• Diastolic dysfunction
• echocardiography-exercise

The diagnosis of heart failure is not always easy. It is even more difficult for the group of patients who have symptoms of heart failure but are found to have a normal left ventricular ejection fraction (LVEF), a condition variously called diastolic heart failure, heart failure with preserved ejection fraction or more precisely heart failure with a normal ejection fraction (HFNEF).1 2 Most patients presenting to a hospital emergency department with heart failure will have obvious classic diagnostic symptoms and signs such as peripheral oedema, raised jugular venous pressure and a third heart sound, and echocardiography will confirm a dilated left ventricle and reduced ejection fraction. These patients with ‘systolic heart failure’ or more accurately heart failure with reduced ejection fraction (HFREF) are simpler to recognise and treatment with inhibitors of the rennin-angiotensin-aldosterone system and β-blockers is well established and supported by large clinical trials. For the HFNEF group, however, the situation is completely different. Although presentation with HFNEF can be acute with obvious pulmonary oedema often it is a complaint of breathlessness on exertion that prompts the referral to a clinic. Breathlessness is of course a non-specific symptom and has many causes, particularly in the elderly, and is often confounded by obesity and smoking. It may also be an angina equivalent especially in diabetics. Thus, there has been considerable debate about the reliability of symptoms in this group of patients. Many have suggested that many with the diagnostic label of HFNEF do not have heart failure at all.3 In one study patients being treated for a clinical diagnosis of HFNEF were found to have the same self-reported symptoms and 6-minute walk test (6-MWT) performance as patients with HFREF yet had normal NT-proBNP levels.4 It was concluded that perception of symptoms is disproportionate to the degree of cardiac pathology, and indeed suggests that these patients did not have heart failure at all as indicated by the normal BNP levels. However, it is with exercise that breathlessness occurs and it would seem more pertinent to assess these patients when they exercise rather than at rest. It is on this point that the paper by Holland, Prasad and Marwick in this issue (see page 1024) is so relevant to the problem of diagnosis of HFNEF.5

According to the recent European Society of Cardiology updated guidelines for the diagnosis of HFNEF four obligatory conditions need to be satisfied: (1) presence of signs or symptoms of congestive heart failure; (2) a ‘normal’ LVEF >50%, (3) a non-dilated LV with LVEDV <97 ml/m², and (4) evidence of diastolic LV dysfunction.6 Two measurement issues arise immediately. First, the problem of accurately measuring the LV ejection fraction which is often ‘eye-balled’ and therefore is critically dependent on the eyes doing the balling.7 Another issue is what exactly is a normal LVEF—new criteria now make >55% a normal LVEF.8 Second, providing evidence of diastolic dysfunction is highly contentious. Classically invasive pressure-volume loops have been considered a ‘gold standard’ but this is not the case as pressure-volume loops are highly variable in HFNEF and the end-diastolic pressure-volume relation (EDPVR) is not a measure of stiffness but the extent to which stiffness depends on volume.9 Mitral inflow velocities, in contrast, are easy to obtain by standard Doppler echocardiography and provide a rough assessment of diastolic function and LV filling pressures. Diastolic function can be graded into broad categories, typically four.10 However, it is well documented that these are preload and afterload dependent. Also these measurements are very age-dependent. However, the preload dependence of the peak early mitral inflow velocity (E) can be useful. With high filling pressures E rises and a restrictive filling pattern (high E and reduced peak atrial velocity) is a reliable predictor or a poor outcome especially if it is persistent. Tissue Doppler imaging allows measurement of the mitral annular velocity, which represents ventricular long-axis function, and a decreased mitral excursion and velocity in early diastole (Em) is one of the earliest markers and is present in all stages of diastolic dysfunction.11 12 Since Em velocity remains reduced and mitral E velocity increases with higher filling pressure, the ratio between transmitral E and Em (E/Em or E/e′) correlates with LV filling pressure or pulmonary capillary wedge pressure (PCWP)13 14: an E/Em ≥15 suggests a PCWP >20 and E/Em <10 a normal PCWP, although recently the accuracy of this relation in those with a reduced LVEF has been doubted. Thus the preload dependence of the mitral inflow velocities can be put to good use. An increased left atrial volume is another good marker of a chronically elevated left atrial pressure. In a population based study LA volume correlated well with the severity and duration of diastolic dysfunction.15

One of the advantages of the measurement E/Em is that it can also be measured on exercise. Burgess et al previously demonstrated that E/Em >13 on exercise could accurately identify a raised LVDP >15 mm Hg.16 Holland et al found that exercise testing with echocardiography in a group of patients with a clinical diagnosis of HFNEF produced some interesting results.5 First, only one quarter of patients with suspected HFNEF had objective evidence of exercise intolerance confirming the poor relation between perceived exercise capacity and measures of circulatory, ventilatory or metabolic abnormalities during exercise.17 It is important therefore to substantiate objectively exercise intolerance in patients with suspected HFNEF. Second, from 148 patients with suspected HFNEF by the ESC criteria, only 63 had evidence of raised E/Em either at rest or with exercise. There were an additional 37 patients with non-conclusive E/Em at rest and who did not meet the supplementary ESC criteria, but developed raised LV filling pressures with exercise and were therefore reclassified as HFNEF. However, only two patients (1.5%) with no evidence of diastolic dysfunction at rest developed raised exercise E/Em. Resting and exercise E/Em correlated with age, exercise capacity, heart rate (rest and peak) and baseline blood pressure. Among patients with raised exercise E/Em, resting E/Em correlated with exercise capacity and heart rate (rest and peak). The main conclusion was that only 102 of the original 148 patients with suspected HFNEF had evidence of raised LV filling pressure at rest or on exercise and when patients with evidence of inducible ischaemia were excluded, the number of patients with confirmed HFNEF according to these criteria decreased to 67. Furthermore, including exercise E/Em, ischaemia and exercise capacity in the criteria only 13 patients from the original 436 would be classified as having HFNEF. This is a startling result. It suggests that either clinical diagnosis is widely inaccurate or that E/Em on exercise is not reflecting the true situation. As noted previously doubts are beginning to be raised about how accurately E/Em does truly reflect LVEDP.18 Perhaps this is not surprising as E and Em are both early diastolic events and it is probably only fortuitous that they correlate with a measurement in late diastole, LVEDP. Certainly in HFREF, E/Em appears to perform badly as a marker of LVEDP or PCWP.18 Furthermore, early diastolic filling is very dependent on the previous systole. Any reduction in systolic function, coordination or dyssynchrony will have a profound impact on the ability of the ventricle to fill adequately at normal pressures.1 Our studies using exercise echocardiography have shown that with even mild exercise there are marked abnormalities of LV rotation, strain, untwisting rate with delay and impaired longitudinal function reserve, both systolic and diastolic, in patients with HFNEF associated with breathlessness and this was not associated with any major or significant changes in E/Em.19 In addition, in the study by Holland et al a maximal exercise protocol on a treadmill was used with measurement of E/Em after peak exercise. The maximal rise in E/Em may have been missed therefore. In addition, in those with fused mitral E and A waves—a common finding with maximal exercise and high heart rates—images were not acquired until the heart rate had fallen to allow separation of the E and A waves. A sub-maximal exercise protocol on a supine bicycle couch, which can tilt and is designed for echocardiography, is preferable.20

The study by Holland and colleagues5 does highlight the importance of including exercise testing in any protocol or algorithm for the diagnosis of HFNEF or excluding heart failure as the main cause of the breathlessness on exercise. Other recent studies using exercise echocardiography have shown that even patients with completely normal echocardiograms at rest can have marked abnormalities of both systolic and diastolic function on mild exercise associated with breathlessness.21 Exercise testing is very illuminating. It makes sense to assess cardiac function when the patient's symptoms occur—that is, on exercise. But it is still not entirely clear which are the best and most accurate criteria for the diagnosis of HFNEF.