Genetic predisposition to high blood pressure and lifestyle factors: associations with

midlife blood pressure levels and cardiovascular events

Running Title: Genetic, lifestyle, and cardiovascular risk

Raha Pazoki1, Abbas Dehghan1,2, Evangelos Evangelou1,3, Helen Warren4,5, He Gao1,2, Mark

Caulfield4,5, Paul Elliott1,2, Ioanna Tzoulaki1,2,3

1Department of Biostatistics and Epidemiology, School of Public Health, Imperial College

London, London, W2 1PG, UK

2MRC-PHE Centre for Environment, School of Public Health, Imperial College London,

London, W2 1PG, UK

3Department of Hygiene and Epidemiology, University of Ioannina Medical School,

Ioannina, Greece

4William Harvey Research Institute, Barts and The London School of Medicine and

Dentistry, Queen Mary University of London, London, UK.

5National Institute for Health Research, Barts Cardiovascular Biomedical Research Center,

Queen Mary University of London, London, UK.

Corresponding authors:

Paul Elliott, Department of Biostatistics and Epidemiology,

School of Public Health, Imperial College London,

London, UK, W2 1PG

Email: p.elliott@imperial.ac.uk

Tel: +44 20 7594 3328

Fax: +44 20 7594 3456

Ioanna Tzoulaki, Department Hygiene and Epidemiology,

University of Ioannina,

Ioannina, Greece, 45110

E-mail: itzoulak@cc.uoi,gr

Tel: +30 26510 0 7605

Twitter handle: @jotzou

Word count 5,915

2

Abstract

Background: High blood pressure (BP) is a major risk factor for cardiovascular diseases

(CVD), the leading cause of mortality worldwide. Both heritable and lifestyle risk factors

contribute to elevated BP levels. We aimed to investigate the extent to which lifestyle factors

could offset the effect of an adverse BP genetic profile, and its effect on CVD risk.

Methods: We constructed a genetic risk score for high BP using 314 published BP loci in

277,005 individuals without previous CVD from the UK Biobank study, a prospective cohort

of individuals aged 40 to 69 years, with median 6.11 years of follow-up. We scored

participants according to their lifestyle factors including body mass index, healthy diet,

sedentary lifestyle, alcohol consumption, smoking, and urinary sodium excretion levels

measured at recruitment. We examined the association between tertiles of genetic risk and

tertiles of lifestyle score with BP levels and incident CVD using linear regression and Cox

regression models respectively.

Results: Healthy lifestyle score was strongly associated with BP (P<10-320 for systolic and

diastolic BP and CVD events regardless of the underlying BP genetic risk. Participants with

favorable compared to unfavorable lifestyle (bottom vs. top tertile lifestyle score) had 3.6,

3.5, and 3.6 mmHg lower systolic BP in low, middle and high genetic risk groups

respectively (P for interaction= 0.0006). Similarly, favorable compared with unfavorable

lifestyle showed 30%, 31%, and 33% lower risk of CVD among participants at low, middle

and high genetic risk groups respectively (P for interaction= 0.99).

Conclusions: Our data further support population-wide efforts to lower BP in the population

via lifestyle modification. Advantages and disadvantages of disclosing genetic predisposition

to high BP for risk stratification needs careful evaluation.

3

Keywords: genetic risk, blood pressure, cardiovascular disease, healthy lifestyle, genetic predisposition to disease

4

What is new?

We show that adherence to a healthy lifestyle (including healthy diet, limited alcohol

consumption, low urinary sodium excretion, low body mass index, and increased

physical activity) is associated with lower blood pressure regardless of the underlying

BP genetic risk.

Adherence to a healthy lifestyle is also associated with lower risk of myocardial

infarction, stroke and composite cardiovascular disease at all levels of underlying

blood pressure genetic risk.

Healthy compared with unhealthy lifestyle showed 30%, 31%, and 33% lower risk of

CVD among participants at low, middle and high genetic risk groups respectively

What are the clinical implications?

Genetically predetermined rise in BP and its complications can be offset at least to

some extent by healthy lifestyle.

Our results further support population-wide efforts to lower BP and subsequent CVD

risk through lifestyle modification.

Introduction

High blood pressure (BP) is the leading single risk factor for mortality and global burden of

disease (9.4 million deaths in 2010) 1. There is a strong graded relationship between BP and

cardiovascular disease (CVD) with even small increments in BP associated with an increased

risk of CVD, the leading cause of death and disease burden worldwide 2.

In the last four decades, efforts to detect and treat elevated BP have been vigorous. However,

there is a need for improved primary prevention of high BP. Heritable and

environmental/lifestyle risk factors both contribute to elevated BP levels 3, 4.There is well

established evidence for independent unfavorable additive effects on BP of excess sodium

intake, unhealthy dietary patterns or adverse calorie balance (body mass), excess alcohol use,

and physical inactivity 5-11. At the same time, more than 314 genetic variants are known to

affect BP levels and could have a cumulative effect of up to 10mmHg higher systolic BP

level by age 503. The mechanism through which these genetic variants affect BP levels is

largely unknown3.

We aimed to investigate the associations between healthy lifestyle adherence and blood

pressure levels and cardiovascular risk in different subgroups of genetic BP risk. We

examined the role of lifestyle in individuals predetermined to have higher BP levels based on

their genetic profile in relation to both BP and future CVD events within UK Biobank.

2

Methods

Additional material is provided in the Online Supplement. Approval for this research was

obtained from the UK Biobank Research Ethics Committee and Human Tissue Authority and

the participants gave informed consent. The calculated genetic risk score will be made

available to other researchers on the UK Biobank website for purposes of reproducing the

results or replicating the procedure.

Study population

The UK Biobank is a national long-term cohort in the UK that recruited 502,638 individuals

aged between 40 and 69 years. The present study is based on a subset of unrelated individuals

with GWAS data12-14 and of European ancestry following quality measures and exclusions

(sex discordance, high missingness/heterozygosity, 1st- and 2nd-degree relatives, and non-

European ancestry excluded) (Supplementary Figure 1).

In detail, we excluded participants who were pregnant or unsure of their pregnancy at

baseline (N=372) as well as those withdrawn consent (N=19). Genetic data were available for

487,409 individuals. After merging genetic and phenotype data, 487,048 individuals

remained. We excluded 147,637 individuals being related to at least one individual in the

genotype data. We further excluded 30,464 individuals of non-European ancestry, 21,134

individuals with CVD events at or before baseline and 10,808 individuals with missing values

on the main covariables of the current study. The final sample for analysis comprised 277,005

participants.

Lifestyle factors and physical measurements

Following informed consent, participants completed a standardized questionnaire on life

course exposures, medical history and treatments and had a range of physical measurements.

We assessed diet based on a self-completed food frequency questionnaire. Participants ranked

3

their daily intake of dietary consumption including alcoholic products, fruits and vegetables,

oily and non-oily fish, processed and unprocessed meat, using touch-screen multiple choice

questions14. We defined smoking based on self-reported information on current (“most of

days” and “occasional” smokers), past smokers and never smokers. Sedentary behavior was

defined by the sum of three questions about the hours per day participants spent (i) driving,

(ii) using a computer and (iii) watching television. We calculated alcohol intake from the

self-reported alcohol drinking information on the touch-screen questionnaire. The quantity of

each type of drink (red wine, white wine, beer/cider, fortified wine, spirits) was multiplied by

its standard drink size and reference alcohol content. Drink-specific alcohol intake during the

reported drinking period (a week for frequent drinkers or a month for occasional drinkers)

was summed up and converted to grams per day for participants with complete response to

the quantitative drinking questions. Grams per day of alcohol consumption for participants

with incomplete response was imputed by bootstrap resampling from the complete responses,

stratified by drinking frequency (occasional or frequent) and sex.

Two BP measurements were taken seated after two minutes rest using an appropriate cuff and

an Omron HEM-7015IT digital BP monitor. Systolic (SBP) and diastolic blood pressure

(DBP) were analyzed. We calculated mean SBP and DBP from two automated (N= 253,419)

or two manual readings (N=15,454) BP measurements. For individuals with one manual and

one automated BP reading (N=7,886), we used mean of these two values. For individuals

with single BP measurement (one manual or one automated BP reading, N=246), we used

that single measurement. For individuals reported to be taking BP-lowering medication

(N=47,438 of individuals), we added 15 and 10 mmHg to SBP and DBP respectively15.

Standing height was measured using a Seca 202 device. Body mass index (BMI) was

calculated as weight (kg) divided by height squared (m2) with weight measured using

electronic weighing scales (Tanita BC-418)13.

4

Sodium and potassium concentrations were measured in stored urine samples by the Ion

Selective Electrode method (potentiometric method) using Beckman Coulter AU5400, UK

Ltd. Analytical range for sodium was 2-200mmol/L and for potassium 10-400 mmol/L.

Details of quality control and sample preparation have been published previously 16.

Healthy Lifestyle Score

A composite healthy lifestyle score adapted from American Heart Association cardiovascular

health recommendations was constructed 17. A healthy lifestyle score was defined by BMI

below median, sedentary hours below median, alcohol intake below median, meat intake

(processed and unprocessed) below median, urinary sodium excretion below median, fruit

and vegetable intake above median, fish intake (oily fish and non-oily) above median, and

never smoking. One point was given for each favorable lifestyle factor (range 0 to 8).

We additionally performed a sensitivity analysis using cut-offs of the American Heart

Association (AHA) for healthy lifestyle 18, 19 presented in Supplementary Table 1.

Cardiovascular events

For all participants, retrospective and prospective linkage to electronic health data is

available, including Hospital Episode Statistics (HES) and Office for National Statistics cause

of death data. HES provide detailed information for participants admitted to hospital and

includes coded data on diagnoses and procedures. CVD was defined as an event of coronary

artery disease (CAD), or stroke or myocardial infarction classified using an in-house

algorithm comprising codes from International Classification of Disease (ICD) 9 and 10

codes and Classification of Interventions and Procedures (OPCS) codes (Supplementary

Table 2).

The recorded “Episode date” was considered as the date of the event. For individuals with

missing of “Episode date”, “admission date” was used as the date of event. For individuals

5

with multiple CVD hospitalizations in HES, the date of the earliest event was used as date of

event. Fatal CVD events were searched in mortality data and were added to the main CVD

variable. We additionally used the algorithmically derived definitions of myocardial

infarction (MI) and stroke as coded by UK Biobank20. History of CVD at baseline was

defined based on self-reported data and/or on HES data with episode date preceding the date

of the visit in the study assessment center. CVD events occurring prior to assessment date and

self-reported CVD events were considered as existing CVD events at baseline. Fatal CVD

events with pre-existing CVD from HES occurring before assessment or self-reported CVD

were considered as existing CVD events at baseline. Individuals with CVD events occurring

before assessment were excluded. Follow-up time was calculated as the time starting from

assessment date for each individual until 31st March 2015. Individuals who died during

follow-up were censored.

Our primary traits and outcomes were a) SBP and DBP measured at baseline and b) the

composite measure of CVD events (See Supplementary Table 2), and separately MI and

stroke (non-fatal and fatal) during follow up (2006-2016). The main exposures studied

included: a) genetic risk score for BP (tertiles), b) lifestyle score (tertiles), and c) the 3×3

matrix of the genetic risk and lifestyle scores.

Genotyping and Imputation

Detailed information about genotyping and imputation in the UK Biobank study has been

provided elsewhere 21, 22. Briefly, DNA samples of the UK Biobank study participants were

genotyped using a custom Affymetrix UK Biobank Axiom array (designed to optimize

imputation performance). Genotype imputation used a special reference panel comprising a

merged sample of UK10K sequencing and 1000 Genomes imputation reference panels to

maximize using of haplotypes with British and European ancestry for imputation. Imputation

6

was carried out centrally by the UK Biobank using an algorithm implemented in the

IMPUTE2 program. Genetic principal components to account for population stratification

were computed centrally by the UK Biobank.

Genetic Risk Score for BP

A weighted genetic risk score was calculated based on previously reported genetic variants

(Supplementary references 1 and 2) for SBP, DBP, and pulse pressure (PP): (i) 267 SNPs

with weights (β coefficients) reported by Warren et al. 1 (we extracted weights from

replication cohorts for novel signals and from UK Biobank for known signals as described in

Warren et al.1) and (ii) 47 SNPs identified and replicated from Hoffmann, et al.23 with

weights (β coefficients) from the International Consortium for Blood Pressure (ICBP)+

Genetic Epidemiology Research on Adult Health and Aging (GERA) meta-analysis. We

calculated a standardized BP genetic risk score for each of SBP, DBP, and PP loci and

combined them into a single BP genetic risk score by averaging the three standardized SBP,

DBP and PP genetic risks. Pairwise-independent, LD filtered (r2 < 0.2) variants were used for

the analysis (Supplementary Table 3).

Statistical Analysis

Genetic risk and lifestyle scores (and their combination) were analyzed with respect to a)

SBP and DBP considered separately using multiple linear regression and b) CVD events

using Cox Proportional Hazards regression with duration of follow-up as the time metric.

Proportional hazards assumptions were tested using Schoenfeld residuals implemented in R

Package Survival, and risk estimates were presented as Hazard Ratio (HR). Analyses were

adjusted for age, sex and, when genetic risk was included in the model, for the first ten

genetic principal components. For the statistical analyses, we also adjusted for BMI,

sedentary lifestyle, smoking status, and healthy diet score (defined as consumption of fruits

and vegetables each ≥ 3 servings a day; fish ≥ 2 servings a week; processed meat ≤ 1 servings

7

a week; and unprocessed meat ≤ 1.5 servings a week17). We tested for interaction between the

genetic risk and lifestyle scores using a likelihood ratio test comparing a model with and

without interaction terms (categorized in tertiles).

We calculated standardized 5-year cumulative incidence rates for CVD, MI, and stroke.

Standardization was performed using the World Health Organization (WHO) Standard

population 24 and the European Standard population (ESP 2013)25.

As sensitivity analyses, we excluded participants receiving BP (N= 47,438) or lipid lowering

(N = 35,155) medication and those with diabetes diagnosis at baseline (N = 10,958) as they

may have changed their lifestyle habits due to the diagnosis of their condition (total

exclusion, N= 64,555). Also, we examined the association between genetic risk score and

each lifestyle variable separately in relation to SBP, DBP and CVD events. Lifestyle

variables included BMI (tertiles), healthy diet score (as above), sedentary lifestyle (tertiles),

smoking status (present, past or never smoker), urinary sodium and potassium excretion and

alcohol intake (tertiles). For sedentary lifestyle, we performed additional sensitivity analysis,

excluding individuals with an event during the 2 years of follow up since low physical

activity could be a marker of subclinical or undiagnosed disease.

All statistical analyses were performed in R software, version 3.3 (R Project for Statistical

Computing).

8

Results

The sample of 277,005 individuals (Supplementary Figure 1) comprised 152,121 (55%)

women and 124,884 (45%) men (Table 1). Median follow-up for CVD was 6.11 years.

During follow-up, 9,278 CVD events occurred (incidence rate 5.55 per 1000 person-years),

of which 2,984 were MI (incidence rate 1.53 per 1000 person-years) and 1919 were stroke

(incidence rate 1.00 per 1000 person-years). Supplementary Table 4 shows the distribution

of various risk factors of CVD per subgroups of genetic risk and healthy lifestyle.

Genetic risk score as a continuous variable (Supplementary Table 5) was significantly

associated with all outcomes examined (HR CVD = 1.11; 95% CI= 1.09 to 1.14; P<10-320) per

unit increase in the genetic risk score. Healthy lifestyle score (per unit) as a continuous

variable was also strongly and inversely associated with both SBP (Beta= -0.88mmHg, 95%

CI, -0.92 to -0.85, P<10-320) and DBP (Beta= -0.71mmHg, 95% CI, -0.73 to -0.69, P<10-320)

(Supplementary Table 5). Similarly, strong inverse associations were observed between

lifestyle score (per unit increase) and incident CVD (HR= 0.92; 95% CI=0.92 to 0.93; P<10-

320). Figures 1 and 2 show a significant risk gradients between tertiles of healthy lifestyle or

tertiles of genetic risk score in relation to SBP levels and incident CVD.

With the exception of alcohol, which was associated with BP, MI, and stroke but not with

CVD (Supplementary Table 5), all other the individual healthy lifestyle factors (diet, BMI,

sedentary lifestyle) were associated with both BP and all CVD events.

Within combined subgroups of healthy lifestyle and genetic risk, healthy lifestyle was

associated with lower SBP and DBP within each tertile of genetic risk (Figure 2). Among

participants at low genetic risk, the estimated mean SBP was 140 mmHg (95% CI 102 to

177) for participants with unfavorable lifestyle and 134 mmHg (95% CI 95 to 172) with a

favorable lifestyle (Figure 2). For those at high genetic risk, the estimated mean SBP was

9

146 mmHg (95% CI 106 to 186) among those with an unfavorable lifestyle and 142 mmHg

(95% CI 100 to 184) among those with a favorable lifestyle. An unfavorable lifestyle as

compared with a favorable lifestyle, was associated with 4.9 mmHg higher SBP among

participants at low genetic risk, 4.3 mmHg higher SBP among participants at intermediate

genetic risk, and 4.1 mmHg higher SBP among participants at high genetic risk

(Supplementary Table 6). There was modest statistical evidence for interaction between

genetic risk and healthy lifestyle scores in relation to SBP (P for interaction=0.0006) but not

for DBP (P for interaction=0.8).

Regarding CVD events, participants with favorable compared with unfavorable lifestyle

showed 30%, 33%, and 31% lower relative risk of CVD among participants at low,

intermediate, and high genetic risk groups respectively (Supplementary Table 7).

Participants with less favorable genetic and lifestyle profiles (top tertile of BP genetic risk,

bottom tertile of healthy lifestyle score) had nearly two-fold greater risk of CVD (HR= 1.75;

95% CI=1.59, 1.93) compared with participants with favorable genetic and lifestyle profiles

(the reference group; Figure 3; Supplementary Table 7). Genetic risk and healthy lifestyle

score were statistically independent of each other and there was no statistically significant

interaction (P for interaction=0.99). Among participants at low genetic risk, the standardized

5-year CVD rates based on WHO World and Europe standard populations were 2.77% and

3.22% respectively among those with an unfavorable lifestyle and they were 1.46% and

1.75% respectively among those with a favorable lifestyle. Among participants at high

genetic risk, the standardized 5-year coronary event rates were 3.53% and 4.11% respectively

among those with an unfavorable lifestyle and 1.76% and 2.09% respectively among those

with a favorable lifestyle. Similar pattern of associations was observed for MI and stroke

(Supplementary Table 8).

10

Sensitivity analyses with exclusion of participants with self-reported diabetes or taking blood

pressure and/or lipid lowering medication, and the individual lifestyle components, showed

similar associations for BP and CVD events (Supplementary Table 6 & 7, Supplementary

Figure 2). Sensitivity analysis for healthy lifestyle score using AHA cut-offs as well as

sensitivity analysis for sedentary lifestyle excluding participants with an event within the first

two years of follow-up did not materially change the results (Supplementary Table 9 & 10).

11

Discussion

In this large study of more than 277,000 individuals, we observed that unhealthy lifestyle and

genetic susceptibility to high BP were associated with higher levels of BP and a greater risk

of subsequent incident CVD events (11% higher risk of incident CVD per unit increase of

genetic risk). The favorable association of lifestyle with BP and CVD was found across all

genetic risk categories, suggesting that the genetically predetermined rise in BP and its

complications can be offset at least to some extent by healthy lifestyle. Our results further

support population-wide efforts to lower BP and subsequent CVD risk through lifestyle

modification.

Consistent with a well-established causal effect of BP on cardiovascular disease27 and with a

lifelong effect of BP variants on BP levels1, 26, we confirmed that a genetic risk score with 314

BP-associated genetic variants is a strong predictor of CVD events including myocardial

infarction and stroke. The association was independent of lifestyle risk factors related to BP

and CVD. Indeed, BP is measured with large measurement error due to BP variability

whereas BP genotypes can be precisely measured, are constant over time, and therefore

capture a fixed component of lifetime BP exposure. Evidence from other areas have shown

that, for example, genetic risk scores using lipid or diabetes mellitus genetic variants predict

CVD and diabetes mellitus, respectively, more consistently over time than standard clinical

biomarkers27, 28.

We showed that the detrimental effect of genes on BP and on subsequent CVD risk can be

largely offset by a healthy lifestyle in support of previous observations on CVD and obesity29,

30. This observation challenges the deterministic interpretation of the genetic risk in

individual-based risk assessment. Genetic risk can be known from birth whereas the other

conventional CVD risk factors usually appear in midlife and given that genetic risk is non-

12

modifiable, it is aligned with lifelong risk prediction. Our observations raise the possibility of

targeting individuals at high genetic risk early in life for lifestyle or pharmacological

modification and primordial prevention strategies31. Yet, more evidence is needed on the

effect of disclosing genetic information to individuals and risk communication in order to

find the most appropriate means to achieve risk modification through lifestyle.

Our study has several strengths including a large sample size and large number of incident

CVD events, up to ~8 years follow-up, and rich baseline phenotyping according to a well-

defined and standardized protocol12. We utilized an updated genetic risk score for BP using

latest published results including data on 314 validated BP loci3, 23. Limitations include the

fact that lifestyle is poorly measured in contradistinction to the genetic information, such that

the associations between lifestyle outcome measures may have been underestimated

(regression dilution32-36). Also, we limited our analyses to food frequency questionnaire data

that lacked information on dairy products, energy intake, and fat consumption. In addition,

lifestyle may be influenced by pre-existing conditions and our results may therefore be

subject to reverse causality. To account for these limitations, we excluded individuals with a

history of CVD and performed sensitivity analyses excluding individuals on treatment for

CVD risk factors.

Conclusion

In our study of 277,005 individuals, we showed that a healthy lifestyle is associated with low

BP levels and a lower risk of subsequent cardiovascular events (i.e. CVD, MI, and stroke)

within each category of BP genetic profile. A high genetic risk was largely offset by a

favorable lifestyle but also people with low genetic risk could lose their inherent protection if

they had an unhealthy lifestyle. While it is possible to modify lifestyle, it is not possible to

alter the genetic makeup stressing the importance of population-wide lifestyle approaches to

13

address the pressing BP problem. Our findings highlight the need for timely lifestyle

interventions to offset the lifetime risk of future high BP and CVD. Given the importance of

population wide lifestyle modification the use of genetic information for risk stratification

merits careful evaluation before it is routinely implemented in clinical practice.

Funding sources

This research has been conducted using the UK Biobank Resource under applications number

10035 and 236 granting access to the corresponding UK Biobank genetic and phenotype data

(released 17 Nov 2016). The research was supported by the British Heart Foundation (grant

SP/13/2/30111) for Large-scale comprehensive genotyping of UK Biobank for

cardiometabolic traits and diseases: UK CardioMetabolic Consortium. P.E. is Director of the

MRC-PHE Centre for Environment and Health and acknowledges support from the Medical

Research Council and Public Health England (MR/L01341X/1). P.E. also acknowledges

support from the NIHR Biomedical Research Centre at Imperial College Healthcare NHS

Trust and Imperial College London, and the NIHR Health Protection Research Unit in Health

Impact of Environmental Hazards (HPRU-2012-10141). P.E. is a UK Dementia Research

Institute (DRI) Professor, UK DRI at Imperial College London. This work was supported by

the UK Dementia Research Institute which receives its funding from UK DRI Ltd funded by

the UK Medical Research Council, Alzheimer’s Society and Alzheimer’s Research UK. This

work used the computing resources of the UK MEDical BIOinformatics partnership (UK

MED-BIO) which is supported by the Medical Research Council (MR/L01632X/1).

Disclosures

None

14

References1. Ezzati M, Obermeyer Z, Tzoulaki I, Mayosi BM, Elliott P and Leon DA. Contributions of risk factors and medical care to cardiovascular mortality trends. Nature reviews Cardiology. 2015;12:508-30.2. Tzoulaki I, Elliott P, Kontis V and Ezzati M. Worldwide Exposures to Cardiovascular Risk Factors and Associated Health Effects: Current Knowledge and Data Gaps. Circulation. 2016;133:2314-33.3. Warren HR, Evangelou E, Cabrera CP, Gao H, Ren M, Mifsud B, Ntalla I, Surendran P, Liu C, Cook JP, Kraja AT, Drenos F, Loh M, Verweij N, Marten J, Karaman I, Lepe MPS, O'Reilly PF, Knight J, Snieder H, Kato N, He J, Tai ES, Said MA, Porteous D, Alver M, Poulter N, Farrall M, Gansevoort RT, Padmanabhan S, Magi R, Stanton A, Connell J, Bakker SJL, Metspalu A, Shields DC, Thom S, Brown M, Sever P, Esko T, Hayward C, van der Harst P, Saleheen D, Chowdhury R, Chambers JC, Chasman DI, Chakravarti A, Newton-Cheh C, Lindgren CM, Levy D, Kooner JS, Keavney B, Tomaszewski M, Samani NJ, Howson JMM, Tobin MD, Munroe PB, Ehret GB, Wain LV, The International Consortium of Blood Pressure GA, The CHDEC, The Exome BPC, The TDGC, The Go TDC, The Cohorts for H, Ageing Research in Genome Epidemiology BPEC, The International Genomics of Blood Pressure C and The UKBCCBPwg. Genome-wide association analysis identifies novel blood pressure loci and offers biological insights into cardiovascular risk. Nat Genet. 2017;49:403-415.4. Chan Q, Stamler J, Griep LM, Daviglus ML, Horn LV and Elliott P. An Update on Nutrients and Blood Pressure. Journal of atherosclerosis and thrombosis. 2016;23:276-89.5. Marmot MG, Elliott P, Shipley MJ, Dyer AR, Ueshima H, Beevers DG, Stamler R, Kesteloot H, Rose G and Stamler J. Alcohol and blood pressure: the INTERSALT study. BMJ (Clinical research ed). 1994;308:1263-7.6. Marmot MG and Elliott P. Public health measures for blood pressure control in the whole community. Clinical and experimental hypertension Part A, Theory and practice. 1989;11:1171-86.7. Asghari G, Yuzbashian E, Mirmiran P, Hooshmand F, Najafi R and Azizi F. Dietary Approaches to Stop Hypertension (DASH) Dietary Pattern Is Associated with Reduced Incidence of Metabolic Syndrome in Children and Adolescents. The Journal of pediatrics. 2016;174:178-184.e1.8. Charakida M, Khan T, Johnson W, Finer N, Woodside J, Whincup PH, Sattar N, Kuh D, Hardy R and Deanfield J. Lifelong patterns of BMI and cardiovascular phenotype in individuals aged 60-64 years in the 1946 British birth cohort study: an epidemiological study. The lancet Diabetes & endocrinology. 2014;2:648-54.9. Pouliou T, Ki M, Law C, Li L and Power C. Physical activity and sedentary behaviour at different life stages and adult blood pressure in the 1958 British cohort. Journal of hypertension. 2012;30:275-83.10. Garcia-Ortiz L, Recio-Rodriguez JI, Puig-Ribera A, Lema-Bartolome J, Ibanez-Jalon E, Gonzalez-Viejo N, Guenaga-Saenz N, Agudo-Conde C, Patino-Alonso MC and Gomez-Marcos MA. Blood pressure circadian pattern and physical exercise assessment by accelerometer and 7-day physical activity recall scale. American journal of hypertension. 2014;27:665-73.11. Elliott P, Stamler J, Nichols R, Dyer AR, Stamler R, Kesteloot H and Marmot M. Intersalt revisited: further analyses of 24 hour sodium excretion and blood pressure within and across populations. Intersalt Cooperative Research Group. BMJ (Clinical research ed). 1996;312:1249-53.12. Elliott P and Peakman TC. The UK Biobank sample handling and storage protocol for the collection, processing and archiving of human blood and urine. International journal of epidemiology. 2008;37:234-44.13. UK Biobank Coordinating Centre; UK Biobank: Protocol for a large-scale

prospective epidemiological resource. Protocol No: UKBB-PROT-09-06 (Main Phase). 21 March 2007 (AMENDMENT ONE FINAL). URL: http://www.ukbiobank.ac.uk/wp-content/uploads/2011/11/UK-Biobank-Protocol.pdf; Access Date: 15 May 2017.14. Sudlow C, Gallacher J, Allen N, Beral V, Burton P, Danesh J, Downey P, Elliott P, Green J, Landray M, Liu B, Matthews P, Ong G, Pell J, Silman A, Young A, Sprosen T, Peakman T and Collins R.

UK biobank: an open access resource for identifying the causes of a wide range of complex diseases of middle and old age. PLoS medicine. 2015;12:e1001779.15. Tobin MD, Sheehan NA, Scurrah KJ and Burton PR. Adjusting for treatment effects in studies of quantitative traits: antihypertensive therapy and systolic blood pressure. Statistics in medicine. 2005;24:2911-35.16. Fry DA, R; Gordon, M; Moffat, S. UK Biobank Biomarker Project Details of assays and quality control information for the urinary biomarker data. 28 October 2016. URL: http://biobank.ctsu.ox.ac.uk/crystal/refer.cgi?id=34972 ; Access Date: 15 May 2017.17. Lloyd-Jones DM, Hong Y, Labarthe D, Mozaffarian D, Appel LJ, Van Horn L, Greenlund K, Daniels S, Nichol G, Tomaselli GF, Arnett DK, Fonarow GC, Ho PM, Lauer MS, Masoudi FA, Robertson RM, Roger V, Schwamm LH, Sorlie P, Yancy CW and Rosamond WD. Defining and setting national goals for cardiovascular health promotion and disease reduction: the American Heart Association's strategic Impact Goal through 2020 and beyond. Circulation. 2010;121:586-613.18. Lichtenstein AH, Appel LJ, Brands M, Carnethon M, Daniels S, Franch HA, Franklin B, Kris-Etherton P, Harris WS, Howard B, Karanja N, Lefevre M, Rudel L, Sacks F, Van Horn L, Winston M and Wylie-Rosett J. Diet and Lifestyle Recommendations Revision 2006. A Scientific Statement From the American Heart Association Nutrition Committee. 2006;114:82-96.19. Bhupathiraju SN, Lichtenstein AH, Dawson-Hughes B and Tucker KL. Adherence index based on the AHA 2006 diet and lifestyle recommendations is associated with select cardiovascular disease risk factors in older Puerto Ricans. The Journal of nutrition. 2011;141:460-9.20. Schnier C and Sudlow C. Definitions of acute myocardial infarction (MI) and main MI pathological types for UK Biobank phase 1 outcomes adjudication. January 2017. URL: http://biobank.ctsu.ox.ac.uk/crystal/docs/alg_outcome_mi.pdf; Access Date: 15 May 2015.21. Genotype imputation and genetic association studies of UK Biobank: Interim Data Release

May 2015. URL: http://www.ukbiobank.ac.uk/wp-content/uploads/2014/04/imputation_documentation_May2015.pdf; Access Date: 17 May 2017.22. Clare Bycroft CF, Desislava Petkova, Gavin Band, Lloyd T. Elliott, Kevin Sharp, Allan Motyer, Damjan Vukcevic, Olivier Delaneau, Jared O’Connell, Adrian Cortes, Samantha Welsh, Gil McVean, Stephen Leslie, Peter Donnelly, Jonathan Marchini. Genome-wide genetic data on ~500,000 UK Biobank participants. bioRxiv. 20 Jul 2017. doi: http://dx.doi.org/10.1101/166298.;20:1-36.23. Hoffmann TJ, Ehret GB, Nandakumar P, Ranatunga D, Schaefer C, Kwok PY, Iribarren C, Chakravarti A and Risch N. Genome-wide association analyses using electronic health records identify new loci influencing blood pressure variation. Nat Genet. 2017;49:54-64.24. Ahmad OB, Boschi-Pinto C, Lopez AD, Murray CJ, Lozano R and Inoue M. Age Standardization of Rates: A New Who Standard. GPE Discussion Paper Series: No.31. EIP/GPE/EBD. World Health Organization 2001. URL: http://www.who.int/healthinfo/paper31.pdf; Access Date: 28 July 2017.25. ISDScotland. GPD Support Population: Standard Population. URL: http://www.isdscotland.org/Products-and-Services/GPD-Support/Population/Standard-Populations/; (access date : 28 July 2017).26. Havulinna AS, Kettunen J, Ukkola O, Osmond C, Eriksson JG, Kesaniemi YA, Jula A, Peltonen L, Kontula K, Salomaa V and Newton-Cheh C. A blood pressure genetic risk score is a significant predictor of incident cardiovascular events in 32,669 individuals. Hypertension (Dallas, Tex : 1979). 2013;61:987-94.27. Kathiresan S, Melander O, Anevski D, Guiducci C, Burtt NP, Roos C, Hirschhorn JN, Berglund G, Hedblad B, Groop L, Altshuler DM, Newton-Cheh C and Orho-Melander M. Polymorphisms associated with cholesterol and risk of cardiovascular events. The New England journal of medicine. 2008;358:1240-9.28. Lyssenko V, Jonsson A, Almgren P, Pulizzi N, Isomaa B, Tuomi T, Berglund G, Altshuler D, Nilsson P and Groop L. Clinical risk factors, DNA variants, and the development of type 2 diabetes. The New England journal of medicine. 2008;359:2220-32.

2

29. Khera AV, Emdin CA, Drake I, Natarajan P, Bick AG, Cook NR, Chasman DI, Baber U, Mehran R, Rader DJ, Fuster V, Boerwinkle E, Melander O, Orho-Melander M, Ridker PM and Kathiresan S. Genetic Risk, Adherence to a Healthy Lifestyle, and Coronary Disease. The New England journal of medicine. 2016;375:2349-2358.30. Tyrrell J, Wood AR, Ames RM, Yaghootkar H, Beaumont RN, Jones SE, Tuke MA, Ruth KS, Freathy RM, Davey Smith G, Joost S, Guessous I, Murray A, Strachan DP, Kutalik Z, Weedon MN and Frayling TM. Gene-obesogenic environment interactions in the UK Biobank study. International journal of epidemiology. 2017;46:559-575.31. Padmanabhan S. Prospects for genetic risk prediction in hypertension. Hypertension (Dallas, Tex : 1979). 2013;61:961-3.32. Berglund L. Regression dilution bias: tools for correction methods and sample size calculation. Upsala journal of medical sciences. 2012;117:279-83.33. Berglund L, Garmo H, Lindback J and Zethelius B. Correction for regression dilution bias using replicates from subjects with extreme first measurements. Statistics in medicine. 2007;26:2246-57.34. Hutcheon JA, Chiolero A and Hanley JA. Random measurement error and regression dilution bias. BMJ (Clinical research ed). 2010;340:c2289.35. Masudi S, Yavari P, Mehrabi Y, Khalili D and Azizi F. Regression dilution bias in blood pressure and body mass index in a longitudinal population-based cohort study. Journal of research in health sciences. 2015;15:77-82.36. Dyer AR, Elliott P and Shipley M. Urinary electrolyte excretion in 24 hours and blood pressure in the INTERSALT Study. II. Estimates of electrolyte-blood pressure associations corrected for regression dilution bias. The INTERSALT Cooperative Research Group. American journal of epidemiology. 1994;139:940-51.

3

Figure Legends

Figure 1. Cumulative hazard rates according to genetic and lifestyle risk tertiles in the

UK Biobank study. The graphs compare different tertiles of genetic risk and lifestyle risk for

hazard of CVD (left hand panels), myocardial infarction (middle panels), and stroke (right

hand panels) (see Supplementary Table 2 for definition of CVD). Cox regression models

were adjusted for age, sex.

Figure 2. Predicted values and 95% CI of systolic blood pressure (SBP) and diastolic

blood pressure (DBP) in the UK Biobank cohort according to genetic and lifestyle risk

tertiles. Predicted values come from linear regression models adjusted for age and sex.

Figure 3. Risk of cardiovascular events (CVD), myocardial infarction (MI), stroke

events in the UK Biobank cohort according to tertiles of genetic risk score (GRS) and

lifestyle score. Low GRS corresponds to lowest tertile of the GRS. Favorable lifestyle

corresponds to top tertile of lifestyle score. Adjusted HRs and 95% CI were derived from Cox

regression models adjusted for age and sex.

4

Table 1. Baseline characteristics and incident events for UK Biobank participants by sex

Characteristics All participants

(N=277,005)

Males

(N=124,884)

Females

(N=152,121)

Age, mean (SD), years 56.3(8) 56.4(8.1) 56.3(7.9)

Males, N (%) 124884(45) NA(NA) NA(NA)

High blood pressure*, N, (%) 133786(52) 69098(59) 64688(46)

Anti-hypertensive medication, N (%) 47438(17) 24332(19) 23106(15)

Lipid treatment, N, (%) 35155(13) 20120(16) 15035(10)

Diabetes mellitus, diagnosed by doctor, N (%) 10958(4) 6462(5) 4496(3)

Body mass index mean (SD), kg/m2 27.2(4.7) 27.7(4.1) 26.8(5.1)

Sedentary lifestyle, mean (SD), hours/day 4.4(2.5) 4.9(2.7) 4(2.2)

Smoking, N (%)

Current 27788(10) 14797(12) 12991(9)

Past 138168(50) 65703(53) 72465(48)

Never 111049(40) 44384(36) 66665(44)

Alcohol intake, mean (SD), gr/day 17.6(20.9) 25.5(25.5) 11.2(12.9)

Fruit pieces (dry or fresh) daily consumption, mean

(SD) 4.8(3.2) 4.5(3.2) 5(3.1)

Vegetable (cooked or raw) consumption, mean (SD)  3(2.5) 2.6(2.5) 3.3(2.5)

Unprocessed meat consumption frequency, mean (SD) 3.7(1.7) 3.9(1.7) 3.5(1.8)

Processed meat consumption frequency, mean (SD) 1.9(1.1) 2.2(1) 1.6(1)

Fish consumption, mean (SD) 3.4(1.4) 3.4(1.4) 3.5(1.4)

Healthy (DASH) diet score ,median[IQR] 3[2,3] 2[2,3] 3[2,4]

Healthy lifestyle score- mean (SD) 6.1(2.2) 5.3(2.1) 6.8(2.1)

Genetic risk category- mean (SD) 0(0.9) 0(0.9) 0(0.9)

Systolic blood pressure, mean (SD), mmHg 140.5(20.5) 144.2(19.3) 137.4(21)

5

Diastolic blood pressure, mean (SD), mmHg 84.1(11.2) 86.4(11) 82.2(11)

Incident outcomes (non-fatal and fatal)

Composite Cardiovascular Disease, N (%) 9278(3.3) 6068(4.9) 3210(2.1)

Myocardial Infarction, N (%) 2984(1.1) 2155(1.7) 829(0.5)

Stroke, N (%) 1919(0.7) 1081(0.9) 838(0.6)

*Systolic blood pressure >140 or diastolic blood pressure >90 or the use of anti-hypertensive medication. SD: standard deviation; IQR: Inter-quintile range.

6