Author + information
- Received October 29, 2008
- Revision received January 27, 2009
- Accepted February 3, 2009
- Published online April 28, 2009.
- Alessandro Giardini, MD, PhD*,‡,* (, )
- Alfred Hager, MD†,
- Astrid E. Lammers, MD‡,
- Graham Derrick, MD‡,
- Jan Müller, MSc†,
- Gerhard-Paul Diller, MD§,
- Konstantinos Dimopoulos, MSc, MD§,
- Dolf Odendaal, MSc‡,
- Gaetano Gargiulo, MD*,
- Fernando M. Picchio, MD* and
- Michael A. Gatzoulis, MD, PhD§
- ↵*Reprint requests and correspondence:
Dr. Alessandro Giardini, Pediatric Cardiology and Adult Congenital Unit, University of Bologna, Via Massarenti 9, 40138 Bologna, Italy
Objectives The goal of this study was to assess the prognostic value of the cardiopulmonary exercise test (CPET) in patients who received a Mustard and Senning (M/S) operation.
Background Patients who received an M/S operation have increased long-term risk of cardiovascular morbidity and mortality. Limited information is available on how to stratify risk in this population.
Methods Between 1996 and 2007, 274 adults (age 26.3 ± 8.9 years, range 16 to 50 years) who had received a Mustard (n = 144) or Senning (n = 130) operation in infancy were studied with CPET. During a follow-up of 3.9 ± 2.3 years (range 0.2 to 10.8 years), 12 patients died at an age of 36 ± 14 years, and 46 patients required a cardiac-related emergency (<24 h from the onset of symptom/condition) hospital admission at an age of 30 ± 11 years.
Results At multivariate Cox analysis, the slope of ventilation per unit of carbon dioxide output (VE/VCO2slope) (hazard ratio: 1.088, p < 0.0001) and percentage of predicted peak oxygen uptake (Vo2%) (hazard ratio: 0.979, p = 0.0136) were the strongest predictors of death/cardiac-related emergency hospital admission among demographic, clinical, and exercise variables. A VE/VCO2slope ≥35.4 (hazard ratio: 10.7, 95% confidence interval [CI]: 7.8 to 24.6), and a peak Vo2% ≤52.3% (hazard ratio: 3.4, 95% CI: 2.5 to 8.2) were associated with an increased 4-year risk of death/cardiac-related emergency hospital admission. Patients who had both a VE/VCO2slope ≥35.4 and a peak Vo2% ≤52.3% of predicted value were at highest risk (4-year event rate: 78.8%).
Conclusions CPET provides important prognostic information in adults with M/S operation. Subjects with enhanced ventilatory response to exercise or those with poor exercise capacity have a substantially higher 4-year risk of death/cardiac-related emergency hospital admission.
The development of atrial redirection procedures represented by the Mustard and Senning (M/S) operations revolutionized the care of patients with transposition of the great arteries (TGA) (1,2). Although these procedures have largely been superseded by the arterial switch operation, there remains a large cohort of young adults who underwent atrial redirection in childhood. There are a number of concerns regarding the long-term outcome in this patient group. Baffle obstruction (3), dysfunction of the systemic right ventricle (RV) (4), exercise intolerance (5,6), and increased risk of arrhythmia and sudden death (7) have all been reported.
As a consequence of these, patients with M/S repair for TGA are at increased risk of premature death. Most recent series show that about 75% to 90% of patients are alive 25 years after the operation (8). Most deaths seem to be related to progressive heart failure (HF) and sudden death (9).
Peak oxygen uptake (Vo2) has been the first cardiopulmonary exercise test (CPET) variable to demonstrate prognostic value (10) in adults with HF related to acquired heart disease. It remains today the most frequently analyzed variable in clinical practice. However, more recently, several investigations have shown that ventilatory efficiency, typically expressed as the minute ventilation/carbon dioxide production (VE/VCO2) slope, is an even stronger prognostic marker in patients with HF (11,12).
Cardiopulmonary exercise testing has been used in M/S patients for the quantification of exercise intolerance (13), but the ability of CPET as a prognostic index in this population has never been assessed. Identification of the highest risk M/S patients has the clear potential to assist in clinical decision-making (e.g., by intensified medical therapy or ultimately heart transplantation). Although recent investigations have shown that VE/VCO2slope and heart rate (HR) reserve may have a prognostic value in the overall group of adults with noncyanotic congenital heart disease (14,15), disease- and operation-specific data are unavailable, therefore limiting the clinical applicability of such finding.
The purpose of the present study was, thus, to assess the potential value of CPET to predict clinically significant cardiovascular end points in adult M/S patients.
This study was designed as a multicenter retrospective investigation. Across the 4 institutions involved in the present study, all M/S patients who are able to exercise routinely undergo CPET as part of clinical follow-up. All consecutive M/S patients with an age at test ≥16 years who underwent a CPET between March 1996 and May 2007 were included. The study consisted of 274 patients with a mean age of 26.3 ± 8.9 years (range 16 to 50 years). The underlying congenital heart lesion was classified as simple versus complex TGA according to the absence or presence of an associated ventricular septal defect. Patients with left or right ventricular outflow tract obstruction were excluded from the present study. Informed consent was obtained from all patients before undergoing exercise testing. The institutional committees on human research approved the present retrospective study.
Follow-up and analysis of survival status and hospitalization
The end point of the study was a combination of mortality and emergency cardiac-related hospital admission. All-cause mortality was used as the end point to eliminate any possibility of bias arising from incorrect classification of cause of death. Cardiac-related hospital admission was defined as any admission directly caused by failure of the cardiac system needing inpatient care to correct. Examples of cardiac admissions are symptomatic cardiac arrhythmias and decompensated HF. Emergency admission was defined as an admission occurring within 24 h of onset of symptom/condition. Subjects in whom hospitalization was of a noncardiac cause, or admission was delayed for >24 h from the onset of symptoms, were treated as censored cases. Patients who underwent elective heart transplantation or elective cardiac surgery were treated as censored cases, and follow-up was censored at the time of the operation.
After the exercise tests, all patients were regularly followed up for cardiac-related events at their respective institutions, which provided for the highest likelihood that all events were captured. Additionally, patients' medical records were reviewed to abstract the dates, timing of, and medical reasons for hospitalization and their survival status. Survival status was further ascertained from phone interviews with patients' primary care physicians.
Cardiopulmonary exercise test
Exercise tests were performed on an electronically braked ergometer cycle (n = 206) or on a treadmill (n = 68). Carbon dioxide elimination, Vo2, and minute ventilation were measured with a computerized breath-by-breath analyzer. Patients and controls performed a maximal exercise test using an incremental protocol that allowed reaching exhaustion in approximately 10 min of exercise. Criterion for test ending was considered patient exhaustion with a respiratory exchange ratio ≥1.09. A 12-lead electrocardiogram and transcutaneous oxygen saturation were also continuously monitored throughout the study, and cuff blood pressure was determined manually every 2 min. The technical details of measurement of peak Vo2and VE/VCO2slope were previously published (16). Resting HR was measured after at least 2 min of complete rest in a seated position, and peak HR was defined as the maximal HR achieved during exercise. Predicted maximum HR was estimated according to the Astrand formula (220 − age) (17). Heart rate reserve was calculated as the difference between peak and resting HR (14). None of the patients had known coronary artery disease or inability to exercise for other reasons. Before exertion, a spirometric measurement was performed to assess forced vital capacity, and forced expired volume in the first second. Standard equations were used to generate predicted values for baseline spirometric and peak exercise parameters (18). Because of age-related differences of normal peak Vo2when expressed in ml O2/kg/min in a patient cohort with a large age range, peak Vo2was expressed as percent of predicted. Cyanosis was defined as arterial blood saturation <90% at rest or during exercise (15).
Assessment of RV function, tricuspid regurgitation, pulmonary arterial hypertension, and baffle status
Right ventricular function was semiquantitatively estimated by echocardiography from parasternal long- and short-axis views and from an apical 4-chamber view. It was quantified as normal, mildly impaired, moderately impaired, or severely impaired. Tricuspid regurgitation was semiquantitatively estimated by echocardiography from the apical 4- and 2-chamber views and was reported as absent/trivial, mild, moderate, or severe. The presence of pulmonary hypertension (PH) was estimated either by right heart catheterization (mean pulmonary arterial pressure >25 mm Hg) or by echocardiography (right ventricular systolic pressure >50 mm Hg, calculated using the modified Bernoulli equation from tricuspid regurgitation jet velocity). Baffle status was routinely assessed by transthoracic echocardiography. Patients with suspected baffle stenosis or residual leak were assessed with cardiac magnetic resonance imaging or cardiac catheterization.
Values are presented as mean ± SD or n (%) as appropriate. Comparisons between subgroups were performed by unpaired tor chi-square tests, as appropriate. The prognostic value of different demographic and exercise variables (age at CPET, type of atrial switch procedure, presence of pacemaker, VE/VCO2slope, peak Vo2%, and HR reserve) was assessed with univariate and multivariate Cox analyses. The hazard ratio with 2-sided 95% confidence interval (CI) is given. Receiver-operator characteristic (ROC) curve analysis was used to identify the cutoff value of VE/VCO2slope and of the peak Vo2% associated with worse outcome, using data censored at 4-year follow-up. The cutoff value was chosen according to the highest likelihood ratio. Kaplan-Meier survival charts were generated to determine a difference in mortality/cardiac-related emergency hospital admission rate between subjects above or below the cutoff values identified. The log-rank test was used to determine differences in event-free survival between patients above or below the cutoff values. The effect of different demographic (age at operation, age at CPET, type of atrial switch procedure, and presence of pacemaker), hemodynamic (presence of PH, and RV function), and exercise variables (HR reserve) on VE/VCO2slope and peak Vo2% was assessed by single and multiple regression. A 2-tailed probability value ≤0.05 was used as the criterion for statistical significance.
The baseline characteristics of the 274 M/S patients included in this study are presented in Table 1.One hundred seventy-seven patients (64%) were males. Simple TGA was diagnosed in 217 patients (79.2%). Patients with a Mustard operation (n = 144) were significantly older at the time of the atrial switch operation (35 ± 37 months vs. 29 ± 32 months, p = 0.017) and at the time of CPET (28.0 ± 8.7 years vs. 24.4 ± 8.8 years, p = 0.0006) than patients with a Senning operation (n = 130), whereas the prevalence of patients requiring a permanent pacemaker was similar (9.7% vs. 7.7%, p = 0.670). No patient was cyanotic at rest. Nine patients (3%) had cyanosis during or at peak exercise.
Follow-up was complete for all patients. At an average follow-up of 3.9 ± 2.3 years (range 0.2 to 10.8 years), 58 patients (21%) reached the composite end point of death or cardiac-related emergency hospital admission (Fig. 1).The annual event rate for the composite end point was 5.9%. The annual death rate was 1.1%. In detail, 12 patients (4.4%) died at a mean age of 36 ± 14 years (range 18 to 50 years), whereas 46 patients (16.8%) needed emergency hospital admission at a mean age of 30 ± 11 years (range 16 to 47 years). Causes of hospital admission were decompensated HF requiring intravenous therapy in 27 patients (no documented arrhythmia at presentation in 16), paroxysmal arrhythmias in 16 patients (atrial in 7 and ventricular in 9), and unexplained syncope in 3 patients.
As a result of the CPET, 5 patients were listed for heart transplantation, 3 patients underwent diagnostic cardiac catheterization, 2 patients underwent tricuspid valve replacement, 2 patients received a biventricular pacemaker, and 1 patient underwent a DDD pacemaker implantation because of exercise-induced complete atrioventricular block. In 3 of the 9 patients with baffle stenosis, the stenosis was considered to be hemodynamically significant. These 3 patients underwent surgical (n = 1) or catheter-based (n = 2) relief of the stenosis. One patient had a residual baffle leak, and the leak was closed percutaneously.
Predictors of mortality and emergency hospital admission
In the univariate analysis, Mustard operation (Fig. 2),greater age at CPET, presence of an implanted pacemaker, VE/VCO2slope, peak Vo2%, and HR reserve were all associated with an increased risk of death/cardiac-related emergency hospital admission (Table 2).
In the multivariate analysis that included all univariate predictors, the VE/VCO2slope was the strongest predictor of the composite end point (Table 2). Peak Vo2% added significant prognostic value and was retained in the model.
Although the statistics on peak Vo2% were significant, the VE/VCO2-based system appeared to be superior, as indicated by differences in Wald chi-square (60.8 for VE/VCO2slope vs. 3.2 for peak Vo2%). After we plotted the annual event rate according to VE/VCO2slope and peak Vo2% quartiles, subjects in the lowest quartile of VE/VCO2demonstrated a favorable prognosis, irrespective of peak Vo2% (Fig. 3).Furthermore, the trends for increasing event rates were more apparent as the VE/VCO2slope increased, compared with decreasing peak Vo2%. Although the number of subjects per subgroup was relatively small, the highest event rate was observed in subjects in the highest VE/VCO2slope quartile and in the lowest peak Vo2% quartile. The combination of the prognostic information provided by VE/VCO2slope and peak Vo2% appeared to be additional and therefore superior to that provided by each single predictor when considered alone.
Cutoff values of VE/VCO2slope and peak Vo2% associated with increased 4-year cumulative event rate
At a 4-year follow-up, 52 of the 174 patients (29.9%) had died (n = 11) or had an emergency hospital admission (n = 41). At ROC analysis, a VE/VCO2slope ≥35.4 had the highest sensitivity and specificity to predict 4-year mortality/cardiac-related emergency hospital admission (sensitivity: 81%; 95% CI: 69% to 91%; specificity: 83%; 95% CI: 75% to 89%). Patients with a VE/VCO2slope ≥35.4 were older at M/S operation (37 ± 35 months vs. 30 ± 31 months, p = 0.0402), had more PH (23.1% vs. 3.4%, p = 0.0001), and had more advanced RV dysfunction (p < 0.0001) than patients with a VE/VCO2slope <35.4, whereas the prevalence of Mustard operation (64.2% vs. 60.7%, p = 0.748), and the age at CPET (27.5 ± 10.0 years vs. 26.6 ± 8.4 years, p = 0.402) was similar.
A second ROC curve analysis for peak Vo2% revealed that a peak Vo2% ≤52.3% of predicted value had the highest sensitivity and specificity to predict 4-year mortality/cardiac-related emergency hospital admission (sensitivity: 76%; 95% CI: 63% to 85%; specificity: 72%; 95% CI: 62% to 80%). Combining the information provided by VE/VCO2slope and peak Vo2% cutoff values, we were able to stratify patients into 4 different risk groups, with those with VE/VCO2slope ≥35.4 and a peak Vo2% ≤52.3% of predicted value being at very high risk (4-year event rate: 78.8%) (Fig. 4).
Determinants of peak Vo2% and VE/VCO2slope
Reduced peak Vo2% was associated with Mustard operation (r = −0.149, p = 0.014), higher age at test (r = −0.304, p < 0.0001), reduced HR reserve (r = 0.365, p < 0.0001), presence of pacemaker (r = −0.235, p < 0.0001), RV dysfunction (r = −0.177, p = 0.007), and presence of PH (r = −0.199, p = 0.002).
A higher VE/VCO2slope was associated with Mustard operation (r = 0.133, p = 0.025), higher age at operation (r = 0.156, p = 0.004), higher age at test (r = 0.095, p = 0.009), reduced peak Vo2% (r = −0.624, p < 0.0001), reduced HR reserve (r = −0.236, p < 0.0001), RV dysfunction (r = 0.265, p < 0.0001), and presence of PH (r = 0.416, p < 0.0001). At multivariable analysis, VE/VCO2slope was related to peak Vo2% (p < 0.0001), presence of PH (p < 0.0001), and RV dysfunction (p = 0.017).
Prognostic value of VE/VCO2slope and peak Vo2%
The present investigation demonstrates that in M/S patients, CPET is a valuable prognostic tool. In particular, we could demonstrate that M/S patients with a VE/VCO2slope ≥35.4 or those with a peak Vo2% ≤52.3% of predicted value have a substantially higher risk of death or cardiac-related emergency hospital admission, even after accounting for age and type of atrial repair. In contrast with the adult with acquired heart disease population, the knowledge about the prognostic value of CPET in adults with congenital heart disease is very limited (14–16). The Royal Brompton group reported in a large group of patients with various forms of congenital heart disease that peak Vo2was an independent predictor of outcome in the medium term expressed as death or hospitalization (14). In that study, patients with a peak Vo2<15.5 ml/kg/min demonstrated a 2.9-fold increased risk of hospital admission or death compared with patients with a peak Vo2≥15.5 ml/kg/min. Dimopoulos et al. (15) from the same group also reported that a VE/VCO2slope >38 identified noncyanotic patients at higher risk of death in the midterm. These findings were supported by a more recent report confirming the prognostic value of VE/VCO2slope in a large group of adults with various congenital heart defects (16). However, since all previous studies have included patients with different underlying anatomy and physiopathology, disease- and operation-specific data are unavailable, therefore limiting the clinical applicability of such findings.
Comparison of prognostic value of peak Vo2and VE/VCO2slope
Since the original report of Mancini et al. (10), peak Vo2has gained considerable notoriety in the HF population as a valuable prognostic marker and has been used extensively in clinical practice to identify candidates for heart transplantation. More recently, however, VE/VCO2slope has emerged as a potentially superior prognostic marker compared with peak Vo2(11). As a consequence of its recognition as a potentially important clinical tool, calculation of VE/VCO2slope has become an integral part of CPET in most physiology laboratories. Indeed, VE/VCO2slope is now readily derived by commercial CPET software packages that operate present-day ventilatory expired gas units, which makes its clinical application as feasible as that of peak Vo2.
The present study demonstrates that there is an association between VE/VCO2slope and prognosis across a wide spectrum of HF severity and suggests the superiority of VE/VCO2slope over peak Vo2for assessing prognosis in adults with the M/S operation. Interestingly, we observed that, as previously seen in the adult HF population (12), this variable holds prognostic significance even when overall exercise performance is not severely compromised. A primary reason for this discrepancy may be the dependence of peak Vo2on subject effort for optimal prognostic value, whereas the VE/VCO2slope is largely effort-independent (19). The data from the present study are consistent with the 2 recent reports supporting a higher prognostic value of VE/VCO2slope when compared with peak Vo2in adults with various forms of congenital heart disease (15,16). We advocate introducing VE/VCO2slope and peak Vo2% in the risk assessment of adults with M/S operation and the utilization of the cutoff values identified in the present study.
Mechanisms of increased VE/VCO2slope and insights on VE/VCO2slope prognostic value
The reasons for increased VE/VCO2slope in adults with M/S operations have not been previously examined. In the present study, higher VE/VCO2slope was associated with presence of PH, RV dysfunction, and reduced exercise capacity.
Several investigations have examined the correlation between VE/VCO2slope and other markers of pathophysiology associated with HF. In these studies, increasing VE/VCO2slopes were related to progressively worsening hemodynamics, perfusion/ventilation mismatch, increased chemoreceptor and ergoreceptor activation, and decreased HR variability (20–22). All of these abnormalities are hallmarks of chronic HF syndrome and have been identified as such in the broad group of adults with congenital heart disease (23). Therefore, the increasingly worse prognosis as the VE/VCO2slope increased in the present study likely reflects greater cardiovascular dysfunction compared with individuals with lower VE/VCO2slope responses. In particular, a previous study showed that PH and cyanosis are associated with increased VE/VCO2slope in adults with congenital heart disease (15). Abnormally elevated VE/VCO2slopes have also been found in patients with primary PH who usually do not show arterial desaturation during exercise (24). In this setting, it is thought that ventilation/perfusion mismatch related to PH is responsible for significant gas exchange inefficiency (24). It is likely that the same underlying hemodynamic abnormality (e.g., PH) is responsible for elevation of VE/VCO2slope in adults with M/S operation. Cyanosis has been demonstrated as an important determinant of VE/VCO2slope (15). However, the low prevalence of cyanosis during exercise observed in the present study seems to rule out a role of cyanosis in the elevation of VE/VCO2slope.
HF and arrhythmias in adults with M/S operation
Classically, CPET is considered unable to predict arrhythmic events in adults with HF (18). However, approximately one-quarter of the emergency cardiac-related hospital admissions in the present study were related to the occurrence of cardiac arrhythmias. This is likely related to the fact that for M/S patients, arrhythmias, as well as HF symptoms, are related to the unique myocardial substrate created by surgical scars in conjunction with abnormal pressure/volume loads of long duration, and RV dysfunction (25). Indeed, HF and cardiac arrhythmias are related to the same underlying substrate, possibly explaining why CPET is able to identify patients at risk of cardiac arrhythmias.
The adult congenital heart disease population studied here represents the current workload of tertiary centers. Therefore, we cannot exclude the possibility that the population studied represents a biased sample, favoring patients with more symptoms and lower perceived functional capacity.
An obvious limitation of the present study is that CPET was performed on an ergometer cycle in 75.2% of patients and on a treadmill in 24.8%. Treadmill exercise testing is known to produce slightly higher values of peak Vo2when compared to an ergometer cycle because of a higher number of muscles that are exercised during the test (18). This might have produced slightly higher values of peak Vo2in the group of patients exercised by means of a treadmill. These higher values of peak Vo2might not translate into improved outcome when compared to slightly lower peak Vo2values obtained by subjects who exercised on a cycle. This issue might also be partly responsible for the overall lower sensitivity and specificity of peak Vo2in predicting the study end points. However, very similar results to those presented in the study, and very similar peak Vo2and VE/VCO2slope cutoff values (51.4% and 35.8, respectively) were observed when those patients who exercised on a treadmill were excluded from the analysis.
We are aware that cutoff selection using ROC analysis might be associated with an increased type I error. This should be taken into account when interpreting the cutoff values proposed. Larger, prospective studies with longer follow-up may identify additional factors that impact on VE/VCO2slope and on peak Vo2, and thus on outcome, and examine the potential effects of physical conditioning in this patient group.
Follow-up length was limited to 3.9 ± 2.3 years in the present study. Despite that, the number of events recorded was sufficient for meaningful prognostic interpretation of the value of CPET in patients with M/S operation.
The CPET provides important prognostic information in adults with M/S operation. Subjects with a VE/VCO2slope ≥35.4 or those with a peak Vo2≤52.3% of predicted value have a substantially higher 4-year risk of death or cardiac-related emergency hospital admission combined.
The employer of Dr. Diller and Prof. Gatzoulis (Imperial College) has received an unrestricted educational grant from Actelion (United Kingdom). Dr. Diller also has received travel support and honoraria from Actelion (United Kingdom), Encysive (United Kingdom), and Schering (Germany).
- Abbreviations and Acronyms
- cardiopulmonary exercise test
- heart failure
- heart rate
- Mustard and Senning
- pulmonary hypertension
- receiver-operator characteristic
- systemic right ventricle
- transposition of the great arteries
- minute ventilation/carbon dioxide production
- oxygen uptake
- Received October 29, 2008.
- Revision received January 27, 2009.
- Accepted February 3, 2009.
- American College of Cardiology Foundation
- Wong K.Y.,
- Venables A.W.,
- Kelly M.J.,
- Kalff V.
- Musewe N.N.,
- Reisman J.,
- Benson L.N.,
- et al.
- Benson L.,
- Bonet J.,
- McLaughlin P.,
- et al.
- Lange R.,
- Horer J.,
- Kostolny M.,
- et al.
- Kammeraad J.A.,
- van Deurzen C.H.,
- Sreeram N.,
- et al.
- Mancini D.M.,
- Eisen H.,
- Kussmaul W.,
- Mull R.,
- Edmunds L.H. Jr..,
- Wilson J.R.
- Ponikowski P.,
- Francis D.P.,
- Piepoli M.F.,
- et al.
- Diller G.P.,
- Dimopoulos K.,
- Okonko D.,
- et al.
- Dimopoulos K.,
- Okonko D.O.,
- Diller G.P.,
- et al.
- Wasserman K.,
- Hansen J.E.,
- Sue D.Y.
- Ponikowski P.P.,
- Chua T.P.,
- Francis D.P.,
- Capucci A.,
- Coats A.J.S.,
- Piepoli M.F.
- Walsh E.P.,
- Cecchin F.