Epidemiological studies and hospital discharge series suggest that many patients diagnosed with heart failure have a normal left ventricular ejection fraction (LVEF).1–6 Data from the EuroHeart Failure Survey suggest a bell-shaped distribution of LVEF.3 The bell is shifted to the right (higher levels) in women and in older patients (see Figure 1). These studies also suggest that the prognosis of patients with a depressed or preserved LVEF is similarly poor both in terms of re-hospitalisation and mortality. Patients with preserved LVEF are older, more likely to be women, more likely to have atrial fibrillation and hypertension and more likely to have more non-cardiac co-morbidities, and it is possible that the reasons for the poor prognosis in this group of patients are different from those with depressed LVEF. Indeed, it is likely that the diagnosis of heart failure is often wrong in patients with heart failure and preserved LVEF.1
It is likely that there is much greater diagnostic heterogeneity in patients with suspected heart failure compared with those with reduced LVEF (see Table 1). In some cases, the echocardiogram has simply been misread and the patient does, in fact, have a depressed LVEF. In other cases, there will be long-axis systolic dysfunction. Some patients will have constrictive pericarditis or restrictive cardiomyopathy, which are easy to miss. Some patients will have amyloidosis, also easy to miss, which is usually rapidly fatal. Some patients may have impaired left ventricular relaxation with normal systolic function – ‘true’ diastolic heart failure. Many patients will have atrial fibrillation, making assessment of left ventricular function more difficult. Mitral regurgitation may be an important factor leading to raised left atrial pressure and heart failure with good left ventricular function. A large proportion will have hypertension and may develop markedly elevated pressures during exercise, which, combined with reduced aortic compliance and arterial conductance, will lead to marked increases in ventricular afterload. Finally, many patients probably do not have much wrong with their heart and have other reasons for breathlessness and ankle swelling (see Table 1). Guidelines for the diagnosis of diastolic heart failure have received much criticism and are not widely endorsed.7,8 Since they depend on measuring the Doppler filling characteristics of the left ventricle, patients with atrial fibrillation are effectively excluded from such guidance.
For practical purposes, a diagnosis of diastolic heart failure remains largely a matter of excluding reasons other than cardiac dysfunction as the cause of symptoms. This is unsatisfactory. New echocardiographic techniques, such as tissue Doppler assessment of long-axis function, may allow positive identification of at least some subgroups of patients with heart failure and normal LVEF,9 and may help identify patients at a high risk of events, but their utility has not yet been proved in a large-scale outcome study. Measurement error will also be higher in patients with atrial fibrillation. The difficulty and heterogeneity underlying a diagnosis of heart failure and a normal LVEF creates a major problem for clinical trials. Compared with trials of patients with a low LVEF, those investigating patients with a preserved LVEF have substantially lower rates of hospitalisation for heart failure and mortality.10–13 This probably reflects recruitment of younger patients with few co-morbidities, partly because sicker, older patients may not wish to be enrolled and partly because protocols exclude patients whose symptoms might be due to non-cardiac disease. Lower event rates mean that trials have to be larger or last longer and are therefore much more expensive.
However, investigating heart failure with a normal LVEF becomes an even greater problem when existing clinical trial data are examined in more detail. The trials show that not only are overall event rates lower, but that cardiovascular events, both hospitalisation and mortality, make up a much smaller proportion of overall events.10–13 This suggests either that other problems are progressing faster than the cardiac disease or that substantial over-diagnosis of heart failure exists, leading to the recruitment of patients who have breathlessness due predominantly to respiratory disease, obesity or both. Mortality is substantially lower in such patients.14 This compounds the problem of lower overall event rates, since a smaller proportion of patients can be influenced by a cardiovascular intervention. Non-cardiovascular events are unlikely to be influenced by cardiovascular therapy and may even be increased by them (e.g. beta-blockers). This creates ‘noise’, which requires even larger studies to show benefit. A tool that would positively identify patients with symptoms due to heart failure and normal LVEF and, just as importantly, patients at risk of cardiovascular events would be a major asset in the study of this condition.15
The brain natriuretic peptide (BNP) family of peptides, which are secreted from the heart in response to transmural stress, could be about to transform our ability to identify and manage patients with heart failure and preserved LVEF.16 Controversy exists about which cardiac chamber is mostly responsible for production and it is likely that this may differ depending on the pathology. BNP is first synthesised as a pre-pro-peptide, which is digested to form active BNP and inactive fragments. Active BNP is removed from the circulation predominantly by clearance receptors, but also by renal filtration. The inactive fragments are largely cleared by the kidney. Accordingly, both markers, but especially N-terminal prohormonal brain natriuretic peptides (NT-proBNP), are increased in patients with important renal disease. A dietary salt-load will also tend to increase levels. Atrial fibrillation may cause a marked increase in BNP, but this could reflect the adverse prognosis of these patients.17 A series of studies investigating the utility of BNP/NT-proBNP for the diagnosis of heart failure suggests that levels are elevated in patients with diastolic heart failure, but to a lesser extent than in patients with reduced LVEF.18–20 Diagnostic accuracy appears modest at best. However, this may well reflect the inadequacy of the existing diagnostic techniques. A good diagnostic marker should not show good agreement with inadequate markers of heart failure with a normal ejection fraction, such as echocardiography and clinical diagnosis. Ultimately, the best test is a marker that identifies patients who have symptoms due to the disease, higher cardiovascular morbidity and higher mortality, and in whom the response to cardiovascular treatment is greater.
For patients with heart failure and reduced LVEF, natriuretic peptides appear to be the most powerful prognostic markers available and are certainly superior to symptom severity, exercise capacity or echocardiographic measures of cardiac function.16 Of the various assays available, those using NT-proBNP are the best proved, but assays are strongly correlated. It appears the same is true for heart failure and preserved LVEF.
Recent data from the PEP-CHF study suggest that such a tool exists.21,22 This study randomised people aged >70 years with a diagnosis of heart failure and preserved LVEF based on clinical and echocardiographic features to perindopril or placebo. The study showed that, over the first year, perindopril improved symptoms and exercise capacity and reduced hospitalisations for heart failure. Mortality was not reduced. NT-proBNP was measured in a subset of 375 patients.
Patients with increased levels were more symptomatic, had mildly impaired LVEF (56% in the highest versus 65% in the lowest quartile) and more left atrial dilatation, but were less likely to have the Doppler characteristics of diastolic heart failure suggested by guidelines. Importantly, the rate of death or hospitalisation was similar to the general older population in the lowest quartile, but four times higher in the highest quartile. From both a clinical and a clinical trial perspective, patients with features of heart failure who have cardiovascular event rates similar to patients with depressed LVEF can now be identified not by exclusion but by a positive marker. These patients are likely to respond to cardiovascular interventions in a similar way to those with reduced LVEF. Interestingly, the patients in the lowest risk quartile were most likely to satisfy conventional Doppler criteria for diastolic dysfunction. Other data on diastolic heart failure are consistent with these observations, but provide little additional information on prognosis. The SWEDIC study included only patients who fulfilled guideline criteria for diastolic heart failure.23 Plasma BNP levels were low and the prognosis of the patients, regardless of medical treatment, was excellent.
In summary, raised plasma concentrations of NT-proBNP, and probably other peptides of this family, predict an increase in cardiovascular events and mortality in patients with a clinical diagnosis of heart failure but a preserved LVEF. Echocardiography, at least using conventional imaging and Doppler, is of limited use in identifying patients whose symptoms and prognosis are driven by cardiac disease. Therefore, an elevated plasma concentration of BNP/NT-proBNP might be used as a key aspect of the definition of diastolic heart failure in future guidelines.