1 Introduction

Hypertension is a major risk factor for Cardiovascular Disease (CVD); in turn CVD is the underlying cause of more than a quarter of deaths in the UK [1]. There are no validated tests that can identify early in the disease process which individuals will develop hypertension-mediated organ damage. Dysfunctional vascular traits represent key pathophysiological processes in the development of hypertension and cardiovascular disease, with both inherited and reversible elements. These traits include stiffness of the large arteries, microvascular abnormalities, endothelial dysfunction, and atherosclerosis, phenotypes often apparent prior to established hypertension or organ damage. Hence the interest in measuring vascular function, and in understanding the relationship between measurement techniques and hypertensive phenotypes, including the relative influence of participant sex and genetics. We explore this topic with reference to large scale genomic studies, population-based cohorts, and candidate gene studies.

1.1 Definitions for the Non-expert

Genome: complete set of genes in an organism including introns (non-coding sequences) and exons (coding sequences).

Genome-wide association study: entire genome surveyed for genetic variants occurring more frequently in cases than in controls.

Candidate gene study: specify fewer variants of interest a priori, and aim to establish if a disease association can be confirmed.

Epigenetics: genetic modification without mutations of the DNA sequence; occur in normal development or induced by environmental factors.

Exome: complete set of exons present in an organism which accounts for all the coding regions of genes present.

SNP: single nucleotide polymorphism—DNA sequence variations with a single nucleotide (adenine, thymine, cytosine, or guanine) in the genome sequence altered.

Common variants: SNPs with minor allele frequency (MAF) of greater than 1%, accounting for over 90% of genetic variation between individuals.

Mendelian Randomization: method of using measured variation in genes of known function to examine the causal effect of a modifiable exposure on disease.

2 Assessment of Vascular Function and Disease

Cardiovascular outcome measures in clinical trials generally relate to coded events such as myocardial infarction or death; alternatively, research trials may employ surrogate markers such as vascular stiffness and endothelial dysfunction—early functional traits known to be predictors of more advanced structural changes and development of cardiovascular disease. Assessment techniques quantifying such traits reflect different aspects of vascular health, assessed in the European Society of Cardiology Working group position paper [2]. First, carotid ultrasound to measure intima-media thickness (CIMT) has clinical utility in diagnosing carotid atherosclerotic vascular disease [3]–[5], but is also linearly associated with blood pressure (BP) [6] and adds prognostic value in the prediction of cardiovascular events and mortality, see Sect. 3.1 [7, 8]. Second, pulse wave analysis (PWA), PWA-derived augmentation index (AIx), and carotid-femoral pulse wave velocity (cfPWV) assess for arterial stiffness, a process characterised by functional changes and structural remodelling within the arterial wall, with associated fibrosis and calcification. These measures of arterial stiffening are independent and reliable predictors of hypertension, myocardial infarction and stroke [9]–[11], with meta-analyses of individual patient data showing the alternative brachial-ankle PWV method also associated with cardiovascular complications [12], and stiffening of the carotid artery with incident stroke [13]. The predictive strength of arterial stiffness is, however, greater in subjects with an established cardiovascular risk [14]. Finally, endothelial function refers to its’ ability to detect physical (shear stress) and biochemical signals, and respond through expression of surface molecules and production of vasoactive and inflammatory mediators. Endothelial dysfunction precedes structural micro-circulatory changes. Hypertension can be both cause and consequence of microcirculatory dysfunction, closely tied to peripheral vascular resistance, with vascular tone in turn regulated by many systems (sympathetic nervous system, endocrine, and local autoregulation), each with polygenic influences [15]. Endothelial function can be assessed using ultrasound of the brachial artery with ‘flow-mediated dilation’ (FMD), dilation predominantly mediated by nitric oxide release from endothelial cells. Alternatively, peripheral arterial tonometry (PAT), commonly quantified by the Endo-PAT2000 device (Itamar Medical) also assesses microcirculatory and endothelial function by measuring arterial tone or ‘hyperaemic response’ in the fingertips in response to proximal occlusion. EndoPAT-2000 device also generates an augmentation index adjusted to a heart rate of 75 bpm (AI@75), similar to PWA but derived from peripheral vessels. Hence, these techniques not only reflect different aspects of the pathophysiology of hypertension and cardiovascular disease but may aid identification of different hypertensive phenotypes [16]–[18]. They are also well accepted as being influenced by age, BP, and sex; factors that should be accounted for when comparing techniques. Less well defined are the effects of underlying genetic differences, i.e. the inherited component, or ‘heritability’ of data pertaining to techniques measuring vascular health. Genotypic effects on these vascular assessment tools are myriad, but key checkpoints where influence may be hypothesised include vascular endothelial cell sensitivity to extracellular stimuli, intra-cellular signalling cascades, and effects on transcription, ultimately influencing production of vasoactive substances, vascular tone, and remodelling.

3 Genetics of Hypertension

Familial and twin studies estimate that the heritable component of BP lies between 22 and 65% [19]–[22]. BP is a complex trait with no single gene playing a dominant role; instead multiple genes demonstrate minor additive effects. These genes encode for a variety of proteins, ion channels, receptors, and enzymes involved in endocrine, cardiac, renal, vascular and neural systems that influence BP regulation. This complexity is illustrated by the heterogeneity of underlying pathology in the (rare) monogenic cases of secondary hypertension, examples of which are discussed in Sect. 2.1.1. Other genes are identified only by genome wide association studies (GWAS); an illustrative example follows in Sect. 2.1.2.

3.1 Single Gene Disorders

Monogenic causes of hypertension are rare and mechanisms varied. For example, children with homocystinuria and familial hypercholesterolaemia develop premature atherosclerosis and early endothelial dysfunction [23]; in AD glucocorticoid-remediable aldosteronism, chimeric genes encoding steroid 11ß-hydroxylase (CYP11B1) and aldosterone synthase (CYP11B2) lead to aldosterone regulation by ACTH rather than angiotensin II [24, 25], salt and water retention, and elevation in BP [26]. Finally, AD hypertension with brachydactyly syndrome results from a gain of function mutation in PDE3A, encoding phosphodiesterase 3A and resulting in cerebral vascular anomalies and baroreceptors hypersensitivity [27]. For an in-depth review of monogenic hypertensive syndromes, we would highlight Burrello et al. [28].

3.2 GWAS

GWAS have identified multitudes of genetic loci associated with BP, covered in-depth elsewhere [29]. For example ATP2B1 encoding PMCA1, a plasma membrane ATPase expressed in vascular endothelium and involved in calcium pumping from the cytosol to the extracellular compartment. GWAS can, however, be susceptible to false positive associations if statistical analysis lacks rigour, if the panel fails to reflect genomic variation, or the study lacks statistical power; points to remain cognizant of.

3.3 Epigenetics

Processes of epigenetic modification include methylation, post-translational histone modification, and small non-coding RNAs. HSD11B2 gene promoter methylation for example has been associated with hypertension onset [30, 31]; acetylation meanwhile promotes gene transcription of NOS3 (eNOS) and other genes affecting vascular tone and salt and water homeostasis [32, 33]. Finally, small non-coding RNAs (miRNA) may conversely downregulate genes by binding the corresponding mRNA resulting in repression of translation [33]. Population-based studies further support the role of epigenetics in hypertension [34].

3.4 Sex and BP Genetics

The male–female difference in BP, vascular traits, and CVD is complex. Mediating factors include X and Y chromosome differences, sex-hormone influences, renin–angiotensin–aldosterone system divergence [35], societal and behavioral impacts, and even epigenetic differences, with females receiving genetic imprints from each parent’s X chromosome, random X inactivation leading to further genetic heterogeneity. Gene-by-sex interactions, and age (menopause)-dependent effects further complicate interpretation.

3.5 Summary

Bringing together the evidence of different phenotypes of hypertension [16,17,18, 36, 37], determined by pathophysiology but characterised by the aforementioned vascular traits; and considering the exponentially increasing data regarding hypertension risk alleles; it becomes important to explore the genotypic and sex associations with vascular techniques used to measure these hypertensive phenotypes.

4 Carotid Intima-Media Thickness

4.1 Heritability

A number of studies have reported heritability estimates for CIMT, though with disparate estimates (21 to 65%) despite similar adjustment for covariates, see Fig. 1 and Table 1 [38, 39]. Sacco et al. for example report 65% heritability in 100 Dominican families (1390 individuals, 61% female, mean age 46 years) after adjustment for age, sex, smoking, and BMI [39]; Cecelja et al. estimate age-adjusted heritability at 49% (95% CI 17–63%) in 762 females of the Twins UK cohort with mean age 58 ± 9 years [40]; whilst only 35% heritability is reported by Sayed-Tabatabaei et al. [41] in their assessment of 930 individuals connected in a single pedigree from an isolated population (participants of the Erasmus Rucphen Family study).

Fig. 1
figure 1

Evidence regarding heritability of techniques assessing vascular function

Full size image
Table 1 Evidence regarding heritability of techniques assessing vascular function
Full size table

4.2 Genes

GWAS identifies numerous genetic loci as having possible significance, and studies of candidate genes approximating to these loci have also been widely reported (Table 2); 16 of the 32 identified (50%) also have evidence of association with BP traits. Figure 2 demonstrates that many genes have a role vascular remodelling, such as MMP9 [42] encoding a gelatinase targeting type IV collagen and gelatin; CXCL12 involved in endothelial and epithelial cell proliferation and migration [43]; and VCAN [44] which encodes chondroitin sulfate proteoglycans (extracellular matrix components), thus regulates cell proliferation, differentiation, and survival [45].

Table 2 Gene polymorphisms relating to techniques measuring vascular health, with consideration of sex differences and heritability estimates
Full size table
Fig. 2
figure 2

Gene polymorphisms relating to techniques measuring vascular health, with genes grouped according to function. Based on data in Table 2

Full size image

4.3 Interactions

Other genes demonstrate the importance of gene-by-environment interactions in determining CIMT, for example MCPH1 encodes a damage response protein regulating cell cycle [44]. Similarly, gene–gene interactions are apparent, for example genes involved in cholesterol biology and inflammation where high-density lipoprotein composition is altered in an inflammatory state, with apolipoprotein-A-I and –A-II displaced by Serum amyloid A (SAA). SAA SNPs rs2468844 and rs12218 alter binding affinity of SAA proteins [46, 47], with implications for reverse cholesterol transport, CIMT, plaque formation [48], and plaque stability [49].

5 Vascular Stiffness: Pulse Wave Analysis and Pulse Wave Velocity

5.1 Heritability

Genes are estimated to account from 26 to 43% of the variability in vascular stiffness as measured by PWV (see Table 1, Fig. 1), with data derived from both population and twin studies [40, 41, 52, 53]. For example, the Georgia Cardiovascular Twin Study of 388 twins (41% black; 49% male) aged 12–30 years; report 53% (42–62%) heritability for dorsalis pedis (foot) PWV [53], with no sex or race differences; additionally, the aforementioned Twins UK cohort of 762 females, mean age 58 ± 9 report heritability estimate of 38% (95% CI 16–63%) after adjustment, with annual progression interestingly demonstrating higher adjusted heritability estimates of 55% (31–64%) over 5 years follow-up [40].

5.2 Genes

Many studies of the genetics of arterial stiffness focus on parameters other than PWV, such as pulse pressure and forward and reflected wave amplitude, covered in detail elsewhere [142, 143]. Looking specifically at PWV as the most commonly used technique, GWAS of 644 individuals involved in the Framingham Heart Study did not find any variants achieving genome wide significance in the primary analysis [91], despite the Mitchell et al.study of 2127 participants (mean age of 60 years, 57% female) also derived from the Framingham cohort reporting moderate heritability for PWV (h2 = 0.40), with suggestive linkage regions in chromosomes 2, 7, 13, and 15 [52]. Informed by GWAS, and based on UK Biobank data, Zekavat et al. [85] generated a six variant polygenic arterial stiffness score, showing a relationship with SBP and DBP, and Mendelian randomization data supporting causality, with genetic predisposition of arterial stiffness preceding hypertension [85].

5.3 Interactions

Fifteen of the 24 genes (62.5%) implicated in arterial stiffness have evidence of BP associations, see Table 2. Many candidate gene polymorphisms studied in greater detail relate to the renin angiotensin aldosterone system; in particular angiotensin-converting enzyme (ACE) gene polymorphisms are known to influence vascular tone, fibrosis, and ultimately arterial stiffness, though with discordant results between healthy, diabetic, and hypertensive populations, despite adjustments for demographic and lifestyle factors [104, 106, 142], suggesting either an additional interaction or confounding factor is involved. Similarly, the A1166C polymorphism of angiotensin II type 1 receptor gene (AGTR1) was associated with arterial stiffness in hypertensive participants [103, 108], but not among normotensive participants of the same study, nor the Rotterdam Study population [103, 110]. Study participant age needs to be considered in such publications as combined effects may be apparent, e.g. C allele carriers showing increased PWV, but only beyond 55 years of age [103], though the Rotterdam study population was over 55 years of age but still did not support the association. Additionally, heterogeneous methods of estimating arterial stiffness limit comparisons of studies. Mayer et al. for example find AGTR1 polymorphism significant in femoral-popliteal PWV but not carotid-femoral [108]; Levy et al. conversely report greater heritability estimates for carotid-femoral than for carotid-brachial PWV, consistent with Salvi et al. reporting carotid-femoral techniques are more reliable [91, 144]. This emphasises the need for standardized technique, with the consensus now favouring carotid-femoral PWV [145]. Finally, the importance of ancestry when extrapolating data is highlighted by the concordance of results derived from a common population e.g. Zekavat et al. and Fung et al. reporting UK BioBank data [85, 86], and discordant results in candidate genes and heritability estimates across disparate populations [52, 110].

6 Endothelial Function: Flow Mediated Dilatation and Peripheral Arterial Tone

6.1 Heritability

The influence of genetics on endothelial function as measured by FMD is supported by an Italian cohort of 40 healthy young people (age 6–30, 19 male) with a family history of premature myocardial infarction, demonstrating lower FMD (5.7 ± 5.0% vs. 10.2 ± 6.6% in control subjects; P = 0.001) [146]; and by a cohort of 50 British young people with a family history of coronary artery disease (31 male, mean age 25 years) also suggesting endothelial dysfunction (FMD 4.9 ± 4.6% vs 8.3 ± 3.5% in control group, P < 0.005) [147]. Among 883 participants of the Framingham cohort (53% female; mean age 61), estimated heritability (accounting for covariates) of brachial artery baseline diameter was 0.33 ± 0.07, and FMD% was 0.14 ± 0.06; for FMD%, there was an age-gender interaction (P = 0.01), females showing steeper age-related FMD% decline [56]. Twin studies tend to be preferred above family studies for heritability estimation, as they allow a more precise separation of environmental influences from genetic effects [148], including controlling for such age effects. Twin studies reporting FMD heritability estimates include a Finnish cohort reporting FMD heritability of 24%, derived from 74 male twin pairs (20 monozygous), aged 42–69 years, with monozygous twins demonstrating improved FMD after migrating to Sweden (7.2 ± 4.4 vs 3.7 ± 2.9%, P = 0.003), a country with lower cardiovascular risk [51]. A higher estimate of 39% was reported from 94 male twin pairs from the USA (58 monozygous pairs), mean age 55 ± 2.8 years, 95% Caucasian [57].

6.2 Genes

Candidate genes linked to FMD are included in Table 2, 5 of the 8 (63%) also linked to hypertension, see Fig. 2. Examples include the Asp/Asp genotype of the endothelial nitric oxide synthase (NOS3) Glu298 → Asp polymorphism, which was associated with reduced vascular nitric oxide (NO) generation (a potent vasodilator), decreased brachial artery FMD, and increased CIMT in a group of young healthy individuals free of traditional cardiovascular risk factors [75, 149]. NOS3 regulation involves receptor-mediated mechanisms (e.g. acetylcholine, bradykinin, and substance P) and mechanical stimuli (shear stress). However, NOS3 Asp298 is not unique; more than 100 polymorphisms in NOS3 have been identified [150], with small effect size and significant interaction with other genes and environmental factors [151]. Further elements of the NO system implicated include PDE3A, a phosphodiesterase with a role in the NO/cGMP pathway.

Other genes have more obscure associations, such as PHACTR1 with a role in actin re-organisation but also possibly regulating vasoconstriction via endothelin-1 gene expression [94]; NFKB1 encoding a protein with diverse roles as a transcription regulator [152], and CYBA encoding p22phox, a component of NADPH oxidase involved in vascular ROS generation [134, 135], see Table 2. Yoshino et al. studying the genetics of endothelial dysfunction report coronary vascular responses to Acetylcholine, finding 1563 SNPs connected with cardiovascular physiology and pathology [122]. Variants in adenosine A1 receptor (ADORA 1) were associated with endothelial dysfunction in the entire cohort, while variants in adenosine A3 receptor (ADORA 3) and lipoprotein A (LPA) had the strongest associations with increased risk of endothelial dysfunction in women, again highlighting that sex differences must be considered within this area of research.

We did not find published heritability estimates regarding the EndoPAT assessment tool of peripheral arterial tone, though both race and sex are known to influence results [54, 55]. Numerous candidate genes have been proposed to influence vascular endothelial function, but only six of them reported have specifically been linked to PAT, five of the six (83 percent) had commonality with BP traits. see Fig. 2. The six linked to PAT include NOS3, already discussed in regard to FMD [117]; APO E, ACE [118], and Sphk1 SNPs/alleles [120]. Siedlinski [120] elegantly combine Sphk1 identification through murine transcriptome analysis with in vivo experiments confirming a role in vasoconstriction and endothelial dysfunction, and correlation of human sphingosine-1-phosphate (S1P) serum levels with arterial tonometry.

7 Heritability Study Considerations

BP regulation and vascular function are complex, polygenic traits, additionally influenced by many environmental factors. Molecular genetic analysis is therefore challenging due to the sheer number of relevant genes and their polymorphic effects, as examples in Table 2 illustrate. There are also certain limitations associated with heritability studies, as follows.

7.1 Family Studies

Classical family study design can overlook non-additive genetic effects and shared environmental factors. Additionally, the underlying assumption regarding the genetic relationship is flawed; offspring tend to inherit long segments of DNA resulting in deviations from the expected 50% DNA inheritance from each parent. Furthermore, family studies often recruit based on participant phenotype, with family members then invited to participate. However, techniques to correct for ascertainment bias should be employed, such as Hopper and Mathews method which adjusts the heritability estimate based on the mean and total variance of the genetic and environmental components for each individual family grouping [153].

7.2 Missing Heritability

Another issue is ‘missing heritability’, i.e., the disparity between heritability estimates derived from genotype data (explaining a low proportion of the variance), and from twin studies (estimating significantly higher heritability). Missing heritability is likely a consequence of restriction of many genetic association studies to SNPs—missing rare mutations. Gene-by-gene interactions, epigenetics, and gene-by-environment interactions also contribute to missing heritability, through assumptions that such interactions are minimal, identifiable, and that variance explained by shared environmental factors is identical in pairs. Such assumptions risk inflating heritability estimates by attributing the contribution of environmental factors to genetics.

7.3 Directionality

Directionality is an inherent challenge when assessing genotypic influences effecting vascular traits: differentiating if an identified gene has a direct impact on e.g., PWV, or alternatively elevates BP which in turn leads to arterial remodeling, stiffness, and results in elevated PWV. The high proportion of identified genetic loci and candidate genes common to both vascular phenotypes and hypertension outlined in Table 2 and Fig. 2 highlights this.

7.4 Design

Most data are cross-sectional in nature, from which change over time in vascular function or BP cannot be inferred. One might also hypothesize that SNPs contributing to vascular ageing for example may influence PWV at 60 years of age, but not at 30. Studies that do report heritability of baseline measures and progression, have found discrepancies [40]; therefore, duration of follow up, or population age of cross-sectional data must be reported in detail. Future studies independently confirming heritability of vascular traits and candidate genes, as well as their independence from each other and from BP are required, and will determine the utility of vascular assessment techniques as surrogate endpoints in trials, separate from their use as predictive risk tools.

8 Vascular phenotype

Various genes in Table 2 appear numerous times suggesting effects on multiple vascular function assessment techniques. For example ACE, which cleaves angiotensin I into angiotensin II with vasoconstrictive effects; ACE also stimulates the production of aldosterone, increasing absorption of salt and water in the kidneys; ACE furthermore causes inactivation of the vasoactive mediator bradykinin. It is therefore not surprising that genetic polymorphisms of ACE impact on many of the vascular assessment techniques described. Similarly, NOS3 (endothelial nitric oxide synthase) has been identified as relevant in multiple assessment tools of vascular function, with local vasodilatory regulation of vascular tone and diameter (see Table 2). Other genes or polymorphisms appear specific to the technique or vascular trait, such as SAA1 in CIMT, COL4A in arterial stiffness (PWV), and CYBA encoding p22phox, a component of NADPH oxidase in FMD. Some furthermore show a gene by sex interaction, such as VCAN locus in females, encoding a chondroitin sulfate proteoglycan of the adventitia and intima in CIMT [44], and NOS3 rs1799983 relating to central pulse pressure and forward wave amplitude parameters again only in females [98]. Others appear to only reach significance in those with hypertension, suggesting gene by gene or gene by environment interactions e.g. CYBA T allele associated with higher FMD only in hypertensive individuals [154]. These highlight the importance of comprehensive demographic reporting and consideration of such factors when comparing data from multiple sources. Finally, fewer studies were identified reporting the genetics of measures of endothelial function (FMD and PAT) compared to those relating to vascular stiffness and remodeling/atherosclerosis; we would propose this as an area for future study. Of note, no single gene or SNP discussed here demonstrates a substantial association with the vascular traits and assessment techniques covered. This is to be expected in polygenic traits, but may also reflect features of study design identified above: necessity for standardised technique with these tools, underpowering and lack of external validation cohorts among many studies, gene–gene or gene– environment interactions. Comparisons between different demographic groups are also complicated if age, sex, race, and BP are not fully adjusted for. Researchers should be cognizant of these in future studies.

9 Sex-Differences

Gene-by-sex interaction may not always be captured by GWAS. Efforts to elucidate sex-specific genomic determinants of BP demonstrated in 120 Canadian families found that one quarter of the 539 hemodynamic, anthropometric, metabolic, and humoral traits studied were both age and sex dependent, and one eighth were exclusively age or sex dependent [155].

A vascular phenotypic divide related to participant sex may also exist, demonstrating greater discrimination between normotensive and hypertensive PWV and augmentation index for females than males [16] and supported by our own unit’s experience (unpublished). Conversely, a collaboration establishing reference values for PWV describe apparent sex differences being almost fully accounted for by age and BP differences [156]. Two points therefore to consider if undertaking or analysing vascular function data, is whether the groups were well matched or adjustments for age and BP applied, and we suggest that researchers should also report outcome data stratified by sex to facilitate interpretation.

10 Conclusion

In conclusion, CIMT, PWV/PWA, FMD and PAT offer utility as surrogate markers of atherosclerosis, arterial stiffening, endothelial and microcirculatory function i.e. vascular function, and are predictive of cardiovascular risk. They may also have an increasing role as surrogate endpoints in genomic studies and clinical trials [157], however sex differences remain contentious, and dissecting genetic associations independent from hypertension is challenging. The genetics underlying these vascular assessment techniques have been variably studied, CIMT more so than PAT. The genetics of hypertension has a broad literature base; the next step is to integrate characterization of vascular and hypertensive phenotypes with genotypes as a natural symbiosis in studying the pathophysiology of hypertension and cardiovascular disease, and to better personalize cardiovascular medicine.