Author + information
- Received August 19, 2002
- Revision received November 11, 2002
- Accepted November 27, 2002
- Published online June 18, 2003.
- Joao J. Leite, MD, PhD*,* (, )
- Alfredo J. Mansur, MD, PhD*,
- Humberto F.G. de Freitas, MD*,
- Paulo R. Chizola, MD*,
- Edimar A. Bocchi, MD, PhD*,
- Mario Terra-Filho, MD, PhD*,
- J.Alberto Neder, MD, PhD† and
- Geraldo Lorenzi-Filho, MD, PhD*
- ↵*Reprint requests and correspondence:
Dr. Joao J. Leite, Instituto do Coração, Rua Dr. Eneas Carvalho de Aguiar 44, CEP: 05403-000, São Paulo-SP, Brazil.
Objectives We hypothesized that exercise-related periodic breathing (EPB) would be associated with poor prognosis in advanced chronic heart failure (CHF).
Background Patients with CHF might present instability of the ventilatory control system characterized by cyclic waxing and waning of tidal volume (periodic breathing [PB]). This condition is associated with several deleterious circulatory and neuro-endocrine responses; in fact, PB in awake and asleep patients has been identified as an independent risk factor for cardiac death. During exercise, however, the prognostic value of PB is still unknown in CHF patients awaiting heart transplantation.
Methods Eighty-four patients with established CHF (65 male, 19 female) were submitted to clinical evaluation, echocardiogram, ventricular scintigraphy, determination of resting serum norepinephrine levels, and an incremental cardiopulmonary exercise test on cycle ergometer. Patients were followed for up to 49.7 months (median = 15.3), and 26 patients (30.9%) died during this period.
Results Twenty-five of 84 patients presented EPB (29.7%). The following variables were related to mortality according to Kaplan-Meier and univariate Cox regression analysis: EPB (p = 0.004), New York Heart Association class (p = 0.04), serum norepinephrine (p = 0.06), peak oxygen uptake (ml·min−1·kg−1and % predicted; p = 0.085 and p = 0.10, respectively), slope of the ratio of change in minute ventilation to change in carbon dioxide output during exercise (p = 0.10), and scintigraphic left ventricular ejection fraction (p = 0.10). Cox multivariate analysis identified EPB as the only independent variable for cardiac death prediction (p = 0.007). Therefore, EPB alone was associated with a 2.97-fold increase in risk of death in this population (95% confidence interval = 1.34 to 6.54).
Conclusions Exercise-related periodic breathing independently predicts cardiac mortality in CHF patients considered for heart transplantation.
Chronic heart failure (CHF), a leading cause of morbidity worldwide, is closely related to increased prevalence of lethal cardiac events (1). Recent developments in surgical treatment of CHF imply that patients at high risk of such catastrophic outcomes should be actively sought for. Accordingly, several predictors of poor clinical prognosis in CHF have been recognized: low left ventricular ejection fraction (LVEF) (2), enlarged left atrium with mitral regurgitation (3), increased left ventricular cavity width (4), and increased sympathetic nerve activity (5). Unfortunately, however, this extensive group of predicting factors still lacks accuracy for outcome prediction, and the prevalence of fatal cardiac events in CHF populations is still excessive (1).
Exercise intolerance is a hallmark of advanced, symptomatic CHF (1). Low maximum oxygen consumption (peak O2), for instance, has been widely used for risk stratification in this patient population (6–8). Peak O2, however, has a limited prognostic power even when circulatory-limited patients are evaluated (9,10). More recently, an increased ventilatory response to the metabolic demand, as inferred by a steeper slope of the ratio of change in minute ventilation to change in carbon dioxide output (ΔE/ΔCO2) during progressive exercise, has been proposed to improve outcome prediction (11,12). In fact, hyperventilation and low arterial CO2at rest and during sleep (13)are associated with cyclic waxing and waning of tidal volume (VT) in CHF patients, a condition known as periodic breathing (PB) (13,14). Interestingly, a severe form of PB characterized by ventilatory oscillations interposed with central apneas or hypopneas (Cheyne-Stokes respiration [CSR]) has been identified as an independent predictor of mortality in awake (15)and asleep patients with CHF (16,17). In this context, Corrà et al. (18)have recently shown that exercise-related PB (EPB) also holds prognostic implications in a clinically heterogeneous group of cardiac patients who have been submitted to cardiopulmonary exercise testing. Considering that the prognostic value of a given variable is highly dependent on specific population characteristics, it is not known whether these findings could be extrapolated for patients at higher risk of cardiac death (i.e., those with end-stage disease considered for heart transplantation).
We therefore sought to investigate the clinical value of EPB in predicting cardiac death in a group of patients with advanced CHF who were under evaluation for heart transplantation. To gain further insight into the pathophysiologic mechanisms underlying this condition, we also contrasted the functional characteristics of the ventilatory control system in patients with and in those without EPB.
The study group was comprised of 84 consecutive patients (mean age [± SD] = 45 ± 10 years, 65 males and 19 females) with an established diagnosis of CHF who had clinical and/or functional indication for cardiac transplantation. Forty-four patients were classified as New York Heart Association (NYHA), class III and the remaining patients as class II. Although 12 patients were initially classified as class IV, no patients remained in this class after optimal clinical management. Inclusion criteria were as follows: more than six-month history of CHF (actual values being 73.9 ± 48.0 months), unchanged medication within the last month, and no acute coronary event within six months preceding the study. Patients with body mass index above 30 kg·m−2and other clinical conditions such as hepatic, hematologic, renal, infectious, or pulmonary diseases (including smokers) were also excluded because they are not considered suitable candidates in our transplantation program (19).
Etiology of heart failure was dilated cardiomyopathy in 48 cases (56.5%), ischemic in 15 (17.7%), Chagas’ disease in 11 (12.9%), alcoholic in 4 (4.7%), rheumatic in 5 (5.9%), and post-partum in 2 patients (2.4%). Data were pooled for analysis only after certification that there was no significant effect of specific disease prevalence on EPB frequency. Current medication included digoxin (92%), angiotensin-converting enzyme inhibitors (88%), furosemide (88%), hydrochlorotyazide (30%), potassium chloride (19%), beta-blockers (6%), amiodarone (11%), spironolactone (7%), and nitrates (7%). Seventy-three patients (86.9%) presented in sinus rhythm, and 11 (13.1%) were in chronic atrial fibrillation. Informed consent as approved by the Institutional Medical Ethics Committee was obtained from all subjects.
Doppler echocardiogram with color flow imaging was performed in A and B modes using a multifrequency electronic transducer (Apogee 800Plus, Advanced Technology Laboratories, Bothell, Washington). Gated blood pool scintigraphy in a semi-recumbent position was carried out with a scintigraphic chamber LEM-Plus (Siemens Gammasonics Inc., Des Paines, Illinois) and a low-energy collimator using intravenous pyrophosphate followed by injection of erythrocytes labeled with technetium-99m (30 mCi). Electrocardiogram-triggered diastolic and systolic images (64 × 64 pixels) were obtained in left anterior oblique view for the LVEF calculation. Serum epinephrine levels were measured after 1 h of rest from antecubital venous blood (n = 64) using a high-performance liquid chromatography-pulsed electrochemical detector (model 464, Waters, Milford, Massachusetts).
Cardiopulmonary exercise tests were performed on an electromagnetic-braked cycle ergometer following a ramp-incremental protocol to the limit of tolerance. Pulmonary gas exchange and ventilatory variables were obtained from calibrated signals derived from rapidly responding gas analyzers and a pneumotachograph (CPX System, Medical Graphics Corp., St. Paul, Minnesota). The following variables were recorded breath-by-breath and expressed as the moving average of eight breaths: pulmonary oxygen uptake (O2, l·min−1), pulmonary carbon dioxide output (CO2, l·min−1), minute ventilation (E, l·min−1), and end-tidal partial pressure for CO2(PETCO2, mm Hg). Peak O2values were compared with those predicted by Hansen et al. (20), considering gender, age, and height. The slope of ΔE/ΔCO2was determined by regression analysis throughout the linear phase of this submaximal relationship (11,12).
Assessment of PB
Periodic breathing during exercise was established [EPB(+)] according to the following criteria present on ventilation measurements: 1) three or more regular oscillations (i.e., clearly discernible from inherent data noise); 2) regularity was defined if the standard deviation of three consecutive cycle lengths (time between 2 consecutive nadirs) was within 20% of the average; 3) minimal average amplitude of ventilatory (E) oscillation of 5 l (peak value minus the average of two in-between consecutive nadirs) (Fig. 1).
Follow-up events on prognosis
The study end point was cardiac death. The four heart transplantation cases were considered as censored data at the day of surgery. The patients were followed for up to 49.7 months (median 11.3 months; interquartile range 6.0 to 21.8 months) on regular visits to the outpatient clinic at least twice a year. Information of the absent patients was obtained by telephone call; all deaths were confirmed through evaluation of medical recordings.
Mean and standard deviations (SD) were obtained for values in subjects grouped according to clinical outcome and presence or not of EPB. Kaplan-Meier log-rank and univariate Cox regression analysis were used to correlate categorical or continuous variables to time-to-event, respectively. Meaningful correlated variables (p < 0.10) were then included in a multivariate stepwise forward conditional Cox regression analysis. Significance of the regression coefficients was assessed by considering the asymptotic normality property of maximum likelihood estimates (Wald statistic). Serum norepinephrine data were excluded from the multivariate analysis because only 62 patients had serum determination. Kaplan-Meier cumulative survival curves were constructed and compared using the Mantel-Haenszel log-rank test. Student ttest and chi-squared tests were applied as appropriate for between-groups comparison. The p value of a type I error was established at 0.05 for all hypothesis tests.
Maximum exercise capacity was consistent with that expected in patients with advanced CHF; actual peak O2values were 16 ± 5.2 ml·min−1·kg−1·min−1or 51 ± 15.1% of predicted. Resting echocardiographic data showed severe heart damage with reduced LVEF (35 ± 6.6%) and increased left ventricle end-diastolic and end-systolic and left atrium diameters (74 ± 11.4 mm, 63 ± 12.0 mm, and 47 ± 7.5 mm, respectively). Similarly, scintigraphic evaluation showed a severely reduced LVEF (22 ± 7.4%), and resting serum norepinephrine levels were typically elevated (582 ± 278.6 μg·dl−1).
Clinical and functional correlates of EPB
Twenty-five of 84 patients (29.7%) presented EPB; cardiac deaths were identified in 26 subjects (30.9%). The one-year and two-year cumulative survival rates were 82% and 57%, respectively. We first investigated possible clinical and functional differences between EPB(+) and EPB(−) patients. As shown in Table 1, EPB(+) patients presented lower peak O2, higher ΔE/ΔCO2slope, and lower pre-exercise PETCO2values than EPB(−) patients (p < 0.01). On the other hand, there were no significant between-group differences in relation to the other variables, including LVEF. We also found similar time delays from initial symptoms of CHF to exercise testing in both patient groups (22.9 ± 7.5 vs. 20.6 ± 5.8 months; p = 0.21). Furthermore, we found a marginally significant association between higher functional classification and EPB: 17 of 25 of EPB(+) patients (68%) were classified as NYHA class III, but only 27 of 59 of EPB(−) patients (45.7%) were in this class (p < 0.06).
EPB and cardiac death
Time-corrected univariate regression analysis depicted the following variables as predictors of mortality: EPB(+) (log-rank = 7.98; p = 0.004); NYHA functional classification (log-rank = 4.00, p = 0.04); serum norepinephrine (Wald =3.50; p = 0.06); O2[ml·min−1·kg−1] (log-rank = 2.95; p = 0.085) and [% predicted] (Wald = 2.68; p = 0.10); ΔE/ΔCO2slope (Wald = 2.65; p = 0.10); and scintigraphic LVEF (Wald = 2.59; p = 0.10). Conversely, age, gender, resting PETCO2values, and echocardiographic LVEF were not significantly associated with mortality (p > 0.10).
The multivariate analysis demonstrated that EPB was the only independent predictor of cardiac death (i.e., no other variable could be added to the final model) (p = 0.007). Therefore, the risk of death for EPB(+) patients was 2.97 times greater than for EPB(−) patients (95% confidence interval = 1.34 to 6.54). Figure 2depicts the Kaplan-Meier survival curves for both groups: one- and two-year cumulative survival rates were 89% to 63% and 67% to 45% for EPB(−) and EPB(+) patients, respectively. Similar results were obtained when we considered transplantation (n = 4) as the final event.
We have prospectively investigated the role of EPB in predicting cardiac death in a group of patients with advanced CHF under evaluation for heart transplantation. The novel findings of our study were: 1) EPB was frequent (about 30%) and easily recognized by using clinically useful criteria; 2) it almost tripled the average risk of cardiac death (Fig. 2), independent of traditional demographic, clinical, and resting functional variables; and 3) EPB was associated with steeper ΔE/ΔCO2slope during exercise and low PETCO2at rest (i.e., excessive ventilatory response). This seems to constitute the first study to demonstrate the practical value of this disorder in predicting mortality in patients with end-stage CHF in evaluation for heart transplantation. Our results, therefore, suggest that evidence of EPB during routine cardiopulmonary exercise testing should be sought and clinically valued for risk stratification in this specific patient population.
Cardiopulmonary exercise testing has an established role for prognostic evaluation of patients with CHF. It is well known, for example, that reduction of peak O2relates to poor prognosis in this population (6,7)—as we could confirm in the present study. However, interpretation of peak O2values is fraught with complexities in patient populations. The multifactorial etiology of exercise limitation in this syndrome implies that limited cardiac output might not be the sole culprit for an eventually reduced peak O2. In fact, other cardio-respiratory and metabolic disturbances are common in CHF: recent data demonstrate that evaluation of submaximal exercise is useful to unravel these disorders (11,12,21). In this context, our findings are particularly relevant because identification of EPB does not depend on patients’ maximal effort or attainment of a truly circulatory-limited peak O2.
Prognosis of EPB
Respiratory instability has long been associated with poor prognosis in CHF (14). In fact, evidence that this disorder is frequent among CHF patients (22,23)and is an independent and powerful predictor of mortality, whether during the day (15)or night (16,17), is accruing. Several studies have now demonstrated that CSR during sleep is associated with recurrent hypoxia, increased number of awakenings, enhanced sympathetic activity, and poor prognosis (13,14,16,17,22,23). In these patients, episodes of PB induce fluctuations in heart rate, blood pressure, and gases that are associated with activation of neuro-endocrine responses and a state of chronic sympathetic over-activity—well-known triggers of lethal ventricular arrhythmias (14). Indeed, Javaheri and Corbett (24)found a high prevalence of ventricular tachycardia in CHF patients with hypocapnia and central apnea, suggesting a possible mechanism for the cardiac deaths related to CSR. The same pathophysiologic mechanisms could be associated with mortality in EPB: several daily episodes of exertion-related ventilatory instabilities would elicit a chronic status of systemic neuro-endocrine and circulatory over-activation with harmful effects on the failing heart.
More recently, Corrà et al. (18)found a significant association between EPB and cardiac death in a study involving a population of CHF patients in different stages of the disease who were referred for CPX evaluation. Our longitudinal study further extends their findings by showing that in a selected group of very limited patients considered for cardiac transplantation, EPB is more frequent (30% vs. 12%) and an even more powerful predictor of mortality than in their series of patients with a broader clinical spectrum. In fact, we were unable to add any other variable in the predictive model for cardiac death when EPB was considered. This finding contrasts with those of Corrà et al., who found that peak O2and LVEF were also predictors of major cardiac events. Interestingly, our data suggest that identification of EPB is not related to duration of the disease: we were unable to find significant differences in time delay from initial symptoms of CHF to exercise testing in EPB(+) versus EPB(−) patients.
Our data, therefore, seem to be particularly relevant in clinical terms, because this sample epitomizes a population under increased risk of sudden cardiac death (i.e., a group for which ominous predictors should be especially sought). In addition, the present study shed light on the pathophysiologic determinants of EPB: an excessive ventilatory response to the metabolic demand is likely to play a role in the genesis of this disorder (Table 1).
Relationship between CHF and periodic breathing
Periodic breathing is characterized by a periodic and cyclic waxing and waning of VT. Cheyne-Stokes respiration, for instance, is a severe form of PB in which apneas or hypopneas are alternated with periods of ventilatory oscillation. The mechanisms involved in the etiology of PB are multifactorial and not completely understood. It seems likely, however, that it represents a clinical manifestation of instability of the respiratory control center (14).
It has long been recognized that the central respiratory controllers are strongly influenced by chemoreceptor activity, especially by the level at which the mean arterial pressure for CO2(PaCO2) is regulated (CO2set-point) (14). In fact, low PaCO2plays a major role in the pathogenesis of CSR in CHF (13,14). It has been demonstrated, for instance, that these patients are typically hypocapnic at rest and asleep (13). Furthermore, inhalation of low concentrations of CO2during sleep is able to abolish CSR in conjunction with a small, but significant, increase in PaCO2(25). In CHF patients, hypocapnia might be caused by over-stimulation of the ventilatory control center due to chronic pulmonary congestion (26,27)and increased central (28)and peripheral (29)chemosensitivity. Circulatory delay could also increase the lag time between pulmonary gas-exchanging capillaries and carotid chemoreceptors with consequent increase in gain of the feedback loop. These effects are prone to further contribute to the underlying instability of the respiratory center (14).
Periodic breathing during exercise
The physiologic adjustments during dynamic exercise can produce additional sources of excessive ventilatory response and, therefore, promote further instabilities (30,31). Exercise stimulation of the over-active ergoreceptors during exercise may excessively stimulate the respiratory centers. Group III and IV afferents can respond to accumulated by-products of muscle metabolism and changes in local vascular peripheral conductance (32). Exercise-induced pulmonary capillary congestion, as we have previously demonstrated (33), may also activate lung stretch receptors (14,27). In addition, an excessive ventilatory response could be associated with increased “wasted” ventilation due to enlarged physiologic dead space volume (Vds) and/or reduced VT (i.e., higher Vds/VT ratio during exercise) (34).
The linear relationship between E and CO2that characteristically develops during most of the incremental phase of dynamic exercise (ΔE/ΔCO2) has been used as an index of the ventilatory efficiency based on the following reasoning (34): From this construct, it is clear that an increased ΔE/ΔCO2slope might be related to reduced CO2set-point (i.e., hyperventilation) and/or an enlarged dead space fraction of the breath (i.e., a tachypneic breathing pattern). In our study, EPB(+) patients presented not only higher ΔE/ΔCO2slope but also lower resting PETCO2(Table 1). Unfortunately, however, excessive Vds is associated with a positive arterial-end tidal CO2pressure gradient (i.e., PETCO2can underestimate PaCO2by an unknown amount when the “wasted” ventilation is increased). Considering that we did not directly measure PaCO2or other indexes of pulmonary gas exchange efficiency, we cannot rule out that Vds/VT was disproportionately increased in the more severe patients (i.e., those who presented higher prevalence of EPB) (Table 1). Even taking into consideration these limitations, we consider that our data are consistent with the hypothesis that excessive ventilation during exercise might further contribute to respiratory instability and, therefore, development of EPB in end-stage CHF.
Although the criteria we used to define EPB could have underestimated its true prevalence, we reasoned that a more sophisticated analysis (such as power spectral analysis) (35)would be prone to hamper the practicality of this approach in clinical settings. Indeed, similar criteria for EPB identification have been recently demonstrated to be reliable and reproducible (18); nevertheless, we recognize that the present study lacks a formal test–retest analysis for EPB reproducibility.
We found only a weak association between LVEF and mortality. This finding could be related to the small variability of this measurement in a relatively homogeneous population with advanced disease. However, it is worth noting that other well-known predictors of mortality, such as functional status, peak O2, and norepinephrine levels, were significantly related to death (see Results). It is conceivable that the introduction of newer, more effective drugs after the initial cardiopulmonary exercise testing may have favorably changed the prognosis and, therefore, may lessen the importance of PB in predicting mortality. In addition, some specific characteristics of our sample should be taken into consideration for analysis of the study results. First, no NYHA class IV patient was evaluated; as discussed in Methods, however, 12 patients had been classified as class IV before optimal clinical management. Second, Chagas’ disease was the specific etiology of CHF in 12.9% of the patients; as mentioned, we were unable to identify an independent effect of CHF etiology on EPB frequency. Finally, these patients were typically younger than those evaluated by other authors. Although the effect of age on EPB is to date unknown, we cannot rule out the possibility that our main outcome was influenced by this demographic characteristic.
Our study provides original evidence that in CHF patients awaiting heart transplantation EPB is common, can be easily recognized, and constitutes a powerful predictor of mortality. Therefore, EPB on routine cardiopulmonary exercise testing should be looked for, reported, and clinically valued for risk stratification in CHF patients considered for heart transplantation.
We thank Leila Marise de Oliveira for her dedicated work at the Laboratory of Cardiopulmonary Exercise Testing of InCor. The authors are also grateful to Mariana Cúri for her assistance in the statistical analysis.
- chronic heart failure
- Cheyne-Stokes respiration
- exercise-related periodic breathing
- left ventricular ejection fraction
- New York Heart Association
- arterial carbon dioxide pressure
- periodic breathing
- end-tidal partial pressure for carbon dioxide
- carbon dioxide output
- dead space volume
- minute ventilation
- oxygen uptake
- tidal volume
- Received August 19, 2002.
- Revision received November 11, 2002.
- Accepted November 27, 2002.
- American College of Cardiology Foundation
- Garg R.,
- Yusuf S.
- Aaronson K.D.,
- Schwartz J.S.,
- Chen T.,
- et al.
- Madsen B.K.,
- Keller N.,
- Christiansen E.,
- et al.
- Mancini D.M.,
- Eisen H.,
- Kussmaul W.R.,
- Mull L.,
- Edmunds H.,
- Wilson J.R.
- Van den Broek A.S.,
- Van Veldhuisen D.J.,
- De Graeff P.A.,
- et al.
- Mudge G.H.,
- Goldstein S.,
- Addonizio L.J.,
- et al.
- Opasich C.,
- Pinna G.D.,
- Bobbio M.,
- et al.
- Chua T.P.,
- Ponikowski P.,
- Harrington D.,
- et al.
- Lorenzi-Filho G.,
- Bradley D.
- Lanfranchi P.A.,
- Braghiroli A.,
- Bosimini E.,
- et al.
- Hollenberg M.,
- Tager I.B.
- Javaheri S.,
- Parker T.J.,
- Liming J.D.,
- et al.
- Solin P.,
- Bergin P.,
- Richardson M.,
- et al.
- Lorenzi-Filho G.,
- Azevedo E.R.,
- Parker J.D.,
- Bradley T.D.
- Ponikowski P.,
- Chua T.P.,
- Anker S.D.,
- et al.
- Piepoli M.,
- Clark A.L.,
- Volterrani M.,
- et al.
- Francis D.,
- Davies L.C.,
- Piepoli M.,
- et al.
- Haouzi P.,
- Hill J.M.,
- Lewis B.K.,
- et al.
- Brasileiro F.C.,
- Leite J.J.,
- Vargas F.S.,
- et al.
- Whipp B.J.,
- Pardy R.L.