Epidemiology/Population
Clinical Effect of Naturally Random Allocation
to Lower Systolic Blood Pressure Beginning Before
the Development of Hypertension
Brian A. Ference, Stevo Julius, Nitin Mahajan, Phillip D. Levy, Kim Allan Williams Sr, John M. Flack
Abstract—Systolic blood pressure (SBP) rises approximately linearly with age in most societies. The cause of this rise
is unclear. We tested the hypothesis that SBP is causally associated with the rate of rise in SBP with age by evaluating
the effect of 12 polymorphisms associated with lower SBP on the age-related rate of rise in SBP in a series of m
­ etaregression analyses involving ≤199 477 participants in 63 studies. We then evaluated the effect of these polymorphisms
on the odds of coronary heart disease in 22 223 case and 64 762 control subjects and compared it with the effect of lower
SBP observed in both prospective cohort studies and blood pressure–lowering randomized trials. All 12 polymorphisms
were associated with both lower SBP and a slower age-related rise in SBP. The weighted mean effect of these 12
polymorphisms was associated with a 0.32-mm Hg lower SBP (P=1.79×10−7) and a 0.093-mm Hg/decade slower ­agerelated rise in SBP (P=3.05×10−5). The effect of long-term exposure to lower SBP on coronary heart disease mediated
by these polymorphisms was 2-fold greater than that observed in prospective cohort studies (P=0.006) and 3-fold greater
than that observed in short-term blood pressure treatment trials (P=0.001). We conclude therefore that SBP seems to
be causally associated with the rate of rise in SBP with age and has a cumulative effect on the risk of coronary heart
disease. (Hypertension. 2014;63:1182-1188.) Online Data Supplement
•
Key Words: hypertension ◼ Mendelian randomization analysis ◼ prevention
S
ystolic blood pressure (SBP) increases approximately
linearly with age in most industrialized societies.1,2 The
cause of this rise is unclear. Importantly, however, SBP does
not seem to rise with age in all societies, thus suggesting that
rising SBP with age is not an inevitable part of aging but rather
may be preventable.2
Indeed, several studies have reported that persons with
higher baseline SBP seem to experience a faster rate of rise
in SBP over time as compared with persons with lower levels of SBP.1–3 This observation has led to the hypothesis that
increased SBP causes vascular injury or dysfunction, which
in turn leads to higher SBP and a cycle of accumulating vascular injury that manifests as increasing SBP with age.3 This
hypothesis has potentially important clinical implications
because if increased SBP causes blood pressure to rise with
age, then lowering SBP beginning before the development of
hypertension may prevent or slow the usual age-related rise
in SBP.
We sought to evaluate whether SBP has a causal effect on
the rate of rise in SBP with age by appealing to the principle
of Mendelian randomization. Multiple polymorphisms are
associated with SBP,4 and each of these polymorphisms is
&
control
allocated approximately randomly at the time of conception in
a process referred to as Mendelian randomization.5 Therefore,
evaluating the effect of higher or lower SBP mediated by these
polymorphisms should provide a naturally randomized and
therefore unconfounded estimate of the causal effect of SBP
on the usual age-related rise in SBP. Specifically, if SBP is
causally related to the rate of rise in SBP with age, then polymorphisms associated with lower SBP should also be independently associated with a slower rise in SBP over time. By
contrast, if SBP is not in this causal pathway, then naturally
random allocation to higher or lower SBP mediated by genetic
polymorphisms should not have any effect on the observed
rate of rise in SBP over time.
We therefore tested the hypothesis that SBP is causally
associated with age-related rise in SBP by evaluating the association between lower SBP mediated by a variety of different
genetic polymorphisms and the rate of rise in SBP over time.
In addition, we constructed genetic SBP scores to estimate
the magnitude of the clinical effect of exposure to lower SBP
beginning before the development of hypertension on both the
usual rate of rise in SBP over time and the risk of coronary
heart disease (CHD).
Received October 31, 2013; first decision November 13, 2013; revision accepted February 6, 2014.
From the Divisions of Translational Research and Clinical Epidemiology (B.A.F., J.M.F.) and Cardiovascular Medicine (B.A.F., N.M., K.A.W.),
Cardiovascular Research Institute (P.D.L.), and Department of Internal Medicine (J.M.F.), Wayne State University School of Medicine, Detroit, MI; and
Division of Cardiovascular Medicine, University of Michigan Medical School, Ann Arbor (S.J.).
The online-only Data Supplement is available with this article at http://hyper.ahajournals.org/lookup/suppl/doi:10.1161/HYPERTENSIONAHA.
113.02734/-/DC1.
Correspondence to Brian A. Ference, Division of Translational Research and Clinical Epidemiology (TRaCE), Division of Cardiovascular Medicine,
Wayne State University School of Medicine, UHC 2E2, Detroit, MI 48202. E-mail bference@med.wayne.edu
© 2014 American Heart Association, Inc.
Hypertension is available at http://hyper.ahajournals.org
DOI: 10.1161/HYPERTENSIONAHA.113.02734
1182
Ference et al Naturally Random Allocation to Lower SBP 1183
Methods
case and 64 762 control subjects in the Coronary ARtery DIsease
Genome-Wide Replication And Meta-Analysis (CARDIoGRAM)
consortium,9 and divided this effect estimate (and its standard error)
by the effect of that exposure allele on SBP as reported in the ICBP.4
We then created a genetic SBP score by combining these weighted effect estimates in a meta-analysis to obtain a summary estimate of the
magnitude of the effect of each mm Hg lower SBP on the risk of CHD.
We compared the magnitude of the effect of genetically mediated lifelong exposure to lower SBP on the risk of CHD with the magnitude
of the effect observed per mm Hg lower SBP in a meta-analysis of
prospective epidemiological cohort studies10 and in an updated metaanalysis of hypertension treatment randomized controlled trials conducted among persons without a history of cardiovascular disease.11
A detailed description of the methods is provided in the ­online-only
Data Supplement. All meta-analyses and meta-regression analyses
were performed using inverse-variance weighted random-effects
models. All analyses used publically available summary data and
therefore inherited all adjustments made in the original analyses (including adding either 10 or 15 mm Hg to the measured SBP among
persons being treated with blood pressure–lowering medications).
Effect estimates were compared using a Z test. All statistical tests
used a 2-sided α <0.05 as the threshold for nominal statistical significance, and all analyses were performed using STATA (version 12).
To begin, we searched PubMed, Embase, and Index of Science, as
well as the supplementary materials, online data repositories, and references of selected articles, to identify individual studies that reported
the association between SBP and ≥1 of the 26 single-nucleotide polymorphisms (SNPs) that were associated with SBP at ­genome-wide
level of significance in the International Consortium for Blood
Pressure Genome-Wide Association Studies (ICBP).4 For each polymorphism, we defined the exposure allele as the allele associated with
lower SBP.
To test the hypothesis that SBP is causally associated with the rise
in SBP with age, we performed a meta-regression analysis measuring the effect modification of age on the association between each
exposure allele and SBP, respectively. Because of the conceptual and
numeric equivalence between a regression term in a meta-regression
model and an interaction term in a general linear model, and because
a single interaction term defines the effect modification between any
2 exposure variables, this analysis provides both an estimate of the
change in the effect of an exposure allele on SBP with increasing age
and an estimate of the difference in the age-related rise in SBP per
copy of the exposure allele.6 We then combined the β-coefficients
representing the difference in age-related rise in SBP per copy of the
exposure allele for each SNP in a meta-analysis to produce a summary estimate of the effect of lower SBP on the rise in SBP over time.
Next, we looked up the reported effect of each exposure allele on
the change in SBP over time (defined as the difference in SBP measured at baseline and after re-examination a mean of 23 years later)
among participants in the Malmo Preventive Project (MPP).7 We then
combined these effect estimates in a meta-analysis to estimate the
effect of lower SBP mediated by the combined effect of these polymorphisms on the rate of rise in SBP over time and compared it with
the results of the meta-regression analyses.
To estimate the magnitude of the effect of each mm Hg lower SBP
on the rate of rise in SBP over time, we standardized the effect of each
exposure allele by dividing its effect on the age-related rate of rise
in SBP (and the corresponding standard error) by the effect of that
SNP on SBP measured in mm Hg at 40 years of age (with this effect
estimate derived by centering the meta-regression analysis around
40 years of age). We then created a genetic SBP score by combining
these weighted effect estimates in a meta-analysis to obtain a summary estimate of the magnitude of the effect of naturally random allocation to each mm Hg lower SBP on the age-related rise in SBP.8
To estimate the effect of lifelong exposure to lower SBP on the risk
of CHD mediated by these polymorphisms, we looked up the effect of
each exposure allele on the risk of CHD, as reported among ≤22 223
Table. Results
Sufficient data to permit detailed meta-regression analyses
were available for 12 of the 26 polymorphisms associated
with SBP in the ICBP (Table S1 in the online-only Data
Supplement). In separate meta-regression analyses involving ≤23 studies and 163 674 participants, the exposure allele
for all 12 polymorphisms was associated with both a lower
SBP (Table) and a slower rate of rise in SBP over time
(Figure 1).4,12–17 Overall, the weighted mean effect of these 12
polymorphisms was associated with a 0.31-mm Hg lower SBP
(95% confidence interval [CI], 0.15–0.48; P=1.34×10−4) at a
mean age of 43.2 years and a 0.092-mm Hg/decade slower rise
in SBP with age (95% CI, 0.045–0.139; P=6.26×10−4).
The effect of 11 of these 12 polymorphisms on the change in
SBP over time was also evaluated in the MPP.7 Among 11 290
persons enrolled in MPP without hypertension at baseline, the
weighted mean effect of these 11 exposure alleles was associated with a 0.32-mm Hg lower SBP (95% CI, 0.14–0.49;
Included Polymorphisms and Their Association With Systolic Blood Pressure
Nearby Gene(s)
Chr
SNP
Coded Allele Frequency
SBP Effect Size,
mm Hg*
SE
P Value
MTHFR-NPPB
1
rs17367504
0.15
−0.903
0.0942
8.72E−22
FGF5
4
rs1458038
0.71
−0.706
0.0705
1.47E−23
NPR3-C5orf23
5
rs1173771
0.40
−0.504
0.0612
1.79E−16
CACNB2(3′)
10
rs1813353
0.32
−0.569
0.0812
2.56E−12
C10orf107
10
rs4590817
0.16
−0.646
0.0931
3.97E−12
CYP17A1-NT5C2
10
rs11191548
0.09
−1.095
0.1041
6.90E−26
PLEKHA7
11
rs381815
0.74
−0.575
0.0876
5.27E−11
ATP2B1
12
rs17249754
0.16
−0.928
0.1059
1.82E−18
SH2B3
12
rs3184504
0.53
−0.598
0.0688
3.83E−18
TBX5-TBX3†
12
rs10850411
0.30
−0.354
0.0651
5.38E−08
CYP1A1-ULK3
15
rs1378942
0.65
−0.613
0.0621
5.69E−23
ZNF652
17
rs12940887
0.62
−0.362
0.0568
1.79E−10
Chr indicates chromosome; SBP, systolic blood pressure; and SNP, single-nucleotide polymorphism.
*From the International Consortium of Blood Pressure Genome-Wide Association Studies.4
†TBX5-TBX3 included because it was only marginally below genome-wide level of significance (P=5.38×10−8).
1184 Hypertension June 2014
Figure 1. Associations between exposure alleles and rate of rise in systolic blood pressure (SBP) over time in meta-regression analyses.
Boxes represent the summary point estimate of effect for the association between each exposure allele on rate of rise in SBP per decade
of age. Bars represent 95% confidence interval (CI). Diamond represents summary estimate of effect weighted by the inverse variance of
each point estimate. ES indicates effect size in mm Hg per decade; and SNP, single-nucleotide polymorphism.
P=3.7×10−4) at a mean age of 45.6 years and a 0.094-mm Hg/
decade slower rise in SBP over time (95% CI, 0.017–0.171;
P=0.017; Table S2, Figure 2). This estimate of the effect of
exposure to lower SBP on the rate of rise in SBP with age
among participants enrolled in MPP agreed closely with the
results of the meta-regression analyses (Figure 3). In an analysis that combined data from the meta-regression analyses and
the MPP, the weighted mean effect of these exposure alleles was
associated with a 0.32-mm Hg lower SBP (95% CI, 0.20–0.43;
P=1.79×10−7) and a 0.093-mm Hg/decade slower rise in SBP
with age (95% CI, 0.05–0.13; P=3.05×10−5).
Using a weighted genetic SBP score composed of all 12
polymorphisms and data from the meta-regression analyses,
naturally random allocation to each mm Hg lower SBP was
associated with a 0.324-mm Hg/decade slower rise in SBP
(95% CI, 0.162–0.486; P=9.32×10−4; Table S3). This effect estimate agreed closely with the 0.349-mm Hg/decade slower rise
in SBP per mm Hg lower SBP observed in the MPP (95% CI,
0.205–0.492; P=1.59×10−6), estimated using a genetic SBP
score composed of 29 polymorphisms associated with either
SBP or diastolic blood pressure.7
To further explore the effect of exposure to lower SBP on
the rate of rise in SBP with age, we plotted the effect of 10
polymorphisms on SBP, where the effect of each polymorphism was measured at multiple different common points in
time.7,17 The absolute weighted mean effect of these polymorphisms on SBP increased approximately linearly over time
from 10 to 63 years of age (Figure 4) by 0.072 mm Hg/decade
Figure 2. Associations between exposure
alleles and rate of rise in systolic blood
pressure (SBP) over time in Malmo
Preventive Project. Boxes represent the
summary point estimate of effect for the
association between each exposure allele
on rate of rise in SBP per decade of age.
Bars represent 95% confidence interval
(CI). Diamond represents summary
estimate of effect weighted by the
inverse variance of each point estimate.
ES indicates effect size in mm Hg per
decade; and SNP, single-nucleotide
polymorphism.
Ference et al Naturally Random Allocation to Lower SBP 1185
Figure 3. Effect of lower systolic blood pressure (SBP) mediated
by combined effect of included polymorphisms on rate of rise in
SBP over time. Boxes represent the summary point estimate of
effect for the association between each exposure allele on rate of
rise in SBP per decade of age. Bars represent 95% confidence
interval. Diamond represents summary estimate of effect
weighted by the inverse variance of each point estimate.
(95% CI, 0.026–0.114; P=1.75×10−3). The slope of this
regression line represents both an estimate of the change in
the mean effect of these polymorphisms on SBP with increasing age and the difference in the age-related rate of rise in
SBP attributable to the effect of lower SBP mediated by these
polymorphisms, because these 2 estimates of effect modification are numerically equivalent. Therefore, this analysis also
provides evidence that lower SBP is associated with a slower
rise in SBP with age.
Using a separate weighted genetic SBP score composed
of the same 12 polymorphisms, naturally random allocation to lifetime exposure to each mm Hg lower SBP was
associated with a 6.1% lower odds of CHD (odds ratio,
0.94; 95% CI, 0.91–0.97; P=3.85×10−4), whereas 5-mm Hg
lower SBP was associated with a 27.0% (odds ratio, 0.73;
95% CI, 0.61–0.87; P=3.85×10−4) lower odds of CHD, and
10-mm Hg lower SBP was associated with a 46.7% (odds
ratio, 0.53; 95% CI, 0.38–0.75; P=3.85×10−4) lower odds
of CHD among participants in the CARDIoGRAM studies
(Figure 5A, Table S4a). This effect estimate was essentially
unchanged when measured using a genetic SBP score consisting of 25 of the 26 SNPs associated with SBP in the
ICBP that were measured in the CARDIoGRAM studies
(Figure 5A, Table S4b).9
The magnitude of the effect of lifetime exposure to lower
SBP observed in the genetic studies was >2-fold greater (on
the log scale) per mm Hg than the magnitude of the effect
observed in a meta-analysis of 61 prospective cohort studies
involving almost 1 million persons (P difference=0.006)10 and
≈3-fold greater (on the log scale) per mm Hg lower SBP than
that observed in a meta-analysis of 27 hypertension treatment
randomized controlled trials involving 109 792 participants
without a prior history of cardiovascular disease (P difference=
0.001; Tables S5 and S6).11 A plot of the proportional risk
reduction against the mean length of follow-up in these 3
analyses suggested an approximately log-linear relationship
between the total length of exposure to lower SBP and the
reduced risk of CHD (Figure 5B).
Discussion
We tested the hypothesis that SBP is causally related to the
rise in SBP with age by evaluating the association between
lower SBP mediated by a variety of genetic polymorphisms
and the average rate of rise in SBP over time. Because genetic
polymorphisms are allocated approximately randomly at the
time of conception, these associations should provide an
Figure 4. Association between effect of exposure alleles and systolic blood pressure (SBP) measured at different ages. Boxes represent
the combined weighted mean effect of 10 exposure alleles on SBP (mm Hg) measured at 4 different ages: 10.7 and 38.7 years in the
Young Finns Cohort17 and 45 and 63 years in the Malmo Preventive Project.7 Vertical lines represent 95% confidence interval (CI). The line
represents the absolute increase in the combined effect of these 10 polymorphisms on SBP per year of increasing age.
1186 Hypertension June 2014
Figure 5. Effect of exposure to 10-mm Hg
lower systolic blood pressure (SBP) on
risk of coronary heart disease (CHD).
A, Boxes represent the summary point
estimate of effect for the association
between 10-mm Hg lower SBP and
the risk of CHD. Bars represent 95%
confidence interval (CI). B, Boxes
represent the proportional risk reduction
(1−odds ratio [OR]) of CHD plotted
against the mean length of follow-up
measured in years for each group of
studies. Follow-up for lifetime exposure
to lower SBP assessed in the genetic
studies was estimated by the mean age
of the case subjects in each analysis.
Vertical lines represent 1 SE above and
below the point estimate of proportional
risk reduction. Studies are plotted in
order of increasing length of ­followup. The line represents the increase in
proportional risk reduction of CHD per
year of increasing exposure to lower
SBP. HTN indicates hypertension; RCT,
randomized controlled trial; and RRR,
relative risk reduction.
unconfounded estimate of the effect of exposure to lower
SBP on the usual rise in SBP with age in a manner analogous to a long-term natural randomized trial.8 We found that
naturally random allocation to lower SBP mediated by multiple different common polymorphisms was associated with
a significantly slower rise in SBP with age and a much larger
reduction in the risk of CHD per mm Hg lower SBP than that
observed in either prospective epidemiological cohort studies
or short-term blood pressure–lowering randomized trials. We
conclude therefore that SBP seems to be causally associated
with the rate of rise in SBP with age and has a cumulative
effect on the risk of CHD.
Multiple different lines of evidence in our study support the
notion that SBP is causally associated with the rate of rise
in SBP. The results of our study thus strongly support the
hypothesis that increased SBP causes vascular injury, which
in turn leads to higher SBP and a cycle of accumulating vascular injury that leads to increasing SBP over time. The direct
implication of this finding is that lowering SBP beginning
soon after it begins to rise with age and before the development of hypertension may potentially interrupt the cycle of
accumulating vascular injury that appears to at least in part
cause increasing SBP over time and thereby prevent or slow
the usual age-related rise in SBP.
It is important to note, however, that we evaluated the effect
of genetically mediated lifelong exposure to lower SBP. It is
not known whether lowering SBP with diet, exercise, weight
loss, or medications will have the same effect as genetically
mediated lower SBP. Further studies are needed to determine
whether lowering SBP beginning soon after it begins to rise
with age and before significant vascular injury has developed
can prevent or slow the usual age-related rise in SBP.
Preventing or slowing the usual rise in SBP with age may
be a particularly important public health goal because SBP
seems to have a cumulative effect on the risk of CHD. We
found that lifetime exposure to each mm Hg lower SBP was
associated with a 2-fold greater reduction in the risk of CHD
than that observed in a meta-analysis of prospective cohort
studies (corrected for regression dilution bias)10 and a 3-fold
greater effect than that observed in a meta-analysis of randomized trials comparing blood pressure–lowering treatment
with placebo (or usual care) among persons with established
hypertension.11 The substantial and graded difference in the
risk of CHD associated with lifetime exposure to lower SBP
assessed in the genetic studies, medium-term exposure to
lower SBP assessed in the epidemiological cohort studies, and
short-term exposure to lower SBP assessed in the randomized
trials suggests that the effect of SBP on the risk of CHD is
Ference et al Naturally Random Allocation to Lower SBP 1187
not a threshold effect, but rather is cumulative. This finding
may explain the observation that antecedent blood pressure
is independently associated with the risk of CHD even after
adjusting for current blood pressure in multiple different
observational epidemiological studies.18,19 It may also explain
the legacy effect observed in blood pressure–lowering trials in
which the clinical benefit observed during the trial persists for
several years after the trial was completed.20
The results of our study have important implications for
interpreting other blood pressure genetic studies. We found
that the absolute magnitude of the measured effect of polymorphisms associated with SBP appear to increase over time
(Figure 4). The apparently increasing effect of genetic polymorphisms on SBP over time can be explained by the fact that
SBP is causally associated with the age-related rise in SBP.
Therefore, the effect of a polymorphism associated with SBP,
measured at any point in time, will actually be a composite of
the effect of that polymorphism on SBP plus the additional
effect that accrues over time because of the effect of SBP
mediated by that polymorphism on the rate of rise in SBP over
time. As a result, estimating the effect of SBP on the rate of
rise in SBP with age, the odds of CHD, or any other outcome
using a genetic score that adjusts the effect of each exposure
allele by its effect on SBP measured in middle age will systematically underestimate the true magnitude of the effect of
SBP on the outcome under study.
Our study has several limitations, including inherent limitations when making causal inferences using the principle of
Mendelian randomization.21 To address these limitations, we
limited our analyses to SNPs that were reported to be associated with SBP at genome-wide level of significance in the
ICBP.4 Some of these polymorphisms, however, have been
reported to have pleiotropic effects in addition to their effect
on SBP.4 We therefore based our conclusions on the weighted
mean effect of these polymorphisms using a genetic SBP
score. Because an SBP genetic score measures the effect of
naturally random allocation to each unit lower SBP mediated
by the combined effect of the SNPs included in the score, the
results of our study should reflect the effect of lower SBP
itself rather than the effect of any particular polymorphism
or pathway. Including multiple polymorphisms each of which
presumably acts through a different mechanism to lower SBP
significantly reduces the likelihood that the analysis will be
confounded by pleiotropy because it is unlikely that each of
the SNPs included in the genetic SBP score acts through similar and directionally consistent pleiotropic effects. However,
the Mendelian randomization paradigm does not allow us to
exclude the possibility that each of the included SNPs is associated with ≥1 broad functional pathways such as inflammation that may independently influence both SBP and the rate
of rise in SBP with age. Further study is needed to elucidate
the mechanisms by which SBP seems to influence the rate of
rise in SBP over time.
Perspective
We found that SBP seems to be causally associated with the
rate of rise in SBP with age and has a cumulative effect on
the risk of CHD. Therefore, the results of our study suggest
that lowering SBP beginning soon after it begins to rise with
age and before the development of hypertension may have
the potential to significantly slow the usual age-related rise in
SBP and substantially reduce the lifetime risk of CHD.
Acknowledgments
B.A. Ference had full access to all of the data in the study and takes
responsibility for the integrity of the data and the accuracy of the data
analysis.
Disclosures
None.
References
1. Franklin SS, Gustin W 4th, Wong ND, Larson MG, Weber MA, Kannel
WB, Levy D. Hemodynamic patterns of age-related changes in blood
pressure. The Framingham Heart Study. Circulation. 1997;96:308–315.
2.Carvalho JJ, Baruzzi RG, Howard PF, Poulter N, Alpers MP, Franco
LJ, Marcopito LF, Spooner VJ, Dyer AR, Elliott P. Blood pressure
in four remote populations in the INTERSALT Study. Hypertension.
1989;14:238–246.
3. Wu M, Ware JH, Feinleib M. On the relation between blood pressure
change and initial value. J Chronic Dis. 1980;33:637–644.
4. Ehret GB, Munroe PB, Rice KM, et al. Genetic variants in novel pathways influence blood pressure and cardiovascular disease risk. Nature.
2011;478:103–109.
5.Lawlor DA, Harbord RM, Sterne JA, Timpson N, Davey Smith G.
Mendelian randomization: using genes as instruments for making causal
inferences in epidemiology. Stat Med. 2008;27:1133–1163.
6. Ference BA, Yoo W, Flack JM, Clarke M. A common KIF6 polymorphism increases vulnerability to low-density lipoprotein cholesterol: two
­meta-analyses and a meta-regression analysis. PLoS One. 2011;6:e28834.
7. Fava C, Sjögren M, Montagnana M, Danese E, Almgren P, Engström
G, Nilsson P, Hedblad B, Guidi GC, Minuz P, Melander O. Prediction
of blood pressure changes over time and incidence of hypertension by a
genetic risk score in Swedes. Hypertension. 2013;61:319–326.
8. Palmer TM, Lawlor DA, Harbord RM, Sheehan NA, Tobias JH, Timpson
NJ, Davey Smith G, Sterne JA. Using multiple genetic variants as instrumental variables for modifiable risk factors. Stat Methods Med Res.
2012;21:223–242.
9. Lieb W, Jansen H, Loley C, et al; CARDIoGRAM. Genetic predisposition to higher blood pressure increases coronary artery disease risk.
Hypertension. 2013;61:995–1001.
10.Lewington S, Clarke R, Qizilbash N, Peto R, Collins R; Prospective
Studies Collaboration. Age-specific relevance of usual blood pressure
to vascular mortality: a meta-analysis of individual data for one million
adults in 61 prospective studies. Lancet. 2002;360:1903–1913.
11. Law MR, Morris JK, Wald NJ. Use of blood pressure lowering drugs
in the prevention of cardiovascular disease: meta-analysis of 147 randomised trials in the context of expectations from prospective epidemiological studies. BMJ. 2009;338:b1665.
12. Levy D, Ehret GB, Rice K, et al. Genome-wide association study of blood
pressure and hypertension. Nat Genet. 2009;41:677–687.
13. Newton-Cheh C, Johnson T, Gateva V, et al; Wellcome Trust Case Control
Consortium. Genome-wide association study identifies eight loci associated with blood pressure. Nat Genet. 2009;41:666–676.
14. Johnson AD, Newton-Cheh C, Chasman DI, et al; Cohorts for Heart and
Aging Research in Genomic Epidemiology Consortium; Global BPgen
Consortium; Women’s Genome Health Study. Association of hypertension drug target genes with blood pressure and hypertension in 86,588
individuals. Hypertension. 2011;57:903–910.
15.Johnson T, Gaunt TR, Newhouse SJ, et al; Cardiogenics Consortium;
Global BPgen Consortium. Blood pressure loci identified with a
­gene-centric array. Am J Hum Genet. 2011;89:688–700.
16. Ho JE, Levy D, Rose L, Johnson AD, Ridker PM, Chasman DI. Discovery
and replication of novel blood pressure genetic loci in the Women’s
Genome Health Study. J Hypertens. 2011;29:62–69.
17. Oikonen M, Tikkanen E, Juhola J, et al. Genetic variants and blood pressure in a population-based cohort: the Cardiovascular Risk in Young Finns
study. Hypertension. 2011;58:1079–1085.
18.Vasan RS, Massaro JM, Wilson PW, Seshadri S, Wolf PA, Levy D,
D’Agostino RB. Antecedent blood pressure and risk of cardiovascular
disease: the Framingham Heart Study. Circulation. 2002;105:48–53.
1188 Hypertension June 2014
19.Allen N, Berry JD, Ning H, Van Horn L, Dyer A, Lloyd-Jones DM.
Impact of blood pressure and blood pressure change during middle age on
the remaining lifetime risk for cardiovascular disease: the cardiovascular
lifetime risk pooling project. Circulation. 2012;125:37–44.
20.Kostis WJ, Thijs L, Richart T, Kostis JB, Staessen JA. Persistence
of mortality reduction after the end of randomized therapy in
clinical trials of blood pressure-lowering medications. Hypertension.
2010;56:1060–1068.
21. Nitsch D, Molokhia M, Smeeth L, DeStavola BL, Whittaker JC, Leon
DA. Limits to causal inference based on Mendelian randomization:
a comparison with randomized controlled trials. Am J Epidemiol.
2006;163:397–403.
Novelty and Significance
What Is New?
• Using naturally randomized evidence, we show for the first time that
systolic blood pressure (SBP) seems to be causally associated with the
rate of rise in SBP with age and has a cumulative effect on the risk of
coronary heart disease.
What Is Relevant?
• The finding that exposure to lower SBP beginning before the development of hypertension is associated with a slower age-related rise in SBP
suggests that lowering SBP beginning before the development of hypertension may prevent or slow the usual age-related rise in SBP.
• The finding that SBP seems to have a cumulative effect on the risk of
coronary heart disease suggests that preventing or slowing the usual
age-related rise in SBP has the potential to substantially reduce the lifetime risk of coronary heart disease.
Summary
The association between SBP genetic variants and the rate of rise
in SBP over time is consistent with a possible causal effect of SBP
on the usual age-related rise in SBP.