Mar 2, 1994 - Abdominal Obesity. *Maijolein Wssel; fJacob C. Seidell, $Hans R E Koppeschaal; $Philip Sinits. Abstract ..... Pawan (22) reported a decrease in ...
No Specific Effect of Fluoxetine Treatment on Fasting Glucose, Insulin, Lipid Levels, and Blood Pressure in Healthy Men with Abdominal Obesity *Maijolein Wssel; fJacob C. Seidell, $Hans R E Koppeschaal; $Philip Sinits Abstract In this paper we investigated the effect of fluoxetine (60 mgd) on serum lipids, glucose and insulin concentrations and blood pressure by means of a randomized, double-blind placebo controlled trial. Thirty-eight overweight @MI: 26-30kdm2), nondiabetic, nonhypertensive men with an abdominal fat distribution (waisffliipratio: > 0.97) received dietary advice and placebo or fluoxetine for 12 weeks.
Our results indicate that fluoxetine treatment has no specific effect beyond that expected for weight loss on serum lipid, glucose and insulin concentrations, and blood pressure in overweight men. Key words: obesity, pharmacological treatment, visceral fat, I>loodlipids (OBESITY RESEARCH 1994;2:152-159)
Introduction Obesity, and especially abdominal fat accumulaThe changes in serum parameters and blood pres- tion, is known to be associated with an increased risk for sure in the fluoxetine treated group were not differcoronary heart disease (8,16) and diabetes mellitus ent from the placebo treated group, despite a signifi(1,21). Several studies showed a positive correlation cantly larger weight loss in the fluoxetine group. In between the amount of visceral fat, and risk factors like both groups serum total-cholesterol concentrations, increased levels of uiglycerides and reduced levels of serum LDL-cholesterol concentrations and tlie HDL-cholesterol (12,17) as well as increased glucose HDLLDL ratio were significantly improved after and insulin levels (23,31). Fluoxetine, a serotonin retreatment. Reductions in fasting glucose concentrauptake inhibitor, causes a reduction in body weight in tion and systolic blood pressure were only significant obese patients (5,10,14,18-20,24). Several studies in the placebo group. A reduction of serum triglycreported that treatment with fenfluramine, also a seroerides and a n increase of HDL-cholesterol were tonin increasing agent, leads to a reduction in blood found in the fluoxetine treated group. lipids (2,22) and blood glucose (27,35). This effect could not always be explained by the decrease in body In the total study population the changes in serum weight (2,35). To our howledge, the effect of fluoxelipids seemed to be more strongly related to the tine on blood lipids has not been reported. There are a change in total body fat or subcutaneous abdominal few studies in which a blood glucose lowering effect of fat (assessed by MRI) compared to the change in visfluoxetine was reported (3,25). Potential favorable ceral fat. The improvement of most of the serum effects of fluoxetine on the risk factors of fasting blood lipids was related to tlie change in total body fat glucose, serum insulin, total cholesterol and HDLLDL independent of the mechanism for attaining this fat ratio, independent of the amount of weight loss, could loss. have an additional beneficial effect in the treatment of Submitted for publication July 16, 1993. abdominal obesity by decreasing the risk for coronary Accepted for publication in final form Octciber 5,1993. heart disease and diabetes mellitus. From the* Department of Human Nutrition. Wageningen Agricultural University, The Netherlands, tDeparIment of Epidemiology, National Institute This study reports the effect of fluoxetine versus of Public HealUi and Envimnmental Pmtection., The Netherlands. $Department placebo treatment on lipid, glucose and insulin concenof Endocrinology, University Hospital Ulrecht. The Netherlands. $Eli Lilly Netherlands. The Netherlands. trations, and the associations with changes in visceral Reprint requests to Dr.Visser, Department of Human Nutrition, Wageningen fat, in overweight men with an abdominal fat distribution. Agricultural University, P . 0 . Box 8129. 6700 EV Wageningen. The Netherlands. Copyright 01994 NAASO.
152 OBESITY RESEARCH Vol. 2 No. 2 Mar. 1994
Fluoxetine and Metabolic Risk Factors in Abdominal Obese Men, Visser et al.
Methods and Procedures Subjects
Moderately overweight men with abdominal obesity were recruited by advertisements in local newspapers. Of the 170 people responding, 132 returned a completed medical questionnaire containing questions about health and medication, body weight, waist and hip circumferences, smoking habits and alcohol consumption. Fortyone male subjects aged between 25-50 years, with a stable body weight for the last two months, a body mass index between 26-30 and a waisthip ratio > 0.97 were selected. Exclusion criteria were: use of any medication, participation in other studies, previous treatment with fluoxetine, alcohol abuse or smoking. The study protocol was comprehensively explained and all subjects gave their written informed consent. The protocol was approved by the Ethical Committee of the University Hospital of Uuecht. The subjects were selected on basis of their selfreported measurements and underwent a medical examination and standard laboratory blood tests. One subject withdrew before start of the study. Forty subjects entered the study. Two subjects withdrew during the study, one for personal reasons (fluoxetine treated group), and one because of disappointment with the rate of weight loss (placebo treated group). Study Design
After a two-week, single-blind, placebo run-in period (week 0-2) the subjects were randomly assigned to
placebo run-in period was started. Endpoint blood samples were collected during week 13 and week 14. Averages of the two determinations were used in the analysis. During the study period there were several control visits during which subjects were asked about any side-effects of the drug treatment. Compliance was checked by counting the returned and supplied capsules during these control visits. Blood Samples and Analysis
Blood samples were collected between 07:30 and 08:30 am, with the subjects fasted. The subjects had been instructed not to eat after 21:OO pm the evening before the blood collection. They were allowed to drink coffee or tea (without sugar and without milk), or (mineral) water until 23:OO pm. Blood was collected from the antecubital vein in vacutainers. Serum was obtained by low-speed centrifugation and stored at -80°C.The samples were analyzed after the complete study was finished. Serum triglycerides were determined using an enzymatic procedure after free glycerol was removed from the reaction mixture (32). Serum total cholesterol was determined by an enzymatic colorimetric method using cholesterol reagents from Boehringer Mannheim GmbH (Mannheim, Germany) (30). Hypercholesterolemiawas defined according to the cholesterol consensus of The Netherlands (9). High-density-lipoprotein cholesterol was measured by the precipitation method described by Warnick et al. (38). Low-density-lipoproteincholesterol was calculated according to the formula of Friedewald et al. (11). Glucose was determined in blood, using a GOD-PAP kit from Boehringer Mannheim GmbH (Almere, The Netherlands). Serum insulin concentration was measured by radioimmunoassay using 1251-insulin, purchased from the Radiochemical Centre (Amersham, United Kingdom), and I.R.P. 66/304 as standard. Serum insulin concentrations c 5 ma1 (limit of detection) were considered as 2.5 mW1. The within-run coefficient of variation of control sera was 0.91% for total cholesterol, 1.3% for HDLcholesterol, 1.9% for triglyceridesand 1.1% for glucose. The intra-assay coefficient of variation for insulin was 2.6% at a level of 55 mIU/l. Mean bias with regard to serum pools provided by the Centers for Disease Control (Atlanta, GA) was -0.07 mmom for total cholesterol and +0.11 mmol/l for uiglycerides. The mean bias with regard to target values of serum pools obtained from the Solomon Park Research Laboratories (Kirkland, WA) was +0.09 mmolh for HDL-cholesterol.
placebo treatment or fluoxetine treatment (60 mg daily) for another 12 weeks. During the double-blind treatment period (week 2-14) subjects took one capsule per day orally. Before start of the study all subjects received information brochures on healthy nutrition and means to loose weight by the Netherlands Heart Foundation and the Netherlands Bureau for Food and Nutrition Education. Also a presentation was given about the health risks of being overweight and having an abdominal fat distribution and about the effect of weight loss on coronary risk factors. This corresponds to a “usual care” treatment of obesity in general practice in The Netherlands. Reasons for adding dietary advice to the double-blind study were an aim to standardize the diets of the participating men and to reduce the number of dropouts in the placebo treated group. Subjects were asked not to change their level of physical activity during the study. Baseline measurements of body composition, body fat distribution and blood pressure were carried out during week one. These measurements were repeated during the last week of the treatment period (week 14) . Blood Pressure Blood pressure was measured on two days during Duplicate baseline blo& samples were collected 5 days apart from each other from 1 to 5 weeks before the week 1 (baseline) and week 14 (endpoint) using an
OBESITY RESEARCH Vol. 2 No. 2 Mar. 1994 153
Fluoxetine and Metabolic Risk Factors in Abdominal Obese Men, Visser et al.
oscillometric blood pressure computer (Boso-oscillomat, Jungingen, Germany). Subjects were at rest in a supine position for 4 hours (during which energy expenditure was measured) before the b l d pressure measurement was carried out on the left arm.The two values within one week were averaged. Body Composition Body weight was measured to the nearest 0.05 kg (Berkel EDdOT, Rotterdam, The Netherlands) with the subjects wearing swimming clothes and after voiding. B,ody height was measured to the nearest 0.1 cm usinga wall-mounted stadiometer. The body mass index (BMI) was calculated by dividing body weight (kg) by height (m2). Body density was determined by underwater weighing as previously described (36).
mm with a plastic tape at the level of waist (at midpoint between lower rib and the iliac crest) and hip (trochanter level) with the subjects standing. Waist-hip circumference ratio (WHR) was calculated. Abdominal fat areas were assessed by magnetic resonance imaging (MRI) while subjects were in the supine position. Three transverse MRI scans at the level of the umbilicus (umbilicus, 1 cm above and 1 cm below) were made with a whole-body scanner (29,36). Means of the three scans were used in the calculations.
Stntistical Analysis The data of the subjects completing the study were analyzed by using the statistical program SAS (Statistical Analyses System; SAS Institute Inc., Cary, USA). The results are expressed as mean values with standard deviations (SD). Since many of the data were not normally distributed non-parametric tests were used. Body Fat Distribution Body fat distribution was assessed using body cir- Mean values of the variables in the weeks before or the cumferences and magnetic resonance imaging (MRI). fist week of the placebo run-in (week 1) were considBody circumferences were measured to the nearest 1 ered as baseline and were compared with the values of
Table 1: Baseline characteristicsof the placebo and the fluoxetine treatment groups Fluoxetine (n=18)
Placebo (n=20) mean
SD
mean
SD
Demographics Age (Y) Height (m) Body weight (kg) Body fat mass (kg) Body mass index (kg/m2)
38.8 1.79 89. I 25.2 27.9
7.7 0.05 6.0 3.9 1.3
42.6 1.77 87.7 24.4 21.9
5.6 0.04 5.9 4.1 1.o
Body fat distribution Total abdominal fat (cm2) Subcutaneous abdominal fat (cm2) Visceral fat (cm2) Visc/Subc fat ratio Waist/Hip ratio
463 254 209 0.83 0.97
90 50 44 0.09 0.04
437 244 193 0.79 0.98
56 30 32 0.10 0.03
Serum parameters Triglycerides (mmoV1) Total-cholesterol (mmoY1) HDL-cholesterol(mmolfl) LDL-cholesterol(mmoVI) HDULDL ratio Glucose (mmol/l)* Insulin (mEll)
1.93 6.81 1.21 4.73 0.27 5.64 9.38
0.84 1.14 0.22 1.01 0.08 0.42 5.68
1.99 6.88 1.12 4.86 0.25 5.33 8.33
0.87 1.48 0.15 1.30 0.07 0.36 4.89
85 139
9 14
86 138
9 14
Blood pressure Diastolic (mmHg) Systolic (mmHg)
* p < 0.05 placebo versus fluoxetine treated group (Wilcoxon rank-sum test). Values of insulin < 5 mE/l were considered.2.5mE/l. 154 OBESITY RESEARCH Vol. 2 No. 2 Mar. 1994
Fluoxetine and Metabolic Risk Factors in Abdominal Obese Men, Visser et al.
the variables in weeks 13 and 14 of the study by Wilcoxon signed-rank tests. Comparisons between the groups were made using the Wilcoxon rank-sum test. Based on earlier studies at the Department, 40 men in total were selected for detecting, with 80% probability, a 10% improvement in serum parameters and blood pressure. Spearman correlations coefficients were calculated between the changes in serum parameters or blood pressure and the changes in body fat distribution or total body fat after treatment. P-values < 0.05 were considered significant.
The glucose, total-cholesteroland LDL-cholesterol concentration and the systolic blood pressure decreased significantly in the placebo-treated patients. The HDULDL ratio increased significantly for the placebo-treated patients. Serum triglycerides, total-cholesteroland LDLcholesterol concentration decreased significantly in the fluoxetine-treated patients, while HDL-cholesterol and the HDL/LDL ratio significantly increased. No treatment differences in the change of serum parameters or blood pressure were observed. The change in HDLLDL ratio and the change in insulin was significantly correlated with the baseline Results values only in the placebo treated group. Differences Baseline values of demographic variables including between the groups in changes of these two parameters the parameters fat distribution, serum variables concen- were therefore analyzed by analysis of covariance, using trations and blood pressure of the subjects are given in the baseline values of these parameters as a covariable. Table 1. Values are presented for both treatment groups However, adjustment for these baseline values did not separately. The two treatment groups differed only in change the results (data not presented). Table 3 shows the relation between the change in fasting glucose level which was higher in the placebotreated group (pc0.05).At the start of the study, 53 per- serum parameters or blood pressure and the change in cent of the subjects had a serum total-cholesterol con- body fat distribution or body fat mass. Since there were no significant differences in the change of serum paracentration of 6.5 mmol/l or more. After treatment body weight was significantly meters between the two treatment groups, the data of the reduced in both the placebo-treated group (-2.4 ? 2.8 kg; two groups were combined. The baseline values and p=O.OOl) and the fluoxetine-treated group (-5.9 ? 2.8 changes in abdominal fat areas are reported elsewhere kg; p=O.OOOl). The fluoxetine-treated patients had a (36). The change in visceral fat mass was only correlatgreater mean weight loss than the placebo-treated ed with the change in the HDLLDL-ratio (r=-0.41; ~~0.05 and ) not with the other serum parameters nor patients (p