Pathophysiological insights into machine-learning-based subphenotypes of acute heart failure with preserved ejection fraction: a biomarker analysis of PURSUIT-HFpEF registry

Abstract

Abstract Background Our understanding of heart failure with preserved ejection fraction (HFpEF) has evolved from mere left ventricular diastolic dysfunction to a multiorgan disease. Purpose This study aimed to investigate the heterogeneous pathophysiology of the several phenotypes of HFpEF with multiple biomarker data. Methods This study is a post-hoc analysis of the PURSUIT-HFpEF study (N=1,231), an ongoing, prospective, multi-referral center, observational study of acute decompensated HFpEF [UMIN000021831]. In this registry, there is a pre-defined sub-cohort in which we perform multi-biomarker tests (N=212). We applied the machine-learning-based clustering model that we have previously established to the sub-cohort and classified them into 4 phenotypes: phenotype 1 labeled as "rhythm trouble" (n=69), phenotype 2 "ventricular-arterial uncoupling" (n=49), phenotype 3 "low output and systemic congestion" (n=41), and phenotype 4 "systemic failure" (n=53). Characteristics of biomarker tests in each phenotype were evaluated. Results The phenotype 2 characterized by hypertension and cardiac hypertrophy showed the highest value of N-terminal pro-brain natriuretic peptide (NT-proBNP), IL-16, high-sensitive C-reactive protein (CRP), tumor necrosis factor (TNF)-α, growth differentiation factor (GDF)-15, troponin T, and cystatin C, whereas the phenotype 1 presented the lowest value of these markers. The phenotype 3 showed the second highest value of GDF-15 and cystatin C. The phenotype 4 presented a low NT-proBNP value and a relatively high GDF-15. Conclusions The specific characteristics of biomarkers in each phenotype confirmed the validity of the machine learning model-based classification. The contribution of inflammation to the pathogenesis considerably differed in different HFpEF phenotypes. Systemic inflammation substantially contributes to the pathophysiology of the classical HFpEF phenotype with cardiac hypertrophy.