Long-term outcomes of phenoclusters in preclinical heart failure with preserved and mildly reduced ejection fraction

Aims: The identification of subjects at higher risk for incident heart failure (HF) with preserved ejection fraction (EF) suitable for more intensive preventive programmes remains challenging. We applied phenomapping to the DAVID-Berg population, comprising subjects with preclinical HF, aiming to refine HF risk stratification. Methods: The DAVID-Berg study prospectively enrolled 596 asymptomatic outpatients with EF > 40% with hypertension, diabetes mellitus or known cardiovascular disease. In this cohort, we performed an unsupervised cluster analysis on 591 patients, including clinical, laboratory, electrocardiographic and echocardiographic parameters. We tested the association between each cluster and a composite outcome of HF/death. Results: The median age was 70 years, 55.5% were males and the median EF was 61.0%. Phenomapping provided three different clusters. Subjects in Cluster 3 were the oldest and had the highest prevalence of atrial fibrillation, the lowest estimated glomerular filtration rate (eGFR), the highest N-terminal pro-brain natriuretic peptide (NT-proBNP) and the largest left atrium. During a median follow-up of 5.7 years, 13.4% of subjects experienced HF/death events (N = 79). Compared with Clusters 1 and 2, Cluster 3 had the worst prognosis (log-rank test: Cluster 3 vs. 1 P < 0.001; Cluster 3 vs. 2 P = 0.008). Cluster 3 was associated with a risk of HF/death 2.5 times higher than Cluster 1 [adjusted hazard ratio (HR) = 2.46, 95% confidence interval (CI) 1.24–4.90]. Conclusions: Based on phenomapping, older patients with lower kidney function and worse diastolic function might represent a subset of preclinical HF with EF > 40% who deserve more efforts to prevent clinical HF.