Donate Help Contact The AHA Sign In Home
American Heart Association
Circulation
Search: search_blue_button Advanced Search
« Previous Article | Table of Contents | Next Article »
Circulation. 1996;94:2159-2170

This Article
Right arrow Abstract Freely available
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow Request Permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Mitchell, B. D.
Right arrow Articles by MacCluer, J. W.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Mitchell, B. D.
Right arrow Articles by MacCluer, J. W.

(Circulation. 1996;94:2159-2170.)
© 1996 American Heart Association, Inc.

Articles

Genetic and Environmental Contributions to Cardiovascular Risk Factors in Mexican Americans

The San Antonio Family Heart Study

Braxton D. Mitchell, PhD; Candace M. Kammerer, PhD; John Blangero, PhD; Michael C. Mahaney, PhD; David L. Rainwater, PhD; Bennett Dyke, PhD; James E. Hixson, PhD; Richard D. Henkel, PhD; R. Mark Sharp, PhD; Anthony G. Comuzzie, PhD; John L. VandeBerg, PhD; Michael P. Stern, MD; Jean W. MacCluer, PhD

the Department of Genetics (B.D.M., C.M.K., J.B., M.C.M., D.L.R., B.D., J.E.H., R.D.H., R.M.S., A.G.C., J.L.V., J.W.M.), Southwest Foundation for Biomedical Research, San Antonio, Tex; and the Department of Medicine/Epidemiology (M.P.S.), University of Texas Health Science Center (San Antonio).

Correspondence to Braxton D. Mitchell, PhD, Department of Genetics, Southwest Foundation for Biomedical Research, PO Box 760549, San Antonio, TX 78245-0549. E-mail bmitchel@darwin.sfbr.org.


*    Abstract
up arrowTop
 *Abstract
down arrowIntroduction
 down arrowMethods
 down arrowResults
 down arrowDiscussion
 down arrowReferences
 
Background The familial aggregation of coronary heart disease can be in large part accounted for by a clustering of cardiovascular disease risk factors. To elucidate the determinants of cardiovascular disease, many epidemiological studies have focused on the behavioral and lifestyle determinants of these risk factors, whereas others have examined whether specific candidate genes influence quantitative variation in these phenotypes.

Methods and Results Among Mexican Americans from San Antonio (Tex), we quantified the relative contributions of both genetic and environmental influences to a large panel of cardiovascular risk factors, including serum levels of lipids, lipoproteins, glucose, hormones, adiposity, and blood pressure. Members of 42 extended families were studied, including 1236 first-, second-, and third-degree relatives of randomly ascertained probands and their spouses. In addition to the phenotypic assessments, information was obtained regarding usual dietary and physical activity patterns, medication use, smoking habits, alcohol consumption, and other lifestyle behaviors and medical factors. Maximum likelihood methods were used to partition the variance of each phenotype into components attributable to the measured covariates, additive genetic effects (heritability), household effects, and an unmeasured environmental residual. For the lipid and lipoprotein phenotypes, age, gender, and other environmental covariates accounted in general for <15% of the total phenotypic variance, whereas genes accounted for 30% to 45% of the phenotypic variation. Similarly, genes accounted for 15% to 30% of the phenotypic variation in measures of glucose, hormones, adiposity, and blood pressure.

Conclusions These results highlight the importance of considering genetic factors in studies of risk factors for cardiovascular disease.


Key Words: epidemiology • genetics • risk factors • lipoproteins • lifestyle


*    Introduction
up arrowTop
 up arrowAbstract
 *Introduction
down arrowMethods
 down arrowResults
 down arrowDiscussion
 down arrowReferences
 
The familial aggregation of coronary heart disease can be accounted for in large part by a clustering of cardiovascular disease risk factors. Prospective epidemiological studies have revealed that a substantial portion of the variation in CHD risk within populations can be attributed to variation in age, gender, obesity, glucose tolerance status, blood pressure, and lipid and lipoprotein levels.1 2 One strategy for identifying the genes that influence CHD risk is to identify the genes that influence the intermediate traits associated with atherosclerosis, ie, traits such as dyslipidemia, obesity, glucose intolerance, and hypertension. Such investigations must account for the fact that these intermediate traits themselves are multifactorial in nature, being influenced both by genes and environmental factors.

The SAFHS was designed to investigate the genetics of heart disease and its risk factors in Mexican Americans. A basic premise of the SAFHS is that the genes that influence normal variation in physiological traits are likely to contribute significantly to CHD risk. Families participating in the SAFHS were therefore selected to be representative of the underlying San Antonio community. The goals of this report are twofold: first, to describe the basic design of the SAFHS; and second, to summarize the overall genetic and environmental contributions to phenotypic variation for a large panel of cardiovascular risk factors.


*    Methods
up arrowTop
 up arrowAbstract
 up arrowIntroduction
 *Methods
down arrowResults
 down arrowDiscussion
 down arrowReferences
 
Enrollment of Probands and Family Members
Data collection for the SAFHS began in January 1992, and by June 1995, a total of 1236 individuals from 42 extended families had been enrolled. Probands were identified from the community at large in several stages. First, a single census tract encompassing a low-income neighborhood of San Antonio was selected for study. Second, all residential addresses within this census tract were identified from the Polk directory.3 Third, households within this tract were approached in random order to determine whether any resident within the household met the study eligibility requirements for a proband. To be eligible, the proband had to be 40 to 60 years old, to have a spouse who was willing to participate, and to have at least six offspring and/or siblings >=16 years old who were residing in San Antonio. Once a proband was enrolled, all first-degree (eg, parents, siblings, and offspring), second-degree (eg, half-siblings, grandparents/grandchildren, aunts/uncles, nieces/nephews, and double cousins), and third-degree (eg, first cousins, great-uncles/great-aunts, and great-nephews/great-nieces) relatives of the proband and of the proband's spouse who were >=16 years old were invited to participate. Spouses of these relatives were also invited to participate.

Probands and participating family members received a medical examination in our clinic. Pregnant women were not eligible to participate; women reporting that they were pregnant were recontacted after their pregnancy to reschedule their examination. Occasionally, examinations were conducted in the participant's home. Examinations were conducted in the morning, after the participant underwent a 12-hour fast. The medical examination included an anthropometric assessment, a glucose tolerance test, and an interview component. All procedures were approved by the institutional review board, and all subjects gave informed consent.

Phenotypic Assessments
An anthropometric assessment was conducted in which height and weight were measured. Additional measurements of adiposity were also obtained, including a series of skinfold thicknesses, measured with Lange calipers, and circumferences, measured with a steel tape measure. These measures were used to derive several different indexes of adiposity, including body mass index (weight in kilograms divided by height in meters squared), ratio of subscapular to triceps skinfold, and ratio of waist to hip circumference. Body composition was also assessed by bioimpedance. The systolic (first phase) and diastolic (fifth phase) blood pressures were measured to the nearest even digit using a random-zero sphygmomanometer (Hawksley-Gelman) on the right arm of the seated participant. Three readings were recorded for each individual, and the subject's blood pressure was defined as the average of the second and third readings.

Subjects were also administered a 75-g glucose equivalent load, and glucose and insulin levels were determined from a blood sample obtained 2 hours after the glucose challenge. Diabetes was diagnosed according to the World Health Organization plasma glucose criteria.4 Subjects were also considered to have diabetes if they were currently taking antidiabetic medication.

Biochemical Methods
Fasting blood samples were drawn for measurement of lipids and lipoproteins, glucose and insulin, and gender hormones. Serum samples were obtained from whole blood after clotting. Plasma samples were obtained from whole blood collected in disodium EDTA (or fluoride tubes for glucose) and kept on wet ice until separation of plasma and cell components. Plasma and cells were separated by centrifugation at 800g for 10 minutes. The resulting plasma supernatant was transferred to Tygon tubing, which was heat-crimped into sections containing {approx}50 µL as described previously5 and stored at -75°C. This method provides multiple samples of serum that are thawed only once and also prevents desiccation and oxidation of samples during prolonged storage.

Total plasma cholesterol, HDL-C, HDL3-C, and triglycerides were directly measured. Cholesterol and triglyceride concentrations were assayed enzymatically with commercial reagents supplied by Boehringer-Mannheim Diagnostics and Stanbio, respectively. HDL-C was measured after precipitation of apoB-containing particles from thawed plasma by the use of dextran sulfate-Mg2+.6 Levels of HDL3-C were obtained using a dual-precipitation method,7 and levels of HDL1+2-C were derived from these measures by subtraction. Levels of LDL-C were derived by subtracting HDL-C and VLDL-C (estimated as one fifth of total triglyceride levels) from total cholesterol.8 LDL-C concentrations were computed only for those individuals with triglyceride levels of <400 mg/dL. The interassay coefficients of variation for control samples in these assays were 2.1% for total plasma cholesterol, 4.6% for HDL-C, 5.4% for HDL3-C, and 3.7% for triglycerides.

Lp(a) concentrations were measured with a commercial assay kit (MacraLp[a] ELISA kit, Strategic Diagnostics) with an EL340 microplate reader (Bio-Tek). Apolipoprotein concentrations were measured in a commercial laboratory (Medical Research Laboratories); apoB and apoAI were determined by the use of nephelometry,9 10 11 apoAII and apoE were determined by competitive immunoassays,12 13 and LpAI was determined by electroimmunoassay.14 LCAT activity was measured using an exogenous substrate system as previously described.15 The interassay coefficients of variation were 4.3% for Lp(a), 3.5% for apoAI, 4.4% for apoAII, 2.9% for apoB, 8.1% for apoE, and 6.8% for LpAI. Four control sera were used for the LCAT activity assay, with coefficients of variation ranging from 2.9% to 4.3%.

Plasma glucose was measured using an Abbott V/P Analyzer, and serum concentrations of insulin and DHEAS were measured using commercial radioimmunoassay kits (Diagnostic Products Corp). Serum concentrations of SHBG were measured using a commercial double-antibody system (Diagnostic Systems Laboratory). The interassay coefficients of variation were 7.9% for DHEAS and 9.0% for SHBG. The coefficient of variation between duplicate aliquots measured in a single laboratory run was 6.5% for the fasting and 8.0% for the 2-hour insulin concentrations.

Interviews
Interviews were administered to determine social, behavioral, and lifestyle factors that might be associated with cardiovascular risk. Information was also obtained regarding the individual's past medical history, educational background, household income level, reproductive history, and smoking and alcohol consumption patterns. Women were classified as postmenopausal if they reported having had both ovaries removed or if >=12 months had elapsed since their last menstrual period.

A food frequency questionnaire was administered to obtain information about usual dietary intake. We developed our food frequency questionnaire, following the approach of Willett et al,16 specifically for use in this Mexican American population.17 The questionnaire consisted of 102 different food items, chosen because in aggregate, they accounted for 80% to 85% of the consumption of the following nutrients: total calories, total protein, total fat, and total carbohydrates. By applying nutrient values obtained from the nutrient tables provided by the US Department of Agriculture18 19 20 and other sources, we were able to compute the daily consumption of the specified nutrients (in g/d). The daily consumption of fat, carbohydrates, and protein can also be expressed as a percentage of total daily caloric consumption.

Physical activity was assessed using a modified version of the Stanford 7-Day Physical Activity Recall Instrument.21 22 Subjects reported the weekly number of hours they sleep and engage in moderately strenuous, heavy, and very heavy physical activities. Examples of activities corresponding to each category are provided to assist the subject's responses. Light physical activity is defined as the difference between the total possible hours of weekly activity (ie, 7 daysx24 h/d=168 hours) and the number of hours accounted for by sleep and moderate, heavy, and very heavy activity. Each category of physical activity is scored in weekly METs, in which 1 MET equals the energy expenditure of 1 kg/body wt per hour or an oxygen uptake of 3.5 mL/kg per minute.

Statistical Methods
The overall aim of these analyses was to determine the extent to which genes, measured environmental factors, and household factors contribute to variation in a large panel of CHD-related traits. Information on the covariance among relatives was used to estimate the polygenic (or additive genetic) component of variance.

The level of trait, y, for individual i (yi) was modeled as follows23 :

where µ is the mean of the trait in males, and ßj is the regression coefficient for covariate j. The quantitative covariates for each individual were scaled by subtracting out the mean covariate value so that the regression coefficient represents the effect associated with a 1-unit deviation from the mean level of the covariate in the population. Measures of the qualitative covariates (eg, female gender, diabetes, medication use, menopausal status, smoking status, and so on) were scaled so that the regression coefficient represents the effect of having the covariate present compared with having it absent. The remaining parameters, gi, hi, and ei, represent the deviations from µ for individual i that are attributable to the residual genetic, household, and unmeasured environmental effects, respectively. Within each pedigree, a unique household number was assigned to all individuals with the same current address. The effects of gi, hi, and ei are assumed to be independent and normally distributed with mean zero and variance {sigma}g2, {sigma}h2, and {sigma}e2. The likelihood of the phenotypes of family members is assumed to follow a multivariate normal distribution with covariance matrix {Pi}, where {Pi} is a function of the coefficient of relatedness between individuals and the residual additive genetic, household, and environmental variances. Estimates of the mean and variance values were obtained using maximum likelihood methods.23

The significance of covariate effects on each phenotype was initially evaluated by considering a model in which phenotypic variation was influenced by the measured covariates and residual environmental factors only (ie, the genetic and household effects were constrained to be zero). All 22 covariates were allowed to enter the model. Robust standard errors were estimated, and the significance of each covariate was evaluated using a robust score test.24 Those covariates whose effects were significant at the .10 level were retained for all subsequent analyses, even if the significance levels decreased after inclusion of genetic and household effects in the model. We used this less-restrictive significance level to increase the probability of including all important covariates in the final models.

After the initial covariate analyses, the constraints on the genetic and household effects were removed, and maximum likelihood methods were used to estimate simultaneously the effects of the covariates, heredity, and household. The significance of the genetic and household effects was also evaluated using the score test. To test whether the set of environmental covariates included in each analysis accounted for a significant component of the phenotypic variation in that trait once the genetic and household effects were accounted for, the likelihood ratio test was performed. This test compares the likelihood of a full model (covariates and additive genetic and household effects) with that of a nested model (additive genetic and household effects only). The likelihood ratio statistic is distributed asymptotically as a {chi}2 statistic with df equal to the difference in number of parameters in the two models being compared.

The relative proportions of the variance explained by the measured environmental covariates, genes, and household were calculated as the variance attributable to that particular component divided by the total phenotypic variance. The residual variance that was not accounted for by the three components corresponds to the residual environmental variance or the proportion of the variance attributable to unmeasured environmental factors.

The analysis of each phenotype was restricted to those individuals for whom all covariate data were complete. Dietary interviews were not obtained from 212 individuals, including 154 who resided in Mexico. This left a total of 1024 individuals for whom dietary information and at least some of the phenotypic information were collected. Seventy-four of these individuals had missing information on one or more of the covariates and were therefore excluded from the subsequent analyses. Differences between the remaining 950 individuals and the sample sizes shown for the multivariate analyses represent the number of individuals for whom a particular phenotypic measurement was not obtained. In addition, lipoprotein phenotypes were not analyzed for individuals currently taking lipid-lowering medications (n=23), and blood pressures were not analyzed for 112 individuals currently taking antihypertensive medications. Insulin values were not analyzed for individuals with previously or newly diagnosed diabetes (n=188). Glucose values were not analyzed for diabetic subjects currently taking antidiabetic medications (n=108). Insulin, triglyceride, SHBG, and Lp(a) levels were log-transformed before analysis to reduce skewness; 0.1 mg/dL was added to each Lp(a) concentration before the transformation. Diabetes, use of diabetes medications, and duration of diabetes were not considered as potential covariates for the analyses of glucose and insulin but were considered in all other analyses.


*    Results
up arrowTop
 up arrowAbstract
 up arrowIntroduction
 up arrowMethods
 *Results
down arrowDiscussion
 down arrowReferences
 
The distribution of the 1236 examined individuals (494 men and 742 women) is shown in Table 1Down according to gender, number of independent households, and pedigree number. Individuals participating in this study resided in 655 separate households, 300 of which were multiplex households. There were a total of 413 full sibships, with sizes ranging from 1 to 11.


View this table:
[in this window]
[in a new window]
  		Table 1. Distribution of Examined Individuals According to Pedigree Number

Because we recruited extended families, the sample of examined individuals included a very large number of relative pair types. The sample included information on 2405 pairs of first-degree relatives (1082 parent-offspring pairs and 1323 sibling pairs), 2851 pairs of second-degree relatives (2393 avuncular pairs [aunt/uncle-niece/nephew], 353 grandparent-grandchild pairs, and 105 half-sibling pairs), and 3199 pairs of third-degree relatives (2317 cousin pairs, 552 great-avuncular pairs, 308 half-avuncular pairs, and 22 great-grandparent-great-grandchild pairs). A representative pedigree is shown in the FigureDown.



View larger version (12K):
[in this window]
[in a new window]
  		Figure 1. Representative pedigree from the San Antonio Family Heart Study.

The mean levels of covariates are shown by age and gender in Table 2Down. The prevalence of diabetes increased from <2% in individuals below the age of 30 to 40% to 45% in individuals >70. Mean caloric intake declined with age in both men and women. Approximately one half of the daily caloric intake was derived from carbohydrate, whereas slightly >30% was derived from fat. Approximately 30% to 35% of men and 12% to 15% of women in this population reported that they currently smoked. The mean number of formal years of education ranged from {approx}5 to 11 years. Approximately 20% of women <30 reported current use of oral contraceptives, and {approx}15% of women between the ages of 50 and 69 reported current use of estrogens.


View this table:
[in this window]
[in a new window]
  		Table 2. Levels of Covariates According to Age and Gender

Mean levels of the unadjusted quantitative phenotypes are shown in Table 3Down according to age and gender. Regression coefficients for the environmental covariates, estimated simultaneously along with the effects of heredity and household, are shown in Tables 4 through 6DownDownDown. For the lipids and lipoproteins (Table 4Down), age and/or gender effects were significant for every phenotype except LCAT activity. Diabetes or a related variable (eg, use of antidiabetic medications and/or duration of diabetes) was significantly associated with concentrations of HDL3-C, HDL1+2-C, triglycerides, apoAII, apoB, apoE, and LpAI. One or more of the dietary variables was significantly associated with concentrations of HDL1+2-C, triglycerides, apoAI, and LpAI. Current smoking was significantly associated with lower concentrations of HDL-C and HDL3-C, whereas alcohol intake was associated with significantly higher concentrations of HDL-C, HDL1+2-C, apoAI, apoAII, and LpAI. Use of oral contraceptives and/or postmenopausal estrogens was associated with significantly higher concentrations of total cholesterol, HDL3-C, triglycerides, apoAI, apoAII, apoB, and LpAI and with significantly lower concentrations of apoE and Lp(a).


View this table:
[in this window]
[in a new window]
  		Table 3. Levels of Cardiovascular Risk Factors According to Age and Gender


View this table:
[in this window]
[in a new window]
  		Table 4. Environmental Correlates of Lipid and Lipoprotein Levels


View this table:
[in this window]
[in a new window]
  		Table 5. Environmental Correlates of Glucose and Hormone Levels


View this table:
[in this window]
[in a new window]
  		Table 6. Environmental Correlates of Anthropometric Measures and Blood Pressure

The environmental correlates of glucose and hormone levels are shown in Table 5Up. Women had significantly higher concentrations of 2-hour glucose and insulin, SHBG, and DHEAS than did men. Concentrations of fasting and 2-hour glucose, 2-hour insulin, and SHBG increased with age, whereas concentrations of DHEAS decreased. Both fasting and 2-hour glucose concentrations were associated with a higher percentage of calories consumed as total protein. Higher physical activity levels were associated with lower concentrations of both fasting and 2-hour insulin, whereas smokers also had significantly lower concentrations of 2-hour insulin than did nonsmokers. Fasting insulin concentrations were significantly associated with total dietary cholesterol and saturated fat intake and with lower consumption of alcohol. Concentrations of SHBG and DHEAS were inversely associated with the presence of diabetes (or use of antidiabetic medications). SHBG concentrations were also higher in smokers and in oral contraceptive users, whereas DHEAS concentrations were higher among those with greater education and were lower among users of oral contraceptives and estrogens.

Table 6Up shows the environmental correlates of the adiposity measures and blood pressure. Women had significantly higher levels of body mass index but lower levels of waist-to-hip ratio, subscapular-to-triceps ratio, and systolic and diastolic blood pressures compared with men. Higher body mass index and waist-to-hip ratio were both significantly associated with the presence of diabetes and with higher total dietary cholesterol intake, whereas waist-to-hip ratio was also associated in women with the onset of menopause. Higher subscapular-to-triceps ratios were significantly associated with physical inactivity. Both systolic and diastolic blood pressures were significantly higher in diabetic than in nondiabetic subjects, whereas systolic blood pressure was also associated with higher total caloric intake.

Table 7Down shows the components of variance for the panel of cardiovascular risk factors. The columns in this table represent the number of individuals with complete data and the proportion of the total phenotypic variance attributable to the measured covariates, additive genetic effects, household effects, and residual environmental factors, respectively. For all of the phenotypes except SHBG, DHEAS, systolic blood pressure, and waist-to-hip ratio, a larger proportion of the phenotypic variance was attributable to genes than to the measured environmental risk factors, including gender and age. For most of the lipid and lipoprotein phenotypes, the proportion of the variance attributable to genes was at least three times as large as the proportion attributable to measured environmental factors. The heritabilities of all traits, with the lone exception of waist-to-hip ratio, were significantly greater than zero.


View this table:
[in this window]
[in a new window]
  		Table 7. Components of Variance for Selected Cardiovascular Risk Factors

The proportion of the total phenotypic variance accounted for by the measured covariates ranged from 1.4% (for Lp[a] concentrations) to 33% (for DHEAS concentrations). For the lipids and lipoproteins, this proportion was <17%. In contrast, the heritabilities for the lipids and lipoproteins ranged from 21% (for LCAT activity) to 69% (for Lp[a] concentrations), with genes accounting for 35% to 45% of the total phenotypic variability in most of these traits. The heritabilities of the glucose and hormone phenotypes were somewhat lower, ranging from 13% (for 2-hour insulin and SHBG concentrations) to 35% (for fasting insulin concentrations). Genes accounted for 18% and 28% of the total phenotypic variation in systolic and diastolic blood pressures, respectively, whereas the heritabilities of body mass index and subscapular-to-triceps ratio were 42% and 32%, respectively.

The proportion of the total phenotypic variability accounted for by household effects tended to be considerably smaller than that accounted for by the genetic effects. In fact, household effects were significant for only nine traits: HDL1+2-C, LDL-C, apoE, LpAI, LCAT activity, fasting insulin, systolic and diastolic blood pressures, and waist-to-hip ratio. Household membership accounted for {approx}5% to 12% of the total phenotypic variability in these traits.


*    Discussion
up arrowTop
 up arrowAbstract
 up arrowIntroduction
 up arrowMethods
 up arrowResults
 *Discussion
down arrowReferences
 
Previous studies have shown that the Mexican American population in San Antonio is characterized by a relatively adverse cardiovascular risk profile.25 The key features of this risk profile include obesity, dyslipidemia, glucose intolerance, and hyperinsulinemia. Compared with non-Hispanic whites in San Antonio, Mexican Americans also have higher mean systolic and diastolic blood pressures but, paradoxically, a lower prevalence of diagnosed hypertension.26 The prevalence of myocardial infarction appears to be lower in Mexican American men than in non-Hispanic white men,27 although whether this deficit reflects a lower incidence of cardiovascular disease or greater case fatality among those experiencing an infarction is controversial.

We have simultaneously estimated the contributions of genes and a variety of epidemiological covariates to phenotypic variability in levels of cardiovascular risk factors. Although numerous studies have attempted to distinguish between the effects of environmental and genetic factors on cardiovascular risk factors, none have considered the wide range of environmental covariates or analyzed the range of cardiovascular risk factors as described in this study. The covariates, genetic effects (heritability), and household effects together accounted for {approx}50% to 60% of the total phenotypic variability in most of the lipid/lipoprotein, blood pressure, and anthropometric phenotypes and 30% to 50% of the total phenotypic variability in glucose and hormones concentrations. Age, gender, and the other covariates together accounted for relatively little of the phenotypic variation in most of these traits (generally <15%, with a few exceptions), whereas variation attributable to genetic effects accounted for a relatively large amount (typically twofold to sixfold more than the measured environmental covariates). The further contribution of household effects toward accounting for variation in these phenotypes was marginal (range, 0% to 12%).

We estimated only one component of the total possible genetic effect on these phenotypes: specifically, the heritability. Heritabilities, which represent the additive effects of genes only, assume no interaction between alleles at a single locus (ie, no dominance effects) or between different loci (ie, no epistatic effects). Both dominance and epistatic effects frequently contribute to phenotypic variation, and more-detailed genetic analyses (eg, complex segregation analysis) are required to characterize these nonadditive genetic effects. Among SAFHS participants, we have already detected evidence for major genes with dominance effects that influence plasma concentrations of HDL-C,28 LDL-C,29 apoAI,30 apoB,29 fat mass by bioimpedance,31 and serum concentrations of 2-hour insulin.32

Direct comparisons of our results with those obtained from other studies are difficult. First, our sample included a wide range of different types of related individuals, and for this reason it provides greater power to discriminate between effects attributable to genes and those due to shared households than do samples based on nuclear families only. Second, the SAFHS study population consists of randomly ascertained families and thus permits inferences to be made about the relative importance of genetic and environmental factors in influencing phenotypic variability at the population level. Such inferences cannot be made from studies in which the probands are ascertained according to extreme values of some phenotype (eg, hypercholesterolemia, glucose intolerance, and so on). When families are ascertained according to "affected" probands, a small number of loci with relatively large effects may contribute disproportionately to the underlying phenotypic variability.

Because heritability represents that portion of the total phenotypic variance that is attributable to variation at the genetic level, the heritability of the same trait will differ between populations that differ in the distribution of environmental risk factors for that trait. For example, with all other factors being equal, the heritability of HDL-C concentrations will be higher in populations with a low frequency of alcohol consumption than in populations in which alcohol consumption varies widely.

Comparisons between studies must also take into account which covariates were included in the analysis. We considered a wide range of potential covariates, but we did not include other phenotypes with strong genetic determinants. For example, we did not adjust for the effects of body mass index on lipoprotein phenotypes because this adjustment would have had the effect of removing from the heritability any pleiotropic effects of obesity genes. Had we adjusted for body mass index, we would have observed a higher proportion of variance explained by measured covariates (at the expense of the overall genetic component of variance).

Finally, the heritability was expressed in our study as the proportion of the total (rather than residual) phenotypic variability. Thus, we estimated that 39% of the total phenotypic variance in serum cholesterol is attributable to additive genetic effects. An equivalent expression of this heritability is that 47%, or 0.392/(1-0.128-0.038), of the residual phenotypic variance can be attributed to additive genetic effects after accounting for the effects of household, age, gender, and other environmental covariates.

With these precautions in mind, it is worth noting that the heritabilities for many traits estimated in our study are consistent with those obtained by other researchers. For example, the heritability of total cholesterol in our population was 39%, which compares with estimates of 40% to 60% reported by others.33 34 35 36 37 38 39 40 41 Similarly, our estimates of the heritability of HDL-C, LDL-C, and triglycerides (46%, 40%, and 40%, respectively) are reasonably close to those estimated from other populations (40% to 60% for HDL-C,36 37 38 39 40 41 42 43 44 45 50% to 60% for LDL-C,37 38 39 40 41 44 and 35% to 50% for triglycerides35 36 38 39 44 ). Hasstedt et al42 estimated that genes accounted for 28% of the phenotypic variability in HDL3-C, whereas the heritability of this trait in our population was 38%. The heritability of apoAI concentrations in our population was 43%, which falls well within the range reported in several other studies,35 38 46 although at least two additional studies failed to find any evidence for genetic transmission of this trait.47 48 The estimated heritability of apoAII concentrations has ranged from 25% to 35%,38 47 48 which, again, is in close agreement with the estimated heritability of 34% in our population. Berg35 estimated that the heritability of apoB concentrations was 66% from a sample of twins, although in another study, Hamsten et al38 estimated a higher heritability in children (51%) than in their parents (14%), prompting them to speculate that apoB concentrations may be heavily influenced by exogenous factors in older individuals. In our study, the overall heritability for apoB concentrations was 31%, although we did not consider any age interactions in our analyses. Nearly all of the phenotypic variability in Lp(a) concentrations is reported to be due to genes,35 49 although the heritability was only 69% in our population.

The heritability of blood pressure estimated from previous studies has ranged from 15% to 40%,33 34 36 50 51 whereas we estimated heritabilities of 18% and 28% for systolic and diastolic blood pressure, respectively, in our population. A wide range in the heritability of body mass index has been reported, ranging from 25% to 60%,36 52 53 compared with 42% in our Mexican American population. We estimated the heritability of fasting glucose concentrations to be 18% compared with estimates of 25% to 30% reported elsewhere.54 55

Also consistent with prior studies, environmental risk factors accounted for a relatively small proportion of the phenotypic variation in most of our analyses. Associations have been reported previously between different measures of physical activity and a variety of cardiovascular risk factors, including lipid and lipoprotein levels, body mass index, and blood pressure,56 57 although in general the reported associations have been weak. In our study, physical activity was not independently associated with any of the above phenotypes, but it was significantly associated with a lower subscapular-to-triceps skinfold ratio and with lower serum insulin levels. This latter result is consistent with several previous studies suggesting a positive association between levels of physical activity and enhanced insulin sensitivity.58 59 60

Dietary variables were significantly associated with several of the phenotypes, although the magnitudes of these associations were modest. Willett61 argues that the reasons most studies observe relatively poor correlations between dietary intake and serum lipid concentrations are that endogenous sources of variability overwhelm exogenous sources and that the exogenous sources (ie, dietary intake) are very difficult to measure. As in other studies, alcohol consumption in our study was significantly associated with elevated concentrations of HDL-C and apoAI62 and with lower concentrations of fasting insulin.63 Alcohol consumption was also significantly associated with elevated concentrations of HDL1+2-C, apoAII, and LpAI. Current smokers had markedly lower concentrations of HDL-C than nonsmokers, as has been previously observed,64 and they had lower concentrations of HDL3-C and 2-hour insulin and higher concentrations of SHBG. Diabetes status was not significantly associated with Lp(a) concentrations in our analysis, although a modest association has been reported previously from the SAFHS once the effects of protein size variation were accounted for in the analysis.65

The extended family structures embedded within our sample enabled us to estimate household effects from the data independent of additive genetic effects. Household effects are attributable to unmeasured nongenetic factors, which are shared more closely by individuals living in the same households than by individuals living in different households. They may represent unmeasured dietary or other lifestyle factors. In our study, household effects were significantly greater than zero for nine phenotypes, although the estimated magnitude of these effects was small, ranging from 5% to 12% of the total phenotypic variation.

After accounting for covariate, genetic, and household effects, the proportion of the variance that remains unexplained for most of the phenotypes in this study ranged from 40% to 60%. For many of these traits, some of this unexplained phenotypic variance can likely be attributed to the nonlinear effects of environmental risk factors on specific genotypes. Although such genotypexenvironment effects can be difficult to detect, these interactions may nevertheless profoundly influence many of these phenotypes.

It is likely that we have underestimated the effects of some of the environmental covariates. For example, the relatively crude instruments that we used to assess dietary intake and physical activity may underestimate considerably the "true" effects of these variables on cardiovascular risk. We also have not considered possible interactions among the measured environmental factors, such as between diabetes and gender. The likely effect of these shortcomings is to underattribute variance to the measured environmental factors and overattribute variance to the unmeasured (residual) environmental factors. In addition, there may remain important environmental determinants of cardiovascular risk that we have not yet attempted to measure.

The results of these analyses are being used to guide our ongoing investigations into the determinants of cardiovascular risk factors. The inclusion of environmental covariates in the analysis will improve our power to detect genetic linkage. By accounting for the known effects of measured covariates on the phenotypic variance, the genetic signal in the remaining "unexplained" portion of the variance will be strengthened. Conversely, the power of epidemiological analyses to detect environmental risk factors should be strengthened by accounting for variability attributable to genes. By considering both genes and environmental risk factors together, we hope not only to identify specific genes that contribute to the high proportion of the variance in these traits but also to determine how the effects of environmental factors influence the expression of these genes.


*    Selected Abbreviations and Acronyms
 
apo = apolipoprotein
CHD = coronary heart disease
DHEAS = dehydroepiandrosterone sulfate
HDL-C = high-density lipoprotein cholesterol
LCAT = lecithin:cholesterol acyltransferase
LDL-C = low-density lipoprotein cholesterol
Lp(a) = lipoprotein(a)
LpAI = HDL particles bearing apoAI but not apoAII
SAFHS = San Antonio Family Heart Study
SHBG = sex hormone–binding globulin
VLDL-C = very low-density lipoprotein cholesterol


*    Acknowledgments
 
This work was supported by grants PO1-HL-45522 and DK-44297 from the National Institutes of Health. We are deeply grateful for the cooperation of the families participating in the SAFHS. We also thank Jane F. VandeBerg, Rosa Rosello, Janie Ludwig, Kathy K. Gruber, and Gilbert Aldrete, Jr, for technical assistance with the biochemical assays; Mary Perez and Irma Gomez, our clinic nurses; Richard Granato, who coordinated the field work; and Laurie Reinhart and Jim Bridges, who provided support with data management.

Received January 30, 1996; revision received April 30, 1996; accepted May 5, 1996.


*    References
up arrowTop
 up arrowAbstract
 up arrowIntroduction
 up arrowMethods
 up arrowResults
 up arrowDiscussion
 *References
 

  1. The Pooling Project Research Group. Relationship of blood pressure, serum cholesterol, smoking habit, relative weight and ECG abnormalities to incidence of major coronary events: final report of the Pooling Project. J Chron Dis. 1978;31:201-306.[Medline] [Order article via Infotrieve]
  2. Castelli WP, Garrison RJ, Wilson PWF, Abbott RD, Kalousdian S, Kannel WB. Incidence of coronary heart disease and lipoprotein cholesterol levels in the Framingham Study. JAMA. 1986;256:2835-2838.[Abstract]
  3. Polk RL. 1988-89 San Antonio City Directory. Dallas, Tex: RL Polk and Co; 1989.
  4. World Health Organization Expert Committee. Second Report on Diabetes Mellitus. Geneva, Switzerland: World Health Organization; 1980, Technical Report Series No. 646.
  5. Cheng M-L, Woodford SC, Hilburn JL, VandeBerg JL. A novel system for storage of sera frozen in small aliquots. J Biochem Biophys Methods. 1986;13:47-51.[Medline] [Order article via Infotrieve]
  6. Warnick GR, Benderson J, Albers JJ. Dextran sulfate-Mg2+ precipitation procedure for quantification of high-density-lipoprotein cholesterol. Clin Chem. 1982;28:1379-1388.[Free Full Text]
  7. Patsch W, Brown SA, Morrisett JD, Gotto AM Jr, Patsch JR. A dual-precipitation method evaluated for measurement of cholesterol in high-density lipoprotein subfractions HDL2 and HDL3 in human plasma. Clin Chem. 1989;35:265-270.[Abstract/Free Full Text]
  8. Friedewald WT, Levy RI, Frederickson DL. Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin Chem. 1972;18:499-502.[Abstract]
  9. Hogle DM, Smith RS, Curtiss LK. Quantitation of plasma apolipoprotein A-I using two monoclonal antibodies in an enzyme-linked immunosorbent assay. J Lipid Res. 1988;29:1221-1229.[Abstract]
  10. Young SG, Smith RS, Hogle DM, Curtiss LK, Witztum JL. Two new monoclonal antibody-based enzyme-linked assays of apolipoprotein B. Clin Chem. 1986;32:1484-1490.[Abstract/Free Full Text]
  11. Smith SJ, Cooper GR, Henderson LO, Hannon WH, Apolipoprotein Standardization Collaborating Group. An international collaborative study on standardization of apolipoproteins A-I and B, part I: evaluation of a lyophilized candidate reference and calibration material. Clin Chem. 1987;33:2240-2249.[Abstract/Free Full Text]
  12. Stein EA, DiPersio L, Pesce AJ, Kashyap M, Kao J-T, Srivastava L, McNerney C. Enzyme-linked immunoabsorbant assay of apolipoprotein AII in plasma, with use of a monoclonal antibody. Clin Chem. 1986;32:967-971.[Abstract/Free Full Text]
  13. Miller JA, Stein EA, Kaplan LA. Development of a competitive binding enzyme-linked immunoabsorbant assay (ELISA) for plasma apolipoprotein E using a monoclonal antibody (Mab). Clin Chem. 1990;36:964. Abstract.
  14. Parra HJ, Mezdour H, Ghalim N, Bard JM, Fruchart JC. Differential electroimmunoassay of human LpA-I lipoprotein particles on ready-to-use plates. Clin Chem. 1990;36:1431-1435.[Abstract/Free Full Text]
  15. Hill JS, O K, Wang X, Paranjape S, Dimitrijevich D, Lacko AG, Pritchard PH. Expression and characterization of recombinant human lecithin:cholesterol acyltransferase. J Lipid Res. 1993;34:1245-1251.[Abstract]
  16. Willett W, Sampson L, Stampfer M, Rosner B, Bain C, Witschi S, Hennekens C, Speizer F. Reproducibility and validity of a semiquantitative food frequency questionnaire. Am J Epidemiol. 1985;122:51-65.[Abstract/Free Full Text]
  17. Stern MP, Gonzalez C, Hernandez M, Knapp JA, Hazuda HP, Villalpando E, Valdez RA, Haffner SM, Mitchell BD. Performance of semiquantitative food frequency questionnaires in international comparisons: Mexico City vs San Antonio, TX. Ann Epidemiol. 1993;3:300-307.[Medline] [Order article via Infotrieve]
  18. Adams CF. Nutritive Value of American Foods. Washington, DC: US Department of Agriculture; 1975, handbook No. 456.
  19. Consumer and Food Economics Institute. Composition of Foods: Dairy and Egg Products, Raw, Processed, Prepared. Washington, DC: US Department of Agriculture; 1976, handbook No. 8-1.
  20. Consumer and Food Economics Institute. Composition of Foods: Fruits and Fruit Juices, Raw, Processed, Prepared. Washington, DC: US Department of Agriculture; 1982, handbook No. 8-9.
  21. Blair SN, Haskell WL, Ho P, Paffenbarger RS, Vranizan KM, Farquhar JW, Wood PD. Assessment of habitual physical activity by a seven-day recall in a community survey and controlled experiments. Am J Epidemiol. 1985;122:794-804.[Abstract/Free Full Text]
  22. Sallis JF, Haskell WL, Wood PD, Fortmann SP, Rogers T, Blair SN, Paffenbarger RS. Physical activity assessment methodology in the Five-City Project. Am J Epidemiol. 1985;121:91-106.[Abstract/Free Full Text]
  23. Lange K, Weeks DE, Boehnke M. Programs for pedigree analysis: MENDEL, FISHER, and dGENE. Genet Epidemiol. 1988;5:471-472.[Medline] [Order article via Infotrieve]
  24. Beaty TH, Self SG, Liang KY, Connolly MA, Chase GA, Kwiterovich PO. Use of robust variance components models to analyze triglyceride data in families. Ann Hum Genet. 1985;49:315-328.[Medline] [Order article via Infotrieve]
  25. Mitchell BD, Stern MP, Haffner SM, Hazuda HP, Patterson JK. Risk factors for cardiovascular mortality in Mexican Americans and non-Hispanic whites: the San Antonio Heart Study. Am J Epidemiol. 1990;131:423-433.[Abstract/Free Full Text]
  26. Haffner SM, Mitchell BD, Stern MP, Hazuda HP, Patterson JK. Decreased prevalence of hypertension in Mexican Americans. Hypertension. 1990;16:225-232.[Abstract/Free Full Text]
  27. Mitchell BD, Hazuda HP, Haffner SM, Patterson JK, Stern MP. Myocardial infarction in Mexican Americans and non-Hispanic whites: the San Antonio Heart Study. Circulation. 1991;83:45-51.[Abstract/Free Full Text]
  28. Mahaney MC, Blangero J, Comuzzie AG, VandeBerg JL, Stern MP, MacCluer JW, Hixson JE. A major locus for high density lipoprotein cholesterol in the San Antonio Family Heart Study: segregation and linkage analyses. Arterioscler Thromb Vasc Biol. 1995;15:1730-1739.[Abstract/Free Full Text]
  29. Kammerer CM, Atwood LD, Hixson JE, Hackleman SM, Mitchell BD, VandeBerg JL, Haffner SM, Stern MP. Major genes affect serum levels of low density lipoprotein cholesterol and apolipoprotein B in Mexican Americans. Circulation. 1994;90(suppl I):I-509. Abstract.
  30. Blangero J, Williams-Blangero S, Mahaney MC, Comuzzie AG, Hixson JE, Samollow PB, Sharp RM, Stern MP, MacCluer JW. Effects of a major gene for apolipoprotein A-I concentration are thyroid hormone dependent in Mexican Americans. Arterioscler Thromb Vasc Biol. In press.
  31. Comuzzie AG, Blangero J, Mahaney MC, Mitchell BD, Hixson JE, Samollow PB, Stern MP, MacCluer JW. A major gene with gender-specific effects influences fat mass in Mexican Americans. Genet Epidemiol. 1995;12:475-488.[Medline] [Order article via Infotrieve]
  32. Mitchell BD, Kammerer CM, Hixson JE, Atwood LD, Hackleman S, Blangero J, Haffner SM, Stern MP, MacCluer JW. Evidence for a major gene affecting post-challenge insulin levels in Mexican Americans. Diabetes. 1995;44:284-289.[Abstract]
  33. Sing CF, Moll PP. Genetics of variability of CHD risk. Int J Epidemiol. 1989;18(suppl 1):S183-S195.
  34. McGue M. Tecumseh and the genetics of coronary heart disease: a review. In: Rao DC, Elston RC, Kuller LH, Feinleib M, Carter C, Havlik R, eds. Genetic Epidemiology of Coronary Heart Disease: Past, Present, and Future. New York, NY: AR Liss; 1984:23-36.
  35. Berg K. Role of genetic factors in atherosclerotic disease. Am J Clin Nutr. 1989;49:1025-1029.
  36. Hunt SC, Hasstedt SJ, Kuida H, Stults BM, Hopkins PN, Williams RR. Genetic heritability and common environmental components of resting and stressed blood pressures, lipids, and body mass index in Utah pedigrees and twins. Am J Epidemiol. 1989;129:625-638.[Abstract/Free Full Text]
  37. Chung CS. Cincinnati Lipid Research Clinic: review of genetic data and design, limitations and prospects. In: Rao DC, Elston RC, Kuller LH, Feinleib M, Carter C, Havlik R, eds. Genetic Epidemiology of Coronary Heart Disease: Past, Present, and Future. New York, NY: AR Liss; 1984:135-147.
  38. Hamsten A, Iselius L, Dahlen G, deFaire U. Genetic and cultural inheritance of serum lipids, low and high density lipoprotein cholesterol and serum apolipoproteins A-I, A-II, and B. Atherosclerosis. 1986;60:199-208.[Medline] [Order article via Infotrieve]
  39. Rice T, Vogler GD, Perry TS, Laskarzewski PM, Rao DC. Familial aggregation of lipids and lipoproteins in families ascertained through random and nonrandom probands in the Iowa Lipid Research Clinics Family Study. Hum Hered. 1991;41:107-121.[Medline] [Order article via Infotrieve]
  40. Rao DC, Laskarzewski PM, Morrison JA, Khoury P, Kelly K, Wette R, Russell J, Glueck CJ. The Cincinnati Lipid Research Clinic Family Study: cultural and biological determinants of lipids and lipoprotein concentrations. Am J Hum Genet. 1982;34:888-903.[Medline] [Order article via Infotrieve]
  41. Beaty TH, Self SG, Chase GA, Kwiterovich PO. Assessment of variance components models on pedigrees using cholesterol, low-density, and high-density lipoprotein measurements. Am J Med Genet. 1983;16:117-129.[Medline] [Order article via Infotrieve]
  42. Hasstedt SJ, Kuida H, Ash KO, Williams RR. Effects of household sharing on high-density lipoprotein and its subfractions. Genet Epidemiol. 1985;2:339-348.[Medline] [Order article via Infotrieve]
  43. Vogler GP, Rao DC, Laskarzewski PM, Glueck CJ, Russell JM. Multivariate analysis of lipoprotein cholesterol fractions. Am J Epidemiol. 1987;125:706-719.[Abstract/Free Full Text]
  44. Feinleib M, Garrison RJ, Fabsitz R, Christian JC, Hrubec Z, Borhani NO, Kannel WB, Rosenman R, Schwartz JT, Wagner JO. The NHLBI twin study of cardiovascular disease risk factors: methodology and summary of results. Am J Epidemiol. 1977;106:284-295.[Abstract/Free Full Text]
  45. Iselius L. Analysis of the family resemblance for lipids and lipoproteins. Clin Genet. 1979;15:300-306.[Medline] [Order article via Infotrieve]
  46. Moll PP, Michels VV, Weidman WH, Kottke BA. Genetic determination of plasma apolipoprotein A1 in a population-based sample. Am J Hum Genet. 1989;44:124-139.[Medline] [Order article via Infotrieve]
  47. Hasstedt SJ, Albers JJ, Cheung MC, Jorde LB, Wilson DE, Edwards CO, Cannon WN, Ash KO, Williams RR. The inheritance of high density lipoprotein cholesterol and apolipoproteins A-I and A-II. Atherosclerosis. 1984;51:21-29.[Medline] [Order article via Infotrieve]
  48. Sistonen P, Ehnholm C. On the heritability of serum high density lipoprotein in twins. Am J Hum Genet. 1980;32:1-7.[Medline] [Order article via Infotrieve]
  49. Hasstedt SJ, Wilson DE, Edwards CO, Cannon WN, Carmelli D, Williams RR. The genetics of quantitative plasma Lp(a): analysis of a large pedigree. Am J Med Genet. 1983;16:179-188.[Medline] [Order article via Infotrieve]
  50. Moll PP, Harburg E, Burns TL, Schork MA, Ozgoren F. Heredity, stress, and blood pressure, a family set approach: the Detroit Project revisited. J Chron Dis. 1983;36:317-328.[Medline] [Order article via Infotrieve]
  51. Morton NE, Gulbrandsen CL, Rao DC, Rhoads GG, Kagan A. Determinants of blood pressure in Japanese-American families. Hum Genet. 1980;53:261-266.[Medline] [Order article via Infotrieve]
  52. Longini IM Jr, Higgins MW, Hinton PC, Moll PP, Keller JB. Genetic and environmental sources of familial aggregation of body mass in Tecumseh, Michigan. Hum Biol. 1984;56:733-757.[Medline] [Order article via Infotrieve]
  53. Moll PP, Burns TL, Lauer RM. The genetic and environmental sources of body mass index variability: the Muscatine Ponderosity Family Study. Am J Hum Genet. 1991;49:1243-1255.[Medline] [Order article via Infotrieve]
  54. Boehnke M, Moll PP, Kottke BA, Weidman WH. Partitioning the variability of fasting plasma glucose levels in pedigrees: genetic and environmental factors. Am J Epidemiol. 1987;125:679-689.[Abstract/Free Full Text]
  55. Laskarzewski PM, Rao DC, Glueck CJ. The Cincinnati Lipid Research Clinic Family Study: analysis of commingling and family resemblance for fasting blood glucose. Genet Epidemiol. 1984;1:341-355.[Medline] [Order article via Infotrieve]
  56. Folsom AR, Caspersen CJ, Taylor HL, Jacobs DR Jr, Luepker RV, Gomez-Marin O, Gillum RF, Blackburn H. Leisure time physical activity and its relationship to coronary risk factors in a population-based sample. Am J Epidemiol. 1985;121:570-579.[Abstract/Free Full Text]
  57. Gibbons LW, Blair SN, Cooper KH, Smith M. Association between coronary heart disease risk factors and physical fitness in healthy adult women. Circulation. 1983;67:977-983.[Abstract/Free Full Text]
  58. Regensteiner JG, Mayer EJ, Shetterly SM, Eckel RH, Haskell WL, Marshall JA, Baxter J, Hamman RF. Relationship between habitual physical activity and insulin levels among nondiabetic men and women. Diabetes Care. 1991;14:1066-1074.[Abstract]
  59. Donahue RP, Orchard TJ, Becker DJ, Kuller LH, Drash AL. Sex differences in the coronary heart disease risk profile: a possible role for insulin: the Beaver County Study. Am J Epidemiol. 1987;125:650-657.[Abstract/Free Full Text]
  60. Mayer EJ, Burchfiel CM, Eckel RH, Marshall JA, Haskell WL, Hamman RF. The role of insulin and body fat in associations of physical activity with lipids and lipoproteins in a biethnic population. Arterioscler Thromb. 1991;11:973-984.[Abstract/Free Full Text]
  61. Willett W. Nutritional Epidemiology. New York, NY: Oxford University Press; 1990.
  62. Schaefer EJ, Lamon-Fava S, Ordovas JM, Cohn SD, Schaefer MM, Castelli WP, Wilson PWF. Factors associated with low and elevated plasma high density lipoprotein cholesterol and apolipoprotein A-I levels in the Framingham Offspring Study. J Lipid Res. 1994;35:871-882.[Abstract]
  63. Mayer EJ, Newman B, Quesenberry CP Jr, Friedman GD, Selby JV. Alcohol consumption and insulin concentrations: role of insulin in associations of alcohol intake with high-density lipoprotein cholesterol and triglycerides. Circulation. 1993;88:2190-2197.[Abstract/Free Full Text]
  64. Criqui MH, Wallace RB, Heiss G, Mishkel M, Schonfeld G, Jones GTL. Cigarette smoking and plasma high-density lipoprotein cholesterol: the Lipids Research Clinics Program Prevalence Study. Circulation. 1980;(suppl IV):IV-70-IV-76.
  65. Rainwater DL, MacCluer JW, Stern MP, VandeBerg JL, Haffner SM. Effects of NIDDM on lipoprotein(a) concentration and apolipoprotein(a) size. Diabetes. 1994;43:942-946.[Abstract]



This article has been cited by other articles:


Home page
 Nephrol Dial TransplantHome page
N. H. Arar, V. S. Voruganti, S. D. Nath, F. Thameem, R. Bauer, S. A. Cole, J. Blangero, J. W. MacCluer, A. G. Comuzzie, and H. E. Abboud
A genome-wide search for linkage to chronic kidney disease in a community-based sample: the SAFHS
Nephrol. Dial. Transplant., April 28, 2008; (2008) gfn215v1.
[Abstract] [Full Text] [PDF]

Home page
 CirculationHome page
W. Post, L. F. Bielak, K. A. Ryan, Y.-C. Cheng, H. Shen, J. A. Rumberger, P. F. Sheedy II, A. R. Shuldiner, P. A. Peyser, and B. D. Mitchell
Determinants of Coronary Artery and Aortic Calcification in the Old Order Amish
Circulation, February 13, 2007; 115(6): 717 - 724.
[Abstract] [Full Text] [PDF]

Home page
 Am. J. Clin. Nutr.Home page
N. F Butte, G. Cai, S. A Cole, and A. G Comuzzie
Viva la Familia Study: genetic and environmental contributions to childhood obesity and its comorbidities in the Hispanic population.
Am. J. Clinical Nutrition, September 1, 2006; 84(3): 646 - 654.
[Abstract] [Full Text] [PDF]

Home page
 Hum Mol GenetHome page
R. Arya, E. Demerath, C. P. Jenkinson, H. H.H. Goring, S. Puppala, V. Farook, S. Fowler, J. Schneider, R. Granato, R. G. Resendez, et al.
A quantitative trait locus (QTL) on chromosome 6q influences birth weight in two independent family studies
Hum. Mol. Genet., May 15, 2006; 15(10): 1569 - 1579.
[Abstract] [Full Text] [PDF]

Home page
 J. Lipid Res.Home page
K. E. North, H. H. H. Goring, S. A. Cole, V. P. Diego, L. Almasy, S. Laston, T. Cantu, B. V. Howard, E. T. Lee, L. G. Best, et al.
Linkage analysis of LDL cholesterol in American Indian populations: the Strong Heart Family Study
J. Lipid Res., January 1, 2006; 47(1): 59 - 66.
[Abstract] [Full Text] [PDF]

Home page
 J. Lipid Res.Home page