Heart Failure and Hypertension: The Diastolic Dilemma

Heart failure is now the leading cause of hospitalization for people aged 65 and over in the United States; unlike other complications of hypertension, its prevalence has been increasing over the past 30 years.1 For a 40-year-old man or woman, the “remaining lifetime risk” of developing heart failure is about 20%, a surprisingly high number. If subjects with known coronary heart disease are excluded, the risk is 11% in men and 15% in women.2 Blood pressure (BP) is a major contributor to this condition; the risk is twice as high in hypertensive men as in normotensives and three times as high in hypertensive women. Ninety percent of new cases of heart failure in the Framingham Heart Study3 had a history of previous hypertension. This risk is more strongly related to systolic than diastolic pressure.4 In the last 10 years or so, it has become apparent that approximately half of the patients who present with classic signs and symptoms of heart failure appear to have normal ventricular function, typically defined by the finding of an ejection fraction of >50% on echocardiography. This group has been variously described as having diastolic dysfunction, diastolic heart failure, or “heart failure with a normal ejection fraction.” This condition, while undoubtedly of major importance, remains a murky area for several reasons. First, assessment of diastolic function is difficult and ideally requires cardiac catheterization; while noninvasive measures are used, they are nonspecific. Second, there are no reliable animal models of diastolic heart failure. Third, there are no treatments that are specifically aimed at improving diastolic function. In an 11-page document outlining practice guidelines for the evaluation and management of heart failure, published recently by the American Heart Association and American College of Cardiology (AHA/ACC),5 diastolic failure received less than a page. A classic example of diastolic heart failure (HF) is a little old lady with systolic hypertension who presents with increasing breathlessness and, on examination, is found to have a high systolic pressure, tachycardia, and signs of congestive failure such as rales in the lungs, and an elevated jugular venous pressure. The chest x-ray shows signs of congestion, and an echocardiogram shows a normal ejection fraction (EF) and mild left ventricular hypertrophy (LVH). Diastolic HF is classically thought of in terms of the pressure-volume loop of the left ventricle, but the key component is that the end-diastolic pressure is high, while the systolic part of the loop is normal. In days gone by, HF was classified as “forward” HF, where the basic problem was thought to be the pumping ability of the heart, for which digitalis was recommended, or “backward” failure, where the problem was fluid retention, for which diuretics are prescribed.6 This distinction is analogous to the more modern concept of systolic and diastolic failure. In practice, of course, patients with systolic failure also develop diastolic failure, as manifested by fluid retention, and have raised end-diastolic pressures, but there are patients who have diastolic failure with normal systolic function. When there is a large number of ways of measuring something, none of which can claim superiority, health care professionals can be sure that, as with weight loss programs, none of them work very well. The ideal measurement of diastolic dysfunction requires cardiac catheterization and high-fidelity micromanometer catheters, which is, of course, impractical for routine clinical evaluation. Regular cardiac catheterization would show an increased end-diastolic pressure and normal systolic function. Several different noninvasive measures of diastolic dysfunction have been proposed.7 The indices that are supposed to indicate diastolic dysfunction are also not straightforward. Thus the traditional method has been the E:A ratio of flow across the mitral valve measured by the Doppler technique. There are two main components: the E wave, corresponding to the early ventricular filling when the valve opens, and a later A wave, corresponding to filling from atrial contraction. Normally, the E wave is larger, but when relaxation is impaired, the down slope of the E wave is prolonged and the A wave is larger. Two other patterns, however, are also indicative of diastolic dysfunction. One is known as a “pseudonormal” pattern and the other as a “restrictive.” The details of these need not concern the readers here, but suffice it to say that when multiple patterns have to be distinguished, particularly when one of them looks normal but really is not, the reliability of the classification is likely to be low. More recently, two other echo techniques have been introduced: measurements of pulmonary vein velocity and Doppler tissue imaging, but they also show multiple patterns that differ in subtle ways.7 Furthermore, in many cases, the patterns vary according to extrinsic factors such as heart rate and BP. Diastolic HF is thought to be responsible for as many as 74% of cases of HF in hypertensive patients.8 Despite this likelihood, it is rarely diagnosed in clinical practice except by default. This is a problem because the symptoms, such as dyspnea and fatigue, are very nonspecific. For most conditions, it is preferable to have a test that shows a specific abnormality rather than an inappropriately normal result. The abnormality we are looking for here is some evidence of diastolic dysfunction, but, as noted, the problem is that there is no consensus as to how this condition can be measured noninvasively, and conventional techniques such as echocardiography do not routinely report any evaluation of diastolic function. Based on their wide experience in the Framingham Heart Study, Vasan and Levy9 have proposed three diagnostic categories: definite, probable, and possible. The diagnosis of definite diastolic failure requires the strong clinical evidence of HF; a normal EF; and objective evidence of diastolic function, defined as “abnormal LV relaxation/filling/distensibility indices on cardiac catheterization.” Since many patients with HF are either not catheterized or measurements of diastolic dysfunction (other than end-diastolic pressure) are not made, the diagnosis is likely to be probable or possible. Two population-based studies have compared the prognosis of systolic and diastolic failure. One of the studies evaluated all 216 patients who were diagnosed with HF in Olmsted County, Minnesota in 1991.10 Of the patients who had their systolic function evaluated, the EF was normal in 43%. The prognosis was similar in the patients with diastolic and systolic failure. The second study is the Framingham Heart Study,11 where it was found that half of 73 HF patients had normal systolic function—65% were women, whereas 75% of patients with impaired systolic function and HF were men. Approximately 75% of both groups had a history of hypertension. Mortality was high in both groups but higher in those with systolic HF.11 Diastolic dysfunction may be a function of age independent of BP and LV mass.12 Elderly patients with borderline systolic hypertension (systolic pressures between 140-159 mm Hg) have been found to have signs of diastolic dysfunction on echocardiography without any impairment of systolic function.13 While diastolic failure typically presents in the elderly, abnormalities of diastolic function have been described in young, healthy subjects with a family history of hypertension. An example comes from a study conducted by Aeschbacher et al.14 which followed a cohort of 25-year-old males for 5 years, half of whom had a family history and half who did not. Diastolic dysfunction was assessed by echocardiography and included measures such as the E:A wave ratio, pulmonary vein flow, and myocardial isovolumic relaxation time, as described above. Over this period, those with a family history developed diastolic dysfunction and increased BP but not LVH. Cross-sectional studies have also shown that diastolic dysfunction may occur in the absence of systolic function changes or increased LV mass in young subjects with a positive family history of hypertension.15 What the relationship is between these findings of diastolic dysfunction and diastolic HF in later life is not yet clear, but in one study of elderly patients the E:A ratio did not discriminate between elderly patients with and without diastolic HF.16 There are two major components of diastolic function: active relaxation, which occurs early in the diastolic cycle, and passive stiffness, which is dominant in the later part. Patients with diastolic HF show both impaired relaxation and increased passive stiffness. The prototypic example of diastolic HF is amyloid, where the heart muscle is directly affected.17 In these patients, the systolic pressure is normal and the ventricles cannot maintain an elevated systolic pressure. Like most diseases, the changes that lead to diastolic HF are almost certainly multifactorial. A number of pathological processes affecting the myocardium have been identified. An obvious candidate is LVH. It is not hard to imagine that a hypertrophied heart would have difficulty expanding during diastole, which has been demonstrated to occur, but while there are correlations between the degree of hypertrophy and diastolic dysfunction in older patients,18 diastolic dysfunction can often occur in the absence of LVH, particularly in younger subjects (as noted above). A less well-appreciated change is fibrosis, which accompanies LVH, but animal studies have indicated that it may be myocardial fibrosis rather than hypertrophy that accounts for impaired diastolic relaxation.18 Myocardial ischemia is another factor and again has been found to affect diastolic function when it is present; however, many patients who present with diastolic HF have no significant coronary heart disease. Most discussions of the underlying causes of diastolic HF have only considered changes occurring in the heart itself. This focus is perhaps not so surprising because the topic has largely been the purview of cardiologists, whose interest falls off markedly beyond the aortic valve. Nevertheless, the obvious primary change in hypertension is what is happening in the arterial circulation. It is generally accepted wisdom that the reason that hypertensive patients develop ventricular hypertrophy and HF is because the increased arterial pressure places an excessive burden on the heart, or afterload. Why should the same explanation not apply to diastolic HF? It is usually thought that increased afterload affects systolic function, but there is evidence from animal studies that diastolic function can also be affected.19 If the aorta is constricted enough to raise the systolic pressure by a modest amount in a healthy dog, then the fall of LV pressure during diastole is initially accelerated, but if the constriction is made tighter and the load on the heart further increased, then LV relaxation in diastole is slowed. Patients with diastolic HF have been found to have stiffer aortas than normal,20 which may result in accelerated wave reflection from the periphery and a combined increase of central aortic systolic pressure and reduced diastolic pressure. The latter will lead to decreased coronary artery perfusion during diastole and hence impair myocardial relaxation. These and other findings have led some investigators to suggest that diastolic HF “has nothing to do with diastolic dysfunction at all.”16 In the same way, preload or venous return may also influence diastolic function. In dogs with experimentally induced cardiomyopathy, reducing venous return by constricting the vena cava can acutely improve diastolic function.19 In an interesting study of patients about to undergo coronary artery bypass surgery, extensive hemodynamic monitoring was carried out while they were anesthetized, looking at the effects of increasing venous return by raising their legs.21 A wide spectrum of responses was observed. In some patients, there was an increase of stroke volume and very little change of LV end-diastolic pressure, but in others stroke volume decreased and end-diastolic pressure increased markedly. In these patients, indices of diastolic dysfunction also worsened. The different patterns may be attributed to where people are on the Starling curve, which is an inverted U-shape. As the filling pressure of the heart is gradually increased, stroke volume increases at first, then levels off, and then starts to decrease as the heart is filled to its limit. As might be expected, definitive trials of the treatment of diastolic HF are lacking; what studies have been done have been either inconclusive or small.22,23 The Digoxin Investigation Group study24 of the effects of digoxin on outcomes in HF included a subgroup of 492 patients with normal EFs, but found no effect of digoxin compared with placebo on mortality or hospitalizations for HF. The AHA/ACC statement (cited above) on the treatment of HF has no specific recommendations for treating diastolic HF and just makes general suggestions such as treating the hypertension, controlling atrial rate if atrial fibrillation is present, and using diuretics if there is venous congestion.5 The only large-scale outcome study to date (published after the AHA/ACC statement) is the Candesartan in Heart-Assessment of Reduction in Mortality and Morbidity (CHARM)-Preserved trial,25 in which 3023 patients with HF and normal EFs were randomized to have either candesartan or placebo added to their other medications (surprisingly, less than 20% were taking angiotensin-converting enzyme inhibitors). There was no significant effect on the primary outcome (deaths plus hospitalizations), although the total numbers of admissions for HF were slightly fewer in the candesartan group. Diastolic HF, if such it should be called, is a condition that is beset with uncertainties and paradoxes. In brief, we do not know how to diagnose it exactly (other than by sophisticated and nonroutine invasive methods); we do not know what causes it; and we do not know specifically how to treat it. In our present state of knowledge, therefore, the average practicing physician may be excused if he or she does not pay much attention to it, other than being aware that HF with apparently normal ventricular (systolic) function is common in older patients with hypertension. There is an urgent need for a convenient and reliable noninvasive test to screen for it and for more outcome trials such as the CHARM study.25