European Journal of Heart Failure 10 (2008) 252 259 www.elsevier.

com/locate/ejheart

BNP and NT-proBNP predict echocardiographic severity of diastolic dysfunction

Jasmine Grewal a , Robert McKelvie a , Eva Lonn a , Peter Tait a , Jonas Carlsson d , Monica Gianni e , Christina Jarnert c , Hans Persson b,
a b

Population Health Research Institute and McMaster University, Hamilton, Ontario, Canada Department of Clinical Sciences, Karolinska Institutet, Danderyd Hospital, Stockholm, Sweden c Department of Cardiology, Karolinska University Hospital, Solna, Sweden d Astra Zeneca, R&D, Mlndal, Sweden e Department of Medicine, University of Insubria, Varese, Italy

Downloaded from http://eurjhf.oxfordjournals.org/ by guest on February 7, 2014

Received 1 May 2007; received in revised form 6 November 2007; accepted 28 January 2008

Abstract Aims: To evaluate the best combination of clinical parameters and brain natriuretic peptide (BNP) or N-terminal pro-BNP (NT-proBNP), to predict diastolic dysfunction (DD) in heart failure with preserved left ventricular ejection fraction (HF-PLEF) as determined by Dopplerechocardiography. Methods and Results: HF patients with EF N 40% in the CHARM Echocardiographic Substudy were included and classified to have normal diastolic function, or mild, moderate or severe diastolic dysfunction. Plasma BNP and NT-proBNP levels were measured and relevant clinical characteristics recorded. 181 participants were included in this analysis, 72 (40%) had moderate to severe DD. A model including age, sex, BNP, body mass index, history of atrial fibrillation, coronary artery disease, diabetes mellitus, hypertension and left atrial volume was highly predictive of moderate to severe DD; AUC 0.81 (0.730.88; p b 0.0001). Similarly, substitution of BNP with NT-proBNP resulted in an AUC 0.79 (0.720.87; p b 0.0001). In these models; BNP N 100 pg/ml (OR 6.24 CI 2.4216.09, p = 0.0002), history of diabetes (OR 3.52 CI 1.43 8.70, p = 0.006) and NT-proBNP N 600 pg/ml (OR 5.93 CI 2.2115.92, p = 0.0004), history of diabetes mellitus (OR 2.75 CI 1.126.76, p = 0.03) respectively remained independent predictors of DD in HF-PLEF. Conclusions: Natriuretic peptides were the strongest independent predictors of DD, as determined by Doppler-echocardiography, in HFPLEF. 2008 European Society of Cardiology. Published by Elsevier B.V. All rights reserved.
Keywords: Natriuretic peptides; Diastolic dysfunction; Heart failure; Diagnosis

1. Introduction Heart failure (HF) is a growing worldwide epidemic and is associated with substantial morbidity and mortality. Left ventricular systolic dysfunction is often considered to be the main abnormality in HF. However, up to 50% of patients with HF have a preserved left ventricular ejection fraction
Corresponding author. Department of Cardiology, Danderyd Hospital, SE-182 88 Stockholm, Sweden. Tel.: +46 8 6556849; fax: +46 8 6226810. E-mail address: hans.persson@ds.se (H. Persson).

(HF-PLEF); suggesting that isolated diastolic dysfunction (DD) is the pathophysiological mechanism underlying the clinical syndrome of HF in these patients [1]. Recent data suggest that mortality rates among individuals with HFPLEF are similar to those with HF and systolic dysfunction [2]. Bhatia et al. recently found that among patients with heart failure, the one year mortality and hospitalisation for HF did not differ among those with an EF N 50% vs. EF b 40% [3]. Interestingly, Owan et al. observed that the prevalence of HF-PLEF has increased over a 15 year period, and the rate of death related to this entity has not decreased [4].

1388-9842/$ - see front matter 2008 European Society of Cardiology. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.ejheart.2008.01.017

J. Grewal et al. / European Journal of Heart Failure 10 (2008) 252259

253

This is in contrast to HF secondary to depressed EF, where a decrease in mortality over time was observed. This is likely related to evolving therapies for systolic heart failure, and underscores the need for improved diagnosis and the development of new therapies for HF-PLEF. Furthermore, even in the absence of clinical HF, DD is associated with increased rates of future hospitalizations, development of HF, and allcause mortality [5]. Worsening stages of DD on echocardiography are associated with incremental risk of adverse outcomes including the development of clinical HF [6]. Accurately diagnosing DD could possibly lead to improved treatments and may have substantial health care implications, both from a clinical and resource utilization perspective. Moreover, reproducible, widely applicable and prognostically meaningful approaches to defining DD are important for clinical trials where the use of complex methods to assess diastolic function may be difficult to standardize and implement. In routine clinical practice Doppler echocardiography is the method of choice to diagnose DD [6]. Numerous algorithms have been proposed, most based on transmitral Doppler patterns. However, transmitral Doppler derived indices of diastolic function are dependent on loading conditions, and accurate measurements are operator dependent. Tissue Doppler imaging is a newer technique that can be used in combination with transmitral Doppler to determine the presence and severity of DD [6]. However, this assessment of DD is more complex and requires expert interpretation. Many parameters have been shown to be associated with DD, including echocardiographic measurements, various clinical characteristics, increased left atrial (LA) volume and elevated levels of B-type natriuretic peptide (BNP) and N-terminal (NT)-proBNP [712]. Identifying simple clinical and/or biochemical and/or echocardiographic measurements that can reliably identify the presence and severity of DD is particularly important for patients with HF-PLEF. Therefore, we aimed to determine the best set of clinical parameters, LA volume and brain natriuretic peptide (BNP) or N-terminal pro-BNP (NT-proBNP) that could accurately predict DD, as evaluated by echocardiography. If indeed a simple set of such parameters could be shown to be strongly associated with DD on echocardiography, use of such parameters could help circumvent the need for detailed, difficult, and costly echocardiographic assessments (in situations where echo is not readily available) to determine the presence of prognostically important DD. 2. Methods 2.1. Study design We conducted a cross-sectional study in patients with HFPLEF (EF N 40%), in which LV diastolic function was evaluated using Doppler echocardiography. In addition, simple clinical characteristics were recorded and LA volume and natriuretic peptides were measured. The current report is part

of the multicenter echocardiographic substudy of the international multicenter randomised controlled CHARMPreserved trial [13,14]. 2.2. Study organization All investigators participating in the CHARM-Preserved study were invited to participate in the echocardiographic substudy. Danderyd University Hospital and Hamilton Health Sciences were the Core Laboratories responsible for the protocol, training of sites, and reading study echocardiograms. Echocardiograms were recorded at the investigator's site and shipped to one of the Core Laboratories, with one single reader at each site. Inter-reader variability for 10 diastolic function measurements was assessed in 25 patients between the 2 core laboratories using intra-class correlation coefficients (median ICC 0.784, range 0.6670.954). All NT-proBNP and BNP measurements were performed at the Western Infirmary, Glasgow, Scotland. 2.3. Ethical considerations The echocardiographic substudy was approved by the ethical review boards of all participating centres and all patients provided written informed consent. The study was conducted according to the rules outlined in the Helsinki declaration. 2.4. Patients Patients participating in the CHARM-Preserved study [13] were asked to participate in the echocardiographic substudy (CHARMES) [14]. The inclusion (NYHA Class IIIV, EF N 40%) and exclusion criteria of the substudy were the same as the main study. Additional exclusion criteria for the CHARMES substudy were a poor quality echocardiographic study, the presence of moderate to severe mitral and/or aortic regurgitation, and a prosthetic mitral valve. Those participants in the original CHARMES study who did not have both BNP and NT-proBNP, pulmonary vein and E/ A Valsalva measurements were also excluded from this analysis. 2.5. Measurement of diastolic function 2.5.1. Doppler echocardiography Echocardiographic assessment was defined as the gold standard for the determination of DD in this study and the measurements performed are described in detail in the original CHARMES paper [14]. 2.5.2. Echocardiographic classification of diastolic function The classification of diastolic function on echocardiography was defined a priori using the algorithm outlined in Fig. 1 adapted from Redfield et al [6]. The classification included the following categories: 1) normal; 2) relaxation

Downloaded from http://eurjhf.oxfordjournals.org/ by guest on February 7, 2014

254

J. Grewal et al. / European Journal of Heart Failure 10 (2008) 252259

Downloaded from http://eurjhf.oxfordjournals.org/ by guest on February 7, 2014

Fig. 1. Algorithm for echocardiographic classification of diastolic function (adapted from [6] E, peak early diastolic transmitral flow velocity; A, peak late diastolic transmitral flow velocity; E/A reversal, E/AE/A during valsalva 0.5; AR, pulmonary venous atrial reversal peak flow velocity; Adur, duration of A wave; ARdur, peak pulmonary venous atrial reversal flow velocity duration; S, peak systolic pulmonary venous flow velocity; D, peak diastolic pulmonary venous flow velocity.

abnormality (mild dysfunction); 3) pseudonormal (moderate dysfunction); and 4) restrictive abnormality (severe dysfunction). Two investigators (HP, JG) blinded to patients' clinical characteristics performed this assessment. Relaxation and restrictive abnormalities were assessed by mitral inflow parameters and the classification was based on the E/A abnormality. To distinguish pseudo normal from normal diastolic function, two of the measures outlined in the algorithm had to be abnormal. In patients with atrial fibrillation, deceleration time was used for classification to abnormal relaxation or restrictive diastolic dysfunction, whereas pulmonary systolic/ diastolic peak velocity ratio [15,16] was used to assess for pseudonormal diastolic dysfunction. 2.6. Left atrial volume LA volume was calculated using the area/length method as previously described [17]. The area was traced and length measured in two orthogonal planes, the apical 4 chamber and 2 chamber views, and was indexed to body surface area (LAVI). Abnormal and enlarged LAVI was defined as N 28 ml/m2 [17]. 2.7. NT-proBNP and BNP Blood for NT-proBNP and BNP was obtained at the time of echocardiography. Plasma NT-proBNP was determined using the Elecsys proBNP sandwich immunoassay on an Elecsys 2010 (Roche Diagnostics, Basel, Switzerland). Two cut offs for NT-proBNP were selected before the analyses, 300 pg/ml and 600 pg/ml, with a value of less than 300 pg/ml shown to be optimal for ruling out HF [18]. BNP was

determined using the Shionoria immunoradiometric assay kit [19]. These cut off values determined to be optimal for ruling out HF were selected to test the most conservative association with the presence of HF-PLEF. 2.8. Clinical variables Simple clinical variables were selected for inclusion in the models to assess prediction of severity of DD on echocardiography: age, sex, body mass index (BMI), heart rate, creatinine, medical history of atrial fibrillation, coronary artery disease, diabetes and hypertension. These clinical factors have all been shown to be associated with DD [712,20,21]. 2.9. Statistical analysis For the assessment of clinical parameters associated with echocardiographic DD, diastolic function on echocardiography was grouped into normal/mild DD and moderate/severe DD. This was done as moderate and severe DD were predictive of adverse outcomes (death and hospitalization for HF) in CHARMES [14], while mild DD and normal diastolic function were not. Similarly, other studies have found overall good outcomes in individuals with normal diastolic function and in those with mild DD [6]. The model building process proceeded in three steps. First, we did a univariate screen of the predictor variables to examine their relationship with the outcome. Second, we used best subset selection with Mallow's Cp as the selection criteria, to choose the minimal predicting combination of predictors. This ensured that the impact of each predictor

J. Grewal et al. / European Journal of Heart Failure 10 (2008) 252259

255

was evaluated individually as well as in multivariate interactions with the other predictors. Age and sex were also kept throughout in the models as they were considered to be important control variables on the basis of subject matter. The results of the logistic regression models were reported as odds ratios and 95% confidence intervals (CI). The models were compared in terms of discriminatory ability, using computed Areas under the Receiver Operating Curve (AUC). Third, we used the likelihood ratio testing to evaluate clinically important hypotheses while keeping the results of the second step in mind (model 7 contains all the predictors chosen in the second step). The relationship between continuous predictors was examined with the Pearson correlation coefficient and between a categorical and a continuous predictor with t-test analysis. These analyses were then redone after exclusion of patients with atrial fibrillation. A p -value b 0.05 was considered statistically significant. 3. Results 3.1. Study participants A total of 312 patients were included in the echocardiographic substudy of CHARM. Of these, 131 were excluded, as they did not have both BNP and NT-proBNP measurements and a full echocardiographic study, therefore, 181 patients were included in this analysis. This represents 6% of the patients in the CHARM-Preserved trial [13] and 58% of participants in the original CHARMES substudy [14]. The characteristics of the participants in the normal/mild DD and moderate/severe DD groups are shown in Table 1. Those participants with moderate/severe diastolic dysfunction tended to be older with higher rates of diabetes, coronary artery disease, and atrial fibrillation. The most common aetiologies of HF were ischaemic heart disease, hypertension and idiopathic cardiomyopathy. The proportions of patients with of NYHA Class IIIV HF were similar between normal/mild and moderate/severe DD groups. Participants with moderate/severe DD had higher levels of BNP/NTproBNP and increased LAVI. 3.2. Systolic and diastolic function The echocardiogram was performed 547 days (median) after randomisation into the original CHARM-Preserved trial. There were 56 patients classified as having normal diastolic function on echocardiography, and DD was found in 125 (69%). Mild DD (impaired relaxation) was present in 53 (30%), moderate DD in 55 (30%) and severe DD in 17 (9%) patients. The mean LVEF measured in the study was 55%. Ejection fractions (EF) in the mild, moderate and severe DD groups were 54 8%, 57 10% and 55 9% respectively (p = NS). Similarly, EF was 54 7% in those identified as having normal diastolic function. Therefore, it is unlikely that symptoms could be attributed to systolic dysfunction.

Table 1 Characteristics of HF-PLEF Patients with normal diastolic function or mild diastolic dysfunction vs. those with moderate or severe diastolic dysfunction Characteristic Normal/mild diastolic dysfunction N (%) n = 109 71 38 75 28 (65) (35) (69) (26) Moderate/severe diastolic dysfunction N (%) n = 72 47 25 43 29 (65) (35) (60) (40)

Sex Male Female History of hypertension History of diabetes mellitus History coronary artery disease History myocardial infarction History of atrial fibrillation Atrial fibrillation at time of echo Medications at baseline ACE-Inhibitors Beta Blockers Aspirin Statins Diuretics Aetiology of heart failure Ischaemic heart disease Idiopathic Hypertension Aortic valve disease Excess alcohol intake Atrial fibrillation NYHA class II III IV Characteristic Age (years) Baseline systolic BP Baseline diastolic BP Heart rate (beats/min) BMI (kg/m2) Ejection fraction (%) LA volume index (ml/m2) BNP (pg/ml) NT-proBNP (pg/ml) Creatinine (umol/L)

83 (76) 56 (51) 18 (16) 7 (6)

59 (82) 41 (57) 27 (38) 16 (22)

Downloaded from http://eurjhf.oxfordjournals.org/ by guest on February 7, 2014

21 70 76 68 73

(19) (64) (70) (62) (67)

15 41 51 37 55

(21) (57) (71) (51) (76)

67 (62) 12 (11) 19 (17) 2 (2) 1 (1) 5 (5) 65 (60) 43 (39) 1 (1) Mean (SD) 65 (12) 134.5 (19.4) 77 (11.2) 68 (12) 30 (5) 54 (7) 36 (10) 60 (84) 376 (638) 93 (30)

51 (71) 4 (6) 9 (12) 1 (1) 0 (0) 6 (8) 43 (60) 28 (39) 1 (1) Mean (SD) 70 (10) 134.6 (19.2) 74 (11.2) 67 (13) 29 (6) 57 (8) 44 (13) 165 (227) 1419 (3423) 110 (50)

NYHA = New York Heart Association Class of heart failure, BMI = body mass index, LA = left atrium.

3.3. Predictors of diastolic dysfunction Age and sex adjusted univariate variables associated with moderate/severe DD are shown in Table 2. NT-proBNP N 300 pg/ml, NT-proBNP N 600 pg/ml, BNP N 100 pg/ml, LAVI, history of atrial fibrillation, and diabetes were all significant predictors of moderate/severe DD. NT-proBNP N 600 pg/ml and BNP N 100 pg/ml were the strongest predictors. Creatinine was also significantly associated with moderate/severe DD in univariate analysis and had a modest but significant association with BNP (r = 0.30, p = 0.0004)

256

J. Grewal et al. / European Journal of Heart Failure 10 (2008) 252259 Table 4 Models predicting moderate/severe diastolic dysfunction with BNP N 100 pg/ml Variable Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Age Sex BNP LAVI BMI AF CAD Diabetes HTN OR (95% CI) p Area under ROCcurve (CI) 0.72 (0.640.80) 0.76 (0.670.83) 0.76 (0.680.80) 0.77 (0.690.85) 0.78 (0.710.86) 0.80 (0.720.87) 0.81 (0.730.88) 0.51 0.60 0.0002 0.12 0.16 0.13 0.18 0.006 0.05

Table 2 Univariate predictors of moderate/severe diastolic dysfunction Variable a OR (95% CI) Area under p ROC-curve (CI) .74 (.66.81) .67 (.58.75) .72 (.64.80) .68 (.59.77) .67 (.59.75) .66 (.57.74) .62 (.53.70) .62 (.54.70) .62 (.54.70) .55 (.46.63) .51 (.42.59) .53 (.45.62) .55 (.46.63) .63 (.53.73) b 0.0001 0.01 b 0.0001 0.001 0.001 0.001 0.35 0.21 0.29 0.30 0.79 0.32 0.17 0.004

NT-proBNP N 600 pg/ml 7.4 (3.416) NT-proBNP N 300 pg/ml 2.2 (1.14.4) BNP N 100 pg/ml 4.9 (2.310.4) LAVI N 28 ml/m2 4.4 (1.215.7) History of atrial fibrillation 2.5 (1.25.1) History of diabetes mellitus 2.2 (1.24.8) History of coronary 1.1 (0.52.4) artery disease History of hypertension 1.0 (0.51.9) BMI N 30 kg/m2 0.8 (0.41.4) Heart rate (beats/min) 0.99 (0.961.01) ACE-Inhibitor therapy 1.1 (0.532.3) Beta blocker therapy 0.74 (0.401.4) Diuretic therapy 1.6 (0.83.1) Creatinine (mol/L) b 3.8 (1.410.1)

NT-proBNP = N-terminal pro brain natriuretic peptide, BNP = brain natriuretic peptide, LAVI = left atrial volume index, BMI = body mass index. a Each variable adjusted for age, sex. b For each 100 umol/L increase.

1.02 (0.971.06) 1.25 (0.552.83) 6.24 (2.4216.09) 2.94 (0.7411.72) 0.53 (0.211.29) 2.20 (0.796.14) 2.44 (0.807.46) 3.52 (1.438.70) 0.42 (0.180.98)

Downloaded from http://eurjhf.oxfordjournals.org/ by guest on February 7, 2014

BNP = brain natriuretic peptide N 100 pg/ml, LAVI = left atrial volume index N 28 ml/m2, AF = atrial fibrillation, CAD = coronary artery disease, BMI = body mass index N 30 kg/m2, HTN = hypertension.

and NT-proBNP (r = 0.20, p = 0.02). We did not find an important association of medications or heart rate with BNP (r = 0.07 p = 0.3) or NT-proBNP (r = 0.01 p = 0.9), and as a result these were not included in the final models. The multivariate models (discussed below) were evaluated with and without creatinine (in addition to other listed variables) and yielded similar results. Creatinine was not significantly related to moderate and severe diastolic dysfunction in these complimentary analyses (p = 0.58 and p = 0.68, respectively). Unfortunately, only 132 of the total 181 patients (74%) had a documented creatinine level, as it was not a mandatory test according to the CHARM protocol. Without accounting for the missing creatinine values, we can surmise that the multivariate interactions of creatinine and the other model predictors does not change the important association of natriuretic peptides and diastolic dysfunction as discussed below. All of these analyses were also repeated

after the exclusion of patients with atrial fibrillation, and the results remained the same. Various models were evaluated to determine those most strongly associated with moderate/severe DD as outlined in Table 3. Model 7, which included the most clinically relevant variables, BNP and LAVI was most predictive for moderate/ severe DD on echocardiography, AUC = 0.81 (95% CI, 0.73 to 0.88) (Table 4). Models 6 and 5 had comparable AUCs of 0.80 and 0.78 respectively (Table 4). Likelihood ratio testing showed that model 7 was only marginally more likely to predict the presence of significant DD than model 6 (LR 4.2 df 1, p = 0.04). Among the variables included in model 7, the
Table 5 Models predicting moderate/severe diastolic dysfunction with NT-proBNP Variable Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Age Sex NT-proBNP LAVI BMI AF CAD Diabetes HTN OR (95% CI) p Area under ROCcurve (CI) 0.74 (0.660.81) 0.76 (0.660.84) 0.76 (0.680.84) 0.76 (0.680.84) 0.77 (0.690.85) 0.78 (0.700.86) 0.79 (0.720.87) 0.53 0.94 0.0004 0.08 0.18 0.51 0.39 0.03 0.11

Table 3 Nested models for predicting moderate/severe diastolic dysfunction Model Xs 1 2 3 4 5 6 7 Age, Sex, BNP/NT-proBNP Age, Sex, BNP/NT-proBNP, LAVI Age, Sex, BNP/NT-proBNP, LAVI, AF Age, Sex, BNP/NT-proBNP, LAVI, AF, Diabetes Age, Sex, BNP/NT-proBNP, LAVI, AF, Diabetes, CAD Age, Sex, BNP/NT-proBNP, LAVI, AF, Diabetes, CAD, BMI Age, Sex, BNP/NT-proBNP, LAVI, AF, Diabetes, CAD, BMI, HTN

BNP = brain natriuretic peptide N 100 pg/ml, NT-proBNP = N terminal pro brain natriuretic peptide N 600 pg/ml, LAVI = left atrial volume index N 28 ml/m2, AF = atrial fibrillation, CAD = coronary artery disease, BMI = body mass index N 30 kg/ m2, HTN = hypertension.

1.01 (0.971.06) 1.03 (0.462.32) 5.93 (2.2115.92) 3.32 (0.8512.95) 0.54 (0.221.32) 1.43 (0.494.16) 1.60 (0.544.72) 2.75 (1.126.76) 0.50 (0.221.32)

NT-proBNP = N-terminal pro brain natriuretic peptide N 600 pg/ml, LAVI = left atrial volume index N 28 ml/m2, AF = atrial fibrillation, CAD = coronary artery disease, BMI = body mass index N 30 kg/m2, HTN = hypertension.

J. Grewal et al. / European Journal of Heart Failure 10 (2008) 252259

257

only independent predictors of moderate/severe DD were BNP N 100 pg/ml (OR 6.24 CI 2.4216.09, p = 0.0002) and a history of diabetes (OR 3.52 CI 1.438.70, p = 0.0006). Models with NT-proBNP N 600 pg/ml were also examined (Table 5). Models with NT-proBNP N 300 pg/ml were also examined but the results are not shown since they were found to be less predictive. In Model 7, which included the most clinically relevant variables, NT-proBNP and LAVI was most predictive of moderate/severe DD, AUC of 0.79 (Table 5). The strongest independent predictors of moderate/ severe DD in this model were NT-proBNP N 600 pg/ml (OR 5.93 CI 2.2115.92, p = 0.0004) and a history of diabetes (OR 2.75 CI 1.126.76, p = 0.03). The other models had comparable AUCs (Table 5) and likelihood ratio testing implied that model 7 was no more predictive of DD than the other simpler models (results not shown). 4. Discussion Our study is the first to identify the best combination of easily measured clinical parameters associated with clinically important echocardiographic DD in the setting of HFPLEF. Our results demonstrate that high levels of natriuretic peptides are the most powerful in diagnosing significant DD in the setting of HF-PLEF as compared to other clinical variables and LAVI. Natriuretic peptides can therefore be used to provide objective evidence of prognostically important DD in HF-PLEF. This would be especially useful in clinical settings where detailed and complex echocardiographic assessments are not possible and in clinical trials, where objective, standardized and reproducible identification of prognostically important DD is needed. Our findings are certainly complimentary to a recent consensus statement published by Paulus et al [22]. This group nicely outlined an algorithm for the diagnosis of HF with normal ejection fraction. The emphasis is on using a combination of elevated filling pressures as determined by tissue Doppler echocardiography and natriuretic peptides (BNP N 200 pg/ml and NT-proBNP N 220 pg/ml). If one or the other is not available, then the physician is advised to look at mitral inflow velocity parameters, pulmonary vein flow parameters, left ventricular mass index, left atrial volume or the presence of atrial fibrillation. Our approach was a little different as we set out to determine the best clinical predictors of HF-PLEF, in an effort to provide an approach for the many medical centres where detailed echocardiographic assessment of diastolic dysfunction is not possible. Like Paulus et al, we used fairly conservative cut-off points for BNP (values designated as being most sensitive rather than specific, as per the diagnosis of systolic heart failure) in determining the association with HF-PLEF, and still found that natriuretic peptides were most valuable in predicting moderate/severe DD in HF-PLEF. Similarly, other studies have also used more sensitive values for the diagnosis of HF-PLEF. For example, in a study of patients with normal systolic function on echocardiogram, plasma BNP 57 pg/mL detected the 28

patients with isolated abnormal diastolic function with 100% positive predictive value [23]. Another study which evaluated 294 patients with normal left ventricular function, reported that plasma BNP 62 pg/mL had a sensitivity of 85% and a specificity of 83% for the diagnosis of diastolic dysfunction [24]. We have confirmed the importance of natriuretic peptides i0n the diagnosis of HF-PLEF and have shown that the specificity can be increased with evaluation of other parameters. The recommendations by Paulus et al. suggest that increased left atrial volumes, atrial fibrillation or increased left ventricular mass in addition to natriuretic peptides would increase the likelihood of HF-PLEF. Our results suggest that the presence of diabetes would also increase the likelihood of HF-PLEF. These findings should prove useful for clinicians in the diagnosis of prognostically important HF-PLEF. Some other important observations have also been made in this group of patients with HF-PLEF. Progressive NYHA Class symptoms were present in a large portion of patients with mild diastolic dysfunction. This was independent of ejection fraction which remained preserved across the spectrum of diastolic function. Also, ejection fraction was similar in those with NYHA Class II, III, and IV symptoms (55 9%, 54 8%, 57 4% respectively). This observation may be analogous to the presence of advanced heart failure symptoms that seem to be out of proportion to the degree of LV depression in low EF HF. Also, in contrast to the low EF HF population, this group of patients with HF-PLEF had higher BMI measurements. A lower BMI in systolic HF studies may be attributed to cardiac cachexia occurring as a result of advanced (NYHA IV) symptoms and more importantly reduced cardiac output. The prevalence of advanced symptoms, and then presumably cardiac cachexia, was approximately 2% in the overall CHARM cohort and 1% in our analysis, less than that seen in other HF studies. We also considered that patients with an increased BMI and symptoms of dyspnoea were incorrectly diagnosed as having HF, resulting in an increase in BMI in the study group. This is a concern, however, we could not find an association between high BMI and normal and mild diastolic dysfunction (Table 2). We demonstrated that both BNP and NT-proBNP are elevated in individuals with moderate and severe DD as compared to normal and mild DD. BNP N 100 pg/ml and NT-pro BNP N 600 pg/ml alone are both strongly associated with clinically important DD. These results are consistent with the known pathophysiology of HF-PLEF where a variety of clinical conditions can ultimately lead to an increased left ventricular end diastolic wall stress. Iwanaga et al. have demonstrated that BNP levels correlate very closely with left ventricular end diastolic wall stress in the setting of HFPLEF (r2 = 0.887) making it a good surrogate marker of worsening DD [25]. Moreover, BNP was shown to strongly correlate with pulsed wave Doppler and tissue Doppler parameters in patients with HF-PLEF [26]. Similarly, LAVI values were larger in the moderate/severe DD group and

Downloaded from http://eurjhf.oxfordjournals.org/ by guest on February 7, 2014

258

J. Grewal et al. / European Journal of Heart Failure 10 (2008) 252259

strongly predicted DD in univariate analysis. LAVI is increasingly recognized as a relatively load-independent marker of LV filling pressures and has been shown to increase with worsening DD [27] and correlates closely with NT-proBNP [28]. Natriuretic peptides and LAVI both reflect increases in wall stress. However, our study demonstrates that BNP N 100 pg/ml and NT-proBNP levels N 600 pg/ml were more strongly associated with DD in HF-PLEF than LAVI. Several previous studies have demonstrated the utility of both BNP and NT-proBNP used in isolation in predicting DD in HF-PLEF, but have not evaluated the use of models including easily measured clinical variables and natriuretic peptides [710]. Our study shows that a history of diabetes is associated with HF-PLEF. This may be related to impairment of myocardial microcirculation, reduction of coronary flow reserve, hyperglycaemia, poor metabolic control, endothelial dysfunction and myocardial fibrosis. Several previous studies have shown LV diastolic filling abnormalities in people with diabetes, thought to be related to LV remodelling and hypertrophy independent of other concomitant risk factors, such as hypertension and epicardial coronary artery disease [29,30]. Among these, the Framingham study found increased rates of HF in people with diabetes and suggested that this association was independent of other CHD risk factors [30]. 5. Study limitations The echocardiography protocol used in our study did not include tissue Doppler imaging for determining the degree of DD due to the multicenter design of the trial and inherent variabilities in tissue Doppler assessment. However, in a previous study using a similar combination of mitral inflow and pulmonary vein Doppler measurements, 93% of patients with HF-PLEF were found to show evidence of DD [31], making us confident of the classification of DD for the purposes of our paper. We included patients in atrial fibrillation at the time of the echocardiogram. We used previously published criteria for assessment of DD in atrial fibrillation; moreover there were few patients in atrial fibrillation at the time of echocardiography. Furthermore, our results were similar when participants with atrial fibrillation were excluded. 6. Conclusions HF-PLEF is a common entity and is most commonly attributed to DD. Advanced degrees of DD are associated with worse outcomes. Currently, the presence and severity of DD is generally ascertained with the use of complex echocardiographic methods using multiple often challenging measurements. These are difficult to standardize and perform reproducibly in routine clinical practice. In such clinical settings, measuring BNP or NT-proBNP, using available standardized assays, provides an alternate simple method of

identifying DD and grading its severity. Combining the measure of BNP or NT-proBNP with a simple clinical parameter such as a history of diabetes mellitus can further help in predicting moderate/severe DD in HF-PLEF. Acknowledgments We gratefully acknowledge the contributing 48 sites in Canada, Iceland, Malaysia, Russia, Sweden and USA, the core laboratory teams, and the Executive Committee of the CHARM program for important contributions to the study. References
[1] Cohen-Solal A, Desnos M, Delahaye F, Emeriau JP, Hanania G. A national survey of heart failure in French hospitals. Eur Heart J 2000;21: 7639. [2] Bursi F, Weston SA, Redfield MM, et al. Systolic and diastolic heart failure in the community. JAMA 2006;296:220916. [3] Bhatia RS, Tu JV, Lee DS, et al. Outcome of heart failure and preserved ejection fraction in a population-based study. NEJM 2006;355:2609. [4] Owan TE, Hodge DO, Herges RM, Jacobsen SJ, Roger VL, Redfield MM. Trends in prevalence and outcome of heart failure with preserved ejection fraction. NEJM 2006;355:2519. [5] Tsutsui H, Tsuchihashi M, Takeshita A. Mortality and readmission of hospitalized patients with congestive heart failure and preserved versus depressed systolic function. Am J Cardiol 2001;88:5303. [6] Redfield M, Jacobsen S, Burnett J, Mahoney D, Bailey K, Rodeheffer R. Burden of systolic and diastolic ventricular dysfunction in the community. JAMA 2003;289:194202. [7] Dahlstrom U. Can natriuretic peptides be used for the diagnosis of diastolic heart failure? Eur J Heart Fail 2004;6:2817. [8] Qi W, Kjekshus J, Klinge R, Kjekshus JK, Hall C. Cardiac natriuretic peptides and continuously monitored atrial pressures during chronic rapid pacing in pigs. Acta Physiol Scand 2000;169:95102. [9] Dokainish H, Zoghbi WA, Lakkis NM, et al. Optimal noninvasive assessment of left ventricular filling pressures: a comparison of tissue Doppler echocardiography and B-type natriuretic peptide in patients with pulmonary artery catheters. Circulation 2004;109:24329. [10] Lubien E, DeMaria A, Krishnaswamy P, et al. Utility of B-natriuretic peptide in detecting diastolic dysfunction: comparison with Doppler velocity recordings. Circulation 2002;105:595601. [11] Tsang TS, Barnes ME, Gersh BJ, Bailey KR, Seward JB. Left atrial volume as a morphophysiologic expression of left ventricular diastolic dysfunction and relation to cardiovascular risk burden. Am J Cardiol 2002;90:12849. [12] Tsang TS, Gersh BJ, Appleton CP, et al. Left ventricular diastolic dysfunction as a predictor of the first diagnosed nonvalvular atrial fibrillation in 840 elderly men and women. J Am Coll Cardiol 2002;40:163644. [13] Yusuf S, Pfeffer MA, Swedberg K, et al. Effects of candesartan in patients with chronic heart failure and preserved left-ventricular ejection fraction: the CHARM-Preserved Trial. Lancet 2003;362:77781. [14] Persson H, Lonn E, Magnus E, et al. Diastolic dysfunction in heart failure with preserved systolic functionneed for objective evidence. Results from the CHARM Echocardiographic Substudy CHARMES. J Am Coll Cardiol 2007;49:68794. [15] Oki T, Tabata T, Yamada H. Evaluation of left atrial filling using systolic pulmonary venous flow measurements in patients with atrial fibrillation. Clin Cardiol 1998;21:16974. [16] Traversi E, Cobelli F, Pozzoli M. Doppler echocardiography reliably predicts pulmonary artery wedge pressure in patients with chronic heart failure even when atrial fibrillation is present. Eur J Heart Fail 2001;3:17381.

Downloaded from http://eurjhf.oxfordjournals.org/ by guest on February 7, 2014

J. Grewal et al. / European Journal of Heart Failure 10 (2008) 252259 [17] Lang RM, Bierig M, Devereaux RB, et al. Members of the Chamber Quantification Writing Group Recommendations for chamber quantification: a report from the American Society of Echocardiography's Guidelines and Standards Committee and the Chamber Quantification Writing Group, developed in conjunction with the European Society of Echocardiography, a branch of the European Society of Cardiology. J Am Soc Echocardiogr 2005;18:144063. [18] Januzzi JL, Camargo CA, Anwaruddin S, Baggish AL, Chen AA. The N-terminal Pro-BNP investigation of dyspnea in the emergency department (PRIDE) study. Am J Cardiol 2005;95:94854. [19] Tsutamoto T, Wada A, Maeda K, et al. Attenuation of compensation of endogenous cardiac natriuretic peptide system in chronic heart failure: prognostic role of plasma brain natriuretic peptide concentration in patients with chronic symptomatic left ventricular dysfunction. Circulation 1997;96:50916. [20] Celentano A, Vaccaro O, Tammaro P, Galderisi M, Marina OM. Early abnormalities in non-insulin-dependent diabetes mellitus and impaired glucose tolerance. Am J Cardiol 1995;76:11736. [21] Muller-Brunotte R, Kahan T, Malmqvist K, Edner M. Blood pressure and left ventricular geometric pattern determine diastolic function in hypertensive myocardial hypertrophy. J Hum Hypertens 2003;17:8419. [22] Paulus WJ, Tschpe C, Sanderson JE, et al. How to diagnose diastolic heart failure: a consensus statement on the diagnosis of heart failure with normal ejection fraction by the heart failure and echocardiography associations of the European society of cardiology. Eur Heart J 2007;28:253950. [23] Maisel AS, Koon J, Krishnaswamy P, et al. Utility of B-natriuretic peptide as a rapid, point-of-care test for screening patients undergoing

259

[24]

[25]

[26]

[27]

[28]

[29]

[30] [31]

echocardiography to determine left ventricular dysfunction. Am Heart J 2001;141(3):36774. Lubien E, DeMaria A, Krishnaswamy P, et al. Utility of B-natriuretic peptide in detecting diastolic dysfunction: comparison with Doppler velocity recordings. Circulation 2002;105(5):595601. Iwanaga Y, Nishi I, Furuichi S, et al. B-type natriuretic peptide strongly reflects diastolic wall stress in patients with chronic heart failure. J Am Coll Cardiol 2006;47:7428. Dokainish H, Zoghbi W, Lakkis N, Quinones M, Nagueh S. Comparative accuracy of B-type natriuretic peptide and tissue Doppler echocardiography in the diagnosis of congestive heart failure. Am J Cardiol 2004;93:11305. Pritchett A, Mahoney D, Jacobsen S, Rodeheffer R, Karon B, Redfield M. Diastolic dysfunction and left atrial volume. J Am Coll Cardiol 2005;45:8792. Lim TK, Ashrafian H, Dwivedi G, Collinson PO, Senior R. Increased left atrial volume is an independent predictor of raised serum natriuretic peptide in patients with suspected heart failure but normal left ventricular ejection fraction: Implication for diagnosis of diastolic heart failure. Eur J Heart Fail 2006;8:3845. Devereux R, Roman M, Paranicas M. Impact of diabetes on cardiac structure and function: the Strong Heart Study. Circulation 2000;101: 22716. Kannel WB, Hjortland MC, William P. Role of diabetes in congestive heart failure: the Framingham study. Am J Cardiol 1974;34:2934. Vasan RS, Benjamin EJ. Diastolic heart failure no time to relax. N Engl J Med 2001;344:569.

Downloaded from http://eurjhf.oxfordjournals.org/ by guest on February 7, 2014