Author + information
- Received May 7, 2005
- Revision received July 4, 2005
- Accepted July 18, 2005
- Published online November 15, 2005.
- ↵⁎Reprint requests and correspondence:
Dr. Marco Guazzi, Cardiopulmonary Laboratory, Cardiology Division, University of Milano, San Paolo Hospital, Via A. di Rudini, 8, 20142 Milan, Italy.
Objectives This study sought to define the relative prognostic value of cardiopulmonary exercise testing (CPET) variables in heart failure (HF) patients with preserved versus reduced systolic function.
Background Cardiopulmonary exercise testing has an established role in the assessment of patients with systolic heart failure (SHF). Two variables, peak Vo2and, more recently, the Ve/Vco2slope, have been shown to be extremely valuable in risk stratification. However, data are lacking in terms of the prognostic value of CPET in patients with diastolic heart failure (DHF).
Methods A total of 409 HF patients underwent CPET. Patients were divided into three groups according to the following left ventricular ejection fraction (LVEF) cutoffs: ≥40%, ≥45%, and ≥50%. The CPET response and the ability of peak Vo2and the Ve/Vco2slope to predict total mortality and hospitalization were examined.
Results At univariate Cox regression analysis, both the peak Vo2and the Ve/Vco2slope were significant predictors in SHF and DHF. Multivariate analysis documented a similar prognostic power of Ve/Vco2slope and peak Vo2in all SHF groups. Conversely, in DHF patients, Ve/Vco2slope outnumbered peak Vo2, remaining the only predictor regardless of LVEF. In DHF, the area under the receiver operating characteristic curve for the Ve/Vco2slope identified a cutoff of 32.6 (74% sensitivity, 52% specificity), 33.1 (76% sensitivity, 62% specificity), and 33.3 (97% sensitivity, 40% specificity) for an LVEF cutoff of ≥40%, ≥45%, and ≥50%, respectively.
Conclusions These results extend the clinical and prognostic applicability of CPET to DHF. An impairment in exercise ventilation rather than peak Vo2holds clinical and prognostic impact in this increasing subset of patients.
Recent epidemiological studies have provided growing recognition that heart failure (HF) with preserved left ventricular ejection fraction (LVEF) recognized as diastolic heart failure (DHF) is a common and costly clinical entity that is increasing in prevalence (1,2). However, because of the paucity of clinical trials addressing DHF patients, scientific evidence regarding the natural history and clinical features of this syndrome remains limited.
Impaired physical performance is an important hallmark of the early stages of HF (3), and it has recently been suggested that abnormalities in exercise metabolism are part of the clinical scenario that limits exercise tolerance in DHF patients (4). Cardiopulmonary exercise testing (CPET) is the gold-standard technique for the evaluation of putative mechanisms that underlie exercise intolerance in HF (5), and CPET-derived indexes of cardiovascular and respiratory limitation have repeatedly emerged as precise predictors of survival rate (6–12). However, studies assessing the clinical significance and prognostic power of CPET-derived variables have, to this point, exclusively involved patients with depressed LVEF or systolic heart failure (SHF). How and whether CPET may add to the backlog of information regarding DHF patients with preserved LVEF remains unknown.
We therefore designed the current study with two objectives: 1) to explore the prognostic value of CPET variables in the subset of DHF patients, and 2) to better examine the physiological response to exercise in symptomatic patients with DHF.
This was a multicenter study consisting of HF patients from the Cardiopulmonary Laboratories at San Paolo Hospital, Milan, Italy; the Virginia Commonwealth University, Richmond, Virginia; and the VA Palo Alto Health Care System and Stanford University, Palo Alto, California. A total of 409 patients with chronic HF were studied during a progressively increasing maximal CPET. Patients who met the Framingham criteria for congestive HF were considered eligible candidates (13). Subjects with significant obstructive lung disease evidenced as a forced expiratory volume in 1 s ≤70% or who were unable to perform a maximal exercise test were excluded from the study. Between 15% and 20% of the whole population were smokers and were similarly distributed across different subgroups. All patients were in New York Heart Association functional class II to III.
As detailed subsequently, diastolic function was assessed by echocardiography and Doppler-derived mitral inflow analysis.
Despite the fact that documentation of a normal systolic function (i.e., preserved LVEF) is part of the definition of DHF, considerable differences in the selection criteria for LVEF cutoffs have been reported across various studies. In a series of reports comparing patients with preserved versus depressed systolic function, the cutoff used was a LVEF ≥40% (14,15). The Working Group of the European Society of Cardiology Guidelines (16) defined systolic function as being normal when LVEF is ≥45%. Subsequent studies aimed at better standardizing the diagnostic criteria of DHF have proposed a cutoff of ≥50% (17–19). It follows that depending on which LVEF cutoff is used, the diagnostic yield of DHF varies greatly. Accordingly, we performed three separate survival analyses based on the three LVEF cutoffs proposed ≥40%, ≥45%, and ≥50%. The LVEF was determined during the initial in-hospital evaluation.
The cause of HF was coronary artery disease in 224 patients, hypertension in 95 patients, and idiopathic cardiomyopathy (cardiac enlargement and absence of a specific reason for that) in the rest of the patients. Other than total mortality, we included hospitalization for cardiac reasons as an end point and performed the analysis considering a follow-up period of one year. Irrespective of etiology, individuals with HF can shift from a stable to an uncompensated status (or vice versa) rather abruptly. Limiting the follow-up period to one year may be clinically optimal given the fluid nature of cardiac function in the HF patient. We recently completed an analysis of the impact of time past CPET on the prognostic characteristics of the Ve/Vco2slope and peak Vo2in subjects with HF (20). This analysis indicated that the sensitivity for predicting outcomes increased modestly, whereas specificity dramatically decreased for both CPET variables after more than one year after exercise testing. A one-year tracking period may therefore represent an appropriate balance between avoiding outdated information and the economic constraints of multiple exercise tests.
Subjects were followed up for cardiac-related mortality and hospitalization via medical chart review at the respective centers. Cardiac-related mortality was defined as death directly resulting from failure of the cardiac system. An example fitting this definition is myocardial infarction followed by cardiac arrest. Cardiac-related hospitalization was defined as a hospital admission directly resulting from cardiac dysfunction requiring in-patient care. An example fitting this definition is decompensated HF requiring the use of an intravenous inotropic agent. Any death or hospital admission with a cardiac-related discharge diagnosis, confirmed by diagnostic test or autopsy, was considered an event. Subjects in whom mortality or hospitalization was of a non-cardiac etiology were treated as censored cases.
Patients lost to follow-up were considered to be anyone who was not tracked for one year and did not have an event; 56 patients were not tracked for one year and were not included in the final analysis.
Standard M-mode and two-dimensional echocardiography and Doppler blood flow measurements were performed in agreement with the American Society of Echocardiography guidelines (21). Septal and posterior LV wall thicknesses were obtained from the parasternal long-axis view. The LV end-systolic and end-diastolic volumes were obtained from two-dimensional apical images. The LVEF was calculated from two-dimensional apical images according to the Simpson method. The LV mass was calculated according to the formula proposed by Devereux et al. (22). Pulsed-wave Doppler echocardiography was used to assess mitral peak early (E) and late (A) wave flow velocity and E-wave deceleration time. Isovolumic relaxation time was also determined.
Doppler mitral in-flow velocities were used to classify diastolic function according to the following classification: normal, rates of mitral E to A ratio >1 and E-wave deceleration time ≤220 ms; mild, rates of mitral E to A ratio <1 and E-wave deceleration time >220 m; moderate, rates of mitral E to A ratio 1 to 2 and E-wave deceleration time 150 to 200 ms; severe, rates of mitral E to A ratio >2 and E-wave deceleration time <150 ms (1,23,24).
CPET procedure and data collection
Symptom-limited CPET was performed in all patients after written informed consent had been obtained. All centers used individualized ramp protocols, and maximal tests were planned to obtain an exercise duration between 8 and 10 min. Centers in the U.S. used treadmills and the Italian center used a cycle ergometer. A potential prognostic bias derived by comparing different exercise modes can reasonably be excluded considering that in a previous study the comparison between these two modes of exercise led to the identical predictive cutoff value for both peak Vo2and Ve/Vco2slope (25). Standard 12-lead electrocardiograms were obtained at rest, each minute during exercise, and for at least five minutes during the recovery phase; blood pressure was measured using a standard cuff sphygmomanometer. Minute ventilation (Ve, body temperature, ambient pressure, saturated with water), oxygen uptake (Vo2, standard pressure and temperature, dry [STPD]), carbon dioxide output (Vco2, STPD) and other cardiopulmonary variables were acquired breath-by-breath, averaged over 30 s, and printed in rolling averages every 10 s. The V-slope method was used to measure the anaerobic threshold (26). Although different methods for calculating the Ve/Vco2slope have been proposed, we measured this variable by including all data points from the beginning to the end of exercise, in agreement with what has recently been shown by our group (27) and others (28). We reasoned that this was the preferable method for preventing variability among laboratories.
Univariate Cox regression analysis was used to determine the prognostic ability of peak Vo2and Ve/Vco2slope. Multivariate Cox regression analysis was then performed using a forward stepwise approach to assess the combined prognostic effects of these variables in predicting total mortality and hospitalization in all HF populations. Entry and removal p values for the multivariate analyses were set at 0.05 and 0.10, respectively.
Receiver operating characteristic (ROC) curves were constructed for the peak Vo2and Ve/Vco2slope classification schemes for the mortality end points. A z test was used to compare peak VO2and the Ve/Vco2slope prognostic power (29). Optimal threshold values (highest combination of sensitivity/specificity) were identified via ROC curve analysis. The threshold value was determined when the z test showed a significant difference in area under the ROC curve between peak Vo2and Ve/Vco2slope. Cox regression analysis retained only that parameter. Two threshold values were determined when the z test showed that the area under the ROC curves between peak Vo2and the Ve/Vco2slope were not significantly different and multivariate Cox regression analysis retained both parameters. Kaplan-Meier analyses and hazard ratio calculations were subsequently performed with the threshold values. The log-rank test was used to compare the equality of survival distributions in the Kaplan-Meier analyses.
Intergroup differences between clinical and exercise variables were compared using unpaired ttest analysis. All data are reported as mean values ± standard deviation. Statistical tests with a p value <0.05 were considered to be significant. All tests were performed using SPSS for Windows version 11.0 (SPSS Inc., Chicago, Illinois).
Over the one-year tracking period, there were 23 deaths and 90 hospitalizations for cardiac reasons. Table 1reports the clinical characteristics and CPET data of the study populations according to LVEF cutoffs. In all groups, SHF and DHF patients had similar age and gender distributions. A significantly lower distribution of diuretics, aspirin, statins, and digoxin occurred in DHF patients with a LVEF ≥45% and ≥50%.
Patients exercised above their anaerobic threshold and achieved a high peak respiratory exchange ratio (RER) (≥1.05 in all groups), suggesting that they developed significant metabolic acidosis and exercised close to maximal intensity. In all groups, patients with SHF had a significantly lower peak Vo2and a significantly higher Ve/Vco2slope. Peak HR, exercise time, and RER were not significantly different between SHF and DHF across identified populations.
Table 2summarizes the echo-derived data. In all groups, DHF patients presented with significantly smaller LV volumes and masses. However, the ratio LV mass/LV end-diastolic volume was significantly lower in SHF. The DHF patients, irrespective of LVEF, showed a significantly lower E/A ratio and prolonged deceleration time.
Univariate Cox regression analysis showed that both peak Vo2, except for the LVEF ≥45% group, and the Ve/Vco2slope were significant predictors of combined mortality and cardiac-related hospitalization in SHF and DHF groups (Table 3).Multivariate Cox regression analysis (Table 4)showed that both peak Vo2and Ve/Vco2slope were predictors of mortality in SHF groups.
In contrast, in DHF patients Ve/Vco2slope outnumbered peak Vo2, remaining the only CPET predictor regardless of LVEF (residual chi-square = 9.9, p = 0.002 for LVEF ≥40%; residual chi-square = 8.6, p = 0.003 for LVEF ≥45%; and residual chi-square = 14.5, p < 0.001 for LVEF ≥50%). The ROC curve analysis results for the peak Vo2and Ve/Vco2slope classifications are presented in Table 5.In all groups of SHF and DHF patients, the area under the ROC curve was greater for the Ve/Vco2slope than for peak Vo2. In DHF, the best Ve/Vco2cutoff identified was 32.6 (sensitivity of 74% and specificity of 52%) for LVEF ≥40%; 33.1 (sensitivity of 76% and specificity of 62%) for LVEF ≥45%; and 33.3 (sensitivity of 97% and specificity of 40%) for LVEF ≥50%.
It is widely accepted that peak Vo2, and, more recently Ve/Vco2slope, are strong prognostic markers in HF populations with impaired LV function (6–12). Previous reports have, however, primarily focused on the prognostic applications of CEPT variables in SHF patients. Although the present results confirm that in SHF patients both peak VO2and Ve/Vco2slope retain a prognostic value, the new finding was that a steep Ve/Vco2slope, compared with peak Vo2, is a more powerful prognostic marker in patients with DHF. Thus, abnormalities in exercise ventilation that can be determined submaximally and reflect the constellation of central and peripheral factors that underlie exercise intolerance in HF powerfully predict outcomes in both SHF and DHF. Although a need for definitive standardization of HF with preserved systolic function or DHF has been repeatedly emphasized and considerable effort has been posed on it (17–19), a precise and accepted definition is still under scrutiny. In previous observational studies, a preserved systolic function was defined on divergent LVEF cutoffs ranging from 40% to 50% (14–19). This obviously precludes a direct comparison between findings in different studies and may explain why most questions and clinical issues related to DHF remain to be answered. To circumvent this problem, we performed the analysis by grouping patients according to the three different LVEF cutoffs. Interestingly, in all groups categorized as DHF, Ve/Vco2slope showed a similar prognostic power regardless of the LVEF cutoff considered. A potential additional strength of the present investigation stands on the use of a combined mortality/cardiac-related hospitalization end point.
Most studies examining the prognostic value of CPET have not used hospitalization as an end point. Given that HF is the primary hospital diagnosis-related group among Medicare patients, and considering that the one-year readmission rate of DHF is nearly identical to that for SHF and approaches 50% (30–32), analysis of measures predicting hospitalization in this population seems warranted.
Ve/Vco2slope versus peak Vo2: prognostic value
In a study by Ponikowski et al. (33), it was observed that a subgroup of patients with depressed LV function but normal peak Vo2(≥18 ml · min−1 · kg−1) and a Ve/Vco2slope ≥34 had a significantly higher mortality rate than patients with a normal peak Vo2and a Ve/Vco2slope <34. This provocative study was the first to introduce the concept that ventilatory inefficiency during exercise predicts mortality in HF even in the presence of a normal peak Vo2. Our results confirm this observation and extend it to HF patients with preserved LVEF. This further challenges the belief that the greater the severity of the syndrome, the steeper the Ve/VCO2slope; an assumption that again has been generated by the analysis of prospective trials exclusively including patients with significant reduction in LVEF.
Interestingly, in all DHF groups the identified cutoffs for Ve/Vco2slope were definitively similar to the cutoff of 34 that has been identified as a reference value in previous studies (6–12) for populations with depressed LV systolic function. Together, these observations suggest that an impairment in ventilatory efficiency should be considered a hallmark for the estimation of risk among patients with HF, and that these risk estimates are equally powerful among patients with reduced and preserved systolic function. As a parallel finding, we observed in all DHF patients that peak Vo2was significantly higher whereas Ve/Vco2slope was significantly lower in DHF compared with SHF, suggesting that there is a significant difference in functional capacity and physical performance between the two conditions. These findings, along with the lower mortality rate, suggest that despite the fact that DHF is a syndrome that is increasing in prevalence and that has a high morbidity, the natural history and clinical course are less severe than those for SHF (2,17).
CPET features of DHF
In a recent study characterizing exercise pathophysiological features of DHF, Kitzman et al. (4) reported that, among 50 patients with DHF, the peak exercise Ve/Vco2ratio was similar to that observed in patients with SHF. Although average peak Vo2was lower than that observed in our population, it was still significantly higher in comparison with that for patients with SHF. Differences between the results of Kitzman et al. (4) and our findings may be explained by considering that the Ve/Vco2ratio does not necessarily parallel the slope of this relationship; identical ratios may occur with markedly different slopes. Another important difference is that the average age of patients was considerably higher than that in our population, which likely accounts for the differences in peak Vo2.
We did not investigate the pathophysiological mechanisms that lead to ventilatory inefficiency in patients with DHF. Specifically, we do not know the extent to which the increased Ve/Vco2slope is secondary to uneven perfusion/ventilation matching, or whether it reflects muscle deconditioning with early metabolic acidosis or ergoreflex activation. This is an unanswered question that needs to be addressed in future studies, and has important implications for optimizing therapeutic strategies in diastolic HF. In addition, the total number of DHF patients was small. However, given the recognized lower mortality rate in these patients and considering our final combined end point, i.e., total mortality and hospitalization, the sample size seems adequate for drawing preliminary conclusions. At least in patients with SHF, beta-blocker distribution was suboptimal. The main reason is that, apart from patients who did not tolerate beta-blocker titration, some patients were already receiving digoxin, and beta-blocker titration to high effective doses was not tolerated. Compared with other DHF populations, the female gender distribution in our study patients was quite low (12%) (4,34,35). This suggests that some caution may be warranted in the extrapolation of our findings to a predominantly female gender population of HF patients with preserved systolic function (4).
These results extend the clinical and prognostic applicability of CPET-derived variables to patients with DHF and preserved LVEF. An impaired ventilation rather than peak Vo2holds clinical and prognostic impact in this population. Given the increasing burden of DHF, our findings may help to better characterize the natural history, treatment, and estimation of risk in DHF.
Presented in part at the 77th American Heart Association Scientific Sessions, New Orleans, Louisiana, November 7 to 10, 2004.
- Abbreviations and Acronyms
- cardiopulmonary exercise test
- diastolic heart failure
- heart failure
- left ventricular ejection fraction
- receiver operating characteristic
- systolic heart failure
- carbon dioxide production
- oxygen uptake
- Received May 7, 2005.
- Revision received July 4, 2005.
- Accepted July 18, 2005.
- American College of Cardiology Foundation
- Zile M.R.,
- Brutsaert D.L.
- Gibbons R.J.,
- Balady G.J.,
- Timothy B.J.,
- et al.
- Wasserman K.,
- Hansen J.E.,
- Sue D.Y.
- Mancini D.,
- Heisen H.,
- Kussmaul W.,
- et al.
- Gitt A.,
- Wasserman K.,
- Kikowski C.,
- et al.
- Francis D.P.,
- Shamin W.,
- Davies L.C.,
- et al.
- Guazzi M.,
- Reina G.,
- Tumminello G.,
- Guazzi M.D.
- Smith G.L.,
- Masoudi F.A.,
- Vaccarino V.,
- Radford M.J.,
- Krumholz M.
- European Study Group on Diastolic Heart Failure
- Vasan R.S.,
- Levy D.
- Zile M.R.,
- Gaasch W.H.,
- Carrol J.D.,
- et al.
- Zile M.R.,
- Brutsaert D.L.
- ↵Arena R, Myers J, Aslam SS, Varughese EB, Peberdy MA. Impact of time past exercise testing on prognostic variables in heart failure. Int J Cardiol 2005. In press.
- Cheitlin M.D.,
- Armstrong W.F.,
- Aurigemma G.P.,
- et al.
- Garcia M.J.,
- Thomas J.D.,
- Klein A.L.
- ↵Arena R, Guazzi M, Myers J, Peberdy A. Prognostic characteristics of cardiopulmonary exercise testing in heart failure: comparing American and European models. Eur J Cardiovasc Prev 2005. In press.
- Beaver W.L.,
- Wasserman K.,
- Whipp B.J.
- Vasan R.S.,
- Larson M.G.,
- Benjamin E.J.,
- et al.
- Dauterman K.W.,
- Massie B.M.,
- Gheorgiade M.
- Ponikowski P.,
- Francis D.P.,
- Piepoli M.F.,
- et al.
- Senni M.,
- Tribouilloy C.M.,
- Rodeheffer R.J.,
- et al.
- Ramachandran S.,
- Vasan R.S.,
- Larson M.G.,
- et al.