Genome-wide association and multi-trait analyses characterize the common genetic architecture of heart failure

Abstract

Heart failure is a leading cause of cardiovascular morbidity and mortality. However, the contribution of common genetic variation to heart failure risk has not been fully elucidated, particularly in comparison to other common cardiometabolic traits. We report a multi-ancestry genome-wide association study meta-analysis of all-cause heart failure including up to 115,150 cases and 1,550,331 controls of diverse genetic ancestry, identifying 47 risk loci. We also perform multivariate genome-wide association studies that integrate heart failure with related cardiac magnetic resonance imaging endophenotypes, identifying 61 risk loci. Gene-prioritization analyses including colocalization and transcriptome-wide association studies identify known and previously unreported candidate cardiomyopathy genes and cellular processes, which we validate in gene-expression profiling of failing and healthy human hearts. Colocalization, gene expression profiling, and Mendelian randomization provide convergent evidence for the roles of BCKDHA and circulating branch-chain amino acids in heart failure and cardiac structure. Finally, proteome-wide Mendelian randomization identifies 9 circulating proteins associated with heart failure or quantitative imaging traits. These analyses highlight similarities and differences among heart failure and associated cardiovascular imaging endophenotypes, implicate common genetic variation in the pathogenesis of heart failure, and identify circulating proteins that may represent cardiomyopathy treatment targets.

Fig. 1. Study overview.

Overview of major study analyses to identify HF-associated genetic variants, shared risk loci with other traits/diseases, prioritize genes/tissues/cell types, and identify potential treatment targets. GWAS genome-wide association study, HF heart failure, LD linkage disequilibrium.

Fig. 2. Genome-wide associations for heart failure.

Results of the multi-ancestry GWAS meta-analysis of all-cause heart failure, performed using a fixed-effect inverse variance weighted model. A Manhattan plot of genome-wide significant (p < 5 × 10⁻⁸) associations. Each point represents a genetic variant. Variants in red are located +/−500 kb of a genome-wide significant locus. The x-axis represents the genomic position, and the y-axis represents the strength of association as represented by −log₁₀(p value). B Candidate genes were assigned to each genome-wide significant variant (p < 5 × 10⁻⁸) in the multi-ancestry and ancestry-specific analyses (based on proximity to the nearest transcription start site). Candidate genes are grouped by chromosome. Previously unreported candidate genes (>500 kb from a previously reported locus) are denoted by stars. The size of each point corresponds to the strength of association as represented by −log₁₀(p value). Where multiple independent variants mapped to the same gene, only the strongest association is shown.

Fig. 3. Associations of heart failure risk variants with common cardiometabolic traits.

Dotplot indicating associations between lead variants at heart failure risk loci (y-axis), with common cardiometabolic traits (x-axis), with summary estimates obtained from GWAS reported in the IEU OpenGWAS Project. Of the 47 lead risk variant for HF, 46 (or a proxy) were reported in at least 1 cardiometabolic trait GWAS. The size of each point denotes the absolute z-score for each trait, with reference to the heart-failure increasing allele. The shading of each point denotes whether the association met an FDR adjustment for multiple testing. Associations exceeding the conventional genome-wide significance threshold are denoted with a white circle. Variants are grouped by chromosome. FDR false discovery rate.

Fig. 4. Shared associations between heart failure and cardiac MRI traits.

A Cross-trait LD score regression was performed to estimate genetic correlations (r_g) between heart failure and cardiac MRI traits. Significant associations using the Bonferroni method to account for multiple testing are noted with a star. B GWAS associations between lead heart failure risk variants and cardiac MRI traits. The size of each point denotes the absolute z-score for each trait, with reference to the heart failure increasing allele. The shading of each point denotes whether the association met an FDR adjustment for multiple testing. Associations exceeding the conventional genome-wide significance threshold are denoted with a white circle. Variants are grouped by chromosome. LVEDV left-ventricular end-diastolic volume, LVEDVi left-ventricular end-diastolic volume indexed for body surface area, LVSEV left-ventricular end-systolic volume, LVESVi left-ventricular end-systolic volume indexed for body surface area, LVEF left-ventricular ejection fraction, FDR false discovery rate.

Fig. 5. Results of multivariate genome wide association study.

Multivariate GWAS and multi-trait colocalization were performed to identify genetic loci associated with HF and cardiac structure/function traits. A Results of multivariate GWAS. The x-axis denotes the multivariate GWAS method, and the y-axis denotes the independent lead variants at each locus. The size of each point denotes the absolute z-score for each trait. The shading of each point denotes whether the association met an FDR adjustment for multiple testing. Associations exceeding the conventional genome-wide significance threshold are denoted with a white circle. Variants are grouped by chromosome. B Results of multi-trait colocalization. The x-axis denotes heart failure and cardiac imaging traits. The y-axis represents the lead variant at each independent locus identified in the multivariate GWAS. Lines connect groups of traits with evidence of colocalization at a given locus. The size of each point represents the posterior probability for colocalization. Evidence for colocalization was determined based on the default variant specific regional and alignment priors $\left(P_{\mathrm{R}}^{*}=P_{\mathrm{A}}^{*}=0.5\right)$, with colocalization identified when $P_{\mathrm{R}}{P}_{\mathrm{A}}\ge 0.25$. FDR false discovery rate.

Fig. 6. TWAS results.

TWAS identified 36 distinct genes (representing 73 gene–tissue pairs) where expression was associated with adverse HF/structure/function traits, and 28 distinct genes (across 111 splicing–tissue pairs) where splicing was associated with adverse HF/structure/function traits. A Dotplot depicting the gene–tissue pairs where gene expression was significantly associated with HF. B Dotplot depicting the gene–tissue pairs where transcript splicing was significantly associated with HF. In A, B, the bubble size corresponds to absolute z-score, with bubbles colored to the direction of effect, while white dots denote associations that were significant after Bonferroni adjustment for multiple testing (p < 0.05/17703 genes). Only the most significant gene–tissue pair is shown when multiple splicing events in a given gene were identified. C Left ventricular gene expression profiling from MAGNet for genes prioritized by TWAS. Red dots represent candidate genes with significant differential expression among failing vs. healthy hearts, after Bonferroni adjustment for multiple testing.

Fig. 7. Proteome-wide Mendelian randomization.

Proteome-wide MR was performed using high-confidence genetic instruments to detect associations between circulating proteins and cardiac endophenotypes. A Associations between circulating protein levels and HF traits estimated using Mendelian randomization. Protein-trait associations passing an FDR (false discovery rate) q < 0.05 are highlighted. B Number of shared associations across HF traits.