International Journal of Obesity (2003) 27, 491–497 & 2003 Nature Publishing Group All rights reserved 0307-0565/03 $25.00 www.nature.com/ijo PAPER Evidence for joint action of genes on status and CVD risk factors in American Indians: the Strong Heart Family Study

KE North1*, JT Williams2, TK Welty3, LG Best4, ET Lee5, RR Fabsitz6, BV Howard7 and JW MacCluer2

1Department of Epidemiology, University of North Carolina, Chapel Hill, NC, USA; 2Department of Genetics, Southwest Foundation for Biomedical Research, San Antonio, TX, USA; 3Aberdeen Area Tribal Chairmen’s Health Board, Rapid City, SD, USA; 4Missouri Breaks Industries Research, Inc., Timber Lake, SD, USA; 5Center for American Indian Health Research, School of Public Health, University of Health Sciences Center, Oklahoma City, OK, USA; 6Epidemiology and Biometry Program, National Heart, Lung, and Blood Institute, Bethesda, MD, USA; and 7MedStar Research Institute, Washington, DC, USA

OBJECTIVES: Previous research among American Indians of the strong heart family study (SHFS) has demonstrated significant heritabilities for CVD risk factors and implicated diabetes as an important predictor of several of the phenotypes. Moreover, we recently demonstrated that genetic effects on CVD risk factors differed in diabetic and nondiabetic individuals. In this paper, we investigated whether a significant genetic influence on diabetes status could be identified, and whether there is evidence for joint action of genes on diabetes status and related CVD risk factors. METHODS AND RESULTS: Approximately 950 men and women, age 18 or older, in 32 extended families, were examined between 1997 and 1999. We estimated the effects of genes and environmental covariates on diabetes status using a threshold model and a maximum likelihood variance component approach. Diabetes status exhibited a residual heritability of 22% (h2 ¼ 0.22). We also estimated the genetic and environmental correlations between diabetes susceptibility and eight risk factors for CVD. All eight CVD risk factors displayed significant genetic correlations with diabetes status (BMI (rG ¼ 0.55), fibrinogen (rG ¼ 0.40), HDL-C (rG ¼À0.37), ln triglycerides (rG ¼ 0.65), FAT (rG ¼ 0.38 ), PAI-1 (rG ¼ 0.67), SBP (rG ¼ 0.57), and WHR (rG ¼ 0.58)). Three of eight traits (HDL-C (rE ¼À0.32), ln triglycerides (rE ¼ 0.33), and fibrinogen (rE ¼ 0.20)) displayed significant environmental correlations with diabetes status. CONCLUSIONS: These findings suggest that in the context of a high of diabetes, still unidentified diabetes genes may play an important role in influencing variation in CVD risk factors. International Journal of Obesity (2003) 27, 491–497. doi:10.1038/sj.ijo.0802261

Keywords: pleiotropy; diabetes status; CVD risk factors; American Indians; strong heart family study

Introduction of CVD in 13 American-Indian communities in Cardiovascular (CVD) is the most common macro- three geographic regions ( (AZ), Oklahoma (OK), and vascular complication of diabetes mellitus, accounting for North and South Dakota (DK)), and to identify CVD- 80% of deaths among diabetic individuals.1,2 The relation of associated risk factors. A pilot family study was added in the diabetes to CVD and associated CVD risk factors has been the third phase, the Strong Heart Family Study (SHFS), which was focus of recent research among American Indians of the designed to localize genes influencing CVD risk factors in 30 Strong Heart Study (SHS). This longitudinal population-based large families (approximately 950 participants). study was initiated in 1988 to determine the prevalence and Among American Indians of the SHS the prevalence of diabetes has reached epidemic proportions3 and is believed to contribute to the increasing rates of CVD morbidity and 4,5 *Correspondence: Dr KE North, Department of Epidemiology, University mortality. Additionally, alarming increases in CVD risk of North Carolina at Chapel Hill, Bank of America Center, 137 E. Franklin factors across time have been reported.3 Many CVD risk Street, Suite 306, Chapel Hill, NC 27514-3628, USA. factor phenotypes are significantly heritable, and diabetes E-mail: [email protected] Received 5 June 2002; revised 5 November 2002; status has been identified as an important correlate of these 6 accepted 9 December 2002 risk factors in American Indians participating in the SHFS. Action of genes on diabetes status and CVD risk factors KE North et al 492 Moreover, we recently demonstrated that genetic effects on Strong heart family study obesity and lipid-related CVD risk factors in the SHFS The design and methods of the SHFS have been described differed in diabetic and nondiabetic individuals.7 elsewhere.7 Approximately 950 men and women, age 18 y or The heritability of diabetes susceptibility has not been older, in 32 extended families, were examined between 1997 examined in American Indians of the SHFS. However, many and 1999 in the three field centers: Arizona, Oklahoma, studies have implicated a strong genetic component to and North and South Dakota. The Arizona Field center is variation in diabetes status.8 Other evidence that diabetes located in Phoenix and has enrolled primarily Pima Indians, has strong genetic determinants includes marked disease- but there are also representatives of the closely related rate differences between populations9,10 and a close corre- Maricopa and Tohono O’odham tribes.24 The Oklahoma spondence between genetic admixture rates and disease center encompasses 11 tribes, primarily Plains and South- prevalence in hybrid populations.11–13 Twin studies have eastern American Indians living in the Lawton-Anadarko, demonstrated a high rate of concordance among mono- Oklahoma area (Apache, Caddo, Comanche, Delaware, Fort zygotic (MZ) twins.14–17 Additionally, family studies have Sill Apache, Kiowa, and Wichita tribes). The Dakota Center established the contribution of a positive family history in has examined participants primarily from the Cheyenne the development of diabetes18,19 and a higher prevalence of River Sioux tribe of the Cheyenne River reservation in South diabetes in first, second, and third degree relatives in Dakota. comparison to controls20,21. All these studies are suggestive of a significant genetic component to diabetes susceptibility. In this study, we wished to determine if a significant Phenotypic, demographic, and lifestyle data genetic influence on diabetes status could be identified and The SHFS exam consisted of a personal interview, physical whether there is evidence for the joint action of genes on examination, laboratory tests, and a carotid ultrasound. diabetes status and related quantitative CVD risk factors in Standard protocols were used for the collection of all data the American Indians of the SHFS. To accomplish these and are described in detail in previous publications24,25. objectives, we first estimated the effects of genes and Body fat mass was measured using an RJL bioelectric environmental covariates on diabetes susceptibility, using a impedance meter and estimated by the RJL formula based threshold model and a maximum likelihood variance on total body water.26 component approach.22 We then used an extension of Blood pressure was measured three times and the mean of variance component methods to determine the genetic and the last two measurements was used for analysis. environmental correlations between quantitative risk factors Fasting blood samples were obtained during the physical for CVD and discrete disease outcomes (diabetes status).23 examination for the measurement of lipids, lipoproteins, Genetic and environmental correlations with diabetes apolipoproteins, insulin, glucose, plasma creatinine, plasma susceptibility were assessed for eight CVD risk factors (body fibrinogen, and PAI-124,27. All phenotypes were assayed at fat (FAT), body mass index (BMI), HDL (HDL-C), MedStar Research Institute, Washington, DC, using standard ln fibrinogen, ln triglycerides, plasminogen activator inhi- laboratory methods as previously described24,27. The b- bitor 1 (PAI-1), systolic blood pressure (SBP), and waist-to-hip quantification procedure for lipoprotein quantification, as ratio (WHR)). described by the Lipid Research Clinics, was applied to these data for family study participants. For cohort participants, the b-estimation procedure for lipoprotein quantification Materials and methods was applied.28 Triglycerides were measured using the Trigly- Strong heart study ceride-GB system pack for Hitachi 717. Fibrinogen was The study design and methods of the SHS have been measured in an automated clot-rate assay based upon the described previously.24 The SHS is a longitudinal study with original method of Clauss.29 The ST4 Instrument three clinical examinations and mortality and morbidity (Diagnostica Stago), with standardization with the CAP surveillance of resident tribal members aged 45–74 y (the reference material, was used to measure fibrinogen levels. cohort). During the baseline examination, conducted be- PAI-1 was measured using a double antibody ELISA, origin- tween 1989 and 1991, 4549 tribal members were examined. ally developed by DeClerck et al.30 This assay is sensitive to The second examination, conducted between 1993 and free PAI-1 (latent and active) and not to PAI-1 in complex 1995, included 89% of the surviving members of the original with t-PA. cohort and the third examination, conducted in 1998–1999, The glucose tolerance test was performed only if the included 88% of surviving members of the cohort. A pilot participant was not on insulin or any oral hypoglycemic family study, the SHFS, was added in 1998 in which 8–12 medications for known diabetes and if the participant had a extended families (more than 300 family members at least fasting glucose value less than 225 mg/dl (as determined by 18 y of age) were recruited and examined at each center. An Acucek II, Baxter Healthcare, Grand Prairie, TX, USA). extension of the pilot family study is currently in progress Diabetes (DM), impaired glucose tolerance (IGT), and and will involve recruitment of approximately 90 additional normal glucose tolerance (NGT) status were determined families, for a total of 2700 new participants. according to the World Health Organization criteria.31

International Journal of Obesity Action of genes on diabetes status and CVD risk factors KE North et al 493 All participants gave informed consent for this study, individuals with impaired glucose tolerance (n ¼ 119) were which was approved by the Institutional Review Boards of all excluded from these analyses. The analysis of each pheno- of the participating institutions. type was restricted to those individuals for whom complete covariate data were available. Lipid phenotypes were not Statistical genetic methods analyzed for individuals currently taking lipid-lowering The phenotypic variance in CVD risk factors was partitioned medications (n ¼ 13), and blood pressure measures were not into additive genetic and environmental variance compo- analyzed for individuals currently taking antihypertensive nents,32 using maximum likelihood variance decomposition medications (n ¼ 143). For each quantitative trait, all outliers methods.33–35 These univariate quantitative genetic analyses (here defined as any value more than 3.5 s.d. from the mean) were performed using the computer package SOLAR.36 were removed prior to analysis to minimize the effect of trait The heritability of diabetes status was estimated using a kurtosis.38 pedigree-based maximum likelihood method that models affection status by a liability threshold model.32,22 Although disease status is a qualitative trait, with individuals scored as Results either affected or unaffected, the threshold model assumes Information on diabetes status was available for 887 there is an underlying quantitative liability to disease. If an participants. The prevalence of diabetes was 30% (nearly all individual’s liability exceeds a specified threshold, the type II diabetes), with a higher proportion of diabetic female individual becomes affected. If an individual’s liability is (168/518, 32%) than male individuals (100/369, 27%). The below the threshold, the individual is unaffected. Liabilities median age of the sample is 39.2 y and the high rate of and the threshold value itself are estimated using age- and diabetes in this young sample is a clear indication of the sex-specific parameters, and the population prevalence of continuing problem of diabetes among American-Indian the disease trait. populations. We next used a mixed discrete-continuous trait variance In total, 31 of 32 of the SHFS families had at least one component method for mixed traits to examine the affected participant, with a mean of 8.6 diabetics per family genetic relation between diabetes susceptibility and quanti- group (range between 4 and 17). The diabetic participants in 23 tative variation in CVD risk factors. Specifically, we these 31 families included a large number of relative pair estimated the genetic and environmental correlations be- types (Table 1). A total of 4079 relation pairs included at least tween diabetes susceptibility and eight quantitative risk one diabetic participant with many instances of first degree factors for CVD. (eg, siblings or parents and offspring), second degree (eg, P-values for estimated effects are obtained using likelihood avuncular pairs, grandparent–grandchild pairs), and third ratio tests, where the likelihood of a given model is estimated degree or greater relative pairs (eg, great-avuncular pairs, first and compared to the likelihood of the model in which the cousin pairs, and second cousins). effect is absent. Twice the difference in the natural logarithmic likelihoods is asymptotically distributed as a 1/2 : 1/2 mixture of a w2 variable with one degree of freedom Heritability of diabetes status 37 and a point mass at zero. Using the threshold model, diabetes status exhibited a Evidence consistent with pleiotropy (in which the same residual heritability of 41% (h2 ¼ 0.4170.09, Po0.00001). gene influences several traits) is indicated by additive genetic Therefore, after correction for age, sex, age-by-sex inter- correlations that are significantly different from zero. The action, age2, and age2-by-sex interaction, 41% of the detection of significant shared genetic effects on multiple phenotypes can lead to hypotheses regarding the genetic Table 1 Strong Heart Family Study: numbers of examined relative pairs regulation of complex phenotypes, identify possible con- including at least one diabetic participant stellations of genes underlying suites of complex pheno- types, and when major locus effects have been detected, Relationship Number of relative pairs generate tentative linkage hypotheses.

Parent–offspring 429 Application to SHFS data Siblings 350 To maximize the power to detect genetic correlations, our Half-siblings 58 analyses used combined data from all three centers. We used Grandparent–grandchild 154 Avuncular 1139 the covariates sex, age, age-by-sex interaction, age2, age2-by- Grand avuncular 416 sex interaction, and center, because these effects are First cousins 542 common to both diabetes susceptibility and the CVD risk First cousins once removed 663 factors examined here. No additional covariate adjustments Second cousins 68 Other 260 were made as previous studies have demonstrated a minimal reduction of the genetic signal for the traits examined here, Total relative pairs 4079 when additionally adjusted for BMI and smoking status.6 All

International Journal of Obesity Action of genes on diabetes status and CVD risk factors KE North et al 494 remaining variation in liability to diabetes in this population lipid-, clotting-, and blood pressure-related phenotypes and can be attributed to additive genetic factors. We also have implicated diabetes status as an important correlate.6 corrected for BMI and the heritability was reduced to 31% Moreover, we recently demonstrated that genetic effects on (h2 ¼ 0.3170.10, Po0.0001). Recognizing that there are obesity- and lipid-related CVD risk factors in the SHFS twice as many diabetic participants in Arizona as compared differed in diabetic and nondiabetic individuals.7 Motivated to Dakota and Oklahoma, we speculated that including the by these findings, we were interested to determine if a center effect might remove some of the genetic component significant genetic influence on diabetes status itself could be in diabetes status. When center was included as a covariate, identified, and whether there is evidence for the joint action diabetes status exhibited a residual heritability of 22% of genes on diabetes status and related quantitative CVD risk (h2 ¼ 0.2270.10, P ¼ 0.002). Nonetheless, it was important factors in the American Indians of the SHFS. In our previous to include center as a covariate in all analyses to obtain an analyses, only five traits (BMI, FAT, WHR, HDL-C, and ln estimate of between-center differences. When larger sample triglycerides) showed evidence for distinct genetic effects in sizes become available, independent analyses will be con- diabetic and nondiabetic individuals. However, high stan- ducted in each center. dard errors of parameter estimates were obtained for the three additional phenotypes, indicating low power of the Correlations between diabetes status and CVD risk sample. factors In this study, after the effects of age, sex, age-by-sex The descriptive statistics for -risk interaction, and center have been accounted for, we have factors in diabetics and nondiabetics are reported in demonstrated a moderate additive genetic effect on diabetes Table 2. The genetic and environmental correlations of status. To our knowledge, there are few published actual diabetes status with CVD risk factors are given in Table 3. heritability estimates for diabetes status8. Nonetheless, there These correlations vary widely, but all CVD risk factors is strong evidence that diabetes has strong genetic determi- display a significant genetic correlation with diabetes status. nants9–21. The identification of moderate genetic influences In total, three of eight CVD risk factors (HDL-C, triglycerides, on diabetes status in the American Indians of the SHFS is and fibrinogen) also show a significant environmental important, not only in its own right, but also because correlation with diabetes status. diabetes status has been implicated as the best predictor of As all eight CVD risk factors displayed significant genetic CVD.5 correlations with diabetes status, we briefly examined We also found evidence for common genetic effects on whether other CVD risk factors would also show significant diabetes status and eight obesity, lipid, clotting, and blood genetic correlations with diabetes status. For apolipoprotein pressure traits. In fact, such pleiotropic action of genes is B and LDL-C, no significant genetic correlation was detected predicted in highly coordinated systems, such as in CVD risk (data not shown). factors (eg, HDL-C and BMI). Although no prior studies have reported genetic correlations between diabetes status and these CVD traits, studies have consistently demonstrated an Discussion effect of diabetes on obesity,39,40 lipids,41–44 clotting Our previous analyses of American Indians from the SHFS traits45,46, and blood pressure measures.47 In the American have demonstrated significant heritabilities for obesity-, Indians of the SHFS, there is a high prevalence of diabetes. In

Table 2 A summary of descriptive statistics of cardiovascular disease risk factors presented by diabetes status: The Strong Heart Family Study

Diabetic Nondiabetic

Phenotype Mean s.d. Sample size Mean s.d. Sample size

Obesity BMI (kg/m2) 33.48 6.32 271 30.31 6.61 499 FAT (%) 38.35 8.76 263 34.82 10.19 499 WHR (ratio) 0.96 0.06 270 0.91 0.08 495

Lipidsa HDL-C (mg/dl) 39.23 11.04 266 44.73 12.93 492 Triglyceride (mg/d1) 170.04 107.44 265 115.83 62.27 499

Clotting Fibrinogen (mg/d1) 368.83 80.92 249 307.46 67.78 491 PAI-1 (ng/ml) 58.91 39.18 254 49.66 35.97 493

Hypertensionb SBP (mmHg) 128.42 16.71 178 119.34 12.62 479

aIndividuals taking cholesterol-lowering medications were not included in these analyses. bIndividuals taking antihypertensive medications were not included in these analyses. For all phenotypes, all outliers were removed prior to analysis.

International Journal of Obesity Action of genes on diabetes status and CVD risk factors KE North et al 495 75 Table 3 Genetic (rG) and environmental (rE) correlations between diabetes diabetes in men with normal glucose tolerance. In status and CVD risk factors among American Indians of the SHFS addition, lifestyle factors such as smoking and physical activity have been shown to affect lipid levels.25,76 CVD risk factor r r G E To maximize the number of individuals entering the analyses, our models included only age, age2, sex, and center BMI (kg/m2) 0.55 (0.14)*** 0.10 (0.13) FAT (%) 0.38 (0.14)* 0.11 (0.14) as covariates. It would have been of interest to estimate the WHR (ratio) 0.58 (0.15)*** 0.13 (0.14) effect of diabetes duration; however, many individuals did a HDL-C (mg/dl) À0.37 (0.27)** À0.32 (0.18)** not provide information on diabetes duration and we were ln Triglyceride (mg/dl)a 0.65 (0.21)*** 0.33 (0.13)** unable to consider this effect. ln Fibrinogen (mg/dl) 0.40 (0.11)** 0.20 (0.11)*** PAI-1 (ng/ml) 0.67 (0.17)*** 0.18 (0.10) Given the multiple comparisons presented in Table 3, it SBP (mmHg)b 0.57 (0.21)** 0.11 (0.11) might be argued that a correction for multiple tests was needed. Using a Bonferroni correction for eight comparisons, a Individuals taking cholesterol-lowering medications were not included in a P-value of 0.05/8 ¼ 0.00625 would be required to ensure the these analyses. b conventional 5% level of statistical significance over all Individuals taking antihypertensive medications were not included in these 77 analyses. For all phenotypes, all outliers were removed prior to analysis. traits ). Using such an approach, only four of the eight E, Environmental; G, Genetic. Numerical values in parentheses are standard genetic correlations would have been considered statistically errors. significant (BMI, WHR, ln triglyceride, and PAI-1). However, Asterisks after numerical values of individual coefficients indicate significance this correction assumes that all of the comparisons are as follows: *(Pr0.05), **(Pr0.01), ***(Pr0.001). independent, which they are not in our study (eg, ln triglyceride and HDL-C). In summary, a significant heritability of diabetes status, the context of this genetic, metabolic, and neuroendocrine and several significant genetic and environmental interac- background, it may be that a common genetic effect in tions between diabetes status and eight CVD risk factors were diabetics, or a diabetes susceptibility gene, is an important demonstrated. We believe that this represents an important influence on variation in CVD risk factors. step in the understanding of the determinants of CVD risk Interestingly, six of the CVD risk factors that are indicative factors. These findings will be important for future research, of gene-by-gene interaction with diabetes status are compo- as statistical genetic models incorporating these interactions nents of the metabolic syndrome, a cluster of metabolic should better approximate the biological reality of the traits, abnormalities including central obesity, abnormal glucose and make it easier to detect, localize, and identify genes tolerance, elevated insulin, and triglycerides, depressed HDL- contributing to variation and covariation of CVD risk C, and .48–50 Previous studies have implicated a factors, and to measure their effects. These results are a first common underlying genetic factor influencing the develop- step in the search for CVD risk factor genes in American ment of the syndrome50–55, and the suggestion of pleiotropy Indians. Future research will determine the chromosomal in diabetes status and some metabolic syndrome traits, is location of CVD risk factors genes and ultimately, functional consistent with those findings. Future research will explore polymorphisms associated with the variability will be this relationship more closely by examining the clustering of identified. the metabolic syndrome traits, and using these clusters in the calculation of heritabilities and in linkage analysis. Three traits, HDL-C, triglycerides, and fibrinogen, dis- Acknowledgements played significant environmental correlations with diabetes We thank the Strong Heart Family Study participants. status. These findings are suggestive of common unmeasured Without their participation, this project would not have residual effects (for example, diet and/or physical activity) been possible. In addition, the cooperation of the Indian on diabetes status and lipid and clotting levels. Although no Health Service hospitals and clinics, and the directors of the other studies have reported an environmental correlation Strong Heart Study clinics, Betty Jarvis, Marcia O’Leary, Dr between diabetes status and lipid or clotting measures, Tauqeer Ali, Alan Crawford, and the many collaborators and relations between diabetes and lipid levels56–58 and between staff of the Strong Heart Study have made this project diabetes and clotting variables59,60 have been consistently possible. We also thank Drs John Blangero, Laura Almasy, demonstrated. Lifestyle factors such as smoking, alcohol Tony Comuzzie, and Lisa Martin for their generous con- consumption, diet, and physical activity may also be tribution to this research. This research was conducted while implicated in these findings, as the residual variance can author (KEN) was a postdoctoral fellow at the Southwest encompass any unmeasured environmental effect. Addition- Foundation for Biomedical Research and was funded by a ally, studies have consistently demonstrated an effect of cooperative agreement that includes Grants U01 HL65520, smoking, alcohol consumption, diet, and physical activity U01 HL41642, U01 HL41652, U01 HL41654, U01 HL65521 on lipid levels58,61–67 and on fibrinogen levels.68–74 Within from the National Heart, Lung, and Blood Institute. The the SHS populations, elevated triglyceride concentrations views expressed in this paper are those of the authors and do were significantly associated with the development of not necessarily reflect those of the Indian Health Service.

International Journal of Obesity Action of genes on diabetes status and CVD risk factors KE North et al 496 References 18 Knowler WC, Pettitt DL, Savage PJ, Bennett PH. Diabetes 1 Carr ME. Diabetes mellitus: a hypercoagulable state. J Diabetes incidence in Pima Indians: contributions of obesity and parental Am J Epidemiol Complicat 2001; 15: 44–54. diabetics. 1981; 113: 144–156. 2 Galloway JM, Alpert JS. Coronary artery disease and hyperten- 19 Lee ET, Anderson PS, Bryan J. Diabetes, parental diabetes, and sion. In: Galloway JM, Goldberg BW, Alpert JS (eds). Primary care obesity in Oklahoma Indians. Diabetes Care 1985; 8: 107–113. 20 Warram JH, Martin BC, Soeldner JS, Krolewski AS. Study of of Native American patients: diagnosis, therapy, and epidemiology. Woburn, MA: Butterwoth-Heinemann, 1999, pp 125–131. glucose removal rate and first phase insulin secretion in the 3 Welty TK, Rhoades D, Lee ET, Yeh F, Cowan LD, Fabsitz R, Robbins offspring of two parents with NIDDM. In: Camerini-Davalos RA, DC, Devereux RB, Henderson JA, Crawford AS, Howard BV. Cole HS (eds). Adv Exp Med Biol 1988; 246: 175–183. Changes in cardiovascular risk factors among American Indians: 21 Hsueh W, Mitchell BD, Aburomia R, Pollin T, Sakul H, Ehm MG, the strong heart study. Ann Epidemiol 2002; 12: 97–106. Michelsen BK, Wagner MJ, St Jean PL, Knowler WC, Burns DK, 4 Lee ET, Cowan LD, Welty TK, Sievers M, Howard WJ, Oopik A, Bell CJ, Shuldiner AR. Diabetes in the old order Amish. Wang W, Yeh J, Devereux RB, Rhoades ER, Fabsitz RR, Go OT, Characterization and heritability analysis of the Amish Diabetes Howard BV. All-cause mortality and cardiovascular disease Study. Diabetes Care 2000; 23: 595–601. mortality in three American Indian populations, aged 45–74 22 Duggirala R, Williams JT, Williams-Blangero S, Blangero J. A years, 1984–88: the strong heart study. Am J Epidemiol 1998; 147: variance component approach to dichotomous trait linkage 995–1008. analysis using a threshold model. Genet Epidemiol 1997; 14: 5 Howard BV, Lee ET, Cowan LD, Devereux RB, Galloway JM, Go 987–992. OT, Howard WJ, Rhoades ER, Robbins DC, Sievers ML, Welty TK. 23 Williams JT, Van Eerdewegh P, Almasy L, Blangero J. Joint Rising tide of cardiovascular disease in American Indians. The multipoint analysis of multivariate qualitative and quantitative strong heart study. Circulation 1999; 99: 2389–2395. traits. I. Likelihood formation and simulation results. Am J Hum 6 North KE, Howard BV, Best L, Lee ET, Fabsitz RR, MacCluer JW. Genet 1999; 65: 1134–1147. Genetic and environmental contributions to cardiovascular 24 Lee ET, Welty TK, Fabsitz RR, Cowan LD, Le N, Oopik AV, disease risk factors in American Indians: the strong heart family Cucchiara AJ, Savage PJ, Howard BV. The Strong Heart Study A study. Am J Epidemiol 157: 303–314. study of cardiovascular disease in American Indians: design and 7 North KE, MacCluer JW, Best L, Welty TK, Lee ET, Fabsitz RR, methods. Am J Epidemiol 1990; 132: 1141–1155. Howard BV. Evidence for distinct genetic effects on obesity and 25 Howard BV, Lee ET, Cowan L, Fabsitz RR, Howard WJ, Oopik AJ, lipid related CVD risk factors in diabetic and non-diabetic Robbins DC, Savage PJ, Yeh JL, Welty TK. Coronary heart disease American Indians: the strong heart family study. Diabetes Metab prevalence and its relation to heart disease in American Indians: Res Rev 2003, (in press). the Strong Heart Study. Am J Epidemiol 1995; 142: 254–268. 8 Duggirala R, Blangero J, Almasy L, Dyer TD, Williams KL, 26 Liedtke RJ. Principles of bioelectric independence analysis. 1997, Leach RJ, O’Connell P, Stern MP. Linkage of www.rjlsystems.com/research/bia-principles.html. mellitus and age on onset to a genetic location on chromo- 27 Howard BV, Welty TK, Fabsitz RR, Cowan LD, Oopik AJ, Le NA, some 10q in Mexican Americans. Am J Hum Genet 1999; 64: Yeh J, Savage PJ, Lee ET. Risk factors for coronary heart disease in 1127–1140. diabetic and non-diabetic Native Americans: The Strong Heart 9 Lee ET, Howard BV, Savage PJ, Cowan LD, Fabsitz RR, Oopik AJ, Study. Diabetes 1992; 41(Suppl 2): 4–11. Yeh J, Go O, Robbins DC, Welty TK. Diabetes and impaired 28 Friedwald WT, Levy RI, Fredrickson DS. Estimation of the glucose tolerance in three American Indian populations aged concentration of low-density lipoprotein cholesterol in plasma 45-74 years. Diabetes Care 1995; 18: 599–610. without the use of the preparative ultracentrifuge. Clin Chem 10 Diehl AK, Stern MP. Special health problems of Mexican 1972; 18: 499–502. Americans: obesity, gallbladder disease, diabetes mellitus, and 29 Geffken D, Keating F, Kennedy M, Cornell E, Bovill E, Tracy R. cardiovascular disease. Adv Intern Med 1989; 34: 73–96. The measurement of fibrinogen in population-based research: 11 Chakraborty R, Ferrell RE, Stern MP, Haffner SM, Hazuda HP, Studies on instrumentation and methodology. Arch Pathol Lab Rosenthal M. Relationship of prevalence of non-insulin depen- Med 1994; 118: 1106–1109. dent diabetes mellitus to Amerindian admixture in the Mexican 30 DeClerck PJ, Alessi MC, Verstreken M, Kruithof EK, Juhan-Vague Americans of San Antonio, Texas. Genet Epidemiol 1986; 3: I. Measurement of plasminogen activator inhibitor 1 in biological 435–454. fluids with a murine monoclonal antibody-based enzyme-linked 12 Brosseau JD, Eelkema RC, Crawford AC, Abe TA. Diabetes among immunosorbent assay. Blood 1988; 71: 220–225. the three affiliated tribes: correlations with degree of Indian 31 Alberti KG, Zimmet PZ. Definition, diagnosis, and classification inheritance. Am J Public Health 1979; 69: 1277–1278. of diabetes mellitus and its complications. Part 1: diagnosis and 13 Knowler WC, Williams RC, Pettit DJ, Steinberg AG. Gm3,5,13,14 classification of diabetes mellitus provisional report of a WHO and type 2 diabetes mellitus: an association in American consultation. Diabetes Med 1998; 15: 539–553. Indians with genetic admixture. Am J Hum Genet 1988; 43: 32 Falconer DS. Introduction to quantitative genetics, 3rd edn.: Long- 520–526. man Essex, UK;1989 . 14 Barnett AH, Eff C, Lesline RDG, Pyke AA. Diabetes in identical 33 Hopper JL, Mathews JD. Extensions to multivariate normal twins: a study of 200 pairs. Diabetologia 1981; 20: 87–93. models for pedigree analysis. Ann Hum Genet 1982; 46: 373–383. 15 Newman B, Sekby JV, King MC, Slemenda C, Fabsitz R, Friedman 34 Lange K, Boehnke M. Extensions to pedigree analysis IV. GD. Concordance for type II (NIDDM) in male twins. Diabetologia Covariance components models for multivariate traits. Am J 1987; 30: 763–768. Med Genet 1983; 35: 816–826. 16 Kaprio J, Tuomilehto J, Koskenvou M. Concordance for type 1 35 Amos CI. Robust variance-components approach for assessing (insulin dependent) and type 2 (non-insulin-dependent) diabetes genetic linkage in pedigrees. Am J of Hum Genet 1994; 54: mellitus in a population-based cohort of twins in Finland. 535–543. Diabetologia 1992; 30: 763–768. 36 Almasy L, Blangero J. Multipoint quantitative-trait linkage 17 Rewers M, Hamman RF. Risk factors for NIDDM. National analysis in general pedigrees. Am J Hum Genet 1998; 62: Diabetes Data Group. Diabetes in America, 2nd edn. National 1198–1211. Institutes of Health, National Institute of Diabetes and 37 Self SG, Liang KY. Asymptotic properties of maximum likelihood Digestive and Kidney , Bethesda, MD. NIH Publication ratio tests under nonstandard conditions. J Am Statist Assoc 1987; no. 95-1468; 1995. 82: 605–610.

International Journal of Obesity Action of genes on diabetes status and CVD risk factors KE North et al 497 38 Blangero J, Williams JT, Almasy L. Robust LOD scores for variance Association of fibrinogen with cardiovascular disease risk factors component-based linkage analysis. Genet Epidemiol 2000; and cardiovascular disease in the Framingham Offspring Popula- 19(Suppl 1): S8–S14. tion. Circulation 2000; 102: 1634–1638. 39 Bjorntorp P, Rosmond R. Visceral obesity and diabetes. Drugs 60 Lam TH, Liu LJ, Janus ED, Lam KS, Hedley AJ. Fibrinogen, Suppl 1999;1: 13–18. other cardiovascular risk factors and diabetes mellitus in Hong 40 Seidell JC. Obesity, insulin resistance and diabetesFa worldwide Kong: a community with high prevalence of Type 2 diabetes epidemic. Br J Nutr 2000; 83(Suppl 1): S5–S8. mellitus and impaired glucose tolerance. Diabetes Med 2000; 17: 41 Reaven GM. Non-insulin-dependent diabetes mellitus, abnormal 798–806. lipoprotein metabolism, and atherosclerosis. Metabolism 1987; 61 Blankenhorn DH, Nessim SA, Johnson RL, Sanmarro ME, Azen SP, 36:(Suppl 1): 1–8. Cashin-Hemphill L. Beneficial effects of combined colestipol- 42 Howard BV, Howard WJ. Dyslipidemia in non-insulin-dependent niacin therapy on coronary atherosclerosis and coronary venous diabetes mellitus. Endocrine Rev 1994; 15: 263–274. bypass grafts. JAMA 1987; 257: 3233–3240. 43 Larsen JL. Dyslipidemia in diabetes: a comprehensive analysis. 62 MacAuley D, McCrumm EE, Stott G, Evans AE, Duly E, Trinick T, Adv Nurse Pract 1998; 6: 36–43. Sweeney K, Boreham CA. Physical activity, lipids, apolipopro- 44 Haffner SM. Management of dyslipidemia in adults with diabetes. teins, and Lp(a) in the Northern Ireland Health and Activity Diabetes Care 1998; 21: 160–178. Survey. Med Sci Sports Exerc 1996; 28: 720–736. 45 Fujii S, Goto D, Zaman T, Ishimori N, Watano K, Kaneko T, Okada 63 Gensini GF, Comeglio M, Colella A. Classical risk factors and H, Makiguchi M, Nakagawa T, Kitabatake A. Diminished emerging elements in the risk profile for coronary artery disease. fibrinolysis and thrombosis: clinical implications for accelerated atherosclerosis. J Atheroscler Thromb 1998; 5: 76–81. Eur Heart J 1998; 19(Suppl A): A53–A61. 46 Bastard JP, Pieroni L, Hainque B. Relationship between plasma 64 Haffner SM. Diabetes, hyperlipidemia, and coronary artery plasminogen activator inhibitor 1 and insulin resistance. Diabetes disease. Am J Cardiol 1999; 83: 17F–21F. Metab Res Rev 2000; 16: 192–201. 65 Wagner AM, Perez A, Calvo F, Bonet R, Castellvi A, Ordonez, J. 47 Hall JE, Brands MW, Henegar JR. Mechanisms of hypertension Apolipoprotein (B) identifies dyslipidemic phenotypes associated and kidney diseases in obesity. Ann NY Acad Sci 1999; 892: with cardiovascular risk in normocholesterolemic type 2 diabetes 91–107. patients. Diabetes Care 1999; 22: 812–817. 48 Ferrannini E, Haffner SM, Mitchell BD, Stern MP. Hyperinsuline- 66 Luc G, Bard J, Evans A, Arveiler D, Ruidavets J, Amouyel P, mia: the key feature of a cardiovascular and metabolic syndrome. Ducimetiere P. The relationship between apolipoprotein AI- Diabetologia 1991; 34: 416–422. containing lipoprotein fractions and environmental factors: the 49 Despres JP, Lamarche B, Mauriege P, Cantin B, Lupien PJ, Dagena PRIME Study. Atherosclerosis 2000; 152: 399–405. GR. Hyperinsulinemia as an independent risk factor for ischemic 67 Imamura H, Uchida K, Kobata D. Relationship of cigarette heart disease. N Engl J Med 1996; 334: 952–957. smoking with blood pressure, serum lipids, and lipoproteins in 50 Mitchell BD, Kammerer CK, Mahaney MC, Blangero J, Comuzzie young Japanese women. Clin Exp Pharmacol Physiol 2000; 27: AG, Atwood LD, Haffner SM, Stern MP, MacCluer JW. Genetic 364–369. analysis of the IRS pleiotropic effects of genes influencing insulin 68 Tuut M, Hense H. Smoking, other risk factors and fibrinogen levels on lipoprotein and obesity measures. Arterioscler Thromb levels. Evidence of effect modification. Ann Epidemiol 2001; 11: Vasc Biol 1996; 16: 281–288. 232–238. 51 Carmelli, Cardon LR, Fabsitz R. Clustering of hypertension, 69 Rimm EB, Williams P, Fosher K, Criqui M, Stampfer MJ. Moderate diabetes, and obesity in adult male twins: Same genes or same alcohol intake and lower risk of coronary heart disease: meta- environments? Am J Hum Genet 1994; 55: 566–573. analysis of effects on lipids and haemostatic factors. Br Med J 52 Edwards KL, Newman B, Mayer E, Selby JV, Krauss RM, Austin 1999; 319: 1523–1528. MA. Heritability of factors of the insulin resistance syndrome in 70 Mennen LI, Balkau B, Vol S, Caces E, Eshwege E. Fibrinogen: a women twins. Genet Epidiol 1997; 14: 241–253. possible link between alcohol consumption and cardiovascular 53 Duggirala R, Williams K, Arya R, Blangero J, Stern MP. Genetic disease. Arterioscler Thromb Vasc Biol 1999; 19: 887–892. factors underlying the Insulin Resistance Syndrome in nondia- 71 James S, Vorster HH, Venter CS, Kruger HS, Nell TA, Veldman FJ, betic Mexican Americans. Am J Hum Biol 2000; 12: 274. Ubbink JB. Nutritional status influences plasma fibrinogen 54 Mahaney MC, Blangero J, Comuzzie AG, Van de Berg JL, Stern concentration: Evidence from the THUSA Survey. Thromb Res MP, MacCluer JW. Plasma HDL cholesterol, triglycerides, and 2000; 98: 383–394. adiposity: a quantitative test of the cojoint trait hypothesis in the 72 Di Minno G, Piemontino U, Cerbone AM. Nutrition and San Antonio Family Heart Study. Circulation 1995; 92: 3240–3248. thrombogenic factors. Nutr Metab Cardiovasc Dis 1999; 9(4 Suppl): 55 Masuzaki H, Paterson J, Shinyama H, Morton NM, Mullins JJ, 16–20. Seckl JR, Flier JS. A transgenic model of visceral obesity and the 73 Koenig W, Ernst E. Exercise and thrombosis. Coronary Artery Dis metabolic syndrome. Science 2001; 294: 2166–2170. 56 Schmidt MI, Watson RL, Duncan BB, Metcalf P, Brancati FL, 2000; 11: 123–127. Sharrett AR, Davis CE, Heiss G. Clustering of dyslipidemia, 74 Vanninen E, Laitinen J, Uusitupa M. Physical activity and hyperuricemia, diabetes, and hypertension and its association fibrinogen concentration in newly diagnosed NIDDM. Diabetes with fasting insulin and overall obesity in a general population. Care 1994; 17: 1031–1038. Atherosclerosis risk in communities study investigators. Metabo- 75 Lee ET, Welty TK, Cowan LD, Rhoades DA, Devereux R, Go O, lism 1996; 45: 699–706. Fabsitz R, Wang W, Howard BV. Incidence of diabetes mellitus in 57 Garvey WT, Hermayer KL. Clinical implications of the insulin American Indians of three geographic areas: the Strong Heart resistance syndrome. Clin Cornerstone 1998; 1: 13–28. Study. Diabetes Care 2002; 25: 49–54. 58 Tenkanen L, Manttari M, Manninen V. Some coronary risk factors 76 Yurgalevitch SM, Kriska AM, Welty TK, Go O, Robbins DC, related to the insulin resistance syndrome and treatment with Howard BV. Physical activity and lipids and lipoproteins in gemfibrozil. Experience from the Helsinki Heart Study. Circulation American Indians ages 45–74. Med Sci Sports Exerc 1998; 30: 543– 1995; 92: 1779–1785. 549. 59 Stec JJ, Silberashatz H, Tofler GH, Matheney TH, Sutherland P, 77 Wallenstein S, Zucker C, Fleiss J. Some statistical methods useful Lipinska I, Massaro JM, Wilson PFW, Muller JE, D’Agostino RB. in circulation research. Circ Res 1980; 47: 1–9.

International Journal of Obesity