Genetics of cardiovascular disease: importance of sex and ethnicity

Abstract

Sex differences in incidence and prevalence of and morbidity and mortality from cardiovascular disease are well documented. However, many studies examining the genetic basis for cardiovascular disease fail to consider sex as a variable in the study design, in part, because there is an inherent difficulty in studying the contribution of the sex chromosomes in women due to X chromosome inactivation. This paper will provide general background on the X and Y chromosomes (including gene content, the pseudoautosomal regions, and X chromosome inactivation), discuss how sex chromosomes have been ignored in Genome-wide Association Studies (GWAS) of cardiovascular diseases, and discuss genetics influencing development of cardiovascular risk factors and atherosclerosis with particular attention to carotid intima-medial thickness, and coronary arterial calcification based on sex-specific studies. In addition, a brief discussion of how ethnicity and hormonal status act as confounding variables in sex-based analysis will be considered along with methods for statistical analysis to account for sex in cardiovascular disease.

Introduction

Sex differences in incidence, prevalence, morbidity and mortality from cardiovascular diseases are well documented and represent important health disparities [1–3]. These sex differences can be classified as those that are sex-specific such as erectile dysfunction in men and hypertensive pregnancy disorders in women, or conditions common to both sexes but which show sex differences in presentation and outcomes such as hypertension, atherosclerosis, angina, and stroke.

Sex differences in cardiovascular diseases result from a complex interaction among genetic, hormonal and environmental factors that provide a profile of individual risk and phenotypic presentation of disease. Therefore, there is an increased interest in identifying the genetic components of disease to optimize treatments and outcomes. However, investigations into the genetics of cardiovascular diseases in men and women are hampered by the failure to include the sex chromosomes in genome-wide association studies (GWAS), to account for sex as variable in targeted genetic analyses, and to examine hormone-gene interactions. In this paper, we will provide general background on the X and Y chromosomes, discuss how they have been ignored in GWAS of cardiovascular disease, and discuss genetic factors contributing to sex differences in conventional cardiovascular risk factors and atherosclerosis with examples of sex-specific studies of carotid intima-medial thickness, and coronary arterial calcification. Confounding factors of ethnicity and hormonal status also will be considered along with some proposed methods for statistical analysis for future studies.

Background: the genetics of sex

In humans, sex is determined via the X and Y sex chromosomes. Each somatic cell of the human body contains 23 pairs of chromosomes, 22 of which are the same in both sexes (autosomes); however females have two copies of the X chromosome, whereas males have one copy of the X and one copy of the Y [4]. Therefore, differences between the sexes result from the particular combination of sex chromosomes an individual possesses, as well as, differing levels of the sex hormones. Often early studies of sex differences compared persons with XX and XY genotypes to those with sex chromosome monosomy (e.g. XO karyotype in Turner syndrome) or trisomy (e.g. XXY karyotype in Klinefelter syndrome) to distinguish whether observed sex differences were due to sex chromosome dosage effects independent of hormonal influences [5].

As genotyping technologies have rapidly improved, studies of the genetic basis of sex differences are no longer limited to investigating global differences in sex chromosome dosage effects, but can directly examine the individual genes and genetic variants on the X and Y chromosomes. In humans, the larger X chromosome contains 813 protein coding genes, whereas the much smaller Y chromosome contains 143 protein-coding genes, although many of the Y genes are multi-copy genes or present in both males and females (Ensemble Genome Browser, http://www.ensembl.org); only 46 genes are on the male-specific region of the Y (MSY). The sex chromosomes evolved from ancestral autosomes [6, 7], and over the course of evolution, the X has retained many of the ancestral genes (98% retained), but the Y lost many (3% retained) [8]. Only 14 ancestral X-Y gene pairs are still present in humans, and many of these have important regulatory functions in regards to transcription, translation, and protein stability [8, 9]. Some of the genes in the ancestral gene pairs retain the same function across the X and Y, but others have diverged. Additionally, both the X and Y have acquired new genes through evolution, with an emphasis on reproductive function [10, 11]. The Y chromosome contains the SRY gene, or sex-determining region Y, which encodes the testes determining factor [12]. Many other genes on the Y chromosome are related to male sexual development and reproduction, including multicopy gene families expressed exclusively in the testes [8, 11]. The X chromosome also includes reproductive-related genes, such as the androgen receptor [13], and genes important for brain development, blood clotting, and visual pigmentation [14].

The X and Y chromosomes have developed unique properties as compared to the autosomes, as illustrated schematically in Figure 1 (this schematic diagram does not reflect the large difference in genetic content between the X and Y chromosomes, such as the large block of non-coding heterochromatin on the Y located in the MSY region designated in the figure at *[11]). Unlike autosomal pairs, the X and Y do not generally undergo recombination due to evolutionary divergence. However, there are two regions on each end of the X and Y chromosomes that are homologous, called the pseudoautosomal regions (PAR), which behave like the autosomes and do recombine [6] (Figure 1A). For most genes in the PAR, both copies are expressed across both sex chromosomes (XX for females or XY for males). However, for other regions, that is, the genes on the non-pseudo-autosomal regions, the aneuploidy causes unequal patterns of expression across the sexes, which requires some form of dosage compensation across males and females [15]. One method is upregulation of X-linked genes [16]. In humans, one of the female copies of the X chromosome is silenced during embryogenesis to achieve dosage compensation during a process called X chromosome inactivation [6] (Figure 1B). The X-linked gene XIST is expressed from the inactive chromosome, triggering DNA methylation that is responsible for the chromosome-wide silencing [17]. The inactive chromosome remains silent in somatic cells, but is reactivated during meiosis. This inactivation is tissue and cell specific, such that the maternally inherited allele may be expressed in some cells, while the paternally inherited allele may be expressed in others [18]. This process occurs randomly so that usually each chromosome is inactivated in 50% of cells, although preferentially (skewed) inactivation of one chromosome can occur either globally or in specific tissues, as demonstrated in cardiac muscle of mice [18]. Skewed X-inactivation has been observed in samples from human arteries, suggesting that the cells in the plaque arose from a single or similar group of cells [19]. To further complicate matters, approximately 15% of X-linked genes escape inactivation in humans [20, 21]. For these `escapee genes,' both copies are active in females (whereas males may have a single active copy if there is no homologous gene on the Y). Mechanisms of dosage compensation, whether concerning X or Y specific genes undergoing upregulation or X inactivation, or shared homologous X-Y gene pairs or pseudoautosomal genes, may be related to differential gene expression across the sexes and are important considerations for sex specific disease risk.

Figure 1.

(A) Schematic of the X and Y chromosomes in males and females showing regions of shared or specific expression. This schematic is not drawn to scale, and does not include the large heterochromatin block within the male-specific region of the Y on the q arm (denoted by *) which is not expressed. (B) Schematic of X chromosome inactivation in multiple cells of the female body.

In contrast to the PAR on the X and Y chromosomes that is present in both sexes, the non-PAR region of the Y chromosome (MSY) encompasses the vast majority of the Y chromosome [11]. This region does not undergo recombination, and is inherited from father to son as a single haplotype, reflecting the paternal lineage only. Many of these genes are widely-expressed (at the RNA level), exhibiting global regulatory functions in transcription, translation, and other biological processes (non-reproductive), and hence may be important to men's health beyond sex determination and sexual reproduction [8].

Both X and Y chromosomes contribute to disease. Many rare Mendelian genetic disorders are linked to the sex chromosomes. According to the Online Mendelian Inheritance in Man (OMIM) catalog of genetic disorders in humans, approximately 7% of cataloged phenotypes are X-linked [22]. XY males are at increased risk compared to XX women for X-linked disorders, such as learning disabilities and mental retardation caused by X-linked genes important in brain development and function [14]. Hearing impairment (DFNY1, MIM 400043) reported in a Chinese family was the only documented Mendelian disorder showing Y-linkage in humans [23]. However, subsequent analysis found that the condition was related to an insertion of chromosome 1 into the Y gene rather than a mutation of a Y-chromosomal gene [24].

Much work to link the sex chromosomes to disease has focused on such rare disorders caused by a single gene, and less is known about the influence on complex traits with multiple genetic and etiological components. In particular, variants on the X chromosome are understudied for complex traits, with significantly fewer genome-wide associations than the autosomes [22] (Figure 2). Of the over 2800 genome-wide significant associations reported for over 300 traits in the NHGRI GWAS Catalog [25], only 15 published associations reside on the X despite representing approximately 5% of the human genome, including 153 million nucleotide base pairs and 1,669 genes [26]. This representation is in stark contrast to the similarly sized chromosome 7, which holds 120 published associations [22]. There are a number of reasons for this discrepancy, including reduced power due to the use of sex-specific analyses or low data quality, reflecting poor genotyping accuracy on current genome-wide arrays and quality control issues. However, an important reason is also that although the X chromosome is routinely included in standard genotyping arrays, it is simply excluded from analysis. Indeed, of the first 53 studies that have released publically available GWAS data, only 31 include the X chromosome [22]. Since the number of copies of X chromosome variants is confounded with sex, special analysis methods and computational tools are required (discussed below); therefore the sex chromosomes have historically been excluded to simplify analyses, setting a critical precedent that needs revision.

Figure 2.

Plot of the number of published GWAS associations studies reported in the NHGRI catalog by number of genes. Each dot represents a pair of autosomal chromosomes. The `X' and `Y' chromosomes are plotted in red.

Important issues regarding the genetics of sex differences are not limited to large scale GWAS studies in humans but also are critical concerning preclinical work involving model organisms. In particular, basic science research utilizes experiments in model organisms as critical tools to investigate human disease. However, the genetics and biology of the sex chromosomes differs from species to species regarding genes that are conserved and mechanisms of dosage compensation [8, 15]. Therefore the choice of model organism may be critical to the study of sex differences. The mouse has become a favorite model organism for preclinical disease research because many genes are shared with humans, although the two species also differ in important areas, highlighted by the work of the Mouse ENCODE Consortium [27]. Mice have fewer conserved X-Y gene pairs [8] and fewer genes that escape X inactivation as compared to humans [21, 28], which may suggest that they may NOT be an ideal model organism to study the genetics of sex differences for all conditions in humans.

Genetics of cardiovascular disease

General considerations

Cardiovascular diseases are complex traits with both sex-specific (i.e. cardiovascular diseases associated with pregnancy, erectile dysfunction) and sex-different in etiology, pathophysiology, symptom presentation, severity and progression [3, 29, 30]. The concept that there are sex by trait interactions is not new [31] and several methods are used to examine genetic components of complex traits including GWAS, linkage analysis, association studies, and candidate-gene association studies. However, not all studies of populations of men and women account for sex in the analysis and if sex is considered, it is treated only as a confounding variable [32–35]. However, when sex and sex chromosomes are considered, there is the potential to gain new insight into etiology of disease, differences in mechanisms of progression and new targets for preventive and therapeutic interventions [36]. Failing to do so, hampers medical progress toward optimizing individual outcomes.

Cardiovascular risk factors

Major risk factors for cardiovascular disease include smoking, high blood pressure, high glucose, central adiposity and high blood lipids. While some psychosocial, economic and environmental conditions as well as life-style choices (i.e. access to health care, stress, smoking, diet, and activity) have physiological consequences, it is important to determine which biological factors (regulation of blood pressure, glucose, and lipid metabolism) reflect inherited components in order to optimize preventive and therapeutic strategies.

Hypertension

Men develop hypertension at younger ages than women, thus carrying an increased life-long burden of disease including the risk of stroke [37–40]. Recall that sexual dimorphism is impacted by the sex chromosomal complement independent of hormonal influences. The Sry locus of the Y chromosome, in addition to being required for the development of the testes, was found in experimental animals to regulate tyrosine hydroxylase, a critical enzyme in the synthesis of norepinephrine [41]. Greater activity of tyrosine hydroxylase in men would pre-disposes them to hypertension differently than women [42] and indeed, several large genetic studies have identified loci on the Y chromosome associated with hypertension in men (for review see, [43]).

In addition to synthesis of adrenergic transmitter, genetic variants in alpha and beta adrenergic receptors affect vascular tone. In women with non-obstructive coronary disease, genetic variants in adrenergic receptors associate with resting hemodynamics and increased risk for stroke, myocardial infarction and heart failure [44–46]. Polymorphisms in the genes encoding enzymes and receptors for the renin-angiotensin system would influence development of hypertension in both women and men but these polymorphisms in women may predispose them to pregnancy related hypertension [47]. Genetic polymorphisms in these receptors and enzyme systems also would affect a woman's response to pharmacological agents that are used in the treatment of hypertension, i.e. beta and alpha adrenergic antagonists, angiotensin receptor antagonists, and angiotensin converting enzyme (ACE) inhibitors. There are no sex-specific guidelines for treatment of hypertension [48, 49].

Inflammation

The Y chromosome also includes genes involved with macrophage activation affecting innate immunity [50, 51]. Systemic inflammation characterized by activation of macrophages and increases in circulating levels of various cytokines contributes to development of atherosclerosis [52]. Thus, in males, genetic risk for development of other cardiovascular disease apart from hypertension may be imparted simply by inheriting the Y chromosome.

Genes on the X chromosome also affect phenotypic expression of inflammatory risk factors, including genes associated with apoptosis, lipid oxidation, and generation of oxygen-derived free radicals produced by the mitochondria. In an analysis of RNA levels of 683 genes on the X chromosome obtained from individuals at specified time intervals following a stroke, X related genes showed greater and differential up regulation in females compared to males. In females, up-regulated genes were associated with post-translational modification of proteins, small molecule biochemistry and cell-cell signaling; in men, up-regulated genes were associated with cellular migration, cell trafficking and cell death [36]. This study provides evidence that cellular pathways involved with manifestation of a particular cardiovascular outcome, in this case stroke, differs between men and women.

Lipids

Much research and clinical preventive strategies have focused on evaluating and managing serum lipids. Indeed, decreases in high density lipoprotein (HDL) cholesterol and increases in low density lipoprotein (LDL) cholesterol and total cholesterol are considered hallmarks of cardiovascular risk. Sex steroids regulate serum lipids and in women, genetic variants in genes for either estrogen receptor α or β associated with serum lipids, but the associations varied by ethnic groups [53]. The extent to which lipid metabolism shows a sex-difference independent of sex steroids is unclear including mechanisms of lipid deposition in vascular lesions. Male pattern of lipid deposition results in obstructive lesions within the coronary arteries while in women the pattern is diffuse and manifests as non-obstructive ischemic disease [54].

In a genome wide association study, genes of the metabolic pathways for HDL, LDL and triglycerides associated with obstructive coronary artery disease [55] and with serum lipids in healthy volunteers [56]. While the details of these reports are beyond the scope of the present review, several key points serve as an example relevant to studying the genetics of sex differences. In these studies, there was large variability in the number of SNPs among the genes studied and some of the variants were specific to either Caucasian or minority ethnic groups. In these studies, sex was not considered in the analysis. In a genome-wide analysis of loci affecting lipid metabolism conducted in samples from approximately 20,000 subjects from European countries, three loci showed sex-specific effects. The strongest effects were in the gene for 3-hyroxy-3-methylglutaryl-Coenzyme A reductase (HMGCoA) and in the gene that encodes chondroitin sulfate proteoglycan (NCAN). The former enzyme is the rate-limiting step in synthesis of cholesterol (greater effect in females, P=0.001) and a target for statin therapy; the latter enzyme is involved in cell adhesion and migration (greater effect in males P=0.002). Both these loci associated with total cholesterol. The gene encoding lipoprotein lipase (LPL) showed greater association with HDL in males compared to females (P= 0.006) [57]. The age distribution among participants in that study was broad (18–104 years). It is unclear at this time whether these associations would remain if hormonal status was considered in females as estrogen modulates lipid metabolism.

Two meta-analyses of genome-wide association studies evaluating loci for lipid metabolism with coronary artery disease, sex-differences, perhaps not surprisingly, were confirmed in genes regulating total cholesterol, LDL, HDL and triglycerides [58, 59]. Both of these studies identified stronger associations of SNPS to serum triglycerides in women compared to men. Because associations in one analysis were based on those that also associated with coronary artery disease, both women and men in those studies would most likely have phenotypic obstructive disease [59].

Statin therapy, which inhibits HMGCoA, has become a main approach to primary prevention of coronary artery disease and stroke. The safety and efficacy of this treatment in a meta-analysis of 27 clinical trials found the treatment of similar effectiveness in women and men who had equivalent cardiovascular risk [60]. The number of women included in this analysis was less than 30%. As stated above, genetic polymorphisms in the gene for HMGCoA had greater association lipid levels in women compared to men. Genetic polymorphisms in that gene associate with development of type 2 diabetes [61] that was shown to be increased with statin use in the Women's Health Initiative [62]. Although there are no sex-specific guidelines for use of statins in primary prevention strategies for cardiovascular disease, the appropriateness related to “equivalent risk” is recommended [63, 64]

Metabolic syndrome

The term “metabolic syndrome” is used to define a constellation of at least three of five risk factors for cardiovascular disease: waist circumference.(>88 cm for female and 102 cm for male), high-density lipoprotein cholesterol (HDL-C) <50 mg/dL, triglyceride level >150 mg/dL, fasting blood glucose >100 mg/dL, and systemic blood pressure (systolic blood pressure ≥130 or diastolic blood pressure ≥85 mmHg)[65]. Although two individuals with “metabolic syndrome” may exhibit a different combination of three of the five characteristics [66], the collection poses greater risk for cardiovascular disease than any characteristic separately. Not surprisingly there are sex differences in incidence and prevalence of metabolic syndrome (independent of the target values to define the syndrome) [67] due to sex differences in incidence of hypertension and hyperlipidemia. Abdominal obesity shows sexual dimorphism distinct from body mass index. In women, abdominal obesity (as measured by waist circumference) is gaining acceptance as a measure to be considered in assessing cardiovascular risk, but this measure may be influenced by race/ethnicity [68, 69]. Data from a genome-wide association study of individuals of European and European-American descent to evaluate phenotypic traits of height, weight, body mass index, waist circumference, hip circumference and wait-to-hip-ratio confirms these assertions as only sex-differences in loci for waist phenotypes were found [70]. Of 7 loci associated with waist phenotype, all were found in women and not men. Of particular interest is the variant in the gene for PPARG encoding peroxisome proliferator-activator receptor γ (PPARγ), a type II nuclear hormone receptor regulating adipocyte-differentiation, susceptibility for obesity and insulin resistance. PPARγ binds to a site in HSD17B4, which encodes the enzyme hydroxysteroid (17-beta) dehydrogenase involved in the conversion of estrogen to estrone and beta-oxidation of fatty acids [70]. As statins also target PPARγ, and as there are documented interactions with estrogen and endocrine disruptors in regulation of PPARγ [71–73], sex differences observed with statins and type 2 diabetes may be related to genetic variants in PPARγ [74]. Collectively, these data identify the interaction between sex and the genetic components of a phenotype associated with sex differences in adiposity, and demonstrate the utility of pathway analysis to assess sex differences in genetic components of cardiovascular risk susceptibility.

Central adiposity is functionally linked to elevated glucose, insulin resistance and type 2 diabetes. Diabetes poses about a 3 to 6 times greater risk for myocardial infarction in women compared to men across several ethnic groups [67]. Numerous studies have identified candidate genes involved in development of type 2 diabetes although discussion is beyond the scope of this review. However, one study utilizing a meta-analysis of genetic variants on the Metabochip identified two sex-differentiated variants in combined case-control populations of European and Pakistani origin: CCND2, which encodes cyclin proteins involved with cell division, was most significant in males and GIPR, which encodes G-protein coupled receptor for gastric inhibitory polypeptide that when activated stimulates release of insulin in response to elevated glucose, was most significant in females [75]. However, in the subsequent pathway analyses, sex was not considered and the constellation of pathways may interact in different ways in females compared to males, given the other interactions among lipid metabolism, glucose, metabolism and adiposity [75].

Using a metabolomics approach to investigate genetic regulatory networks associated with complex disease traits, 102 of 131 serum metabolites showed sex differences in expression from more than 3,300 population based samples (Cooperative Health Study in the region of Augsburg, KORA). A sex-specific GWAS identified a locus of the carbamoyl-phosphate synthase 1 gene (CPS1) for glycine, a key component in amino acid synthesis [76]. A sex-based analysis revealed sex specific SNPs that have a greater effect on glycine in females compared to males. CPS1 encodes a mitochondrial enzyme involved in hepatic nitrogen urea metabolism and synthesis of arginine, a precursor to formation of nitric oxide, an endothelium-derived relaxing factor the absence of which is considered to contribute to endothelial dysfunction and progression of vascular disease. These studies underscore the ability to analyze existing genetic data sets by sex and their potential to provide a better understanding of mechanisms of cardiovascular disease that may extend to understanding sex-specific risks such as gestational diabetes and insulin resistance associated with polycystic ovarian syndrome (PCOS) in women [77].

Cardiovascular anatomical pathologies

As non-invasive imaging techniques are used to evaluate the presence and progression of asymptomatic vascular lesion characteristic of cardiovascular disease, genetic components of these complex traits warrant mention.

Carotid intima-medial thickness

CIMT is a quantitative trait showing some degree of heritability in different ethnic populations [78]. Using candidate gene analysis, variants associated with CIMT cluster into several gene classes: hemostasis, extracellular matrix remodeling, endothelial function, the renin angiotensin system, inflammation and antioxidation [79]. In recently menopausal women, a candidate gene analysis of genes related to the anticoagulant, procoagulant, fibrinolytic, and innate immunity pathways, only variants in genes of the innate immunity pathway associated with CIMT [80]. These women had endogenous estradiol levels <40pg/ml, and it is unknown if these associations may change with use of menopausal hormone treatments.

Genetic contributions to carotid thickening differ along segments of artery, which may reflect differences in exposure to physical forces at the carotid bifurcation [78]. Vascular stiffness in women but not men was nominally associated with variants in endothelial nitric oxide synthase (NOS3) gene [81]. NOS3 is modulated by estrogen, thus, phenotypic expression of NOS3 variants should vary by hormonal status. With an average age of study participants of 62 years, most of the women would have been postmenopausal and estrogen deficient (see section below).

A meta-analysis of GWAS studies from the Cohorts for Heart and Aging Research in Genomic Epidemiology (CHARGE) consortium consisting of nine-community-based studies (age 44–76 years) identified four genetic variants associated with CIMT: a transcriptional factor (rs11781551), APOC1 (rs445925) associated with dyslipidemia, a telomerase inhibitor involved in chromosomal segregation in mitosis (rs6601530) and a regulator of uric acid (rs4712972). Contrary to these results, another GWAS study of the Framingham community failed to identify genes associated with increases in CIMT, in spite of large sample sizes and equal representation of men and women [34]. Neither of these studies stratified analysis by sex. However, overall heritability for intima-medial thickness of the internal carotid artery was higher for men than women, whereas heritability for intima-medial thickness in the common carotid artery was greater for women than men [40]. Analyses that do adjust for but do not stratify by both age and sex preclude identification of sex-specific genetic variance contributing to progression of CIMT. For example, apolipoprotein E4 (APOE4) associated with CIMT in several studies [78] but this gene may have sex-specific considerations, since it is a more critical risk factor for Alzheimer's disease in women compared to men [82]

Several studies suggest sex-specific genetic factors contribute to CIMT. CIMT is an independent predictor of stroke [83] and incidence, prevalence and outcomes of stroke show sex disparities. One sex specific analysis examined the association between CIMT and genetic variants of the phosphodiesterase 4D (PDE4D) gene, a susceptibility gene for stroke where CIMT would provide a phenotype for stroke risk with reduced complexity [84]. One homozygote variant had a significant odds ratio for increased CIMT in the overall study population. However, when these data were analyzed by sex, the odds ratio remained significant for men but not women [84]. PDE4E metabolizes cyclic AMP, a key signaling molecule involved with prostaglandin signaling. As this variant also correlated with stroke in young men, it is possible that the lack of efficacy of aspirin to reduce stroke risk in men may be related to this signaling pathway [85].

Other genetic variants showing sex specific associations with subclinical atherosclerosis measured by CIMT include variants in the gene for interleukin-6, an inflammatory cytokine, associated with blood pressure and CIMT in Japanese women but not men [78], for cyclin-dependent kinase inhibitor 2A (CDKN2A), a negative regulator of cell proliferation, which showed association with CIMT in men, while the genetic variation in the gene for oxidized low density lipoprotein receptor 1 (OLR1) showed association with CIMT only in women.

Coronary calcification

Prevalence of coronary calcification is greater in men than women and varies by ethnic group [86]. Cardiovascular disease risk factors such as hypertension, hypercholesterolemia and type 2 diabetes can exacerbate accumulation of calcium in the coronary arteries. For example, in individuals at high risk for hypertension, coronary calcification is associated with genes regulating collagen formation, allograft inflammatory factor 1 and bone morphometric protein receptor type 1A [35]. Other genes implicated in coronary calcification are those for matrix metalloproteinase 9 (MMP9) [87] matrix Gla protein (MGP) and osteopontin (OPN) [88]. How these genetic associations might differ with hormonal status (e.g. estrogen) and risk for calcification in women is unknown.

In diabetics, two SNPs of the gene encoding the inflammatory signaling molecule CD40 were negatively associated with coronary calcification in a candidate gene study in European Americans [89]. However, a GWAS of African Americans did not uncover any genetic variants significantly associated with coronary calcification [90]. This negative finding may reflect failure to examine sex-specific effects or consider that multiple genetic pathways converge to promote calcification. Clinically relevant calcification also can exist in persons without significant conventional cardiovascular risk factors. For example, in women being screened for the Kronos Early Estrogen Prevention Study (KEEPS), about 14% were excluded because of the presence of clinically relevant coronary calcification (Agatston score >50) [91]. Women included in KEEPS had Agatston score <50 with most having no calcium in their coronary arteries. However, in those women who had Agatston score >0 but <50, a candidate gene study associated variants in genes related to innate immunity, elastin remodeling and post-translational modification of proteins with coronary calcification [80]. Because these women were estrogen deplete, it is unknown if these genetic variants would associate with calcification following the use of menopausal hormone therapy, as use of conjugated equine estrogen slowed coronary calcification in women of the Women's Health Initiative [92].

Confounding variables: ethnicity and hormonal status

Ethnicity

Although the genetics of ethnicity are not considered in depth in this review, some mention is warranted given that development of cardiovascular disease is influenced by ethnicity/race [33, 39, 79, 93, 94]. As indicated in the preceding sections of this review, some studies account for race and ethnicity with specific designations, e.g. Caucasian of European ancestry, Mexican-American, Han Chinese, Japanese, Dominican Republic, etc. However, studies of populations which may consist of mixed racial origin, fail to adjust for population stratification in genetic analyses. For example, studies of individuals identifying as “Hispanic” are by self-report, unrelated individuals from Northern Manhattan [93] or families of specific background (i.e. the Dominican Republic or Mexican-American) [39, 79, 94]. The type of analyses varied among studies but none studied the effect of ancestry. Self-reported ethnicity is not necessarily the same as the genotypic ethnicity [80]. For instance, “Hispanic” designation from the USA West Coast is most likely to be a mixture of Asian, Native-American and European ancestry, whereas “Hispanic” from the USA East Coast is most likely to be a mixture of Native American, European and African ancestry (Figure 3). Therefore, studies that evaluate ethnicity/race as a contributor to cardiovascular risk require attention to the geographical origins of ancestry using genetic data [95]. A validation study of susceptibility loci for coronary artery disease is a case in point [96]. Although 4 loci associating with coronary artery disease were common between the Han Chinese population and European populations, 4 were unique to the Han Chinese (case control study and the number of males or females was not included in the report) [96]. Due to globalization, persons with multiracial and multiethnic backgrounds are increasingly common place; therefore, consideration of ancestral components of disease may become more relevant to understanding inherited cardiovascular risk. In the era of pharmacogenomics, biological factors will drive response to treatment while self-report of ethnicity may be driven by socio-political factors independent of biological factors.

Figure 3.

Ancestry informative markers (492) were used to stratify participants of the Kronos Early Estrogen Prevention Study (KEEPS) by ethnicity using the program STRUCTURE [80]. The STRUCTURE program uses ancestry informative markers that are evenly distributed across the genome to identify the ancestry of each subject by using the three HapMap 2 populations (CEU: Utah residents from the CEPH; YRI: Yoruba from Ibudan, Nigeria; and CHB/JPT: Han Chinese from Beijing, China, and Japanese from Tokyo, Japan) or the 1000 Genomes populations [121]. For KEEPS, CHB/JPT represents 89 HapMap Asian samples, URI represents 60 HapMap African samples and CEU represents 60 HapMap Caucasian samples. Each circle represents an individual; the colors represent self-identified ethnicity. E and W represent site center as from either the east coast or west coast of the United States, respectively. Women who self-report as USA Hispanic from the West Coast are a mixture of Native Americans and European Ancestry, while those from the East Coast were a mixture of African and European Ancestry.

Hormonal status

Nuclear receptors for the sex steroids affect gene transcription. Therefore, if a gene carrying a particular SNP has a sex steroid receptor in the promoter region, the relative contribution of that SNP to expression of the gene product will be affected by 1) hormonal status; 2) genetic variants in receptors for sex steroids; 3) genetic variants in enzymes required for metabolism (synthesis or degradation) of sex steroids.

The first studies suggesting that hormonal genetic variant accelerate development of atherosclerosis was in a man with a deletion of estrogen receptor alpha [97, 98]. As estrogen modulates endothelial nitric oxide synthase, this pathway is considered an important mechanism by which estrogenic hormones slow progression of atherosclerosis. Variants in the gene for estrogen receptor alpha also associated with levels of high density lipoprotein cholesterol and some inflammatory cytokines in women using estrogen for secondary prevention of cardiovascular disease [99, 100]. However, neither variants of estrogen receptors alpha nor beta significantly associate with cardiovascular events in women of the Women's Health Initiative where the use of estrogenic treatments was used to evaluate primary prevention of cardiovascular disease [101]. There were no consistent relationships between genetic variants in either estrogen receptor alpha or beta and HDL levels in perimenopausal women of various ethnicities [53]. Collectively, these data suggest that perhaps variants in estrogen receptor genes impact cardiovascular disease to a greater extent in men compared to women.

The gene for the androgen receptor, which resides on the X chromosome, contains a functional polymorphism consisting of highly variable numbers of CAG repeats. Given the mosaic pattern of X inactivation in women, variants in this gene will have a greater impact in males than in females. There is significant heterogeneity in the CAG repeats among ethnic groups [102] which may explain, in part, the inability to detect an association between these repeats and heart disease in Caucasian men [103].

These CAG repeats in hypothalamic/pituitary regions of the brain have a small effect on the feedback control for testosterone production in men such that the higher number of repeats, the higher the serum free testosterone levels [104]. These repeats are associated with male infertility. In addition, they may impart some cardiovascular risk as low numbers of repeats associated with abdominal adiposity and higher sympathetic modulation of vascular tone in boys than in girls. Data are conflicting regarding the association of CAG repeats with serum levels of high density lipoprotein cholesterol (HDL) and development of type 2 diabetes in men [105]. In women, CAG repeats associate with polycystic ovarian syndrome, characterized by male-pattern adiposity and propensity for development of type 2 diabetes [106, 107].

In addition to hormonal receptors, there are genetic variants in enzymes involved with synthesis, conversion and degradation of the sex steroids. However, little attention has focused on these variants with hormonal modulation of cardiovascular disease processes in men or women. For example, SULT1A1, an ubiquitous enzyme that is the predominant sulfotransferase present in the liver, is highly polymorphic including copy number and multiple single nucleotide polymorphisms that affect enzyme activity [108, 109]. Thus, a gene “dose” effect of SNP by copy number could affect response to various hormonal therapies impacting vascular function including endothelium-derived nitric oxide. In addition, biological effects of SNPS on this enzyme seem to vary by ethnicity as a SNP in the promoter region decreased SULTA1A1 activity in African Americans but increased activity in Caucasians [110]. Variation in genes for other enzymes in the estrogen metabolic pathway and their association with menopausal vasomotor symptoms also show differences among ethnic groups [111–113]. However, studies linking menopausal symptoms and genotypes to cardiovascular disease are lacking.

Hormonal status affects phenotypic consequences of SNP expression. For example, in unpublished data from our group, the significance of an association of the G allele in SNP of interleukin regulatory factor 4 with change in carotid intima medial thickness in women of the KEEPS trial varied depending on the treatment assignment (Figure 4).

Figure 4.

Association of allele variants for the SNP rs7768807 of the gene from interleukin regulatory factor 4 involved in leukocyte function with changes in carotid intima-medial thickness varied by the type of hormone treatment in women of the Kronos Early Estrogen Prevention Study (KEEPS) compliant to treatment with either placebo (PL), transdermal 17β estradiol (E2) or oral conjugated equine estrogen (oCEE). Minor allele frequency for rs7768807 is 27%.

These observations point to several important considerations in evaluating genetic contributions to cardiovascular disease. First, attention should be given to ethnicity verified by genetic data rather than self-identification. Second, since hormonal status varies across the life span in both men and women, hormonal status, age and sex should be considered when evaluating specific genetic contributions to cardiovascular risk. Finally, studies to examine genetic basis of hormonal influences on cardiovascular function should continue given the on-going use of hormonal products by women and increased use of exogenous testosterone by men.

Statistical considerations for future studies

As genetic studies go forward, there are several important considerations that need to be incorporated into analyses. For example, population geneticists study the distribution and changes of allele frequency in a given population due to natural selection, genetic drift, mutation and gene flow by taking into account recombination, population subdivision and population structure. Therefore, in genetic studies of admixed populations, it is relevant to assess if the self-reported ancestry is the same as genotypic ancestry. As discussed above, this can be accomplished by calculating the proportion of ancestry derived from each geographical region using ancestry informative markers (see Figure 3).

It is also important to consider the statistical analysis method in genetic association studies because of possible confounding of the number of X chromosomal variants by sex. Univariate linear or logistic regression or standard chi-square tests are typically used for the autosomes and these tests can still be applied for the pseudoautosomal regions of the X and Y chromosomes. However, different analysis methods are necessary when analyzing X chromosome variants outside of this region. For the regions undergoing X-inactivation, the statistical analysis should account for this process.

However, some methods to account for X-inactivation have not yet been adopted by the research community because they are not readily available in statistical software packages. Those currently available include PLINK for genome-wide analysis [114] and R/Bioconductor package `snpMatrix' (http://www.bioconductor.org) [115] which both allow for statistical tests of X chromosome variants, and IMPUTE2 [116], which allows for genome-wide imputation of untyped/missing variants on the X chromosome.

Most methods to analyze X chromosome data have been developed and evaluated univariately, rather than considering multivariate models with multiple genetic factors. As complex trait etiology likely involves a large number of genetic and environmental components, this is a critical research area. Furthermore, studies of the genetics of sex differences should not be limited to variants on the sex chromosomes. Autosomal genetic effects can also be modified by sex, and gene-sex interactions effects may underlie observed sex differences in disease traits. Interactions between autosomal variants and sex may be surrogates for gene-gene interactions between variants on the autosomes and the X or Y chromosomes, as well as gene by hormone interaction effects [117]. In fact, many autosomal sequence variants that are associated with gene expression (eQTLs) display sex differences, and such sex-biased effects on expression are observed for important regulatory genes [118]. This suggests that studies of gene-sex interactions may be critical tools to explore the genetics of sex differences. Recently many statistical methods have been developed to explore gene-environment interactions [119, 120]. Such methods could be applied to study sex-specific genetic effects, and represent an important opportunity for future research especially related to cardiovascular disease.

Summary/Conclusions

Cardiovascular diseases represent complex genetic traits with phenotypes being influenced by genetic, hormonal, environmental, and cultural variables. Although many studies have attempted to gain insight into etiology of various cardiovascular diseases and their risk factors through genetic analyses, most have ignored the basic genetics contributed by the sex chromosomes and have failed to conduct sex-stratified analyses, despite emerging tools to study the genetics of sex differences for complex traits. Furthermore, the confounding effects of ancestry and hormonal status across the lifespan also need to be considered. Given that sex differences in incidence, prevalence, morbidity and mortality from cardiovascular diseases are documented around the world, sex must be considered in future investigations of genetic components of the disease if preventive, diagnostic and treatment strategies are to be individualized to optimize outcomes.

Highlights.

• Incidence and prevalence of and morbidity and mortality from cardiovascular diseases differ between men and women.

• Cardiovascular diseases are complex traits.

• Genetic analysis of cardiovascular disease traits often fail to include the sex chromosomes.

• Hormonal status and ethnicity are confounding variables in sex-based genetic analyses.

• Understanding genetic contributions to various cardiovascular diseases will be advanced by considering the interactions among genetic sex, genetic race and hormonal status.