Genome-wide meta-analysis of echocardiographic left ventricular size and function in over 100,000 individuals
S Rand, J S Bundgaard, C Zheng, A Teumer, R S Vasan, M Dorr, EchoGen Consortium, J E Moeller, J J Linde, P Natarajan, J Ghouse, A M Small, H BundgaardAbstract
Background
Genome-wide association studies (GWAS) of cardiac magnetic resonance imaging (cMRI) traits have revealed a complex polygenic architecture underlying left ventricular (LV) structure and function. These traits are powerful endophenotypes for myocardial diseases like hypertrophic and dilated cardiomyopathy (HCM and DCM). Using echocardiography, which is more readily available than cMRI, for GWAS offers a complementary approach to advance genetic discovery in these myocardial disease endophenotypes.
Methods
We performed meta-analyses of GWAS of LV structure and function traits using data from the Copenhagen Hospital Biobank (CHB) and EchoGen consortium. The final meta-analysis will include additional six biobanks, to collectively form the ImaGENE consortium. Meta-analyses were performed for end-diastolic LV internal diameter (LVIDd), LV mass (LVM), posterior wall thickness (PWT), and interventricular septum thickness (IVSDd). Traits exclusive to CHB included LV end-diastolic and end-systolic volume (LVEDV and LVESV), LV end-systolic internal diameter (LVIDs), relative wall thickness (RWT), and ejection fraction (EF) were analyzed separately. We assessed genetic correlation (rg) with cMRI traits, HCM, DCM, and heart failure with reduced (HFrEF) or preserved (HFpEF) EF. Results yet to be generated will include polygenic risk prediction of subtypes of heart failure and cardiomyopathies in UK Biobank.
Results
Sample sizes for the final ImaGENE meta-analysis will include up to 180,000 individuals. For this analysis we included up to 106,684 individuals, and identified 150 genome-wide significant associations in 83 independent loci associated with at least one LV structure or function trait. Of these, 43 were not reported in cMRI studies. Notable novel loci were found near TM2D1, a beta-amyloid peptide-binding protein expressed in cardiac and vascular tissues, and TBX5, encoding a transcription factor crucial for cardiac chamber formation. Several loci were found near cardiomyopathy genes (BAG3, CDKN1A, TTN, HSPB7, LSM3) and others near genes not previously linked to structural heart disease (HLF and DERL3). Echocardiography traits were strongly correlated with their counterpart MRI traits (e.g., LVEDV rg = 0.75; LVESV rg = 0.78; LVEF, rg = 0.99; all P < 5×10-10), cardiomyopathies (e.g. LVEDV and DCM rg = 1.00; LVIDd and HCM rg = 0.69; all P < 1×10-30), and HF subtypes (e.g., LVIDd and HFrEF rg = 0.82; PWT and HFpEF rg = 0.50, all P < 5×10-4). Overall, this raises the possibility that most myocardial disease loci will be present in the data generated by the ImaGENE consortium.
Conclusion
We identified 43 previously unreported genetic loci and replicated 40 additional loci associated with LV structure and function. Our final results will extend beyond the highlights above but currently support the use of echocardiographic traits as scalable, genetically informed endophenotypes of subclinical and clinical heart failure.