Obesity prevalence has been increasing among children and adolescents as it has in adults(Reference Ogden, Carroll, Curtin, McDowell, Tabak and Flegal1). In the USA, the obesity and overweight prevalence in 2003–4 was 17·1 and 33·6 % respectively among children and adolescents aged 2–19 years(Reference Ogden, Carroll, Curtin, McDowell, Tabak and Flegal1). In Europe, Portugal has the second highest overweight prevalence in children, 31·5 %, immediately after Italy, and 11·3 % are obese(Reference Padez, Fernandes, Mourão, Moreira and Rosado2).
Leptin is a 16 kDa peptide hormone codified by the ob gene which is located at chromosome 7 (7q31.1)(Reference Baptista3). It is synthesised mainly by adipose tissue and secreted to plasma, but other tissues such as the stomach, intestines, placenta and testes also secrete leptin(Reference Moran and Phillip4). It regulates the expression of hypothalamic neuropeptides involved in energy metabolism by suppressing food intake and stimulating energy expenditure(Reference Moran and Phillip4–Reference Pilcová, Sulcová, Hill, Bláha and Lisá6).
At first, it was thought that human obesity might occur as a result of lack of leptin (as in the ob/ob mouse). However, it was found that the majority of humans did not lack leptin; in contrast they had circulating leptin concentrations that were highly correlated with adipose tissue mass(Reference Johnson, Huang, Figueroa-Colon, Dwyer and Goran7).
Leptin has effects on different systems of the body: central nervous system (inhibition of food intake, reduction of adipose mass, increased thermogenesis), glucose homeostasis, reproductive system (signalling onset of puberty, maintenance of reproductive function), autonomic nervous system, haematopoietic system, skeletal system, oncogenesis and transplantation (contributing to rejection of transplants)(Reference Moran and Phillip4).
The aim of the present study was to determine the relationship between serum leptin levels and clinical and biochemical features in overweight children and adolescents.
Subjects and methods
Overweight children and adolescents followed in a Paediatric Gastroenterology, Hepatology and Nutrition Unit from 1 February 1999 to 13 November 2006 who had had serum leptin levels determined were included in the study.
Clinical assessment included detailed history and physical examination: weight, height, Tanner stage, systolic and diastolic blood pressure, waist circumference, acanthosis nigricans and stretch marks. Weight measurement was performed using two different electronic equipments according to age (above 2 years, SECA model 763; under 2 years, SECA model 727; SECA GmbH and Co., Hamburg, Germany). Blood pressure was measured three times with a cuff appropriate to each arm size after 5 min rest. High blood pressure was considered when it was above the 95th percentile for age and sex. Waist circumference was measured by the same physician directly over the skin at a level midway between the lower rib margin and the iliac crest. Blood samples were collected after an overnight fast between 08.00 and 10.00 hours and the parameters analysed included glucose, aminotransferases (alanine aminotransferase, aspartate aminotransferase), total cholesterol, HDL and LDL, TAG, serum leptin, C-peptide and insulin. Leptin was measured by immunoradiometric assay (DSL-23100i ACTIVE® Human Leptin IRMA; Diagnostic Systems Laboratories Inc., Webster, TX, USA). Abdominal ultrasound was performed by the same physician to search for steatosis: increased hepatic parenchymal echotexture and vascular blurring(Reference Ramesh and Sanyal8). BMI was calculated by the standard formula (kg/m2). TAG, total cholesterol, HDL and LDL were analysed according to age and sex(Reference Kliegman, Behrman, Jenson and Stanton9).
Overweight was defined as BMI above the 85th percentile and obesity above the 95th percentile for age and sex according to the Centers of Disease Control(10). BMI and waist circumference z-score were calculated(Reference McDowell, Fryar, Hirsch and Ogden11). Insulin resistance was determined by the homeostasis model assessment of insulin resistance (HOMA-IR)(10). Non-alcoholic fatty liver disease was considered in the presence of elevated aminotransferases and/or steatosis in abdominal ultrasound, when other causes were excluded(Reference Schwimmer, Deutsch, Rauch, Behling, Newbury and Lavine12).
Statistical analysis was performed in the Statistical Package for Social Sciences (version 15.0; SPSS Inc., Chicago, IL, USA) using t tests, χ2, Pearson's correlation and linear regression. One outlier of serum leptin was excluded to perform correlation and regression. Significance was considered when P was below 0·05.
The Ethical Committee of S. Marcos Hospital approved the present study.
Results
The present study included 357 overweight children and adolescents with serum leptin levels determined; 53·2 % were girls. The median age of obesity onset was 3 years (range 1–13 years). The majority of patients were previously healthy; thirty-five patients had a previous diagnosis of asthma and one patient had a previous diagnosis of diabetes mellitus. There was a history of overweight in first-degree relatives in 62 %, arterial hypertension in 27·7 %, type 2 diabetes mellitus in 8·8 % and dyslipidaemia in 32·9 %. At the first visit, the mean age was 9·5 (sd 3·2) years and its distribution was similar in both sexes (girls, 9·2 years; boys, 9·7 years, P = 0·154). The mean BMI z-score was 1·72 (sd 1·34), in girls it was 1·71 (sd 1·16) and in boys it was 1·72 (sd 1·11).
Obesity was found in 330 children and adolescents (176 girls and 154 boys) and BMI was between the 85th and 95th percentiles in twenty-seven children (fourteen girls and thirteen boys); 19·4 % had a BMI above 30 kg/m2 and 5·9 % above 35 kg/m2. The mean waist circumference z-score was 1·58 (sd 1·05), 30·5 % had acanthosis nigricans and 17 % had stretch marks. There was high systolic blood pressure in 21·6 % and high diastolic blood pressure in 6·7 % (n 274).
The blood analysis results are shown in Table 1(Reference Kliegman, Behrman, Jenson and Stanton9, Reference Fisher13). There was hyperglycaemia (>5·6 mmol/l) in 4·5 %, high total cholesterol in 17·5 %, low HDL in 13·2 %, high LDL in 10·7 %, high TAG in 22·1 %, insulin resistance in 54·9 %, C-peptide above 0·73 nmol/l in 47·9 % and increased leptin levels, by sex and Tanner stage, in 93·6 %.
Table 1 Associations (r; Pearson's test) between serum leptin levels (ng/ml) and clinical and laboratory characteristics
(Mean values and standard deviations and correlation coefficients)
HOMA-IR, homeostasis model assessment of insulin resistance.
* P < 0·05.
† See Fisher(Reference Fisher13).
‡ See Kliegman et al. (Reference Kliegman, Behrman, Jenson and Stanton9).
The abdominal ultrasound revealed steatosis in 26·7 % (n 195). Non-alcoholic fatty liver disease was present in 46·4 % (n 224).
Serum leptin levels were significantly related to sex, Tanner stage, systolic blood pressure, BMI z-score, C-peptide, insulin, HOMA-IR and aspartate aminotransferase (Tables 1 and 2). We found no correlation between serum leptin and lipid profile or blood glucose.
Table 2 Associations of serum leptin levels (ng/ml)
(Mean values and standard deviations)
* P < 0·05.
† By t test.
In boys (n 166), leptin was significantly related to insulin (r 0·364; P < 0·001) and HOMA-IR (r 0·313; P < 0·001); it was not significantly related to BMI z-score (r 0·142; P = 0·068). In girls (n 190), leptin levels were significantly related to Tanner stage (mean: I–II 40·4 ng/ml, III–V 53·7 ng/ml; P = 0·009), BMI z-score (r 0·145; P = 0·046), C-peptide (r 0·309; P < 0·001), insulin (r 0·335; P < 0·001) and HOMA-IR (r 0·248; P = 0·001).
In the multivariate analysis, with leptin as the dependent variable and the variables with significant associations as independent variables, being female (P = 0·012) and greater BMI (P = 0·006) were determinant factors (Table 3). The associations between leptin and sex and between leptin and BMI z-score are shown in Figs. 1 and 2.
Table 3 Linear regression analysis with serum leptin as the dependent variable
HOMA-IR, homeostasis model assessment of insulin resistance.
* P < 0·05.
Fig. 1 Relationship between serum leptin concentration (ng/ml) and sex. The central line is the mean; the box represents the lower and upper quartiles; the whisker shows the maximum and minimum values; ○, outliers; * outlier excluded from analysis. The mean leptin concentration for girls was 48·0 ng/ml; for boys it was boys 34·4 ng/ml (P = 0·003; t test).
Fig. 2 Relationship between serum leptin concentration (ng/ml) and BMI z-score (r 0·136; P = 0·010; Pearson's test).
After 6 months of follow-up (n 286), with dietary and exercise recommendations and without any pharmacological treatment, the mean z-score BMI was 1·58 (sd 1·08); in girls it was 1·57 (sd 1·11) and in boys it was 1·59 (sd 1·05). Both in girls and in boys it was significantly lower than that at the first visit (P < 0·001 and P = 0·017 respectively) but the difference had no correlation with serum leptin levels.
Discussion
Our work shows several significant associations of clinical and biochemical parameters with serum leptin, but in a multivariate analysis only being female and increased BMI remain significant. As serum leptin correlates only with clinical variables, it was excluded in the obesity protocol of the Gastroenterology and Nutrition Unit of our Hospital after this result.
We exhibit evidence that girls have higher leptin levels than boys, as previously reported in obese(Reference Pilcová, Sulcová, Hill, Bláha and Lisá6, Reference Wabitsch, Blum, Muche, Braun, Hube, Rascher, Heinze, Teller and Hauner14) and non-obese(Reference Garcia-Mayor, Andrade, Rios, Lage, Dieguez and Casanueva15) children and adolescents. This difference increases through puberty as leptin increases from Tanner stage I to V in girls(Reference Carlsson, Ankarberg, Rosberg, Norjavaara, Albertsson-Wikland and Carlsson16); this might be explained at least in part by the higher testosterone levels in males which have a negative effect on leptin concentrations(Reference Wabitsch, Blum, Muche, Braun, Hube, Rascher, Heinze, Teller and Hauner14, Reference Blum, Englaro and Hanitsch17). Our data also reveal significantly higher leptin in Tanner stages III–V when compared with Tanner stages I–II.
The present study confirms previous findings of a positive correlation between serum leptin levels and BMI(Reference Steinberger, Steffen, Jacobs, Moran, Hong and Sinaiko5, Reference Pilcová, Sulcová, Hill, Bláha and Lisá6, Reference Wabitsch, Blum, Muche, Braun, Hube, Rascher, Heinze, Teller and Hauner14, Reference Carlsson, Ankarberg, Rosberg, Norjavaara, Albertsson-Wikland and Carlsson16–Reference Stylianou, Galli-Tsinopoulou, Farmakiotis, Rousso, Karamouzis, Koliakos and Nousia-Arvanitakis18) even after adjustment for age. Although it seems contradictory, because high leptin concentrations should inhibit additional increases in fat mass, this might be indicative of leptin resistance(Reference Johnson, Huang, Figueroa-Colon, Dwyer and Goran7).
If we compare males and females separately, BMI z-score no longer has significant association with leptin in boys but it can be due to the association of a weaker relationship in boys (r 0·142) compared with girls (r 0·145) and a smaller number of patients (166 compared with 190 girls). After adjustment, BMI z-score remains with a positive correlation with leptin levels.
We observed a positive association between serum leptin concentrations and waist circumference as previously described(Reference Steinberger, Steffen, Jacobs, Moran, Hong and Sinaiko5, Reference Pilcová, Sulcová, Hill, Bláha and Lisá6) but it disappeared after adjustment for age.
The relationship between leptin levels and high systolic blood pressure as described in boys in a former study(Reference Steinberger, Steffen, Jacobs, Moran, Hong and Sinaiko5) was no longer present after adjustment.
In contrast to a significant relationship between leptin and insulin resistance independently of body fatness described(Reference Steinberger, Steffen, Jacobs, Moran, Hong and Sinaiko5), we observed a significant association between leptin and fasting insulin and insulin resistance, but it also disappeared after multivariate analysis, as these parameters were associated with higher BMI.
It was important to perform multivariate analysis to exclude relationships between variables other than leptin, as verified with fasting insulin, insulin resistance, high systolic blood pressure and greater BMI.
We found no association between leptin and lipid profile as previous reported(Reference Pilcová, Sulcová, Hill, Bláha and Lisá6); nevertheless, there are data describing a significant correlation with TAG(Reference Steinberger, Steffen, Jacobs, Moran, Hong and Sinaiko5).
Although we did not find any relationship between BMI variation at 6 months and leptin, there are small investigations that found a negative correlation between leptin and weight gain(Reference Byrnes, Baur, Bermingham, Brock and Steinbeck19) and some other studies found higher leptin levels to predict weight gain over time(Reference Johnson, Huang, Figueroa-Colon, Dwyer and Goran7, Reference Fleisch, Agarwal, Roberts, Han, Theim, Vexler, Troendle, Yanovski and Yanovski20, Reference Savoye, Dziura, Castle, DiPietro, Tamborlane and Caprio21), which could be explained by the relationship between higher initial BMI and less adherence to diet and exercise and consequently weight gain, but there is one large study (n 263) in which this relationship was independent of initial BMI(Reference Byrnes, Baur, Bermingham, Brock and Steinbeck19).
In conclusion, we confirmed that girls have higher leptin levels than boys independently of the degree of obesity and that fasting serum leptin has a positive correlation with BMI independently of sex. Although leptin levels should not be regarded a less valuable parameter because of the associations found with anthropometrics or sex as it implicates endocrine changes, we do not consider that serum leptin levels should be included in the investigation protocol of our obese patients. In this large paediatric cohort other associations with leptin described in the literature can be discharged.
Acknowledgements
There are not any conflicts of interests in the present study. H. A. is the specialist that follows the patients included in the study and is the head of the Gastroenterology, Hepatology and Nutrition Unit. H. A. designed the study, performed the statistical analysis and wrote and reviewed the manuscript. C. S. reviewed the statistical analysis. S. C. entered some of the patients into the database and measured the leptin levels, performed the statistical analysis and wrote the manuscript.
