Blood samples were taken with subjects having fasted for at least 12 hours. Serum leptin levels were determined by enzyme-linked immunosorbent assay (ELISA) using a DSL-10-23100 Active human leptin ELISA kit (Diagnostic Systems Laboratories, Webster, Texas).

Biochemical parameters were quantified by colorimetric enzymatic methods in a Hitachi 911 auto-analyzer (Boehringer Mannheim, Kobe, Japan), using reagents from Roche Diagnostics GmbH. Serum insulin concentrations were determined in an IMMULITE analyzer using a solid-phase chemiluminescence enzyme immunometric assay with IMMULITE reagents from Diagnostic Products Corporation.

Body mass index (BMI) was calculated as weight in kg divided by height in m2. The homeostasis model assessment of insulin resistance (HOMA-IR) score was calculated using the formula proposed by Matthews et al.2 (fasting blood insulin concentration in μU/ml×fasting blood glucose in mmol/l/22.5)2.

Hyperglycemia was defined as fasting glucose >110mg/dl (6.1mmol/l), and hyperinsulinemia as fasting insulin above the 75th percentile of the respective distributions: 16.3μU/ml (117pmol/l) for men and 11.7μU/ml (84pmol/l) for women.

The 75th percentile for the HOMA-IR score in our population was 3.0, and insulin resistance was accordingly defined as HOMA-IR score ≥3 and/or glucose intolerance (fasting glucose ≥110mg/dl [≥6.1mmol/l] and/or an oral glucose tolerance test ≥126mg/dl [≥7.8mmol/l] twohours after ingestion of 75g glucose).

Individuals were defined as dyslipidemic if they had one or both of the following: HDL cholesterol <35mg/dl (<0.9mmol/l) for men and <39mg/dl (<1.0mmol/l) for women, and triglycerides ≥150mg/dl (≥1.7mmol/l).

Obesity was defined as BMI ≥30kg/m2.3

The World Health Organization criteria4 for the metabolic syndrome were used – insulin resistance plus two of the following: BMI ≥30kg/m2, triglycerides ≥150mg/dl (≥1.7mmol/l), and HDL cholesterol <35mg/dl (<0.9mmol/l) for men and <39mg/dl (<1.0mmol/l) for women.

In the comparison between the three tertiles of leptin levels, one-way ANOVA was used to compare continuous quantitative variables with normal distribution and the non-parametric Kruskal–Wallis test otherwise. Correlations between data sets were calculated by Pearson's correlation coefficient.

The strength of associations was expressed as odds ratios (OR) and 95% confidence intervals (CI) through logistic regression analysis. Variables showing a statistically significant association with elevated leptin levels on univariate binary logistic regression were then analyzed by multivariate logistic regression. Receiver operator characteristic (ROC) curves were constructed for leptin and the MS components under study.

SPSS for Windows version 14.0 (Chicago, Illinois, USA) was used for the statistical analysis. A p value <0.05 was considered statistically significant.

Leptin is produced mainly in adipose tissue in quantities proportional to its mass and is governed by the Ob gene. It is now known to be an important regulator of body weight, triggering various physiological mechanisms according to the state of the body's energy balance: serum leptin levels act as sensors of the energy balance, communicating information on energy accumulated in adipose tissue to the hypothalamus. High levels lead to reduced appetite and increased energy expenditure, while low levels, reflecting weight loss, increase appetite and reduce energy expenditure. However, in obese individuals, these mechanisms are impaired: high leptin levels do not have these effects. This is leptin resistance.

The physiological mechanisms behind leptin synthesis, secretion and receptor binding and its regulation of appetite and energy expenditure involve numerous hormones and neurotransmitters. These mechanisms are complex and are not fully understood.1,5,6

In this study BMI was used as an indicator of general obesity. In both sexes the rise in BMI from the first to the third tertile of leptin levels was statistically significant (p<0.05 in men and <0.001 in women), practically mirroring the cutoff for obesity (30.9kg/m2 in men and 29.1kg/m2 in women) for the highest tertile (Table 1). Similarly, the correlation coefficients of this association were very strong (r=0.524 in men and r=0.603 in women), with high statistical significance (p<0.001) (Table 2).

BMI was also the strongest predictor of high leptin levels, as shown by AUC of 0.81 in men and 0.84 in women (Figure 2 and Table 4).

The same relation between leptin and obesity was seen in the strong association between elevated leptin levels and the different risk factors and MS components under study, with BMI ≥30kg/m2 showing ORs of 10.11 in men and 6.00 in women on univariate analysis and 9.30 in men and 8.21 in women on multivariate analysis (Table 5).

The results of this study show that hyperinsulinemia and IR strongly influence leptin levels (Figures 1 and 2, Tables 3 and 4); leptin was the main predictor of IR in both sexes (AUC=0.89 in men and 0.85 in women), while after obesity, IR in men and hyperinsulinemia in women were the strongest predictors of high leptin levels (AUC=0.79 and 0.78, respectively).

The correlations between leptin levels and HOMA-IR and insulinemia were also strong (Table 2), particularly insulinemia in men (OR=11.71) and IR in women (OR=21.22) (Table 5).

Hyperinsulinemia and IR are major elements of MS, reflecting the strong interconnections between metabolic risk factors for CVD (especially coronary disease), type 2 diabetes and hypertension; when present together, these risk factors have multiplicative rather than additive effects, greatly increasing the morbidity and mortality caused by these conditions.7 This inter-relation can be seen in Table 1, which shows that, with the exception of HDL cholesterol, the study variables (which are in fact MS components), all rose in a statistically significant fashion from the first to the third tertile of leptin levels.

The relationship between leptin, hyperinsulinemia, and other MS components is so clear that serum leptin could be considered an additional component of MS and a new cardiovascular risk factor.8 The prevalence of MS continues to rise (27.5% in Portugal9) in parallel with the obesity pandemic and is now seen in children and adolescents10,11; for some authors it is the main mechanism in the pathogenesis of CVD.7

The precise mechanisms behind the development of IR in the adipose tissue characteristic of general obesity, and the role of leptin, are still the subject of debate, despite considerable research and the clinical implications.

Adipocytes are among the cells most affected by the action of insulin, which induces preadipocytes to differentiate into adipocytes and then stimulates energy storage in the form of triglycerides. In obesity-related IR this mechanism is impaired, in both skeletal muscle and the liver.

The location of adipose tissue influences the development of IR, with abdominal deposits being more strongly associated with IR (and hence with risk for type 2 diabetes and CVD) than peripheral (gluteal or subcutaneous) fat deposits. One possible but as yet unproven explanation is that intra-abdominal adipocytes are more lipolytically active, which would increase the flow of free portal fatty acids, leading to IR by acting on insulin receptors, reducing their sensitivity and that of subsequent metabolic pathways.12

Adipose tissue is no longer considered merely as storing energy in the form of triglycerides; adipocytes in fact function as endocrine glands, producing and secreting protein hormones including leptin, adiponectin, and adiposin, cytokines such as TNF-α and PAI-1 that are important for homeostasis, angiotensinogen (which regulates vascular tone), steroid hormones like estrogens and cortisol, and substrates such as free fatty acids. Through these products adipocytes influence not only local biology but also overall metabolism in different areas including the brain, liver, muscle, pancreatic beta cells, gonads, lymph glands and blood vessels.

At the same time, some of the products secreted mean that adipose tissue plays a part in the development of IR. This is the case with TNF-α, leptin, peroxisome proliferator-activated receptor gamma (PPAR-γ)13 and cAMP response element-binding protein (CREB).14 PPAR-γ, a transcription factor expressed abundantly in adipose tissue, has a central role in adipocyte differentiation and in the development of obesity caused by fat-rich diets and of IR. CREB is activated in adipocytes in obesity, leading to IR through reduction of adiponectin production and hence the loss of its protective effect against IR, and in atherosclerosis, which also modulates insulin sensitivity. Increased adipose tissue, together with probable changes in liver function, will raise TNF-α levels and promote lipolysis, leading to the release of free fatty acids, which would also trigger IR. Investigation of the possible role of leptin in the different stages of non-alcoholic fatty liver disease, which is also linked to obesity and IR, has led to the conclusion that there is no association, although leptin seems to have some effect in preventing progression of the disease.15,16

Leptin has pleiotropic effects, and besides its important role in energy homeostasis described above, it contributes to maintenance of insulin sensitivity via the hypothalamus, acting in tissues that are sensitive to insulin, and profoundly influencing glucose and lipid metabolism. Defects in these mechanisms that reduce leptin's action (leptin resistance) are linked to the etiopathogenesis of obesity and IR, but the precise relationship is not fully understood.

In this study, serum leptin levels ranged from 1.8 to 47.1ng/ml in men and from 6.4 to 88.4ng/ml in women, with first and third tertiles respectively of <9 and >15.2ng/ml, in men and <24.6 and >47.1ng/ml in women (Table 1), while cutoff values based on the 75th percentile for leptin levels were 19 and 50ng/ml in men and women, respectively. We thus found, as have many authors, that serum leptin levels in females are approximately three times higher than in males, a ratio that is found in children and adolescents, pre- and post-menopausal women, and in the elderly. This difference raises various questions.17–19

In our study, obesity defined by BMI was the strongest predictor of high leptin levels (Figure 2 and Table 4). It is suggested, though not without controversy, that the influence of fat mass on leptin levels depends on its location, with greater expression of leptin mRNA in the subcutaneous adipose tissue characteristic of women than in the abdominal adiposity that predominates in men. Some authors relate this gender difference in fat distribution to differences in lipid profile and glucose–insulin homeostasis that are more favorable in women, which was also found in our study (Table 1), since for similar BMI values, women had lower levels of lipids, glucose and insulin, which would make them less vulnerable to the effects of obesity.19