Delivering more evidence for the specificities of heart failure with improved ejection fraction: New answers, new questions

This article refers to ‘In-hospital course of patients with heart failure with improved ejection fraction in the DELIVER trial’ by M.A. Pabon et al., published in this issue on pages 2532–2540. Definitions of heart failure (HF) have been established to pinpoint specific entities. Yet, timely fluctuations of the underlying specific impairments can add further complexity.1, 2 The traditional subcategorization depends on left ventricular ejection fraction (LVEF) defining HF with reduced ejection fraction (HFrEF) when clinical symptoms and signs of HF are associated with an LVEF ≤40%. It is backed by strong evidence for prognosis and treatment benefit and reflected by distinct recommendations.3, 4 HF with preserved ejection fraction (HFpEF) is almost identically defined by the current European Society of Cardiology (ESC) and American Heart Association/American College of Cardiology (AHA/ACC) guidelines: clinical symptoms and signs of HF, an LVEF ≥50% and evidence of increased left ventricular filling pressures/left ventricular diastolic dysfunction (e.g. elevated natriuretic peptide, non-invasive or invasive haemodynamic measurements).3, 4 Results from observational studies and meta-analyses vary from almost identical mortality rates (overall, cardiovascular and non-cardiovascular) for HFrEF and HFpEF patients in the Framingham Heart Study and Cardiovascular Health Study5 to a lower risk of cardiovascular cause in HFpEF versus HFrEF (and a similar risk of non-cardiovascular death) in the Olmsted County Study6, while the MAGGIC meta-analysis of data from both phenotypes identified lower ejection fraction as a highly significant predictor of mortality.7 The focus on both ends of the HF spectrum and the resulting ‘grey area’8 required the introduction of a further HF type: HF with mildly reduced EF (HFmrEF, LVEF 41–49%)3, 4 instead of speaking of ‘HFpEF, borderline’ with patient characteristics, treatment patterns, and outcome considered similar to those of patients with HFpEF.9 Clinical studies, however, positioned HFmrEF patients closer to HFrEF patients regarding aetiology10-12 and amenability to pharmacologic therapy13-16 with an outcome between HFrEF and HFpEF.17 These definitions are further complicated by issues with determination of ejection fraction. Significant variability has been observed in measuring LVEF as a result of difficulties in obtaining ideal measurements in simple unenhanced two-dimensional (2D) echocardiographic, as well as inter- and intra-assessor variability. In some cases, this variability may reach or even surpass 10% in terms of absolute LVEF and therefore can easily reclassify a patient as HFrEF, HFmrEF, or HFpEF. Therefore, determination of HF with improved ejection fraction (HFimpEF) can be within the boundaries of two readings of the same 2D unenhanced echocardiographic assessment by two assessors.18, 19 With HFimpEF, a first dynamic HF type has found its way into the guidelines. In the 2022 AHA/ACC guidelines it is defined as an improvement from LVEF ≤40% to LVEF >40%.3 The 2021 ESC guidelines demand a change from LVEF ≤40% to LVEF ≥50%, whereas patients that improve from LVEF ≤40% to LVEF 40%. Notably, the authors report that there was no significant difference in length of stay or in-hospital mortality between patients with LVEF consistently >40% versus HFimpEF in patients with a first, HF hospitalization in DELIVER. In a subset of participants with at least one LVEF measurement available during HF hospitalization, 66% of those with HFimpEF and 29% of patients with LVEF consistently >40% experienced a reduction in their LVEF to ≤40% from the time of enrolment (p < 0.001). Dapagliflozin reduced total uncomplicated and complicated HF hospitalizations irrespective of whether HFimpEF or ‘true HFpEF’ was present (Figure 1). This is the biggest dataset on HFimpEF outcomes from a randomized controlled trial, although the original trial was not specifically designed for studying this disease entity. However, with the results of the DELIVER trial, the authors warn that too much optimism about an HFimpEF status might be negligent. These findings strongly support the need for further studies to investigate HFimpEF, the mechanisms underlying LVEF trajectories to identify specific patient groups and risk factors. The HFimpEF status could be more fragile than we think, with lower-risk patients quickly transitioning into high-risk groups with worsened ejection fraction.2 Underlying non-compliance with current recommendation for HF medication appears unlikely, as the patients with existing LVEF data were more likely treated with multi-drug guideline-directed medical therapy. In the present analysis, HFimpEF patients exhibited a higher risk of worsening of LVEF as compared to the composite group with a mean baseline LVEF of 55 ± 9%, indicating a high proportion of HFpEF patients in which (per definition) HF severity can be detached from mere LVEF numbers. The fact that in the present study a larger proportion of patients with HFimpEF were found later on to have HFrEF and the outcomes of patients with HFimpEF were closer to those with HFrEF could suggest that in many cases the improved ejection fraction might have been a measurement variation and not a real improvement of ejection fraction. These findings should raise the awareness in the scientific community of the limitations of site based assessment of ejection fraction as an inclusion criteria to a clinical study. In summary, the findings of this paper of Pabon et al. call for a randomized trial of HFimpEF. A confirmation of the observed odds ratio of 2.65 for a decrease in LVEF to less than 40% by ≥10% when comparing HFimpEF patients to a pure HFmrEF cohort, would have strong clinical implications. Such a study should utilize central assessment of left ventricular function to ascertain that HFimpEF patients have a true improvement in ejection fraction that is not related to measurement variability. Conflict of interest: none declared.