Integrated multiomics of pressure overload in the human heart prioritizes targets relevant to heart failure

Figures

• 

  Fig. 1. Integration of the circulating proteome and tissue sequencing to prioritize targets of cardiac remodeling. Overview of our study design. AS aortic stenosis, CMR cardiac magnetic resonance imaging. PC1 = LV Volumes; PC2 = LV Systolic function; PC3 = LV Diastolic function.

• 

  Fig. 2. Principal component analysis generates 3 phenotypic components of cardiac remodeling. We used principal component analysis (PCA) on 12 echocardiographic measures of cardiac structure and function in the discovery cohort derivation sample (N = 519) and identified 3 composite axes of remodeling that accounted for 65% of the variance. A Loadings for each of the 3 principal components. B Heatmap demonstrating individual level data on the 12 echocardiographic measures and participants’ corresponding principal component value. Source data are provided as a Source Data file. PC principal component, LV left ventricular, RWT relative wall thickness, SVI stroke volume index, AV aortic valve, LVEF left ventricular ejection fraction, LV DTI LV tissue Doppler S lateral annulus, LVESDI left ventricular end-systolic diameter index, LVEDDI left ventricular end-diastolic diameter index, LVESVI left ventricular end-systolic volume index, LVEDVI left ventricular end-diastolic volume index, LVMi left ventricular mass index, LA Vol left atrial volume.

• 

  Fig. 3. Validation of proteomic markers of remodeling in a CMR-based cohort. We examined the relations of the 270 proteins related to any echocardiographic component of remodeling (Supplemental Fig. 2) with cardiac magnetic resonance imaging (CMR) based measures of remodeling and fibrosis. For visualization, proteins related to any CMR measure with an FDR < 0.1 (Benjamini–Hochberg) are presented. Source data are provided as a Source Data file. ECV extracellular volume, LVMI left ventricular mass index, LAVI left atrial volume index, DT deceleration time, EF left ventricular ejection fraction. * FDR < 0.1; ** FDR < 0.05.

• 

  Fig. 4. Proteomic signatures of remodeling are related to clinical outcomes in both an AS population and a general “at-risk” population. A Forest plot of Cox regression results from the discovery cohort relating the proteomic scores to post-TAVI all-cause mortality. Derivation sample N = 500 in unadjusted models. Validation sample N = 302 in unadjusted models. B Forest plot of Cox regression results from the UK Biobank (N = 36,668 in unadjusted models) relating proteomic scores of remodeling to all-cause mortality and incident heart failure. Source data are provided as a Source Data file.

• 

  Fig. 5. Proteo-transcriptional architecture of human aortic stenosis. UMAP of cell types (A) and condition (B) defined by single-nuclear RNA-sequencing (114,288 nuclei across 7 broad clusters). C Cell composition by condition identifies a significant increase in endothelial cells in AS hearts; *** represents significance. D Heatmap representing the differentially expressed genes associated with remodeling across pseudobulk and cell-specific analyses. Numbers adjacent to the cell type label represent the total number of significant genes, inclusive of those which overlap across comparisons. E Select terms from Gene Ontology analysis of differentially expressed genes associated with remodeling, which highlight cell type-specific pro-angiogenic and pro-inflammatory processes. Full tabular results for all analyses are available in Supplementary Data 9. F Quantile-quantile plot showing the association of genetically determined circulating protein expression with HF. A leftward shift from the gray line shows departure from the null (standard uniform) distribution. The red dashed line denotes \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts}…