1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. References

Diastolic dysfunction can be diagnosed on equilibrium radionuclide angiocardiography (ERNA) by a low peak filling rate (PFR) in the setting of a normal left ventricular ejection fraction (LVEF). The authors evaluated the relationship between diastolic dysfunction, LVEF, and end-diastolic volume (EDV). A total of 408 predominantly asymptomatic patients with an LVEF ≥50% by ERNA were studied. LVEF of patients with a low PFR was compared with the LVEF of patients with a normal PFR. Correlation analyses to evaluate the association between PFR and EDV were also performed. The LVEF of patients with a low PFR was lower than the LVEF of patients with normal PFR (59±7 vs 63%±7%; P<.01). There was no correlation between EDV and PFR (r=−0.04; P=.32). The results did not change when the EDV indices were used. In patients who had repeat scans, there was no correlation between the change in EDV and the change in PFR (r=0.16; P=.2). In asymptomatic patients undergoing ERNA who have normal systolic function, a low PFR can be associated with a lower LVEF, but it is not associated with changes in EDV. This suggests that diastolic dysfunction is associated with mild systolic dysfunction.

Diastolic dysfunction refers to abnormal mechanical properties of the myocardium and includes abnormal left ventricular (LV) diastolic distensibility, impaired filling, and slow or delayed relaxation—regardless of whether the LV ejection fraction (LVEF) is normal or depressed and whether the patient is asymptomatic or symptomatic.[1, 2] These diastolic abnormalities are usually present when systolic function is abnormal, but may also occur in isolation in persons with normal or near-normal systolic ventricular function. A recent population-based study reported a high prevalence (approximately 28%) of isolated diastolic dysfunction by echocardiogrpahy in a population older than 45 years, and its presence was associated with increased mortality.[3] The cause of this increased morbidity and mortality remains unclear.

The ventricular relaxation abnormalities can be diagnosed by a low peak filling rate (PFR) on equilibrium radionuclide angiocardiography (ERNA) and correlate well with changes seen on Doppler echocardiography.[4] Isolated diastolic filling abnormalities are frequently seen in patients with hypertension (HTN) and coronary artery disease[5, 6] and are seen in patients undergoing chemotherapy.[7, 8] We tested whether abnormalities in diastolic function are associated with changes in LVEF[9, 10] or changes in end-diastolic volume (EDV) or both. Because heart failure with preserved ejection fraction (HFPEF) was first diagnosed using ERNA, the criteria for diagnosis are well established, even though today the diagnosis of diastolic dysfunction is more frequently made by echocardiography.[11] Even though ERNA is still used today in situations where precise quantification of LV volumes (and ejection fraction [EF]) with high reproducibility is required, it has been supplanted by echocardiography (especially using harmonic imaging, contrast imaging, and 3-dimensional imaging) in most routine clinical situations.[12, 13] However, because the evaluation of LV volumes with ERNA is less operator-dependent and more reproducible compared with echocardiography, the former represents an optimal technique to evaluate the relationship between diastolic dysfunction and ventricular EF and volumes.


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. References

The Yale cardiovascular nuclear imaging database was examined for ERNAs performed between 1995 and 1997. The database in the form used for this study was started in 1995. Because in subsequent years we found that the overwhelming majority of patients in the database were oncology patients, we limited the study population to patients examined between 1995 and 1997. Only patients with normal (≥50%) EF and with normal regional wall motion were included in the study. When multiple studies were available, only the first was included in the analysis. ERNA images were obtained in the anterior, left anterior oblique, and lateral views after labeling of the patients' own red blood cells.[14] Studies were formatted at 16 frames per cardiac cycle. The time activity curve was derived using a validated software program (Yale MGA) and the PFR was calculated from the resulting LV volume curve. This method has been compared with high temporal resolution acquisitions using the nuclear probe, and good correlations have been found with no differences in calculated PFR when framing rates were increased from 16 to 64 frames per cardiac cycle.[15] The EDV was calculated using the count proportional method of Massardo.[16]

Pearson correlations were calculated to assess the relationship between PFR and EDV. Partial correlations were also estimated using multiple regression with adjustment for age and sex and the analysis was repeated adjusting for age, sex, sex, LVEF, HTN, previous chemotherapy, coronary artery disease (CAD), and congestive heart failure (CHF). In the subgroup of patients with weight and height recorded, the correlation was reassessed with adjustment for body mass index and the above-mentioned variables. Lastly, the correlation between PFR and EDV/body surface area (EDV index) was assessed in the same subgroup of patients.

A PFR normalized to EDV of <2.5 EDV/s was considered abnormal.[4, 5] The EDV of patients with an abnormal PFR was compared with the EDV of patients with unimpaired PFR using a t test. Multiple regression was then performed to evaluate abnormal PFR vs normal PFR differences with adjustment for age, LVEF, and sex. These analyses were repeated using a PFR normalized to the stroke volume (SV), with a PFR <4SV/s considered abnormal.[15] Moreover, we also repeated these analyses using the European Study Group on Diastolic Heart Failure age group–adjusted diagnostic criteria of impaired diastolic filling by radionuclide angiocardiography.[6] Also, the EDV of patients with HTN and of patients who had undergone chemotherapy was compared with the EDV of the patients without HTN and of the patients who had not undergone chemotherapy. Additionally, we compared the first and the last scans of the patients who had multiple scans during the selected period to determine whether the changes in EDV correlated with changes in PFR over time. Statistical analysis was performed with commercially available software (SPSS, Chicago, IL, and GraphPad, San Diego, CA). Two-tailed tests were performed throughout. A P<.05 was considered statistically significant. The study was approved by the Yale University human investigation committee. Because the database did not include echocardiography results and because we used only human investigations committee approval to review the database results, we did not have echocardiography results to compare with the ERNA results.


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. References

A total of 408 patients were included in the study. The patients' characteristics are shown in Table 1. The patients were predominantly women, the average age was 50 years, and the mean LVEF was 62%. Fifteen (4%) patients were taking diuretics, 12 (3%) were taking angiotensin-converting enzyme inhibitors, 10 (3%) were taking digoxin, 8 (2%) were taking calcium channel blockers, and 7 (2%) were taking β-blockers. Less than 10% of patients had a history of heart failure. The clinical indications for the ERNA are shown in Table 2. Most patients were referred for EF evaluation before chemotherapy initiation and a smaller number were referred during chemotherapy for surveillance of LV function. Only a small number of patients (68 [17%]) were referred for symptoms of dyspnea that could be consistent with a diagnosis of heart failure (but not necessarily with an established diagnosis of heart failure) for evaluation of LVEF and cavities. Thus, the majority of patients were asymptomatic.

Table 1. Clinical Characteristics and Medical History of Study Patients (n=408)
  1. Abbreviations: EDV, end-diastolic volume.

  2. a

    By history.

Men, No. (%)150 (37)
Age, y51.5±17
Previous chemotherapy, No. (%)a69 (16.9)
Hypertension, No. (%)a47 (11.5)
Coronary artery disease, No. (%)a19 (4.7)
Congestive heart failure, No. (%)a40 (9.8)
Ejection fraction, %62.4±7
Peak filling rate, EDV/s3.1±1
End-diastolic volume, mL156±47
End-systolic volume, mL59±21
Table 2. Indications for Equilibrium Radionuclide Angiocardiography in the Study Patients (n=408)
Pre-chemotherapy, No. (%)219 (54)
Chemotherapy surveillance, No. (%)69 (17)
Dyspnea/lung congestion/edema, No. (%)68 (17)
Hypertension, No. (%)20 (5)
Coronary disease, No. (%)12 (3)
Miscellaneous, No. (%)20 (5)

When patients were classified as having normal and abnormal PFR based on a threshold of 2.5 EDV/s, 104 patients were found to have diastolic dysfunction. Patients with an abnormal PFR were older and had a lower LVEF (59%±7% vs 63%±7%; P<.01), but did not have significantly higher EDV than patients with normal PFR (152±44 vs 156±48; P=.35). However, the end-systolic volume of the patients with abnormal PFR was significantly higher (Table 3). When the PFR was normalized to the SV and analyses were repeated using a PFR <4 SV/s as abnormal,[15] the number of patients with diastolic dysfunction was reduced to 88. The patients with PFR <4 SV/s did not have significantly different LVEF or a significantly different EDV than patients with PFR ≥4 (Table 4). Repeating the analysis using the European Study Group on Diastolic Heart Failure threshold values for impaired diastolic function by radionuclide angiocardiography (PFR <2 EDV/s for persons younger than 30 years, PFR <1.8 EDV/s for persons between 30 and 50 years, and PFR <1.6 for persons older than 50 years) markedly reduced the number of patients with diastolic dysfunction to 18. Despite this, the analysis yielded similar results (Table 5): the patients with low PFR had a lower LVEF but EDV was not significantly different between the two groups.

Table 3. Comparison Between Patients With Normal and Abnormal Peak Filling Rate
 Peak Filling Rate ≥2.5 EDV/s (n=304)Peak Filling Rate <2.5 EDV/s (n=104)P Value
  1. a


Men, No. (%)101 (33)49 (47).02
Age, y49±1561±15<.01
Body mass index, kg/m2a26.8±6.227.9±8.3.61
Ejection fraction, %63±759±7<.01
End-diastolic volume, mL156±48152±44.35
End-diastolic volume index, mL/m2a84.6±2983.7±26.8
End-systolic volume, mL57±2163±23.03
Previous chemotherapy, No. (%)50 (16.4)19 (18.2).65
Hypertension, No. (%)32 (10.5)15 (14.4).28
Coronary artery disease, No. (%)11 (3.6)8 (7.6).11
Congestive heart failure, No. (%)21 (6.3)19 (20.1).002
Table 4. Comparison Between Patients With Normal and Abnormal Peak Filling Rate Normalized to Stroke Volume
 Peak Filling Rate ≥4 SV/s (n=320)Peak Filling Rate <4 SV/s (n=88)P Value
  1. a


Men, No. (%)116 (36)34 (39).7
Age, y52±1653±15.56
Body mass index, kg/m2a26.6±6.128.7±8.8.13
Ejection fraction, %62±763±7.5
End-diastolic volume, mL158±48152±43.27
End-diastolic volume index, mL/m2a85.7±2980.2±23.28
End-systolic volume, mL59±2256±19.12
Previous chemotherapy, No. (%)60 (18.8)9 (10.2).07
Hypertension, No. (%)30 (9.3)17 (19.3).01
Coronary artery disease, No. (%)14 (4.3)5 (5.6).57
Congestive heart failure, No. (%)32 (10)9 (10.2)1.0
Table 5. Comparison Between Patients With Normal and Abnormal Peak Filling Rate Adjusted for Age (According to the European Study Group on Diastolic Heart Failure)
 Peak Filling Rate Normal for Age (n=390)Peak Filling Rate RateAbnormal for Age (n=18)P Value
  1. a


Men, No. (%)145 (37)5 (28).46
Age, y52±1646±16.11
Body mass index, kg/m2a27.1±727±7.96
Ejection fraction, %63±759±7.04
End-diastolic volume, mL157±47155±32.83
End-diastolic volume index, mL/m2a84.2±2887.1±23.69
End-systolic volume, mL59±2256±19.12
Previous chemotherapy, No. (%)64 (16.4)5 (27.8).2
Hypertension, No. (%)44 (11.3)3 (16.7).45
Coronary artery disease, No. (%)19 (5)0 (0).33
Congestive heart failure, No. (%)39 (10)1 (6).83

There was no correlation between PFR and EDV (r=−0.04; P=.32). The lack of correlation persisted after adjusting for age and sex (P=.31) and after adjustment for age, sex, LVEF, HTN, previous chemotherapy, CAD, and CHF (P=.43). Repeating the analysis after adjustment for the above variables and for BMI in the 281 patients whose weight and height were available also failed to reveal a correlation between PFR and EDV (P=.22). In the same subgroup of 281 patients, PFR did not correlate with the EDV index either (r=0.025; P=.66).

Patients with HTN had significantly lower PFR than patients without HTN (2.9±0.8 vs 3.1±0.9 EDV/s; P=.03), while patients who underwent chemotherapy did not have significantly lower PFR compared with patients who did not undergo chemotherapy (3.2±1 vs 3.1±0.9 EDV/s; P=.34). EDV did not differ significantly between patients with HTN and patients without HTN (155±39 vs 157±48 mL; P=.77) or between patients who underwent and did not undergo chemotherapy (156±37 vs 157±48 mL; P=.77).

Sixty-six patients had repeat ERNA scans. The time between the first and the last scan was 8±4 months. The PFR was 3.26±0.9 EDV/s at the time of the first scan and 3.27±0.7 EDV/s at the last scan (P=.84). The EDV was 149±38 mL at the first scan and 155±49 mL at the time of the last scan (P=.24). The EF was 63.5%±0.9% at the first scan and 62.5%±0.9% at the time of the last scan (P=.16). The mean change in EDV was 5.8±38.4 mL and the change in PFR was 0.01±0.71 EDV/s. There was no correlation between the change in EDV and the change in PFR (r=0.16; P=.2).


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. References

The results of this study did not establish a significant association between an increased EDV and an abnormal PFR in this population of patients referred for ERNA who were mostly asymptomatic and who had normal global and regional systolic LV function. Since most of the patients in our study were asymptomatic, the patients with abnormal filling rates would fit the definition of patients with diastolic dysfunction. However, the LVEF of patients with decreased PFR was significantly lower compared with patients with normal PFR in this population. This suggests that while there is no significant change in EDV associated with diastolic dysfunction as assessed by a decreased PFR on radionuclide angiocardiography,[6] end-systolic volume is increased, thus yielding a lower LVEF. These findings are in agreement with data from a recent report that did not show changes in EDV measured by echocardiography in patients with HFPEF.[17] Haykowsky and colleagues also reported that elderly patients with HFPEF have a lower EF at rest compared with healthy controls.[18] Moreover, Bonow and colleagues[5] demonstrated that in healthy patients, neither PFR nor the time to PFR correlated with echocardiographic LV end-diastolic transverse dimension. Our results do not rule out the possibility that an increase in EDV is associated with a decrease in EF below normal limits[19, 20] because we included only patients with a normal EF in our study. Our results are in agreement with a recent study by cardiac magnetic resonance and by magnetic resonance spectroscopy of patients with stable diabetes mellitus, which reported an altered diastolic function and a decreased LVEF together with altered phosphocreatine metabolism in these patients compared with healthy controls.[21]

The fact that the PFR was not significantly lower in patients undergoing chemotherapy underscores the fact that only a fraction of patients receiving chemotherapy develop cardiac abnormalities[7] (and that the incidence of these abnormalities depends on type of chemotherapy, dose, and length of follow-up), while HTN is more likely to cause changes in cardiac function.[6, 22] The results of the patients who had repeat ERNA scans also suggest that changes in PFR are not related to changes in EDV. We did not have the heart rate values of the patients in our database, and thus we could not correct for this variable, which is linked to the PFR.[23, 24] However, because our analysis revealed no significant correlation between PFR and EDV, it is unlikely that including the heart rate into the model would have transformed the correlation into a significant one. Moreover, the use of the PFR/SV, which is not as dependent on the heart rate,[15, 24] did not significantly alter our findings.

A recent study found that EDV is higher in hypertensive patients with HFPEF compared with nonhypertensives,[25] a finding that could not be reproduced by our study, which was conducted in a population that was generally healthy from a cardiac standpoint. The same group also reported an increased EDV in elderly hypertensive patients with HFPEF compared with hypertensive patients without HFPEF in the Cardiovascular Health Study.[26] We found an abnormal PFR but not an increased EDV in the hypertensive subgroup of our patient population; however, our patients were mostly asymptomatic and on average 20 years younger. Very few studies have examined the relationship between diastolic function and EDV or LVEF. These studies were performed mostly in patients with hypertrophic cardiomyopathy and had sample sizes that were much smaller compared with ours.[4, 6, 15, 27-29] Also, because we have only information about the EDV but not about ventricular dimensions and wall thickness, we cannot make statements about the geometry of the ventricle and ventricular mass, which may be important factors in the progression from diastolic dysfunction to clinically manifest heart failure.[9, 30] Finally our data are in agreement with published echocardiographic data using tissue Doppler and speckle tracking, which suggest that subtle systolic abnormalities are frequently associated with diastolic dysfunction. Yu and colleagues showed lower peak regional myocardial sustained systolic velocities in patients with diastolic dysfunction;[20] Vinereanu and associates[31] showed that patients with diastolic dysfunction have decreased long-axis systolic function; Garcia and coworkers showed lower systolic tissue velocities in patients with HFPEF;[32] and, finally, Imbalzano and colleagues[33] showed that patients with HTN and diastolic dysfunction by tissue Doppler have systolic abnormalities by speckle tracking.

We also noted a significant change in the prevalence of diastolic dysfunction depending on the cutoff used to define abnormal PFR. These findings are in agreement with echocardiographic studies in an elderly population, where the prevalence of diastolic dysfunction varied depending on the cutoff used.[34] The proportion of patients with an abnormal diastolic function is also variable in the population depending on the criteria used. The number of persons with abnormal diastolic filling is smallest when corrected for age, but the prevalence reported here is consistent with larger population studies.[34] In addition, in an echocardiographic population study, Redfield and coworkers[3] found a prevalence of diastolic dysfunction of 28%, which is similar to the prevalence of diastolic dysfunction defined by an abnormal PFR in our study.


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. References

Our data suggest that LV end-diastolic volume is unchanged in most patients with diastolic dysfunction but suggests that diastolic dysfunction may be accompanied by (or may appear simultaneously with) a subtle impairment of systolic function.


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. References
  • 1
    Zile MR, Brutsaert DL. New concepts in diastolic dysfunction and diastolic heart failure: part I: diagnosis, prognosis, and measurements of diastolic function. Circulation. 2002;105:13871393.
  • 2
    Gaasch WH, Zile MR. Left ventricular diastolic dysfunction and diastolic heart failure. Annu Rev Med. 2004;55:373394.
  • 3
    Redfield MM, Jacobsen SJ, Burnett JC Jr, et al. Burden of systolic and diastolic ventricular dysfunction in the community: appreciating the scope of the heart failure epidemic. JAMA. 2003;289:194202.
  • 4
    Aggarwal A, Brown KA, LeWinter MM. Diastolic dysfunction: pathophysiology, clinical features, and assessment with radionuclide methods. J Nucl Cardiol. 2001;8:98106.
  • 5
    Bonow RO, Bacharach SL, Green MV, et al. Impaired left ventricular diastolic filling in patients with coronary artery disease: assessment with radionuclide angiography. Circulation. 1981;64:315323.
  • 6
    How to diagnose diastolic heart failure. European Study Group on Diastolic Heart Failure. Eur Heart J. 1998;19:9901003.
  • 7
    Jain D. Cardiotoxicity of doxorubicin and other anthracycline derivatives. J Nucl Cardiol. 2000;7:5362.
  • 8
    Gharib MI, Burnett AK. Chemotherapy-induced cardiotoxicity: current practice and prospects of prophylaxis. Eur J Heart Fail. 2002;4:235242.
  • 9
    Levy D, Larson MG, Vasan RS, et al. The progression from hypertension to congestive heart failure. JAMA. 1996;275:15571562.
  • 10
    Berenji K, Drazner MH, Rothermel BA, Hill JA. Does load-induced ventricular hypertrophy progress to systolic heart failure? Am J Physiol Heart Circ Physiol. 2005;289:H8H16.
  • 11
    Paulus WJ, Tschope C, Sanderson JE, et al. How to diagnose diastolic heart failure: a consensus statement on the diagnosis of heart failure with normal left ventricular ejection fraction by the Heart Failure and Echocardiography Associations of the European Society of Cardiology. Eur Heart J. 2007;28:25392550.
  • 12
    Baur LH. Evaluation of left ventricular function: does the imaging technique matter? Int J Cardiovasc Imaging. 2008;24:803805.
  • 13
    Lim TK, Senior R. Noninvasive modalities for the assessment of left ventricular function: all are equal but some are more equal than others. J Nucl Cardiol. 2006;13:445449.
  • 14
    Callahan RJ, Froelich JW, McKusick KA, et al. A modified method for the in vivo labeling of red blood cells with Tc-99m: concise communication. J Nucl Med. 1982;23:315318.
  • 15
    Bowman LK, Lee FA, Jaffe CC, et al. Peak filling rate normalized to mitral stroke volume: a new Doppler echocardiographic filling index validated by radionuclide angiographic techniques. J Am Coll Cardiol. 1988;12:937943.
  • 16
    Massardo T, Gal RA, Grenier RP, et al. Left ventricular volume calculation using a count-based ratio method applied to multigated radionuclide angiography. J Nucl Med. 1990;31:450456.
  • 17
    Kitzman DW, Little WC, Brubaker PH, et al. Pathophysiological characterization of isolated diastolic heart failure in comparison to systolic heart failure. JAMA. 2002;288:21442150.
  • 18
    Haykowsky MJ, Brubaker PH, John JM, et al. Determinants of exercise intolerance in elderly heart failure patients with preserved ejection fraction. J Am Coll Cardiol. 2011;58:265274.
  • 19
    Clements IP. Combined systolic and diastolic dysfunction in the presence of preserved left ventricular ejection fraction. Eur J Heart Fail. 2005;7:490497.
  • 20
    Yu CM, Lin H, Yang H, et al. Progression of systolic abnormalities in patients with “isolated” diastolic heart failure and diastolic dysfunction. Circulation. 2002;105:11951201.
  • 21
    Diamant M, Lamb HJ, Groeneveld Y, et al. Diastolic dysfunction is associated with altered myocardial metabolism in asymptomatic normotensive patients with well-controlled type 2 diabetes mellitus. J Am Coll Cardiol. 2003;42:328335.
  • 22
    Drazner MH. The transition from hypertrophy to failure: how certain are we? Circulation. 2005;112:936938.
  • 23
    He ZX, Darcourt J, Camous JP, et al. Correlations of left ventricular diastolic parameters and heart rate: assessment through right ventricular pacing. Eur J Nucl Med. 1992;19:343348.
  • 24
    Lee KJ, Southee AE, Bautovich GJ, et al. Normalised radionuclide measures of left ventricular diastolic function. Eur J Nucl Med. 1989;15:123127.
  • 25
    Maurer MS, El Khoury Rumbarger L, King DL. Ventricular volume and length in hypertensive diastolic heart failure. J Am Soc Echocardiogr. 2005;18:10511057.
  • 26
    Maurer MS, Burkhoff D, Fried LP, et al. Ventricular structure and function in hypertensive participants with heart failure and a normal ejection fraction: the Cardiovascular Health Study. J Am Coll Cardiol. 2007;49:972981.
  • 27
    Federmann M, Hess OM. Differentiation between systolic and diastolic dysfunction. Eur Heart J. 1994;15(Suppl D):26.
  • 28
    Cuocolo A, Storto G, Izzo R, et al. Effects of valsartan on left ventricular diastolic function in patients with mild or moderate essential hypertension: comparison with enalapril. J Hypertens. 1999;17:17591766.
  • 29
    De Teresa E, Gonzalez M, Camacho-Vazquez C, Tabuenca MJ. Effects of bisoprolol on left ventricular hypertrophy in essential hypertension. Cardiovasc Drugs Ther. 1994;8:837843.
  • 30
    Krumholz HM, Larson M, Levy D. Prognosis of left ventricular geometric patterns in the Framingham Heart Study. J Am Coll Cardiol. 1995;25:879884.
  • 31
    Vinereanu D, Nicolaides E, Tweddel AC, Fraser AG. “Pure” diastolic dysfunction is associated with long-axis systolic dysfunction. Implications for the diagnosis and classification of heart failure. Eur J Heart Fail. 2005;7:820828.
  • 32
    Garcia EH, Perna ER, Farias EF, et al. Reduced systolic performance by tissue Doppler in patients with preserved and abnormal ejection fraction: new insights in chronic heart failure. Int J Cardiol. 2006;108:181188.
  • 33
    Imbalzano E, Zito C, Carerj S, et al. Left ventricular function in hypertension: new insight by speckle tracking echocardiography. Echocardiography. 2011;28:649657.
  • 34
    Pedersen F, Raymond I, Madsen LH, et al. Echocardiographic indices of left ventricular diastolic dysfunction in 647 individuals with preserved left ventricular systolic function. Eur J Heart Fail. 2004;6:439447.