Author + information
- Received November 6, 2012
- Revision received January 14, 2013
- Accepted February 5, 2013
- Published online May 7, 2013.
- Thomas Weber, MD⁎,†,⁎ (, )
- Siegfried Wassertheurer, DI‡,§,
- Michael F. O'Rourke, MD∥,
- Anton Haiden, MD⁎,
- Robert Zweiker, MD¶,
- Martin Rammer, MD⁎,
- Bernhard Hametner, DI‡ and
- Bernd Eber, MD⁎
- ↵⁎Reprint requests and correspondence:
Dr. Thomas Weber, Cardiology Department, Klinikum Wels-Grieskirchen, Grieskirchnerstrasse 42, 4600 Wels, Austria
Objectives This study sought to test whether measures of pulsatile arterial function are useful for diagnosing heart failure with preserved ejection fraction (HFPEF), in comparison with and in addition to tissue Doppler echocardiography (TDE).
Background Increased arterial stiffness and wave reflections are present in most patients with HFPEF.
Methods Patients with dyspnea as a major symptom were categorized as having HFPEF or no HFPEF, based on invasively derived filling pressures and natriuretic peptide levels. Pulse wave velocity (PWV) was measured invasively (aortic PWV). Aortic pulse pressure (aoPP) and its components (incident pressure wave height, forward wave amplitude; augmented pressure; backward wave amplitude [Pb]) were quantified noninvasively.
Results Seventy-one patients were classified as HFPEF and 65 as no HFPEF (223 patients had intermediate results). Patients with HFPEF were older, more often had hypertension and diabetes, and had larger left atria and higher left ventricular mass. Brachial pulse pressure (bPP), aoPP, and all measures of arterial stiffness and wave reflections were higher in HFPEF patients. Receiver-operating curve analysis–derived area under the curve (AUC) values for separating HFPEF from no HFPEF were 0.823 for E/E′ at the medial annulus, the best TDE parameter; 0.816 for bPP; and 0.867, 0.851, and 0.825 for aortic PWV, aoPP, and Pb, respectively. Adding measures of pulsatile function to TDE resulted in an increase in AUC to 0.875 (bPP; p = 0.03) and 0.901 (aoPP; p = 0.005). In comparison with a TDE-based algorithm, net reclassification improvement was 32.9% (p < 0.0001).
Conclusions Measures of pulsatile arterial hemodynamics may complement TDE for the diagnosis of HFPEF. (Pulsatile and Steady State Hemodynamics in Diastolic Heart Failure; NCT00720525)
- arterial stiffness
- arterial wave reflections
- exertional dyspnea
- heart failure with preserved ejection fraction
- pulsatile hemodynamics
An increase in the pulsatile components of blood pressure (systolic blood pressure [SBP], pulse pressure [PP]) is a major risk factor for developing heart failure in general (1) and heart failure with preserved ejection fraction (HFPEF) in particular (2). An impairment of arterial pulsatile function in HFPEF patients has been described repeatedly (3–5). Mechanistic studies have identified a relationship between diastolic dysfunction and increased arterial stiffening or increased arterial wave reflections in different populations (6–8). At present, HFPEF is usually described by echocardiogram on the basis of impaired left ventricular filling and increased filling pressures (9,10). However, measurement of blood pressure or arterial properties is not recommended in recent guidelines of major cardiovascular societies on the evaluation of diastolic dysfunction and on the diagnosis of HFPEF (9,10).
In this trial, we evaluated whether simple (brachial pulse pressure [bPP]) or more precise measures of pulsatile arterial function can be used for diagnosis of HFPEF in patients with dyspnea on exertion and preserved systolic function. All patients underwent invasive cardiac examination, giving us the opportunity to base our diagnosis upon invasively determined left ventricular end-diastolic pressures (LVEDPs), supported by plasma levels of N-terminal pro–B-type natriuretic peptides (NT-proBNP) (10). Therefore, we were able to test the diagnostic properties of markers of pulsatile arterial function against and in combination with commonly used, recommended echocardiographic parameters (9,10).
We prospectively included patients undergoing cardiac catheterization for suspected coronary artery disease (CAD) at our institutions, presenting with a major symptom of dyspnea on exertion, if left ventricular ejection fraction was normal (>50%) (10). Exclusion criteria were unstable clinical conditions, rhythm other than stable sinus rhythm or stable paced rhythm, more than mild valvular disease, pericardial constriction, and primary pulmonary hypertension. The study was conducted according to the Declaration of Helsinki, approved by our regional ethics committees, and all patients gave written informed consent. The trial was registered at ClinicalTrials.gov (NCT00720525).
Hypertension was present with repeated measurements ≥140 mm Hg SBP and/or ≥90 mm Hg diastolic blood pressure or permanent antihypertensive drug treatment. Diabetes mellitus was defined as a fasting blood glucose concentration >7 mmol/l or antihyperglycemic drug treatment. Creatinine clearance was estimated using the Cockcroft-Gault formula. For this study, we defined significant CAD as at least 1 diameter stenosis 50% or greater in at least 1 coronary vessel, or prior coronary revascularization. The extent of CAD was defined using a modified scoring system (“angioscore”) (11). Pulmonary functional testing was a regular part of the diagnostic workup of our patients, as well as chest x-ray.
Diagnosis of HFPEF
On the basis of LVEDP and plasma levels of NT-proBNP, with cutoff levels according to recently published recommendations (10) and cardiology textbooks (12), we categorized patients as having HFPEF with LVEDP >16 mm Hg, and NT-proBNP levels >220 pg/ml, the HFPEF group. We excluded the condition with LVEDP ≤12 mm Hg and NT-proBNP levels <120 pg/ml, the no HFPEF group. All other patients fell into the intermediate category, which we labeled possible HFPEF.
LVEDP was measured automatically using the built-in software of our coronary angiography systems (Siemens Artis Zee with AXIOM Sensis hemodynamic recording system, Siemens Healthcare, Erlangen, Germany, in Wels, Austria, and Philips Allura Xper FD 10/10, Philips, Eindhoven, the Netherlands, in Graz, Austria) during cardiac catheterization before contrast cineangiography, using 6-F fluid-filled pigtail catheters. All pressure tracings were visually inspected for compliance with the standard definition of LVEDP, i.e., the pressure after the “A” wave at the onset of left ventricular isovolumetric contraction, coincident with the electrocardiogram's R-wave (13). We carefully performed flushing and zeroing, and tried to avoid artifacts. In 21 patients, we compared LVEDPs simultaneously measured with our fluid-filled system and with high-fidelity sensor-tip pressure catheters (5-F Millar SPC-454D, Millar Instruments, Houston, Texas), which were adjusted to baseline electronically, calibrated under saline, connected to a Millar PCU-2000 unit. Values were in excellent agreement, with a mean difference of 0.5 ± 3.2 mm Hg (Online Fig. 1). In addition, reproducibility of LVEDP was good. In 20 patients, mean difference between repeated measurements was 0.5 ± 2.1 mm Hg (Online Fig. 2).
Plasma levels of NT-proBNP were measured, using the commercially available electrochemiluminescence immunoassay “ECLIA” on the Elecsys 1020 analyzer (Roche Diagnostics, Mannheim, Germany).
Arterial pulsatile function
Brachial blood pressure was measured with a validated (14), automated, oscillometric, sphygmomanometer (Omron M5-I, Omron Healthcare, Kyoto, Japan). Estimation of central (aortic) blood pressures and arterial wave reflections was performed noninvasively with the SphygmoCor system Version 9 (AtCor Medical, Sydney, Australia), using the technique of pulse waveform analysis (11) by nurses not involved in other aspects of the study. In addition, the absolute heights of the forward and the backward pressure waves were quantified, using wave separation analysis with the recently developed and validated ARCSolver method (15). Arterial stiffness was assessed invasively (aortic pulse wave velocity [aoPWV]), and noninvasively (carotid–femoral pulse wave velocity). A detailed description of our methods to assess pulsatile arterial function is available in the Online Appendix.
A detailed 2-dimensional and Doppler echocardiogram according to the recommendations of the American Society of Echocardiography (16) was obtained in all patients immediately before or after measurement of arterial stiffness/wave reflections, using a Philips iE33 (Philips) or a Vivid 7 (GE Healthcare, Waukesha, Wisconsin) machine in Wels and Graz, respectively. For pulsed wave tissue Doppler imaging, the sample volume was located at the medial and at the lateral border of the mitral annulus in the apical 4-chamber view, where we obtained early diastolic mitral annulus velocity (E′), late diastolic velocity (A′), and peak systolic velocity (S′).
Data are presented as mean ± SD if normally distributed, median (lower and upper quartile) if non-normally distributed, and as numbers (percentages) if categorical. Continuous and categorical data were compared across all 3 diagnostic groups, using Kruskal-Wallis analysis of variance and the chi-square test, respectively. In order to perform the comparison between pulsatile hemodynamics and echocardiography in clear-cut diagnostic groups, the main analysis was carried out in patients with established diagnosis (HFPEF) and established exclusion (no HFPEF) of HFPEF: we constructed receiver-operating characteristic (ROC) curves to illustrate the diagnostic performance of measures of pulsatile arterial function and of echocardiographic parameters and to compare both methods, using the area under the curve (AUC) and z-statistic. Because this approach may not represent clinical routine ideally (where we have to deal with intermediate cases as well), we performed additional analyses, testing the combination of no HFPEF and possible HFPEF groups against HFPEF patients (for diagnosis of HFPEF) and the combination of HFPEF and possible HFPEF groups against no HFPEF patients (for exclusion of HFPEF). Because possible HFPEF patients actually may represent suffering from a milder course (or an earlier stage) of HFPEF, we also performed multiple linear regression analysis, coding no HFPEF as 0, possible HFPEF as 1, and HFPEF as 2.
To determine the most effective combination of diagnostic tests to predict the presence or absence of HFPEF, we applied multiple stepwise logistic regression with the use of a p value of 0.1 or less for entry into the model and a p value of 0.2 or higher for removal from the model. The predicted and the actually observed values for the presence/absence of HFPEF were cross-classified, and the percentage of cases correctly classified was obtained. The AUC's for uni-, bi-, and multivariable models to predict HFPEF were compared, using the z-statistic.
Net reclassification improvement (NRI), as compared with echocardiography, as recommended by recent guidelines (10), was calculated according to the method of Pencina et al. (17). In addition, to provide a potentially usable classification, we applied a stepwise approach to determine cutoff values leading to 95% sensitivity and specificity for each parameter and parameter set, respectively. Starting (according to the guidelines) with E/E′ at the medial annulus (E/E′med) as the initial marker, we modified the upper and lower limits incrementally, followed by a recount, until we achieved the supposed sensitivity and specificity. Subjects remaining between the upper and lower limits had then been reclassified using various pulsatile measures, following the same procedure. This was done for any parameter combination as well as any parameter order, because a modified order leads to different cutoffs. The huge number of calculations and comparisons was performed by customized software. Statistical analysis was performed using MedCalc software version 11.6.0 (MedCalc Comp, Mariakerke, Belgium), and Matlab 7.8.0 (MathWorks, Natick, Massachusetts).
Mean patient age was 63.4 years, 70.7% were men, and 83.5% patients had hypertension, 23.7% diabetes, and 49.4% CAD. Pulmonary comorbidities were present in 32.1%; 93.8% of our patients were prospectively categorized as New York Heart Association (NYHA) functional class II, and 6.2% as NYHA functional class III, with no significant differences among our diagnostic groups (Table 1,Online Table 1).
Applying our diagnostic criteria, 71 patients were diagnosed with HFPEF, in 65 patients we excluded the condition, and 223 patients fell into the intermediate category (possible HFPEF).
We observed a graded increase of age, presence of hypertension, number of antihypertensive drugs used, and extent of CAD across our diagnostic groups, with the highest values observed in the HFPEF group.
We found a statistically highly significant, graded increase of brachial SBP, mean blood pressure, and PP across diagnostic groups, with the highest values observed in the HFPEF group, which was even more pronounced when blood pressures were measured invasively during cardiac catheterization. In the subgroup of patients undergoing right heart catheterization (n = 129), we found the same differences for right-sided pressures and total pulmonary resistance. Heart rates as well as cardiac output did not differ across groups.
All parameters of pulsatile arterial function showed the same statistically highly significant increase across diagnostic groups. In addition, the duration of ejection was prolonged in HFPEF.
Again, we observed a statistically highly significant distribution of measurements: patients with HFPEF had higher values for wall thickness, left ventricular mass, left atrial dimensions, and estimated filling pressures (E/E′) and lower values for tissue Doppler echocardiography (TDE) measurements, as compared with patients in the no HFPEF group. Patients with possible HFPEF fell into the intermediate category.
Diagnostic performance—ROC analysis
The AUCs in ROC analysis, when measures of arterial function and echocardiographic parameters were used to differentiate HFPEF from no HFPEF, are shown in Table 2 and Online Table 2. The best diagnostic performance of an echocardiographic parameter was observed for E/E′med with an AUC of 0.823. The AUC for bPP, which is readily available even in primary care, was 0.816 (p = 0.88 for comparison with E/E′med). Measures of central arterial function, including the antegrade pressure wave (incident pressure wave height [P1], amplitude of the forward wave [Pf]), the reflected pressure wave (augmented pressure [AP], amplitude of the backward wave [Pb]), their combination (aortic PP), and aortic stiffness (aoPWV), had diagnostic capacities in the same range (Fig. 1). The highest AUC was found for aoPWV (0.867); the highest AUC of a noninvasive measurement was found for aoPP (0.851). According to the z-statistic, the AUCs for the arterial parameters were not statistically significantly different from E/E′med, with p values >0.20 for all, respectively.
When we investigated clinically relevant subgroups, we observed a similar diagnostic performance of E/E′ and the arterial parameters in men and women, patients with and without CAD, and patients up to and above 62 years of age (Table 2, Online Table 2). When we excluded patients from the analysis in whom coronary revascularization was performed (leaving 56 patients in the HFPEF and 56 patients in the no HFPEF-group), the diagnostic performance of measures of pulsatile function and of echocardiographic parameters was even better, but still comparable (AUCs for aoPP and E/E′med were 0.863 and 0.867, respectively) (Online Table 3).
In exploratory analysis, when HFPEF was diagnosed or excluded only on the basis of LVEDP, AUCs were somewhat lower in general, but measures of pulsatile arterial function remained as good as or even better (aoPP) than echocardiographic parameters for the differentiation between HFPEF and no HFPEF (Online Tables 4 and 5).
When we included the possible HFPEF group and tested the diagnostic value of the various measures to diagnose HFPEF (HFPEF group versus the combination of the possible HFPEF and no HFPEF groups), AUCs were lower for all parameters, but measures of pulsatile arterial function and echocardiographic parameters again did not differ significantly (Online Table 6). The same was true when we included the possible HFPEF group and tested the diagnostic value of the various measures to exclude HFPEF (no HFPEF group versus the combination of the possible HFPEF and HFPEF groups) (Online Table 7).
Applying multivariable logistic regression analysis with HFPEF or no HFPEF as outcome variable, E/E′med correctly classified 77% of patients. In comparison, bPP correctly classified 75.7% of patients. Values for central arterial parameters were in the same range (Fig. 2, Table 3). Adding 1 parameter of arterial function to the model with E/E′med improved model fit, up to 86.5% of patients could be correctly identified, and AUCs were significantly higher (Figs. 1 and 2, Online Table 8). bPP in addition to E/E′med significantly increased the AUC (p = 0.03), but was inferior to the combination of E/E′med and aoPP (p = 0.009 for comparison). In the final models, including clinical parameters, medications, echocardiographic measures, and 1 parameter of pulsatile function, up to 90.1% of patients could be correctly classified. Selecting aoPP or Pb as parameters of pulsatile function led to the highest AUCs (Fig. 2, Table 4,Online Table 9).
In multiple linear regression models including possible HFPEF patients (with intermediate coding, see the previous text), pulsatile hemodynamics significantly predicted the degree of HFPEF (Online Table 10).
Incremental diagnostic value—NRI
When we used all echocardiographic parameters with cutoff values as suggested in recent guidelines (10) in the subset of 58 HFPEF patients and 57 no HFPEF patients in whom complete datasets including left ventricular mass and left atrial volumes were available, 42 patients were classified correctly, 13 incorrectly, and 60 patients remained unclassified. The combination with 1 measure of pulsatile arterial function led to a highly significant NRI of up to 32.9 % for noninvasive parameters (aoPP) (Table 5).
Using only E/E′med as a first step to diagnose or exclude HFNEF, as suggested in recently published guidelines (10), and the combination with 2 parameters of pulsatile arterial function led to similar results and highly significant NRIs (Online Tables 11 to 13, Online Fig. 3).
In this study, we observed in a cohort of middle-age and elderly patients presenting as clinically stable, without overt congestion, with a major symptom of exertional dyspnea, that pulsatile arterial hemodynamics are abnormal in those with HFPEF. The principle findings were: 1) that noninvasively measured indexes of arterial function may provide a clue for the diagnosis of HFPEF with equal precision to echocardiography and TDE; and 2) when performed in combination with echocardiography and TDE, indexes may improve accuracy of diagnosis.
We based our diagnosis on invasively derived left ventricular filling pressures with cutoff values according to generally accepted guidelines (10) and major textbooks (12), in combination with plasma levels of natriuretic peptides (10). This approach allowed us to investigate the diagnostic performance of pulsatile hemodynamics independently of echocardiographic measures and to compare pulsatile hemodynamics and echocardiography against each other and in combination. Of note, although echocardiography was not part of the diagnostic algorithm, we observed typical structural (left atrial enlargement, left ventricular hypertrophy) and functional (mitral inflow, TDE patterns, right-sided pressures) abnormalities in our HFPEF patients, strongly supporting the internal validity of our data and the assumption that the patients' symptoms were due to heart failure. In the absence of overt congestion, otherwise well-validated epidemiological criteria for diagnosing heart failure were less useful in our population with early/less severe heart failure, as these criteria mainly depend on clinical signs of volume overload.
We did not exclude patients with pulmonary disease, because this can be an important comorbidity in heart failure patients (18), a fact that is indeed confirmed by our data. We also did not exclude patients with CAD from our analysis, because HFPEF and CAD frequently coexist (19). The diagnostic value of pulsatile hemodynamics in our study was not substantially different in patients with and without CAD, and the extent of CAD (angioscore) did not reach statistical significance in the multivariable models to predict HFPEF. In addition, 16.2% of our patients (statistically not different among our diagnostic groups) underwent coronary revascularization as a consequence of diagnostic testing and symptoms (dyspnea was taken as angina-equivalent, or typical angina was present in addition to dyspnea). When we excluded these patients from the analysis, the diagnostic performance of measures of pulsatile function was even improved, corroborating our main findings, but also suggesting that treatable causes of exertional dyspnea (in this case: myocardial ischemia) should be excluded before labeling patients as primarily HFPEF.
Pathophysiological mechanisms underlying our findings are known, at least to some extent: diastolic function is inversely related to arterial stiffness (PWV) and wave reflection (AIx) (6,7). Against this background, various groups have reported increased arterial stiffness (8,20,21), decreased aortic distensibility (4), and increased wave reflections (5,6) in HFPEF patients. Recently, increased wave reflections were strong, independent predictors of new-onset heart failure in a large, population-based study (22). This may be related to unfavorable myocardial–arterial coupling with deleterious consequences for cardiac cells (23). Conversely, the exact cause of the key symptoms of stable, compensated HFPEF—exertional dyspnea and fatigue—is an area of active research: Prior studies (3,24) suggest that increased arterial stiffness is a contributor. However, the mechanisms are likely complex and multifactorial, perhaps also including impaired chronotropic reserve and vasodilation (25), and even factors beyond the heart (reduced arterial–venous oxygen content difference) may play a role (26).
In our study, HFPEF patients were older than the no HFPEF patients, and increased arterial stiffness (and pulse pressures [PPs]) are a hallmark of arterial aging. The diagnostic performance of pulsatile hemodynamics, however, was not different between younger and older patients, and—in the multivariable models—independent of age. Still, before incorporating these measures into formal diagnostic algorithms, it needs to be established that arterial stiffness/wave reflections are elevated in HFPEF patients as compared with age- and blood pressure–matched asymptomatic controls. This has been shown up to now only for age-matched controls in 2 previous series (21,24).
If these results are confirmed in subsequent studies with more typical HFPEF patients and closely age- and sex-matched controls, then how might they be applied in clinical routine? aoPWV is an invasive measurement, but the finding that models including aoPWV yielded the best diagnostic accuracy is of interest as a proof of concept. Conversely, bPP, a crude estimate of pulsatile load, is readily available in primary care and might be useful as a first-line test in symptomatic patients. In addition, in combination with echocardiography, bPP improves diagnostic accuracy. Among the more precise measures of pulsatile hemodynamics, aoPP seems to be most useful in the diagnostic process. Its superiority over bPP probably reflects the fact that aortic pressures are directly “seen” by the heart and are different from brachial pressures due to the phenomenon of PP amplification, which in turn depends on the stiffness of the aorta and the large arteries, and on wave reflections. Indeed, measures of wave reflections (Pb, AP) approximate the diagnostic value of aoPP, either alone or in combination with TDE, and can be combined with aoPP for further refinement of the diagnostic algorithm. Recently, methods for noninvasive measurement of central blood pressures, using regular oscillometric sphygmomanometers (27), have been developed and validated, which may enhance the availability of measures of pulsatile arterial function. Of note, effective arterial elastance, which has been used to describe arterial properties (7), was less useful to discriminate our diagnostic groups. This is in line with previous reports (28) and may be due to the fact that this measure is not a pure measure of pulsatile hemodynamics, but lumps together pulsatile and steady components of the arterial load (29).
Our study focuses on the role of impaired pulsatile hemodynamics as markers of long-standing/uncontrolled hypertension in the diagnosis of HFPEF. With respect to the treatment of HFPEF, some aspects may be of interest: in trials with uncontrolled hypertension and high PPs (in the range of our HFPEF group), the reduction in blood pressure was related to the improvement of diastolic function (30), whereas in patients with controlled blood pressure and relatively low PPs (in the range of our intermediate group) (31), treatment with antihypertensive drugs (inducing only a minor decrease of PP of 1.7 mm Hg) showed no benefit on clinical outcomes.
Some limitations need to be considered: first, we excluded patients with atrial fibrillation due to the inability to perform stable hemodynamic measurements. Second, although NT-proBNP levels were used for HFPEF case definition in the present study, we recognize there is significant controversy in this regard, and that NT-proBNP levels can be relatively low in HPFEF patients who are clinically noncongested and have nondilated left ventricles (32). Third, we did not perform exercise testing. This should have given additional information (33), not limited to, but particularly regarding the relatively large group with possible HFPEF. In these patients (euvolemic, preserved EF, and unexplained exertional dyspnea), marked increases in LVEDP, PCW, and systolic PA pressures during exercise can establish the diagnosis of HFPEF (32). With respect to pulsatile hemodynamics, a major increase in proximal afterload and arterial stiffening with moderate exercise has been found in HFPEF patients (34), and changes in estimated filling pressures have been related to changes in wave reflection during a submaximal exercise test in HFPEF patients (35). More work is needed to address which of the measures of pulsatile function (and their changes during exercise) are related to the fundamental factors that comprise exercise function (peak oxygen uptake, cardiac output, arteriovenous oxygen difference). Due to the referral to cardiac catheterization, we included more men than women into the study, leading to a relatively high proportion of men with HFPEF. The diagnostic value of pulsatile hemodynamics was independent from sex, however. Still, our findings should be replicated in other populations, particularly in typical HFPEF populations (elderly, predominantly female) with appropriate age- and sex-matched controls and in clinically diagnosed HFPEF patients. In the present study, we focused on stable patients without overt signs of congestion, where the diagnosis was less obvious and had to be based on invasive and laboratory measures. This may possibly explain the high rate of patients in the intermediate diagnostic group in our study, although the presence of a gradual onset of HFPEF rather than a simple presence or absence could be another explanation. Finally, extreme group analysis (considering only HFPEF and no HFPEF patients) allows for comparison between clearly established diagnostic groups, but likely overestimates the diagnostic utility estimates (AUCs, NRIs) for clinical routine use.
We have found that measures of pulsatile arterial function, notably bPP and aoPP and measures of wave reflection, may complement TDE in patients with unexplained exertional dyspnea in stable, nondecompensated patients.
For supplementary figures and tables and an expanded Methods section, please see the online version of this paper.
This work and its publication was partly supported by a grant of the government of Lower Austria and European Funds for Regional Development (EFRE), contract number WST3-T-81/015-2008. Dr. Wassertheurer is an employee of the AIT Austrian Institute of Technology, which is developing algorithms to measure vascular stiffness; he is also the co-inventor of a patent that is partly used in the ARCSolver. Prof. O'Rourke is a director of AtCor Medical, manufacturer of systems for pulse wave analysis and pulse wave velocity measurements; he has also received financial support from Aortic Wrap, Merck, and Novartis Pharmaceuticals. Dr. Hametner is an employee of the AIT Austrian Institute of Technology. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- aortic pulse pressure
- aortic pulse wave velocity
- augmented pressure
- area under the curve
- brachial pulse pressure
- coronary artery disease
- E/E′ at the medial annulus
- heart failure with preserved ejection fraction
- left ventricular end-diastolic pressure
- Net reclassification improvement
- N-terminal pro–B-type natriuretic peptide
- New York Heart Association
- incident pressure wave height
- amplitude of the backward wave
- amplitude of the forward wave
- pulse pressure
- pulse wave velocity
- receiver-operating characteristic
- systolic blood pressure
- tissue Doppler echocardiography
- Received November 6, 2012.
- Revision received January 14, 2013.
- Accepted February 5, 2013.
- American College of Cardiology Foundation
- Klapholz M.,
- Maurer M.,
- Lowe A.M.,
- et al.
- Hundley W.G.,
- Kitzman D.W.,
- Morgan T.M.,
- et al.
- Kawaguchi M.,
- Hay I.,
- Fetics B.,
- Kass D.A.
- Weber T.,
- Auer J.,
- O'Rourke M.F.,
- Punzengruber C.,
- Kvas E.,
- Eber B.
- Weber T.,
- O'Rourke M.F.,
- Ammer M.,
- Kvas E.,
- Punzengruber C.,
- Eber B.
- Borlaug B.A.,
- Melenovsky V.,
- Redfield M.A.,
- et al.
- Fernandes V.R.S.,
- Polak J.F.,
- Cheng S.,
- et al.
- Nagueh S.F.,
- Appleton C.P.,
- Gillebert T.C.,
- et al.
- Paulus W.J.,
- Tschöpe C.,
- Sanderson J.E.,
- et al.
- Weber T.,
- Auer J.,
- O'Rourke M.F.,
- et al.
- Davidson C.J.,
- Bonow R.O.
- Kern M.J.,
- Appleby S.
- Le Jemtel T.H.,
- Padeletti M.,
- Jelic S.
- Narayanan A.,
- Aurigemma G.P.
- Kang S.,
- Fan H.-M.,
- Li J.,
- et al.
- Chirinos J.A.,
- Kips J.G.,
- Jacobs D.R.,
- et al.
- Shah S.J.,
- Wasserstrom J.A.
- Borlaug B.A.,
- Melenovsky V.,
- Russell S.D.,
- et al.
- Haykowsky M.J.,
- Brubaker P.H.,
- John J.M.,
- et al.
- Lam C.S.P.,
- Roger V.L.,
- Rodeheffer R.J.,
- et al.
- Segers P.,
- Stergiopulos N.,
- Westerhof N.
- Borlaug B.A.,
- Nishimura R.A.,
- Sorajja P.,
- Lam C.S.P.,
- Redfield M.M.
- Guazzi M.,
- Adams V.,
- Conraads V.,
- et al.
- Tartière-Kesri L.,
- Tartière J.-M.,
- Logeart D.,
- Beauvais F.,
- Cohen Solal A.