestrol acetate for HIV-associated wasting [32]. In this study, the BACRI was reported not only to differentiate positive effects of treatment versus placebo but also to differentiate dose effects [15]. The use of this instrument in the present study clearly demonstrated a significant benefit of nandrolone over both placebo and testosterone for subjective improvements in recovery of symptoms from anorexia/cachexia and patient perception of benefit. This questionnaire was designed to target the aspects of quality of life that are specific to a nutritional/appetite intervention, such as body image, enjoyment of food, appetite per se and the actual intervention. These areas are not covered in traditional quality of life questionnaires and therefore a treatment effect may be missed.
 
Clinically, this study demonstrates unequivocally the superiority of nandrolone decanoate over placebo for increasing fat-free mass, weight, body mass index, and perception of treatment benefit when compared with placebo. In addition, nandrolone was superior to testosterone for increasing weight and perception of treatment benefit and tended to increase fat-free mass. This study provides important information for decision making in clinical practice. Further research should investigate the longer term effectiveness of these agents.
 
METHODS
 
Subjects

Potential subjects were recruited through contact with clinicians and multidisciplinary HIV health care workers. To participate, subjects had to be HIV-positive male adults with mild/moderate wasting and an energy intake greater than 75% of requirements, estimated using the Harris & Benedict equation [13] by a 24-h dietary recall, on stable antiretroviral therapy including at least two agents, with a CD4 count >50 cells/_L. For this study, mild/moderate wasting was defined as a weight loss of 5-15% in the last 12 months, or a body mass index of 17-19 kg/m2, or a body cell mass/height ratio <13.5 kg/m (13.5 kg/m was defined as the 5th percentile of the normal population based on 43 measures in HIV-positive weight-stable patients). Exclusion criteria included a weight increase of >3% in the last 2 months; hypersensitivity to anabolic/androgenic agents; prostate or mammary cancer; use of anabolic/appetite-stimulating agents in the last 3 months; chronic use of systemic corticosteroids (except topically); significant cardiac, renal, hepatic or other disease which may prevent study completion; an AIDS-defining illness (other than HIV wasting syndrome) within the last 2 months; malignancy other than Kaposi's Sarcoma localized to the skin; involvement in a vigorous resistance exercise training programme (weight training) in the last 2 months; liver function tests (aspartate transaminase/alanine transaminase) five times the upper limit of normal; fasting total cholesterol, triglycerides, insulin or serum creatinine≥three times the upper limit of normal; platelets≦50 000 cells/_L; haemoglobin≦8.0 g/dL; prostate specific antigen (PSA)≥4 ng/mL; active drug abuse; alcohol consumption>four standard drinks/day; administration of any investigational drug in the last 3 months. Each participant gave written informed consent and institutional ethics approval was obtained at each study site.
 
Protocol
Eligible subjects were randomized per site, with each consecutive subject recruited to receive the next study identification number from lowest to highest. For emergencies, a drug identification record was supplied to the investigator at each site. The randomization was weighted 2:1:1 with 150 subjects to receive 150 mg nandrolone decanoate, 75 to receive 250 mg testosterone (30 mg propionate, 60 mg phenylpropionate and 100 mg decanoate) and 75 to receive placebo. Each treatment was given as a 1-mL oil-based intramuscular injection in the gluteal region every fortnight. All injections were given by qualified clinical staff and compliance was confirmed by the study out-patient records and vial count. The treatment period was 12 weeks.
 
Clinical endpoints
The primary endpoint was a comparison of absolute change in fat free mass at 12 weeks between the nandrolone decanoate and placebo groups, and between the nandrolone decanoate and testosterone groups. Secondary outcome measures included change in weight, body fat, intracellular and extracellular water, body cell mass, muscle strength, perception of treatment, HIV disease markers and safety.
 
Body composition
Age, body weight and height were recorded. Body composition was assessed using bioelectrical impedance analysis (RJL101; RJL Systems Inc., Clinton Township, MI, USA). Fat free mass was assessed using the published equation of Kotler et al [14], which has been validated for use in HIV-positive subjects. Body composition was measured at baseline, week 6 and week 12. Body cell mass, intracellular and extracellular water and fat mass were also calculated using validated and published equations from the same source.
 
Perception of treatment
The Bristol-Myers Anorexia Cachexia Recovery Instrument (BACRI) [15], an eight-item questionnaire, was used to examine change in subjective recovery from symptoms of anorexia/cachexia (seven items, considered together as BACRI 7) and patient perception of benefit (one item, termed BACRI 1). BACRI 1 scores are measured on a scale ranging from 0 to 10, where 10 denotes the largest improvement. BACRI 7 scores are measured on a scale of 0-70, where 70 denotes the largest improvement. The BACRI was administered at week 12.
 
Laboratory parameters
Laboratory parameters were measured in a fasting state at baseline and week 12. Liver function tests included ALAT/SGPT, ASAT/SGOT, bilirubin and creatinine. Standard biochemistry included total protein, albumin, haemoglobin, leucocyte count and platelet count. Cholesterol [total, high-density lipoprotein (HDL) and low-density lipoprotein (LDL)], glucose, insulin and glycosylated haemoglobin (HbA1c) were measured as markers of lipid and glucose metabolism. CD4 T-cell count and HIV RNA viral load were measured as markers of HIV disease progression. Prostate-specific antigen (PSA) was also assessed. All sites conducted their own laboratory investigations.
 
Data analysis
Data were analysed using SAS version 8 or higher (SAS Institute, Cary, NC) using the general linear model (proc glm). An analysis of variance was applied with change from baseline to 12 weeks in fat-free mass as the dependent variable. Treatment and centre were entered as factors. Interaction between treatment and centre was investigated. Confidence intervals and estimated means for the contrasts of nandrolone decanoate versus placebo and nandrolone decanoate versus testosterone and summary statistics are presented. Analysis was performed on an intention-to-treat basis with the last observation carried forward. In order to be included in the intention-to-treat analysis, subjects must have had at least one post-baseline fat free mass assessment. Differences between groups in changes in serum lipids, glucose, insulin and PSA were assessed using nonparametric statistics. The superiority of nandrolone or testosterone for eliciting an increase in fat-free mass at various levels of baseline testosterone was investigated by incorporating an additional interaction term (baseline free or total testosterone level in halves, tertiles, quartiles or quintiles by treatment group) into the general linear model discussed above.
 
Author Affiliations
1The Albion Street Centre, Surry Hills, NSW, Australia, 2The University of Wollongong, Wollongong, NSW, Australia, 3Organon International, Oss, The Netherlands, 4Infectious Diseases Service, State Servant Public Hospital, Sao Paulo, Brazil, 5Institute of Infectology Emilio Ribas, Sao Paulo, Brazil, 62nd Internal Medicine Clinic, 1st IKA Hospital, Melissia, Greece, 7Department of Endocrinology, Cisanello Hospital, Pisa, Italy, 8Division of Allergy and Clinical Immunology, Department of Clinical Medicine, University of Rome "La Sapienza", Rome, Italy, 9Department of Medicine, Hospital Kuala Lumpur, Kuala Lumpur, Malaysia, 10AIDS Diagnosis and Therapy Center, Warsaw, Poland, 11Infectious Diseases and AIDS Unit, Hospital Clinic, Barcelona, Spain, 12Thai Red Cross AIDS Research Center, Chulalongkorn Hospital, Bangkok, Thailand, 13Bamrasnaradura Hospital, Nonthaburi, Thailand, 14Chelsea & Westminster Hospital, London, UK, 15Holdsworth House, Darlinghurst, NSW, Australia and 16Academic Medical Center, Amsterdam, The Netherlands