Genetic Variations Associated With Echocardiographic Left Ventricular Traits in Hypertensive Blacks
Echocardiographic measures of cardiac target organ damage, including left ventricular mass and relative wall thickness, are powerful predictors of heart disease morbidity and mortality. The aim of this study is to investigate whether single nucleotide polymorphisms in candidate genes for hypertension and heart disease have effects on quantitative measures of hypertensive cardiac target organ damage, independent of their actions on blood pressure levels, in a cohort of hypertensive black sibships. To detect replication of genetic effects across samples, this study took advantage of the affected sibling pair design and created 2 samples, each with 448 unrelated individuals. As part of the Genetic Epidemiology Network of Arteriopathy Study, subjects were screened using 2D echocardiography, and 395 single nucleotide polymorphisms in 80 candidate genes were genotyped. Linear regression was used to test for single nucleotide polymorphisms significantly associated with left ventricular mass index (g/m2.7) or relative wall thickness after adjusting for associated covariates. Significant single nucleotide polymorphisms were subsequently tested for consistent directionality in genotype–phenotype relationships across samples. Three single nucleotide polymorphisms, 1 each in the APOE, SCN7A, and SLC20A1 genes, were significantly associated in both samples with left ventricular mass index and had replicate genotype–phenotype relationships. One in the ADRB1 gene was significantly associated with relative wall thickness with replicate effects in both samples. We identified genetic variation that significantly influences left ventricular traits with replicable effects in a cohort of hypertensive, black siblings.
- single nucleotide polymorphism
- left ventricular hypertrophy
- ventricular remodeling
Cardiac disease, including myocardial infarction, hypertensive and ischemic heart disease, and heart failure, is the leading cause of mortality and morbidity, accounting for 28% of the 2.4 million deaths in the United States each year.1 Understanding predictors of cardiac events is particularly important for the black population. In 2003, the age-adjusted cardiac death rate for blacks was higher compared with the entire population, 364 vs 232 per 100 000 persons per year.1 Furthermore, hypertension occurs in 40% to 60% of blacks2,3 and prevalence of cardiac target organ damage may be nearly twice that in the non-Hispanic white population.4
Hypertension is a primary risk factor for cardiac disease.5 However, because persons with similar blood pressure levels have varied cardiac outcomes, hypertension is not an exclusive predictor of cardiac risk or of damage to cardiac muscle.6,7 Echocardiographic measures of cardiac target organ damage (increased left ventricular mass [LVM] and high relative wall thickness [RWT]) have been well documented as powerful, independent predictors of cardiac morbidity and mortality.6,7 Echocardiography is a validated, noninvasive method for measuring cardiac function and structure to identify preclinical cardiac damage.8–10
Echocardiographically measured LVM and RWT are complex quantitative traits with both genetic and environmental components. Correlates of LVM and RWT are well documented in multiethnic studies and include older age, male sex, diabetes and increased body mass index, systolic blood pressure, stroke volume, and dietary sodium intake.4,11–13 However, ≤75% of interindividual variation in LVM and RWT remains unexplained by these established risk factors,14 prompting studies of genetic influences on echocardiographic measures of preclinical cardiac disease. The Framingham Heart Study estimated the heritability of LVM at 0.24 to 0.32,15 and a study involving 110 twin pairs obtained an LVM heritability estimate of 0.69.16 Identification of genetic loci involved in determining LVM and RWT and, therefore, predisposing to cardiac disease is highly warranted.
Linkage studies conducted in rats have identified several chromosomal regions associated with increased LVM.17,18 An alternative strategy to linkage analysis are gene–disease association studies, which provide greater statistical power for identifying genetic variation underlying diseases of multifactorial etiology.19 Although replication of results in independent data sets is critical for validation, most published gene–disease associations have not been replicated.20,21 To examine the effects of single nucleotide polymorphisms (SNPs) on echocardiographic measures of cardiac target organ damage, we examined associations of SNPs with LVM and RWT in hypertensive black adults using 2 samples of unrelated individuals to replicate results.
The National Heart Lung and Blood Institute established the Family Blood Pressure Program (FBPP) in 1996, joining established research networks investigating hypertension and cardiac diseases. One of the 4 networks in Family Blood Pressure Program is the Genetic Epidemiology Network of Arteriopathy (GENOA), which recruited hypertensive black, Hispanic, and non-Hispanic white sibships for linkage and family-based association studies to investigate genetic contributions to hypertension and hypertensive target organ damage. Subject recruitment for GENOA was population based in 3 geographic locations: Jackson, Miss; Starr County, Tex; and Rochester, Minn. Subjects for this particular GENOA substudy were blacks from Jackson. GENOA recruited sibships containing ≥2 individuals with clinically diagnosed essential hypertension before age 60 years. Participants were diagnosed with hypertension if they had a previous clinical diagnosis of hypertension by a physician with current antihypertensive treatment or an average systolic blood pressure ≥140 mm Hg or diastolic blood pressure ≥90 mm Hg on the second and third clinic visit. Exclusion criteria included secondary hypertension, alcoholism or drug abuse, pregnancy, insulin-dependent diabetes mellitus, or active malignancy.
Data for GENOA was collected over 2 phases. Phase I began in 1996, collecting blood pressure readings, information regarding family history, hypertension risk factors, and blood samples for genotyping and laboratory tests. Study visits were conducted in the morning after an overnight fast of ≥8 hours. Blood pressure was measured with random o sphygmomanometers and cuffs appropriate for arm size. Three readings were taken in the right arm after the participant rested in the sitting position for ≥5 minutes; the last 2 readings were averaged for the analyses. In Jackson, this was done for 1854 people coming from 683 sibships. Phase II began in 2001 with the goal of measuring target organ damage, specifically via echocardiography. Approximately 80% of phase I participants were successfully rerecruited for phase II. Blood pressure, hypertension risk factor information, and blood samples were reassessed. Informed consent was obtained from all of the subjects, and approval was granted by participating institutional review boards.
To detect SNP effects that replicated in another sample, this study took advantage of the affected sibling pair design, originally used for linkage analysis, and created 2 samples, each with 448 unrelated individuals. This was done by randomly sampling 1 sibling from each hypertensive sibship without replacement to create the first sample. From the remaining people, we randomly sampled a second sibling from each sibship to establish the second sample. Singletons were equally divided between the 2 samples. These 2 large samples enabled us to test for replication of SNP associations in samples with similar genetic and environmental backgrounds.
The left ventricular phenotypes of interest, LVM and RWT, were derived using phased-array echocardiographs with M-mode, 2D and pulsed, continuous wave, and colorflow Doppler capabilities. Standardized methods, along with training and certification, were used by field-center technicians to achieve high-quality recordings. Readings were performed at the New York Presbyterian Hospital–Weill Cornell Medical Center and verified by a single highly experienced investigator. To measure LVM and RWT, the parasternal acoustic window was used to record ≥10 consecutive beats of 2D and M-mode recordings of the left ventricular internal diameter and wall thicknesses at, or just below, the tips of the anterior mitral leaflet in long- and short-axis views. Correct orientation of planes for imaging and Doppler recordings was verified using standardized protocols. Measurements were made using a computerized review station equipped with digitizing tablet and monitor screen overlay for calibration and performance of each measurement. Left ventricular internal dimension and interventricular septal and posterior wall thicknesses were measured at end diastole and end systole according to the recommendations of the American Society of Echocardiography in ≤3 cardiac cycles.22 Calculations of LVM were made using a necropsy-validated formula,9 RWT was calculated as 2*(posterior wall thicknesses)/left ventricular internal dimension. Left ventricular phenotypes have excellent reliability when measured through echocardiography; for example, the correlation between repeated measures of LVM was 0.93 between paired echocardiograms in hypertensive adults.8
Genotyping of 395 SNPs in 80 candidate genes for subjects from Jackson was conducted at the GENOA central genotyping center at the University of Texas-Houston (see the data supplement available online at http://hyper.ahajournals.org for a list of SNPs). Genes were selected to represent biological pathways or positional candidate genes from systems known to be associated with hypertension and heart disease, including ion transport, inflammation, vascular wall biology, renin–angiotensin system, and lipid metabolism. SNPs provide a measure of genetic variation occurring between individuals of a population, because they measure single nucleotide base changes in a gene. SNPs were selected with a minor allele frequency >0.05 if nonsynonymous and >0.1 if marking 5′, 3′, or intronic regions of the gene. On average, 5 SNPs per gene (range: 1 to 18) were selected to represent variation in the region. This was done using the public National Center for Biotechnology Information database (http://www.ncbi.nlm.nih.gov) and the private Celera database (http://www.celeradiscoverysystem.com). SNP genotyping was obtained using a combination of 2 genotyping platforms: mass spectrometer–based detection system implemented on a Sequenom MassARRAY system and the fluorogenic TaqMan assay implemented on an ABI Prism 7900 Sequence Detection System. Primer and probe sequences are available on request.
High-throughput data analyses were conducted using the statistical software R. Descriptive statistics for covariates, outcome variables, and SNPs were generated. Continuous variables are presented as mean±SD. Student’s t test was used to confirm that the covariates and outcome variables in the 2 samples were not significantly different. LVM index (g/m2.7; LVMI) and RWT were transformed using the natural logarithm. Linear regression was used to identify covariates associated with LVMI and RWT. Multivariable models for both LVMI and RWT were then constructed based on the univariate modeling results and known risk factors for the outcome. Genetic descriptive statistics were calculated, including either a χ2 test or the Fisher’s exact test for Hardy–Weinberg equilibrium (HWE).
We applied a multistage analysis strategy to identify and validate SNP associations replicating in 2 samples and to reduce the implications of false-positives on our inferences. First, linear regression was used in both samples to identify SNPs that were significantly associated (α=0.10) with the respective outcome after adjustment for covariates. Only SNPs statistically significant in both samples were eligible for further analysis. Second, the genotype-specific means of LVMI and RWT associated with each SNP were determined in both samples. Two-sample t tests were applied to all of the pairwise comparisons of genotype-specific means within the SNP to determine whether the mean outcome differed between genotypes.23,24 SNPs with the same pairwise results in both samples were further analyzed. Third, ANCOVA was used to determine whether the genotype-specific mean of LVMI or RWT within each SNP was parallel and coincident across samples, thus ensuring homogeneity of effect across samples.23
General descriptive statistics of the covariate and outcome variables for the 2 samples are shown in Table 1. No significant differences were found between the 2 samples for any of the covariates or outcomes. Using the partition value of 51 g/m2.7,25 15% of participants had LVH in sample 1 and 16% in sample 2. Using the threshold of 0.43 for RWT to detect concentric left ventricular geometry,26,27 3% of participants had concentric left ventricular geometric patterns in sample 1 and 4% in sample 2.
Table 2 outlines the covariate associations in the univariate and final multivariable model of each outcome in detail. LVMI was adjusted for age, body mass index, gender, systolic blood pressure, diabetes, and stroke volume. RWT was adjusted for age, body mass index, systolic blood pressure, and stroke volume. The residuals from the adjusted models provided the dependent variable for the SNP associations.
In sample 1, 34 SNPs in 23 genes were significantly associated (α=0.10) with LVMI, and 19 SNPs in 14 genes were associated with RWT. In sample 2, 33 SNPs in 20 genes were significantly associated with LVMI, and 34 SNPs in 23 genes were associated with RWT. Of these, 3 SNPs were significantly associated with LVMI in both samples: APOE_rs449647 (P=0.014 and 0.055, respectively), SCN7A_cv356952 (P=0.089 and 0.024, respectively), and SLC20A1_cv9546580 (P=0.095 and 0.001, respectively; Table 3). One SNP was significantly associated with RWT in both samples: ADRB1_Arg389Gly (P=0.019 and 0.065, respectively; Table 3).
Genotype–Phenotype Relationships for LVMI
The 3 SNPs replicating associations with LVMI exhibited significant and consistent genotype–phenotype relationships in both samples (Table 4). There was a significant difference in mean LVMI between the APOE_rs449647 AA genotype and the combined AT/TT genotypes in samples 1 and 2 (P=0.010 and 0.004, respectively). The AT and TT genotypes were combined, because t tests revealed no significant differences between mean LVMI at these genotypes in either sample (P=0.644 and 0.876, respectively). The major allele, T, of APOE_rs449647 was significantly associated with higher LVMI by a mean of 4.1 g/m2.7 in sample 1 and 3.4 g/m2.7 in sample 2.
A significant difference in mean LVMI was also found between the AA homozygote in SCN7A_cv356952 and the combined AT/TT genotypes in samples 1 and 2 (P=0.041 and 0.025, respectively). The AT and TT genotypes were combined, because mean LVMI between those genotypes was not significantly different in either sample (P=0.427 and 0.055). The major allele, T, of SCN7A_cv356952 was significantly associated with higher LVMI by a mean of 1.9 g/m2.7 in sample 1 and 2.3 g/m2.7 in sample 2.
The AA homozygote in SLC20A1_cv9546580 had a significantly lower mean LVMI compared with the heterozygote AG in samples 1 and 2 (P=0.037 and 0.005, respectively). The difference in mean LVMI between genotypes AG and GG was only significant in sample 2 (P=0.162 and 0.0003). The AG genotype of SLC20A1_cv9546580 was associated with the highest values of LVMI in both samples, exhibiting a pattern of overdominance.
In addition, all 3 of the SNPs were separately tested using ANCOVA for genotype–sample interactions. There was no statistical evidence (α=0.05) that sample influenced the genotype-specific means of LVMI (data not shown). Therefore, the consistency of directionality in the genotype–phenotype relationships for LVMI was confirmed across samples (shown without log transformations in Figures 1 to 3⇓⇓).
The 3 SNPs found to be significantly associated with LVMI were added to the multivariable model, and the R2 significantly increased in both samples. In sample 1, the R2 increased from 0.316 to 0.381 (P=0.003). In sample 2, the R2 increased from 0.316 to 0.368 (P<0.001). Jointly, these SNPs explain 5% to 7% additional variability in LVMI beyond traditional risk factors.
Genotype–Phenotype Relationship for RWT
The Arg389Gly SNP in ADRB1 was significantly associated with RWT in both samples and exhibited significant, consistent genotype–phenotype relationships within and across samples (Table 4). The CC homozygote in ADRB1_Arg389Gly had a significantly lower mean RWT compared with the combined CG and GG genotypes in samples 1 and 2 (P=0.028 and 0.019, respectively). The CG and GG genotypes were combined, because the mean RWT between those genotypes was not significantly different in either sample (P=0.079 and 0.869). The major allele, G, of ADRB1_Arg389Gly was significantly associated with higher RWT by a mean of 0.01 in sample 1 and 0.009 in sample 2.
ADRB1_Arg389Gly genotypes were also tested using ANCOVA for genotype–sample interactions. There was no statistical evidence (α=0.05) that sample influenced the genotype-specific means of RWT (data not shown). Therefore, the consistency of directionality in the genotype–phenotype relationships for RWT was confirmed across samples (Figure 4).
The ADRB1 SNP found to be significantly associated with RWT was added to the multivariable model, and the R2 increased in both samples. In sample 1, the R2 increased from 0.124 to 0.142 (P=0.019). In sample 2, the R2 increased from 0.122 to 0.132 (P=0.067). This indicates that ADRB1_Arg389Gly explains 1% to 2% additional variability in RWT beyond traditional risk factors.
Genetic Descriptives for Significant SNPs
All 395 of the SNPs were tested for HWE using either the χ2 test or the Fisher’s exact test. Of the 4 SNPs identified as significantly associated with LVMI or RWT, ADRB1_Arg389Gly violated HWE in both samples (P=0.015 and 0.038, respectively), and SLC20A1_cv9546580 violated HWE in sample 1 (P=0.007). This could be because of various reasons, including population stratification,28 type 1 error, or because selection is working against ≥1 genotype, thereby indicating a true association.29 Furthermore, we would not expect this hypertensive sample to follow HWE distributions if the SNP is associated with hypertension.30 None of the 4 SNPs were in linkage disequilibrium, and more information specific to individual SNPs can be found on Table S1.
Replication of Results
To assess whether the effects of these genetic polymorphisms replicated beyond the black Jackson cohort, we used the already existing Hispanic cohort of GENOA from Starr County, Texas (n=1228). Of the 4 SNPs reported above, 2 also showed significant associations with LVMI in the Hispanic cohort: SCN7A_cv356952 (AA versus AT and TT: P=0.029) and SLC20A1_cv9546580 (AA/AG versus GG: P=0.074). We noted that the allele frequency distribution was different (α=0.05; χ2 degrees of freedom=1) between the black cohort from Jackson and the Hispanic cohort from Starr County in 3 of the 4 SNPs, SCL20A1_cv9546580 being the only SNP with similar allele frequencies. The minor allele frequencies for APOE_rs449647, SCN7A_cv356952, and ADRB1_Arg389Gly in blacks are 0.29, 0.46, and 0.38 (respectively) compared with 0.23, 0.24, and 0.21 in Hispanics. These differences in allele frequencies indicate that we had a different power to detect the same SNP effect in these 2 ethnic groups. Allele frequencies affect the power of single gene association tests because they are related to the number of individuals in a particular genotype class.
In this study of hypertensive black siblings, we identified 3 SNPs (APOE_rs449647, SCN7A_cv356952, and SLC20A1_cv9546580) that replicated association with LVMI in 2 black samples and showed consistent phenotypic effects. In addition, we identified 1 SNP (ADRB1_Arg389Gly) that replicated association with RWT in 2 black samples and also demonstrated consistent phenotypic effects. Furthermore, 2 of the SNPs showed significant associations with LVMI in an independent population-based sample of Hispanic participants in GENOA (SCN7A_cv356952 and SLC20A1_cv9546580).
The process by which we identified these results carefully addressed 2 main issues of conducting gene–disease association studies: replication of results and the issues of multiple hypothesis testing. Both of these issues will require increasing attention in the near future as the field of genetic epidemiology transitions toward genome-wide association studies. Genome-wide association studies will increase the number of nonreplicable results, because the number of potentially positive results will increase exponentially. A meta-analysis of >600 reported gene–disease associations found few replications.31 Of the 600 associations, 166 had been studied ≥3 times, and only 6 were consistently replicated.31 Although this meta-analysis included studies of many outcomes, the same issue applies to gene association studies for left ventricular traits. It is possible that the lack of replicated results in the published literature is because of differences among study populations in distributions of genetic and environmental factors,32 as well as in sampling techniques and analysis strategies. Therefore, creating 2 samples from a single population provides an opportunity to compare results across samples with similar genetic backgrounds and environmental exposures.
The second issue of association studies that we addressed is that of multiple hypothesis testing. A variety of methods are available to correct for multiple hypothesis testing: the conservative Bonferroni adjustment of a single P value, controlling for false discovery rates, cross-validation techniques, or replicating results in an independent sample. Most of these methods operate on the principle of lowering the α-level necessary for rejecting the null hypotheses. α-Levels and the power to detect associations inherently take into account factors such as allele frequencies and the size of each allele effect. It is highly likely that alleles conferring an effect in complex, common diseases will have small effect sizes. Therefore, overly conservative adjustments of the α-level may miss a large portion of true associations. Genome-wide association studies magnify the problems of correcting for multiple hypothesis testing, because thousands, instead of hundreds, of hypothesis tests are conducted, making conservative correction techniques even more certain to exclude many true associations. To address the issue of multiple testing and to reduce the number of false-positives, we implemented a multistage analysis strategy to identify and validate SNP associations replicating in both samples. Ultimately, all 4 of SNPs reported in this article did the following: (1) were significantly associated with LVMI or RWT in both samples; (2) exhibited consistent, significant differences between the same genotype-specific mean outcome in both samples; and (3) showed consistent directionality and magnitude of genotype–phenotype relationships across samples.
Hypertension, cardiac target organ damage, and overt cardiac disease all occur at increased rates in the black population. Of the 4 SNPs identified through the multistage testing, ADRB1_Arg389Gly, associated with RWT, is of particular interest for the black population. ADRB1 is the gene for the β1-adrenergic receptor that is the receptor for norepinephrine on the cardiomyocyte. Arg389Gly SNP is a nonsynonymous mutation resulting in a missense substitution within amino acid 389. This amino acid change modifies protein function by increasing contractile response at the myocyte.33 The G allele has been found to be more frequent in the black population and may contribute to increased hypertension and heart failure rates in blacks.34 We identified the G allele of the Arg389Gly SNP to be significantly associated with increased RWT, which is a predictor of cardiac outcomes, such as heart failure. Our current results support the hypothesis of the Arg389Gly mutation playing a role in increased heart disease in blacks.
The direct functional implications of the 3 SNPs found to be associated with LVMI are not well studied in terms of their impact on LVM, but the genes are etiologic candidates for greater LVM and heart disease. Apolipoprotein E is a lipoprotein implicated in dyslipidemia and associated with a variety of heart disease outcomes. SCN7A is a voltage gated sodium ion channel, and SLC20A1 is the gene for a solute carrier protein, both of which may play a role in LVMI through cell signaling and regulation of hemodynamic load.
LVMI and RWT are complex quantitative traits that are influenced by both genetics and the environment. This study identified SNPs with replicated main effects for LVMI and RWT while adjusting for other risk factors. However, context dependency of the main effects within the genome and environment are critical to fully understand these complex traits. Therefore, future directions for our research are to further explore these SNP main effects along with potential interactions with other known risk factors and/or with other SNPs. In addition, we feel it is worthwhile to further explore the ADRB1 SNP given that it is a nonsynonymous SNP that has been implicated previously in heart disease in blacks.
The field of genetic epidemiology is growing rapidly with the increasing availability of genomic data. This article highlights statistical and methodologic issues that are emerging from this exponential growth and presents a novel approach for replicating gene–disease association results. This study provides insight into genetic contributions of quantitative measures of hypertensive cardiac organ damage. Consideration of context dependency and interactive genetic effects is an obvious progression for this research.
We thank Jian Chu, Todd Greene, Reagan Kelly, Jennifer Smith, and Yan Sun for their scientific input throughout the development of this article.
Sources of Funding
This work was supported by the National Institutes of Health and the National Heart Lung and Blood Institute grants HL054481 and HL54457.
This paper was sent to Curt D. Sigmund, associate editor, for review by expert referees, editorial decision, and final disposition.
- Received October 6, 2006.
- Revision received October 23, 2006.
- Accepted February 7, 2007.
National Center for Health Statistics. CDC Health, United States, 2005. Hyattsville, MD: US Government Printing Office, 2005.
Kramer H, Han C, Post W, Goff D, Diez-Roux A, Cooper R, Jinagouda S, Shea S. Racial/ethnic differences in hypertension and hypertension treatment and control in the Multi-Ethnic Study of Atherosclerosis (MESA). Am J Hypertens. 2004; 17: 963–970.
Kizer JR, Arnett DK, Bella JN, Paranicas M, Rao DC, Province MA, Oberman A, Kitzman DW, Hopkins PN, Liu JE, Devereux RB. Differences in left ventricular structure between black and white hypertensive adults: the Hypertension Genetic Epidemiology Network Study. Hypertension. 2004; 43: 1182–1188.
Asia Pacific Cohort Studies Collaboration. The impact of cardiovascular risk factors on the age-related excess risk of coronary heart disease. Int J Epidemiol. 2006; 35: 1025–1033.
Palmieri V, Dahlof B, DeQuattro V, Sharpe N, Bella BN, de Simone G, Paranicas M, Fishman D, Devereux RB. Reliability of echocardiographic assessment of left ventricular structure and function: the PRESERVE study. Prospective Randomized Study Evaluating Regression of Ventricular Enlargement. J Am Coll Cardiol. 1999; 34: 1625–1632.
Bella JN, Wachtell K, Palmieri V, Liebson PR, Gerdts E, Tlitalo A, Koren MJ, Pederson OL, Rokkedal J, Dahlof B, Roman MJ, Devereux RB. Relation of left ventricular geometry and function to systemic hemodynamics in hypertension: the LIFE study. Losartan Intervention for Endpoint Reduction in Hypertension Study. J Hypertens. 2001; 19: 127–134.
du Cailar G, Ribstein J, Mimran A. Dietary sodium and target organ damage in essential hypertension. Am J Hypertens. 2002; 15: 222–229.
Post WS, Larson MG, Myers RH, Galderisi M, Levy D. Heritability of left ventricular mass: the Framingham Heart Study. Hypertension. 1997; 30: 1025–1028.
Swan L, Birnie DH, Padmanabhan S, Inglis G, Connell JM, Hillis WS. The genetic determination of left ventricular mass in healthy adults. Eur Heart J. 2003; 24: 577–582.
Innes BA, McLaughlin MG, Kapuscinski MK, Jacob HJ, Harrap SB. Independent genetic susceptibility to cardiac hypertrophy in inherited hypertension. Hypertension. 1998; 31: 741–746.
Risch N, Merikangas K. The future of genetic studies of complex human diseases. Science. 1996; 273: 1516–1517.
Lang RM, Bierig M, Devereux RB, Flachskampf FA, Foster E, Pellikka PA, Picard MH, Roman MJ, Seward J, Shanewise JS, Solomon SD, Spencer KT, Sutton MS, Stewart WJ. Recommendations for chamber quantification: a report from the Am Society of Echocardiography’s Guidelines and Standards Committee and the Chamber Quantification Writing Group, developed in conjunction with the European Association of Echocardiography, a branch of the European Society of Cardiology. J Am Soc Echocardiogr. 2005; 18: 1440–1463.
Sokal RR, Rohlf FJ. Biometry: The Principles and Practice of Statistics in Biological Research. 3rd ed. New York, NY: WH Freeman and Company, 1994.
Falconer DS. Introduction to Quantitative Genetics. 2nd ed. New York, NY: John Wiley & Sons, Inc, 1981.
Nunez E, Arnett DK, Benjamin EJ, Liebson PR, Skelton TN, Taylor H, Andrew M. Optimal threshold value for left ventricular hypertrophy in blacks: the Atherosclerosis Risk in Communities Study. Hypertension. 2005; 45: 58–63.
Wachtell K, Bella JN, Liebson PR, Gerdts E, Dahlof B, Aalto T, Roman MJ, Papademetriou V, Ibsen H, Rokkedal J, Devereux RB. Impact of different partition values on prevalences of left ventricular hypertrophy and concentric geometry in a large hypertensive population: the LIFE study. Hypertension. 2000; 35: 6–12.
Deng HW, Chen WM, Recker RR. Population admixture: detection by Hardy-Weinberg test and its quantitative effects on linkage-disequilibrium methods for localizing genes underlying complex traits. Genetics. 2001; 157: 885–897.
Lee WC. Searching for disease-susceptibility loci by testing for Hardy-Weinberg Disequilibrium in a gene bank of affected individuals. Am J Epidemiol. 2003; 158: 397–400.