| Summary: | Hypertension is a common, highly heritable trait of complex aetiology. Multiple environmental and lifestyle factors contribute to blood pressure variation. Hence the study of hypertension causality is not straightforward. Genetic linkage studies have implicated a number of loci involved in blood pressure regulation and the development of hypertension. Candidate gene association studies, however, have not reported any reproducible associations. Early genome-wide association studies (GWAS) showed remarkable success in identifying validated common variants associated with common diseases such as coronary artery disease and type 1 diabetes. However, the first GWAS of hypertension showed little success. This was largely because of a lack of statistical power and insufficient genomic coverage. Furthermore, it is widely believed that the failure of one GWAS of hypertension was partly due to misclassification of controls that were not phenotyped for blood pressure. Subsequently, two large international consortia-run GWAS of blood pressure as a quantitative trait produced tangible results. The current study is a GWAS of hypertension using an extreme case-control design. It employed intensive phenotyping and extreme case-control definitions to select a sample of individuals from a restricted geographical area of relative homogeneity. The aim was to reduce misclassification bias and increase the likelihood of detecting any genetic effects. Cases were sampled from the Nordic Diltiazem study, and defined as individuals younger than 60 years with at least two consecutive measurements of systolic blood pressure (SBP) ≥ 160 mmHg or diastolic blood pressure (DBP) ≥ 100 mmHg. Controls were sampled from the prospective Malmö Diet and Cancer Study, and defined as individuals aged at least 50 years with SBP ≤ 120 mmHg and DBP ≤ 80 mmHg with no evidence of cardiovascular disease during ten years of follow-up. The groups represent, respectively, the upper 1.7% and lower 9.2% of the Swedish blood pressure distribution. Comparison of groups from the extreme tails of distribution increased statistical power by inflating observed effect sizes. With genome-wide SNP coverage we were able to adjust for population stratification using principal components analysis. Following quality control exclusions, a final set of 521,220 single nucleotide polymorphisms was available for analysis in 1,621 cases and 1,699 controls. Seventeen SNPs were associated with hypertension at a P < 1 × 10-5 threshold of significance, of which three attained genome-wide significance, defined as P < 5 × 10-7. The top hit, rs13333226, underwent a two stage validation process in a total of 14 independent cohorts. The combined odds ratio for the discovery cohort and all replication cohorts meta-analysed was 0.87 (95% CI 0.84 – 0.91, P = 3.67 × 10-11) with the minor G allele associated with a lower risk of hypertension. In total 21,466 cases and 18,240 controls were included. After adjustment for age, age2, sex, and BMI, and when the discovery cohort was excluded from analysis, the association remained significant. Estimated glomerular filtration rate (eGFR), a measure of kidney function, was available in seven of the cohorts. When the analysis was repeated with adjustment for eGFR the effect was marginally strengthened. rs13333226 is located in close proximity, at -1617 base pairs, to the uromodulin (UMOD) transcription start site. UMOD encodes uromodulin, also known as the Tamm-Horsfall protein. Uromodulin is produced predominantly in the thick ascending limb of the loop of Henle and is the most abundant protein in urine. Its function is unclear; however, variants in UMOD have been associated with chronic kidney disease. Clinical functional studies were conducted in three separate populations. The minor G allele of rs13333226 (associated with a lower risk of hypertension) was associated with lower urinary uromodulin excretion. Furthermore, in one sample following a low salt diet urinary uromodulin excretion was significantly lower in the presence of the G allele, whereas after a high salt diet genotype was no longer associated with urinary uromodulin. If this were verified, this would entail a gene-environment interaction. Our combined results suggest that UMOD may have a role in regulating blood pressure, possibly through an effect on sodium homeostasis. There is ample evidence of a strong, graded relationship between blood pressure and subsequent renal disease. Hence the current finding is biologically plausible. Information on kidney disease was not available for the discovery samples so this could not be explored. However, the association between rs13333226 and hypertension was not substantively altered by adjustment for eGFR in the seven validation cohorts in which it was recorded, suggesting that it is independent of renal function. In conclusion, we have performed a GWAS of hypertension using an extreme case-control design. The most significant hit was validated in a meta-analysis of the discovery sample and 14 additional cohorts. Moreover, functional studies showed a relationship between genotype and urinary protein excretion. Overall, we demonstrate that with careful methodological planning and phenotyping it is possible to generate replicable hypertension GWAS results in a relatively small sample size.
|
|---|
