LETTERS jects were monitored to measure sex hor- by centrifugation immediately after sam- OBSERVATIONS mones in order to adapt the treatment pling and frozen at Ϫ20°C until the end of dose. The study protocol was approved the trial, then all of the samples were by the Henri Mondor Hospital Ethics thawed and the analyses performed in a Androgen Therapy Committee. All of the included subjects single batch. All of the analysts were blind gave written informed consent. to the treatment allocation. Insulin was Improves Insulin Men with low levels of PTT (con- determined by the immunoradiometric Sensitivity and firmed by two measurements) were se- assay method (Medgenix Diagnostics, Decreases Leptin lected from a large occupation-based Fleurus, Belgique), leptin by a commer- population. The inclusion criteria were as cial radioimmunoassay (RIA) (Linco Re- Level in Healthy follows: 1) either PTT Յ3.4 ng/ml [5th search, St. Charles, MO), follicle- Adult Men With Low percentile value of PTT distribution in the stimulating and luteinizing hormone by Plasma Total 1,718 men of the TELECOM Study (7)] the Automated Chemiluminescence Sys- from 1985 to 1987 and Ͻ4.0 ng/ml (13th tem 180 (Ciba Corning), and androgens Testosterone percentile value) from 1992 to 1993 (3) and estradiol by RIA (7). The only missing or PTT Ͻ4.0 ng/ml from 1992 to 1993 datum was one 2-h plasma insulin mea- A 3-month randomized placebo- and Ͻ4.0 ng/ml a few days before inclu- surement at 3 months in a subject treated controlled trial sion; 2) no history of vascular thrombosis by DHT. or ischemic heart disease; 3) no treatment The primary end points to assess by androgens, anti-androgens, and an- insulin sensitivity were fasting plasma tidiabetic or antithrombotic drugs; 4) insulin–to–fasting plasma glucose ratio n men, an association between lower normal values of plasma prolactin, estra- and homeostasis model assessment plasma total testosterone (PTT) and in- diol, and thyroxin; 5) no current prostatic (HOMA) index. Plasma leptin, 2-h I sulin resistance has been found in disease and a normal PSA value. A total of plasma glucose and insulin, and blood cross-sectional studies (1,2) and in one 18 healthy men with stable low plasma pressure were taken as secondary criteria. nested case-control study (3) without any androgens (Table 1) and a range of PTT Treatment tolerance was assessed by in- possible conclusion in terms of causality from 1.4 to 3.7 ng/ml at baseline were terview, by prostatic examination, and by or direction of the relationship. Indeed, to included. PSA, as well as by weight, electrocardio- obtain such information, randomized The 18 selected men were randomly gram (ECG), lipid, hemoglobin, hemato- controlled trials are needed. Until now, assigned to one of three treatment groups: crit, fibrinolysis markers, and hepatic only one clinical trial has suggested that testosterone, DHT, or placebo. The ran- enzyme variations. testosterone therapy improves insulin domization code was known only to the A sample size of 36 subjects was sensitivity in obese men (4). Cross- study manager. Treatment was a gel ad- needed to detect a difference of 5 mg/dl sectional studies concerning leptin regu- ministered every morning by percutane- for the decrease of fasting plasma glucose, lation by androgens have provided no ous route. The daily dose during the first assuming SD ϭ 5 mg/dl, using a two- definitive conclusions as to whether the weeks was 125 mg for the testosterone tailed Student’s t test with ␣ϭ0.05 and negative association between androgens and 35 mg for the DHT treatment groups. ␤ϭ0.20. However, we could not reach and leptin level is independent (5) or de- The adaptation of treatment doses be- that number, and the recruitment was pendent (6). This randomized controlled tween days 10 and 20 aimed at obtaining closed after having included 18 subjects. trial was designed to assess the role of an- a trough level of PTT between 4 and 10 To evaluate the treatment effect, the dif- drogens on insulin sensitivity and leptin ng/ml for the testosterone group and a ference between the values at entry and at regulation in healthy adult men. trough level of plasma DHT between 4 the end of the treatment period was cal- This study was a randomized, dou- and 10 ng/ml for the DHT group. To culated for each subject, and then the ble-blind, unicentric, controlled, clinical maintain the double blinding, the study Kruskal-Wallis nonparametric test was trial. Three treatments (testosterone, di- manager also sometimes changed the used. When statistical significance (P Յ hydrotestosterone [DHT], and placebo) dose of placebo. The subjects were asked 0.05) was reached for any overall three- were compared in parallel groups during not to change their dietary and physical group comparison, two-by-two compari- a 3-month period. All of the examinations activity. Compliance to treatment was as- sons were performed using the Bonferroni were performed by only two physicians, sessed by interview and by measuring sex test to correct for multiple comparisons. using a standardized protocol. Blood was hormones and gonadotropins at the end At baseline, the three treatment drawn between 8:00 A.M. and 9:30 A.M. of the trial. groups were similar with respect to age, after an overnight fast to determine fasting Plasma glucose, total cholesterol, BMI, waist-to-hip ratio (WHR), blood plasma glucose, insulin, leptin, sex hor- HDL cholesterol, triglycerides, apoli- pressure, plasma glucose, lipids, insulin, mones, lipids, coagulation and fibrinoly- poprotein (apo)-A1, apoB, hepatic en- leptin, androgens, and sex hormone– sis parameters, hepatic enzymes, and zymes, and blood cell count were assayed binding globulin, as well as hemoglobin, prostate-specific antigen (PSA) and blood on the same day of venipuncture. PSA and hematocrit, coagulation, and fibrinolysis cell count. Then, a standard 75-g oral glu- fibrinolysis markers were measured parameters (data not shown). At the end cose tolerance test and a digital rectal ex- within 3 days after venipuncture. For hor- of the trial, a significant difference was amination were performed. In addition, mone measurements at baseline and at shown for the variation of fasting plasma between days 10 and 20, all of the sub- the end of the trial, plasma was separated insulin (P Ͻ 0.05), fasting plasma insu- DIABETES CARE, VOLUME 24, NUMBER 12, DECEMBER 2001 2149 Letters Table 1—Baseline characteristics and variations in the three treatment groups (after minus before) Testosterone DHT Placebo P n 666 Age (years) 52.8 Ϯ 4.2 51.2 Ϯ 3.9 55.4 Ϯ 3.6 0.80 BMI (kg/m2) 29.9 Ϯ 0.9 27.8 Ϯ 0.9 28.0 Ϯ 1.1 0.84 WHR 0.95 Ϯ 0.02 0.96 Ϯ 0.02 0.96 Ϯ 0.03 0.99 Systolic blood pressure (mmHg) 152 Ϯ 5 143 Ϯ 7 126 Ϯ 8 0.08 Diastolic blood pressure (mmHg) 93 Ϯ 488Ϯ 280Ϯ 5 0.17 Fasting plasma glucose (mg/dl) 101 Ϯ 597Ϯ 299Ϯ 4 0.97 Total cholesterol (mg/dl) 212 Ϯ 14 228 Ϯ 11 221 Ϯ 14 0.70 HDL cholesterol (mg/dl) 45 Ϯ 444Ϯ 442Ϯ 6 0.81 Triglycerides (mg/dl) 126 Ϯ 20 142 Ϯ 18 123 Ϯ 18 0.64 Fasting plasma insulin (␮U/ml) 14 Ϯ 418Ϯ 413Ϯ 3 0.52 Leptin (ng/ml) 10.1 Ϯ 4.5 6.4 Ϯ 1.3 6.2 Ϯ 1.5 0.81 Plasma total testosterone (ng/ml) 2.4 Ϯ 0.1 2.9 Ϯ 0.3 2.7 Ϯ 0.3 0.20 Plasma bioavailable testosterone (ng/ml) 0.6 Ϯ 0.1 0.8 Ϯ 0.1 0.6 Ϯ 0.1 0.06 Plasma SHBG (nmol/l) 16.5 Ϯ 1.9 16.9 Ϯ 2.6 21.0 Ϯ 3.3 0.32 ⌬ Fasting plasma glucose (mg/dl) 4 Ϯ 3 Ϫ1 Ϯ 33Ϯ 4 0.42 ⌬ Fasting plasma insulin (␮U/ml) Ϫ0.8 Ϯ 2.0 Ϫ6.2 Ϯ 2.2 2.7 Ϯ 1.6 0.02* ⌬ Fasting plasma insulin/fasting plasma glucose Ϫ0.23 Ϯ 0.32 Ϫ1.09 Ϯ 0.29 0.43 Ϯ 0.35 0.003* ⌬ HOMA index Ϫ0.09 Ϯ 0.53 Ϫ1.54 Ϯ 0.69 0.73 Ϯ 0.39 0.012* ⌬ Leptin (ng/ml) Ϫ1.2 Ϯ 1.7 Ϫ1.8 Ϯ 0.6 0.4 Ϯ 0.4 0.05† ⌬ Total cholesterol (mg/dl) Ϫ7 Ϯ 7 Ϫ4 Ϯ 5 Ϫ12 Ϯ 5 0.66 ⌬ HDL cholesterol (mg/dl) Ϫ1 Ϯ 2 Ϫ5 Ϯ 11Ϯ 2 0.09 ⌬ Triglycerides (mg/dl) 6 Ϯ 18 11 Ϯ 12 19 Ϯ 24 0.78 ⌬ Systolic blood pressure (mmHg) 4 Ϯ 5 Ϫ3 Ϯ 521Ϯ 7 0.052† ⌬ Diastolic blood pressure (mmHg) 4 Ϯ 55Ϯ 514Ϯ 3 0.22 ⌬ Weight (kg) 3.3 Ϯ 1.1 1.4 Ϯ 1.0 Ϫ0.3 Ϯ 1.0 0.09 Data are means Ϯ SEM. *P Ͻ 0.01 for DHT vs. placebo; †P Ͻ 0.05 for DHT vs. placebo. lin–to–fasting plasma glucose ratio (P Ͻ three groups for lipids (Table 1), PSA, he- increased) the response to DHT treatment 0.01), and HOMA index (P Ͻ 0.05), patic enzymes, coagulation, and fibrino- and blood pressure, probably explaining which all decreased under androgens. lysis parameters, but hemoglobin and the nearly significant improvement of sys- The two-by-two comparisons showed a hematocrit increased under androgens tolic blood pressure under androgens by a significant improvement only for DHT (P Ͻ 0.05 and P Ͻ 0.01, respectively), regression to the mean phenomenon in compared with placebo (P Ͻ 0.01 for all mainly with testosterone (data not shown).