Distinct Exercise Response Patterns in Patients With Heart Failure With Preserved Ejection Fraction

Figures

• 

  Figure 1. Patient selection flow chart. Of all 2306 patients referred for CPETecho, 913 met eligibility criteria. CPETecho indicates combined cardiopulmonary exercise testing and exercise echocardiography; HFpEF, heart failure with preserved ejection fraction; and LVEF, left ventricular ejection fraction.

• 

  Figure 2. Visualization of HFpEF exercise phenotypes after dimensionality reduction in the derivation cohort. A, Grouping of phenotypes projected onto a 2‐dimensional space using UMAP, a technique used to simplify complex data by reducing it to 2 or 3 dimensions. This 2‐dimensional embedding provides an initial visual grouping of the phenotypes, allowing observation of broad patterns between clusters. B, Grouping of phenotypes based on 3‐dimensional UMAP embeddings. The additional dimension helps to reveal clearer separation and more explicit discrimination between the phenotypes, highlighting finer structural differences that may be less apparent in the 2‐dimensional representation. See Video S1. HFpEF indicates heart failure with preserved ejection fraction; and UMAP, uniform manifold approximation and projection.

• 

  Figure 3. HFpEF phenotypes based on exercise limitations. Heatmap of clinical and CPETecho characteristics (rows) according to HFpEF exercise phenotypes (columns). Darker rectangles indicate higher values of the respective characteristic. AVO2 diff indicates peripheral oxygen extraction; BMI, body mass index; CO, cardiac output; E/A, ratio of early (E) to late (A) mitral valve flow velocity; E/e′, transmitral blood flow (E) over early diastolic septal mitral annular velocity (e′); eGFR, estimated glomerular filtration rate; Hb, hemoglobin; highest, highest value obtained from either submaximal or maximal exercise stages; LAVI, left atrial volume index; LV, left ventricular; mPAP/CO slope, mean pulmonary artery pressure over cardiac output slope; NT‐proBNP, N‐terminal prohormone B‐type natriuretic peptide; PAPs, systolic pulmonary artery pressure; RV S′, peak systolic velocity of the lateral tricuspid annulus; SVi, indexed stroke volume; TAPSE, tricuspid annular plane systolic excursion; and VO2, oxygen uptake.

• 

  Figure 4. External validation of exercise phenotypes. A, Grouping of phenotypes in validation cohort based on 2‐dimensional UMAP embeddings. All clusters could be reproduced in the validation cohort. B, Consistency of variable expression in derivation and validation data sets. Each dot represents a variable that was identified as significantly associated with 1 of the 5 HFpEF exercise phenotypes used in the clustering analysis. The x axis shows the log2 fold change of the variable in the derivation cohort, indicating how strongly that variable is elevated or reduced in a specific phenotype compared with others. The y axis shows the log2 fold change of the same variable in the validation cohort. All variables demonstrate consistent elevation or reduction in each cluster in both data sets. Specifically, variables in the upper right quadrant are elevated in a given cluster in both data sets, whereas those in the lower left quadrant are reduced in a given cluster in both data sets. The absence of dots in the upper left and lower right shaded quadrants indicates that no variable shows opposing patterns between data sets. HFpEF indicates heart failure with preserved ejection fraction; and UMAP,…

• 

  Figure 5. Survival free of all‐cause death or unplanned cardiovascular hospitalization, stratified by HFpEF exercise phenotype. A, Kaplan–Meier curves showing differences across HFpEF exercise phenotypes for the combined clinical end point of all‐cause death and unplanned cardiovascular hospitalization in the combined derivation and validation cohort of 913 patients with HFpEF. B, Adjusted Cox survival curves showing differences across HFpEF exercise phenotypes for the combined clinical end point, accounting for age, sex, body mass index, hemoglobin, estimated glomerular filtration rate, N‐terminal prohormone B‐type natriuretic peptide, and HFpEF risk scores. HFpEF indicates heart failure with preserved ejection fraction.