Background

Cardiovascular disease (CVD) is the leading cause of world-wide mortality, with more than 42 million people1 affected by forms of the disease including stroke, atherosclerosis, and hypertension. In 2008, an estimated 17.3 million people died from CVD, representing more than 30% of all global deaths2. Hypertension is a major risk factor for CVD3 and causes ~50% of ischemic heart disease and increases the risk of stroke4. The World Health Organisation estimates hypertension to contribute to 7.1 million deaths every year2, where the prevalence was estimated to be 40% in adults over 25 years old in 20082. Hypertension becomes a greater risk for CVD with age, where systolic blood pressure (SBP) becomes of an important predictor of CVD risk4.

Dietary sodium intake is one of the most common and important risk factors for hypertension5-8. Combinations of observational studies9,10, clinical trials11,12 and meta-analyses13-15 have shown a positive association between salt intake and sodium excretion with blood pressure (BP) and hypertension risk16.

Conducted in 1988, the clinical study INTERSALT17 found that sodium excretion was positively associated with SBP within 10,079 men and women aged 20-59 across 48 centres from 32 countries. The estimated effect of 100mmol/d lower dietary sodium intake corresponded to a 2.2mmHg lower average population SBP17-19. The World Heart Federation estimates that a universal reduction in dietary sodium intake of about 1g a day (approximately 3g of salt) would lead to a 50% reduction in required hypertensive treatment, 22% fewer deaths resulting from strokes and 16% fewer deaths from coronary heart disease (CHD)20.

A recent meta-analysis of randomized controlled trials (RCTs)13 identified seven studies where reductions in sodium excretion of between 27-39mmol/24h were associated with a decrease in SBP of 1-4mmHg. However, there was no strong evidence supporting decreases in CVD morbidity with salt restriction.

Animal models have shown an association between high salt intake and risk of hypertension. For example, chimpanzees fed a diet containing 35 versus 120mmol of sodium per day had significantly lower BP21. And after introducing a diet containing ~248mmol of sodium per day for 2 years, subsequent reduction to ~126mmol of sodium per day reduced BP compared to animals maintained on an increased salt diet21. Additionally, Dahl salt-sensitive rats fed on a high NaCl diet have a greater rise in BP compared with salt-resistant rats fed on the same diet22. Within 3 generations of selection, the salt-sensitive rats and salt-resistant rat strains show clear differences, suggesting that salt sensitivity is inherited.

However, some studies have also shown negative results suggesting that low salt intake could even be harmful23-25, highlighting the inconsistencies within this area. Moreover, observational studies are known to suffer from bias, confounding and reverse causation, potentially providing misleading results, which could prove costly when identifying specialised drug targets for reducing the prevalence of CVD. CVD is a complex trait, where the maintenance of BP is influenced by multiple environmental and genetic determinants, as well as their interactions.

The mechanisms underlying the relationship between dietary salt intake and BP are many and complex, achieved by paracrine, neural, endocrine and systematic control. For example, once plasma volume decreases, renin is released from the juxtaglomerular cells (JGCs) within the kidney and activates the renin-angiotensin system (RAS). Renin transforms angiotensinogen to angiotensin I (ANG I), which is then converted to ANG II by the angiotensin-converting enzyme (ACE)26. ANG II then stimulates the adrenal cortex to release aldosterone (Aldo), which stimulates sodium and water absorption through many mechanisms, restoring plasma volume. Another pathway involves transforming growth factor-beta (TGF-beta), where excess salt intake rapidly increases endothelial production of TGF-beta, causing arterial stiffness, peripheral vasoconstriction and arterial hypertension27. Nitric oxide (NO) normally counterbalances the effects of TGF-beta, but is depleted in salt-sensitive individuals, increasing susceptibility to hypertension. Salt-sensitivity involves a genetic defect, combined with renal injury and other environmental factors. Therefore, in genetically predisposed individuals, high salt intake reduces renal vascular function and increases the risk of hypertension27.

The prevalence and complexity of CVD has driven genetic studies that identify genes associated with sodium/water homeostasis and exploit the properties of these variants with the hope to develop greater understanding the pathways of salt-induced high BP and salt-sensitive hypertension.

GenSalt28,29 is a prospective intervention study in rural China of 3,153 participants in 658 families, with untreated pre-hypertension or hypertension. The study has identified many candidate genes for the underlying pathway explaining the association between salt intake and BP. These genes encompass the renin-angiotensin system (REN, AGT, AT2R1/2, ACE, RENBP, ACE2, APLN and AGTRL1), the aldosterone system (CYP11B1/2, MLR and HSD11B1/2), and the endothelial system (ET1/EDN1, NOS3 and SELE), epithelial sodium channels (ENaC, SCNN1B and SCNN1G)29. Additional genome-wide association studies (GWAS) have identified more than 47 genetic variants at 40 loci associated with BP and hypertension risk30-41. Additionally, 8 genetic variants have been associated with BP in more than 25,000 individuals: at the MTHFR-NPPB, AGT, NPR3, HFE, NOS3, LSP1/TNNT3, SOX6 and ATP2B1 loci3. Using a multi-stage design in 200,000 individuals of European descent, the International Consortium for Blood Pressure GWAS on SBP, DBP and hypertension risk identified 29 loci robustly associated with BP, 16 of which were novel. Of these, six loci contained genes previously thought to regulate BP: GUCY1A3-GUCY1B3, NPR3-c5orf23, ADM, FURIN-FES, GOSR2 and GNAS-EDN3) and revealed ten new loci providing clues on BP aetiology including SLC4A7, SLC39A8, MOV10, EBF1 and JAG1. However, these only explain a small amount of the heritability of this trait.

Much effort has focused on understanding the 'missing heritability' in complex traits such as CVD, which can be attributable to small effects of genetic variants, rare variants of moderate penetrance and gene-environment interactions. Additionally, there is a considerable lack of causal and mechanistic analyses on the association between dietary sodium intake and BP. Evidence suggests that the salt/water balance is in part mediated by epigenetic mechanisms, specifically histone modification and DNA methylation (in animals and plants), which alters expression levels of important regulatory genes such as epithelial Na+ channels (ENaCs) in response to varying salt intake42-46. These epigenetic mechanisms also play a role in the prenatal imprinting of postnatal-specific feeding behaviours and intergenerational transmission of salt appetite from mother to offspring47, a pathway of which is evident in rats and has been verified in human newborn infants47. Furthermore, data suggests that environmental, particularly nutritional, factors during pregnancy can lead to changes in DNA methylation of the offspring48, supporting the hypothesis that origins of diseases in adults begin in utero48-50.

Epigenetic association studies are beginning to accumulate51-53. Importantly, differentiating between causative and consequential differential epigenetic modifications is difficult but necessary, as epigenetic marks are more similar to phenotypes than genotypes, and can thus be confounded by other environmental factors.

Our knowledge in this area is far from complete and inconsistencies between previous studies drives the importance of understanding the causal and mechanistic pathways underlying this known association between dietary sodium intake and BP. Therefore, I intend to examine the longitudinal, intergenerational, intrauterine, and epigenetic contributions of dietary sodium intake to BP regulation, assessing the underlying causal pathways and mechanisms.

Objectives

1. Identify genes involved in sodium regulation and excretion through GWAS, focusing on genes with the greatest diversity of effect on circulating sodium levels.

2. Use a novel MR approach to assess the causal links between dietary sodium intake and BP with known genetic variants.

3. Multi-level model of how BP changes longitudinally in children as a result of dietary sodium intake, in terms of genetic and epigenetic pathways.

4. Examine the intergenerational and intra-uterine effects of maternal diet and offspring BP, taking into account paternal, maternal-prenatal and maternal-postnatal dietary intake and the challenges with this type of analysis.

5. Examine the differential methylation in BP of children and mothers, prospectively comparing BP of children with varying diets.

6. Assess differential transcription/expression using mRNA levels due to differential methylation found in (5).

7. Identify differences in BP due to varying diets in urban vs. rural populations and exploit the genetic information available on BP to assess rural vs. urban interaction.

Study Design

Participants and Variables

GWAS will be based on the ~10,000 children within ALSPAC54 who have been genotyped on the 610K Illumina SNP Chip. Averages from repeated BP measurements will be used to increase the reliability and sensitivity of this trait. BP has been collected from children at 37, 49 and 61 months and 7, 9, 10, 11, 12, 13, 15, and 17 years at rest and after exercise. Maternal BP measurements were collected at each pregnancy trimester, and when the children were 11, 13, 15 and 17 years old. Information on family history of hypertension, CHD and stroke from parents is also available.

Salt intake will be derived from 3-day nutritional diaries and food frequency questionnaires (FFQ) from a sub-sample from ALSPAC children collected at 4, 8, 18, 43, and 61 months, and at 7.5, 10.6 and 13.9 years and from mothers and 32 weeks gestation, and at 47 months, 97 months and 13 years. Nutrient intakes and food groups consumed will be derived from the FFQs and diet diaries using similar methods to previous studies16,55. Information on sodium excretion and sodium-related metabolites56 will be obtained from the ~1,000 children who have urine spot samples taken at age 10, 15 and 17. Epigenomic information will be obtained from ARIES in 1,000 ALSPAC mothers and children, of whom have phenotypic data. Intergenerational data will primarily use ALSPAC (and COCO90s, if available). Assessment of urban vs. rural interaction will use the Indian Migration Study57.

Rare variants analysis will utilize ~2,000 ALSPAC children who have whole-genome sequencing data and will be used as a reference set for imputing the rest of the cohort. Significant associations will be performed for replication in other cohorts where suitable data are available.

Significant associations will be followed up in available data from cohorts including EPIC, BWHHS, BRHS, Birth to Twenty, Indian Migration Study, Hellenic Isolated Cohorts (HELIC), Born in Bradford, Danish National Birth Cohort, Generation R and the German Infant Study on the Influence of Nutrition Intervention.

Statistical Analysis

Genome-wide data will initially be imputed to HapMap Phase II. Rare variant (1-5% MAF) will be using genome-wide SNP data from ~10,000 ALSPAC children imputed into UK10K dataset (or 1,000 genomes if UK10K is not available).

Multivariate regression will be used to assess associations between exposures and outcomes, where analyses will be adjusted for appropriate confounders (eg. age, sex, BMI). Intergenerational associations between maternal diet and offspring BP will take into account additional covariates (eg. maternal nutrition before pregnancy, supplementation, birthsize, intake after birth, child growth from birth and sex). Longitudinal multivariate modelling of BP in ALSPAC children will use MLwiN to fit multi-level models, which describe BP variation over time as well as covariation between BP phenotypes. MR methodology will be used to assess causal associations between salt intake and BP, using dietary data and sodium-related metabolites as the exposures. Associations between paternal, maternal post-natal and maternal pre-natal dietary information and offspring BP will be examined to assess whether associations between maternal diet and offspring BP is due to intrauterine effects.

Assessment of epigenetic associations will involve examining differential methylation and using the two-step MR approach58. Expression levels will be investigated using genome-wide expression data on 1,000 ALSPAC children measured from lymphoblastoid cell lines using the Illumina Human-6 v2 Beadchip (48,000 transcripts). All analysis will be using STATA, R, PLINK and MLwiN.