Associations Between Adiposity and Metabolic Syndrome Over Time: The Healthy Twin Study

Abstract

Background:

We evaluated the association between changes in adiposity traits including anthropometric and fat mass indicators and changes in metabolic syndrome traits including metabolic syndrome clustering and individual components over time. We also assessed the shared genetic and environmental correlations between the two traits.

Methods:

Participants were 284 South Korean twin individuals and 279 nontwin family members had complete data for changes in adiposity traits and metabolic syndrome traits of the Healthy Twin study. Mixed linear model and bivariate variance-component analysis were applied.

Results:

Over a period of 3.1 ± 0.6 years of study, changes in adiposity traits [body mass index (BMI), waist circumference, total fat mass, and fat mass to lean mass ratio] had significant associations with changes in metabolic syndrome clustering [high blood pressure, high serum glucose, high triglycerides (TG), and low high-density lipoprotein cholesterol] after adjusting for intra-familial and sibling correlations, age, sex, baseline metabolic syndrome clustering, and socioeconomic factors and health behaviors at follow-up. Change in BMI associated significantly with changes in individual metabolic syndrome components compared to other adiposity traits. Change in metabolic syndrome component TG was a better predictor of changes in adiposity traits compared to changes in other metabolic components. These associations were explained by significant environmental correlations but not by genetic correlations.

Conclusions:

Changes in anthropometric and fat mass indicators were positively associated with changes in metabolic syndrome clustering and those associations appeared to be regulated by environmental influences.

Introduction

Changes in anthropometric indicators are established risk factors for changes in metabolic syndrome clustering and individual components of metabolic syndrome.^1

–4 Anthropometric indicators such as body mass index (BMI) and waist circumference (WC) can indirectly estimate adiposity and distribution of adiposity, while fat mass indicators can characterize specific biologic body components.⁵ Longitudinal changes in the fat mass indicators in relation to changes in metabolic syndrome clustering and individual metabolic syndrome components may provide insight into the functional impact of body components.⁵

Given the reported inter-individual variations in adiposity and metabolic trait changes,⁶ genetic and environmental factors may influence their associations. A few studies have addressed the impact of heritability and common environmental influences on the changes in adiposity and metabolic syndrome-related traits,^7,8 and the effect of shared genetic and environmental factors on changes in anthropometric indicators and in leptin and insulin levels.⁹ However, the impact of shared genetic factors and environment on changes in adiposity traits and changes in metabolic syndrome-related traits remains to be elucidated.

In this study, in a cohort of participants from the Healthy Twin study, we evaluated the associations between changes in adiposity traits including anthropometric indicators, and fat mass indicators, and changes in metabolic syndrome traits including metabolic syndrome clustering, and individual components over time. We further assessed for possible shared genetic and environmental factor-mediated influences on the correlations between both traits.

Subjects and Methods

Study population

The study subjects were participants in the Healthy Twin study, a multi-center cohort study established in 2005 for South Korean adult (>30 years of age) same-sex twins and their first-degree adult family members.^10,11 The trait measurements were recorded for 1624 individuals at baseline and 794 individuals at follow-up, of the total 3479 study participants. A total of 563 individuals from 161 families (191 men and 372 women; 239 monozygotic twin individuals, 45 dizygotic twin individuals, and 279 nontwin family members) with two sets of measurements (baseline and follow-up, follow-up interval, 3.1 ± 0.6 years) for adiposity and metabolic syndrome-related traits were included in the present work.

Compared to the excluded individuals (n = 2916), those included in our study (n = 563) were more likely to be women (66.1% vs. 57.9% P < 0.001) and older (45.1 ± 12.4 years vs. 43.9 ± 13.8 years, P = 0.037). However, there was no significant difference in BMI between the included and excluded individuals. This study was carried out in accordance with the Declaration of Helsinki. Informed consent along with conflict of interest disclosure was obtained from the study participants. All study procedures were approved by the respective institutional review boards of the participating institutions.

Measures of adiposity and metabolic syndrome-related traits

Adiposity traits including BMI, WC, total fat mass (TFM), body fat percentage (BFP), and fat mass-to-lean mass ratio (FLR) were measured at baseline and follow-up visits. BMI was calculated as measured weight (kg) divided by measured height squared (m²), using a digital balance (Tanita Co., Seoul, Korea), and stadiometer (Samwha Co., Seoul, Korea), respectively. WC was measured at the mid-point between the lower margin of the rib cage and the iliac crest using a stretch-resistant tape. TFM and total lean mass (TLM) were measured using a bioelectrical impedance analyzer (InBody 2.0 and 3.0; Biospace, Seoul, South Korea) using multi-frequency segmental bioelectrical method. BFP was calculated as the percentage of TFM over measured bodyweight. FLR was calculated as percentage of TFM divided by TLM. All participants fasted overnight for at least 12 hrs and did not exercise for at least 4 hrs before measurements. Changes in adiposity traits were calculated as follows: (value at follow-up–value at baseline) × 100/value at baseline.

The measured metabolic syndrome traits were blood pressure (BP), fasting serum glucose (FSG), triglycerides (TG), and high-density lipoprotein cholesterol (HDL-C). BP was manually assessed using a standard mercury sphygmomanometer under standard conditions. The levels of FSG (measured using hexokinase enzymatic assay), TG (measured using enzymatic assay), and HDL-C (measured using enzymatic or homogeneous assay) were measured using ADVIA 1650 (Siemens, Germany) or HITACHI 7600-210/HITACHI 7180 (HITACHI, Japan) equipment. Inter-assay coefficient of variation (the average of the high and low coefficient of variation) for those measurements was set below 7%. Blood samples for the biochemical analyses were collected from each subject after a 12-hr overnight fast, and the analyses were conducted in a central laboratory authorized by the Korean Association of Quality Control for Clinical Laboratory Examination.

The criteria of metabolic syndrome traits were adapted from the Harmonized definition¹² and are as follows: high BP, BP ≥130/85 mmHg or a history of hypertension; high FSG, FSG ≥5.6 mmol/L (100 mg/dL) or a history of diabetes mellitus; high TG, TG ≥1.7 mmol/L (150 mg/dL); and low HDL-C, HDL-C <1.03 mmol/L (40 mg/dL) in men or <1.29 mmol/L (50 mg/dL) in women. Because we included WC as one of the adiposity traits, WC was not considered as a metabolic syndrome component. To make normal distribution of the changes in individual metabolic syndrome components, levels of individual metabolic syndrome components at baseline and follow-up were divided into quintiles where the value for each quintile was assigned from 1 (first quintile) to 5 (fifth quintile). The changes in individual metabolic syndrome components were defined as quintile value at follow-up minus quintile value at baseline (score range, −5 to 5). The metabolic syndrome clustering score was defined as the number of individual criteria present at baseline and follow-up (0–4, 0 = no metabolic traits; 4 = all four traits present). The change in metabolic syndrome clustering was calculated as clustering score at follow-up minus clustering score at baseline (score range, −4 to 4).

Self-reported questionnaires were used to assess medical history at baseline and follow-up, socioeconomic status at follow-up, alcohol consumption (current user vs. current nonuser), cigarette smoking (never vs. ever smoker), and physical activity with regular moderate to high intensity (yes vs. no) at follow-up.

Statistical analysis

Paired t-tests were applied to compare means of adiposity and metabolic syndrome traits at baseline and follow-up. McNemar tests were applied to compare proportion of socioeconomic status and health behaviors at baseline and follow-up. Those comparison tests were conducted separately in twin individuals and nontwin family members. In addition, those traits and variables at baseline and follow-up between the two groups were compared using t-test for continuous variables and chi-squared test for categorical variables. The outcome variables (change in each metabolic syndrome trait and change in metabolic syndrome clustering score) were assessed for normality of distribution. A mixed linear model was then applied to evaluate the associations between change in adiposity traits and change in metabolic syndrome traits. Appropriate adjustment were made to control for random (intra-familial correlations and twin effects) and fixed effects (age, sex, baseline score of metabolic syndrome trait, and health behaviors, educational level, and income level at follow-up). All statistical analyses were performed using IBM Statistical Package for the Social Sciences (SPSS) software version 22.0.0.0 (IBM Corp., Armonk, NY).

A bivariate variance-component analysis based on maximum likelihood ratio and variance component decomposition was used to evaluate the effect of shared genetic and environmental factor influences on the relationship between relative changes in adiposity phenotypes with relative change in metabolic syndrome phenotype components. The genetic and environmental correlations were analyzed using the Sequential Oligogenic Linkage Analysis Routines (version 6.6.2 package). The phenotypic correlations were partitioned into relationships explained by genetic (ρ _G) and environmental sharing (ρ _E) after adjusting for age and sex. Values of ρ _G or ρ _E that deviated significantly from zero (P < 0.05) were regarded as a genetic or environmental association between the two phenotypes, respectively. Genetic correlations indicated the extent to which the additive effects of the same set of genes influenced more than one phenotype, or common genetic factors that influenced both phenotypes through shared pathways.¹³ Environmental correlations indicated the environmental and nonadditive genetic effects.¹⁴

Results

The medical, socioeconomic, and educational information of the study participants are summarized in Table 1. Twin individuals were more likely to have higher education and income, be younger, and have lower adiposity levels and better metabolic traits compared to nontwin family members at baseline and follow-up. There were significant increases in WC, BP, and HDL level with decreasing in FSG and the sum of metabolic syndrome traits over the follow-up period of 3.1 ± 0.6 years in twin individuals and nontwin family members. TFM and BFP were significantly increased in twin individuals, while no significant changes in BMI and FLR in both groups over the follow-up period.

Table 1.

Comparison of Characteristics Over Time in Twins and Family Members (n = 563)

	Twins (n = 284)			Nontwin family members (n = 279)
	Baseline	Follow-up	P	Baseline	Follow-up	P
Men	84 (29.6)^a	84 (29.6)^b	1.000	107 (38.4)	107 (38.4)	1.000
Education
<High school graduation	29 (10.2)^a	33 (11.7)^b	0.167	85 (30.8)	84 (30.3)	0.286
Graduated high school	110 (38.7)	101 (35.8)		74 (26.8)	79 (28.5)
>High school graduation	145 (51.1)	148 (52.5)		117 (42.4)	114 (41.2)
Income
<1,500,000 won	57 (20.7)^a	48 (17.1)^b	0.002	88 (33.2)	87 (32.1)	0.132
1,500,000–<3,000,000 won	121 (44.0)	106 (37.7)		96 (36.2)	94 (34.7)
≥3,000,000 won	97 (35.3)	127 (45.2)		81 (30.6)	90 (33.2)
Current smoker	70 (24.6)	47 (16.5)	0.697	85 (30.5)	52 (18.6)	0.039
Physical activity (moderate or high)	170 (59.9)	149 (52.5)	0.055	160 (57.3)	150 (53.8)	0.395
Current alcohol user	179 (63.9)	168 (60.0)	0.229	173 (62.5)	150 (54.9)	0.002
Age (years)	40.1 ± 8.1^a	43.1 ± 8.1^b	<0.001	50.3 ± 13.9	53.4 ± 13.9	<0.001
BMI (kg/m²)	22.9 ± 2.7^a	23.0 ± 2.8^b	0.780	24.3 ± 3.0	24.2 ± 3.1	0.130
WC (cm)	77.8 ± 7.7^a	80.0 ± 8.2^b	<0.001	83.2 ± 8.3	84.5 ± 8.6	<0.001
TFM (kg)	17.0 ± 4.6^a	17.4 ± 5.0^b	0.045	19.0 ± 5.4	18.9 ± 6.0	0.680
BFP (%)	28.8 ± 6.2^a	29.5 ± 7.0	0.039	30.0 ± 7.5	29.9 ± 8.4	0.878
FLR (%)	41.6 ± 13.0^a	41.9 ± 13.0	0.636	44.6 ± 15.5	44.0 ± 16.5	0.409
Systolic blood pressure (mmHg)	107.3 ± 14.7^a	110.6 ± 14.9^b	<0.001	113.4 ± 16.6	116.9 ± 15.8	<0.001
Diastolic blood pressure (mmHg)	67.2 ± 9.8^a	72.3 ± 10.0	<0.001	68.9 ± 10.1	72.9 ± 9.5	<0.001
FSG (mmol/L)	5.06 ± 0.56^a	4.91 ± 0.80^b	<0.001	5.40 ± 1.30	5.10 ± 0.96	<0.001
TG (mmol/L)	1.17 ± 0.72^a	1.36 ± 0.87	<0.001	1.40 ± 0.88	1.44 ± 0.82	0.465
HDL-C (mmol/L)	1.24 ± 0.28^a	1.40 ± 0.31	<0.001	1.18 ± 0.27	1.38 ± 0.32	<0.001
Metabolic syndrome clustering	0.99 ± 0.91^a	0.78 ± 0.96^b	<0.001	1.42 ± 1.06	1.17 ± 1.05	<0.001

Metabolic syndrome clustering is computed as presence ( = 1) or absence ( = 0) of high blood pressure, high FSG, high TG, and low HDL-C. Data are presented as number (%) or mean ± standard deviation. P value represents comparison between baseline value and follow-up value in each group using McNemar test (for categorical variables) or paired t-test (for continuous variables).

P < 0.05 for comparison of baseline values between groups using t-test (for continuous variables) or chi-squared test (for categorical variables).

P < 0.05 comparison of follow-up values between groups using t-test (for continuous variables) or chi-squared test (for categorical variables).

BFP, body fat percentage; BMI, body mass index; FLR, fat mass-to-lean mass ratio; FSG, fasting serum glucose; HDL-C, high-density lipoprotein cholesterol; TFM, total fat mass; TG, triglycerides; WC, waist circumference.

Table 2 presents the associations between changes in adiposity traits and changes in metabolic syndrome clustering and individual metabolic syndrome traits after adjusting for random and fixed effects (as described in the Methods section). Increases in BMI, WC, TFM, and FLR were associated with increased metabolic syndrome clustering over time. In assessing the relationships between individual adiposity and metabolic syndrome traits, increase in TG was found to be associated with increases in all adiposity traits and increase in BMI was associated with increase in all individual metabolic syndrome traits except for FSG.

Table 2.

Associations of Changes in Metabolic Syndrome Clustering and Components with Changes in Adiposity Traits (n = 563)

	Change in MetS clustering	Change in SBP	Change in DBP	Change in FSG	Change in TG	Change in HDL-C
Change in BMI	0.033 (0.020, 0.046)^*	0.021 (0.002, 0.040)^*	0.021 (0.001, 0.041)^*	0.018 (−0.004, 0.041)	0.046 (0.026, 0.066)^*	−0.037 (−0.053, −0.020)^*
Change in WC	0.011 (0.001, 0.022)^*	0.011 (−0.004, 0.027)	0.007 (−0.010, 0.023)	0.020 (0.002, 0.038)^*	0.024 (0.007, 0.040)^*	−0.013 (−0.026, 0.001)
Change in TFM	0.003 (0.001, 0.005)^*	0.002 (−0.001, 0.006)	0.003 (−0.001, 0.007)	−0.001 (−0.005, 0.003)	0.006 (0.002, 0.010)^*	−0.004 (−0.007, −0.001)^*
Change in BFP	0.002 (−0.001, 0.005)	0.002 (−0.002, 0.006)	0.003 (−0.001, 0.007)	−0.002 (−0.006, 0.003)	0.005 (0.001, 0.009)^*	−0.003 (−0.006, 0.001)
Change in FLR	0.002 (0.001, 0.004)^*	0.001 (−0.001, 0.004)	0.003 (0.001, 0.006)^*	−0.001 (−0.004, 0.003)	0.004 (0.001, 0.007)^*	−0.002 (−0.004, 0.001)

Values are estimates (95% confidence interval) assessed by mixed linear model after adjusting for random effects (intra-familial and twin effects) and fixed effects (age, sex, baseline MetS trait, and education and income levels, smoking status, physical activity, and alcohol use at follow-up). Change in MetS clustering = MetS clustering at follow-up–MetS clustering at baseline; MetS clustering is computed as presence ( = 1) or absence ( = 0) of high BP, high FSG, high TG, and low HDL-C. Change in MetS component = quintile at follow-up-quintile at baseline. Change in adiposity trait = (value at follow-up–value at baseline) × 100/value at baseline.

P < 0.05.

BP, blood pressure; DBP, diastolic BP; MetS, metabolic syndrome; SBP, systolic BP.

Bivariate analysis showed significant environmental correlations between changes in all adiposity traits and change in metabolic syndrome sum (Table 3). Significant common environmental factor correlations (ρ _E) were found for the relationships between change in FSG and HDL-C with changes in all adiposity traits, change in diastolic BP with changes in BMI, TFM, and BFP, and change in TG with changes in BMI, WC, TFM, and FLR. Common genetic correlation (ρ _G) was only found for the relationship between change in TFM and change in diastolic BP.

Table 3.

Cross-Traits Correlations of Changes in Metabolic Syndrome Clustering and Components with Changes in Adiposity Traits (n = 563)

	Change in MetS clustering		Change in SBP		Change in DBP		Change in FSG		Change in TG		Change in HDL-C
	ρG	ρE	ρG	ρE	ρG	ρE	ρG	ρE	ρG	ρE	ρG	ρE
Change in BMI	0.72 ± 0.26	0.25 ± 0.06^*	0.12 ± 0.43	−0.04 ± 0.06	0.24 ± 0.27	0.13 ± 0.07^*	0.47 ± 0.38	0.31 ± 0.06^*	−0.03 ± 0.33	0.27 ± 0.06^*	0.48 ± 0.22	0.55 ± 0.05^*
Change in WC	0.81 ± 0.58	0.13 ± 0.06^*	−0.38 ± 0.85	−0.10 ± 0.06	0.59 ± 0.63	0.02 ± 0.07	1.00	0.17 ± 0.05^*	−0.03 ± 0.85	0.45 ± 0.05^*	0.10 ± 0.51	0.33 ± 0.06^*
Change in TFM	−0.04 ± 0.48	0.18 ± 0.06^*	−1.00	0.10 ± 0.05	−0.82 ± 0.52^*	0.17 ± 0.07^*	−0.33 ± 0.82	0.19 ± 0.06^*	−0.39 ± 0.56	0.23 ± 0.06^*	−0.39 ± 0.52	0.20 ± 0.06^*
Change in BFP	0.43 ± 0.39	0.49 ± 0.05^*	−0.35 ± 0.68	0.11 ± 0.06	0.01 ± 0.44	0.22 ± 0.06^*	−0.21 ± 0.93	0.5 1 ± 0.05^*	−0.36 ± 0.45	−0.08 ± 0.06	0.25 ± 0.48	0.14 ± 0.06^*
Change in FLR	0.01 ± 0.27	0.17 ± 0.07^*	−0.64 ± 0.49	0.06 ± 0.06	−0.09 ± 0.25	−0.02 ± 0.07	0.18 ± 0.41	0.17 ± 0.06^*	0.43 ± 0.26	0.17 ± 0.06^*	0.14 ± 0.25	0.24 ± 0.06^*

Values represent correlation coefficient ± standard error assessed by bivariate analysis after adjusting for age, sex, and baseline MetS trait. Change in MetS clustering = MetS clustering at follow-up–MetS clustering at baseline; MetS clustering is computed as presence (=1) or absence (=0) of high BP, high FSG, high TG, and low HDL-C. Change in MetS component = quintile at follow-up-quintile at baseline. Change in adiposity trait = (value at follow-up–value at baseline) × 100/value at baseline.

P < 0.05.

ρG, genetic correlation coefficient; ρE, environmental correlation coefficient.

Discussion

In the present subset of South Korean twins and their family members, we found that the changes in adiposity traits over time were positively associated with clustering of metabolic syndrome traits and that these associations were mainly regulated by shared environmental factors. The associations between changes in individual adiposity traits and changes in individual metabolic syndrome traits were variable. Nevertheless, change in BMI tended to be consistently associated with changes in individual metabolic syndrome traits compared to other adiposity traits, while change in metabolic syndrome trait TG was consistently associated with changes in all adiposity traits.

Previous studies have shown positive relationships between change in BMI and body weight with change in metabolic syndrome clustering^1
–3 and change in individual metabolic syndrome traits.^2,4 There is evidence for change in visceral fat predicting incident metabolic syndrome.¹⁵ However, comparable studies for the associations with changes in general fat mass have rarely been conducted. On the other hand, other population studies have revealed that anthropometric measures such as BMI and WC could well predict metabolic risk factors, as compared to the adiposity specific measures.^16,17 Given the findings from this and other previous studies, changes in adiposity-specific indicators such as TFM, BFP, and FLR may not be superior to anthropometric measures when assessing associations with changes in metabolic syndrome traits. In addition, change in BMI was the only adiposity trait to significantly affect changes in most individual metabolic syndrome component changes over time, suggesting it to be a more sensitive indicator compared to other adiposity traits. Similarly, change in TG may reflect earlier changes in adiposity traits compared to changes in other metabolic syndrome traits.

The phenotypic correlation coefficient between two traits is determined by the genetic and environmental correlation coefficient along with the heritability of the two traits. The phenotypic correlation coefficient between two traits is thus mainly determined by the environmental correlation coefficient under low heritability, while genetic correlation coefficient exerts a more determinant effect under high heritability.¹⁸ In our bivariate analysis to assess shared genetic and shared environmental effects on studied traits, we did not observe significant genetic correlations except for the relationship between change in TFM and change in diastolic BP whereas shared environmental correlations between most changes of adiposity traits and metabolic syndrome traits were significant. Our findings support those from the STANSLAS family study in which the common environmental effects such as familial environment and lifestyle had a significant effect on the short-term changes in metabolic syndrome-related factors.⁸

Studies on shared genetic or environmental correlations between changes in adiposity traits and changes in metabolic syndrome traits are scarce. Although some cross-sectional studies have shown genetic correlations between WC and metabolic syndrome features,¹⁸ we could find only one longitudinal study on genetic and environmental correlations between changes in anthropometric indices and changes in leptin and insulin.⁹ The significant genetic correlation we found between change in TFM and change in diastolic BP suggests that the genetic effect on one trait may be completely mediated by the effects of another trait, or two independent genetic effects may exist. In addition, shared environmental factors may cause overestimation of genetic correlation coefficients as has been reported in other studies.^18,19 Further longitudinal studies will be necessary to better elucidate the genetic and environmental correlations between changes in adiposity traits and changes in metabolic syndrome traits.

Current findings from our observational study highlight the importance of adiposity reduction in prevention of metabolic syndrome clustering and suggest potential environmental factors controlling these traits. However, there are some limitations to be noted for the study. First, we analyzed the metabolic syndrome traits using quintiles to take into account the normal distribution and adjusted for baseline metabolic syndrome traits to correct a potential of regression to the mean effects³; however, there may be residual confounding factors such as concurrent medications and diet. For instance, improved metabolic syndrome clustering and FSG and HDL-C over time may be explained by our study design. We sent the reports about measurement results including clinical interpretation of those findings and helpful health behaviors to the participants. Therefore, the participants may make efforts to improve their health status using nonpharmacological or pharmacological therapy. However, no data are available regarding the pharmacological and nonpharmacological interventions that may have been implemented by the participants. This is an important limitation of the present study.

Second, although our results were based on a subset of a population-based sample, volunteer bias cannot be excluded. Finally, longer time interval between measurements points may result in increased changes in those traits and then, current short follow-up period has a limitation for assessing stability or changes in genetic and environmental correlations over time. The limitations notwithstanding, our findings add to the scarce knowledge base of evidence on the role of genetic and environmental effects on the correlations between changes in adiposity and metabolic syndrome traits.

Considering the results of this study and previous knowledge, it would be reasonable to conclude that the positive associations between changes in anthropometric and fat mass indicators and changes in metabolic syndrome clustering appear to be mediated by shared environment influences rather than shared genetic influences. The findings from our study suggest that environmental control could be effective therapeutic intervention for obesity and metabolic syndrome.

Footnotes

Acknowledgment

This work was supported by the 2015 Inje University research grant.

Author Disclosure Statement

No conflicting financial interests exist.

References

Maison

, Byrne

, Hales

, et al. Do different dimensions of the metabolic syndrome change together over time? Evidence supporting obesity as the central feature. Diabetes Care, 2001; 24:1758–1763.

Hillier

, Fagot-Campagna

, Eschwege

, et al. Weight change and changes in the metabolic syndrome as the French population moves towards overweight: The D.E.S.I.R. cohort. Int J Epidemiol, 2006; 35:190–196.

Rutter

, Sullivan

, Fox

, et al. Baseline levels, and changes over time in body mass index and fasting insulin, and their relationship to change in metabolic trait clustering. Metab Syndr Relat Disord, 2014; 12:372–380.

Srinivasan

, Myers

, Berenson

. Rate of change in adiposity and its relationship to concomitant changes in cardiovascular risk variables among biracial (black-white) children and young adults: The Bogalusa Heart Study. Metabolism, 2001; 50:299–305.

Muller

, Lagerpusch

, Enderle

, et al. Beyond the body mass index: Tracking body composition in the pathogenesis of obesity and the metabolic syndrome. Obes Rev, 2012; 13 Suppl 2:6–13.

Lagerpusch

, Bosy-Westphal

, Kehden

, et al. Effects of brief perturbations in energy balance on indices of glucose homeostasis in healthy lean men. Int J Obes (Lond), 2012; 36:1094–1101.

Golla

, Strauch

, Dietter

, et al. Quantitative trait linkage analysis of longitudinal change in body weight. BMC Genet, 2003; 4 Suppl 1:S7.

Herbeth

, Samara

, Ndiaye

, et al. Metabolic syndrome-related composite factors over 5 years in the STANISLAS family study: Genetic heritability and common environmental influences. Clin Chim Acta, 2010; 411:833–839.

Friedlander

, Meiner

, Sharon

, et al. Leptin, insulin, and obesity-related phenotypes: Genetic influences on levels and longitudinal changes. Obesity (Silver Spring), 2009; 17:1458–1460.

10.

Sung

, Cho

, Lee

, et al. Healthy Twin: A twin-family study of Korea—protocols and current status. Twin Res Hum Genet, 2006; 9:844–848.

11.

Gombojav

, Song

, Lee

, et al. The Healthy Twin Study, Korea updates: Resources for omics and genome epidemiology studies. Twin Res Hum Genet, 2013; 16:241–245.

12.

Alberti

, Eckel

, Grundy

, et al. Harmonizing the metabolic syndrome: A joint interim statement of the International Diabetes Federation Task Force on Epidemiology and Prevention; National Heart, Lung, and Blood Institute; American Heart Association; World Heart Federation; International Atherosclerosis Society; and International Association for the Study of Obesity. Circulation, 2009; 120:1640–1645.

13.

Almasy

, Blangero

. Multipoint quantitative-trait linkage analysis in general pedigrees. Am J Hum Genet, 1998; 62:1198–1211.

14.

MacCluer

, Scavini

, Shah

, et al. Heritability of measures of kidney disease among Zuni Indians: The Zuni Kidney Project. Am J Kidney Dis, 2010; 56:289–302.

15.

Shah

, Murthy

, Abbasi

, et al. Visceral adiposity and the risk of metabolic syndrome across body mass index: The MESA Study. JACC Cardiovasc Imaging, 2014; 7:1221–1235.

16.

Bosy-Westphal

, Geisler

, Onur

, et al. Value of body fat mass vs anthropometric obesity indices in the assessment of metabolic risk factors. Int J Obes (Lond), 2006; 30:475–483.

17.

Lee

, Song

, Sung

. Which obesity indicators are better predictors of metabolic risk?: Healthy twin study. Obesity (Silver Spring), 2008; 16:834–840.

18.

Povel

, Boer

, Feskens

. Shared genetic variance between the features of the metabolic syndrome: Heritability studies. Mol Genet Metab, 2011; 104:666–669.

19.

Akesson

, Bensch

, Hasselquist

, et al. Estimating heritabilities and genetic correlations: Comparing the ‘animal model’ with parent-offspring regression using data from a natural population. PLoS One, 2008; 3:e1739.