The Year in Cardiology

Heart Failure: the Year in Cardiology 2019

John G.F. Cleland; Alexander R. Lyon; Theresa McDonagh; John J.V. McMurray


Eur Heart J. 2020;41(12):1232-1248. 

In This Article

The Diversity of Heart Failure Phenotypes

Precision-medicine, which should also be accurate, requires patients to be classified in a way that informs management. For oncology, this has focused on the genetic cause, tumour location, and spread. For heart failure, a multi-system disorder, it is much more complex.[41–47]

Current, therapeutically relevant classifications of heart failure include the severity of congestion (based on symptoms, signs, blood biomarkers, and imaging), CAD, heart rate and rhythm and QRS duration, blood pressure, serum potassium, renal function, indices of iron deficiency, mitral regurgitation, infiltrative myocardial disease (e.g. amyloid), and ventricular phenotype.[41,48] Optimal management of heart failure, with a few rare exceptions, requires only a modest amount of information but this still creates many thousands of patient-subgroups or clusters that might have different therapeutic needs.[45,46] Such subgroups will increase exponentially with the introduction of each new class of treatment. Despite this heterogeneity of substrate and wealth of interventions, precision-medicine is in its infancy in heart failure.

One therapeutically relevant classification of heart failure is by LVEF, a surrogate for left ventricular (LV) dilation. Prior to the 1980s, imaging of cardiac function was available only in expert centres. Clinical trials relied on the chest X-ray rather than the echocardiogram to support a diagnosis of heart failure. The success of trials such as SOLVD, MERIT, and CHARM, which all had a reduced LVEF as an inclusion criterion, led to the adoption of LVEF <40% as the European Society of Cardiology (ESC) Guideline definition for HFrEF.[49] Values ≥40% were termed HFpEF, comprising patients with a mid-range or mildly-reduced (HFmrEF), normal (HFnEF) and, perhaps, supra-normal (HFsnEF) LVEF.[50] Analyses of >350 000 routinely collected echocardiograms suggested that the nadir of risk, whether or not the patient has a diagnosis of heart failure, lies in the range 60–65% both for men and women. Interestingly, an LVEF of >70% was associated with similar risk as an LVEF of 30–40% (Figure 2).[50]

Figure 2.

All-cause mortality according to left ventricular ejection fraction reported on >350 000 routine echocardiograms stratified by age and sex. HFmrEF, heart failure with mildly reduced ejection fraction; HFnEF, heart failure with normal ejection fraction; HFrEF, heart failure with reduced ejection fraction; HFsnEF, heart failure with supra-normal ejection fraction. Reproduced with permission from ref.50

The ESC Guidelines of 2016 introduced the concept of HFmrEF, for two main reasons. Firstly, because of imprecision, an echocardiographic measurement could not reliably distinguish between two measurements of LVEF within 10% of each other. Creating a buffer-zone between HFrEF and HFnEF meant that misclassification was less likely. This innovation meant that a trial of HFpEF could not claim benefit for all patients with an LVEF >40% based solely on an effect in those with an LVEF 40–49%. Secondly, the introduction of HFmrEF challenged the convention that an LVEF <40% was the correct threshold for HFrEF. Some analyses subsequent to the ESC 2016 Guideline suggest that patients with an LVEF <50% may respond to treatment similarly to those with an LVEF <40%.[51] However, this interpretation could reflect confirmation-bias amongst enthusiastic proponents of HFmrEF (Table 1). The evidence is not so consistent when looked at in its entirety, especially if mortality is considered a key outcome. In the future, many trials will probably include both HFrEF and HFmrEF, others will include HFmrEF, HFnEF, and HFsnEF, but NT-proBNP should be used routinely to stratify risk and potentially exclude low-risk patients who have little to gain from yet another 'pill'. Assuming we continue to use LVEF to classify patients, which seems likely since we cannot undo the past, then the major issue is where to set thresholds. For HFrEF, these have ranged from <25% in COPERNICUS, <30% in MADIT-II, and RAFT to <35–40% for the bulk of other trials.[51] For HFpEF, LVEF has generally been set at >40% or >45% with no upper limit. Analyses of recent trials have led some to suggest that, for patients with an elevated NT-proBNP, the upper limit of LVEF for HFmrEF should be increased to 55% or even 60% but this seems premature until consistency is demonstrated across multiple interventions and end-points and measurement precision for LVEF improves.

In a substantial observational study of patients with HFpEF and pulmonary hypertension, progression of right rather than left ventricular dysfunction was observed and was associated with an increased risk of atrial fibrillation (AF) and death.[52] Although right ventricular (RV) dysfunction is a powerful prognostic marker, remarkably few trials focusing on RV dysfunction have been done (SERENADE: