Introduction

A substantial variability in serum leptin concentrations remains among subjects with equivalent amount of adipose tissue1,2 and obviously factors other than adipose tissue mass are important in the regulation of serum leptin. In accordance, investigations have shown that prolonged fasting decreases serum leptin and leptin expression,3 whereas only excessive food intake increases serum leptin.4 Furthermore, hormones such as dexamethasone5,6 and insulin lead to an increase in leptin concentrations,7,8 and adrenergic agonists to a decrease in serum leptin.9

The majority of investigations indicate a gender difference in leptin concentrations. The increased leptin concentration in females has been confirmed in several large studies and persists even when leptin is controlled for body mass index (BMI),10 percentage of body fat,11 total fat mass,12 different fat depots13 or skin-fold thickness. In addition, the sex difference is evident at the ob mRNA level.14 In a study by Wabitsch et al,15 the fat mass corrected leptin concentrations were at the highest in boys during early puberty and declined thereafter, whereas the leptin concentrations in girls increased during the whole of puberty. In trans-sexual subjects, leptin concentrations were reversed according to sex steroid-induced change in phenotype, suggesting an influence of sex hormones on plasma leptin.16 This had led to the suggestion that sex hormones are important in regulating serum leptin. Furthermore, some investigations have shown that the fluctuation in estrogen during the menstrual cycle has an influence on leptin concentrations.17 However, other studies have not been able to confirm any significant difference during the menstrual cycle,18,19 in pre-or postmenopausal states20 or in postmenopausal women treated with hormone replacement therapy.20,21,22

The investigations regarding the effects of testosterone have been more convincing. In hypogonadal men, leptin concentrations adjusted for BMI, were elevated when compared with those for healthy men, and normalized on testosterone substitution.23 However BMI is not a good estimate of adipose tissue mass and especially not when androgen hormones are administered because this is followed by changes in both lean mass and adipose tissue mass.24 However, association studies have confirmed a negative correlation between leptin and different androgens.13

In the present study we have investigated the association between leptin and sex hormones in 36 women. In addition, the direct effects of testo-sterone, DHT and 17 β-estradiol were investigated in female subcutaneous adipose tissue fragments in vitro. Finally, the correlation between leptin and estimates of body composition, evaluated by anthropometric measurements and by DEXA scans, was investigated.

Materials and methods

Subjects

Thirty-six healthy adult women with an age span from 23 to 65 y with a wide range of BMI (19.3–61.9 kg/m2) were selected to the study. Thirty-four were premenopausal and two were postmenopausal. Twelve were considered as non-obese (BMI<30), 12 as obese (30<BMI<40) and the last 12 as severely obese (BMI>40). All the subjects were healthy apart from being obese and none were diabetic. Body weight was stable for at least 2 months prior to the study as indicated by history. The subjects were included in the study after informed consent in accordance with Helsinki Declaration II and the local Ethics Committee approved the study.

Anthropometric measurements.

Height and body weight were measured and body mass index (BMI=weight in kg divided by the height in meters squared) was calculated. Waist circumference was measured in the supine position midway between the lower rib and the iliac crest. The hip circumference was measured at the widest part of the hip region and the waist–hip ratio (WHR) was calculated from the measurements. The sagittal diameter (SD) was measured in maximal expiration with a calliper halfway between the lower rib and the iliac crest.

Dual energy X-ray absorbtiometry (DEXA) scans

Body composition was evaluated using a QDR-1000 densiometre (Hologic, Waltham, MA, USA). Absorbance was measured in the patient alone and together with a soft tissue and a bone standard, respectively. The DEXA scanner has an accuracy of 3.3% in determining lean body mass compared with underwater weighting and a reproducibility of 98.7–99.7%. Each scan can later be manipulated to give body composition in specifically chosen areas of interest. The following measurements were used to evaluate the effects of adipose tissue distribution on leptin measurements: total fat mass (FMDEXA), percentage fat tissue=FMDEXA/body weight×100% (FM%), lean body mass (LBMDEXA), fat tissue localized on the trunk (TFDEXA), percentage of fat localized on the trunk=TFDEXA/FMDEXA×100% (TF%), subcutaneous fat tissue localized on arms and legs (PERIFATDEXA) and percentage of subcutaneous fat localized on arms and legs (PERIFATDEXA%).

Incubation of adipose tissue

Nine women (25.5<BMI<36.0 kg/m2) had a subcutaneous adipose tissue biopsy taken at the level of the umbilicus. The tissue was removed using sterile technique, placed in Medium 199 without phenol red and all subsequent procedures were carried out under a laminar airflow hood. The tissue was minced into fragments of less the 10 mg each. All samples were washed free of blood clots and free lipid and placed in organ culture as previously described.22,25 In brief, 500 mg adipose tissue fragments floated freely in 16 ml serum-free Medium 199 without phenol red in 50 ml plastic tubes. The cultures were placed in a humidified incubator and maintained at 37°C in an atmosphere of 5% CO2 in O2. The medium was supplemented with 25 mM Hepes, 5% bovine albumin and 1 nM insulin (Novo Nordisk, Demark). The adipose rissue inplants were incubated in duplicates, and the medium collected at least every 24 h the medium was changed and the cumulative leptin secretion calculated. Because of the pronounced response to dexamethasone, all incubations with adipose tissue were made with dexamethasone as a control to assure the responsiveness of the cells. The culture medium was kept at −20°C until leptin was measured. All reagents were obtained from Sigma Chemical Co., St Louis, MO, USA.

Blood parameters

Leptin was measured as a single measurement after fasting for 8 h in serum samples with a RIA method from Linco Research Ltd., St Charles, MO. The range of the standard curve in this assay was 0.5–100 ng/ml. Intraassay coefficient of variation was 3.7%. Full sex hormone status was measured at the Statens Serum Institute, Copenhagen, Denmark. The following hormones were measured: sexual hormone binding globulin (SHBG), 17b-estradiol, free estradiol, estrone, free testosterone, testosterone, dihydrotestosterone (DHT), androstendione and dehydroepiandrosterone sulfate (DHAS). Insulin was measured with an RIA method.

Statistical analysis

Data are presented as means±s.e.m. All analyses were performed using the Statistical Package for the Social Science (SPSS/PC+, SPSS, Chicago, IL). The distributions of leptin, insulin and the various sex hormones were normalized by log transformations. Differences in mean values between obesity groups were assessed by ANOVA, followed by post hoc correction (Bonferroni). Differences between the two groups in the in vitro assay were assessed by students unpaired t-test. Simple and multiple regression analysis were used to identify the independent effects of variables associated with variations in leptin concentrations. P values below 0.05 were taken as statistically significant.

Results

Associations between serum leptin and body composition

Data on subjects participating in the study are given in Table 1. As seen, when the level of obesity increased, a number of well-known metabolic characteristics associated with adiposity were displayed, such as increased concentrations of plasma leptin and insulin. The mean serum leptin (s-leptin) concentrations in the obese (30<BMI<40 kg/m2) was 35.1±2.9 ng/ml and in the very-obese (BMI>40 kg/m2) 50.1± 3.3 ng/ml, corresponding to about three and four times higher concentrations compared to the non-obese (BMI<25 kg/m2; 11.9±2.5 ng/ml), respectively. The range in the very obese group was varying from 37.2 to 72.4 ng/ml and in the non-obese group from 1.9 to 34.6 ng/ml.

Table 1 Subjects characteristics
Full size table

In Table 2 the correlations between s-leptin and different estimates of obesity are shown. s-Leptin was highly correlated to estimates of obesity in the total group and especially with the FM% determined by DEXA scans (r=0.88). The correlations between leptin and the different fat localizations determined by DEXA scan were in the same range (Table 2). According to the anthropometry the highest correlation was obtained to total fatness determined by BMI (r=0.83) and the weakest to abdominal fatness determined by WHR (r=0.52). However, the correlations to waist (r=0.76) and sagittal diameter (r=0.78) were high (Table 2). The association between FM% and serum leptin is plotted in Figure 1.

Table 2 Pearson's correlation coefficient between leptin and measurements of adiposity
Full size table
Figure 1
figure 1

The correlation between leptin and dihydrotestosterone (DHT), insulin and FM% in 36 women.

Full size image

Dividing our subjects according to BMI into three groups revealed that the correlation between s-leptin concentrations and measurements of adipose tissue were only significantly correlated in the non-obese individuals (results not shown). If all obese subjects (BMI>30 kg/m2) were investigated in one group, leptin correlated with BMI (r=0.48, P=0.02) and FM% (r=0.56, P=0.01), whereas no other estimate was significantly correlated with leptin. If the leptin: FM-DEXA ratio was calculated as an index of the plasma leptin per unit of adipose tissue, the leptin secretion almost doubled from non-obese to obese (P<0.05), whereas the increase from obese to very obese was only 17% (NS) (Table 1).

Correlations between leptin, insulin and sex hormones

In Table 3 the correlations between leptin, insulin and sex hormones are shown. In the correlation analysis there was a significant negative correlation between leptin and SHBG, total estradiol and free estradiol and among the androgens a significantly negative correlation was found between leptin and DHT (r=−0.57, P<0.001; Figure 1) whereas no significant correlations were found to total and free testosterone or androstendione. A negative correlation was found between leptin and DHAS (−0.31, P<0.05).

Table 3 Correlations between leptin with levels of various sex hormones
Full size table

Since the fat mass and the level of insulin may be interplayers in the associations between leptin and sex hormones we also made the correlation analysis after controlling for these two variables. When the correlations between leptin and sex hormones were controlled for insulin, only minor changes were seen in the correlation coefficients but when leptin was controlled for fat mass using FM% no correlations were found (Table 3). In multiple regression analysis using various sex hormones, DHT, SHBG, insulin and FM%, it was found that only insulin and FM% remained independently and significantly correlated to leptin without any significant influence of the sex hormones (Table 4).

Table 4 In multiple regression only insulin and FM% were significantly correlated with leptin
Full size table

In vitro incubation of subcutaneous adipose tissue

Incubations with testosterone (50, 500 and 1000 nM), 17β-estradiol (50 and 500 nM) and DHT (50 and 500 nM) were made for 24 h, but no effects on leptin secretion were found. To assure that adipose tissue explants were viable all incubations were made in parallel with a 10 nM dexamethasone incubation, which caused an almost 3-fold (P<0.001) increase in leptin secretion after 24 h. To further investigate the influence of sex hormones 17 β-estradiol (500 nM), testosterone (500 nM) and DHT (500 nM) were added together with dexamethasone (10 nM) to the culture medium but no changes were seen in leptin secretion compared to dexamethasone without sex hormone addition (Figure 2). In a pilot study (n=3) incubations for 72 h were made to investigate whether longer incubation time was necessary with both testosterone and 17 β-estradiol but without any influence on the results.

Figure 2
figure 2

Effects of sex hormones on leptin production in vitro. Subcutaneous adipose tissue (500 mg) was incubated for 24 h with testosterone, dihydroestosterone (DHT) and 17 β-estradiol in the presence or absence of dexamethasone (10 nM). Dexamethasone stimulated leptin production 3-fold (P<0.001), whereas the sex hormones were without effect on leptin secretion (mean±s.e.m., n=9).

Full size image

Discussion

There are well-known gender differences in the serum concentrations of leptin and several studies have indicated that these differences are due to the influence of sex hormones on leptin production.15,16,17 In the present study we have investigated the relationship between leptin and sex hormones in a group of women with a range of BMI from 19 to 48 kg/m2 as well as in subcutaneous adipose tissue in vitro.

In the in vivo investigations we found by simple correlation analysis significant correlations between several of the sex hormones and leptin. Previously, it has been shown that insulin is an important regulator of leptin secretion and leptin mRNA expression. Therefore the correlation between leptin and sex hormones were controlled for both insulin and FM and consequently the correlations disappeared (Table 4). The in vitro studies with incubations of subcutaneous adipose tissue confirmed that neither estrogen nor androgens (DHT or testosterone) had any direct effects on adipose tissue leptin production.

A few other studies have investigated whether androgens have direct effects on adipose tissue leptin secretion in humans. In subcutaneous preadipocytes leptin secretion and leptin mRNA have been reported to be moderately inhibited by testosterone and DHT in vitro.15 However, these studies were made in preadipocytes from the female breast and the doses used for the experiments were supra-physiological (100–1000 ng/ml testosterone). In addition, leptin secretion and leptin mRNA expression in maximally differentiated preadipocytes are very low and consist of only approximately 1% of that in mature adipocytes26 making the relevance of this regulation questionable. Recently, a study in adipose tissue explants from human omental fat cells has been performed. Omental fat cells did not change in leptin secretion after testosterone addition in females or males, which is in agreement with our study using subcutaneous adipose tissue. However, several other androgens (DHT, stanozolol, androstenedione and DHAS) inhibited leptin secretion in female omental adipose tissue samples by 27–48% whereas no effects were found in male samples in vitro. 27 This is hard to interpret as long as no sex differences in androgen receptor density28 or function have been reported in omental fat. Furthermore, the reported gender difference between men and women cannot be explained by androgens, as they were reported not to affect male adipose tissue leptin secretion. The present finding that sex hormones may not be directly involved in the sexual dimorphism in serum leptin is in agreement with most recent studies in both rats30 and humans.31,32,33

The significance of leptin regulation in omental fat can also be discussed. Omental fat is a relatively small fat depot in comparison with subcutaneous fat. Investigations in women with different degrees of intra-abdominal fat depots have not revealed any differences in circulating leptin levels.29 This may reflect the fact that omental adipocytes secrete less leptin compared to subcutaneous adipocytes14 and therefore have a rather small influence on total amount of circulating leptin.

The in vivo studies, which indicate that sex hormones may affect leptin production, seem to often have problems in relation to controlling for differences in fat mass. Very often BMI is used instead of more direct measurements of the fat mass.20,23,34 Moreover, it has been well described that insulin stimulates human adipose tissue leptin production,8 and, in addition, that insulin may inhibit the production of SHBG in the liver.35,36 Insulin level increases with obesity and the level of SHBG decreases.37 As most of the androgens are tightly bound to SHBG, investigations with measurement of total testosterone may be confounded by the lower SHBG level associated with obesity. Thus, a negative correlation between leptin and a total measurement of androgens is likely to arise because of decreased SHBG levels. The present finding of reduced DHT in obesity might actually reflect the reduction in SHBG, as DHT is tightly bound to SHBG and is measured as total amount of DHT in plasma.

The present study reveals that s-leptin is closely associated with different estimates of fat mass as seen in most other studies. We found the strongest correlations in the lean individuals whereas no correlations were seen in obese or very obese subjects. These findings might indicate that the regulation of s-leptin differs in lean and obese individuals or may be due to the greater variability in s-leptin in obese subjects. In obese women we found a 2-fold greater leptin level per fat mass compared to non-obese (P<0.05), which might be explained by an increase in leptin/fat cell and/or an increase in fat cell numbers. However, no differences were found in leptin/kg fat mass between obese and the very obese subjects. These results might indicate that in some very obese subjects an insufficient leptin level is present compared to the amount of adipose tissue, which may contribute to the development of obesity. Ioffe et al 38 have tested the possibility that a relative low leptin production leads to obesity by mating animals carrying a weakly expressed adipocyte-specific human transgene to ob/ob mice. The mice carrying the transgene had a plasma leptin of approximately one-half that found in wild-type mice and were markedly obese, though not as obese as ob/ob mice. However, taking the 36 women in one group, there was a clear linear relationship between FM% and leptin (Figure 1), as previous reported by others,2 but an exponential rise39 was not found.

In conclusion, by simple correlation analysis there was a negative association between leptin and DHT. However, in multiple regression analysis, only insulin (partial correlation coefficient=0.55, P<0.004) and FM% (partial correlation coefficient=0.72, P<0.001) were significantly correlated with leptin without any significant effect of sex hormones. These findings were in agreement with in vitro studies where neither estrogen nor androgens (testosterone and DHT) affected subcutaneous adipose tissue leptin production. This indicates that neither estrogens nor androgens may directly be involved in the observed gender difference in s-leptin.