Key result
Candesartan reduced the rate of cardiovascular death or heart failure hospitalization across all CHARM patients, with consistent benefits observed in the EF 40-49% subgroup.
Why the study?
Does candesartan reduce cardiovascular death or heart failure hospitalization in patients with heart failure and mid-range ejection fraction?
Does candesartan reduce cardiovascular death or heart failure hospitalization in patients with heart failure and mid-range ejection fraction?
Answers are grounded in this paper and cite their sources.
This editorial argues that while candesartan may benefit patients with HFmrEF, categorizing heart failure patients solely by ejection fraction is flawed and future trials should target specific underlying pathobiologies.
This article refers to ‘Heart failure with mid-range ejection fraction in CHARM: characteristics, outcomes and effect of candesartan across the entire ejection fraction spectrum’ by L.H. Lund et al., published in this issue on pages 1230–1239. Since the onset of randomized controlled trials of treatments for chronic heart failure (HF), in the 1980s, patients have been categorized and selected for enrolment based on left ventricular (LV) ejection fraction (EF). Through the 1990s the focus was almost entirely on HF patients with reduced EF (HFrEF). Subsequently, attention was given to HF patients with normal or preserved EF (HFpEF).1-3 Patients with EF in the range of 40–49% (now being referred to as HF with mid-range EF, or HFmrEF)4 were generally either excluded from trials or lumped in with patients with higher EF. As such, there appeared to be a knowledge gap for treating patients with EFs in this range. The analysis by Lund et al.,5 published in this issue of the Journal, contributes to filling this gap by observing a favourable outcome response to the angiotensin receptor blocker, candesartan, among patients with EF in this mid range. These findings are at least hypothesis-generating for the appropriateness of such treatment within this patient group. But this analysis begs the questions: specifically, does HFmrEF define a sufficiently homogeneous population (a disease?) to justify uniform treatment based on this finding alone, and, more broadly, how useful is EF in stratifying treatment in patients with HF? In this retrospective analysis, Lund et al.5 observed that the 1322 patients within the Candesartan in Heart failure Assessment of Reduction in Mortality and Morbidity (CHARM)-Overall programme6 with EF 40–49% had some baseline characteristics similar to those of patients with lower EF and others that were intermediate between patients with lower EF and those with higher EF. The incidence rate for the primary outcome of this analysis—cardiovascular death or HF hospitalization—in these patients was similar to that of patients with higher EF. But patients with EF in the mid range showed reduction in the rate of primary outcome events within the candesartan group, compared with the control group, with a similar estimated hazard reduction to that observed in patients with reduced EF. In this analysis, patients with EF ≥50% showed no treatment benefit with candesartan. Results of randomized controlled trials cannot predict the impact of a therapy on an individual patient. Rather, such results are used to anticipate the aggregate efficacy of a treatment within a new population with characteristics similar to the investigational population. Such a conclusion requires a degree of certainty beyond that attainable from the present analysis, which is best viewed as a post-hoc subgroup analysis of the CHARM-Overall programme. Since such analyses are confounded by their multiplicity, their principal purpose is generally to explore for consistency of the primary finding across multiple groups by testing for treatment-by-subgroup interaction. Across all CHARM patients, candesartan reduced the rate of cardiovascular death or HF hospitalization, with a hazard ratio of 0.84 (95% confidence interval 0.77–0.91; P<0.0001).6 This finding was consistent across all prespecified subgroups, including EF ≤40% vs. EF >40%, with no statistically significant treatment-by-subgroup interaction. Likewise, in the present analysis, which examined patients across three ranges of EF, there was no statistical interaction between EF group and treatment effect (P = 0.27). For these reasons the authors appropriately conclude that the present findings must be considered hypothesis-generating, rather than definitive. Nevertheless, they provide the best evidence yet that the benefit of angiotensin receptor blocker treatment may extend to patients with HF and EF ≥40%. The larger question is whether EF is the best means of grouping patients for the purpose of treatment strategy. The 2016 European Society of Cardiology HF guidelines4 reinforce this approach and identifies patients with EF of 40–49%, or HFmrEF, as a ‘separate group’ and as a ‘diagnosis’, implying that this EF range may delineate a population with relatively homogeneous therapeutic responsiveness. Such a conclusion is not warranted. Although the historical restriction of enrolment in HF trials to patients with reduced EF has forced us into clinical recommendations that are tied to a specific EF range, EF is unlikely to represent the best tool for sorting patients into groups of common therapeutic response. Going forward in performing clinical trials, we should move away from relying on designations such as HFpEF or HFmrEF to identify patient groups that meet this expectation. The initial rationale for focusing on EF was the assumption that reduced EF identifies a patient with a primary abnormality of LV contractility, thought to be the principal basis for HF. As a result, the earliest investigations explored inotropes as therapeutic agents.7, 8 Over the next two decades we learned that it was not inotropes that improved clinical outcomes in these patients, but rather neurohormonal antagonists. These agents concomitantly improved EF, not by directly augmenting inotropy, but rather by reversing adverse LV remodelling with dilatation.9-12 In fact, EF is much less a marker of contractility than it is of dilatational remodelling.13 As the left ventricle dilatates, while maintaining stroke volume within a given range, EF, the ratio of stroke volume to LV end-diastolic volume, secondarily falls. Even as a marker of remodelling, EF is secondary to direct measures of LV volume. It may have been fortuitous that the two principal diagnostic groups of patients with low EF—ischaemic cardiomyopathy and non-ischaemic dilated cardiomyopathy—have similarly favourable outcome benefits from multiple agents explored, although the exact mechanism of benefit may differ in these two diagnostic groups. Non-ischaemic patients may receive proportionally more benefit through reverse remodelling, whereas patients with an ischaemic aetiology may have relatively greater benefit through the coronary protective and anti-ischaemic properties of at least some of these agents. In contrast, outcome trials in patients with HFpEF have generally yielded neutral results. One possible explanation is that the multiple diagnostic and/or pathophysiologic categories of patients represented among trial participants with HFpEF manifest varied responses to the therapies investigated. For example, it seems likely that these trials included patients with restrictive myopathy and primary pulmonary vascular disease, as well as those with the syndrome of metabolic-senescent cardiovascular disease (MetS-CVD).13 Smaller, single-centre trials are more likely to have enrolled a more homogeneous population with the latter syndrome. The Reduction of Endpoints in the NIDDM with the Angiotensin II Antagonist Losartan (RENAAL) trial, which showed improvement in its primary endpoint of worsening kidney function in patients with type 2 diabetes and nephropathy, also showed a 32% reduction in the relative risk of first hospitalization for HF.14 RENAAL excluded patients with a history of HF, but this population had numerous characteristics of MetS-CVD, including a mean systolic blood pressure of 153 mmHg and 93% baseline prescription of anti-hypertensive agents. These findings strongly suggest that alternate sets of patient characteristics may define a particular disease and thereby markedly exceed the ability of EF measurement to establish a population with a fairly uniform therapeutic response. There is presently great interest in sub-categorizing patients into specific phenotypes, based on in-depth analysis of multiple clinical and laboratory markers.15 It would be ideal if the categories generated would actually represent discrete diseases (rather than mere phenotypes), associated with relatively uniform response to therapeutics. RENAAL findings14 point toward a potential first step: investigating patients with MetS-CVD, identified by some combination of specific characteristics—e.g. type 2 diabetes, hypertension, obesity, advanced age, and imaging characteristics, such as left atrial enlargement and LV hypertrophy. In designing clinical trials, we would do far better to seek patients with a common disease, or common pathobiology, rather than a common EF range. It seems likely that any particular EF range, ‘reduced EF’, ‘preserved EF, or now ‘mid-range EF’, contains patients with a spectrum of diseases, which will not necessarily respond to a common therapeutic intervention. Conversely, as schematically postulated in Figure 1, it seems likely that the distribution of EFs within a population with a specific disease will span beyond any given EF range. The characteristics that Lund et al.5 observed, seemingly intermediate between groups with lower and higher EFs, do not represent those of a distinct population worthy of aggregating within future clinical trials. Rather, they likely portray a shifted distribution of various underlying diagnoses, in proportions falling between those represented by HFpEF and HFrEF. Although these ‘mid-range’ patients appear, as a group, to favourably respond to candesartan, that response is likely to be more heterogeneous than we would see if we could segregate the individual diseases represented. Lund et al.5 have taken us a step forward in characterizing overall therapeutic response within patient populations defined by EF range. We now should go deeper and further, beginning to investigate populations with common diseases across the EF range. In so doing, we are likely to provide therapeutic direction in a far more precise manner than we have done to date, using EF as our primary stratification tool. Conflict of interest: M.A.K. is consultant and/or data monitoring committee member for Amgen, BMS, Ironwood, Livanova, Boehringer-Ingelheim, Array, Novartis, Takeda, Luitpold, Alnylam, SC Pharma.
No takes yet. Share an insight, caveat, or question.
Marvin A. Konstam (2018) conducted an editorial in Heart failure with mid-range ejection fraction (HFmrEF) (n=1,322). Candesartan vs. Control was evaluated on Cardiovascular death or HF hospitalization (HR 0.84, 95% CI 0.77-0.91, p=<0.0001). Candesartan reduced the rate of cardiovascular death or heart failure hospitalization across all CHARM patients, with consistent benefits observed in the EF 40-49% subgroup.
Synapse has enriched 10 related papers. Consider them for comparative context: