The Sieverling paper reports Dr. Cheney believes diastolic heart failure lies at the heart of CFS. This papers presents an overview of what diastolic dysfunction is, how it is diagnosed, and how it is treated.
The diastolic phase
The first thing to note is that the heart is an elastic muscle; it distends to fill up with enough blood and then explosively contracts to pump the blood to the rest of the body.
First the deoxygenated blood from the veins enters into the right chamber (ventricle) of the heart where it is gently pumped to the lungs and oxygenated. From the lungs it flows into the upper chamber of the left side of the heart called the left atrium.
Upon opening of the mitral valve it enters the left ventricule. As ventricle reaches its maximum point of expansion the mitral valve opens and the blood from the left atrium begins to fill it. This wave of blood entering the left ventricule is called the ‘E” wave.
When the two chambers have an equal amount of blood a contraction occurs and that forces more blood nto the left ventricle. The wave of blood caused by this contraction is called the “A” wave. The process of ventricular relaxation and filling is called the diastolic phase.
Once the left ventricle is filled ventricular contraction occurs and oxygenated blood is pumped through the aorta into the body. This is the systolic phase. In a blood pressure reading such as 120/80, 120 indicates the peak pressure produced by the contraction of the ventricle as it pumps blood through the body during systole; 80 represents the lowest pressure reached as the left ventricle dilates to fill with blood. The force of the systolic contraction can be felt in the pulse.
The problem in diastolic dysfunction, then, is not a problem with pumping but with filling. Diastolic dysfunction occurs when the left ventricle of the heart is unable to fill with normal amounts of blood.
There are essentially two components to diastolic dysfunction: (1) the left ventricle becomes unable to relax enough to receive proper amounts of blood during the first phase of filling (relaxation), and/or (2) it becomes too stiff to accommodate the blood propelled into it by the atrial contraction during last phase. The heart, for a time, can make up for impaired ventricular relaxation by forcing more blood into the left ventricle but eventually the left ventricle becomes so stiff that no level of increased contraction can compensate and reduced cardiac output occurs. It is only in the last stages of diastolic dysfunction, stroke volume is impacted.
Diastolic dysfunction and symptom expression
Clinically diastolic heart failure presents in a similar way as systolic heart failure. The cardinal symptoms are shortness of breath, first upon exertion, but eventually extending to standing and while at rest. ‘The dominant and most recognizable symptom of congestive heart failure is ‘shortness of breath’.
‘The other typical complaint is fatigue’ (Francis et. al 2000). Other common symptoms include wheezing and coughing as well as nausea and vomiting and edema (bloating) in the extremities. Swelling of the ankles, particularly at days end, is often the first symptom that brings a patient into a physician’s office.
The veins of the neck are often distended and the upper right abdomen may ache. A cardiovascular examination often finds distinctive heart (‘gallop’, knock, murmurs, ‘heave’) and lung sounds (‘rales’) (Francis et. al. 2000).
These are all signs of blood backing up and causing problems either in the lungs or in the lower extremities. As the left ventricle becomes less and less able to accept the blood presented to it, the blood backs up into the venous circulation causing edema.
As left ventricular hypertrophy and/or stiffening proceeds blood can back up into the left atrium inhibiting circulation to the lungs and causing shortened breath as well as coughing, wheezing and lung sounds (rales) (Hanes et. al. 2005).
Some signs are more common to diastolic dysfunction than systolic dysfunction (ventricular hypertrophy, edema, lung sounds) and vice versa but it is impossible to differentiate the two simply during a clinical examination.
Note that the extent of heart failure can be estimated simply through an analysis of physical activity limitations:
- Stage I – no limitations of physical activity, ordinary physical activity does not lead to undue fatigue or shortness of breath
- Stage II – slight limitation of physical activity is seen. Patient is comfortable at rest or ordinary physical activity results in fatigue, palpitations or shortness of breath
- Stage III – marked limitations of physical activity are seen. Patient is comfortable at rest but even slight physical activity causes fatigue, palpitations or shortness of breath.
- Stage IV – symptoms of cardiac insufficiency are present at rest, and discomfort is increased with any physical activity.
Note that it is possible to be in the initial stages of heart failure and be asymptomatic. If, as Dr. Cheney suggests, heart failure is the cause of the activity limitations present in CFS then most CFS patients appear to be in stages III or IV.
Diagnosing diastolic dysfunction: laboratory tests
While systolic heart dysfunction is easily characterized using tests of ejection fraction diagnosing diastolic dysfunction is more difficult. Ejection fraction is the percentage of blood expelled by the heart during systole. Only one test – left ventricular end diastolic pressure (LVEDP) – can reliably determine diastolic dysfunction (Zile et. al. 2001).
LVEDP is the pressure in the left ventricule just prior to systole. Measuring LVEDP, however, requires cardiac catheritization, an invasive and expensive procedure, and is not commonly done.
This does not mean diagnosing diastolic dysfunction (DD) is not achievable, it simply means it is not simple. Since there are no other single measures other than LVEDP that are universally abnormal in patients with heart failure it has been difficult for the medical community to come to agreement over which set of tests reliably confirm it (Zile et. al. 2001, Francis et. al. 2000, Skaluba and Irwin 2004).
Taking a very conservative approach, an influential American group, the Farmington Heart Study Group, asserts diastolic heart failure is definite only when invasive tests show diastolic abnormalities shortly after an acute cardiac event.
Taking a more liberal approach the European Group on Diastolic Heart Failure posits that diastolic heart failure can be reliably addressed based on symptoms and signs (shortened breath during exercise, particular heart and lung sounds, and measures taken mostly by echocardiography (E/A ratio, deceleration time, IVRT, etc.) (Galdierisi 2005).
Next to cardiac catheritization echocardiography is the best means of assessing diastolic dysfunction. Unfortunately none of the measures of echocardiography are consistently abnormal in all patients with diastolic heart failure. Nor as we shall see are the abnormalities measured by echocardiography well correlated with independent measures of functioning such as exercise capacity or with the gold standard of diastolic dysfunction, LVEDP.
Given the entire battery of tests provided by echocardiography, however, most sources suggest a physician should be able to fairly confidently diagnose diastolic dysfunction in patients with symptoms of heart failure.
Two measures commonly used are ‘transmitral velocity’ and deceleration. As noted above the blood flow across the mitral valve into the left ventricle occurs in two phases; first during ‘E’ wave or ‘relaxation’ phase, it passively flows from the full left atrium into the near empty left ventricle; second, during the ‘A’ wave, after an equilibrium between the chambers is reached the left atrium contracts and forces more blood into the left ventricle. The E and A wave tests measure the velocity of blood across the mitral valve.
(1) Reduced relaxation stage – In this first stage of diastolic dysfunction the left ventricule is unable to relax enough to accept normal amounts of blood during the first (E) phase of ventricular filling. A reduced relaxation phase manifests itself as reduced E/A wave ratio.
An E/A wave ratio of less than .75 is considered to be indicative of diastolic dysfunction. Some controversy exists, however, how important a low E/A ratio, are. Because only a small fraction of people (2-3%) with a low E/A ratio have signs or symptoms of heart failure a low E/A ratio is not necessarily evidence of significant diastolic dysfunction or diastolic heart failure (Skaluba and Irwin 2004). (See below).
(2) Pseudonormal stage– Interestingly, as diastolic dysfunction proceeds the E/A ratio normalizes; i.e. achieves a pseudonormal appearance (between .75 and 1.5). This occurs when the increased stiffness of the left ventricule inhibits blood flow during the later phase (A wave).
Since reduced ventricular relaxation has already contributed to a reduced E wave, both waves now appear to be in the proper proportion and the E/A ratio is normal. Distinguishing a normal E/A wave ratio from a pseudonormal one requires looking at other measures.
Patients with a pseudonormal pattern and moderate diastolic dysfunction almost always have an enlarged left atrium. These patients also typically experience shortened breath during exertion and have moderate functional impairment (NYHA IIa-III).
The amplitude of their Ar wave is usually larger than 25 cm/sec. and its duration usually exceeds that of the A wave. Loading tests such as the Valsalva maneuver should indicate abnormalities.
(3) Restrictive stage – In the final stage of diastolic dysfunction called the restrictive filling stage, as left ventricular enlargement and fibrosis prevents the left ventricule from expanding during the atrial contraction phase, the E/A ratio reverses and a high E/A ratio is seen (>1.5).
In this stage the left ventricle has become so stiff that no amount of atrial contraction can force more blood into it. Because of this almost all the blood it receives occurs during the first (relaxation) phase.
In this stage blood build up in the left atrium increases blood pressure to the point that the mitral valve opens before a peak pressure gradient is produced. This further reduces filling during the relaxation period.
Between .75 & 1.5
1 (reduced relaxation)
Between .75 and 1.5
E-wave deceleration time
Deceleration time indicates how rapidly the early diastolic phase of filling proceeds. This is the phase, remember, in which filling is dependent on a pressure gradient from a left atrium that is full of blood to a left ventricule that has just ejected its blood and is mostly empty.
Deceleration times increase as left ventricular relaxation decreases; if the left ventricule is not distended as much as usual then the pressure gradient between it and the left atrium decreases and it will take longer to fill.
Isovolumic Relaxation time (IVRT, IRT)
IRT measures the time from the closing of the aortic valve – which halts blood flows out of the heart (systole) – to the time the mitral valve opens and blood begins to refill the left ventricle. Essentially it measures the time during which the ventricle relaxes just before it starts to receives blood again.
Hearts with diastolic dysfunction usually take longer to fully relax. Healthy hearts usually take about 70 +/- 12 ms (microseconds?) while hearts with diastolic dysfunction usually take about 110 ms to fully relax.
Atrial Flow Reversal
Even in healthy people the force of the atrial contraction during the last phase of diastole often causes some blood to back up into the pulmonary veins. This pulse of blood can be picked up on an echocardiogram and is called an Ar wave.
As left ventricular pressure builds up in heart failure due to increased stiffness of the heart muscle more and more blood is forced into the pulmonary veins and the amplitude and the duration of the Ar wave increases. Diastolic dysfunction is indicated when the amplitude of the A-wave is over 25 cms. and its duration lasts longer than that of the A wave.
Based on the information given above a generalized chart can be developed (Koren 2002, Galderisi 2005, Pollard and Pozzoli 2001).
Degree of dystolic dysfuncton
E wave deceleration
Ar vs. A wave
None – normal
*70 +/- 12
Mild to moderate
20 ms >
25 ms >s
*> 40 years of age = 80 +/- 12
As the load on the heart cells increases during diastolic heart failure they begin to shift position and enlarge causing left ventricular hypertrophy and/or stiffness. Increased collagen deposition and fibrous tissue production due to activation of the renin-aldosterone-angiotensin system further stiffens the ventricule.
Left ventricular hypertrophy or enlargement is is well correlated with increased left ventricular pressure (LVEDP) and is, therefore, a good indicator of serious diastolic dysfunction (Zile et. al. 2001). Left ventricular enlargement is much common in diastolic than in systolic heart failure (Shamsham and Mitchell 2000).
Another commonly found structural abnormality is left atrial dilation (Shamsham and Mitchell 2000, Torosoff and Philbin 2003). Left atrial dilation occurs when a stiff left ventricule forces blood to back up into the left atrium. Chest x-rays can also distinguish a third abnormality of significant diastolic dysfunction called pulmonary congestion.
This is all very complicated but the upshot is that structural changes occurring in diastolic dysfunction often cause heart enlargement or hypertrophy, and these can be readily found.
Although it appears that most physicians and reference sources accept the reliability of non-structural measures of diastolic dysfunction two recent papers with contrasting positions indicate a real disparity exists in how valuable researchers believe these tests are.
Interestingly, both papers suggest these tests are unnecessary; one because the authors assert a more accurate test is available and one that asserts tests of systolic functioning are all that is needed to diagnose diastolic dysfunction.
Traditional measures of diastolic dysfunction are unreliable? A key process in mapping out the cause of any disease is correlating symptom exacerbation with measures of dysfunction. If lactic acid buildup, for instance, contributes to muscle fatigue then as the muscles get more and more fatigued lactic acid levels should rise.
Since exercise intolerance is one of the cardinal symptoms of heart failure one study examined indices of diastolic dysfunction to determine if they were correlated with measures of exercise intolerance (Skaluba and Litwin 2004)
The authors noted how inconsistently measures of mitral valve flow (E/A) or pulmonary vein flow (atrial flow reversal) have correlated in past studies with left ventricular filling pressure – the gold standard for diastolic dysfunction.
Remember that diastolic dysfunction is synonymous with stiffness of the left ventricle. If it is unable to relax enough to allow filling the heart will use enormous pressure to fill it up – thus left ventricular filling pressure is indicative of diastolic dysfunction.
The results are even poorer when individuals with intact systolic functioning (i.e. those with putative diastolic dysfunction such as CFS patients) are included. This suggests that E/A and atrial flow reversal tests are relatively poor indicators of diastolic function.
The problem with the E measure is its sensitivity to both changes in preload and afterload means it may reflect changes in those conditions rather than in reduced left ventricular relaxation or increased stiffness.Preload is the amount of tension in the ventricular wall just prior to when it contracts and propels blood into the aorta.
Simply put it is a function of anything (blood volume, pressure) that effects filling (high BV, BP = high preload, . Afterload is the force in the arteries the heart must push against in order to expel its blood.
Afterload is effected by peripheral arterial resistance, (high vasoconstriction = high afterload), viscosity of the blood (high viscosity=high afterload), etc. Thus conditions which cause increased preload, such as high blood pressure or reduced preload such as low blood volume (which many CFS patients have) may skew the results of these tests.
This study found that traditional indices of mitral valve inflows such as the E/A ratio, E-deceleration time and isovolumetric relaxation time were not significantly correlated with exercise capacity (Skaluba and Litwin 2004). (Isovolumetric relaxation time was almost (p<.069) significantly correlated with exercise capacity).
That left ventricular mass correlated well with exercise capacity suggested left ventricular hypertrophy is a good indicator of serious diastolic dysfunction.
The measure which was best correlated with exercise capacity, however, was called E/Ea. E/Ea is the velocity of blood flows past the mitral valve divided by the early diastolic velocity of the mitral valve annulus. As Ea is less affected by preload than E/A it appears to be a more effective measure of left ventricular relaxation rates.
E/Ea has been shown to be well correlated with both left ventricular diastolic pressure and with pulmonary capillary wedge pressure (PCWP). PCWP is the pressure found in the pulmonary arteries during diastole. As pressure builds up in the left ventricule it begins to cut off the circulation to the lungs, causing difficulty in breathing. It is increased LVEDP not reduced diastolic relaxation which produces the symptoms of heart failure.
Tests of diastolic dysfunction are unnecessary?
On the other side of the spectrum some researchers have posited that diagnosing diastolic heart failure does not require tests of diastolic dysfunction but can be done through exclusion; i.e. diastolic heart failure occurs when heart failure can be reliably diagnosed based on symptoms and no systolic dysfunction (reduced ejection fraction) can be found.
A recent study found that while great variability was found in the prevalence of almost every test of diastolic dysfunction, almost every patient with at least mild hypertrophy and heart failure exhibited abnormal diastolic function in at least one of the tests (Zile et. al. 2001).
Only increased LVEDP was almost universally found in all patients. Almost all the indices occurred in from 40-60% of all the patients. The authors concluded that given a reliable diagnosis of heart failure based on symptom presentation and normal ejection fraction (no systolic dysfunction), diastolic dysfunction can be assumed and there is no need for further testing (Zile et. al. 2001).
This is, however, a controversial position as only one indice of abnormal diastolic dysfunction was needed to assume DHF was present and no control study group was used. Note also that having at least mild hypertrophy was a pre-condition for being included in the study.
Prevalence and significance of mild diastolic dysfunction
Since the symptoms CFS patients display do not always correlate well with those found in heart failure an important question may be how significant findings of diastolic dysfunction, in particular, diastolic relaxation, are in other people with objective evidence of diastolic dysfunction but without strong symptomatic evidence of it.
Several studies that have examined the prevalence of diastolic dysfunction in the middle aged and more elderly population have found that it is is not at all uncommon. Using echocardiography a large study in Minnesota found that 21%, 7% and 1% of middle-aged and older adults had mild (impaired relaxation only), moderate (pseudonormal filling) and severe diastolic dysfunction (restrictive) respectively (Redfield et. al. 2003).
Most of those found to have moderate to severe diastolic functioning had not been diagnosed with heart failure.
Were the largely asymptomatic people with mild or moderate diastolic dysfunction at risk? A large study, the Strong Heart Study, found that after confounding factors such as age, sex, cholesterol, blood pressure, diabetes, etc. were taken into account, that E/A ratio’s less than 0.6 (= impaired relaxation, mild diastolic dysfunction) were not associated with increased mortality in an older population.
High E/A ratio’s (>1.5), however, were associated with an increased risk of mortality (Bella et. al. 2002). Another five year follow up study found mild disastolic dysfunction conferred a significantly increased risk of mortality (Redfield et. al. 2003). This study, however, accounted for many fewer confounding factors that the Strong Heart Study. These studies suggest findings of restrictive filling may place someone with few symptoms of heart failure at risk.
CFS patients should bear in mind the sample groups in these studies were much older than the typical CFS patient and that if heart failure is common in CFS it appears, in contrast to most people with heart failure, to be largely non-progressive.
Given these contrasting position it is no wonder that the Hurst Manual of The Heart, a standard cardiac reference text, states ‘there is no agreement as to what constitutes abnormal diastolic dysfunction” and that the ‘recognition, evaluation and treatment of diastolic heart failure remains an obvious challenge’ (Francis et. al. 2000).
What can a patient take away from this maze of sometimes conflicting information? One’s own physician must, of course, be the guide in interpreting the tests regarding this complex matter. This review left me – a layman – with the following conclusions:
Since findings of mild diastolic dysfunction (impaired relaxation) are not well correlated with the gold standard of diastolic functioning, LVEDP , or aerobic capacity, their significance regarding heart failure is unclear. Since it is not uncommon for healthy people to have low E/A ratio’s, and atrial flow reversal, these measures, in particular, should be treated with caution. On the other hand findings that indicate restrictive filling is occurring appear to be significant.
The E/Ea flow test appears to be a more solid indicator of diastolic heart failure than the E/A ratio. Similarly, structural abnormalities such as left ventricular hypertrophy and left atrial dilation are signs of significant diastolic dysfunction as are indications of pulmonary congestion on a chest x-ray.
While there are questions regarding the efficacy of several measures of diastolic dysfunction it must be noted that most authors and reference sources appear comfortable in recommending these tests to verify diastolic heart failure when symptoms of heart failure are present and tests of ejection fraction are normal.
What test abnormalities mean in the absence of such symptoms is entirely unclear to me. One study found mild diastolic dysfunction was not uncommon in the people without symptoms of heart failure (about 20%).
CFS patients appear to inhabit a kind of in-between ground with regard to heart failure symptoms. Every patient is different, of course, but CFS patients as a group (a) do not appear to exhibit the symptoms of heart failure to the same degree as heart failure patients (i.e. shortened breath, edema are not hallmarks of CFS), (b) they do not commonly exhibit many of the symptoms of heart failure (edema, rales, particular heart sounds, upper right abdominal quadrant soreness, nausea, coughing, wheezing), and (c) they exhibit many symptoms that are not common to heart failure patients (increased allergies, chemical sensitivities, lymph node swelling, muscle pain, sore throat, impaired cognition,etc.). With regard to symptoms they present an atypical form of heart failure.
Both Peckerman and Lerner have suggested CFS patients may have a ’subtle’ form of heart failure that has slipped below the physicians and researchers radar screen. Even though it is ‘subtle’ they suggest it still may have disabling consequences.
It appears clear that if heart failure commonly exists in CFS it does not (a) follow the traditional course of heart failure, (b) that it is accompanied by a set of dysfunctions unique to CFS patients, and that (c) some of the most common compensatory mechanisms triggered by heart failure are not found in CFS (RAA activation, increased blood volume, increased uric acid levels).
While this paper has been about diastolic dysfunction it is still far from clear that tests of systolic functioning; that is, ejection fraction, are normal in CFS patients. In his 2005 presentation Dr. Cheney reportedly states systolic dysfunction is not commonly found in CFS.
A study by Peckerman to be published sometime in 2005 is expected, however, to indicate that ejection fraction is dramatically reduced in some CFS patients. This could indicate both diastolic and systolic dysfunction is present in at least some CFS patients.
There is clearly still much to learn about the heart and CFS. Hopefully the funding will be available for researchers to pursue this new arena of interest.
Several classes of drugs are commonly used to treat diastolic dysfunction.
Angiotensin converting enzyme inhibitors (ACEI’s – enalapril, lisinopril) and angiotensin receptor blockers(ARB’s – candesartan, losartan) are among the most common pharmaceuticals used in treating diastolic dysfunction. ACEI’s can decrease left ventricular hypertrophy and increase left ventricular relaxation. ARB’s reduce hypertension and improve exercise intolerance and quality of life.
Beta blockers (atenolol, metoprolol) are the other main treatment for diastolic dysfunction. By slowing the heart rate BB’s cause increased left ventricular filling time. They also improve hypertension, reduce left ventricular hypertrophy and inhibit renin release.
Calcium channel blockers can reduce left ventricular hypertrophy and improve stage one filling and exercise capacity.
Diuretics are commonly used to reduce total blood volume, which is often raised in heart failure. Reducing blood volume reduces diastolic pressure. In an attempt to compensate for increased diastolic pressure heart cells increase in size. Over time this leads to increased ventricular stiffness and the inability to accommodate normal flows of blood. The problem with diuretics is that, not surprisingly, given their reduction in blood volume, they often result in reduced cardiac output, which is already decreased in some patients with diastolic dysfunction. A subset of CFS patients have reduced not increased blood volume.
Exercise is effective in reducing the symptoms of diastolic heart failure (Haney et. al. 2005). The Hurst Manual states “Importantly, exercise intolerance can improve with training, which should be encouraged in patients with classes I-III (of IV) heart failure symptoms” (Francis et. al. 2000).
- A Guide to Cardiovascular Issues in CFS: Part I – Testing the Heart, Stroke Volume, Future Research
- A Guide to Cardiovascular Issues in Chronic Fatigue Syndrome. Part II: the Cheney Theory – An Inquiry
- A Guide to Cardiovascular Issues in Chronic Fatigue Syndrome Part IVa: Superoxide and the Heart
- A Guide to Cardiovascular Issues in CFS IVb: Peroxynitrite and the Heart
Bella, J., Palmieri, V., Roman, M., Liu, J., Welty, T., Lee, E., Fabsitz, R., Howard, B. and R. Devereux. 2002. Mitral valve ratio of early to late diastolic filling velocity as a predictor of mortality in middle-aged and elderly adults: the Strong Heart Study. Circulation 105, 1928-1933.
DeBececker, P. Dendale, P., De Meirleir, K., Campine, I., Vandenborne, K. and Y. Hagers. 1998. Autonsomic testing in patients with Chronic Fatigue Syndrome. American Journal of Medicine 105: 22S-26S/
Francis, G., Gassler, J., Sonnenblick, E. Pathophysiology and diagnosis of heart failure.2001. In Hurst’s The Heart, 10th e.d., 655-679. Ed. Fuster, V., Alexander, R. and R. Rourke. McGraw-Hill, New York
Koren, M.J. Diastolic Congestive Heart Failure. Medical Director, Non-Invasive Cardiology, Memorial Hospital, Jacksonville, Florida.
Haney, S., Sur, D. and Z. Xu. 2005. Diastolic heart failure: a review and primary care perspective. Journal of American Board of Family Practices 18: 189-98.
Mosterd, A., Hoes, A., de Bruyne, M., Deckers, J., Linkert, D., Hofman, A. and D. Grobbee. 1999. Prevalence of heart failure and left ventricular dysfunction in the general population. The Rotterdam Study. European Heart Journal 20, 447-455.
Otto, M., Balohlavek, M., Khandheria, B., Gilman, G., Svartikova, A. and V. Somers. 2004. Comparison of right and left ventricular function in obese and nonobese men. American Journal of Cardiology 93: 1569-1572.
Redfield, M., Jacobsen, S., Burnett, J., Mahoney, D., Bailey, K. and R. Rodeheffer. 2003. Burden of systolic and diastolic ventricular dysfunction in the community. JAMA 289, 194-202.
Roelandt, J. and M. Pozzoli. Non-invasive assessment of left ventricular diastolic function and filling pressure.
Skaluba, S. and A. Irwin. 2004. Mechanisms of exercise intolerance. Circulation 109: 972-977.
Torosoff, M. and Philbin, E. 2003. Improving outcomes in diastolic heart failure. Postgraduate Medicine 113: 51-58.
Zile, M., Gaasch, W., Carroll, J., Feldman, M., Aurigemma, G., Schaer, G., Ghali, J. and P. Liebson. 2001. Heart failure with a normal ejection fraction. Is measurement of diastolic function necessary to make the diagnosis of diastolic heart failure? Circulation 104: 779-782.