Large scale genome-wide association analyses identify novel genetic loci and mechanisms in hypertrophic cardiomyopathy

Abstract

Hypertrophic cardiomyopathy (HCM) is an important cause of morbidity and mortality with both monogenic and polygenic components. We here report results from the largest HCM genome-wide association study (GWAS) and multi-trait analysis (MTAG) including 5,900 HCM cases, 68,359 controls, and 36,083 UK Biobank (UKB) participants with cardiac magnetic resonance (CMR) imaging. We identified a total of 70 loci (50 novel) associated with HCM, and 62 loci (32 novel) associated with relevant left ventricular (LV) structural or functional traits. Amongst the common variant HCM loci, we identify a novel HCM disease gene, SVIL, which encodes the actin-binding protein supervillin, showing that rare truncating SVIL variants cause HCM. Mendelian randomization analyses support a causal role of increased LV contractility in both obstructive and non-obstructive forms of HCM, suggesting common disease mechanisms and anticipating shared response to therapy. Taken together, the findings significantly increase our understanding of the genetic basis and molecular mechanisms of HCM, with potential implications for disease management.

Figure 1:. Study Flowchart.

Abbreviations: DCM, dilated cardiomyopathy; HCM, hypertrophic cardiomyopathy; LV, left ventricular; maxFDR, upper bound of the estimated false discovery rate computed using MTAG; MRI, magnetic resonance imaging; N_eff, effective sample size (see methods); UKB, UK Biobank.

Figure 2:. GWAS and rare variant association analyses identify SVIL as a novel HCM gene.

A) GWAS in HCM and DCM identify a subthreshold locus near SVIL. GWAS-PW analysis identifies this locus as sharing the same causal variant (model 3) in both DCM and HCM (posterior probability of model 3, PPA3, 0.98). B) Summary statistics of the lead HCM variant (rs6481586) showing effect and non-effect alleles (EA/NEA) and opposite directions of effect (regression coefficient) in HCM and DCM. C) Forest plot showing excess of rare loss of function (LoF) variants in SVIL in HCM vs. controls in the Rare Disease Bioresource (BRRD), Genomics England (GeL) and Oxford laboratory. D) Schematic of the rare LoF SVIL variants in HCM cases (top, total N=1,845) and controls (bottom, total N=37,481) along the linear structure of SVIL. The coordinates reflect the codon numbers, and the coloured bars are the exons. The height of the exons reflects expression in cardiac isoforms and is not to scale. Detailed variant annotation appears in Supplementary Table 6.

Figure 3:. LV traits and HCM genetic correlations and use of MTAG to empower locus discovery.

Pairwise genetic correlation between left ventricular (LV) traits shown in heatmap as absolute values (∣rg_LV∣) ranging from 0 (white) to 1 (red). LV traits are sorted based on ∣rg_LV∣ along the x and y axes using Euclidean distance and complete hierarchical clustering into 3 clusters: LV contractility (blue), volume (green) and mass (dark red). See dendrogram on top. The table in the middle shows the individual LV trait common variant heritability (h²_SNP) and genetic correlation with HCM (rg_HCM), with corresponding standard errors (SE). The trait with the strongest correlation (based on rg_HCM) in each of the 3 clusters was carried forward for multi-trait analysis of GWAS (MTAG) to empower locus discovery in HCM. MTAG resulted in an increase of the effective sample size (N_eff, based on number of cases and controls and increase in mean χ² statistic) from 21,816 to 28,224, with an estimated upper bound of the false discovery rate (maxFDR) of 0.027. Other abbreviations: LVconc, LV concentricity index (LVM/LVEDV); LVEDVi, LV end-diastolic volume indexed for body surface area; LVEF, LV ejection fraction; LVESVi, LV end-systolic volume indexed for body surface area; LVMi, LV mass indexed for body surface area; maxWT, maximal LV wall thickness; meanWT, mean LV wall thickness; strain^circ, global LV circumferential strain; strain^long, global LV longitudinal strain; strain^rad, global LV radial strain. Note: Since strain^circ and strain^long are negative values where increasingly negative values reflect increased contractility, we show -strain^circ and -strain^long to facilitate interpretation rg_HCM sign. Full rg_LV and rg_HCM results are shown in Supplementary Table 9.

Figure 4:. Circular Manhattan plot of HCM summary statistics from MTAG analysis.

Previously published loci are identified in black (N=20), novel loci discovered by single trait all-comer GWAS meta-analysis are identified in blue (N=13) and additional novel loci from MTAG are identified in green (N=35). Two other loci reaching GWAS significance threshold in the single trait HCM GWAS meta-analysis but not reaching significance in MTAG are not shown (mapped to TRDN/HEY2 and CHPF, see Table 1). Results with P<1x10⁻¹⁵ are assigned P=1x10⁻¹⁵. Variants with P<5x10⁻⁸ are shown as black triangles. Locus naming was performed primarily by OpenTargets gene prioritisation considering FUMA and prior gene association with Mendelian HCM. See Supplementary Table 10 for loci details.

Figure 5:. Mendelian randomization (MR) analysis of LV contractility and blood pressure on risk of obstructive (oHCM) and non-obstructive (nHCM) hypertrophic cardiomyopathy (HCM).

Odds ratio (OR) represented are those inferred from the inverse variance weighted (IVW) two-sample MR per standard deviation increase (SD). The error bars represent the 95% confidence interval of the OR. A) MR suggests causal association of LV contractility (exposure) with HCM, oHCM and nHCM (outcomes), where increased contractility increases disease risk. Genetic instruments for LV contractility were selected from the present GWAS of left ventricular ejection fraction (LVEF), and strain in the radial (stain_rad), longitudinal (stain_long) and circumferential (strain_circ) directions in 36,083 participants of the UKB without cardiomyopathy and with available CMR. To facilitate interpretation of effect directions, OR for strain_circ and strain_long reflect those of increased contractility (more negative strain_circ and strain_long values). The outcome HCM GWAS included 5,927 HCM cases vs. 68,359 controls. Of those, 964 cases and 27,163 controls were included in the oHCM GWAS, and 2,491 cases and 27,109 were included in the nHCM GWAS. Note a logarithmic scale in the x-axis. B) MR suggests causal associations of systolic (SBP) and diastolic (DBP) blood pressure with HCM, nHCM and oHCM. Genetic instruments for SBP, DBP and pulse pressure (PP = SBP-DBP) were selected from a published GWAS including up to 801,644 individuals. See Supplementary Table 16 for full MR results.