AMERICAN JOURNAL OF HUMAN BIOLOGY 21:84–90 (2009)
Original Research Article
Relationship Between Obesity, Adipocytokines, and Blood Pressure: Possible Common Genetic and Environmental Factors IA PANTSULAIA,1,2 SVETLANA TROFIMOVA,1 EUGENE KOBYLIANSKY,1 AND GREGORY LIVSHITS1,3* 1 Human Population Biology Research Unit, Department of Anatomy and Anthropology, Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel 2 Department of Biomedicine, Institute of Medical Biotechnology, Ministry of Education and Sciences of Georgia, Tbilisi, Georgia 3 YORAN Institute for Human Genome Research, Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel
ABSTRACT Adipokines may link adipose tissue to the inflammatory, metabolic, and immune dysregulation. The variation of adipokine levels within individuals, intercorrelations, and relationships to well-established measures of adiposity are incompletely defined. The main goal of the present study was quantitative evaluation of the genetic interrelationships between obesity and adipokines in normal human population. The study sample comprised 272 families of various sizes, including 530 men and 531 women aged 18–80 years, randomly recruited in rural population living in Russia. Various fatness and fat distribution measures (OB), blood pressure (BP), and plasma levels of several adipokines (AC), such as adiponectin, leptin, resistin, and IGFBP-1, have been measured. The likelihood ratio tests clearly revealed that genetic effect for all studied phenotypes was highly significant (P < 0.001) and accounted for 45.9% 6 8.1%, 33.7% 6 7.9%, 35.7% 6 9.8% of variation for AC, OB, and BP, respectively. The pairwise bivariate analyses showed that strong phenotypic correlation between the obesity (OB) and adipocytokines (AC) was caused by both common genetic and environmental factors (rG 5 0.597 6 0.116, rE 5 0.671 6 0.051). The phenotypic correlation between BP and OB is explained by shared genetic factors only (rG 5 0.532 6 0.109), whereas the phenotypic correlation between BP and AC has only common environment basis (rE 5 20.212 6 0.081) and was mostly due to the correlation observed in females. Our results suggest that genetic factors play a significant role in regulation of variation of the examined traits. The variation of OB traits is almost fully due to genes influencing variation of AC, whereas the correlation between BP and AC is only marginally significant and caused only by shared environment. Am. J. Hum. Biol. 21:84–90, 2009. ' 2008 Wiley-Liss, Inc. Obesity (OB), a new pandemic, is associated with an increased risk of death, morbidity, and accelerated aging (Smith and Haslam, 2007; Walley et al., 2006). It often speeds the onset, increases the prevalence, and intensifies the severity of all of the major diseases of old age (Bruunsgaard and Pedersen, 2003; Ogden et al., 2003). Moreover, OB is strongly associated with the risk of death in both genders and in all racial/ethnic groups and at all ages (Adams et al., 2006; Janssen and Mark, 2007). For example, Adams et al. (2006) found that the risk among obese subjects (BMI > 30 kg/m2) was two to three times higher than that among participants who had a BMI of 23.5–24.9. OB has important functional implications in the older population. Many studies have highlighted that the aging is associated with considerable changes in body composition. After 20–30 years of age, fat-free mass (FFM) progressively decreases, whereas fat mass increases till 60– 70 years of age, and declines thereafter (Kyle et al., 2001). Aging is also associated with a redistribution of both body fat and FFM, in particular, increasing the intraabdominal fat and decreasing the peripheral because of the loss of skeletal muscle (Beaufrere and Morio, 2000). Fat tissue is increasingly viewed as an active endocrine organ with a high metabolic activity. Adipocytes produce and secrete several proteins, such as leptin, adiponectin, and resistin, that act as veritable hormones, responsible for the regulation of energy intake and expenditure (Ahima, 2006). Therefore, hormonal changes that occur during aging may be enhancing the accumulation of fat, the reduction of FFM, and energy balance. It has been found that aging is associated with a decrease in adiponectin secretion, increased leptin levels and reduced resistance to leptin (Gomez et al., 2003; Milewicz et al., 2005). C 2008 V
Wiley-Liss, Inc.
Leptin is one of the major adipocytokines (AC), which regulate body weight, energy metabolism, and the occurrence of obesity and related disorders in humans (Sader et al., 2003). The existing data indicate that leptin correlates strongly with the OB and may represent a link between excess adiposity and increased cardiovascular sympathetic activity (Aneja et al., 2004). Similarly, plasma levels of resistin were found to be significantly higher in obese individuals (Azuma et al., 2003). In contrast, adiponectin levels are inversely related to OB, and are more closely related to visceral than subcutaneous fat (Kazumi et al., 2004). These data clearly suggest that circulating levels of AC may be important risk factors for overweight and OB. However, the function(s) and directionality of the relationship between their expressions in different age groups are not completely clear. Furthermore, we know very little regarding the common genetic factors influencing these two facets, despite the fact that the variation of each of them significantly influenced by genes (Bell et al., 2005; Pantsulaia et al., 2007; Smith and Haslam, 2007).
Contract grant sponsor: Israel National Science Foundation; Contract grant number: 1042/04. *Correspondence to: Gregory Livshits, Human Population Biology Research Unit, Department of Anatomy and Anthropology, Sackler Faculty of Medicine, Tel Aviv University, Ramat Aviv 69978, Tel Aviv, Israel. E-mail:
[email protected] Received 15 April 2008; Revision received 16 July 2008; Accepted 28 July 2008 DOI 10.1002/ajhb.20821 Published online 11 September 2008 in Wiley InterScience (www. interscience.wiley.com).
RELATIONSHIP BETWEEN OB, AC, AND BP
The main goals of the present study were as follows: (1) to assess quantitatively the genetic interrelationships between OB and AC in normal human population, and (2) because the correlation between OB and blood pressure (BP) is well known phenomenon (Aneja et al., 2004; Francischetti and Genelhu, 2007; Livshits and Gerber, 2001), to determine whether the putative genes affecting OB and AC also affect BP. MATERIALS AND METHODS Sample The sample included 1,061 apparently healthy individuals belonging to 272 households. Every participated subject signed an informed consent document, which was approved by The Tel Aviv University ethics committee. The individuals are representative of Caucasian population sampled in several villages along the Volga River (Russia). We gave the detailed description of the sample recently elsewhere (Kalichman et al., 2006). The collected information entailed gender, age, basic socioeconomic parameters, standard anthropometric measurements, as well as data on chronic morbidity and medical treatment. The individuals were free from acute and/or chronic metabolic illnesses and were not taking any lipid-lowering, antihypertensive, or hypoglycemic medications. Blood sampling and biochemical assays The plasma samples were collected after an overnight fast into the tubes with EDTA and the cells were removed by centrifugation and stored at 2708C until analysis. The AC (adiponectin, resistin, and leptin) levels were measured by a sandwich enzyme immunoassay (ELISA) technique using the set of specific antibodies and standards from R&D Systems (Minneapolis, MN). The inter- and intra-assay coefficients of variation were less than 5.0 and 8.7% for all studied variables. Previous studies also showed significant association of leptin and OB variation with GH/IGF-1 hormonal system (Liew et al., 2005). Consequently, we include IGFBP-1 in analysis as a useful component potentially affecting BP. The mean intra- and inter-assay coefficients of variation for IGFBP-1 were 1.5 and 4.8% (Pantsulaia et al., 2005). Blood pressure and obesity characteristics Blood pressure was measured by a standard mercury sphygmomanometer (Livshits and Gerber, 2001). OB measures taken from each individual included: (1) Body mass index, BMI (kg/m2); (2) Waist to hip circumferences ratio, WHR; (3) Eight skinfolds (chest, abdomen, subscapular, hip, upper, medial and dorsal, lower arm, and calf); (4) Nine circumferences (mesosternal chest minimal, waist, hip, upper arm, lower arm, wrist, thigh, calf, ankle). All the data were obtained with standard anthropometric techniques and reported in details by us earlier (Livshits et al., 1998). Statistical and genetic analysis of the data Preliminary statistical analyses. Preliminary statistical analyses were performed using the STATISTICA 6.0 for PC (Statsoft, Minneapolis, USA). Since the distributions of the studied molecules were markedly skewed, the data on biochemical markers were log-transformed. Differences
85
in means of the plasma AC between male and female groups were determined by the Mann-Whitney U-test. In the female sample, all correlations were first examined separately for pre and postmenopausal women, and then for the total group. To identify the main age-related trends, the studied subjects were arbitrarily divided into a number of age groups (70 years). Multivariate analyses. Since the traits within each group of BP, OB, and AC are intercorrelated, to avoid redundancy and repetition in data analysis, principal component (PCA) and canonical correlation analyses of the data were undertaken. PCA is a statistical technique that exploits multicollinearity in a larger number of variables to arrive at a smaller number of variables (principal components, PC) that are independent linear combinations of the original set. According to Jolliffe (2002), this is achieved by linear transformation of the original measurements to a new set of variables—PCs, which are uncorrelated, and which are ordered so that the first few PCs explain major part of the variation present in all original variables. In this study, PCA was conducted on each of the trait categories to reveal the complex relationships between OB, AC, and BP. The factors were constructed with varimax rotation of principal components to maximize the sum of squares of the loading for each factor. Then factor scores were computed and expressed quantitatively for each individual according to the design of their construction. Since the bilateral measurements of skinfolds were especially strongly correlated, corresponding values for each individual were averaged. Thus, eight initial skinfolds and nine circumference measurements were subjected to PCA. The eigen value 1 criterion was utilized to calculate the principal components scores for each individual in their pedigreed data set. As a result, new composite variables, PC-SKF and PCCRF, were used in our subsequent analyses. The detail description of the method is given in Jolliffe (2002). Canonical correlation analysis was used to examine the relations between the obesity measures and each of the other two sets of variables, AC and BP. Canonical correlation is a multivariate procedure for assessing the linear relationship between two multidimensional (two sets) variables, called canonical variables, one representing a set of independent variables, the other a set of dependent variables (Thompson, 1984). This procedure allows the simultaneous examination of the effect of several predictor variables (e.g. OB) on the criterion (outcome) variables (e.g. BP). The logic here is that variables that are highly correlated with a canonical variate have more in common with it and they should be considered more important when deriving a meaningful interpretation of the related canonical variate. The criterion for choosing the important variables in each canonical variate is the structure coefficients (loadings). As a rule of thumb for meaningful loadings, an absolute value equal to or greater than 0.3 is often used (Thompson, 1984). With canonical analysis, it is also possible to examine the correlation of each of the variables within the group of predictor variables (e.g. OB traits) to the outcome variables (e.g. BP). In this study we examined in pair-wise fashion, canonical correlations between all three sets of traits. Since, AC is produced by adipose tissue, they were considered as inAmerican Journal of Human Biology
86
I. PANTSULAIA ET AL. TABLE 1. Descriptive statistics of studied OB, BP, and AC variables according to generation and sex Mean 6 SD Father
Mother
Son
Daughter
Age (years) 65.21 6 9.13 (43.00–92.00) 63.68 6 9.73 (41.00–90.00) 36.86 6 10.92 (18.00–79.00) 35.67 6 10.92 (17.00–76.00) SBP (mm/Hg) 135.82 6 21.74 (80.00–230.00) 139.77 6 23.40 (90.00–240.00) 120.82 6 14.28 (90.00–200.00) 114.69 6 15.13 (80.00–190.00) DBP (mm/Hg) 83.04 6 10.29 (60.00–120.00) 83.22 6 11.42 (60.00–120.00) 77.84 6 8.43 (60.00–115.00) 74.28 6 8.62 (50.00–110.00) PC-SKF 20.58 6 0.58 (21.58 to 1.29) 0.74 6 0.99 (21.29 to 4.38) 20.63 6 0.52 (21.53 to 1.51) 0.48 6 0.92 (21.13 to 4.21) PC-CRF 20.10 6 1.00 (23.04 to 2.80) 0.16 6 1.12 (22.56 to 4.34) 0.14 6 0.81 (21.68 to 2.55) 20.21 6 1.02 (22.31 to 2.58) 2 BMI (kg/m ) 23.71 6 4.01 (15.33–39.75) 27.08 6 4.92 (15.62–44.54) 23.11 6 3.07 (16.38–34.03) 23.93 6 4.60 (15.78–43.47) WHR 0.94 6 0.07 (0.76–1.17) 0.88 6 0.08 (0.72–1.12) 0.87 6 0.06 (0.75–1.12) 0.79 6 0.06 (0.67–1.06) ADIP (lg/ml) 5.63 6 2.76 (0.94–13.95) 6.75 6 4.12 (0.08–27.05) 4.43 6 2.35 (0.58–14.93) 5.55 6 2.80 (0.08–16.63) IGFBP-1 (ng/ml) 18.26 6 27.84 (0.56–326.12) 11.19 6 16.40 (0.37–170.84) 14.40 6 23.20 (0.48–239.24) 8.48 6 8.85 (0.37–64.35) LEPT (ng/ml) 2.32 6 2.62 (0.082–22.28) 9.08 6 7.99 (0.56–55.56) 1.55 6 1.61 (0.053–13.29) 8.27 6 5.90 (0.99–35.87) REST (ng/ml) 3.27 6 2.51 (1.00–15.30) 3.27 6 2.20 (1.00–12.43) 3.00 6 2.38 (1.00–17.68) 3.96 6 2.85 (1.00–15.97)
U-test
2 2 1, 2 1, 2 1 1, 2 1, 2 1, 2 1, 2 2
U, Mann-Whitney test; significant differences between mothers and fathers indicated by 1 as well as between daughters and sons by 2. SBP, systolic blood pressure; DBP, diastolic blood pressure; PC, principal component; SKF, skinfolds; CRF, circumferences; BMI, body mass index; WHR, waist/hip ratio; ADIP, adiponectin; IGFBP, insulin-like growth factor binding protein; LEPT, leptin; REST, resistin.
dependent predictors of OB, while OB variables in turn as a predictor of BP. The statistical null hypothesis that the canonical correlations are zero in the population is tested. Only the correlations with the first significant canonical variable were examined. Statistical genetic analysis. To estimate the extent of familial and possible genetic influences on each variable, two mutually complementary analyses was undertaken: (a) parents-offspring correlations were computed using the statistical package MAN-5 (Malkin and Ginsburg, 2003); and (b) the heritability (h2) of each trait was calculated using standard variance components analysis, using statistical package ‘‘FISHER’’. The analysis is based on a quantitative genetic theory and allows one to distinguish the different independent components forming the variation of the studied trait (Falconer and Mackay, 1996). Accordingly, the total variation of the phenotype (VPh) can be divided into genetic (VG) and environmental (VE) components. The later includes components attributable to common environmental effects shared by spouses (SP), household (HS), and siblings (SB), respectively. The remaining unexplained residual variation was defined as VRS. Heritability is a proportion of phenotypic variation attributable to putative genetic effects, i.e. h2 5 VG/VPh. This program finds the best fitting and most parsimonious linear genetic model for the trait variability and produces maximum-likelihood estimates of genetic parameters on the basis of pedigree data. To distinguish between the genetic and environmental sources of covariation at the next stage of the analysis, a bivariate variance component model was fitted to PC variables that showed significant phenotypic cross-correlation. The bivariate mixed model, as utilized in the FISHER package calculates the variance components for each of the two traits and also evaluates an additive genetic correlation (rG), as well as an environmental correlation (rE), between them. While genetic correlation is a quantitative measure of the shared effects of genes on each of the two traits under analysis, environmental correlation estimates the extent to which these traits share a common environment (Falconer and Mackay, 1996). As a regular correlation, both rG and rE may theoretically range between 21.0 and 1.0. The statistical significance of (rG) and (rE) was studied using nested models and by examining the change in chi-square values between the models. American Journal of Human Biology
RESULTS Descriptive statistics and sex differences Table 1 provides the basic descriptive statistics of each of the studied OB, AC, and BP traits according to generation and sex of the participants. The data for AC are presented before log-transformation in the original units. The mean levels of OB and ACs displayed a marked sex dependency for the majority of variables. Women tended to have higher OB measures, including leptin and adiponectin. IGFBP-1 concentrations, however, were significantly higher in men. Sex differences in BP and REST values were observed only between sons and sisters. Age-dependent trends of studied variables To ascertain the possible age-dependence of each studied trait, several piecewise linear models were examined. To choose the best-fitting curve, maximum likelihood of the parameter estimates for different statistical models were obtained using the software CURVEF (Malkin et al., 2002). Age effects were statistically significant for all variables, except REST in both genders, though the magnitude of the effects varied considerably. For example, obesity characteristics, such as WHR and BMI, increased with age and were significantly higher in older individuals (>60 years) than younger (Fig. 1a,c, P < 0.001). This is in particular true with respect to WHR, where consistent trend observed till the very old age. However, BMI reached its maximum at the vicinity of sixth decade of life and then showed the signs of diminution in both sexes. In females, skinfold (PC-SKF) and circumference (PC-CRF) measurements showed age-dependent pattern very similar to BMI. In males, however, although the age-dependent changes were also observed, both traits showed no very clear pattern (Fig. 1b, P > 0.05). Our data also demonstrated that the circulating levels of ADIP changed significantly with the age in both sexes. Age explained 7.4 to 9.0% of ADIP variation, but contributed between 19.0 and 37.0% to SBP variation (P < 0.001). Finally, LEPT and IGFBP-1 levels variation were significantly associated with age only in males, although the magnitude of correlation was low (0.10–0.22, P < 0.05). Pre and postmenopausal women showed virtually the same pattern of age relationship for all variables. After adjustment for age, the differences between pre and
RELATIONSHIP BETWEEN OB, AC, AND BP
87
nificant, in particular between the SBP and DBP, between various measures of OB, as well as between them and the AC. Relations were especially strong between LEPT and OB, ranging from 0.56 (WHR in men, P < 0.001) to 0.70 (skinfolds in women, P < 0.001). The correlations were also significant between ADIP and LEPT and BP in female’s cohort (P < 0.01), but negligible in males. At the next step, for appraising the association between the set of variables (AC, BP, and OB), the canonical correlation analyses were performed. In the canonical correlation analysis between AC and OB on one hand and BP on the other hand, we considered BP as a dependent variable and OB and AC as independent predictor variables. We were also interested in the extent of the dependence of the OB variation on AC. As seen in Table 2, the canonical correlations between the OB and AC sets were very substantial (0.71–0.80, P < 0.001), regardless of sex. Hence, canonical correlation suggests that some 50% or more of the OB variation is attributable to combined AC effects. It is of interest to mention that the canonical weights, representing respective variable’s unique contribution to the observed canonical correlation were of similar magnitude for various OB/AC combinations. Additionally, Table 2 indicates a significant relationship between the BP and OB, although magnitude of the effect was much lower, 0.26–0.29. Finally, canonical correlation between the AC and BP sets was significant only in female’s cohort (0.221, P < 0.001, Table 2). Because the interpretation of separately taken canonical variable is complicated, we also examined the relationships between the studied traits using PC. PC of each of the three sets of the traits (AC, OB, and BP, Table 2) according to sex showed very similar results in each of the sets. In each analysis only first principal component (PC) with eigenvalue >1.0 was retained. It explained some 86, 82, and 49% of the total variation of BP (SBP, DBP), OB (WHR, BMI, CRF, SKF), and AC (ADIP, IGFBP-1, LEPT) variables, respectively. Figure 2 shows that the pairwise correlations between the PCs closely followed the pattern observed in the canonical analysis. For example, associations between PC-OB and PC-AC (Fig. 2a) as well as between PC-BP and PC-OB (Fig. 2b) were statistically highly significant. Nevertheless, the correlations between PC-AC and PC-BP were not significant in both genders (r 5 0.05, P > 0.05). As seen from Figure 3a,c, PC-OB and PC-BP showed comprehensible age-dependent change across the whole range of ages, but PC-AC-levels remained about the same till the sixth decade of life and then rapidly and significantly decreased in both sexes (Fig 3b). Consequently, PC variables were selected in the statistical-genetic analysis. Familial correlations and variance decomposition analysis The sex and age dependent trends of obesity characteristics (a) WHR, (b) PC-CRF/PC-SKF, (c) BMI. Fig. 1.
postmenopausal women were statistically nonsignificant for any variable. The interrelationship between the studied traits Analysis of the correlations between the studied traits revealed that they all (except REST) were statistically sig-
Familial effects for all examined single traits and PCs were statistically significant (P < 0.01). The corresponding regression coefficients ranged between 0.221 for LEPT and 0.519 for ADIP, evidently suggesting contribution of the genetic factors to variation of each studied trait. At the next step of the study, the model-fitting analysis was applied to ascertain the extent of the genetic and environmental influences on variation of PC variables. It also simultaneously takes into account the effect of the potential covariates. The likelihood ratio test revealed that additive genetic effect for PC-AC was highly signifiAmerican Journal of Human Biology
88
I. PANTSULAIA ET AL. TABLE 2. The canonical correlations and regression coefficients between OB, AC, and BP sets Men
Women
SBP Canonical R PC-SKF PC-CRF BMI WHR Canonical R ADIP IGFBP-1 LEPT Canonical R PC-SKF PC-CRF BMI WHR
DBP 0.28
SBP
DBP
*
0.25 0.25 0.27 0.21
0.27 0.22 0.20 0.22 0.17
ADIP 20.24 20.21 20.23 20.27
0.71* IGFBP-1 20.32 20.33 20.35 20.28
SBP
DBP
*
0.19 0.22 0.21 0.19
0.26 0.23 0.25 0.25 0.21
20.05 20.04 0.05 LEPT 0.68 0.54 0.59 0.53
20.15 20.08 0.15 ADIP 20.19 20.12 20.15 20.17
0.80* IGFBP-1 20.39 20.34 20.39 20.34
*
0.21 0.24 0.24 0.20
0.22 0.23 0.24 0.19
0.22*
0.07 20.04 20.06 0.03
Whole
0.14* 20.13 20.13 0.18 LEPT 0.78 0.70 0.74 0.46
20.09 20.07 0.08 ADIP 20.21 20.16 20.18 20.22
0.73* IGFBP-1 20.35 20.33 20.37 20.31
20.09 20.08 0.11 LEPT 0.70 0.61 0.65 0.49
SBP, systolic blood pressure; DBP, diastolic blood pressure; PC, principal component; SKF, skinfolds; CRF, circumferences; BMI, body mass index; WHR, waist/hip ratio; ADIP, adiponectin; IGFBP, insulin-like growth factor binding protein; LEPT, leptin; REST, resistin. * P < 0.01.
cant (P < 0.001), and accounted for 45.9% 6 8.1% of the total phenotypic variance, adjusted for age (Table 3). The remaining was due to random environmental effects and common household environment (36.3% 6 7.3% and 17.8% 6 6.1%, respectively). Genetic effects for two other PC-variables were also statistically significant and ranged from 35.6% 6 7.3% (for PC-BP) to 33.7% 6 7.9% (for PCOB). Significant contribution of common environment factors was also encountered in variation of these two variables (Table 3). The outcomes of the pairwise bivariate analysis suggested that the phenotypic correlation between PC-BP and PC-OB is explained by shared genetic factors only (rG 5 0.532 6 0.109, rE 5 0.00), whereas between PC-BP and PC-AC has only common environment basis (rE 5 20.212 6 0.081) and was mostly due to correlation observed in females. Finally, the analysis of PC-OB and PC-AC revealed that the correlation between these two variables caused by both common genetic and shared environmental factors (rG 5 0.597 6 0.116, rE 5 0.671 6 0.051). DISCUSSION The incidence of obesity and its associated disorders is growing markedly worldwide. It increases the likelihood of death from all causes by 20%, and more specifically death from coronary artery disease and stroke are increased by 25 and 10%, respectively (Adams et al., 2006; Janssen and Mark, 2007). It is of note that most of the conditions associated with obesity are also associated with aging. The identification of causative links between the natural aging process and obesity could therefore be of considerable importance. Many of the interactions between the obesity and related diseases seem to be orchestrated by a complex network of soluble mediators derived from adipocytes. Recent studies convincingly demonstrated that adipose tissue functions as a highly specialized, endocrine and paracrine organ producing an array of adipokines (Ronti et al., 2006). In the present study, we attempted to assess the age-related changes of OB, AC, and BP and to evaluate the extent of the putative common genetic and environmental factors that may potentially cause the correlation between them. American Journal of Human Biology
Fig. 2. The relationships between the obesity (PC-OB), adipocytokines (PC-AC), and blood pressure (PC-BP) traits.
The measurements of the circulating AC observed in this study fall well within the range of variation for presumably healthy subjects reported by other investigators (Papadopoulos et al., 2005; Wolfe et al., 2004). Our data
89
RELATIONSHIP BETWEEN OB, AC, AND BP
TABLE 3. Variance decomposition analysis of principal components
for unadjusted BP, OB, and AC variables Variance components VAD VPR VHS VSB VRS
PC-BP
PC-OB
PC-AC
35.6% 6 7.3% 31.4% 6 6.0% 0 0 33.0% 6 9.4%
33.7% 6 7.9% 0 16.0% 6 6.0% 0 50.1% 6 7.7%
45.9% 6 8.1% [0] 17.8% 6 6.1% 0 36.3% 6 7.3%
[0], parameter was fixed at the indicated value; VAD, additive genetic; VPR, common parent environment; VHS, common household; VSB, common sibling; VRS, residual environment.
Fig. 3. The sex- and age-related changes of PCs extracted from the corresponding analyses of obesity (a: PC-OB), adipocytokines (b: PCAC), and blood pressure (c: PC-BP) variables.
unequivocally suggested strong and significant correlation between age and studied variables (Fig. 3a,c). Previous studies that have analyzed the effects of age on AC have obtained controversial results; some failed to find a correlation with age or described positive/negative correlations (Adamczack et al., 2005; Gannage-Yared et al., 2006).
However, these studies have included only relatively small samples (153 healthy nondiabetic men). The higher number of subjects included in our study (>1,000 individuals), with a broad range of age and body anthropometry, might have facilitated the finding of these differences. We also revealed strong and significant correlation between AC (except resistin) and OB variables (r 5 0.578, P < 0.001). The canonical correlation between AC (excluding resistin) and OB varied between 0.71 in males and 0.80 in females (P < 0.001, in both instances). These correlations were not unexpected, since significant correlations between variety of OB measures and AC were reported in a number of publications (Kaprio et al., 2001; Menzaghi et al., 2006). Abundant evidence from twin and family studies has demonstrated genetic influence for familial clustering of adiposity traits including AC (Kaprio et al., 2001; Menzaghi et al., 2006). They also reported significant genetic and environmental correlations between OB and AC, consistently suggesting substantial pleiotropic genetic effects on circulating leptin and BMI. Some of these studies estimated that BMI genetic variation is virtually indistinguishable from the genetic effects on leptin (Kaprio et al., 2001), while others suggest the existence of both common and variable-specific genetic effects (Livshits et al., 2003). The results of our study confirm that the circulating levels of AC do not randomly vary in a healthy population, but rather strongly aggregate in families, where the heritability estimates account for 45.9% 6 8.1% of the total variation (Table 3). The results of our bivariate analyses were also in agreement with previously published results. For example, both genetic and environmental correlation estimates were highly significant (rG 5 0.597 6 0.116, rE 5 0.671 6 0.051) between AC and OB variables in our study. These estimates suggest that the major portion (>80%) of the univariate heritability estimates, for both OB and AC, are explained in pleiotropic genetic effects. The shared environmental factors contributed about 45% to nongenetic variation of each OB and AC. At the same time, a very substantial proportion (75%) of the univariate genetic variance of BP in this sample was attributable to common genetic influences with OB. However, we observed only modest correlation between BP and AC, which was fully explained in environmental correlation observed in female cohort. In general, while contribution of genetic factors to correlation between BP and OB had been discussed (Livshits and Gerber, 2001; Vinck et al., 1999), very few studies examined genetic correlations between OB and AC (e.g. Kaprio et al., 2001), and we are not aware of any study that considered genetic correlation between BP and AC. American Journal of Human Biology
90
I. PANTSULAIA ET AL.
There are number of major strengths of this study: (1) we examined ethnically and socioeconomically homogeneous population, with phenotype measurements nonmodified by the medication or vitamin supplement consumption; (2) the examined individuals were free from chronic metabolic disease and specifically none of them was diabetic, and (3) each individual was assessed simultaneously for all the examined variables. As any other study, this article has a number of limitations and deficiencies. This includes the absence of the longitudinal data to test the possible bias in the estimates of the age-dependent trends inferred using cross-sectional data, as it is often done in other studies of this kind. Another problem is the potential bias in analysis of the familial data to reveal the association between the traits, assuming the randomly collected sample. Note, however, both packages that we implemented in this study, MAN and FISHER, are specifically designed for the family based data and able to account for nonindependence due to kinship. Nevertheless, not all the analyses are possible with these programs. Specifically, canonical correlation analysis is prone to bias to nonindependence of the individuals. Therefore, significance of the results obtained in this analysis and in some of the preliminary analyses concerning bivariate correlations between the traits of interest, should be considered as ‘‘tendencies.’’ Yet, in final and most important part of the article, where we used variance component analysis (both uni and bivariate) implemented in the FISHER package, the size and structure of the families was taken into account for all the parameter estimates, including age effect, and bivariate correlations. In conclusion, our results suggest that the major (but not entire) genetic component of OB variation is attributable to genes affecting variation of leptin, adiponectin, and IGFBP-1, while the genetic covariation between BP and OB is caused by other genes.
LITERATURE CITED Adamczack M, Rzepkat E, Chudek J, Wiecek A. 2005. Ageing and plasma adiponectin concentration in apparently healthy males and females. Clin Endocrinol 62:114–118. Adams KF, Schatzkin A, Harris TB, Kipnis V, Mouw T, Ballard-Barbash R, Hollenbeck A, Leitzmann MF. 2006. Overweight, obesity, and mortality in a large prospective cohort of persons 50 to 71 years old. N Engl J Med 355:763–778. Ahima RS. 2006. Adipose tissue as an endocrine organ. Obesity (Silver Spring). Suppl 5:242S–249S. Aneja A, El-Atat F, McFarlane SI, Sowers JR. 2004. Hypertension and obesity. Recent Prog Horm Res 59:169–205. Azuma K, Katsukawa F, Oguchi S, Murata M, Yamazaki H, Shimada A, Saruta T. 2003. Correlation between serum resistin level and adiposity in obese individuals. Obes Res 8:997–1001. Beaufrere B, Morio B. 2000. Fat and protein redistribution with aging: Metabolic considerations. Eur J Clin Nutr 54(suppl):S48–S53. Bell CG, Walley AJ, Froguel P. 2005. The genetics of human obesity. Nat Rev Genet 2005; 3:221–234. Bruunsgaard H, Pedersen BK. 2003. Age-related inflammatory cytokines and disease. Immunol Allergy Clin North Am 23:15–39. Falconer D, Mackay T. 1996. Introduction to quantitative genetics, 4th ed. London: Longman. Francischetti EA, Genelhu VA. 2007. Obesity-hypertension: an ongoing pandemic. Int J Clin Pract 61:269–280. Gannage-Yared MH, Khalife S, Semaan M, Fares F, Jambart S, Halaby G. 2006. Serum adiponectin and leptin levels in relation to the metabolic syndrome, androgenic profile and somatotropic axis in healthy non-diabetic elderly men. Eur J Endocrinol 155:167–176. Gomez JM, Maravall FJ, Gomez N, Navarro MA, Casamitjana R, Soler J. 2003. Interactions between serum leptin, the insulin-like growth factor-I
American Journal of Human Biology
system, and sex, age, anthropometric and body composition variables in a healthy population randomly selected. Clin Endocrinol (Oxf) 58:213– 219. Janssen I, Mark AE. 2007. Elevated body mass index and mortality risk in the elderly. Obes Rev 8:41–59. Jolliffe IT. 2002. Principal component analysis, 2nd ed. New York: Springer. Kalichman L, Livshits G, Kobyliansky E. 2006. Indices of body composition and chronic morbidity: a cross-sectional study of a rural population in central Russia. Am J Hum Biol 18:350–358. Kaprio J, Eriksson J, Lehtovirta M, Koskenvuo M, Tuomilehto J. 2001. Heritability of leptin levels and the shared genetic effects on body mass index and leptin in adult Finnish twins. Int J Obes 25:132–137. Kazumi T, Kawaguchi A, Hirano T, Yoshino G. 2004. Serum adiponectin is associated with high-density lipoprotein cholesterol, triglycerides, and low-density lipoprotein particle size in young healthy men. Metabolism 5:589–593. Kyle UG, Genton L, Hans D, Karsegard VL, Michel JP, Slosman DO, Pichard C. 2001. Total body mass, fat mass, fat-free mass, and skeletal muscle in older people: cross-sectional differences in 60-year-old persons. J Am Geriatr Soc 49:1633–1640. Liew CF, Wise S, Yeo K, Lee K. 2005. Insulin-like growth factor binding protein-1 is independently affected by ethnicity, insulin sensitivity, and leptin in healthy, glucose-tolerant young men. J Clin Endocrinol Metab 90:1483–1488. Livshits G, Gerber LM. 2001. Familial factors of blood pressure and adiposity covariation. Hypertension 37:928–935. Livshits G, Pantsulaia I, Trofimov S, Kobyliansky E. 2003. Genetic variation of circulating leptin is involved in genetic variation of hand bone size and geometry. Osteoporos Int 14:476–483. Livshits G, Yakovenko K, Ginsburg E, Kobyliansky E. 1998. Genetics of human body size and shape: pleiotropic and independent genetic determinants of adiposity. Ann Hum Biol 25:221–236. Malkin I, Ginsburg E. 2003. Program Package for Mendelian Analysis of Pedigree Data (MAN, Version 5). Tel Aviv, Israel: Department of Anatomy and Anthropology, Sackler Faculty of Medicine, Tel Aviv University. Technical Report. Malkin I, Karasik D, Livshits G, Kobyliansky E. 2002. Modelling of age related bone loss using cross-sectional data. Ann Hum Biol 29:256– 270. Menzaghi C, Coco A, Salvemini L, Thompson R, De Cosmo S, Doria A, Trischitta V. 2006. Heritability of serum resistin and its genetic correlation with insulin resistance-related features in nondiabetic Caucasians. J Clin Endocrinol Metab 91:2792–2795. Milewicz T, Krzysiek J, Janczak-Saif A, Sztefko K, Krzyczkowska-Sendrakowska M. 2005. Age, insulin, SHBG and sex steroids exert secondary influence on plasma leptin level in women. Endokrynol Pol 56:883– 890. Ogden CL, Carroll MD, Flegal KM. 2003. Epidemiologic trends in overweight and obesity. Endocrinol Metab Clin North Am 32:741–760. Pantsulaia I, Livshits G, Trofimov S, Kobyliansky E. 2007. Genetic and environmental determinants of circulating resistin level in a community-based sample. Eur J Endocrinol 156:129–135. Pantsulaia I, Trofimov S, Kobyliansky E, Livshits G. 2005. Genetic regulation of the variation of circulating insulin-like growth factors and leptin in human pedigrees. Metabolism 54:975–981. Papadopoulos DP, Makris TK, Krespi PG, Poulakou M, Stavroulakis G, Hatzizacharias AN, Perrea D, Votteas VV. 2005. Adiponectin and resistin plasma levels in healthy individuals with prehypertension. J Clin Hypertens (Greenwich) 12:729–733. Ronti T, Lupattelli G, Mannarino E. 2006. The endocrine function of adipose tissue: an update. Clin Endocrinol (Oxf) 64:355–365. Sader S, Nian M, Liu P. 2003. Leptin: A novel link between obesity, diabetes, cardiovascular risk, and ventricular hypertrophy. Circulation 108:644–646. Smith SC, Haslam D. 2007. Abdominal obesity, waist circumference and cardio-metabolic risk: awareness among primary care physicians, the general population and patients at risk—The Shape of the Nations survey. Curr Med Res Opin 23:29–47. Thompson B. 1984. Canonical correlation analysis: uses and interpretation. In: Quantitative applications in the Social Sciences, 47. Thousand Oaks, CA: Sage Publications. p. 8–21. Vinck W, Vlietinck R, Fagard R. 1999. The contribution of genes, environment and of body mass to blood pressure variance in young adult males. J Hum Hypertens 3:191–197. Walley A, Blakemore A, Froguel P. 2006. Genetics of obesity and the prediction of risk for health. Hum Mol Genet 2:R124–R130. Wolfe B, Jimerson D, Orlova C, Mantzoros C. 2004. Effect of dieting on plasma leptin, soluble leptin receptor, adiponectin and resistin levels in healthy volunteers. Clin Endocrinol (Oxf) 61:332–338.
