Data-Driven Approach to Identify Subgroups of Heart Failure With Reduced Ejection Fraction Patients With Different Prognoses and Aldosterone Antagonist Response Patterns

BACKGROUND: Patients with heart failure with reduced ejection fraction have a poor prognosis. The identification of subgroups with different outcomes and treatment response patterns may help to tailor strategies to each individual patient. We present an exploratory study of patients enrolled in the EMPHASIS-HF trial (Eplerenone in Patients With Systolic Heart Failure and Mild Symptoms) using latent class analysis with validation using the EPHESUS trial (Eplerenone, a Selective Aldosterone Blocker, in Patients With Left Ventricular Dysfunction After Myocardial Infarction) to identify subgroups of patients with different prognosis and response to eplerenone therapy. METHODS AND RESULTS: Latent class analysis identifies mutually exclusive groups of individuals maximizing within-group similarities and between-group differences. In the EMPHASIS-HF trial, 2279 heart failure with reduced ejection fraction patients were randomized to eplerenone or placebo and were characterized according to 18 clinical features. Subgroup definitions were applied to 6472 patients enrolled in the EPHESUS trial to validate observations. Event-free survival and effect of eplerenone on the composite of cardiovascular death and heart failure hospitalization were determined for each subgroup. Four subgroups were identified with significant differences in event-free survival (P=0.002). The subgroup C had the worst event-free survival in both studies and was characterized by older age, lower body mass index, worse renal function, higher baseline potassium levels, high prevalence of anemia, diabetes mellitus, previous revascularization and higher rates of eplerenone discontinuation, and hyperkalemia during follow-up. Two subgroups (B and C) showed a poorer response to eplerenone in both studies and these groups shared common features such as lower body mass index and high prevalence of anemia. Clinical profiles, prognosis, and treatment response patterns of the 4 subgroups applied in EPHESUS trial presented similarities to those observed in EMPHASIS. CONCLUSIONS: Using a data-driven approach, we identified heart failure with reduced ejection fraction subgroups with significantly different prognoses and potentially different responses to eplerenone. However, these data should be regarded as hypothesis-generating and prospective validation is warranted, to assess the potential clinical implications of these subgroups.