Between April 2005 and January 2006, breast milk and venous blood samples were obtained from healthy lactating women who gave written, informed consent to participate in the study. The study protocol, the contents of the written information sheet, and the consent form were approved by the Ethics Committee of Uludag University Medical School, Bursa, Turkey.

The study design was longitudinal and cross sectional. Twenty-two breastfeeding women [age (years) = 29 ± 2 (mean ± standard error of the mean); weight (kg) = 69 ± 3; height (cm) = 161 ± 2 cm; BMI (body mass index, kg/m²) = 26 ± 2] participated in the longitudinal part of the study. In the longitudinal component of the study, changes of milk leptin concentration were investigated only during the first 30 days of the lactation period due to recruitment and retainment problems of subjects for a longer period. Breast milk samples were obtained from the same women at 1–3 days, 4–14 days and 15–30 days after birth.

The cross-sectional component of the study was designed to investigate changes in milk leptin concentrations during a longer period, from 0 to 180 lactation days. Five groups of exclusively breastfeeding women participated in the cross-sectional study. The first group consisted of 37 breastfeeding women [age (years) = 30 ± 2; weight (kg) = 70 ± 3; height (cm) 160 ± 2 cm; BMI = 27 ± 2] enrolled at days 1 to 3 after birth (expressing colostrum). The second group consisted of 27 breastfeeding women [age (years) = 31 ± 2; weight (kg) = 67 ± 2; height (cm) 159 ± 2 cm; BMI = 26 ± 1] enrolled at days 4–14 after birth (expressing transitional milk). The third group consisted of 16 breastfeeding women [age (years) = 29 ± 1; weight (kg) = 69 ± 3; height (cm) 159 ± 2 cm; BMI = 27 ± 1] enrolled at days 15–30 after birth (expressing early mature milk). The fourth group consisted of 37 breastfeeding women [age (years) = 29 ± 2; weight (kg) = 66 ± 3; height (cm) 161 ± 2 cm; BMI = 25 ± 1] enrolled at days 31–90 after birth (expressing mature milk). The fifth group consisted of 43 breastfeeding women [age (years) = 30 ± 2; weight (kg) = 65 ± 3; height (cm) 159 ± 2 cm; BMI = 26 ± 1] enrolled at days 91–180 after birth (expressing late mature milk). These 5 cross-sectional maternal groups were demographically and biologically equivalent with regard to age [F(4,155) = 0.140; P = 0.967], weight [F(4,155) = 0.517; P = 0.723], height [F(4,155) = 0.193, P = 0.942] and BMI [F(4,155) = 0.364, P = 0.834]. The power of performed test with alpha equal to 0.05 was 0.049 in all of these analyses.

Inclusion criteria were women who gave birth at term after uncomplicated pregnancies. Exclusion criteria were lactation failure, any existing disease (i.e. maternal disease due to pregnancy or unrelated and pre-existing or neonatal disease), under hormonal and any drug therapy, and body mass index greater than 35.0 at enrolment.

Fore milk sample collection was standardized by milk volume and time: 5 ml were collected in plastic tubes by hand expression 3 hours (h) after breakfast (bread, butter, olive, cheese and tea). Samples were vortexed, divided into 500 μl aliquots (1.0 ml Eppendorf tubes) and frozen at -20°C until analysis. Maternal blood (5 ml) was collected into a plain evacuated glass tube at the same time as the milk sample. Blood samples were centrifuged within 30 min at 2000 × g for 10 min at 4°C. Serum was divided into 500 μl aliquots and stored at -20°C until analysis. Milk samples were thawed overnight in the refrigerator and vortexed continuously to ensure sample uniformity. Given evidence that leptin levels are higher in whole milk than in skim milk 4511 and that sonication yields significantly increased levels 511, we analyzed whole milk samples after sonication (Bandelin Electronic, UW 70, D-1000 Berlin 45, Germany) with 8–10 bursts of 10 sec in duration each.

We used a human leptin immunoradiometric assay kit (Human Leptin IRMA, DSL-23100; Webster, Texas, USA) for the measurements of leptin in breast milk and serum. All milk and serum samples from the individual subjects were assayed in duplicate in the same test battery. Intra- and inter- assay coefficients of variation (CV) for the leptin IRMA were < 5% and < 6.7 %, respectively, with a detection limit 0.1 ng/ml using 100 μl samples.

Chemiluminescent microparticle immunoassay (CMIA) was used for the quantitative determination of estradiol, prolactin, and thyroxine by using Architect 8200^® autoanalyser (Architect 8200, Abbott). Serum cortisol was measured by radioimmunoassay using a commercially available kit (Diagnostic Products Corporation, Los Angeles, CA, USA). Insulin was measured by chemiluminescent immunometric assay with an Immulite 2000^® Analyzer (Diagnostic Products Corporation, Los Angeles, CA, USA).

Statistical analysis was performed using SigmaStat V2.03 (SPSS Science Software GmbH, Erkrath, Germany) for Windows. The Kolmogorov-Smirnov test was used for normality. Differences in the log values of whole milk leptin concentrations and their changes with the lactation periods were examined with one-way ANOVA followed by Tukey's pairwise multiple comparison method. In the longitudinal study, changes in the log values of serum leptin concentrations in lactating women at various time points during 1 to 30 days lactation period were conducted with one-way repeated measure ANOVA followed by Tukey's multiple comparison method. Longitudinal changes in serum hormone levels during 180 days of lactation were examined with one-way ANOVA followed by Tukey's pairwise multiple comparison method. The relationships between two variables were determined by Pearson's correlation and multiple regression analysis. Data are given as mean ± standard error of the mean (SEM); p values less than 0.05 were considered significant in all tests.

Leptin concentration in breast milk showed a lognormal distribution. There was a wide range of leptin concentrations in colostrum (0.16–7.0 ng/ml) and mature milk (0.11–4.97 ng/ml).

In the longitudinal study, leptin levels in colostrum (collected at 1–3 days) were 3.35 ± 0.25 ng/ml and then fell significantly [F(2,42) = 30.058, p < 0.001; power of the performed test with alpha = 0.050: 1.000] to 2.65 ± 0.21 ng/ml (p < 0.05) in transitional milk (expressed at 4–14 days) and to 1.63 ± 0.18 ng/ml (p < 0.001) in early mature milk (Figure 1A).

In the cross sectional study, the profile of the changes in breast milk leptin concentrations with time was similar to the profile observed in the longitudinal study. Breast milk leptin concentrations were highest in colostrum (3.28 ± 0.21 ng/ml), breast milk expressed 1 to 3 days after birth, and then fell significantly [F(4,155) = 14.34, p < 0.001; power of the performed test with alpha = 0.050: 1.000] during the 180 days lactation period (Figure 1B). Leptin concentrations in breast milk expressed 31–90 days or 91–180 days were lower than the observed concentrations in colostrum (p < 0.001) and in transitional milk (p < 0.01). Regression analysis revealed a significant inverse relation (r = -0.694, p < 0.001; power of the performed test with alpha = 0.050: 1.000) between breast milk leptin concentrations and lactation days.

The mean concentrations of serum leptin, cortisol, insulin, thyroxine, estradiol, and prolactin in the five study groups are shown in Table 1. Serum leptin concentrations were highest during days 1–3, and then fell slightly [F(4,155) = 3.67, p < 0.05; power of the performed test with alpha = 0.050: 1.000] during the 180 days lactation period (Table 1).

Lactating women expressing transitional (postpartum days 4–14) and mature milk (postpartum days 15 to 180) had significantly lower (p < 0.05–0.001) serum cortisol, estradiol and prolactin concentrations than lactating women expressing colostrum at postpartum days 1 to 3 (Table 1).

The relationships between colostrum (expressed at 0–3 days) and mature milk (expressed at 15–180 days) leptin concentrations and maternal serum hormones were determined in the cross-sectional component of the study. Regression analysis revealed significant correlations between colostrum leptin concentrations and maternal serum leptin (r = 0.425, p < 0.01), cortisol (r = 0.549, p < 0.01) and thyroxine (r = -0.0530, p < 0.01) concentrations (Figure 2A, B, C). There was no correlation between colostrum leptin concentrations and serum insulin (r = 0.159, p = 0.57), prolactin (r = 0.120, p = 0.66) or estradiol (r = -0.180, p = 0.29) concentrations (data not shown). Correlation analysis between log values of colostrum leptin concentrations and maternal serum hormones revealed similar results (data not shown).

Regression analysis revealed significant correlations between breast milk leptin concentrations during days 15–180 and maternal serum leptin (r = 0.547, p < 0.001), thyroxine (r = -0.329, p < 0.01) and insulin (r = 0.331, p < 0.01) concentrations (Figure 2D, E, F). There was no correlation between mature milk leptin concentrations and maternal serum cortisol (r = 0.119, p = 0.25) prolactin (r = 0.130, p = 0.24) or estradiol (r = 0.061, p = 0.56) concentrations (data not shown). Correlation analysis between log values of mature milk leptin concentrations and maternal serum hormones revealed similar results (data not shown).

Serum leptin concentrations were correlated with serum thyroxine (r = -0.327, p < 0.05), insulin (r = 0.648, p < 0.001), estradiol (r = 0.639, p < 0.001) and prolactin (r = 0.530, p < 0.001) during days 1–3 postpartum (Figure 3). In the group expressing mature milk, serum leptin concentrations were also correlated with serum thyroxine (r = -0.345, p < 0.001) and insulin (r = 0.271, p < 0.01), but not with serum estradiol (r = 0.115, p = 0.27) or prolactin (r = -0.134, p = 0.20). There was no correlation between serum leptin and cortisol concentrations in breastfeeding women either expressing colostrum or mature milk (data not shown).