The Q223R Polymorphism of the Leptin Receptor Gene Is Significantly Associated with Obesity and Predicts a Small Percentage of Body Weight and Body Composition Variability

Genetic variation at the leptin receptor gene locus may play an important role in the pathophysiology of human obesity, a leptin-resistant state. Previous studies exploring potential associations between leptin receptor gene polymorphisms and obesity have reported conflicting results. The aim of this study was to evaluate a genetically homogeneous population for associations between body composition variables and three common leptin receptor gene polymorphisms (K109R, Q223R, and K656N) that have potential functional significance as well as to assess the contributions of these polymorphisms to the variability of obesity. One hundred and eighteen consecutively enrolled subjects (62 women: mean age, 17.5 ± 1.6 yr; body mass index range, 16.2–30.1; 56 men: mean age, 17.8 ± 1.8 yr; body mass index range, 15.4–35.9) were genotyped for the three polymorphisms, and their body mass index, sum of 4 skinfolds, fat-free mass, percent fat mass, serum leptin levels, caloric intake, fat intake, and exercise patterns were determined. Allele frequencies were estimated by the gene-counting method, and genotype distributions between 89 normal weight (body mass index, ≤25 kg/m²) and 29 overweight-obese (body mass index, >25 kg/m²) subjects were compared using χ² test (using codominant, dominant, and recessive models). Analysis of covariance was also performed to evaluate associations between the polymorphisms and body composition variables after controlling for potential confounders. For the Q223R polymorphism, there was a higher prevalence of the R223 allele in the homozygous form among overweight-obese subjects vs. normal weight subjects (20.7% vs. 4.5%; P = 0.01). Furthermore, simple and multiple regression analyses revealed that the R223 allele in the homozygous form is a significant predictor of both body mass index (P = 0.015) and percent fat mass (P = 0.02) even after adjusting for age and gender and explains 4.5% of the variance in percent fat mass and 5% of the variance in body mass index. There was no significant difference in allele frequencies or genotype distributions for the K109R or K656N polymorphisms. These findings support the hypothesis that the Q223R polymorphism (but not the K109R or K656N polymorphism) of the leptin receptor gene is associated with obesity and predicts a small percentage of body weight and body composition variability in a genetically homogeneous population.

OBESITY IS DUE to the combined effects of genes, environment, lifestyle, and the interactions of these factors (1, 2). The discovery of leptin, the protein product of the ob gene, and its receptor has greatly advanced our understanding of the mechanisms regulating energy homeostasis and body weight (3, 4). Leptin is an adipocyte-derived hormone that acts to reduce food intake and increase energy expenditure by binding and activating its specific receptor in the hypothalamus (2, 4, 5). The leptin receptor is a single transmembrane protein belonging to the superfamily of cytokine receptors and has several alternatively spliced isoforms (one long isoform and several short isoforms) that are distributed in many tissues (2, 3–6). The biologically active long isoform is abundantly expressed in the hypothalamus, where it activates the Janus kinase-signal transducer and activation of transcription (STAT) system to alter the expression of hypothalamic neuropeptides (5, 6). Single nucleotide mutations of the leptin gene (resulting in a truncated protein) (7) or the leptin receptor (LEPR) gene (resulting in a premature termination of the intracellular domain) (8) are responsible for the morbid obesity seen in mouse models of obesity (ob/ob and db/db mice, respectively) (7, 8) as well as a few rare cases of obesity in humans (9–11). Although human obesity is generally not thought to be a monogenic disorder, leptin levels increase with increasing amounts of fat mass (12, 13), suggesting that obesity is a leptin-resistant state in humans. Therefore, it has been previously suggested that genetic variation at the leptin receptor locus and/or postreceptor defects may play a significant role in the pathophysiology of human obesity.

Several common polymorphisms and rare variants of the long isoform of the human LEPR gene have been reported, and the potential associations of these polymorphisms with obesity have been evaluated in different populations. Although two linkage and/or association studies have found significant associations between body composition variables and polymorphisms of the LEPR gene (14, 15), five previous studies failed to find significant results (16–20). It is possible that these conflicting results may be due in part to differences in the genetic backgrounds of the studied populations. In addition, a weak association might have been missed due to the relatively limited numbers of subjects studied (16, 21) or to suboptimal statistical analysis (e.g. lack of genotype evaluation using several different genetic models). Faced with the above conflicting data, we used restriction fragment length polymorphisms detecting amino acid substitutions in exons 4, 6, and 14 of the LEPR gene (K109R, Q223R, and K656N polymorphisms, respectively) to test for associations with body composition variables in 120 young subjects from a genetically homogeneous population, and we assessed the potential contributions of these variables to variability of obesity, an analysis that has not been performed in any previous study.

Subjects and Methods

Subjects

One hundred and twenty consecutively enrolled Greek students participated in this study. After being informed of the purpose and procedures of the study, all subjects and their parents signed a consent form. The study protocol was approved by the ethics Committee of Harokopio University and the institutional review board of the Beth Israel Deaconess Medical Center. Blood samples were obtained after an overnight fast from all but 2 subjects, leaving 118 subjects (56 males and 62 females) for the subsequent analysis.

Subjects completed self-administered questionnaires on demographic characteristics, general health status, daily exercise, duration and intensity of smoking, alcohol-drinking habits, and eating behavior. Using 3-d food records, subjects recorded the type and amount of food and beverages consumed for 2 consecutive weekdays and 1 weekend day using standard household measures. A trained interviewer reviewed the records with the respondent to clarify entries, servings, and forgotten foods.

Anthropometry and body composition measurements

For all subjects, weight and height were measured by the same observer to the nearest 0.5 kg and 0.5 cm, respectively. Body mass index (BMI) was calculated as weight (kilograms)/height (meters)². Waist, hip, and arm circumferences were measured to a precision of 0.1 cm, and the waist to hip ratio was calculated (22). Triceps, biceps, subscapular, and suprailiac skinfolds were measured twice by one observer on the right side of the body (23) to a precision of 0.2 mm, and the average of the two measurements was used. Bioelectrical impedance analysis was also used for assessment of body composition, as described previously by Lukaski et al. (24). Fat-free mass (FFM) and percent fat mass (%FM) were calculated using formulas developed by Deurenberg et al. (25). Subjects in this study were classified as normal weight, overweight, or obese according to BMI cut-off points of 25 or 30 kg/m² (26). Thus, 20.4% of the subjects were overweight (15 males and 9 females), and 4.2% (4 males and 1 female) were obese.

Leptin measurements

Plasma leptin levels (nanograms per ml) were evaluated in one run using a commercially available RIA (Linco Research, Inc., St. Louis, MO), with a sensitivity of 0.5 ng/ml.

DNA extraction and genotyping

Buffy coats of nucleated cells were obtained from blood samples, and genomic DNA was extracted from leukocyte nuclei by the salting-out method (27). The DNA was dialyzed against TE buffer (10 mmol/liter Tris-HCl and 1 mmol/L Na₂EDTA, pH 7.4).

Genotyping of the three common DNA sequence variants in exons 4, 6, and 14 of the human LEPR gene was carried out by restriction enzyme analysis of PCR-amplified DNA, using the specific pairs of oligonucleotide primers previously described by Gotoda et al. (18). Reverse primers for the codon 109 and 656 variants have a sequence mismatch in their 3′-region to create an artificial restriction site. After amplification, the PCR products were digested with 5 U of the restriction endonucleases HaeIII and MspI (New England Biolabs, Inc., Beverly, MA) for 12 h at 37 C for the K109R and Q223R polymorphisms, respectively, and with 5 U BstUI (New England Biolabs, Inc.) for 6 h at 60 C for the K656N polymorphism. The samples were then electrophoresed on 4% agarose gels, prestained with ethidium bromide (0.5 μg/ml) to visualize the different genotypes (18).

Statistical analysis

The descriptive characteristics of the group variables were expressed as the mean and sd. Allele frequencies were estimated by the gene-counting method, and the genotype distribution of the polymorphisms was tested for Hardy-Weinberg equilibrium byχ ² analysis. Analysis of covariance was performed for comparisons across genotypes for the K109R, Q223R, and K656N polymorphisms, with age and gender as covariates for all variables, except for FFM, which was adjusted for gender and height, and %FM, which was adjusted for gender only, because the regression formulas that have been used for the prediction of FFM and %FM were age-adjusted. Simple and multiple regression analyses were performed to test for associations between independent variables (including gene polymorphisms) and body composition variables (used as dependent variables). In these analyses, leptin values were log-transformed to normalize its distribution. The level of significance was defined at P < 0.015 to adjust for the multiple comparisons made in this study (Bonferroni correction). This study had more than 80% power to demonstrate statistically significant associations at the conventional α = 0.05 level if the underlying associations were strong or moderate, i.e. they had an r ≥ 0.20.

Results

Descriptive statistics for the different phenotypic variables in this study for males and females are shown in Table 1. The mean age of all study participants was 17.7 ± 1.7 yr (range, 14–26 yr). Females, compared with males, had significantly lower BMI (t = -3.466; P = 0.001), lower FFM values (t =− 21.286; P < 0.001), higher sum of four skinfolds (t = 3.101; P = 0.002), higher %FM (t = 9.333, P < 0.001), and higher plasma leptin levels (t = 7.723; P < 0.001).

TABLE 1.

Descriptive characteristics of the study participants (n = 118)

Variables	Females (n = 62)			Males (n = 56)
	Mean	sd	Range	Mean	sd	Range
Age (yr)	17.5	1.6	14–26	17.8	1.8	15–25
BMI (kg/m2)	21.3	3.1	16.2–30.1	23.4a	3.8	15.4–35.9
Sum of skinfolds (mm)	48.7	17.4	23.0–95.0	38.6b	18.4	17.0–106.0
FFM (kg)	39.5	3.8	28.4–47.9	59.6a	6.1	45.4–72.5
FM (kg)	15.5	4.8	7.9–29.1	13.3	7.3	3.0–35.0
FM (%)	27.6	5.0	17.6–38.4	17.3a	6.9	6.0–32.6
Leptin (ng/ml)	9.7	6.0	2.1–30.0	3.2a	2.6	1.2–12.8

^a

Statistically different from females, P < 0.001.

^b

Statistically different from females, P < 0.05.

The genotype distribution and allele frequencies for the K109R, Q223R, and K656N polymorphisms of the LEPR gene are presented in Table 2A. All polymorphisms were in Hardy-Weinberg equilibrium. Genotype distribution and allele frequencies were not different between male and female subjects (data not shown). The allele frequencies in this Greek population are, in general, comparable with those previously reported for other Caucasian populations (14, 15, 17, 18).

TABLE 2A.

Genotype distribution and allele frequencies of the three exonic polymorphisms in the LEPR gene

	Polymorphisms
	K109R	Q223R	K656N
Genotypes
1/1	0.763 (90)	0.440 (52)	0.593 (70)
1/2	0.229 (27)	0.475 (56)	0.339 (40)
2/2	0.008 (1)	0.085 (10)	0.068 (8)
Alleles
1	0.877	0.678	0.763
2	0.123	0.322	0.237

Proportions and number of cases (in parentheses) are given for the genotypes. Only proportions are given for the alleles. 1, Wild-type alleles; 2, variant alleles.

Genotype and allele frequency distributions for the three exonic polymorphisms were also compared using a χ² test between subjects with BMI of 25 kg/m² or less and those with BMI more than 25 kg/m² (Table 2B). The genotype distribution of the Q223R polymorphism was different between normal weight and overweight-obese subjects (χ² = 8.1; df = 2; P = 0.02), as a result of a higher prevalence of R/R homozygotes among overweight-obese subjects, implying a recessive model of inheritance. More specifically, ANOVA models with least significant difference post-hoc tests revealed that %FM and sun of skinfolds were significantly higher in R/R homozygotes (for %FM: R/R, 28.6 ± 1.8%; Q/R, 21.2 ± 1.1%; Q/Q, 23.1 ± 1.0%; P < 0.05; for sum of skinfolds: R/R, 56.8 ± 6.5 mm; Q/R, 40.4 ± 2.1 mm; Q/Q, 44.9 ± 2.8 mm; P < 0.05). In similar ANOVA models, R/R homozygotes had higher leptin levels (R/R, 9.7 ± 2.2 ng/ml; Q/R, 5.6 ± 0.6 ng/ml; Q/Q, 7.2 ± 0.9; P = 0.07; Table 3), with the mean leptin levels in R/R homozygotes being significantly higher (P = 0.04) by post-hoc least significant difference analysis. No significant differences in genotype or allele frequency distributions were detected for the K109R and K656N polymorphisms.

TABLE 2B.

χ^b test results for genotypic and allelic variations at three different exonic polymorphisms in the LEPR gene between normal weight (BMI, ≤25 kg/m²) and overweight-obese (BMI, >25 kg/m²) subjects

	Polymorphisms
	K109R			Q223R			K656N
	BMI ≤25	BMI >25	χb/P	BMI ≤25	BMI >25	χb/P	BMI ≤25	BMI >25	χb/P
Genotypes
Codominant model
1/1	69 (77.5)	21 (72.4)	3.20 /0.21	39 (43.8)	13 (44.8)	8.10 /0.02	54 (60.7)	16 (55.2)	0.30 /0.86
1/2	20 (22.5)	7 (24.1)		46 (51.7)	10 (34.5)		29 (32.6)	11 (37.9)
2/2	(0.0)	1 (3.5)		4 (4.5)	6 (20.7)		6 (6.7)	2 (6.9)
Dominant model
1/1	69 (77.5)	21 (72.4)	0.32 /0.62	39 (43.8)	13 (44.8)	0.01 /1.0	54 (60.7)	16 (55.2)	0.27 /0.67
1/2+ 2/2	20 (22.5)	8 (27.6)		50 (56.2)	16 (55.2)		35 (39.3)	13 (44.8)
Recessive model
1/1+ 1/2	89 (100)	28 (96.5)	3.10 /0.25	85 (95.5)	23 (79.3)	7.40 /0.01	83 (93.3)	27 (93.1)	0.98 /1.0
2/2	(0.0)	1 (3.5)		4 (4.5)	6 (20.7)		6 (6.7)	2 (6.9)
Alleles
1 (%)	88.8	84.5	0.74 /0.37	69.7	62.1	1.16 /0.33	77.0	74.1	0.19 /0.72
2 (%)	11.2	15.5		30.3	37.9		23.0	25.9

The number of cases and proportions (in parentheses) are given for the genotypes. Only proportions are given for the alleles. 1, Wild-type alleles; 2, variant alleles. Significant results (P ≤ 0.05) are in bold.

TABLE 3.

Analysis of covariance of the body composition variables across genotypes for three exonic polymorphisms of the LEPR gene

Variable	K109R				Q223R				K656N
	Geno (n)	Mean	se	P	Geno (n)	Mean	se	P	Geno (n)	Mean	se	P
BMI (kg/m2)	K/K (90)	22.1	0.4	0.42	Q/Q (52)	22.3	0.5	0.04	K/K (70)	22.4	0.4	0.88
	K/R (27)	22.6	0.7		Q/R (56)	21.8	0.5		K/N (40)	22.2	0.6
	R/R (1)	26.2			R/R (10)	24.8	1.1		N/N (8)	21.9	1.2
Sum skinfolds (mm)	K/K	43.2	1.9	0.08	Q/Q	44.5	2.5	0.07	K/K	44.2	2.2	0.95
	K/R	44.3	3.4		Q/R	44.1	2.4		K/N	43.3	2.9
	R/R	84.3			R/R	55.3	5.7		N/N	42.5	6.5
FM (%)	K/K	22.5	0.6	0.82	Q/Q	22.6	0.8	0.06	K/K	22.1	0.7	0.65
	K/R	23.0	1.2		Q/R	22.0	0.8		K/N	22.1	0.9
	R/R	25.8			R/R	26.9	1.9		N/N	21.7	2.1
Leptin (ng/ml)	K/K	6.5	0.5	0.75	Q/Q	7.2	0.9	0.07	K/K	7.0	0.6	0.70
	K/R	7.2	0.9		Q/R	5.6	0.6		K/N	6.2	0.8
	R/R	7.7			R/R	9.7*	2.2		N/N	6.4	1.7

The mean and se of all variables are adjusted for gender and age, except for FFM, which is adjusted for gender and height, and FM, which is adjusted for gender only. Geno, Genotypes; n, number of subjects; Significant results (P ≤ 0.05) are in bold.

^a

Leptin levels in the R/R group are significantly higher (P = 0.04) by least significant difference post hoc tests.

Analyses of covariance of several body composition variables for the K109R, Q223R, and K656N polymorphisms are shown in Table 3. No significant associations were observed between any of the three polymorphisms with BMI, %FM, or plasma leptin levels, except for a weak association between the Q223R polymorphism and BMI (P = 0.04). Similarly, no significant associations were observed when the sum of skinfolds was examined.

In the post-hoc analysis, carriers of the Q223 allele had lower values for BMI (−2.7 U, 22.1 ± 0.3 vs. 24.8 ± 1.1 kg/m², P = 0.01) and %FM (−4.6%, 22.3 ± 0.6% vs. 26.9 ± 1.9%, P = 0.02) than noncarriers. Similarly, they exhibited lower plasma leptin levels, although this difference was not statistically significant (6.5 ± 0.5 vs. 8.5 ± 1.5 ng/ml, P = 0.20).

Regression analysis revealed that the presence of the R223 allele in the homozygous form is a significant predictor of both BMI and %FM (Table 4). Moreover, 5% of the variance in the %FM is explained by the presence of the R223 allele (P = 0.01). Similarly, this allele explains 3% of the variance in BMI (P = 0.05). After adjusting for age and gender, the R223 allele in the homozygous form was still found to be a statistically significant predictor, explaining 4.5% of the variance in %FM and 5% of the variance in BMI. Further adjustment for smoking, energy intake, or fat intake did not alter the results significantly (data not shown). Finally, there were no significant interactions of genetic polymorphisms with diet or exercise in predicting obesity (data not shown).

TABLE 4.

Simple and multiple regression coefficients as well as corresponding r^b values of the study dependent variables regressed on the presence or absence of the R223 allele in the homozygous form with and without adjustment for gender and age

Dependent variables	ra	rab	rb	rbb	% Variability (adjusted rb)
BMI	0.18	0.03a	0.22	0.05b	0.13
% FM	0.23	0.05d	0.21	0.04d	0.45

r, Regression coefficient; r^b, percent variability of the dependent variables explained by R223 allele; r^b, percent variability explained by all independent variables in the model; r_a, bivariate regression coefficients; r_b, adjusted for gender and age.

^a

P = 0.05 vs. r_a.

^b

P = 0.015 vs. r_a.

^c

P = 0.01 vs. r_a.

^d

P = 0.02 vs. r_a.

Discussion

In this study we investigated the effects of three common exonic polymorphisms in the LEPR gene, namely the K109R, Q223R, and K656N polymorphisms, on body composition variables in a Greek population of young subjects. All three polymorphisms are associated with amino acid substitutions in the extracellular region of the LEPR and have potential functional consequences. One polymorphism causes a conservative change [lysine (K) to arginine (R) at codon 109], whereas the other two result in changes in charge [glutamine (Q) to arginine (R) at codon 223 and lysine (K) to asparagine (N) at codon 656] and therefore are the most likely to have functional consequences (28). Other allelic variations in coding and noncoding sequences of the LEPR gene have also been reported, some of which result in silent changes or represent rare mutations (14, 17, 18, 20, 21, 28). These were not studied herein because they were unlikely to have any functional significance.

For K109R and K656N polymorphisms, no significant differences in allele frequency or genotype distribution were observed between normal weight and overweight-obese subjects in this study, consistent with previous reports (14, 17–21). Similar to our findings, no association of these two polymorphisms with body composition variables has been reported previously (19), although a few studies were suggestive of sibling pair linkages between the K109R polymorphism with BMI and FM (15) or weak associations between the K656N substitution and either BMI or FFM in subgroups of lean British subjects (18) and/or overweight Caucasian females (14), respectively.

For the Q223R polymorphism, a higher percentage of homozygotes for the R223 allele was found among heavier subjects in our study. R/R homozygotes had higher BMI (+2.7 U; P = 0.01) and %FM (+4.6%; P = 0.02) values than carriers of the Q223 allele. Moreover, unlike the K109R and the K656N polymorphisms, the Q223R polymorphism was a significant predictor of 5% of the body composition variability. These results are not surprising considering the polygenic nature of most human obesity, according to which each individual gene is expected to contribute in a minor way to the phenotypic variation, and combinations of several genes are likely to contribute or predispose to obesity (1).

The effects of the R223 allele remained unchanged when the other two polymorphisms were taken simultaneously into consideration in the regression models (data not shown). Thus, a compound heterozygotic mechanism involving these three polymorphisms does not appear to predispose to obesity in our population. This does not rule out the possibility, however, that a missense mutation in one allele may cause obesity in combination with an as yet unidentified and/or not studied mutation in the other allele as previously suggested (20).

Evidence of a significant effect of the Q223R polymorphism on human body composition has also been reported in two recent studies (14, 15). In the Québec Family Study, Chagnon et al. observed a significant sibling pair linkage between the Q223R polymorphism and FM (14). Although no association between body composition variables and the Q223R polymorphism was observed, there was a weak, but significant, association between the Q223 allele and FFM in lean males when the analysis was performed by BMI and gender groups (14). In the HERITAGE Family Study cohort, stronger evidence of an association between the Q223R polymorphism and human adiposity was reported among Caucasians, although no reciprocal linkages were detected (15). In particular, middle-aged Caucasian males who were carriers of the R223 allele had significantly higher BMI, %FM, and plasma leptin levels than noncarriers (15). Similarly in our population, R/R homozygotes had significantly higher BMI and %FM values, whereas carriers of the Q223 allele exhibited lower mean adiposity values.

On the other hand, negative results have been reported for the Q223R polymorphism in different Caucasian populations, including American (19), British (18), and Danish (17) groups. However, these investigators did not evaluate this polymorphism using varying genetic models (codominant, heterozygous, and homozygous), as was performed in our study, and no previous study evaluated these polymorphisms as predictors of percent variability of body weight or composition. Negative results have also been reported for other racial populations, including Japanese (20), blacks (15), and Pima Indians (21). Again, this may be due to differing statistical models or analysis, the fact that allelic frequencies vary significantly among different races (15, 20, 21), and/or the possibility that the physiological effect of the Q223R polymorphism has a significant racial component.

A new finding of this study was the quantification of percent variability of body weight and composition predicted by the Q223R polymorphism. Given that the prevalence of R/R homozygotes in this population is approximately 10%, a significant fraction of the population may be at risk for developing obesity with this susceptible genotype. The functional significance of this allele genotype is not entirely known. However, several lines of evidence suggest that this polymorphism may play a role in the pathogenesis of obesity. The Q223R substitution in exon 6 is located in the extracellular region of the LEPR within the first cytokine domain (C domain), which represents a leptin-binding site. It has been previously suggested that this single amino acid change, a glutamine (Q) for an arginine (R) with a change in charge from neutral to positive, could affect the functionality of the receptor and alter its signaling capacity (14, 15, 28), similarly to what has been observed for the fa mutation in the fatty Zucker rat (29). In this rat model of obesity, the fa mutation in the leptin receptor is also located in the first C domain, which is highly conserved among species, and results in a single amino acid substitution (Q269P). This substitution affects the functionality of the receptor, with a significant reduction in cell surface expression of the receptor as well as changes in signal transduction, including constitutive activation of STAT1 and STAT3 and highly impaired ligand-induced STAT5B activation (30). The altered signaling caused by the fa mutation may generate a state of leptin resistance, which, in turn, leads to the observed obese phenotype (31). The glutamine to arginine substitution at position 223 of the human LEPR clearly does not cause a null mutation, because lean subjects homozygous for this polymorphism exist. However, the proximity and similarity of the Q223R polymorphism to the Q269P mutation, which causes obesity in the Zucker mouse model, raise the distinct possibility that the Q223R polymorphism may lead to subtle changes in signaling pathways that also predispose to a leptin-resistant state. Although the higher leptin levels in R/R homozygotes provide supportive evidence for this hypothesis, future in vitro experiments involving expression of wild-type and mutant leptin receptors in CHO cells are needed to evaluate the effect of the Q223R substitution on the functionality of the long isoform of the human LEPR.

Potential limitations of this study include its moderate size. However, the inclusion of only young subjects, which limits the effect of nongenetic determinants of obesity, as well as the homogeneous population base of this study provide a distinct advantage when performing genetic studies and evaluating the corresponding associations. Large ongoing cohort studies, such as the Nurses Health Study, or randomized clinical trials, such as the Look Ahead Study, provide appropriate vehicles for testing these associations in large population samples. We did not explore associations between body composition variables and polymorphisms separately in subgroups divided by gender and BMI because of the expected small size of the subgroups. However, we adjusted for gender and BMI (considered as both a dichotomous and a continuous variable) in the statistical analysis. Finally, in addition to performing measurements using state of the art techniques (in the laboratory analysis and DNA extraction), we performed detailed statistical analysis using several different genotype models as well as relevant linear regression analysis to assess predictors of percent variability of obesity.

In conclusion, our results support the hypothesis that the Q223R polymorphism of the LEPR gene is associated with obesity and predicts a small percentage of body weight and composition variability in a Mediterranean population. In contrast, the K109R and K656N polymorphisms do not appear to play a role of comparable significance. As our knowledge of obesity genes advances with new discoveries, further study of well defined obesity phenotypes and associated gene mutations is necessary to fully elucidate the specific genetic factors predisposing to obesity in humans.

Acknowledgements

This work was supported by in part a grant (KA0008) from Harokopio University and by the Hershey Family Award from Harvard Medical School and a Clinical Research grant from the American Diabetes Association (to C.S.M.).

Abbreviations:

 
• BMI,

  Body mass index;

 
• FFM,

  fat-free mass;

 
• %FM,

  percent fat mass;

 
• LEPR,

  leptin receptor gene;

 
• STAT,

  signal transducer and activation of transcription.