M13 - Genome-wide Association Studies of Treatment Response in Randomized Clinical Trials (HT GWAS)
Investigator Names and Contact Information
Alex Reiner, MD and Charles Kooperberg, PhD
Introduction/Intent
(From the Study Proposal) Much of the inter-individual variation in response to drug therapy has been hypothesized to have a genetic basis. Yet, the specific genetic factors that contribute to vascular risk in response to HT are unknown. Because it requires fewer a priori assumptions, a genome-wide association study (GWAS) approach may increase the likelihood of identifying clinically important variants and reveal novel mechanisms (hypothesis-generation). With the breadth and depth of genomic linkage disequilibrium coverage, state-of-the-art whole-genome SNP platforms increasingly also allow for more focused testing of candidate genes based on prior biologic hypotheses (hypothesis-testing). In this regard, oral estrogen and progestin have multi-factorial effects on atherosclerosis, inflammation, thrombosis, lipids, and glucose metabolism. Therefore, a comprehensive evaluation of SNPs using current generation GWAS technology is the next logical step to identifying and confirming the genetic variants and mechanisms that influence cardiovascular risk in response to HT.
For our primary GWAS, we propose to analyze a case-cohort sample including 744 CHD, 600 strokes, 544 VTE, and 1,677 incident diabetes cases and 3,565 matched controls from the WHI-HT. To help separate chance findings from associations that are likely to be biologically important, we propose to perform replication genotyping for our top hits in an independent sample of cases and controls selected from the WHI-Observational Study (OS) cohort. Without any additional genotyping cost, we will be able to link the GWAS data from the WHI-HT and –OS cases and controls to relevant intermediate outcomes such as blood pressure, glucose, lipids, coronary artery calcium, hormone concentrations, and biomarkers of vascular inflammation and thrombosis, thereby providing important information relevant to the biological basis of replicated associations.
We hypothesize that (a) it is possible to identify common variants that alter the effects of HT (E- alone or E+P) on the risk of cardiovascular disease (CVD) events (CHD, stroke, and VTE) and incident diabetes in postmenopausal women; (b) there are some common genetic determinants of stroke and VTE and that these genetic determinants differ from those related to CHD; and (c) the overall clinical vascular risks and benefits depend on how genes interact to influence the balance of pro- vs. anti-atherosclerotic and thrombotic effects of HT. Therefore, we propose the following specific aims:
1) To identify common genetic variants that reproducibly alters risk of CHD, stroke, VTE, and incident diabetes after exposure to HT (E-alone or E+P) in postmenopausal women from WHI-HT. Particularly, we will prioritize the GWAS data on the basis of prior knowledge and findings from existing GWAS of CVD traits (hypothesis-testing), with the remaining SNPs assigned a lower-priority (hypothesis-generation).
a) To maximize statistical power to detect HT-gene interaction effects on vascular events, we will perform a secondary analysis using a composite clinical endpoint of stroke + VTE under the assumption that stroke and VTE share common pathogenic determinants.
2) To replicate significant gene-HT interactions on CHD, stroke, VTE, or diabetes risk in independent study populations, including a nested case-control sample from WHI-OS. Moreover, the GWAS conducted in the WHI-HT will serve as a rich resource for future replication of other GWAS of CVD and metabolic disorders.
3) To begin to explore potential causal pathways underlying the risk of HT associated with CVD events, we will integrate replicated associations as “causal instruments” with relevant quantitative intermediate phenotypic outcomes available in WHI, including blood pressure, coronary artery calcium, and blood concentrations of lipids, insulin, glucose, hormones, and inflammation and coagulation biomarkers.
4) In conjunction with the Program Coordinating Center and NHGRI, to develop and disseminate innovative analysis methods for adding genome-wide technologies to randomized clinical trials and interpreting the results in the context of a randomized treatment assignment.
Information generated from this study will be critical to determine the impact of genetic variants on women’s health and to prioritize them for intervention studies aimed to maximize overall clinical benefit of HT and minimize their associated cardiovascular risk. Findings may also provide valuable insights into disease pathways and mechanisms, and identify novel targets for screening, prevention, and treatment of vascular disorders (CHD, stroke, VTE) and diabetes in post-menopausal women.
Study population (synopsis) | Analytes |
GARNET (HT GWAS**) N~ 5,062 HT (dbGaP-eligible) a. European American (EA): N~4,416 b. AA: N~176 c. H: N~56 d. Other race: N~214 e. QC samples: 100 AA + 100 H from SHARe Outcome Distribution a. CHD (MI or coronary death): N~517, CHD controls: N~517 b. Stroke: N~349; Stroke controls: N~349 c. VTE: N~313; VTE controls: N~313 d. Self-report treated diabetes: N~1,078 e. Multiple outcomes: N~174; Multiple outcomes controls: N~174 f. Outcome-free Controls: ~2,431 g. SHARe ppts (for QC): N=200 h. Diabetes: N=1080; diabetes controls, N=1078 Exome Chip: ~ 3,074 of the GWAS participants were genotyped using residual DNA. Replication: ~9,034 EA OS cases; 4,000 OS controls | GWAS: Illumina Omni Quad 1.0M Exome Chip: [Illumina Infinium Exome Array] Replication: Fluidigm Array |
M13-GARNET was included in the 2013 GWAS Imputation project (imputed to 1000G data). See the table in Key Genetics and Biomarkers Studies for information about the GARNET GWAS included in the imputition data.
Replication (Fluidigm Array): For details, see GARNET Phase II Case Control Summary in Study Documentation
The final sample consist of 9034 cases and 4000 controls. In this final sample, there are 3182 (1477+1705) CHD case/control pairs, 1602 (716+886) VTE case/control pairs, and 3797 (2041+1756) Diabetes case/control pairs. Only 932 cases (208 CHD, 724 Diabetes) did not have a matched controls.
Exome Chip:
In Summer 2012, residual DNA from ~3,074 GARNET GWAS participants were genotyped on the Illumina Human OmniExpress + Exome Array.
Results/Findings
Phase I: Illumina Omni 1M Array
Phase II: Custom Fluidigm Array
Publications: