PAPER Segregation analysis of waist circumference, hip circumference and waist-to-hip ratio in the Korean Nationwide Family Study

Transcription

1 (2002) 26, ß 2002 Nature Publishing Group All rights reserved /02 $ PAPER, hip circumference and waist-to-hip ratio in the Korean Nationwide Family Study SH Jee 1 *, MT Kim 2, SY Lee 3 and TH Beaty 4 1 Department of Epidemiology and Disease Control, Graduate School of Health Science and Management, Yonsei University, Seoul, Korea; 2 School of Nursing, Johns Hopkins University, Baltimore, MD, USA; 3 Department of Preventive Medicine and Public Health, School of Medicine, The Ajou University, Suwon, Korea; and 4 Department of Epidemiology, School of Hygiene and Public Health, Johns Hopkins University, Baltimore, MD, USA AIM: A central distribution of adipose tissue is frequently associated with cardiovascular disease and its risk factors. In this study, we investigated environmental, familial and genetic influences on waist circumference (WC), hip circumference (HC) and waistto-hip ratio (WHR) in 2507 members of 435 families who had participated in the Korean Nationwide Health Examination Survey. METHOD: Maximum likelihood methods were used to fit several genetic and nongenetic models of inheritance to these data to determine whether an unobserved Mendelian major gene could explain the familial distribution of WC, HC and WHR. Adjustments for age, age 2, body mass index, smoking, alcohol consumption and exercise were carried out separately for males and females by multiple regression procedures for WC, HC and WHR phenotypes prior to segregation analysis. Regression models were used to test genetic and non-genetic models in these 435 families. RESULTS: Segregation analysis did not provide statistical evidence of a major gene controlling either HC or WHR. Mendelian single-locus models with two underlying genotypic distributions were best supported by these data on WC, and this putative major gene explained the 22.4% of variance in adjusted WC. CONCLUSION: Future linkage studies may be worthwhile to further clarify the mechanisms controlling WC. (2002) 26, DOI: =sj=ijo= Keywords: waist; hip; waist-to-hip ratio; segregation analysis Introduction Waist circumference (WC) and waist-to-hip ratio (WHR) are anthropometric measures of obesity and are believed to be better predictors of coronary heart disease than body mass index (BMI). 1 3 However, the genetic underpinnings of WC and WHR remain to be determined. During the past decade, evidence has accumulated that higher WHR values represent a risk factor for cardiovascular disease. 4 6 Variation in WHR may reflect an increase of *Correspondence: SH Jee, Department of Epidemiology and Disease Control, Graduate School of Health Science and Management, Yonsei University, PO Box 8044, Seoul , South Korea. jsunha@yumc.yonsei.ac.kr Received 22 February 2001; revised 3 September 2001; accepted 1 October 2001 visceral adipose tissue and subcutaneous fat affecting WC, as well as a decrease in the gluteofemoral muscle affecting hip circumference (HC) alone. Although previous studies 7,8 have suggested a role for a major gene in influencing WHR, these results are still actively debated, especially for WC. Recent works have suggested that WC is more closely associated with measures of abdominal visceral fat than WHR However, there is little data available concerning possible genetic control of WC. Knowing the model of inheritance for any major gene influencing WC, HC and=or WHR would be helpful for directing linkage studies using either candidate genes or a genome-wide array of markers. In the present study, we have investigated patterns of familial correlation and have tested a series of major gene and non-genetic models of inheritance for three quantitative measures, WC, HC and WHR, in randomly ascertained Korean families from the Korean Nationwide Health Examination Survey.

2 Materials and methods Study sample Study subjects participated in the Korean Nationwide Health Examination Survey, 12 a population-based family study designed to identify and evaluate both genetic and nongenetic determinants of cardiovascular disease. In the present analysis, only families with three or more members were included. Families in which only one member had valid waist or hip measurement data were not included. After exclusions, this data set consisted of 2507 individuals in 435 families (1209 males and 1298 females). Details concerning data collection have been described previously. 12 The median number of individuals in a family was 5.76; 191 families had < 5 members, 148 had 6 members, 71 had 7 members, 22 had 8 members and 3 had > 9 members. Data collection protocols Data concerning age, gender, cigarette smoking, alcohol consumption, and exercise were obtained from an interviewer-administered questionnaire. Standing height was measured using a wall-mounted vertical metal meter, and body weight was recorded to the nearest kilogram using a balance scale. Body mass index (BMI) was also calculated as weight in kg divided by height in m 2. WC was measured midway between the lower rib and iliac crest; HC was measured at the level of the trochanters. Using a standardized questionnaire, participants were asked to report their smoking habits, including the number of cigarettes smoked per day and the duration of cigarette smoking in years, along with other health habits, including alcohol drinking and exercise. They were classified as current smokers if they smoked currently and had done so for at least 1 y, nonsmokers if they never smoked, and ex-smokers if they had smoked but had quit. Likewise, they were classified as current drinkers, nondrinkers and ex-drinkers. Exercise was just classified into two categories: yes or no. Statistical analysis Multiple linear regression was used to summarize the relationship between WC, HC, WHR, BMI, age, cigarette smoking, alcohol consumption and exercise. This analysis formed the basis for selecting which variables were used to adjust WC, HC and WHR levels prior to further genetic analysis. All of these variables were adjusted for age, age 2, BMI, cigarette smoking, alcohol consumption, and exercise. Standardized residuals for WC, HC and WHR were calculated separately for males and females. Familial correlation and heritability analysis Familial correlations in WC, HC and WHR were estimated by using the FCOR program in Statistical Analysis Genetic Epidemiology package version Correlation coefficients for spouse pairs (r sp ), parent offspring (r po ) and sibling pairs (r ss ) were computed, adjusting for age, BMI, cigarette smoking, alcohol drinking and exercise. The heritabilities were computed the correlation coefficients by using Rice s method. 14 The algebraic equation for the heritability (h 2 ) is: h 2 ¼ (r ss þ r po )(1 þ r sp )=(1 þ r sp þ (2r sp *r po )). Segregation analysis To investigate the role of genetic and environmental influences in determining inter-individual variability in WC, HC and WHR, a series of class D regressive models proposed by Bonney 15 and implemented in the SAGE 13 were used. These models assume that variation in a quantitative trait among individuals is the result of a major gene effect, residual variation that may show familial correlations and individual variation. The class D models used in this analysis assume that, given a common parentage, additional sibling effects are equal among all sibs in a nuclear family. The most general model describes the distribution of WC, HC and WHR values as due to independent contributions of a single factor with a major effect (which may or may not represent a single Mendelian gene), residual familial correlations and individual error. This general model considers two alleles at a single locus (denoted L and H), resulting in three possible types of individuals (LL, LH and HH), termed ousiotypes by Cannings et al. 16 For example, the mean WC level associated with each ousiotype is denoted m LL, m LH or m HH, with a variance of s 2 (assumed to be equal among all three types ). Transmission of the L or H factor may be attributable to an unobserved Mendelian gene or some non-genetic environmental factor. The parameter P is defined as the frequency of allele L; q ¼ (1 7 P) equals the frequency of allele H in the population from which these families are drawn. Founder individuals are generally assumed to come from a reference population in Hardy Weinberg equilibrium. Individuals with types LL, LH and HH are assumed to transmit the allele L with probabilities t LL, t LH and t HH, respectively. These transmission probabilities are used to calculate probabilities of all three types among non-founder individuals whose parents are included in the pedigree. In addition, the parameters r sp, r po (or r mo, r fo ) and r ss denote spouse, parent offspring (mother offspring or father offspring), and sib sib residual correlation, respectively. We fit a series of explicit genetic and arbitrary nongenetic models of inheritance to these family data to identify the most appropriate model of inheritance and to test each sub model against the general model (which has 11 parameters). The likelihood ratio test (LRT) was computed as twice the difference in log likelihood values between any two models (a reduced model and the general model) and served to test hypotheses regarding these models. As the number of pedigrees increases, this LRT statistic will approximate a w 2 distribution with degrees of freedom equal to the difference in the number of estimated parameters between the reduced and general models. Under this approach, genetic or nongenetic models giving a poor fit to the data will have a high 229

3 230 w 2 value and will be rejected. We also compared non-nested models using Akaike s Information Criterion (AIC). The AIC is computed as ( 7 2 log likelihood þ 2 (no. of estimated parameters)) to incorporate a penalty for models having more parameters. The most conservative model can be easily identified as that having the lowest AIC value. Results Study population The mean age was 36.6 y (males, 36.1 and females, 37.1 y), with a wide range (10 94 y) (Table 1). Overall, the study population showed a high prevalence of cigarette smoking among males (48.7%). WHR was normally distributed and ranged from 0.67 to 1.36, with a mean of 0.84 (males, 0.82 and females, 0.87). WC ranged from 55.5 to cm in the entire sample, with a mean of 77.3 cm; the distribution was slightly positively skewed (a coefficient of skewness, 0.02). HC ranged from 70.2 to cm, with a mean of 92.0 cm. The distribution of HC, before and after log transformation, was negatively skewed (a coefficient of skewness, ). Adjustment of the WC, HC and WHR In adjusting the observed WC, HC and WHR, we chose to include potential confounding variables (age, age 2, BMI, cigarette smoking, alcohol consumption and exercise) in a multiple linear regression model. For WC, these five variables accounted for 79.4% in males and 80.1% in females of the Table 1 General characteristics of study subjects: the Korean Nationwide Family Study Variables Males (s.e.) (n ¼ 1209) Females (s.e.) (n ¼ 1298) Age (y) (17.64) (19.23) Waist (cm) (10.47) (10.68) Hip (cm) (7.96) (7.82) Waist-to-hip ratio 0.87 (0.08) 0.82 (0.09) Weight (kg) (12.23) (9.79) Height (cm) (8.45) (6.81) Body mass index (3.55) (3.72) Current cigarette smoking 48.69% 2.93% Current alcohol drinking 60.30% 37.40% Exercise 15.70% 8.16% observed variation in WC. For HC, these five variables accounted for 62.9% in both females and males (Table 2). For WHR, these five variables accounted for 44.0% in males and 44.6% in females of the observed variation in WHR. Correlations among three anthropometric measures We found a high correlation (r ¼ 0.79, P < 0.001) between observed WHR and WC, but a much lower correlation (r ¼ 0.18, P < 0.001) between WHR and HC. All correlations shown in Table 3 are significantly different from zero. The correlation between WHR and WC decreased slightly after adjustment for covariates such as age, BMI and lifestyle, but remained highly significant. Familial correlations for WC, HC and WHR Table 4 shows estimated familial correlations for adjusted WC, HC and WHR. Adjusted WC, HC and WHR showed a correlation pattern consistent with possible genetic control, ie a low spouse (mother father) correlation and higher parent offspring and sibling correlations. Heritabilities in this study were estimated to be 60.0% for WC, 45.4% for HC and 42.2% for WHR, respectively. Segregation analysis of WC We first conducted segregation analysis on WC values adjusted for age and gender; the results did not provide statistical evidence of a major gene (results not shown). We then repeated the segregation analysis after adjusting age, age 2, gender and BMI. A series of nine regressive models were fit to the data on WC. Table 5 shows parameter estimates, 7 2 log-likelihood and LRT statistics for the following models: (1) a sporadic model, which allows random environmental effects but assumes no genetic transmission or variation; (2a c) models with familial correlations, but no major gene effect; (3a c) Mendelian single locus models, which allow a major gene effect, as well as residual familial correlation; and (4) an equal transmission model, which allows a major effect (three distinct phenotypic subgroups Table 2 Variance in adjusted waist circumference (WC), hip circumference (HC) and waist-to-hip ratio (WHR) explained by covariates in the Korean Nationwide Family Study Covariates in model WC HC WHR Women Men Women Men Women Men Age variables a Age þ BMI Age þ BMI þ lifestyle variables b a Age variables: age and age 2. b Lifestyle variables: cigarette smoking, alcohol drinking and exercise. Table 3 Correlation between waist-to-hip ratio (WHR), waist circumference (WC) and hip circumference (HC) in the Korean Nationwide Family Study WC HC r P-value r P-value Crude Age variables a Age þ BMI Age þ BMI þ lifestyle variables b a Age variables: age and age 2. b Lifestyle variables: cigarette smoking, alcohol drinking and exercise.

4 Table 4 Familial correlation for adjusted waist circumference (WC), hip circumference (HC) and waist-to-hip ratio (WHR) among 2307 members of 435 families 231 WC HC WHR Number of pairs Model 1 Model 2 Model 3 Model 1 Model 2 Model 3 Model 1 Model 2 Model 3 Spouse Parents offspring Mother son Mother daughter Father son Father daughter Siblings (intra class) Sister sister Brother brother Sister brother Model 1: adjusted for age. Model 2: adjusted for age and body mass index. Model 3: adjusted for age, body mass index, cigarette smoking, alcohol drinking and exercise. or types) but no transmission of the underlying factor from parents to offspring, along with residual familial correlation. We assessed the goodness of fit for each of these four restricted models (1 4) compared to the general or unrestricted model (5). The sporadic (LRT ¼ 115.8, d.f. ¼ 9, P < 0.001) and familial correlation models were both strongly rejected, suggesting that multiple underlying distributions do exist. The model of equal transmission of the major effect (LRT ¼ 19.9 with 3 d.f., P < 0.001) was also rejected. The Mendelian recessive model (3b) with residual familial correlations was also rejected. However, the Mendelian dominant (model 3a, LRT ¼ 2.5 with 4 d.f., P ¼ 0.780) and the codominant model (3c) with residual familial correlations could not be rejected (LRT ¼ 2.5 with 3 d.f., P ¼ 0.999). Both these Mendelian models gave the same 7 2lnL value, suggesting that a model with two or three underlying distributions provided an adequate fit to these data. When model 3a was compared to 3c in terms of the AIC value, the two-distribution dominant model was more conservative. In the dominant model (3a), the predicted WCs for genotype LL and HH were quite different (76.6 and 87.3, respectively). Maximum-likelihood estimates for the transmission parameters under the general model (LL ¼ 1.00, LH ¼ 0.35, Table 5 in 2507 members of 435 families q L m LL (s.e.) m LH (s.e.) m HH (s.e.) s 2 (s.e.) t LL t LH t HH SP (s.e.) PO (s.e) SS (s.e.) 7 2lnL w 2 d.f. P AIC NP a Sporadic < (0.12) (0.94) Familial correlation (a) Unimodel (sp) < (0.12) (0.94) (0.05) (b) Unimodel (sp, po) < (0.13) (0.95) (0.05) (0.02) (c) Unimodel (sp, po, ss) < (0.13) (0.96) (0.05) (0.02) (0.03) Mendelian (a) Dominant m LL > (0.03) (0.16) (s.e) (0.99) (1.06) (0.09) (0.03) (0.04) (b) Recessive m HH < (0.08) (0.15) (0.99) (s.e) (0.99) (0.05) (0.03) (0.03) (c) Codominant > (0.03) (0.52) (0.86) (0.94) (1.09) (0.05) (0.04) (0.04) Equal transmission q L q L q L < (0.04) (0.94) (1.94) (1.49) (1.03) (0.05) (0.03) (0.04) General (t free) (1.00) (0.04) (0.36) (0.88) (1.07) (1.10) (0.01) (0.01) (0.06) (0.04) (0.04) The transmission probability estimate was fixed at the upper bound. a NP, number of parameters.

5 232 HH ¼ 0.00) closely resembled the probabilities expected for Mendelian transmission of two alleles at a single major gene. Maximum-likelihood estimates from this best-fitting dominant model were used to estimate the proportion of variance in adjusted WC that could be explained by this putative major gene. Multiple linear regression analyses reduced the variance of WC from to 28.8 (74.8%) when age, age 2, BMI, cigarette smoking, alcohol consumption and exercise were considered as covariates. The variance in adjusted WC was further reduced to 22.3 when a major gene effect was included (model 3a). These results suggested that this putative major gene can explain 22.4% of the total variance in residual WC, ie s 2 ¼ P (m i 7m) 2 f i =s 2 Total Adj. Segregation analysis of HC and WHR We conducted segregation analysis on HC and WHR separately, both before and after adjustment for age, BMI and lifestyle variables, in the same manner as presented for WC. These analyses did not provide statistical evidence of a major gene influencing WHR or HC (results not shown). Discussion Data from this family study indicated that WC, HC and WHR show significant familial correlation, after adjusting for age, age 2, BMI, cigarette smoking, alcohol consumption and exercise. These residual correlations reflect the combined effects of polygenic heritability (h 2 ) adjusted for age, BMI, cigarette smoking, alcohol drinking and exercise and any unobserved major genes, showing overall heritabilities estimated at 60.0% for WC, 45.4% for HC and 42.2% for WHR. Segregation analysis showed strong evidence of a Mendelian major gene locus for WC, but not for HC or WHR. Environmental (cigarette smoking, alcohol consumption and exercise) and anthropometric (age and BMI) factors together explained about 80% of the total variance in WC. Our results suggested that adjusted WC may be under control of a major gene, although other measures of obesity (HC and WHR) may be under more complex genetic control. There have been a number of twin studies addressing genetic control of WC. 8,17 Selby et al, 18 using the National Heart, Lung and Blood Institute s (NHLBI) Twin Study, found heritability estimates for WC ( ¼ 0.46) after controlling for BMI. Rose et al 17 reported that age and BMI-adjusted heritability estimates ranged from 0.72 to 0.82 for WC and from 0.36 to 0.61 for WHR in their study. Nelson et al 8 also reported that WC had higher heritabilities among males and females than did WHR. All these studies demonstrated that the estimated heritability was higher for WC than for WHR. They have also suggested the importance of further study to elucidate the mode of inheritance for both WC and WHR. In contrast to our results, a recent family study 7 has provided evidence for the presence of a major gene controlling WHR after adjustment for age and BMI, suggesting that it is an appropriate trait for further genetic linkage analysis. However, this group did not control the effect of lifestyle factors such as smoking, drinking and exercise. Moreover, they did not find evidence of a major gene for WC. One similar finding from the study of Feitosa et al 7 and our study is that both studies concluded a higher heritability of WC than that of WHR, and suggesting that WC was more likely to be explained by single locus models. In our study, we also tried to examine whether there was evidence of a major gene for WHR, after controlling for age and BMI. For accurate comparison, we used a method similar to that employed by Feitosa et al, 7 for the NHLBI family heart study, both before and after controlling for lifestyle factors. We could not find any evidence of a major gene for WHR. The possible reasons for these different findings between our study and theirs are crucial to improve an understanding of genetic control of WHR. One difference concerned the levels of WHR. Our population was relatively lean, and WHR levels in this study (0.82 for males and 0.87 for females) were lower than those in the study of Feitosa et al 7 (0.96 for males and 0.88 for females). Moreover, two samples showed very different distributions in both BMI and WC. The mean BMI of subjects from the NHLBI family heart study 19 was for females and for males, while the corresponding BMI levels in our study were and 22.32, respectively. The overall WC from the NHLBI family study were 85 cm for females and 95 cm for males, while the corresponding WC in our study were and cm, respectively. Another difference was the mean family size and its effect on the statistical power of the study sample. In our study, the average number of family members was 5.76, whereas in the Feitosa study it was only 2.61 per family. Finally, this difference may or may not be racial or ethnic in origin. Therefore, further study is needed with regard to genetic control of both WHR and WC. To our knowledge, our data provides the first strong evidence for a major gene for WC, after controlling for age and BMI. In addition, we controlled for the effect of lifestyle factors on WC, and then performed segregation analysis again to evaluate whether a major gene might still influence WC. There was still strong evidence of a major gene controlling WC, and this putative major gene accounted for 22.4% of the variation in WC in this sample. The findings of this study may partially validate the major gene theory for abdominal visceral fat In 1996, Bouchard 20 and his colleagues reported that the variability in visceral fat area was accounted for according to Mendelian expectation, although the support for the major gene theory was less strong after controlling for total fat mass. In the present study, we controlled BMI, instead of total fat mass, to test the possibility that WC and BMI are influenced by the similar etiological factors. Unlike Bouchard et al s findings, we had the major gene evidence for WC after the adjustment of BMI as substantiated by a Mendelian single-locus models with two underlying genotypic distributions. For HC, we also performed segregation analysis to examine the evidence of a major gene. There was no evidence of a

6 major gene for HC, before or after transformation to reduce the skewness. In our study, WC was highly correlated with WHR (r ¼ 0.79), whereas HC was only weakly correlated with WHR (r ¼ 0.18). Although we found evidence of a major gene for WC, we saw no such evidence for WHR in this study. Our study has several limitations. First, our family sizes were relatively small and many families spanned only two generations; therefore, the data are less informative than those derived from larger pedigrees of three or more generations. Furthermore, there is considerable error associated with single measurements of waist and hip circumference; the variability due to measurement error for WC and WHR might be high in our data. Since residual environmental variance inflates the total variance of these obesity factors, the genetic effects estimated here may underestimate both familial correlation and genotypic means in these Mendelian models because these data are cross-sectional, as with all family studies. In summary, we have found a clear familial aggregation of three anthropometric measures of obesity: WC, HC and WHR. In this study of relatively lean Korean population, only WC was consistent with a major gene effect, even after controlling for the effect of age, BMI, smoking, drinking and exercise. The putative major gene described here may provide guidance for further study of the molecular basis for elevated WC. Acknowledgements This study was partially supported from an HMP grant (no. HMP-98-M ) of the 1998 Good Health Research and Development Project, Ministry of Health and Welfare, Republic of Korea. The results of this paper were obtained by using the program package SAGE, which is supported by a US Public Health Service Resource Grant (1 P41 RR03655) from the National Center for Research Resources. References 1 Donahue R-P, Abbott RD, Bloom E, Reed DM, Yano K. Central obesity and coronary heart disease in men. Lancet 1987; i: Lapidus L, Bengtsson C, Larsson B, Pennert K, Rybo E, Sjiistriim L. Distribution of adipose tissue and risk of cardiovascular disease and death: a 12 y followup of participants in the population study of women in Gothenburg. Sweden. Br Med J 1984; 288: Rimm EB, Stampfer MJ, Giovannucci E et al. Body size and fat distribution as predictors of coronary heart disease among middle-aged and older US men. Am J Epidemiol 1995; 141: Onat A, Sansoy V, Uysal O. Waist circumference and waist-to-hip ratio in Turkish adults: interrelation with other risk factors and association with cardiovascular disease. Int J Cardiol 1999; 70: Megnien JL, Denarie N, Cocaul M, Simon A, Levenson J. Predictive value of waist-to-hip ratio on cardiovascular risk events. Int J Obes Relat Metab Disord 1999; 23: Gupta R, Majumdar S. Correlation of waist-hip ratio with coronary heart disease and risk factor prevalence in a rural male population. Indian Heart J 1994; 46: Feitosa MF, Borecki I, Hunt SC, Arnett DK, Rao DC, Province M. Inheritance of the waist-to-hip ratio in the National Heart, Lung, and Blood Institute Family Heart Study. Obes Res 2000; 8: Nelson TL, Vogler GP, Pedersen NL, Miles TP. Genetic and environmental influences on waist-to-hip ratio and waist circumference in an older Swedish twin population. Int J Obes Relat Metab Disord 1999; 23: Seidell J, Cigolini M, Charzewska J, Ellsinger BM, Deslypere J, Cruz A. Fat distribution in European men: a comparison of anthropometric measurements in relation to cardiovascular risk factors. Int J Obes Relat Metab Disord 1992; 16: Pouliot MC, Despres JP, Lemieux S et al. Waist circumference and abdominal sagittal diameter: best simple anthropometic indexes of abdominal visceral adipose tissue accumulation and related cardiovascular risk in men and women. Am J Cardiol 1994; 73: van der Kooy K, Seidell J. Techiques for the measurement of visceral fat: a practical guide. Int J Obes Relat Metab Disord 1993; 17: Ministry of Health and Welfare Korean National Health Examination Survey. Korean Institute of Health and Social Affairs; p SAGE. Statistical analysis for genetic epidemiology, release 3.1. Computer program package available from Department of Epidemiology and Biostatistics, Rammelkamp Center for Education and Research, Metrohealth campus, Case Western Reserve University, Cleveland, OH, USA, Rice T, Despres JP, Daw EW et al. Familial resemblance for abdominal visceral fat: the Heritage family study. Int J Obes Relat Metab Disord 1997; 21: Bonny GE. On the statistical determination of major gene mechanisms in continuous human traits: regressive models. Am J Med Genet 1984; 18: Cannings E, Thompson EA, Skolnick MH. Probability functions on complex pedigrees. Adv Appl Prob 1978; 10: Rose KM, Newman B, Mayer-Davis EJ, Selby JV. Genetic and behavioral determinants of waist hip ratio and waist circumference in women twins. Obes Res 1998; 6: Selby J, Newman B, Quesenberg C, Fabsitz R, Carmelli D, Meaney J, Slemenda C. Genetic and behavioral influences on body fat distribution. Int J Obes 1990; 14: Wilk JB, Djousse L, Borecki I et al. Segregation analysis of serum uric acid in the NHLBI family study. Hum Genet 2000; 106: Bouchard C, Rice T, Lemieux S, Despres JP, Perusse L, Rao DC. Major gene for abdominal visceral fat area in the Quebec Family Study. Int J Obes Relat Metab Disord 1996; 5: Rice T, Borecki IB, Bouchard C, Rao DC. Segregation analysis of body mass index in an unselected French Canadian sample: the Quebec Family Study. Obes Res 1993; 1: Rice T, Borecki IB, Bouchard C, Rao DC. Segregation analysis of fat mass and other body composition measures derived from underwater weighing. Am J Hum Genet 1993; 52: