topeeng.gif (8383 bytes)

[ Scientific Activity - Actividad Científica ] [ Brief Communications - Temas Libres ]

Prognostic Stratification of Patients with Congestive Heart Failure by Cardiopulmonary Exercise Testing

Philippe Sellier, Marie Christine Iliou, Laurent Prunier, Jean Claude Verdier, Olivier Charon, Pierre Corona, Petr Kolar.

Hôpital Broussais, 96 Rue Didot, Paris, France.

Material and Methods


Background: prognostic value of peak VO2 has been demonstrated in patients (pts) with congestive heart failure (CHF). However, cut-off values are debated. Other variables may be useful to improve the detection of high risk pts.
Objective: to determine the prognostic value of other variables measured during exercise testing in pts with CHF.
Methods: 124 patients (115 male, mean age : 56.5 ± 12 years) with a history of CHF and depressed left ventricular function (radionuclide LV ejection fraction, LVEF < 0.45) underwent a symptom limited cardiopulmonary exercise test on a bicycle. Measured variables were peak VO2, anaerobic threshold (AT) , % of predicted peak VO2 (%PPVO2) and indexed peak oxygen pulse (IPOP). The main criteria was total mortality. Survival was determined by mailing questionnaires to all the patients, with a minimum of 6 months of follow-up.
Results: During the follow-up (mean : 22 ± 13 months), 16 patients died and 4 were lost to follow-up. Exercise variables at inclusion were as follows:

table.gif (3204 bytes)

The prognosis was significantly different (p = 0.004) between pts below or under the cut-off value of 0.65 for %PPVO2.
Conclusions: Cardiopulmonary exercise variables, especially %PPVO2, are useful to stratify the prognosis of patients with CHF. A low exercise capacity is in favour of an increased risk of future death.


Introduction: Heart failure is a disease whose frequency is increasing, associated to an important mortality and morbidity. Despite significant advances in medical therapy, the survival of these patients remains limited. Furthermore, the cost of care (especially rehospitalisation) is important. Cardiopulmonary exercise testing has reached an important place during the last years as a risk stratification tool for patients with heart failure. Among the evaluation variables, peak oxygen uptake appears as a valuable predictive factor of survival. More recently, it has been demonstrated that the use of indexed parameters improved the results of this categorization. However, cut-off values are still debated. The use of other variables may be useful to improve the detection of high risk patients.


Objectives: The aim of this study was to determine the prognostic value of the variables measured during exercise testing with gas exchange evaluation in patients with congestive heart failure.


Material and Methods: 124 successive patients were included in this study (115 men and 9 women, mean age : 56.5 ± 12 years. All the patients having a history of stable heart failure and an alteration of the left ventricular function, defined as a radionuclide left ventricular ejection fraction (LVEF) less than or equal to 0.40 without significant valvular disease were included in this study. The diagnosis of heart failure was made on the basis of usual criteria : exertional dyspnoea, peripheral oedema or pulmonary congestion. After a clinical evaluation (history, risk factors, clinical examination, chest X-ray), patients were submitted to a non-invasive testing of their cardiac status associating :

- a radionuclide assessment of the left ventricular ejection fraction by a technetium ventriculography.

- an exercise test with gas exchange evaluation. The symptom limited exercise protocol included a starting level of 30 watts and an increment of 10 watts by minute. Respiratory gas were analyzed using a CPX/D Medical Graphic system (St Paul, Minnesota, USA). The equipment was calibrated with a standard gas mixture before each test. The peak VO2 cut-off values used were those proposed by Mancini : = 10 ml/kg per min, > 10 to14 ml/kg per min, > 14 to 18 ml/kg per min and > 18 ml/kg per min. Contra-indications to cardiopulmonary exercise testing were the presence of uncontrolled heart failure, anginal pain or ventricular arrhythmias. The exercise test was performed without modification of the preexisting medical treatment and before any active exercise training.

Occurrence of events during the follow-up was evaluated by mailing questionnaires to all the included patients, at least six month after the last inclusion, in June 1998. The main evaluation criteria was total mortality. In case of death, the date and the cause were documented with the help of the general practitioner and the cardiologist.

Numerical values are expressed as means ± standard deviation. Differences between the means of groups were tested by the Student t test. The comparison of proportions was made by using the x2 test. A p value of 0.05 or less was considered as significant. Survival curves were made by the Kaplan-Meier method. The comparison of survival curves was made by the Log-Rank test. All the available data were submitted to an univariate analysis. The influence of a continuous variable on survival was analyzed by using a cut-off value corresponding to the median of this parameter. Multivariable regression analysis was performed on significant univariate predictors.


Results: The etiology of heart failure was ischaemic in 98 (79%) and non ischaemic in 26 patients (21%) of cases. At the inclusion, 32 patients (26%) were in class I of the NYHA, 47 (38%) in class II, 40 (32%) in class III and 54 % in class IV. Six patients (5 %) were in atrial fibrillation. The left ventricular function was clearly depressed with an average ejection fraction of 26.15 ± 8 %. Sixty six percent of the patients were on diuretics, 19 % on digitalics, 50 % on betablockers and 97.6 % on ACE inhibitors. The peak VO2 obtained at the maximum of the effort was on the average 18.7 ± 6 ml/kg per min. The anaerobic threshold (reached in 98 patients) was 13.5 ± 4.6 ml/kg per min. Using the classification of Mancini, 3 patients (2.4 %) were in class I, 25 patients (20.2 %) in class II, 47 patients (37.9 %) in class III, and 49 patients in class IV (39.5 %). Average oxygen pulse was 10.25 ± 3.4 ml/beat.

The mean follow-up was 22.1 ± 13 months. Four patients were lost to follow-up. During this period, there were 16 deaths (13 %), among which 14 were of cardiac origin. Furthermore, 4 patients underwent a cardiac transplantation. Twelve patients were rehospitalized for heart failure.

The clinical characteristics and data about LV function in surviving and non-surviving patients are summarized in Table 1. Some variables have a significant influence on prognosis such as the NYHA classification (p = 0.00002) and the presence of atrial fibrillation (p = 0.031). Interestingly, the level of LV dysfunction was not significantly different between survivors and non-survivors, although there was a non-significant trend (p = 0.07). Non-survivors had a significantly heavier treatment (93.8 % on diuretics vs 62 % in survivors, p = 0.26 and 43.8 % on digitalics vs 15.7 % in survivors, p = 0.21).

img001.gif (15998 bytes)

The comparison of cardiopulmonary exercise test data between survivors and non-survivors is described in Table 2. The patients who deceased during the follow-up had a shorter duration of effort (4.96 ± 2 min vs 6.62 ± 3 min, p = 0.03). The peak VO2 was significantly lower (15.2 ± 3.3 ml/kg per min vs 19.2 ± 6.3 ml/kg per min vs, p = 0.016). The anaerobic threshold was not significantly different between the two groups. The percentage achieved of predicted peak VO2 was clearly lower in non-survivors (56 ± 12 % vs 70 ± 21 %, p = 0.007). The oxygen pulse and the oxygen pulse indexed to the weight were also lower in non-survivors (8.6 ± 2.5 ml/beat vs 10.5 ± 3.5 ml/beat, p = 0.031 and 0.12 ± 0.02 ml/Kg per beat vs 0.15 ± 0.04 ml/Kg per beat, p = 0.012).

img002.gif (17303 bytes)

Peak oxygen consumption, using the Mancini classification, has no significant predictive value of death (p = 0.284), because of the overlap between different classes (Figure 1). At the opposite, indexed peak oxygen consumption (Figure 2), using a cut-off value defined by the median of the values, had a clear prognostic value (p = 0.0046). The indexed peak oxygen pulse (cut-off value of 0,142 ml/kg per beat) was also predictive of survival (Figure 3), but with a lower statistical significance (p = 0.030). Furthermore, in the group of patients having the lower indexed peak VO2 (< 65 % of predicted values), all the patients who died (Figure 4)were in the lower class of indexed peak oxygen pulse (< 0.142).

img001.gif (15998 bytes)
Fig. 1

img002.gif (17303 bytes)
Fig. 2

img005.gif (11852 bytes)
Fig. 3

img006.gif (13132 bytes)
Fig. 4

Multivariate Cox regression analysis of exercise variables (Table 3)showed significant results for peak exercise systolic blood pressure (p = 0.002), percentage achieved of predicted V02 (p = 0,0302). There was a non-significant trend in favor of the prognostic value of indexed peak oxygen pulse (p = 0,0659).

img003.gif (12250 bytes)


Discussion: This study shows that in patients with heart failure and LV dysfunction, the best independent exercise predictor of death is the indexed peak V02. In our experience, the usual classifications based on peak VO2 itself, such as the one proposed by Mancini, has a lower predictive value, due to overlap between classes. The indexed peak oxygen pulse may help to detect patients who are at risk of future death, but this trend needs further confirmation. This study also remind us that peak systolic blood pressure measured during the exercise has also a powerful predictive value of future deaths. Indexed peak oxygen pulse, more dependant on the cardiac pump function, has an additive value to peak oxygen consumption to better detect high risk patients. Interestingly, the clinical variables (age, gender, etiology of the disease), except the NYHA classification, have no predictive value of survival in this study. At the opposite, the cardio thoracic ratio is related to the risk of future death. The absence of any significant influence of LV function on the prognosis may be explained by the fact that all this population was selected on the basis of a confirmed LV dysfunction.

Conclusions: We found that indexed peak oxygen consumption was the best predictor of survival at long term in patients with heart failure and LV dysfunction. Other variables, such as indexed peak oxygen pulse, may help to further improve risk stratification.


Questions, contributions and commentaries to the Authors: send an e-mail message (up to 15 lines, without attachments) to , written either in English, Spanish, or Portuguese.




These Companies contributed to the Congress

bagocard.gif (2719 bytes)glicar.gif (3465 bytes)

klchlogo4.gif (6328 bytes)