Phenotypic clustering of heart failure with preserved ejection fraction reveals different rates of hospitalization

Abstract

Aims Approximately 50% of patients with heart failure have preserved (≥50%) ejection fraction (HFpEF). Improved understanding of the phenotypic heterogeneity of HFpEF might facilitate development of targeted therapies and interventions. Methods This retrospective study characterized a cohort of patients with HFpEF based on similar clinical profiles and evaluated 1-year heart failure related hospitalization. Enrolment, medical and pharmacy data were used to identify patients newly diagnosed with heart failure enrolled in a Medicare Advantage Prescription Drug or commercial healthcare plan. To identify only those patients with HFpEF, we used natural language processing techniques of ejection fraction values abstracted from a linked free-text clinical notes data source. The study population comprised 1515 patients newly identified with HFpEF between 1 January 2011 and 31 December 2015. Results Using unsupervised machine learning, we identified three distinguishable patient clusters representing different phenotypes: cluster-1 patients had the lowest prevalence of heart failure comorbidities and highest mean age; cluster-2 patients had higher prevalence of metabolic syndrome and pulmonary disease, despite younger mean age; and cluster-3 patients had higher prevalence of cardiac arrhythmia and renal disease. Cluster-3 had the highest 1-year heart failure related hospitalization rates. Within-cluster analysis, prior use of diuretics (cluster-1 and cluster-2) and age (cluster-2 and cluster-3) was associated with 1-year heart failure related hospitalization. Combination therapy was associated with decreased 1-year heart failure related hospitalization in cluster-1. Conclusion This study demonstrated that clustering can be used to characterize subgroups of patients with newly identified HFpEF, assess differences in heart failure related hospitalization rates at 1 year and suggest patient subtypes may respond differently to treatments or interventions.

Evaluating the adverse outcome of subtypes of heart failure with preserved ejection fraction defined by machine learning: a systematic review focused on defining high risk phenogroups

“…Of the ten studies three suggested that there were six phenotypes and six studies suggest that there were only three phenotype (Table 2 (Tab. 2) ; References in Table 2: Casebeer et al, 2021[ 12 ]; Cohen et al, 2020[ 18 ]; Gu et al, 2021[ 37 ]; Hedman et al, 2020[ 43 ]; Kao et al, 2015[ 52 ]; Nouraei and Rabkin, 2021[ 79 ]; Schrub et al, 2020[ 107 ]; Segar et al, 2020[ 109 ]; Shah et al, 2015[ 112 ]; Woolley et al, 2021[ 131 ]). One study reported that the 'optimal number of clusters was six', but due to the small size of two clusters ( n = 3 and n = 2), they focused on the (remaining four) patient clusters (Woolley et al, 2021[ 131 ]).…”

“…Phenogroup 1 patients had the lowest prevalence of heart failure comorbidities and highest mean age; Phenogroup 2 patients had higher prevalence of metabolic syndrome and pulmonary disease, despite younger mean age. Phenogroup 3 patients had higher prevalence of cardiac arrhythmia and renal disease (Casebeer et al, 2021[ 12 ]).…”

“…There were 10 studies that applied machine learning, artificial intelligence or computational biology to human data on HFpEF (Table 1 (Tab. 1) ; References in Table 1: Casebeer et al, 2021[ 12 ]; Cohen et al, 2020[ 18 ]; Gu et al, 2021[ 37 ]; Hedman et al, 2020[ 43 ]; Kao et al, 2015[ 52 ]; Nouraei and Rabkin, 2021[ 79 ]; Schrub et al, 2020[ 107 ]; Segar et al, 2020[ 109 ]; Shah et al, 2015[ 112 ]; Woolley et al, 2021[ 131 ]). Nouraei and Rabkin examined 196 patients with HFpEF from an ambulatory clinic population and employed a non-hierarchical cluster analysis that used partitioning around medoids (PAM), on 47 variables (Nouraei and Rabkin, 2021[ 79 ]).…”

“…Casebeer et al (2021[ 12 ]) identified 1515 patients newly diagnosed with heart failure and an ejection fraction of 50 % or greater who were enrolled in a US Medicare Advantage Prescription Drug healthcare. Hierarchical clustering assigned each observation to their own cluster and then iteratively merges clusters based on their distance to other clusters.…”

“…A high-performance clustering procedure was used to perform the cluster analysis. For categorical variables, the Hamming method was used to create a distance matrix on the reduced set of baseline characteristics, and then the Ward method was used to apply hierarchical clustering to the distance matrix data source (Casebeer et al, 2021[ 12 ]).…”