Heart Failure with Mid-Range Ejection Fraction: Causes and Consequences

This article refers to ‘Predictors and outcomes of heart failure with mid-range ejection fraction’ by V. Bhambhani et al., published in this issue on pages XXX. Heart failure (HF) with mid-range ejection fraction (HFmrEF) has garnered a lot of attention since being highlighted in the current 2016 European Society of Cardiology (ESC) guidelines.1-3 Studies thus far have focused on prevalent HFmrEF, with the main findings that (i) prevalent HFmrEF constitutes 10–20% of the HF population; (ii) characteristics of patients with HFmrEF are largely intermediate between those of HF with preserved (HFpEF) and reduced ejection fraction (HFrEF); (iii) the notable exception to (ii) is with regard to ischaemic aetiology or ischaemia as a precipitating factor for HF decompensation,4, 5 which is at least as common in HFmrEF as HFrEF; (iv) circulating biomarker profiles are also intermediate between HFpEF and HFrEF;6, 7 and (v) reported outcomes are variable, with some studies reporting similar long-term event rates,8 others reporting intermediate medium-term outcomes,9 and yet others reporting lower in-hospital mortality rates in HFmrEF.4 These published data on prevalent HFmrEF have deepened our understanding but are limited by survival bias and inability to assign causality to risk factors studied cross-sectionally. To understand the true natural history of HFmrEF, we need to study new-onset HFmrEF, i.e. incident cases. Furthermore, data from community-based and multinational studies on incident HFmrEF are particularly scant. In this issue of the Journal, the study by Bhambhani et al.10 addresses an important gap in prior knowledge by studying biochemical and clinical predictors of incident HFmrEF and comparing these to predictors of incident HFrEF and HFpEF. This is the first study ascertaining the true incidence rate of HFmrEF in the community: among a total of 28 820 participants from four large community-based studies, 200 incidence cases of HFmrEF were reported within a median follow-up of 12 years, translating to an incidence rate of 6.7 cases per 10 000 person-years (compared to 26.9 and 34.9 cases per 10 000 person-years in HFpEF and HFrEF, respectively). A particular strength of this study was the prospective adjudication of incident HF in all four cohorts. Additionally, data on echocardiography were available relatively close to date of diagnosis of HF in the majority of cases.11 This careful study of incident cases allowed new insights into antecedent predictors of HFmrEF and prognosis of HFmrEF from the point of diagnosis. Whereas previous studies hinted towards an important role of ischaemia in the aetiology of HFmrEF, data from this study extend the previous cross-sectional data showing that prior myocardial infarction is an important risk factor for developing new HFmrEF. This provides additional evidence for the existence of an “ischaemic HFmrEF” phenotype, which carries therapeutic implications. First of all, evidence has shown that patients with HFmrEF and coronary artery disease have worse outcomes, more similar to patients with HFrEF.5 Results from the Swedish HF registry showed that the presence of ischaemic heart disease modified the effect of beta-blockers on outcomes in patients with HFmrEF.5 When treated with beta-blockers, patients with HFmrEF and ischaemic heart disease derived a similar treatment benefit to patients with HFrEF, but not those without ischaemic heart disease. Furthermore, a sub-analysis of the CHARM programme suggested that patients with HFmrEF (67% of whom had ischaemic aetiology) might benefit from treatment with angiotensin receptor blockade (Lund L.H., unpublished data). Additionally, a post-hoc analysis from the TOPCAT trial suggested that patients in the lower end of the left ventricular ejection fraction (LVEF) spectrum within the trial observed the greatest benefit of spironolactone.12 Patients in this lowest LVEF stratum in TOPCAT had more than twice the prevalence of a history of myocardial infarction, compared with patients in the highest LVEF stratum. Taken together with the new evidence provided by Bhambhani et al., these data suggest that ischaemic HFmrEF is an important phenotype. By extrapolation, patients with ischaemic HFmrEF may benefit from specific therapy targeted at the cause (e.g. coronary revascularization) and may be more likely to respond to guideline-directed therapy for ischaemic HFrEF including beta-blockers and renin–angiotensin–aldosterone system inhibitors. Results also showed that following a first onset of HF, overall mortality rates of patients with HFmrEF were similar to those of patients with HFrEF and slightly worse than those of patients with HFpEF (50 vs. 46 and 39 deaths per 1000 person-years in HFmrEF vs. HFrEF and HFpEF, respectively). Previous reports on long-term outcomes in patients with HFmrEF varied. The ESC HF Long-Term Registry showed that 1-year all-cause mortality in HFmrEF was intermediate (7.6%) between HFrEF (8.8%) and HFpEF (6.4%).9 In the CHARM programme, the risk of cardiovascular death or HF hospitalization was lower in HFmrEF than HFrEF (15.9 vs. 8.5 per 100 patient-years), and more similar to HFpEF (8.9 per 100 patient-years) (Lund L.H., unpublished data). Data from the TIME-CHF trial suggested a similar overall survival for HFrEF, HFmrEF, and HFpEF (62%, 58% and 61%, respectively) over a median follow-up of 794 days.8 These mixed results may have been due to patients being recruited at different stages of disease in prior studies, in contrast to the inception cohort followed by Bhambhani et al.10 As the authors have indicated, however, echocardiographic assessment of LVEF was neither standardized nor classifiable in more than a quarter of patients, resulting in potential differential bias. Furthermore, information on HF-specific treatments and changes in LVEF following the diagnosis of HF was not available. Lastly, in the study by Bhambhani et al.,10 biochemical predictors of incident HFmrEF included natriuretic peptides, cystatin-C and high-sensitivity troponin, suggesting a possible role for ventricular stretch, renal dysfunction and myocardial injury/ischaemia, respectively, in the pathophysiology of HFmrEF. Natriuretic peptides had similar associations for new-onset HFmrEF and HFpEF, whereas the association with HFrEF was greater than for HFmrEF. Adjusting for both clinical variables and natriuretic peptides, high-sensitivity troponin and cystatin-C remained significant predictors of HFmrEF. These data raise the intriguing possibility of using biomarkers for risk prediction on top of clinical risk factors for HFmrEF. For instance, high-sensitivity troponin may identify patients without a known history of myocardial infarction, who may still be at risk of ischaemic HFmrEF. The distinction of ischaemic HFmrEF from non-ischaemic HFmrEF may, in turn, allow more targeted therapy with anti-ischaemic therapies. By extension, HFmrEF patients whose biomarker profiles resemble HFrEF mechanistically may be more likely to respond to HFrEF-specific therapies. Novel tools in biomarker research such as network analyses can potentially provide valuable insights into underlying mechanisms, e.g. protein–protein interactions in HFrEF, HFmrEF and HFpEF.7, 13, 14 A previous study using network analyses based on a panel of 37 biomarkers in acute HF patients with HFrEF, HFmrEF and HFpEF revealed that biomarker profiles in HFrEF were associated with cardiac stretch. In contrast, biomarker profiles in HFpEF were associated with inflammation and endothelial dysfunction.7 Notably, patients with HFmrEF shared some characteristics of the HFrEF biomarker profile and others of the HFpEF profile.7 Indeed, when we revisit data from the COACH trial,14 we observe that age- and sex-matched patients with HFmrEF share biomarker characteristics of both HFrEF and HFpEF (Figure 1). This suggests that biomarkers might be a useful tool in identifying patients with HFmrEF who have mechanistic pathways similar to HFrEF, and might therefore benefit from proven guideline-directed treatment for HFrEF. In conclusion, results from the study by Bhambhani et al.10 provide us with important population-based data on new-onset HFmrEF clarifying its incidence rate in the community (four to five times lower than HFpEF and HFrEF), the importance of antecedent myocardial ischaemia in its pathogenesis, and its dismal outcomes which rival that of HFrEF (and worse than HFpEF). The study also opens the door to more comprehensive approaches using biomarkers that may be employed to further characterize the pathophysiology of HFmrEF as well as identify patients with HFmrEF who might benefit from guideline-directed treatment for HFrEF. Conflict of interest: none declared.