ET (Endothelin)-1 and Ischemic Heart Disease
A Mendelian Randomization Study
- coronary artery disease
- endothelin receptor antagonists
- genome-wide association study
- myocardial infarction
ET (endothelin)-1 is a vasoconstrictive peptide. ET receptor antagonists are used to treat pulmonary hypertension. Whether ET-1 is a target of intervention in cardiovascular disease or a symptom of vascular damage is unclear. A genetic variant promoting ET-1 expression was recently found positively associated with ischemic heart disease (IHD).1 Here, we used separate sample Mendelian randomization (MR), that is, instrumental variable analysis with genetic instruments, to assess whether ET-1 is associated with IHD and for completeness with myocardial infarction (MI) and some of their major risk factors.
We obtained 3 single nucleotide polymorphisms (SNPs; rs4253238 [KLKB1], rs2731672 [F12], and rs5370 [EDN1]) strongly (P value<5×10−8) and independently (r2<0.01) associated with CT-pro-ET-1 (C-terminal-pro-endothelin-1), a biomarker of ET-1, from a genome-wide association study in 6674 people of European ancestry, mean age 49 years, 47% men, adjusted for age and sex, with no evidence of population stratification.2 These 3 SNPs explained 14.3% of the variance in CT-pro-ET-1.2 rs2731672 was not available for all outcomes, so was replaced with a correlated (r2=0.93 using LDlink with 1000 Genomes catalog for Europeans) proxy, rs2545801 (F12). We applied these genetic predictors of CT-pro-ET-1 to the largest publically available extensively genotyped IHD case (n≤76014)–control (n≤264785) study (largely based on the UK Biobank SOFT CAD, CARDIoGRAMplusC4D 1000 Genomes, and the MIGen/CARDIoGRAM Exome chip study) and to the CARDIoGRAMplusC4D 1000 Genomes MI case (n=~42 561)–control (n=123 504) study, mainly of people of European descent, adjusted for study-specific covariates and genomic control. We also applied them to the DIAbetes Genetics Replication And Meta-analysis diabetes mellitus case (n=34 840)–control (n=114 981) study overwhelmingly of people of European descent adjusted for age, sex, and study with genomic control, mean age 57.4 years, and to the Global Lipids Genetics Consortium Results (n=196 475), also comprised mainly of people of European descent, mean age 55.2 years, adjusted for age, age,2 sex, and genomic control. We aligned the SNPs on allele as none were palindromic.
As our primary analysis, we obtained unconfounded estimates using fixed effects inverse variance weighted (IVW) meta-analysis of SNP-specific Wald estimates (ratio of SNP on outcome to SNP on exposure), which assumes all SNPs are valid instrumental variables, and a weighted median method, valid if <50% of the weight is contributed by invalid SNPs. As a sensitivity analysis, we used the MR-Egger method, valid if the instrument strength independent of direct effect assumption holds. MR-Egger provides a check on the validity of the IVW estimate, with a nonzero intercept indicating an invalid IVW estimate. We checked the selected SNPs for pleiotropic effects using PhenoScanner (http://www.phenoscanner.medschl.cam.ac.uk/phenoscanner), which can provide phenotypes for a given SNP and its correlates. To balance comprehensiveness and false positives, we searched for phenotypes strongly associated (P value<5×10−8) with the selected SNPs or with highly correlated (r2>0.9) SNPs. All statistical analyses were conducted using R version 3.4.2 (R Foundation for Statistical Computing, Vienna, Austria), and the R package MendelianRandomization. No Institutional Review Board approval was required because the data are publicly available.
Genetically instrumented CT-pro-ET-1 was positively associated with IHD using IVW (odds ratio [OR], 1.05; 95% confidence interval [CI], 1.02–1.09) and weighted median (OR, 1.05; 95% CI, 1.02–1.09) as was MI (IVW: OR, 1.06; 95% CI, 1.01–1.10; weighted median: OR, 1.06; 95% CI, 1.01–1.11; Table). Heterogeneity was low. In the sensitivity analysis, the MR-Egger point estimates were similar, but the wider CIs included the null value (IHD: OR, 1.05; 95% CI, 0.89–1.25; MI: OR, 1.10; 95% CI, 0.91–1.34), possibly because of the similarity in the SNP-specific associations with CT-pro-ET-1, which makes the MR-Egger estimates imprecise. The MR-Egger intercepts did not differ from zero (P values 0.91 [IHD] and 0.66 [MI]), suggesting some validity of the IVW estimates. Genetically predicted CT-pro-ET-1 was not clearly associated with diabetes mellitus, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol or triglycerides, although heterogeneity was high for low-density lipoprotein cholesterol and triglycerides (Table). After excluding rs2545801 for IHD and MI because of its associations with potentially IHD-related attributes identified using PhenoScanner, such as activated partial thromboplastin time and Factor XII antigen, the IVW estimates were similar (Table).
In this first MR study of the effects of a biomarker of ET-1 on IHD, we found higher CT-pro-ET-1 might cause IHD, but was not related to diabetes mellitus or lipids. We took advantage of very large genetic studies, nevertheless, limitations exist. We used only 3 SNPs for CT-pro-ET-1, but they explained a substantial proportion of the variance. Confounding by population stratification is possible, but the genetic associations are from separate studies in ancestrally similar populations2 with genomic control as appropriate. The genetic predictors of CT-pro-ET-1 might affect IHD other than via CT-pro-ET-1, but we obtained similar results after excluding rs2545801 (Table). Whether the metabolites found by PhenoScanner associated with rs2545801 (phenylalanine and alanine to phenylalanine ratio) and rs4253237 (alanine to histidine ratio, histidine, histidine to valine ratio, phenylalanine to valine ratio, B-type natriuretic peptide, and bradykinin) affect IHD other than via CT-pro-ET-1 is unknown, but these metabolites are not clearly known to affect IHD. Adrenomedullin is highly correlated with and stimulated by ET-1.2,3 The role of adrenomedullin in IHD is uncertain, replicated SNPs predicting adrenomedullin are too limited to determine an effect on IHD or MI using MR.2,3 MR-Egger did not give any statistical indications of pleiotropy, but with only 3 SNPs power to detect pleiotropy may be low. We rely on the validity of the conduct and presentation of the underlying studies.2 We could not assess whether associations with IHD varied by age, sex, ethnicity, or baseline levels of CT-pro-ET-1. However, causal factors should act consistently, but might not be relevant in all settings.
Our findings have some consistency with the recently discovered effects of ET-1 expression on IHD,1 and with ET-1 promoting atherosclerosis in mice.4 Notably, ET-1 is also regulated by several potential targets of intervention, including dietary, environmental, and endocrine factors,5 as well as existing therapeutics. Our study provides further evidence that factors regulating ET-1 might be targets of intervention to combat IHD.
Data on myocardial infarction have been contributed by CARDIoGRAMplusC4D investigators and have been downloaded from http://www.CARDIOGRAMPLUSC4D.ORG for the CARDIoGRAMplusC4D 1000 Genomes MI study. Data on coronary artery disease have been contributed by the CARDIoGRAMplusC4D and UK Biobank CardioMetabolic Consortium CHD working group who used the UK Biobank Resource (application number 9922). Data on diabetes mellitus have been contributed by the Diagram Consortium and downloaded from http://diagram-consortium.org/downloads.html. Data on lipids have been contributed by GLGC Results and downloaded from http://csg.sph.umich.edu/abecasis/public/lipids2013/. The program is available from the corresponding author.
Circ Genom Precis Med is available at http://circgenetics.ahajournals.org.
- © 2018 American Heart Association, Inc.
- Beygui F,
- Wild PS,
- Zeller T,
- Germain M,
- Castagné R,
- Lackner KJ,
- et al
- Li MW,
- Mian MO,
- Barhoumi T,
- Rehman A,
- Mann K,
- Paradis P,
- et al