The factors that regulate leptin production in adipose tissue are still largely unknown. The previous studies that showed a positive correlation between body mass index or percentage of body fat and serum leptin levels have two shortcomings. First, they were done in unrelated persons; thus, a possible contribution of genetic factors to this association cannot be excluded. Second, these studies did not examine a possible role of body fat distribution in serum leptin levels. This is of interest because high serum insulin levels are accompanied by increased leptin levels [11], and obesity of the upper body involving a large amount of abdominal visceral fat is associated with high insulin levels [12-14]. To avoid the potentially confounding influence of genetic factors, we examined the effect of adiposity and body fat distribution on plasma leptin levels in 23 pairs of identical twins who were discordant for obesity.

Methods

Top Methods Results Discussion Author & Article Info References

Participants

The Finnish Twin Cohort includes all pairs (4307 monozygous, 9581 same-sexed dizygous pairs) of adult Finnish twins who were born before 1958 and were alive in 1975 [15]. To obtain information on the twins' height and weight, questionnaires were sent to the twins in 1975, 1981, and 1990. On the basis of responses to the 1990 questionnaire, we identified 50 pairs of identical twins who were born between 1932 and 1957 and were discordant for obesity. Discordance for obesity was defined as a difference in body mass index of at least 4 kg/m²; in addition, the body mass index had to be greater than 27 kg/m² in the obese twin and less than 25 kg/m² in the lean twin. From these 50 pairs of twins, we excluded persons who had a history of thyroid disorders, psychiatric diseases, diabetes, major musculoskeletal problems, or any other disease affecting insulin sensitivity and persons who were taking medications (for example, diuretics or ß-blockers) that may affect glucose metabolism. On the basis of a response letter, all eligible twin pairs were invited to take part in our study if they still met the criteria. Twenty-eight twin pairs were examined.

Two pairs were excluded after physical examination showed the difference in body mass index to be less than 3 kg/m². Three pairs that had a body mass index difference of 3 to 4 kg/m² were included in the final study sample. One pair was excluded after the obese twin was found to have previously undiagnosed overt diabetes mellitus.

Zygosity of the twin pairs was originally determined on the basis of responses to a self-administered questionnaire on similarity during the school years and was validated by examination of 11 blood group markers [16]. To confirm the monozygosity of the pairs that we studied, an expert did dermatoglyphic analysis of fingertip prints [17, 18]. All but six pairs were confirmed to be monozygotic. Samples of DNA from the six pairs in which zygosity was uncertain were typed for markers at six different polymorphic gene loci (DIS80, APOB, D17S30, COL2A1, VWA, and HUMTH). Four of the six pairs were found to be monozygotic; the other two were dizygotic and were excluded. Thus, the final study sample consisted of 23 nondiabetic identical twin pairs (14 female pairs and 9 male pairs) who had a difference in body mass index greater than 3 kg/m², had no diseases, and were not receiving continuous treatment with medication. Patient characteristics are shown in Table 1.

The local ethics committee approved the study, and all patients gave informed consent.

Biochemical Analysis

Fasting plasma leptin levels were determined in duplicate by a radioimmunoassay kit (Linco Research, Inc., St. Charles, Missouri) as previously described [7]. Briefly, plasma samples (100 micro L) were incubated overnight with ¹²⁵I-labeled leptin and with leptin antibody at 4 °C. The next day, bound fraction was separated by using a precipitating reagent with a 20-minute incubation; the fraction was centrifuged, the supernatant was decanted, and the pellets were counted. The calibration range was 0.5 µg/L to 100 µg/L, and the within-assay variation was 8%.

To estimate insulin resistance, an oral 2-hour glucose (75 g) tolerance test was done; insulin levels were measured at 0, 30, 60, 90, and 120 minutes. Plasma insulin levels were measured by using a commercial radioimmunoassay method (Pharmacia Diagnostics, Uppsala, Sweden). The antiserum of this kit is specific for insulin and does not cross-react with proinsulin or C-peptide (cross-reactivities < 0.1%). We used the area under the plasma insulin curve as an estimate of insulin resistance.

Analysis of Adiposity

Adiposity was expressed as body mass index (kg/m²) and the percentage of body fat. The latter was determined using the four-component method [19]; this method is based on the division of body mass into four components, each with a different density: fat tissue (density, 0.9007 g/cm³), water (density, 0.994 g/cm³), minerals (density, 3.042 g/cm³), and proteins (density, 1.34 g/cm³). Water mass was estimated by using the bioelectrical impedance method (BIA-101A/S, RJL System, Inc., Clemens, Michigan) [20]. Mineral mass was estimated by using dual-energy x-ray absorptiometry (Norland XR26, Norland Corp., Fort Atkinson, Wisconsin). The density of the whole body was estimated by underwater weighing and was corrected for information on body water and mineral mass. The proportion of fat tissue was calculated from the density of the whole body according to the formula of Siri [21].

The distribution of body fat was measured by using magnetic resonance imaging [22]. Imaging was done at 0.1 T (Mega4, Instrumentarium Co., Helsinki, Finland). Axial and sagittal localizers were used to obtain a transaxial T1-weighted image (TR [relaxation time]/TE [echo time], 155/20; slice thickness, 10 mm) at the level of the fourth lumbar vertebra. Visceral and subcutaneous fat areas were measured. Magnetic resonance imaging was not done in three pairs of twins (two female pairs, one male pair) either because the participants had claustrophobia or because the imaging equipment was temporarily malfunctioning. To classify the twins according to the degree of visceral fat accumulation, the twin pairs were divided according to the obese twin's median abdominal visceral fat area, as described elsewhere [23]. Because men have a higher proportion of visceral fat than women, separate median values were used (109 cm² for men and 58 cm² for women).

Statistical Analysis

We used a paired t-test to compare means between obese and lean twins. Because of its skewed distribution, leptin was analyzed after logarithmic transformation. Pearson correlation coefficients were calculated to quantify the association between intrapair differences in leptin levels and corresponding differences in other variables. When comparing leptin levels between men and women, we used analysis of covariance to adjust for percentage of body fat. Multiple regression analyses were done to determine the independent contribution of intrapair differences in visceral or abdominal subcutaneous fat to the difference in leptin levels between obese and nonobese twins. The data are presented as the mean ±SD.

Results

Top Methods Results Discussion Author & Article Info References

The average difference in weight between the obese and lean twins was 16 kg in men and 19 kg in women (Table 1). Accordingly, the intrapair difference in body mass index and percentage of body fat was greater in women than in men. Despite the moderate differences between the sexes in body mass index, women had a much higher percentage of body fat than did men, among both the obese and the lean twins. In both men and women, the abdominal visceral fat area was more than twice as great in the obese twins as in the lean twins. The difference in the subcutaneous abdominal fat area between obese and lean twins was also significant in both sexes but was greater among female pairs than male pairs.

For both men and women, plasma leptin levels were approximately threefold higher in obese twins than in their nonobese siblings (Table 1). Obese and lean women had leptin levels that were three times higher than those in men. To determine whether these higher leptin levels were explained by the higher percentage of body fat in women, we compared leptin values in the men and women who had overlapping body fat percentages ranging from 24.5% to 29.5%. There were no twin pairs in which both members were in this subgroup. The mean percentage of body fat in these seven men (26.4% ± 1.6%) and seven women (26.9% ± 1.6%) was similar. Plasma leptin levels were 8.3 ± 6.9 µg/L in men and 9.4 ± 4.0 µg/L in women (P > 0.2). Moreover, when all men and women were compared by analysis of covariance, with sex and percentage of body fat as covariates, adjusted leptin levels were similar in men and women (12.9 ± 4.2 µg/L and 11.5 ± 3.2 µg/L, respectively).

To characterize the contribution of fat distribution to leptin levels, we divided the twin pairs into subgroups according to whether the obese twin had a high or low visceral fat area (Table 2). In obese twins with a high visceral fat area, plasma leptin levels were 3.7-fold higher than those in their nonobese siblings; in obese twins that had a low visceral fat area, plasma leptin levels were 2-fold (in men) or 2.6-fold (in women) higher than those in their nonobese siblings.

To further characterize the relation between plasma leptin levels and body fat distribution, we determined the correlation between intrapair differences (that is, obese compared with nonobese twins) in adiposity indices and intrapair differences in leptin levels (Table 3 and Figure 1). In both sexes, intrapair differences in abdominal visceral fat area correlated positively with differences in plasma leptin levels. In women only, differences in body mass index and percentage of body fat correlated with differences in leptin levels. In men only, intrapair differences in the area under the plasma insulin curve, an estimate of insulin resistance, correlated with differences in plasma leptin levels. In neither sex did differences in abdominal subcutaneous fat area correlate with differences in leptin levels.

View larger version (17K): [in this window] [in a new window]   Figure 1. Correlation between intrapair differences in percentage of body fat (top) or abdominal visceral fat area (bottom) and intrapair differences in plasma leptin levels. \#9679; = women; = men.

To ascertain whether abdominal visceral or subcutaneous fat area is associated with plasma leptin levels independent of total body fat, we did two sets of multiple regression analyses. The intrapair difference in plasma leptin levels was the dependent variable, and intrapair differences in body fat distribution measures and percentage of body fat were explanatory variables. In the first model, differences in visceral fat area and percentage of body fat were included. In men, the regression coefficient was significant (P = 0.029) for the intrapair difference in visceral fat area but was not significant (P = 0.069) for the difference in the percentage of body fat. The whole model explained 73% (P < 0.001) of the variation in the intrapair difference in plasma leptin levels. In women, the regression coefficient was not significant for the difference in visceral fat area (P > 0.2) but was highly significant for the difference in percentage of body fat (P < 0.001). The whole model explained 81% (P < 0.001) of the variation in the intrapair differences in plasma leptin levels.

In the second model, the regression coefficient was significant for the difference in percentage of body fat (P = 0.003 in men and P < 0.001 in women) and was not significant for the difference in abdominal subcutaneous fat area (P > 0.2 for men and women). The whole model explained 61% (P = 0.002) of the variation in the intrapair difference in plasma leptin levels in men and 82% (P < 0.001) of the variation in women.

Discussion

Top Methods Results Discussion Author & Article Info References

We found that serum leptin levels were higher in obese twins than in nonobese twins, that women had higher leptin levels than did men, and that visceral adiposity had a greater effect on serum leptin levels than did peripheral adiposity. The differences in leptin levels that were related to overall obesity are similar to results of previous studies of persons who were not monozygotic twins [6, 7]. Because the twins we studied were genetically identical, our results indicate that the association between adiposity and leptin levels is a consequence of differences in adipose mass and is not confounded by genetic factors. This idea is further supported by the close correlation between the intrapair differences in plasma leptin levels and the intrapair differences in abdominal visceral fat area in both sexes and in percentage of body fat in women. Our data indicate that the increased leptin levels seen with mild to moderate obesity are not genetically determined.

Serum leptin levels were approximately threefold higher in women than in men, both among the obese and nonobese twin subgroups and in the face of similar body mass indices. Other studies [6, 7] have reported that leptin levels are higher in women that in men. In our study, the obese women had almost 14 percentage points more body fat than did the obese men, and the lean women had 7 percentage points more fat than did the lean men. Subgroups of men and women with equal percentages of body fat had almost identical plasma leptin levels. In addition, serum leptin levels adjusted for percentage of body fat were similar in men and women. These data indicate that the difference between the sexes in plasma leptin levels can be explained by differences in adiposity.

Several observations suggest that the effect of visceral fat rather than peripheral fat is greater in the regulation of plasma leptin levels; this is particularly true in men. First, the difference in plasma leptin levels between obese and nonobese twins was greater among persons with high amounts of visceral fat than among those with low amounts. Second, simple correlation analysis showed that in men, intrapair differences in plasma leptin levels correlated with differences in visceral fat but not with subcutaneous abdominal fat or total body fat. Third, in the multiple regression analysis, differences in visceral fat area alone contributed significantly to variations in serum leptin levels in men but not in women. In women, visceral fat probably makes up less than 10% of the total fat mass [24]. The large amount of subcutaneous fat in women may have masked the influence of visceral fat on leptin levels, particularly because intrapair differences in leptin levels in women also correlated with differences in body mass index and percentage of body fat. Fourth, a close correlation was seen in men between intrapair differences in leptin levels and areas under the plasma insulin curve. Because visceral fat is primarily associated with insulin resistance, this association indirectly supports the importance of visceral fat in the regulation of leptin levels. In contrast to our data, the expression of leptin messenger RNA is equally elevated in omental and subcutaneous fat cells of obese persons [25]. It is not known, however, whether similar expression of leptin messenger RNA also results in equal production of protein.

Our study included 23 pairs of identical twins who were discordant for obesity. The discordance allowed us to compare intrapair differences in obesity-related factors and leptin levels. However, when the study group was divided according to sex, the final number of participants became small. The group was also highly selected because discordance for obesity was required. However, the well-defined group of identical twins allowed us to draw conclusions; if the participants had been genetically different, we would have needed to examine a much larger sample.

Accurate measurement of total visceral fat mass by using magnetic resonance imaging requires scanning the entire abdominal region [26]. We used quantitation of visceral fat area at the level of the fourth lumbar vertebra. The area measured at this level correlates well with total visceral fat mass but is slightly less accurate than the area measured by scanning the entire abdominal region [24]. Additional studies are needed to show whether total visceral fat mass is an even stronger determinant of plasma leptin levels than the cross-sectional visceral fat area used in our study.

Regarding the physiologic and clinical implications of our findings, it is worth noting that visceral rather than peripheral fat is the main tissue for short-term storage or release of energy [24]. It is therefore plausible that leptin production by visceral fat could have a major role in the regulation of fuel homeostasis. Our results also indicate that in obese persons, high leptin levels caused by leptin resistance are probably not genetically determined.

In conclusion, our findings in identical twins who were discordant for obesity indicate that plasma leptin levels are increased in persons with mild to moderate obesity independent of genetic background. Higher leptin levels in women can be explained by their greater percentage of adipose tissue. Visceral fat may be of major importance in the regulation of leptin levels, but its contribution in women is probably less than that in men because of the greater amount of peripheral fat in women.

Dr. Karonen: Department of Clinical Chemistry, Helsinki University Central Hospital, FIN-00290 Helsinki, Finland.

Dr. Rissanen: Department of Psychiatry, Helsinki University Central Hospital, FIN-00180 Helsinki, Finland.

Dr. Koskenvuo: Department of Public Health, University of Turku, FIN-20520 Turku, Finland.

Dr. Koivisto: Department of Medicine, Helsinki University Central Hospital, FIN-00290 Helsinki, Finland.