Heart failure is ejection fraction in motion

‘Everything changes but change itself. Everything flows and nothing remains the same… You cannot step twice into the river, for other waters and yet others go flowing ever on.’ Heraclitus of Ephesus, ∼500 BC. Well into the 21st century, left ventricular ejection fraction (EF) measurement continues to be the mainstay for classification of patients with heart failure (HF). Although EF has many recognized limitations, it is easy to obtain and non-invasive, and thus it remains universally used for HF stratification, surveillance, and management. Clinical decisions are often based on a single EF value, in accordance with the results of pivotal clinical trials and recommendations of international guidelines. However, several studies indicate that bidirectional EF transitions occur during the course of HF, potentially leading to patient misclassification and failure to implement appropriate treatment.1 Following clinical guidelines, patients with HF have historically been classified as reduced EF (HFrEF) or preserved EF (HFpEF),2, 3 with each group considered an exclusive and unalterable compartment. In 2016, the European Society of Cardiology (ESC) guidelines on HF officially adopted an intermediate group, termed HF with mid-range EF (HFmrEF; EF of 40–49%).4 The 2021 universal definition of HF includes HFrEF (EF ≤40%), HF with mildly reduced EF (HFmrEF; EF of 41–49%), HFpEF (EF ≥50%), and HF with improved EF (HFimpEF; baseline EF ≤40%, a ≥10-point increase from baseline EF, and a second EF measurement of >40%).5 This new guideline is the first to introduce the dynamic EF behaviour. In the current issue of the Journal, Strange et al.6 report the largest study on the prognostic importance of dynamic changes (up or down) in EF over long-term follow-up. The authors retrospectively analysed the patterns among ∼34 000 deaths during >500 000 person-years of surveillance of ∼117 000 adults who had multiple EF measurements at intervals of ≥6 months. The main results were that one-third of cases exhibited a decline in EF, and that each 5-unit decline was associated with increasingly higher mortality. In contrast, both stable and increasing EF levels were associated with preserved survival. Among the 17 000 cases referred for HF investigation, there was a clear gradient in overall cardiovascular (CV) mortality across the full spectrum of changes in left ventricular EF. This study builds on prior reports of EF trajectories in patients with HFrEF and HFpEF.7-10 Moreover, it is a unique report with important clinical relevance due to the large number of patients and events, and the analysis of a large spectrum of upward and downward EF changes. The long-term trajectory of EF in HFrEF is the best studied of all groups. Overall, at long-term follow-up, ∼50% of HFrEF patients continued to show an EF of <40% (equal to or below baseline EF), ∼25% exhibited an increase to HFmrEF, and ∼25% jumped to HFpEF (Figure 1A). The EF trajectory in HFrEF forms an inverted U-shape, with lower EF levels at both ends of the distribution.7 The data provided by Strange et al.6 indicate that each 5-unit decline of EF during follow-up is associated with higher CV mortality, while increasing EF values were associated with a survival benefit (Figure 1B). Extensive research has explored whether EF improvement represents myocardial remission or true myocardial recovery or ‘myocardial cure’ in patients with HFrEF, which is beyond the scope of the present editorial comment. Since the term HFmrEF was introduced in the 2016 ESC guidelines,4 a vast number of studies have been performed to better characterize this group of patients (PubMed March 2021: 266 documents). Overall, the evidence suggests that HFmrEF is not actually a true category by itself, but rather comprises HF patients with mildly reduced EF.11 During long-term follow-up, around one-third of these patients remain in HFmrEF, one-third show an EF decline to HFrEF, while one-third show an upward trend to HFpEF7 (Figure 1C). Regarding outcomes in this patient sub-group, Strange et al. found slightly different mortality trends compared to HFrEF.6 As in HFrEF, an EF decline during follow-up was associated with higher CV mortality; however, CV mortality did not decline in patients with an EF increase of up to +10 EF units (Figure 1D). The study by Strange et al.6 was robust in size, as it used the National Echo Database Australia (NEDA), a large observational database that is valuable for a ‘big data’ approach; however, it is unable to provide granularity and pathobiological insight for such mortality disparities. Most data regarding long-term EF longitudinal trajectories and outcomes in patients with HFpEF come from retrospective analyses, in which subsequent EF assessments were clinically driven and are thus susceptible to indication bias.9, 10 The study by Strange et al.6 is also subject to these limitations. A recent prospective study in patients with HFpEF included pre-defined echocardiograms at scheduled intervals, and showed that ∼88% of patients remained with EF of ≥50% during at least a decade of follow-up, while ∼10% of patients evolved to HFmrEF and <2% to HFrEF8 (Figure 1E). However, in the sub-group of patients with HFpEF of ischaemic aetiology, nearly 30% showed evolution to HFmrEF during follow-up. The prognosis of patients with HFpEF relative to EF dynamics is heterogeneous and may be interpreted in at least two sub-groups. The first sub-group includes patients with a baseline EF of 50–59%, who showed CV mortality trends similar to those of HFmrEF6 (Figure 1D). Indeed, in recent sub-analyses of clinical trials that were designed to examine the full HFpEF spectrum, a positive effect was observed for an EF of up to 60%, suggesting that the 50–59% EF range may clinically benefit from neurohormonal blockade, and should be further examined as a separate entity in future clinical trials. A post-hoc analysis of the Treatment of Preserved Cardiac Function Heart Failure With an Aldosterone Antagonist (TOPCAT) trial suggested the potential efficacy of spironolactone in patients with an EF of up to 60%, with declining efficacy associated with increasing EF.12 In a post-hoc analysis of the Candesartan in Heart Failure: Assessment of Reduction in Mortality and Morbidity (CHARM) trial, candesartan significantly reduced the primary outcome (CV death and HF hospitalizations) up to and beyond an EF of ∼50%, and reduced recurrent HF hospitalizations up to an EF of ∼60%.13 More recently, a putative placebo analysis across the full spectrum of EF supported the clinical benefits of sacubitril/valsartan in HF, with the most pronounced effects observed at an EF of up to 60%.14 The second sub-group comprises patients whose baseline EF is ≥60%, who showed high and flat mortality trends regardless of whether EF subsequently increased or decreased. An increase in CV mortality was only shown at extreme downward EF changes6 (Figure 1F). A new phenotype has recently been proposed for patients with HF and EF ≥70% – HF with supra-normal EF (HFsnEF) – which is associated with high mortality rates.15 In patients with EF ≥60%, the introduction of new disease-modifying drugs has been quite unsuccessful, with the notable exception of transthyretin amyloidosis. A more personalized and mechanistic approach may improve this situation. Overall, HF is a heterogeneous syndrome in which EF can change during disease progression, in a patient- and aetiology-specific fashion. Current evidence shows that EF-based HF sub-groups are more overlapping than previously appreciated, and such dynamism has prognostic implications. The ever-changing nature of EF in HF will be progressively incorporated into the design of clinical trials, and in future guideline recommendations. Conflict of interest: none declared.