Abstract Body (Do not enter title and authors here):
Background: Highly precise definition of high-risk features associated with HFpEF may guide targeted treatments and inform biological studies. The aim of this two-step study is to 1) define a high risk HFpEF cluster with unsupervised machine learning approach using cardiac magnetic resonance (CMR), 2) define novel pulmonary vascular mechanics at rest and with exercise in low- vs. high-risk phenotypes. Vascular mechanics defines vessel- and cardiac cycle-specific flow dynamics in pulmonary circulation.
Methods: 48 HFpEF participants underwent CMR and invasive cardiopulmonary exercise testing. With unsupervised K-means clustering analyses using CMR data, two specific clusters were identified with different survival outcomes at 12-months (mortality and heart failure hospitalizations): HR=5.4 (CI:1.7-17.4), log-rank p<0.01. Within this cohort (n=48), additional flow data (right ventricular outflow tract pulse wave doppler) was acquired in 13 participants at rest and 25 watts exercise, to obtain simultaneous pressure (pulmonary artery)-flow data for vascular mechanics analyses in time domain.
Results: Features of low- (n=32) and high-risk (n=16) clusters are summarized in Table. High-risk cluster displayed decreased right ventricular ejection fraction and worse biventricular longitudinal strain. Among wave mechanics, characteristic impedance (proximal pulmonary artery stiffness) and backward decompression wave (pulmonary vascular reflections in diastole) were elevated in high-risk HFpEF phenotype.
Conclusions: We report a high-risk HFpEF phenotype with reduced RV ejection fraction and biventricular strain, indicating myocardial limitations. Vascular mechanics revealed proximal pulmonary artery stiffness and increased wave reflections in diastole as key features in high-risk HFpEF, indicating pulmonary vascular limitations. These multimodal methods reveal that mechanism of RV failure in HFpEF likely involves both pulmonary vascular and intrinsic myocardial abnormalities.