We provide results of genome-wide association study for 6 blood pressure and 5 arterial stiffness phenotypes in a carefully characterized study sample. Association analyses and linkage reveal a number of intriguing results. For the GEE model of additive genetic effects (Table ) there were 7 SNPs with p values < 10-5 for blood pressure and 5 for arterial stiffness phenotypes. Among the GEE additive effect model results the most likely candidate genes were MEF2C, SYNE1, and TNFSF11, which were associated with arterial stiffness. We have not yet attempted replication of our results. Follow-up genotyping of the top SBP and DBP SNPs reported in our study sample in additional Framingham participants is planned; additional replication attempts will be needed in independent samples to confirm any of the association results we report.
FBAT (Table ) identified association of COL8A1
with arterial stiffness (p value 6 × 10-6
for rs792833). This gene codes for type VIII collagen, which is produced by aortic endothelial cells [17
], suggesting a biologically plausible association.
Linkage yielded a LOD score of 3, approaching genome-wide significance, for long-term SBP on chromosome 15. A meta-analysis of blood pressure and hypertension linkage studies did not identify this as a region of interest [9
]. The lower LOD scores for long-term SBP on chromosome 17 (~67 cM) in this investigation compared with our prior findings [6
] appears to be largely due to differences in phenotype definition of long-term SBP with the exclusion of early examination BP values in the original cohort participants and the inclusion of offspring cohort examination 7 blood pressures in this analysis. When linkage analyses were repeated with the inclusion of the early original cohort exams using the prior phenotype definitions, the same linkage peak on chromosome 17 emerged (LOD > 4).
For tonometry phenotypes, we found LOD scores for reflected wave amplitude of 5.0 (chromosome 8 at 19 Mb) and 3.2 (chromosome 4, 169 Mb) near peaks for this phenotype that we previously reported in a largely overlapping study sample [7
]. Similarly, we once again identified a linkage peak for carotid-femoral pulse wave velocity (LOD 3.0; chromosome 2 at 74 mb).
Compared with the primary GEE model for additive genetic effects (Table ), a different set of SNPs was identified in secondary GEE general effects models (Table ) for long-term DBP and SBP, including 2 SNPs with p values < 10-6
. Differences in model results may be due to the greater sensitivity of the general model to detect recessive genotype effects. SNPs in CCL20
, and LPP
were associated with both long-term SBP and DBP. GEE general genetic effects models for arterial stiffness phenotypes yielded the lowest p value (p = 1.99 × 10-7
) for rs3773643 in TGFBR2
, which has been implicated in aortic aneurysm and Loeys-Dietz syndrome [18
]. Disruption of the aortic wall would be expected to affect arterial stiffness.
Due to high correlations of SBP and DBP (within examination r = 0.77; long-term r = 0.82), joint analyses of SBP and DBP added little to what was identified in individual phenotype analyses. In contrast, joint analyses of the five tonometry phenotypes, which are less highly correlated, identified LOXL2
, and MEF2C
as attractive candidates. LOXL2
is a member of the lysyl oxidase family of enzymes that initiate cross-linking of collagens and elastin, and alter arterial elasticity [20
]. Collagen and elastin cross-links are critical to tensile strength of the extracellular matrix. Mice null for lysyl oxidase (LOX
) die perinatally from aortic aneurysm [21
is involved in cardiac morphogenesis and extracellular matrix remodeling [22
is involved in aortic vascular smooth muscle differentiation [23
]. To our knowledge, genetic variation in these genes has not previously been shown to be associated with alterations in arterial properties in humans. Whether our results provide nominal evidence of such association or merely chance findings remains to be determined.
Since none of the primary associations attained genome-wide significance, this investigation should be viewed as hypothesis generating. Association analyses for SNPs in six renin-angiotensin-aldosterone pathway genes showed weak evidence of association. Negative results for these candidate genes may be due in part to incomplete linkage disequilibrium coverage of these genes by the SNPs in this genome-wide scan. It is likely that the vast majority of low p values from association analyses are due to chance. Replication studies in other populations, using a genome-wide approach or selective genotyping is needed to establish if any of our results are indicative of true positive associations.
We provide results of genome-wide association testing for blood pressure and arterial stiffness phenotypes obtained in a carefully described community-based sample of adults who were recruited without regard to disease status. Additional studies are needed to validate these results. Finding genetic variants associated with hypertension or altered arterial properties may aid in the identification of high risk individuals and in the development of new targeted therapies for hypertension. Our report is one of the earlier genome-wide association studies of blood pressure. Several additional studies, some with larger sample size and others with more dense genome-wide coverage of common variation will follow. In that regard, a 550 k SNP genome-wide association study in approximately 9400 Framingham Heart Study participants across three generations is underway and results from that study will help in the interpretation of the findings we report in this manuscript.