Common Genetic Components of Obesity Traits and Serum Leptin

Transcription

1 nature publishing group articles Common Genetic Components of Obesity Traits and Serum Leptin nn L. Hasselbalch, Beben Benyamin 2, Peter M. Visscher 2, Berit L. Heitmann,3, Kirsten O. Kyvik 4 and Thorkild I.. Sørensen To estimate common and distinct genetic influences on a panel of obesity-related traits and serum leptin level in adults. In a cross-sectional study of 625 Danish, adult, healthy, monozygotic, and same-sex dizygotic twin pairs of both genders, we carried out detailed anthropometry (height, weight, waist and hip, and skin-fold thickness, body composition assessment by bioimpedance (fat mass and fat-free mass), and measurement of serum leptin level. Bivariate variance component analyses estimated the additive genetic correlations between these measurements. The genetic correlations between the traits for overall fatness (BMI and fat mass index, kg/m 2 ) were 0.94 in men and 0.98 in women, and their correlations with the various local fatness measures ranged from 0.49 to 0.83 in men and from to 0.87 in women. The correlations between the truncal measures (waist circumference and truncal skin folds) and between the peripheral measures (hip circumference and peripheral skin folds) were and 0.47 in men and 0.7 and in women, respectively. The correlations between the truncal and peripheral measures ranged between 0.49 and in men and between and 0.82 in women. For leptin vs. the various measures of overall and local fatness the correlations ranged from 0.54 to in men and from 0.48 to in women. ll correlations were significantly <.00. The study supports control of overall fat mass and peripheral and truncal fat mass by both shared and different genetic components, which suggests that it is important to distinguish between the different phenotypes in the search for genes involved in the development of obesity. Obesity (2008) 6, doi:0.038/oby Introduction The causes of obesity have both genetic and environmental components as shown in a number of family, twin, and adoption studies (). Large family studies have consistently demonstrated a familial correlation in adult overall fat mass measured as BMI (weight/height 2 ), at ~0.2 between parents and offspring and at ~0.3 among siblings (). Twin and adoption studies show these correlations are attributable mainly to genetic influences rather than to the effects of shared environment (). The proportion of phenotypic variance ascribed to genetic variance (the heritability) is ~0.7 for BMI in twin studies, and the magnitude of heritability is about the same in the two genders (). However, the obesity phenotype appears to be a complex and composite phenotype in which different compartments of adipose tissue exhibits different biology and hence may have different genetic and environmental determinants of their sizes. The obvious differences between men and women in average body composition and body fat distribution reflect the biological complexity. Particularly, the difference between the trunk, and especially intra-abdominal fat mass, and the peripheral fat mass is important because of its implications for risk of cardiovascular disease and death (2). Moreover, the waist circumference for given levels of BMI is positively correlated with mortality, even within the normal range (3); thus, mortality is higher the greater the waist circumference for given BMI and lower the greater the BMI for given waist circumference. limited number of studies have examined whether genes influencing one obesity-related phenotype may also have an effect on other obesity-related traits, so-called pleiotropic effects. Belgian longitudinal study of adolescent twins has found such effects for BMI and skin-fold thicknesses and has found that the fat distribution track from adolescence to adulthood, mainly explained by additive genetic influences (4,5). On the other hand, there may be common determinants of the overall degree of fatness as well, and these may be reflected in the serum levels of leptin, which is secreted by the adipose tissue and assumed to play a major role in providing a feed back signal important in central body-weight regulation (6). Serum leptin level has been shown to be under the influence of genetic factors with heritability estimates ranging from 0.28 (ref. 7) to Institute of Preventive Medicine, Copenhagen University Hospital, Copenhagen, Denmark; 2 Queensland Institute of Medical Research, Brisbane, Queensland, ustralia; 3 Research Unit for Dietary Studies at the Institute of Preventive Medicine, Copenhagen University Hospital, Copenhagen, Denmark; 4 Institute of Regional Health Services Research and The Danish Twin Registry, University of Southern Denmark, Odense, Denmark. Correspondence: Thorkild I.. Sørensen (tias@ipm.regionh.dk) Received 8 July 2007; accepted 24 June 2008; published online 6 October doi:0.038/oby obesity VOLUM 6 NUMBR 2 DCMBR

2 0.55 (ref. 8). study of Finnish adult twins found that BMI and leptin were genetically highly correlated in both men and women (9). study of German male and female twins found a high genetic correlation between body fat % and leptin (7). In this study, we assessed, for men and women separately, to what extent the phenotypic correlations between a number of obesity traits may be under the influence of common genetic components. In our study, we included (i) the overall fat mass, measured as BMI and fat mass index (FMI; fat mass/height 2 ); (ii) the truncal fat mass, measured as waist circumference and sum of truncal skin-fold thicknesses (subscapular and suprailiac sites); (iii) the peripheral fat mass, measured as hip circumference and sum of extremity skin-fold thicknesses (biceps and triceps sites); (iv) fat-free mass index (FFMI; fat-free mass/ height 2 ); and (v) serum leptin level. Methods and Procedures This study is part of the larger nation-wide, population-based Danish Twin Registry, and is based on a study, entitled GMINKR, on the metabolic syndrome and related components. The GMINKR sample includes 625 complete same-sex twin pairs (53 monozygotic male pairs (MZM), 46 dizygotic male pairs (DZM), 58 monozygotic female pairs (MZF), and 68 dizygotic female pairs (DZF)) who at the time of examination had no known diabetes or cardiovascular disease, alcohol/drug abuse or other conditions (e.g., pregnancy and breast-feeding) making it difficult to participate in the clinical examination, which also included a bicycle test. The age range of the twins was 8 67 years with a mean age of 38 years. The sampling strategy secured that all twin pairs were complete. Zygosity was determined by polymorphic DN markers. The method of ascertainment and characteristics of the cohort are described in detail elsewhere (0,). The GMINKR study also included 3 opposite-sex dizygotic pairs, who were not included in the present analysis (see Statistical methods section). The participants height, weight, biceps, triceps, subscapular, and suprailiac skin-fold thicknesses were measured to the nearest 0. cm, and waist and hip circumferences to the nearest.0 cm. Measurements of height, weight, waist, and hip circumference were carried out on all twins. BMI was calculated as weight (kg)/height 2 (m 2 ). Measurements of skin folds were carried out in 49 twin pairs. The sum of extremity skin folds was calculated as the sum of the biceps and triceps measurements. The sum of truncal skin folds was calculated as the sum of the subscapular and suprailiac measurements. The methods are described in detail elsewhere (0). The bioelectrical impedance was measured using a BI-03 RJL-system analyser (RJL Systems, Detroit, MI) with a 50 khz, 800 micro mpere device, following the instructions given by the manufacturer. Body fat assessed by bioelectrical impedance was estimated for women and men using the following equation, where impr50 is the electrical impedance with a 50 khz, 800 micro mpere device, and where sex is 0 for women and for men: fat mass = 0.89 weight sex weight (height 2 /impr50) 0.23 height age (ref. 2). Since the machine in one of the centres broke down and had to be sent for repair measurements of bioelectrical impedance were carried out only on 384 twin pairs. Fat-free mass was calculated by subtracting the fat mass from the body weight. FMI and FFMI were calculated as fat mass (kg)/ height 2 (m 2 ) and fat-free mass (kg)/height 2 (m 2 ), respectively. The sample on which skin-fold measures and/or bioelectrical impedance measures were not obtained did not differ from the sample that had complete measures with regard to age, gender, or anthropometry (data not shown). Blood was taken after 0 2 h overnight fast. Plasma leptin concentration was derived from the blood samples using the BIOMOL Leptin (human) LIS K-53 kit (BIOMOL, Plymouth Meeting, P). Statistical methods To examine trait distributions descriptive statistics of the data were explored using SPSS statistical package version 3.0 for Windows (SPSS, Chicago, IL). To reduce skewness the variables BMI, serum leptin concentration, sum of extremity skin folds, sum of truncal skin folds, FMI and FFMI were ln-transformed for the analyses, except for the descriptive statistics. nalyses of twin data assume that intrapair variance of MZ twins is due to unique environmental factors and measurement errors, while intrapair variance in DZ twins in addition is affected by differences in genetic factors. It is also assumed that MZ and DZ twins share common environmental factors to a similar extent. Comparison of the intraclass correlation of the obesity traits in MZ twin pairs with that of DZ twin pairs can, therefore, provide a means of determining the genetic contribution to observed variation in the traits. Genetic, shared, and unique environmental contributions to the following groups of obesity-related phenotypes were investigated: (i) the overall fat mass, measured as BMI and FMI; (ii) the truncal fat mass, measured as waist circumference and sum of truncal skin folds (subscapular and suprailiac skin-fold thicknesses); (iii) the peripheral fat mass, measured as hip circumference and sum of extremity skin folds (biceps and triceps skin-fold thicknesses); (iv) FFMI; and (v) the serum leptin level. Univariate analyses were necessary before progressing to the bivariate analyses because there were model assumptions, which needed to be tested (such as equal phenotypic variance for MZ and DZ twins among men and women) and trait parameters to be estimated (such as intraclass correlations, variance components, and heritability). It also gave the opportunity to compare different causal variance models for each trait to see which provided the best overall fit for the bivariate analyses. Mx, a structural equation modeling package, was used to compute the goodness of fit of the models and maximum-likelihood parameter estimates (3). In this study, the tendency for DZ correlations to be about half or more than half of the MZ correlations led to the consistent use of a model partitioning the total phenotypic variance into additive genetic, shared environmental, and individual unique environmental variance, the latter including variance due to measurement error. lthough we have modeled and estimated variances under the assumption of no nonadditive genetic effects, our results do not rule out such effects. Bivariate analyses were carried out to examine whether the genetic and environmental effects on the obesity traits are correlated. The analyses explored to what extent the observed covariance between the traits can be accounted for by a correlation between the additive genetic effects on the traits. The analyses were carried out using the principles of bivariate covariance decomposition (Figure ). Because MZ twins are genetically identical, the correlation between their respective -component is.0. For DZ twins, who like other siblings, share on average 50% of their segregating genes, the corresponding value is 0.5. The correlation between the C component is by definition.0 in both MZ and DZ twins. The correlation between component is by definition zero and represents the environmental influence unique to the individual. The genetic correlation, r, expresses to what extent the same genotype, or common genetic component, contribute to the phenotypic correlation between the traits. r r r r.0/0.5.0/0.5 x z x z x z x z Trait Trait 2 Trait Trait 2 Twin Twin 2 Figure Bivariate correlation model (see Statistical methods for further explanation) VOLUM 6 NUMBR 2 DCMBR

3 Likewise, the environmental correlation, r, expresses the extent to which environmental influences are shared between the traits. s the anthropometric measures are correlated with age (4), the age of the twins was controlled for in the model (3,5). ffects of age were incorporated in the means model by linear regression of the phenotypes on age. ll analyses were conducted separately for men and women, and the outcome of the analyses were not compared statistically between the two genders for several reasons. First, as mentioned, there are obvious differences between men and women in average distributions of body composition and body fat distribution (see Table ), which have a genetically determined biological and physiological basis. Second, including the two genders in the same analyses would require that gender-specific effects were taken into account by analyzing also the opposite-sex dizygotic twin pairs, and this would rest on the untestable assumption that differences in intraclass correlations between the same-sex and oppositesex twin pairs were due to either genetic or environmental sources of variance. Third, the inclusion of gender-specific effects in the bivariate models would make them too complex and lead to too unreliable estimates with the present sample size. Results The means and the variances of the phenotypes were not significantly different between the zygosity groups, neither in men nor in women (Table ). The intraclass correlations for DZ twins tended to be more than half of the MZ correlations (Table 2). This suggests that the phenotypes are influenced by additive genes (), the common environment (C), and the individual environment (). This leads to consistent use of an C model in the univariate analyses. The univariate analyses demonstrated genetic and environmental influences for all the anthropometric measurements in both men and women (Table 3). For most traits the majority of the variation could be explained by the additive variation in genetic effects. The common environmental influence (C) on the traits was close to zero. The only trait for which the common environmental influence was significant was hip circumference. For all other traits the common environmental influence was nonsignificant, and because the exclusion of the C component in the model for these traits did not result in a significantly worse model fit, the presented heritability estimates are the result of an model, which is, therefore, also the basis for the bivariate models. ll anthropometric measurements included in this study correlate at the phenotypic level (Table 4). Generally the phenotypic correlations, r P, among the anthropometric measures Table Descriptive statistics (mean ± s.d.) of all traits across sex and zygosity MZM DZM MZF DZF Number of twin pairs with skin-fold measurements a with body composition a ge (years) 37.7 ± ± ± ± 0.0 Height (cm) 79.6 ± ± ± ± 6. Weight (kg) 80.6 ± ± ± ±.3 BMI (kg/m 2 ) 25.0 ± ± ± ± 3.9 FMI (kg/m 2 ) 5.4 ± ± ± ± 3.0 Hip circumference (cm) 96.3 ± ± ± ± ± ± ±. 3.3 ± 0.9 Waist circumference (cm) 89.3 ± ± ± ± ± ± ± ± 5.6 FFMI (kg/m 2 ) 9.3 ±. 9.3 ± ±. 6.7 ±.2 Leptin (mmol/l) 0.30 ± ± ±.08 ± DZF, dizygotic female twins; DZM, dizygotic male twins; FFMI, fat-free mass index; FMI, fat mass index; MZF, monozygotic female twins; MZM, monozygotic male twins. a The number of twin pairs with measurements of skin fold and body composition were less for technical reasons (see Methods and Procedures). Table 2 Intraclass correlations with 95% confidence intervals in parenthesis across sex and zygosity adjusted for age MZM DZM MZF DZF BMI (kg/m 2 ) 0.66 ( ) 0.40 ( ) ( 0.8) 0.49 (0.37 ) FMI (kg/m 2 ) 0.62 ( ) 0.32 ( ) ( 0.8) 0.46 (0.30 ) Hip (cm) 0.79 ( ) 0.5 ( ) 0.73 ( ) 0.5 (0.39 ) ( ) 0.37 ( ) 0.65 (0.54 ) 0.28 ( ) 0.55 ( ) 0.34 ( ) 0.65 ( ) 0.46 (0.33 ) 0.69 (0.58 ) 0.32 ( ) ( ) 0.5 ( ) FFMI (kg/m 2 ) 0.92 ( ) 0.56 ( ) 0.87 ( ) 0.53 ( ) Leptin (mmol/l) 0.35 ( ) 0.2 ( ) 0.5 ( ) 0.4 ( ) DZF, dizygotic female twins; DZM, dizygotic male twins; FFMI, fat-free mass index; FMI, fat mass index; MZF, monozygotic female twins; MZM, monozygotic male twins. obesity VOLUM 6 NUMBR 2 DCMBR

4 Table 3 stimates of standardized variance components with 95% confidence intervals in parentheses from the univariate genetic analyses Phenotype h 2 c 2 e 2 BMI (kg/m 2 ) Men 0.7 (0.7 ) 0.29 ( ) Women 0.76 ( 0.8) 0.24 ( ) FMI (kg/m 2 ) Men 0.66 (0.54 ) 0.34 ( ) Women ( ) 0.26 ( ) Hip (cm) Men 0.55 ( ) 0.25 ( ) 0.20 ( ) Women 0.50 (0.28 ) 0.24 ( ) 0.26 ( ) Men (0.62 ) 0.30 ( ) Women 0.65 ( ) 0.35 ( ) Men ( ) 0.40 ( ) Women (0.59 ) 0.33 ( ) Men ( ) 0.30 ( ) Women ( 0.78) 0.28 ( ) FFMI (kg/m 2 ) Men 0.92 ( ) 0.08 ( ) Women 0.88 ( ) 0.2 ( ) Leptin (mmol/l) Men 0.38 ( ) 0.62 ( ) Women 0.59 (0.48 ) 0.4 ( ) The estimates of C were close to zero for all traits, except for hip circumference. were moderate to high. The phenotypic correlations were very high (>0.85) between the measures of overall fatness (BMI and FMI) and waist circumference. High phenotypic correlations would be expected among the measures of peripheral fat mass as well as among the measures of truncal fat mass, but these correlations were at the same level as among the phenotype groups. Bivariate genetic analyses were performed on the traits with the aim of estimating the amount of common genetic and environmental influences among the traits (Figure ). The results of the bivariate variance component analyses are shown in Table 5 (additive genetic correlations, r ) and in Table 6 (environmental correlations, r ). The estimates for men are shown in the upper diagonal of the matrices and for women in the lower diagonal of the matrices. The additive genetic correlation, r, expresses to what extent the same genotype or common genetic component contribute to the phenotypic correlation between the two traits. It defines the correlation among the underlying additive genetic values of the traits and can, therefore, be greater than the heritability of either of the traits. The additive genetic correlations between the measures of overall fatness (BMI and FMI, kg/m 2 ) were very high, 0.94 in men and 0.98 in women. The genetic correlation between the measures of overall fatness and waist circumference tended to be high for men, indicating that the same genes are involved in the distribution of overall fatness and local fat mass centered around the waist. In women the measures of overall fatness had a high genetic correlation with both waist and hip circumferences, and in addition waist and hip demonstrated a high genetic correlation. The correlations between the traits for overall fatness and the various local fatness measures ranged from 0.49 to 0.83 in men and from to 0.87 in women. For the measures of peripheral fat mass (hip Table 4 Matrix of phenotypic correlations of obesity-related traits with 95% confidence intervals in parentheses; results for men above and for women below the diagonal BMI (kg/m 2 ) FMI (kg/m 2 ) Hip (cm) BMI (kg/m 2 ) 0.95 ( ) FMI (kg/m 2 ) 0.97 ( ) Hip (cm) ( ) ( ) 0.87 ( ) skin folds(cm) ( 0.8) FFMI (kg/m 2 ) 0.78 ( 0.82) Leptin (mmol/l) ( ) 0.66 ( ) ( ) 0.92 ( ) ( ) 0.90 ( ) ( ) ( ) (0.65 ) ( ) (0.58 ) 0.52 ( ) 0.69 ( 0.73) 0.83 ( ) ( ) ( ) 0.45 ( ) 0.87 ( ) 0.92 ( ) (0.65 ) 0.58 (0.52 ) 0.63 ( ) (0.65 ) 0.25 ( ) ( ) ( 0.76) ( 0.8) 0.63 ( ) ( 0.78) (0.65 ) (0.66 ) 0.58 (0.5 ) 0.62 ( ) FFMI (kg/m 2 ) ( ) ( ) ( 0.78) 0.52 ( ) 0.76 ( 0.80) 0.50 (0.43 ) 0.29 ( ) 0.62 (0.56 ) Leptin (mmol/l) 0.56 ( ) 0.65 (0.58 ) ( ) (0.56 ) ( ) ( 0.7) 0.22 ( ) 0.46 ( ) 2726 VOLUM 6 NUMBR 2 DCMBR

5 Table 5 Matrix of additive genetic correlations of obesity-related traits with 95% confidence intervals in parentheses; results for men above and for women below the diagonal BMI (kg/m 2 ) FMI (kg/m 2 ) Hip (cm) BMI (kg/m 2 ) 0.94 ( ) FMI (kg/m 2 ) 0.98 ( ) Hip (cm) 0.8 ( 0.85) ( ) 0.87 ( 0.90) ( 0.79) FFMI (kg/m 2 ) ( ) Leptin (mmol/l) 0.7 ( ) (0.5 ) 0.73 ( 0.80) ( ) ( ) ( ) 0.79 (0.73 ) 0.59 (0.49 ) (0.45 ) 0.49 ( ) ( ) 0.47 (0.35 ) ( ) 0.82 ( 0.86) 0.76 ( ) 0.0 ( ) 0.48 ( ) 0.83 ( ) 0.82 ( 0.87) 0.68 ( ) 0.49 (0.36 ) (0.5 ) ( ) 0.2 ( ) ( ) ( ) 0.73 ( ) ( ) ( 0.79) ( ) 0.7 (0.62 ) 0.22 ( ) (0.48 ) FFMI (kg/m 2 ) ( ) 0.52 ( ) 0.22 ( ) 0.3 ( ) 0.3 ( ) 0.38 ( ) 0.3 ( ) 0.68 (0.52 ) Leptin (mmol/l) (0.45 ) 0.54 ( ) 0.69 ( ) ( ) 0.65 (0.55 ) ( ) 0.32 ( ) 0.28 ( ) Table 6 Matrix of unique environmental correlations of obesity-related traits with 95% confidence intervals in parentheses; results for men above and for women below the diagonal BMI (kg/m 2 ) FMI (kg/m 2 ) Hip (cm) BMI (kg/m 2 ) 0.95 ( ) FMI (kg/m 2 ) 0.94 ( ) Hip (cm) 0.89 ( ) ( ) 0.86 ( ) ( 0.8) FFMI (kg/m 2 ) ( ) Leptin (mmol/l) 0.66 ( ) 0.82 ( ) ( ) 0.83 ( 0.87) ( ) 0.79 ( ) 0.8 ( 0.86) ( ) 0.59 ( ) 0.73 ( 0.80) ( 0.80) ( ) 0.68 ( ) 0.8 ( ) ( ) 0.49 ( ) 0.44 ( ) 0.9 ( ) 0.88 ( ) 0.78 ( ) 0.7 ( ) ( ) ( ) 0.47 (0.29 ) 0.39 ( ) skin fold (cm) ( ) ( ) ( ) ( ) 0.85 ( ) 0.66 (0.56 ) 0.63 (0.48 ) 0.55 ( ) FFMI (kg/m 2 ) ( ) 0.62 (0.47 ) (0.56 ) 0.49 (0.35 ) ( ) 0.47 ( ) 0.69 ( 0.78) 0.50 ( ) Leptin (mmol/l) 0.50 (0.37 ) 0.34 ( ) ( ) 0.59 ( ) 0.56 ( ) (0.53 ) 0.28 ( ) 0.42 ( ) and sum of extremity skin folds) the genetic correlation was 0.47 in men and in women. For the measures of truncal fat mass (waist and sum of truncal skin folds), the genetic correlation was in men and 0.7 in women. The correlations between the truncal and peripheral measures ranged from 0.49 to in men and from to 0.82 in women. For leptin vs. the various measures for overall and local fatness the correlations ranged from 0.54 to in men and from 0.48 to in women. The additive genetic correlations were high between FFMI and the measures of overall fatness, but moderate to low for the other anthropometric measures. s judged from the confidence intervals, all correlations were significantly <.00 and significantly >0.00 except for hip circumference vs. FFMI in women. The environmental correlations, r, express the extent to which common environmental factors contribute to the phenotypic correlation between the traits. These correlations were moderate to high between all other measures (Table 6). Discussion The study showed that the phenotypic correlations among various obesity traits were high. However, the extent to which the variances in the traits were explained by a common genetic component varied, but all were significantly <.00 and, with a obesity VOLUM 6 NUMBR 2 DCMBR

6 single exception, significantly >0.00. The results indicated that the phenotypes of BMI, FMI, and waist circumference were closely genetically related in men. In women also hip circumference seemed to be closely genetically correlated with BMI, FMI, and waist circumference. For the peripheral and truncal fat masses the genetic correlations between the traits were high (>0.5), but somewhat lower for the measures of overall fatness. The genetic correlations between the measures of peripheral fat mass as well as between the measures of truncal fat mass were not higher than many of the genetic correlations between the phenotypic groups. These results suggest that there exists strong common genetic factors for all the obesity traits, but also that there is a residual genetic influence on the various local fatness measures that cannot be explained exclusively by the genetic influence on overall fatness. Similar pattern of results were obtained in the analyses of the unique environmental influences. This study is based on a relatively large population-based twin sample with detailed physical assessment and DN-based zygosity determination. The precision of the direct measurements is phenotype dependent. In the existing literature, several studies have shown lower repeatability of skin-fold measurements than for other anthropometric measurements, which may lead to overestimation of the nonshared environment, and conversely, underestimation of the genetic influence on skin folds (6,7). In this study we have attempted to reduce the measurement error by repeating the measure at each skinfold site three times and subsequently used the average of three measures in the analyses. In general, however, biases in the various estimates due to errors of measurements, which may also be correlated, cannot be excluded although we expect them to be of minimal influence. In our attempt to remove the confounding of disease status and/or medical treatment on the intraclass correlations, we excluded subjects diagnosed with diabetes or cardiovascular disease. The smaller studies examining the common genetic influences on measures of overall fatness and leptin concentration, although using measured phenotypic traits, were usually based on selected groups of subjects with an unclear relation to the background population (7,8). The heritability of fasting serum leptin concentration and its relation to other obesity-related traits were in agreement with the result of the study by Kaprio et al. (9), but in contrast to the findings in a previous much smaller study, based on only 33 pairs (8,9). FMI ad FFMI proved to be fairly highly correlated, both regarding the phenotype and the underlying genetic and environmental effects. The increased FFMI may be a consequence of the various endocrine and metabolic disturbances that follows development of obesity, and it seems likely that underlying pleiotropic mechanisms may operate at some steps in the pathways from the genes through the phenotypes. Previous studies have suggested gender differences in genetic effects on BMI (8,9). Our results may suggest gender differences also on other obesity-related traits as well as of the underlying common genetic and environmental influences on these traits. However, as mentioned, this study is not suitable for a more detailed assessment of gender differences of the genetic architecture behind the obesity-related traits. With the increasing knowledge of the genetic influences on body weight and body composition, the ongoing challenge will be to identify the genes responsible for the genetic variance and map their functions. The results of this study indicate that there are significant genetic correlations between different measures of fatness. However, for all the obesity traits included in this study the genetic correlation was significantly below.00, indicating significant residual genetic influence specific for each traits. This stresses the importance of taking the specific obesity phenotype into account when searching for obesity-related genes. similar conclusion can be drawn from the analyses of the unique environmental influences. cknowledgments The GMINKR project was supported by grants from the Danish Medical Research Fund, the Danish Diabetes ssociation, the NOVO Foundation, the Danish Heart Foundation, and potekerfonden. K.O.K. has obtained support from the uropean Union Contract no. QLG2-CT This study was supported by the DIOGNS which is the acronym for Diet, Obesity and Genes supported by the uropean Community (Contract no. FP ), Disclosure The authors declared no conflict of interest The Obesity Society RFRNCS. Maes HH, Neale MC, aves LJ. Genetic and environmental factors in relative body weight and human adiposity. Behav Genet 997;27: Larsson B, Svärdsudd K, Welin L et al. bdominal adipose tissue distribution, obesity, and risk of cardiovascular disease and death: 3 year follow up of participants in the study of men born in 93. Br Med J (Clin Res d) 984;288: Bigaard J, Tjønneland, Thomsen BL et al. Waist circumference, BMI, smoking, and mortality in middle-aged men and women. Obes Res 2003;: Beunen G, Maes HH, Vlietinck R et al. Univariate and multivariate genetic analysis of subcutaneous fatness and fat distribution in early adolescence. Behav Genet 998;28: Peeters MW, Beunen GP, Maes HH et al. Genetic and environmental determination of tracking in subcutaneous fat distribution during adolescence. m J Clin Nutr 2007;86: Paracchini V, Pedotti P, Taioli. Genetics of leptin and obesity: a HuG review. m J pidemiol 2005;62: Jordan J, Brabant G, Brinsuk M et al. Heritability of free and receptor-bound leptin in normal twins. m J Physiol Regul Integr Comp Physiol 2005;288: R4 R Narkiewicz K, Szczech R, Winnicki M et al. Heritability of plasma leptin levels: a twin study. J Hypertens 999;7: Kaprio J, riksson J, Lehtovirta M, Koskenvuo M, Tuomilehto J. Heritability of leptin levels and the shared genetic effects on body mass index and leptin in adult Finnish twins. Int J Obes Relat Metab Disord 200;25: Schousboe K, Visscher PM, rbas B et al. Twin study of genetic and environmental influences on adult body size, shape, and composition. Int J Obes Relat Metab Disord 2004;28: Skytthe, Kyvik K, Holm NV, Vaupel JW, Christensen K. The Danish Twin Registry: 27 birth cohorts of twins. Twin Res 2002;5: Heitmann BL. valuation of body fat estimated from body mass index, skinfolds and impedance. comparative study. ur J Clin Nutr 990;44: VOLUM 6 NUMBR 2 DCMBR

7 3. Neale MC, Boker SM, Xie G, Maes H. Mx: Statistical Modeling. 6th edn. Department of Psychiatry: Richmond, V, Korkeila M, Kaprio J, Rissanen, Koskenvuo M. ffects of gender and age on the heritability of body mass index. Int J Obes 99;5: Neale MC, Cardon LR. Methodology for Genetic Studies of Twins and Families. Kluwer: Dordrecht, the Netherlands, Hansen S, Cold S, Petersen PH, Rose C. stimates of the sources of variation (variance components) of bioelectric impedance and anthropometric measurements in an epidemiological case-control study of breast cancer. ur J Clin Nutr 997;5: Nordhamn K, Sodergren, Olsson et al. Reliability of anthropometric measurements in overweight and lean subjects: consequences for correlations between anthropometric and other variables. Int J Obes Relat Metab Disord 2000;24: Schousboe K, Willemsen G, Kyvik KO et al. Sex differences in heritability of BMI: a comparative study of results from twin studies in eight countries. Twin Res 2003;6: Korkeila M, Kaprio J, Rissanen, Koskenvuo M. ffects of gender and age on the heritability of body mass index. Int J Obes 99;5: obesity VOLUM 6 NUMBR 2 DCMBR