Author + information
- Received September 27, 2002
- Revision received December 10, 2002
- Accepted January 24, 2003
- Published online July 2, 2003.
- Hans Rickli, MD*,†,
- Wolfgang Kiowski, MD*,
- Manuel Brehm*,
- Daniel Weilenmann, MD*,‡,
- Christoph Schalcher, MD*,
- Alain Bernheim, MD*,‡,
- Erwin Oechslin, MD* and
- Hans Peter Brunner-La Rocca, MD*,‡,* ()
- ↵*Reprint requests and correspondence:
Dr. Hans Peter Brunner-La Rocca, Division of Cardiology, University Hospital, Petersgraben 4, 4031, Basel, Switzerland.
Objectives This study investigated the combination of maximal and low-intensity exercise testing in predicting prognosis in chronic heart failure (CHF), using one single exercise test (two-step protocol).
Background Risk assessment based on any single factor has limited accuracy and reproducibility.
Methods Treadmill exercise testing was performed in 202 consecutive CHF patients (174 male; mean age 52 ± 11 years) using “breath-by-breath” gas exchange monitoring. Oxygen uptake (Vo2) kinetics were defined as oxygen deficit (ΔVo2× time [rest to steady state] − Σ Vo2[rest to steady state]) and mean response time (MRT = oxygen-deficit/ΔVo2). Peak Vo2(Vo2max) was defined as the highest Vo2. Mean follow-up was 873 ± 628 days. The primary end point was cardiac mortality and the need for urgent heart transplantation.
Results Forty-four patients (22%) died and 15 (7%) were urgently transplanted. In both univariate and multivariate analyses, MRT >50 s was the most powerful predictor of the primary end point (hazard ratio [HR] 4.44), followed by predicted Vo2max <50% (HR 3.50) and resting systolic blood pressure <105 mm Hg (HR 2.49, all p < 0.001). A majority (n = 130 [64%]) had one or none of these risk factors, with a one-year event rate of only 3%. Patients with two risk factors (n = 45 [22%]) were at medium risk (one-year event rate of 33%). Twenty-seven patients (13%) had all three risk factors, with a one-year event rate of 59%. The area under the curve, using the number of risk factors, was 0.86 ± 0.04 for the primary end point at one year. These results were independent of medication, in particular, beta-blockade.
Conclusions A combination of low-intensity and maximal exercise test results improves assessment of prognosis in patients with CHF.
Exercise testing has become a mainstay for the prognostic assessment of patients with chronic heart failure (CHF) (1). Various studies have shown that patients with impaired exercise capacity, as measured by low peak oxygen consumption (Vo2), have an unfavorable outcome, independent of other parameters (2). However, risk assessment based on any single factor has limited accuracy and reproducibility. Therefore, prognostic scores, including various clinical and hemodynamic parameters, were introduced to improve prediction of outcome in CHF (3).
Recent studies have investigated whether ventilatory and gas exchange responses other than peak Vo2(Vo2max) may have prognostic value (4,5)or tried to substitute peak exercise testing by submaximal tests (6). Thus, assessment of Vo2kinetics at low-intensity exercise differed between heart failure patients with poor versus preserved left ventricular function, whereas Vo2max did not (7). Because Vo2kinetics and Vo2max appear to provide physiologically different information, we wondered whether a combination of parameters obtained at low intensity as well as maximal exercise testing in the same test may provide a more complete risk assessment in CHF patients.
A total of 219 consecutive CHF patients referred to our CHF clinic with a left ventricular ejection fraction <40% were considered for this study. Of these, 17 (8%) did not tolerate spirometry during exercise or had technically inadequate recordings. Thus, 202 patients form the basis for this report. Patients with complex congenital heart disease, CHF with a preserved ejection fraction, or decompensated heart failure were not included in the study. Patients were also excluded if either severe pulmonary or orthopedic disease limited their exercise capacity (<5% of the entire population referred to our clinic). Correctable causes of CHF were treated before inclusion in the study. Conditions had to be stable for at least two months.
The cause of CHF was coronary artery disease in 106 (53%), dilated cardiomyopathy in 69 (34%), valvular disease in 5 (2%), hypertension in 4 (2%), and other or combined cardiac etiologies in 18 (9%). All but two patients (99%) were taking either an angiotensin-converting enzyme inhibitor or an angiotensin II receptor blocker, 182 (90%) were on diuretics, 100 (50%) on digitalis, 91 (45%) on beta-blockers, 45 (22%) on nitrates, and 47 (23%) on amiodarone. One hundred forty-eight patients (73%) were receiving oral anticoagulation, 25 (12%) aspirin, and 11 (5%) both drugs.
All patients were assessed clinically and had a standard multipanel laboratory screening. Subsequently, a treadmill exercise test was performed. Other diagnostic tests, such as right heart catheterization and radiography, were performed according to the discretion of the treating physicians. Because these data are not complete, they were not included in the analysis.
Gas exchange was assessed breath by breath, using a CPX/D system (Medical Graphics Corp., St. Paul, Minnesota), calibrated before each test. There was no familiarization test before the study, but patients were carefully instructed about how to walk on the treadmill and became familiarized with breathing connected to the spirometry system. Patients started walking after reaching steady-state gas exchange while standing quietly. A two-step protocol was used (8). Initially, they walked at 1.0 mph with an elevation of 6% for 6 min, corresponding to ∼0.5 W/kg body weight. Thereafter, both speed and elevation were increased to augment workload by ∼0.15 W/kg body weight per minute until exhaustion. Workload was assessed by rearrangement of the formula by the American College of Sports Medicine (9):where elevation is expressed in %, speed in m/s (i.e., mph × 1,609/3,600), and gas the gravitation constant (i.e., 9.81).
The oxygen deficit was calculated according to the formula (8): where tis the time from rest to steady state (min); ΔVo2is the difference in oxygen uptake between rest and steady state (ml/min); and ΣVo2is the sum of consumed oxygen (ml). To calculate the oxygen deficit, software called BreezeEx, version 3.02 (Medical Graphics Corp.) was used.
The mean response time of oxygen uptake (MRT; i.e., the time constant of oxygen uptake) was calculated using the formula (10): Peak Vo2was defined as the highest Vo2by averaging five of seven consecutive breaths during any stage. In most instances, this corresponded to the highest workload that was sustained for 1 min (i.e., peak work load). Peak Vo2is reported after correction for body weight (ml/min/kg) and as a percentage of predicted normal values accounting for age, height, weight, and gender (11). The ventilatory response to exercise (VE/Vco2) was defined as previously described (12). One case of nonsustained ventricular tachycardia occurred immediately after the test.
General practitioners were primarily responsible for the care of the patients. However, all surviving patients had structured follow-up visits in our clinic. Patient status was determined by one of us (Dr. Rickli) by interviewing referring physicians and/or patients and by chart review, without knowledge of other patient data. A cardiac cause of death was assumed if the patient either died with symptoms of progressive CHF or died suddenly. One patient committed suicide and his follow-up was censored at the time of the event. No other noncardiac deaths were noted. Urgent transplantation, defined by the need for intensive care treatment with positive inotropic agents or a left ventricular assist device (LVAD), was considered as the primary end point. Patients were listed for elective transplantation according to the judgment of the treating physicians, based on clinical data, hemodynamics, and ergospirometry. Therefore, elective transplantation was not considered as an end point and follow-up was censored at that time.
Data are expressed as the mean value ± SD and frequencies, as indicated. Survival and survival free of urgent transplantation were calculated using Kaplan-Meier curves. The hazard ratio (HR) was assessed by univariate Cox regression analysis. Multivariate Cox regression analysis was performed, including significant predictors of outcome in the univariate analysis in a stepwise manner, accounting for potential interrelations between variables. An unadjusted multivariate model was used as well as a model adjusted for age, gender, and ejection fraction. Because both models provided identical results, only the unadjusted model is presented.
Receiver operating characteristics curves for one-year survival and the primary end point of survival free of urgent transplantation were analyzed by nonparametric Zstatistics. Statistical analysis was performed using the software package SPSS, version 9.0 (SPSS Inc., Chicago, Illinois).
Baseline characteristics are listed in Table 1. Forty-four patients (22%) died from a cardiac cause and 15 (7%) needed urgent transplantation (mean follow-up 873 ± 628 days; median 737 days). Another 15 patients underwent elective heart transplantation.
Table 2shows significant predictors of outcome in the univariate analysis. The most significant predictor of all end points was MRT. The predictive value of Vo2max was better if it was normalized to the predicted value rather than just body weight.
Stepwise multivariate Cox regression analysis yielded MRT, percent predicted Vo2max, and resting systolic blood pressure (BP) as independent predictors of outcome (Table 3). These independent predictors of outcome were expressed as dichotomous variables to separate patients into low-, medium-, and high-risk groups. Thus, patients with MRT >50 s (HR 4.44 for the primary end point and HR 3.52 for cardiac mortality; p < 0.001), Vo2max <50% of the predicted value (HR 3.29 and 2.96, respectively; p < 0.001), and resting BP <105 mm Hg (HR 2.49 and 2.25, respectively; p < 0.01) were at particular risk of a poor outcome. Importantly, MRT predicted outcome in patients with percent predicted Vo2max <50% and >50% (Fig. 1).
One hundred thirty patients (64%) had none or only one of these risk factors. These patients had a good prognosis (Fig. 2), whereas patients with two risk factors (n = 45 [22%]) were at medium risk of a poor outcome. Twenty-seven patients (13%) had all three of these risk factors. These patients were at high risk, with a one-year mortality rate of 50% (Figs. 2 and 3). ⇓The area under the curve, using the number of risk factors, was 0.86 ± 0.04 (95% confidence interval [CI] 0.78 to 0.94) for the primary end point at one year and 0.83 ± 0.05 for one-year cardiac mortality (95% CI 0.75 to 0.92; both p < 0.0001). Risk assessment was not superior using the score of Aaronson et al. (3), as compared with the model described herein.
Because of the known impact of beta-blockade on prognosis, patients were separated into those with and without beta-blockade. Apart from the expected trend in heart rate (at rest: 80 ± 17 vs. 84 ± 16 beats/min, p = 0.11; peak: 131 ± 27 vs. 138 ± 25 beats/min, p = 0.05), baseline characteristics did not differ between patients with and without beta-blockade. Patients treated with beta-blockers had a better outcome than those without (primary end point HR 0.56, p < 0.05; cardiac mortality HR 0.51, p = 0.06), but the number of risk factors still predicted the outcome in both subgroups (Fig. 4). Importantly, high-risk patients had an equally poor outcome, irrespective of whether they were on beta-blockers or not (p = 0.8). The incremental value of each of the three risk factors to predict outcome did not differ between those treated with beta-blockers and those without.
This study suggests that the combination of parameters obtained at the beginning of and during maximal exercise testing may be helpful in assessing prognosis in CHF. The prognostic information of low-intensity and maximal exercise testing was comparable to previously described complex scores for prognostic assessment in CHF patients (3,13), but was obtainable with one single exercise test. Importantly, a relatively small subgroup of patients (13%) with a 50% one-year mortality rate could be identified, whose poor prognosis was independent of treatment with beta-blockers. These patients may be candidates for heart transplantation or LVAD implantation.
During the last two decades, medical therapy has progressed significantly (14), thereby reducing the generally accepted survival benefit of heart transplantation. Furthermore, a shortage of organ donors limits its therapeutic potential (15). Therefore, identification of patients who profit most from heart transplantation is essential (16). Recently, survival benefits have been described with the use of LVAD in patients with a very poor prognosis (17). However, the adverse effects and cost of these devices are enormous, highlighting the importance of identification of CHF patients with the poorest prognosis.
During the last decade, numerous studies have assessed prognosis in CHF. Various parameters have been found to be related to outcome in CHF (3,9). Among them, the results of cardiopulmonary exercise testing has gained widespread acceptance (18). Early results have shown that Vo2max <14 ml/kg/min identifies patients with severe CHF who benefit from heart transplantation, whereas in those with Vo2max >14 ml/kg/min, transplantation could be safely deferred (1). As Vo2max is affected by age, gender, and stature, the percent predicted Vo2max, rather than only the weight-adjusted value, may be superior in predicting survival. This approach, however, added only minimal prognostic information in some studies (19), whereas our study and previous data (20)suggest that a cutoff point of 50% of predicted Vo2max was the best value. This discrepancy may be explained, at least in part, by the inclusion of differing patient populations and the use of different exercise protocols.
Individual assessment based on any single factor has limited accuracy and reproducibility (3). Thus, CHF patients in stable conditions were found to have a two-year survival rate as good as 86%, despite significantly reduced Vo2max of 12.3 ± 1.5 ml/kg/min (21). On the other hand, the annual mortality rate may be up to 10%, even in patients with preserved exercise capacity (22). Thus, there is a need for further risk stratification in addition to measurement of Vo2max. Various studies incorporated other parameters to improve risk stratification (19). However, only a few studies have tried to integrate parameters other than Vo2max in the same exercise test (23,24). Moreover, in CHF patients with preserved exercise capacity, an enhanced ventilatory response at peak exercise predicted a poor prognosis and added independent prognostic information to Vo2max (23,25). However, this differs from studies in which the ventilatory response at a submaximal workload provided no prognostic information to Vo2max (26). Previously, attempts have also been made to substitute Vo2max with parameters obtained during submaximal exercise tests (6). However, a correlation of Vo2max to parameters of submaximal exercise tests was often only moderate (8,26–28)and was particularly poor in patients with an intermediate Vo2max value (10 to 20 ml/kg per min) (r = 0.28) (28). This may not be surprising, as low-intensity exercise and peak exercise represent different cardiopulmonary responses in terms of the increment of cardiac output and ventilatory response (7). At the start of exercise, stroke volume increases first, and a further increase in cardiac output predominantly depends on an increase in heart rate (11). In CHF, chronotropic incompetence is common and related to a poor outcome, but not independent of Vo2max. Thus, altered Vo2max may be influenced primarily by the heart rate response, whereas Vo2kinetics at low-intensity exercise are primarily influenced by changes in pump function. Our data suggest that the prognostic value of Vo2kinetics at low-intensity exercise is complementary to Vo2max, and that this may still be true, even if the heart rate response is altered by beta-blockade.
Additionally, we found resting systolic BP to be an independent predictor of prognosis. The relationship between low resting systolic BP and an adverse outcome was found in some but not all previous studies (3,24).
A number of models have been developed for improved risk stratification of patients considered for transplantation (3,29). Our study shows that powerful risk stratification may be possible with the use of one simple test using a two-step protocol. Combining low-intensity and maximal exercise testing is quick and inexpensive and does not require the patient to undergo additional testing. The data indicate that prognosis is very poor in patients with prolonged oxygen kinetics, low Vo2max, and low resting BP. The one-year mortality rate was 50% in this minority of patients (13%). In all other groups, the prognosis was not substantially different from what would have been expected in transplanted patients or those who had LVAD implantation. Patients with two of these risk factors may, however, deserve close monitoring for potential listing for transplantation.
Some limitations apply to the present study. The percentage of women was low (14%). Peak Vo2has been described to be gender-related, and it is unknown whether this also applies to Vo2kinetics in female patients with CHF. Also, the data were derived from patients referred for evaluation of possible heart transplantation. Accordingly, the results may not apply to less severely affected patients. However, their prognosis usually is better and the need for risk assessment less urgent.
Peak Vo2, MRT, and BP are continuous variables; therefore, it is difficult to define a single cutoff value beyond which risk of an event is considered high. Data from different studies examining populations of patients with different severities of heart failure have yielded various thresholds of Vo2max (30). However, stratifying patients above and below Vo2max of 50% of predicted maximum has demonstrated marked differences in survival in this and other studies (19). We have chosen a cutoff value of 50 s for MRT, which lies between the previously proposed values of 40s and 60s (31). The BP cutoff value was chosen to separate patients into approximately equally large subgroups. Using other cutoff values did not change our model.
There was no familiarization test before the study, which may lead to slightly lower Vo2max (11). However, inclusion of a second test, which has been performed in ∼60% of patients, did not influence the results (data not shown). Also, some of the sick patients did not reach steady state during low-intensity exercise, thereby prolonging MRT. However, this did not influence the prognostic value, as these patients were correctly classified.
Advances in medical therapy have significantly improved the prognosis in CHF patients. Thus, prognostic scores based on patient populations not treated with current therapeutic regimens may no longer be valid. Exercise testing had a prognostic impact, irrespective of whether patients were on or off beta-blockade (32). However, the prognosis of high-risk patients on beta-blockade was not worse than that of transplant candidates off beta-blockers, thus questioning the use of peak exercise testing for heart transplantation listing. We confirmed a better prognosis in patients receiving beta-blockers. However, the improvement in prognosis was primarily found in low- to medium-risk patients. The high-risk subgroup defined in this study did not benefit from beta-blocker therapy. Still, it must be pointed out that the number of high-risk patients classified into subgroups was small, and larger studies are needed to better define the prognostic impact of our model in patients receiving beta-blockade as well as angiotensin-converting enzyme inhibition. Importantly, other drug therapy did not influence the outcome or predictive power of our model (data not shown).
We used a protocol with a fixed treadmill speed and elevation rather than a fixed workload. Protocols for treadmill exercise testing generally use steps with fixed speed and elevation, and adjusting our results for body weight did not have any influence. Nevertheless, methodologic differences make a direct comparison between our present study and previous studies difficult, as may the lack of an accepted standard for assessment of Vo2kinetics. One could speculate that the higher the workload, the more oxygen kinetics become heart rate-dependent. This may explain why the correlation between MRT and Vo2max is better with a higher (33)than with lower workload (27,31,34).
The combination of BP and parameters of low-intensity and maximal exercise testing allows better risk assessment of CHF patients, as compared with assessment based on any single factor. Patients with a delayed increase in Vo2at the onset of exercise, low Vo2max (expressed as percent of predicted maximum), and low resting systolic BP are at a particularly high risk. The two-step exercise protocol test employed in this study is quick and no more expensive than determination of Vo2max alone. Prospective applications of the score will be required to assess its clinical value for risk assessment of CHF patients. Furthermore, it is obvious that a risk score such as the one provided here must be used in the clinical context of each individual patient only.
☆ This study was supported in part by Swiss National Funds grants NF 3200-056913 (Dr. Kiowski) and 32-59437-99 (Dr. Brunner-La Rocca).
- blood pressure
- chronic heart failure
- hazard ratio
- left ventricular assist device
- mean response time of oxygen uptake at onset of exercise
- ventilatory response to exercise
- oxygen consumption
- Received September 27, 2002.
- Revision received December 10, 2002.
- Accepted January 24, 2003.
- American College of Cardiology Foundation
- Mancini D.M.,
- Eisen H.,
- Kussmaul W.,
- Mull R.,
- Edmunds L.H. Jr..,
- Wilson J.R.
- Aaronson K.D.,
- Schwartz J.S.,
- Chen T.M.,
- Wong K.L.,
- Goin J.E.,
- Mancini D.M.
- Ponikowski P.,
- Francis D.P.,
- Piepoli M.F.,
- et al.
- Schalcher C, Rickli H, Bräm M, et al. Prolonged oxygen uptake kinetics during low intensity exercise are related to poor prognosis in patients with congestive heart failure. Chest 2003. In Press
- Sharma R.,
- Anker S.D.
- Koike A.,
- Hiroe M.,
- Adachi H.,
- et al.
- Brunner-La Rocca H.P.,
- Weilenmann D.,
- Follath F.,
- et al.
- Franklin B.A.,
- Whaley M.H.,
- Howley E.T.
- Wasserman K.
- Chua T.P.,
- Ponikowski P.,
- Harrington D.,
- et al.
- Hulsmann M.,
- Berger R.,
- Sturm B.,
- et al.
- Moser D.K.,
- Mann D.L.
- Costanzo M.R.,
- Augustine S.,
- Bourge R.,
- et al.
- Stelken A.M.,
- Younis L.T.,
- Jennison S.H.,
- et al.
- Pina I.L.
- Osada N.,
- Chaitman B.R.,
- Miller L.W.,
- et al.
- Francis D.P.,
- Shamim W.,
- Davies L.C.,
- et al.
- Pardaens K.,
- Van Cleemput J.,
- Vanhaecke J.,
- Fagard R.H.
- Lewalter T.,
- Rickli H.,
- MacCarter D.,
- et al.
- Zugck C.,
- Haunstetter A.,
- Kruger C.,
- et al.
- Koike A.,
- Yajima T.,
- Adachi H.,
- et al.