Abstract Background Studies examining changes in cardiac features and long-term outcomes across heart failure (HF) phenotypes, particularly for normal and supranormal left ventricular ejection fraction (LVEF), are limited. Objectives Our objective was to assess changes in echocardiographic parameters and biomarkers over three years and investigate long-term outcomes across LVEF phenotypes in chronic HF. Methods Patients from the SUPPORT trial (mean age 66 years, 25% female, mean LVEF 54%), a prospective, randomized, open-label blinded endpoint study performed in Japan to determine the additional benefit of olmesartan on top of standard therapy in hypertensive patients with HF, were classified into four HF phenotypes based on baseline LVEF: reduced (HFrEF [LVEF ≤40%], n=200 [17.6%]), mildly reduced (HFmrEF [41% ≤ LVEF <50%], n=229 [20.1%]), normal (HFnEF [51% ≤ LVEF <65%], n=403 [35.4%]), and supranormal (HFsnEF [LVEF ≥65%], n=306 [26.9%]). Changes in echocardiographic parameters and biomarkers were assessed in patients with data available at baseline and three years. The composite outcome of HF hospitalization or all-cause death was analysed from baseline and three years (landmark analysis). Results Over three years, all-cause mortality rates were 15% in HFrEF, 10.0% in HFmrEF, 5.5% in HFnEF, and 4.2% in HFsnEF. The most significant decrease in left ventricular (LV) end-diastolic and -systolic diameter occurred in HFrEF (median percent change; LVDd -5.8% and LVDs -8.8%). Left atrial diameter did not change across HF phenotypes. N-terminal pro-brain natriuretic peptide (NT-proBNP) decreased in HFrEF (-35.7%) but increased in HFnEF and HFsnEF (12.5% and 18.2%, respectively). Troponin increased in all groups except for HFrEF. Growth differentiating factor-15 (GDF15) increased consistently across all HF phenotypes (18.5% in HFrEF, 18.0% in HFmrEF, 16.0% in HFnEF, 22.8% in HFsnEF) (Figure 1). Over a median follow-up of 8.6 years, compared to HFnEF, the risk of the composite outcome was higher in HFrEF (adjusted HR 1.40, 95% CI [1.08-1.82]), but was not different in HFmrEF (1.15, [0.90-1.47]). This trend persisted in the landmark analysis. Up to three years, the risk in HFsnEF compared to HFnEF was not different (1.13, [0.70-1.82]), but it was lower after three years (0.73, [0.55-0.95]). The incident curves show a diverging time point at around three years between HFnEF and HFsnEF (Figure 2). Conclusions In patients with chronic HF and three-year follow-up data, there was minimal decrease in LV dimension, most pronounced in HFrEF. Changes in biomarkers were more diverse, except for consistently elevated GDF15 across all HF phenotypes. The risk of HF hospitalization or all-cause death diverged between HFnEF and HFsnEF after three years.Figure 1Figure 2