The Fallacy of Resting Echocardiographic Parameters of Cardiac Function in Heart Failure with Preserved Ejection Fraction: Add Global Longitudinal Strain to the List

This article refers to ‘Impaired left ventricular global longitudinal strain in patients with heart failure with preserved ejection fraction: insights from the RELAX trial’, by A.D. DeVore et al. published in this issue on pages 893–900. Heart failure (HF) is defined as a clinical syndrome that is due to cardiac dysfunction, and is characterized haemodynamically by an imbalance between cardiac output delivery and demand, in a context of elevated left ventricular (LV) filling pressures.1 Thus, objective evidence of cardiac dysfunction is critical for the diagnosis of HF. For practical reasons, landmark studies in the ‘80s and the ‘90s included patients with severe, symptomatic HF, because these patients had poor prognosis with a high rate of cardiovascular events, which allowed for high statistical power.2 In these studies, objective evidence of HF was achieved by documenting cardiomegaly at plain chest X-ray and poor LV ejection fraction (LVEF), defined as less than 40% or 45% by nuclear scan or conventional echocardiography. However, in the ‘90s, it became apparent that approximately half of patients with HF have preserved LVEF, a condition initially labelled as ‘diastolic’ HF and described in patients with LV hypertrophy or concentric remodelling. Thus, the artificially created dichotomy between HF with reduced ejection fraction (HFrEF) and HF with preserved ejection fraction (HFpEF) was born, and implemented widely into clinical practice. It is now well established that patients with HFpEF have in fact subtle cardiac regional systolic dysfunction3 that can be identified at rest in many but not all individuals, and that becomes more obvious during exercise.1 Amongst new echocardiographic parameters that have been proposed to detect this regional systolic dysfunction, resting LV global longitudinal strain (GLS) has been indicated as one of the most promising indices.4 Small studies suggested that decreased resting GLS correlates with exercise intolerance,5 and that it can identify patients at higher risk of adverse events,6 leading some investigators to argue that resting GLS may become a major parameter for detection and assessment of prognosis of patients with HFpEF.7 In this issue of the journal, DeVore et al. report a secondary analysis of echocardiographic data from 187 patients included in the ‘PhosphodiesteRasE-5 Inhibition to Improve Clinical Status And EXercise Capacity in Diastolic Heart Failure’ (RELAX) trial, with the aim of describing the prevalence and clinical significance of abnormal resting GLS in patients with HFpEF.8 A major strength of the RELAX study is the careful non-invasive definition of HFpEF. Thus, patients were enrolled if they had the following cumulative criteria: LVEF more than 50%, admitted or treated for HF, with left atrial enlargement on echocardiography, with decreased exercise capacity, assessed by peak oxygen uptake (VO2 max), and with either elevated NT-proBNP or increased LV filling pressures, determined invasively. Another strength is that echocardiograms were analysed by an expert core centre. The authors found abnormal resting GLS in approximately two-thirds of the patients with HFpEF. However, contrary to their hypothesis, the authors found no correlation of GLS with symptoms, quality of life, or exercise capacity (VO2 max and 6 min walk test), while GLS had moderate correlation with NT-proBNP levels and weak correlation with markers of fibrosis [collagen III N-terminal propeptide (PIIINP) levels].8 There are some important issues which need to be discussed, such as: (i) whether these results are surprising, and (ii) why these results are different from other previous ‘positive’ studies. In our opinion, these results are expected, and some reasons are presented below. Except in very advanced stages, HF manifests during exercise and not at rest, and both HFrEF and HFpEF are characterized by reduced cardiac functional reserve during exercise. In patients with HFrEF it is not necessary to prove the presence of reduced functional reserve, since they have dilated, poorly contracting hearts. However, in patients with HFpEF, heart dimensions and resting parameters of cardiac function lie frequently within the normal limits. Thus, amongst patients with HFpEF, approximately 40% do not have LV hypertrophy or concentric remodelling, 30% do not have increased left atrial size, and one-third have normal echocardiographic indices of diastolic function, at rest.9 Moreover, only a quarter of subjects with moderate or severe diastolic dysfunction, based on resting echocardiography, develop symptoms of HF.10 Even resting haemodynamics, such as cardiac output and LV filling pressures, are most often normal in patients with HFpEF.11 Current European Society of Cardiology guidelines suggest non-invasive criteria for diagnosing HFpEF, based on resting echocardiographic parameters of diastolic LV dysfunction and increased NT-proBNP levels.12 However, these criteria identify less than 60% of patients with HFpEF when compared with the invasive documentation of elevated cardiac filling pressures during exercise.13 The study of DeVore et al. suggests that resting GLS falls in the same category of echocardiographic indices, being normal in approximately one-third of patients with HFpEF.8 Indeed, previous data from the same RELAX study showed that, in patients with HFpEF, resting LV structural and functional parameters, and arterial functional parameters, accounted for only less than 15% of the variance of VO2 max, whereas 64% of the variance of VO2 max was explained by age, sex, body mass index, haemoglobin level, and chronotropic incompetence.14 Therefore, it is not surprising that, using the same echocardiographic database, DeVore et al. found that resting GLS was not predictive of exercise-defined endpoints, such as symptoms, 6 min walk test, and VO2 max. Apparently, these data contradict previous studies that suggested GLS to be a promising parameter for the diagnosis of patients with HFpEF. However, those ‘positive’ studies were small and reported wide confidence intervals of sensitivity and specificity of GLS in predicting outcomes. For example, in a study of 85 patients, low resting GLS predicted increased pulmonary wedged capillary pressure during exercise, but the reported sensitivity [mean 57%, 95% confidence interval (CI) 18–90%] and specificity (mean 69%, 95% CI 56–80%) clearly underlined the limits of resting GLS as a useful parameter.6 In addition, NT-proBNP levels may be completely normal in one-fifth to one-third of patients with definite HFpEF,13, 15 and even when increased, NT-proBNP levels are much lower in HFpEF than in HFrEF.15 Meanwhile, approximately one-third of patients with HFpEF do not have increased myocardial fibrosis, and the increased myocardial stiffness, which is characteristic of HFpEF, is probably due to factors intrinsic to cardiac myocytes.16 Thus, in the study of DeVore et al., it is also not surprising that GLS had moderate correlation with NT-proBNP levels (r = 0.29, P < 0.001), and weak correlation with markers of fibrosis (PIIINP) (r = 0.19, P < 0.01). In fact, this latter weak, but significant correlation was probably driven by the presence of five patients with very low PIIINP levels and normal GLS (Figure 2C of their article, individual plots at lower left). HFpEF is a highly heterogeneous syndrome with many aetiologies, where various pathophysiological pathways interplay. For example, ventricular–arterial coupling is a major determinant of LV performance in HFpEF. It has been proved that in HFpEF, LV function, arterial function, and ventricular–arterial coupling deteriorate during exercise,17 and patients with increased arterial stiffness and consequent decreased LV systolic and diastolic functional reserve are predisposed to develop acute pulmonary oedema.18 In accordance with these data, DeVore et al. report that patients with the highest tertile of resting GLS had the highest effective arterial elastance (i.e. the lowest arterial stiffness). However, data on resting ventricular stiffness and ventricular–arterial coupling (ventricular–arterial elastance ratio), which might have added some useful insights into the mechanisms of HFpEF, are not provided. It is within this context, of poorly performing resting echocardiographic parameters for detection and assessment of prognosis of patients with HFpEF, that stress echocardiography has been recently suggested as a potentially better modality of evaluating patients with HFpEF. In conclusion, the overall picture suggests that resting echocardiographic parameters, including GLS, perform poorly in identifying patients with HFpEF. DeVore et al. provide further important data to support these findings in their carefully performed study. Future studies should focus on identifying useful exercise parameters, and to define simple, reproducible, and standardized exercise-based echocardiographic protocols, in order to improve the diagnosis and prediction of prognosis of patients with HFpEF. Conflict of interest: none declared.