- Email: [email protected]
Determinants of survival and length of survival in adults with Eisenmenger syndrome
Determinants of Survival and Length of Survival in Adults With Eisenmenger Syndrome Warren J. Cantor, MD, David A. Harrison, MD, Jack S. Moussadji, Michael S. Connelly, MBBS, Gary D. Webb, MD, Peter Liu, MD, Peter R. McLaughlin, MD, and Samuel C. Siu, MD
MD,
A retrospective study of adults with Eisenmenger syndrome assessed at a tertiary referral center was performed to identify clinical characteristics and establish prognostic determinants. Mortality and cause of death are reported with baseline clinical data correlated with mortality to identify predictors of death. Clinical events during follow-up, including heart failure, atrial arrhythmia, hemoptysis, and syncope were also reviewed. A total of 109 adults with Eisenmenger syndrome (mean ⴞ SD age 29 ⴞ11 years, 43% men) were followed for a median of 6.3 years. Sixty-six patients (61%) had simple cardiac anatomy (13 atrial septal defect, 43 ventricular septal defect, 10 patent ductus arteriosus). The remainder (43 patients) had complex cardiac anatomy (including atrioventricular septal defect, truncus arteriosus, univentricular heart, and transposition of the great
arteries). There were 33 deaths and 9 transplantations during follow-up. Median survival was 53 years. Multivariate Cox regression analysis identified age at presentation (hazard ratio [HR] 0.90), supraventricular arrhythmia (HR 3.44), precordial electrocardiogram voltage (HR 1.61/mV increase), and poor New York Heart Association functional class (HR 2.60) as independent predictors of mortality. There is a large variation in the life expectancy for adults with Eisenmenger syndrome. Baseline characteristics associated with increased mortality include younger age at presentation (associated with complex anatomy), functional class, supraventricular arrhythmia, and an electrocardiogram index for right ventricular hypertrophy. 䊚1999 by Excerpta Medica, Inc. (Am J Cardiol 1999;84:677– 681)
ver the past 4 decades our knowledge of the management of patients with the Eisenmenger O syndrome has improved. Previously identified pre-
artery bidirectional (predominantly right to left) shunt. Patients were excluded if pulmonary vascular disease was present in only 1 lung (e.g., after a unilateral systemic to pulmonary artery shunt) or if the intracardiac shunt had been surgically closed. Patients with small shunts that were unlikely to be the cause of pulmonary hypertension were not included. There were 13 patients who had only 1 assessment at our center, 5 of whom have died. For those patients who had not had follow-up since January 1994, current status was obtained through referring physician’s records. If recent follow-up remained unavailable, then records of the Registrar General were searched for certificates of death. Five patients remained unaccounted for, and have been included until the time of their last available contact. Data collection: In- and outpatient records were reviewed to extract major anatomic lesions, surgical history, baseline characteristics (age, gender, presenting symptoms, New York Heart Association functional class, medications, electrocardiograms), mortality, and other clinical events during follow-up (hemoptysis, cerebrovascular accident, hospital admission for congestive heart failure, supraventricular arrhythmia, syncope, cerebral abscess, and pregnancy). Supraventricular arrhythmias were only considered significant if they required medical therapy. Admissions for congestive heart failure occurring in other institutions were included based on a discharge diagnosis of congestive heart failure. Objective clinical findings were often poorly documented in such cases, precluding a distinction between right- and left-sided heart failure.
1– 4
dictors and cause of death may therefore have changed sufficiently to alter the natural survival pattern. Accurate survival information is important to select patients who are in greatest need of lung transplantation, especially in view of the recent data suggesting an increased risk of lung transplantation in this population.5 We have reviewed our experience in the management of patients with the Eisenmenger syndrome to identify the current clinical characteristics of the population and to document the causes and predictors for mortality.
METHODS
Population: The study population was obtained from the databases of the adult congenital heart clinic and of cardiac patients assessed for lung transplantation at a single tertiary care institution. Inclusion criteria included age ⬎18 years, at least 1 assessment since 1980, and clinical evidence of severe pulmonary hypertension in the presence of intracardiac or great From The University of Toronto Congenital Cardiac Centre for Adults, The Toronto Hospital, University of Toronto, Toronto, Ontario, Canada. This study was supported in part by operating grants from the Heart & Stroke Foundation of Ontario and Medical Research Council of Canada, Ontario, Canada. Manuscript received December 11, 1998; revised manuscript received April 28, 1999, and accepted April 29. Address for reprints: David A. Harrison, MD, Division of Cardiology, The Toronto Hospital, 12EN 221, 200 Elizabeth Street, Toronto, Ontario, Canada, M5G 2C4. ©1999 by Excerpta Medica, Inc. All rights reserved. The American Journal of Cardiology Vol. 84 September 15, 1999
0002-9149/99/$–see front matter PII S0002-9149(99)00415-4
677
TABLE I Total Population Compared With Subgroup With Hemodynamic Data Available Total Population (n ⫽ 109)
Parameter Age first visit (yrs) Women New York Heart Association functional class* I II III IV Trisomy 21 Simple anatomy Pregnancy Patients with event-free survival Heart or heart-lung transplantation Deaths
Hemodynamic Data (n ⫽ 59)
No Hemodynamics (n ⫽ 50)
28.6 ⫾ 10.8 62 (57%)
30.3 ⫾ 10.6 37 (63%)
26.6 ⫾ 10.6 25 (50%)
16 44 41 6 34 66 14 22 9 33
10 22 24 2 15 42 10 12 7 14
6 22 17 4 19 24 4 10 2 19
(15%) (41%) (38%) (6%) (31%) (61%) (23% of women) (20%) (8%) (30%)
(17%) (38%) (42%) (3%) (25%) (71%) (27% of women) (20%) (12%) (24%)
(12%) (45%) (35%) (8%) (38%) (48%) (16% of women) (20%) (4%) (38%)
*Functional class missing in 2 patients.
Electrocardiographic data was reported using an index for right ventricular hypertrophy defined as a continuous variable calculated as the sum of the R wave in lead V1 and the maximum depth of the S wave in leads V5 or V6 (modified from Chou6). Data analysis: Data are presented as mean ⫾ SD unless specified as median in the text. Comparison of means between groups was performed using 2-tailed Student’s t test. For purposes of analysis, patients were considered to have “simple” cardiac anatomy if the primary anatomic defect was limited to atrial septal defect, ventricular septal defect, or patent ductus arteriosus. All other patients were considered to have “complex” anatomy. The survival of patients since age 18 years is displayed using Kaplan-Meier curves. Patients who underwent lung or heart-lung transplantation were censored from the data analysis as alive at the time of their transplantation because their subsequent survival would no longer reflect the survival pattern for Eisenmenger syndrome. Risk factors for adverse outcome were determined using proportional-hazards analysis. A Cox multivariate analysis was undertaken using the following variables: age at presentation, gender, simple versus complex anatomy, trisomy 21, baseline New York Heart Association functional class, events before first assessment (including syncope, hemoptysis, congestive heart failure, supraventricular arrhythmia), phlebotomy, and electrocardiogram index. Oxygen saturation was not available in enough patients to allow inclusion in the analysis. The multivariate model was analyzed using a backward elimination algorithm with the level of significance set at 0.05.
RESULTS
Population: A total of 109 patients with Eisenmenger syndrome were identified (47 men). Hemodynamic data confirming systemic pulmonary artery pressure was available for 59 of these patients, and the remaining 50 patients were judged to have Eisenmenger physiology from clinical assessment. A comparison between the total population of 109 patients, 678 THE AMERICAN JOURNAL OF CARDIOLOGY姞
VOL. 84
the 59 patients with hemodynamic evaluation available, and the 50 patients without hemodynamic evaluation is shown in Table I. Patients without hemodynamic data were less likely to have simple anatomy, and there were no other differences between the 2 subgroups. For the total population the distribution of cardiac anatomy was ventricular septal defect (n ⫽ 43), atrioventricular septal defect (n ⫽ 19), atrial septal defect (n ⫽ 13), univentricular morphology (n ⫽ 11), patent ductus arteriosus (n ⫽ 10), transposition complexes (n ⫽ 7), truncus arteriosus (n ⫽ 3), pulmonary atresia with Pott’s shunt (n ⫽ 2), and tricuspid atresia with ventricular septal defect (n ⫽ 1). Average age at first clinic assessment was 28.6 ⫾ 10.8 years. Most patients (96 of 109) were symptomatic at referral, having either exertional dyspnea or fatigue (84% of total group). Other symptoms (not mutually exclusive) included: palpitations (21%), cough (13%), hemoptysis (11%), and syncope (9%). There were 16 (15%), 44 (41%), 41 (38%), and 6 (6%) patients in New York Heart Association functional classes I through IV, respectively (data not available in 2 patients). Trisomy 21 was present in 34 patients (31%), 17 of whom had simple cardiac anatomy, 16 had atrioventricular septal defect, and 1 had univentricular connection. Phlebotomy had already been initiated in 30 patients before presentation. Most patients (n ⫽ 71) were taking at least 1 medication including diuretics (n ⫽ 20), digoxin (n ⫽ 38), nitrates (n ⫽ 2), angiotensin-converting enzyme inhibitors (n ⫽ 3), antiarrhythmic drugs (n ⫽ 2), aspirin (n ⫽ 5), calcium antagonists (n ⫽ 5), other cardiac medications (n ⫽ 10), and other noncardiac medications (n ⫽ 37). Despite the fact that pregnancy is considered to be associated with prohibitive risk for patients with Eisenmenger syndrome, 14 of the 62 women in our group have had at least 1 pregnancy (26 total pregnancies). In many cases pregnancy was undertaken several years before referral to our center. Eisenmenger physiology at the time of pregnancy could only be confirmed in 5 pregnancies; the evolution of SEPTEMBER 15, 1999
TABLE II Summary of Mortality by Primary Anatomy Diagnosis Atrial septal defect Ventricular septal defect Patent ductus arteriosus Atrioventricular septal defect Other*
Total Patients 13 43 10 19 24
Death (%) 4 13 2 5 9
(31%) (30%) (20%) (26%) (38%)
Age at Death (mean ⫾ SD) 45.3 43.3 42.4 27.8 28.2
⫾ ⫾ ⫾ ⫾ ⫾
15.8 12.4 20.1 7.9 7.9
*Other diagnoses include: univentricular heart (n ⫽ 11), transposition of the great arteries (n ⫽ 7), truncus arteriosus (n ⫽ 3), and tricuspid or pulmonary atresia (with surgical aortopulmonary shunts; n ⫽ 3).
pulmonary artery hypertension in some or most patients was relatively early in the pregnancy. Of the 26 pregnancies, 6 were electively terminated, 3 ended in spontaneous abortion, and 17 continued to delivery. Primary anatomy for the 10 patients who completed at least 1 pregnancy was: atrial septal defect (n ⫽ 4), ventricular septal defect (n ⫽ 3), patent ductus arteriosus (n ⫽ 1), atrioventricular septal defect (n ⫽ 1), and single ventricle (n ⫽ 1). There were no maternal pregnancy-related deaths in this group. However, by the nature of this study, maternal deaths before 1980 would not have been included. Patients were followed for a median of 6.3 years (range 1 to 40). Nine patients underwent transplantation at average age 37 ⫾ 6.2 years (7 lung transplantation, 2 heart-lung transplantation) of whom 4 died (2 lung and 2 heart-lung). All post-transplantation deaths occurred during hospital admission for transplantation. Mortality was high in nontransplanted patients, with 33 deaths. Average age at death was 37 ⫾ 13.3 years. The distribution of mortality by underlying anatomy is given in Table II. Identified causes of death included: congestive heart failure (n ⫽ 14), sudden (n ⫽ 7), endocarditis (n ⫽ 2), cerebrovascular event (n ⫽ 2), postappendectomy (n ⫽ 1), massive hemoptysis (n ⫽ 1), noncardiac (n ⫽ 4), and unknown (n ⫽ 2). Morbidity was also high, with only 22 patients (20%) alive and free of any clinical event at last assessment. Documented clinical events included: hemoptysis (n ⫽ 34), admission for congestive heart failure (n ⫽ 29), syncope (n ⫽ 26), atrial fibrillation or atrial flutter (n ⫽ 18), and intracerebral abscess (n ⫽ 7). Figure 1 shows the Kaplan-Meier survival curve for the population. Median survival was 52.6 years. All patients were assumed to be enrolled at age 18 and transplanted patients were censored as alive at time of transplantation. These 2 assumptions increase the predicted survival of our population. If transplanted patients are assumed to have died at the time of transplantation, then median survival would be 49.0 years. Approximately half of the patients presented to the clinic before age 25. The median survival for these patients presenting as young adults was 38.2 years. Figure 2 shows the Kaplan-Meier survival curve for the subgroup of 59 patients for whom hemodynamic data were available (group 1) and is compared with
FIGURE 1. The survival curve for the total population of 109 patients with Eisenmenger syndrome is shown. Median survival is 52.6 years, assuming age at enrollment to be 18 years, and transplanted patients are censored as alive at the time of transplantation (see text). The number of patients alive are provided at each decade.
FIGURE 2. The survival curve of the subgroup of patients for whom complete hemodynamic data were available (solid line) is compared with survival for the group without hemodynamic data (broken line). Median survival was 52.6 years for the group with hemodynamic data available (mean ⴞ SE 51 ⴞ 2.1) and 51.4 years for those without hemodynamic data (mean 47.5 ⴞ 3.0).
the survival curve for the remaining population (group 2). The 2 curves almost overlap, with no difference in median survival for the 2 groups (group 1, 52.6 years, group 2, 51.4 years). A Cox multivariate logistic regression was performed in the total population using the 11 clinical variables as listed under the Methods section. Significant predictors for mortality included: age at presentation (hazard ratio [HR] 0.90/year), indicating a worse prognosis for younger age at presentation, increased precordial voltage (HR 1.61/mV), New York Heart Association class III or IV (HR 2.60), and a history of supraventricular arrhythmia requiring treatment (HR 3.44). Confidence intervals and significance levels are presented in Table III. The electrocardiogram index ranged from 0.2 to 6.4 mV. Mortality for patients with an electrocardiogram index ⬍3.0 mV was 21% (14 of 67 patients) compared with a mortal-
CONGENITAL HEART DISEASE/SURVIVAL IN THE ADULT EISENMENGER
679
TABLE III Multivariate Analysis Predictor Age at referral* RV1 ⫹ SV5 or V6 Atrial arrhythmia New York Heart association class III–IV Hemoptysis Phlebotomy Complex anatomy Gender Syncope Trisomy 21 Heart failure
Hazard Ratio
95% Confidence Interval
Significance Level
0.90 1.61 3.44 2.60
0.86–0.94 1.25–2.08 1.35–8.76 1.14–5.93
0.0001 0.0003 0.02 0.03
0.35 2.32 1.96 0.56 0.38 0.58 0.67
0.10–1.21 0.76–7.08 0.70–5.48 0.21–1.48 0.08–1.95 0.21–1.65 0.18–2.44
0.09 0.14 0.20 0.24 0.25 0.31 0.55
*Age at referral indicates increased risk for earlier age at referral (hazard ratio 0.90/year).
ity rate of 45% (19 of 42 patients) for an index ⱖ3.0 mV. Complex anatomy was not found to be a significant multivariate predictor for poor outcome. However, a relation was present between age at presentation and patient anatomy, with complex patients presenting early (23.7 ⫾ 7.2 years) versus those with simple anatomy (31.6 ⫾ 11.8 years), p ⫽ 0.001.
DISCUSSION
Estimate of survival: In this study of adult patients with the Eisenmenger syndrome we found a predicted median survival of 52.6 years based on a population of 109 patients. There were 33 deaths at an average age of 37.0 ⫾ SD 13.3 years. In previous studies, Wood7 reported 56 deaths with an average age at time of death of 33 years for aortopulmonary and ventricular defects and 26 years for atrial septal defects. Three other studies have observed a mortality rate of 10% to 44%, with an average age at death ranging from 25 to 32 years.8 –10 The lower age at death in 2 of these studies9,10 may be a result of the inclusion of pediatric patients. Our population has a substantially higher age at death than these previous reports, including the adult subpopulation of Young and Mark’s publication.8 Lung and heart-lung transplantations are the most significant changes to the management of the Eisenmenger patient in the 25 years that have passed since Young’s study. However, this does not appear to be the reason that our population has an improved survival over Young’s study (the average age at death or transplantation was 37.7 ⫾ 12.1 years), which still exceeds his observed age at death). Our clinic specializes in the adult patient (age ⬎18 years); therefore, part of our higher age at death may reflect a referral bias. However, we postulate that some of the improved survival reflects changes in contemporary management that may include more appropriate use of phlebotomy, improved perioperative anesthetic management,11 and appropriate counseling regarding pregnancy, all of which may contribute to improved survival in Eisenmenger patients. 680 THE AMERICAN JOURNAL OF CARDIOLOGY姞
VOL. 84
Cause of death: Cause of death for the adult patient is also slightly different from that reported in previous studies. Wood2,7 stressed the potential risk of hemoptysis, the cause of death in 29% of the deaths in his population. Death due to hemoptysis is similarly reported in 11% to 15% of other studies.9,10 In contrast, we found only 1 of 33 deaths (3%) to be attributable to massive hemoptysis, despite a high prevalence of hemoptysis within the population (31%). In many previous studies, the most common mode of death is sudden (range from 19% to 47% of reported deaths).8 –10 Sudden death represented 21% of the deaths in our study, but in contrast to other studies congestive heart failure was the predominant mode of death (observed in 42%). Mortality risk factors: Earlier age at presentation, the electrocardiogram index for right ventricular hypertrophy, a history of supraventricular arrhythmia, and poor functional class were significant predictors of mortality. The observation that patients who presented to the clinic at a younger age were at increased risk for early mortality is presumably a function of both a referral bias for complex patients obtaining early referral and for patient referral being biased to the sicker patient. We found the modified electrocardiogram index for right ventricular hypertrophy to be predictive of decreased survival. We believe that the index reflects the degree and duration of pulmonary hypertension. However, it is possible that the index may reflect right ventricular dilation, altered conduction in a failing ventricle (a proportion of these patients had right bundle branch block and were included with no adjustment of the index), or reflect underlying complex anatomy. The presence of supraventricular arrhythmia as a significant predictor of mortality may be a result of a failing right ventricle or could have contributed to clinical deterioration through the loss of preload. Previous studies have found higher mortality rates in patients with ventricular or aortopulmonary shunts than with atrial shunts.3,8 We could not confirm this because of the small patient numbers. However a similar age at death was detected for patients with atrial septal defect (45.3 ⫾ 15.8), ventricular septal defect (43.3 ⫾ 12.4), and patent ductus arteriosus (42.4 ⫾ 20.1), all of whom appeared to be older than those with atrioventricular septal defect (27.8 ⫾ 7.9) or other more complex anatomy (28.2 ⫾ 7.9). Other potential risk factors for death that have been identified include: syncope,10 noncardiac surgery,12 trisomy 21,9 hemoptysis,2,13 elevated right atrial pressure, and low arterial oxygen saturation.10 The variation of factors between these studies likely reflects the relatively small sample sizes of each study, the different management periods under observation, variation of study population age, and different center referral and follow-up bias. There is a need for a prospective evaluation of previously identified predictors of mortality optimize patient selection for intervention such as lung or heart-lung transplantation. The substantial mortality rate of our patients after transplantation SEPTEMBER 15, 1999
highlights the importance of selection of only patients at highest risk for death from Eisenmenger syndrome. Study limitations: As with all retrospective studies our results are limited by variation in patient care, follow-up, and investigation. However, most patients were followed through our center so that consistent clinical data are available. In addition, mortality, the primary outcome of the study, was available through provincial registries. Eisenmenger syndrome is infrequent; therefore, we have a small sample size that reduces the power of the study to recognize all potential important risk factors. However, only 1 study has a larger sample size.10 Our center is a tertiary care facility so that referral bias affects the composition of our population and the time of patient presentation (as is recognized by the importance of age at presentation as an important predictor for survival). However, as the major site for evaluation of adults with congenital heart disease in our region, we expect a large proportion of patients with this unusual problem to have been reviewed here. As a retrospective study we were not able to obtain hemodynamic data to confirm Eisenmenger physiology in 50 patients who were included based on clinical diagnosis. The inclusion of these patients had the potential to alter the survival of the total population. However, baseline characteristics between these 2 groups was minor, with a possibly higher proportion of complex patients and trisomy 21 patients in the group lacking catheterization data (potentially biasing
the results to a worse prognosis for patients without catheterization data). Analysis revealed that the outcome for these 2 groups was not different by either direct comparison or by multivariate analysis.
1. Eisenmenger V. Die angeborenen defecte der kammerscheidewand des herzen. Z Klin Med 1897;32:1–28. 2. Wood P. The Eisenmenger syndrome or pulmonary hypertension with reversed central shunt. Br Med J 1958;2:701–709. 3. Touzet AD, Corone A, Jullien JL, Corone P. Evolution spontanee et complication des syndromes d’Eisenmenger. Arch Mal Coeur Vaiss 1983;5:563–567. 4. Hallidie-Smith KA, Tortoledo R, Goodwin JF. Presentation and natural history of Eisenmenger syndrome. Br Heart J 1971;33:610. 5. Hosenpud JD, Bennett LE, Keck BM, Fiol B, Boucek MM, Novick RJ. The registry of the international society for heart and lung transplantation: fifteenth official report–1998. J Heart Lung Transplant 1998;17:656 – 668. 6. Chou T. Right ventricular hypertrophy. In: Electrocardiography in Clinical Practice. Philadelphia: WB Saunders, 1996:53– 68. 7. Wood P. The Eisenmenger syndrome or pulmonary hypertension with reversed central shunt. Br Med J 1958;2:755–762. 8. Young D, Mark H. Fate of the patient with Eisenmenger syndrome. Am J Cardiol 1971;28:658 – 669. 9. Corone S, Davido A, Lang T, Corone P. Outcome of patients with Eisenmenger syndrome. Apropos of 62 cases followed-up for an average of 16 years. Arch Mal Coeur Vaiss 1992;85:521–526. 10. Saha A, Balakrishnan KG, Jaiswal PK, Venkitachalam CG, Tharakan J, Titus T, Kutty R. Prognosis for patients with Eisenmenger syndrome of various aetiology. Int J Cardiol 1994;45:199 –207. 11. Ammash NM, Connolly HM, Abel MD, Warnes CA. Noncardiac surgery in Eisenmenger syndrome. J Am Coll Cardiol 1999;33:222–227. 12. Hallidie-Smith KA, Goodwin JF. The Eisenmenger syndrome. Prog Cardiol 1974;3:211–225. 13. Clarkson PM, Frye RL, Dushane JW, Burcheu HB, Wood EH, Weiman WH. Prognosis of patients with ventricular septal defect and severe pulmonary vascular obstructive disease. Circulation 1968;38:129 –135.
CONGENITAL HEART DISEASE/SURVIVAL IN THE ADULT EISENMENGER
681
MD,
A retrospective study of adults with Eisenmenger syndrome assessed at a tertiary referral center was performed to identify clinical characteristics and establish prognostic determinants. Mortality and cause of death are reported with baseline clinical data correlated with mortality to identify predictors of death. Clinical events during follow-up, including heart failure, atrial arrhythmia, hemoptysis, and syncope were also reviewed. A total of 109 adults with Eisenmenger syndrome (mean ⴞ SD age 29 ⴞ11 years, 43% men) were followed for a median of 6.3 years. Sixty-six patients (61%) had simple cardiac anatomy (13 atrial septal defect, 43 ventricular septal defect, 10 patent ductus arteriosus). The remainder (43 patients) had complex cardiac anatomy (including atrioventricular septal defect, truncus arteriosus, univentricular heart, and transposition of the great
arteries). There were 33 deaths and 9 transplantations during follow-up. Median survival was 53 years. Multivariate Cox regression analysis identified age at presentation (hazard ratio [HR] 0.90), supraventricular arrhythmia (HR 3.44), precordial electrocardiogram voltage (HR 1.61/mV increase), and poor New York Heart Association functional class (HR 2.60) as independent predictors of mortality. There is a large variation in the life expectancy for adults with Eisenmenger syndrome. Baseline characteristics associated with increased mortality include younger age at presentation (associated with complex anatomy), functional class, supraventricular arrhythmia, and an electrocardiogram index for right ventricular hypertrophy. 䊚1999 by Excerpta Medica, Inc. (Am J Cardiol 1999;84:677– 681)
ver the past 4 decades our knowledge of the management of patients with the Eisenmenger O syndrome has improved. Previously identified pre-
artery bidirectional (predominantly right to left) shunt. Patients were excluded if pulmonary vascular disease was present in only 1 lung (e.g., after a unilateral systemic to pulmonary artery shunt) or if the intracardiac shunt had been surgically closed. Patients with small shunts that were unlikely to be the cause of pulmonary hypertension were not included. There were 13 patients who had only 1 assessment at our center, 5 of whom have died. For those patients who had not had follow-up since January 1994, current status was obtained through referring physician’s records. If recent follow-up remained unavailable, then records of the Registrar General were searched for certificates of death. Five patients remained unaccounted for, and have been included until the time of their last available contact. Data collection: In- and outpatient records were reviewed to extract major anatomic lesions, surgical history, baseline characteristics (age, gender, presenting symptoms, New York Heart Association functional class, medications, electrocardiograms), mortality, and other clinical events during follow-up (hemoptysis, cerebrovascular accident, hospital admission for congestive heart failure, supraventricular arrhythmia, syncope, cerebral abscess, and pregnancy). Supraventricular arrhythmias were only considered significant if they required medical therapy. Admissions for congestive heart failure occurring in other institutions were included based on a discharge diagnosis of congestive heart failure. Objective clinical findings were often poorly documented in such cases, precluding a distinction between right- and left-sided heart failure.
1– 4
dictors and cause of death may therefore have changed sufficiently to alter the natural survival pattern. Accurate survival information is important to select patients who are in greatest need of lung transplantation, especially in view of the recent data suggesting an increased risk of lung transplantation in this population.5 We have reviewed our experience in the management of patients with the Eisenmenger syndrome to identify the current clinical characteristics of the population and to document the causes and predictors for mortality.
METHODS
Population: The study population was obtained from the databases of the adult congenital heart clinic and of cardiac patients assessed for lung transplantation at a single tertiary care institution. Inclusion criteria included age ⬎18 years, at least 1 assessment since 1980, and clinical evidence of severe pulmonary hypertension in the presence of intracardiac or great From The University of Toronto Congenital Cardiac Centre for Adults, The Toronto Hospital, University of Toronto, Toronto, Ontario, Canada. This study was supported in part by operating grants from the Heart & Stroke Foundation of Ontario and Medical Research Council of Canada, Ontario, Canada. Manuscript received December 11, 1998; revised manuscript received April 28, 1999, and accepted April 29. Address for reprints: David A. Harrison, MD, Division of Cardiology, The Toronto Hospital, 12EN 221, 200 Elizabeth Street, Toronto, Ontario, Canada, M5G 2C4. ©1999 by Excerpta Medica, Inc. All rights reserved. The American Journal of Cardiology Vol. 84 September 15, 1999
0002-9149/99/$–see front matter PII S0002-9149(99)00415-4
677
TABLE I Total Population Compared With Subgroup With Hemodynamic Data Available Total Population (n ⫽ 109)
Parameter Age first visit (yrs) Women New York Heart Association functional class* I II III IV Trisomy 21 Simple anatomy Pregnancy Patients with event-free survival Heart or heart-lung transplantation Deaths
Hemodynamic Data (n ⫽ 59)
No Hemodynamics (n ⫽ 50)
28.6 ⫾ 10.8 62 (57%)
30.3 ⫾ 10.6 37 (63%)
26.6 ⫾ 10.6 25 (50%)
16 44 41 6 34 66 14 22 9 33
10 22 24 2 15 42 10 12 7 14
6 22 17 4 19 24 4 10 2 19
(15%) (41%) (38%) (6%) (31%) (61%) (23% of women) (20%) (8%) (30%)
(17%) (38%) (42%) (3%) (25%) (71%) (27% of women) (20%) (12%) (24%)
(12%) (45%) (35%) (8%) (38%) (48%) (16% of women) (20%) (4%) (38%)
*Functional class missing in 2 patients.
Electrocardiographic data was reported using an index for right ventricular hypertrophy defined as a continuous variable calculated as the sum of the R wave in lead V1 and the maximum depth of the S wave in leads V5 or V6 (modified from Chou6). Data analysis: Data are presented as mean ⫾ SD unless specified as median in the text. Comparison of means between groups was performed using 2-tailed Student’s t test. For purposes of analysis, patients were considered to have “simple” cardiac anatomy if the primary anatomic defect was limited to atrial septal defect, ventricular septal defect, or patent ductus arteriosus. All other patients were considered to have “complex” anatomy. The survival of patients since age 18 years is displayed using Kaplan-Meier curves. Patients who underwent lung or heart-lung transplantation were censored from the data analysis as alive at the time of their transplantation because their subsequent survival would no longer reflect the survival pattern for Eisenmenger syndrome. Risk factors for adverse outcome were determined using proportional-hazards analysis. A Cox multivariate analysis was undertaken using the following variables: age at presentation, gender, simple versus complex anatomy, trisomy 21, baseline New York Heart Association functional class, events before first assessment (including syncope, hemoptysis, congestive heart failure, supraventricular arrhythmia), phlebotomy, and electrocardiogram index. Oxygen saturation was not available in enough patients to allow inclusion in the analysis. The multivariate model was analyzed using a backward elimination algorithm with the level of significance set at 0.05.
RESULTS
Population: A total of 109 patients with Eisenmenger syndrome were identified (47 men). Hemodynamic data confirming systemic pulmonary artery pressure was available for 59 of these patients, and the remaining 50 patients were judged to have Eisenmenger physiology from clinical assessment. A comparison between the total population of 109 patients, 678 THE AMERICAN JOURNAL OF CARDIOLOGY姞
VOL. 84
the 59 patients with hemodynamic evaluation available, and the 50 patients without hemodynamic evaluation is shown in Table I. Patients without hemodynamic data were less likely to have simple anatomy, and there were no other differences between the 2 subgroups. For the total population the distribution of cardiac anatomy was ventricular septal defect (n ⫽ 43), atrioventricular septal defect (n ⫽ 19), atrial septal defect (n ⫽ 13), univentricular morphology (n ⫽ 11), patent ductus arteriosus (n ⫽ 10), transposition complexes (n ⫽ 7), truncus arteriosus (n ⫽ 3), pulmonary atresia with Pott’s shunt (n ⫽ 2), and tricuspid atresia with ventricular septal defect (n ⫽ 1). Average age at first clinic assessment was 28.6 ⫾ 10.8 years. Most patients (96 of 109) were symptomatic at referral, having either exertional dyspnea or fatigue (84% of total group). Other symptoms (not mutually exclusive) included: palpitations (21%), cough (13%), hemoptysis (11%), and syncope (9%). There were 16 (15%), 44 (41%), 41 (38%), and 6 (6%) patients in New York Heart Association functional classes I through IV, respectively (data not available in 2 patients). Trisomy 21 was present in 34 patients (31%), 17 of whom had simple cardiac anatomy, 16 had atrioventricular septal defect, and 1 had univentricular connection. Phlebotomy had already been initiated in 30 patients before presentation. Most patients (n ⫽ 71) were taking at least 1 medication including diuretics (n ⫽ 20), digoxin (n ⫽ 38), nitrates (n ⫽ 2), angiotensin-converting enzyme inhibitors (n ⫽ 3), antiarrhythmic drugs (n ⫽ 2), aspirin (n ⫽ 5), calcium antagonists (n ⫽ 5), other cardiac medications (n ⫽ 10), and other noncardiac medications (n ⫽ 37). Despite the fact that pregnancy is considered to be associated with prohibitive risk for patients with Eisenmenger syndrome, 14 of the 62 women in our group have had at least 1 pregnancy (26 total pregnancies). In many cases pregnancy was undertaken several years before referral to our center. Eisenmenger physiology at the time of pregnancy could only be confirmed in 5 pregnancies; the evolution of SEPTEMBER 15, 1999
TABLE II Summary of Mortality by Primary Anatomy Diagnosis Atrial septal defect Ventricular septal defect Patent ductus arteriosus Atrioventricular septal defect Other*
Total Patients 13 43 10 19 24
Death (%) 4 13 2 5 9
(31%) (30%) (20%) (26%) (38%)
Age at Death (mean ⫾ SD) 45.3 43.3 42.4 27.8 28.2
⫾ ⫾ ⫾ ⫾ ⫾
15.8 12.4 20.1 7.9 7.9
*Other diagnoses include: univentricular heart (n ⫽ 11), transposition of the great arteries (n ⫽ 7), truncus arteriosus (n ⫽ 3), and tricuspid or pulmonary atresia (with surgical aortopulmonary shunts; n ⫽ 3).
pulmonary artery hypertension in some or most patients was relatively early in the pregnancy. Of the 26 pregnancies, 6 were electively terminated, 3 ended in spontaneous abortion, and 17 continued to delivery. Primary anatomy for the 10 patients who completed at least 1 pregnancy was: atrial septal defect (n ⫽ 4), ventricular septal defect (n ⫽ 3), patent ductus arteriosus (n ⫽ 1), atrioventricular septal defect (n ⫽ 1), and single ventricle (n ⫽ 1). There were no maternal pregnancy-related deaths in this group. However, by the nature of this study, maternal deaths before 1980 would not have been included. Patients were followed for a median of 6.3 years (range 1 to 40). Nine patients underwent transplantation at average age 37 ⫾ 6.2 years (7 lung transplantation, 2 heart-lung transplantation) of whom 4 died (2 lung and 2 heart-lung). All post-transplantation deaths occurred during hospital admission for transplantation. Mortality was high in nontransplanted patients, with 33 deaths. Average age at death was 37 ⫾ 13.3 years. The distribution of mortality by underlying anatomy is given in Table II. Identified causes of death included: congestive heart failure (n ⫽ 14), sudden (n ⫽ 7), endocarditis (n ⫽ 2), cerebrovascular event (n ⫽ 2), postappendectomy (n ⫽ 1), massive hemoptysis (n ⫽ 1), noncardiac (n ⫽ 4), and unknown (n ⫽ 2). Morbidity was also high, with only 22 patients (20%) alive and free of any clinical event at last assessment. Documented clinical events included: hemoptysis (n ⫽ 34), admission for congestive heart failure (n ⫽ 29), syncope (n ⫽ 26), atrial fibrillation or atrial flutter (n ⫽ 18), and intracerebral abscess (n ⫽ 7). Figure 1 shows the Kaplan-Meier survival curve for the population. Median survival was 52.6 years. All patients were assumed to be enrolled at age 18 and transplanted patients were censored as alive at time of transplantation. These 2 assumptions increase the predicted survival of our population. If transplanted patients are assumed to have died at the time of transplantation, then median survival would be 49.0 years. Approximately half of the patients presented to the clinic before age 25. The median survival for these patients presenting as young adults was 38.2 years. Figure 2 shows the Kaplan-Meier survival curve for the subgroup of 59 patients for whom hemodynamic data were available (group 1) and is compared with
FIGURE 1. The survival curve for the total population of 109 patients with Eisenmenger syndrome is shown. Median survival is 52.6 years, assuming age at enrollment to be 18 years, and transplanted patients are censored as alive at the time of transplantation (see text). The number of patients alive are provided at each decade.
FIGURE 2. The survival curve of the subgroup of patients for whom complete hemodynamic data were available (solid line) is compared with survival for the group without hemodynamic data (broken line). Median survival was 52.6 years for the group with hemodynamic data available (mean ⴞ SE 51 ⴞ 2.1) and 51.4 years for those without hemodynamic data (mean 47.5 ⴞ 3.0).
the survival curve for the remaining population (group 2). The 2 curves almost overlap, with no difference in median survival for the 2 groups (group 1, 52.6 years, group 2, 51.4 years). A Cox multivariate logistic regression was performed in the total population using the 11 clinical variables as listed under the Methods section. Significant predictors for mortality included: age at presentation (hazard ratio [HR] 0.90/year), indicating a worse prognosis for younger age at presentation, increased precordial voltage (HR 1.61/mV), New York Heart Association class III or IV (HR 2.60), and a history of supraventricular arrhythmia requiring treatment (HR 3.44). Confidence intervals and significance levels are presented in Table III. The electrocardiogram index ranged from 0.2 to 6.4 mV. Mortality for patients with an electrocardiogram index ⬍3.0 mV was 21% (14 of 67 patients) compared with a mortal-
CONGENITAL HEART DISEASE/SURVIVAL IN THE ADULT EISENMENGER
679
TABLE III Multivariate Analysis Predictor Age at referral* RV1 ⫹ SV5 or V6 Atrial arrhythmia New York Heart association class III–IV Hemoptysis Phlebotomy Complex anatomy Gender Syncope Trisomy 21 Heart failure
Hazard Ratio
95% Confidence Interval
Significance Level
0.90 1.61 3.44 2.60
0.86–0.94 1.25–2.08 1.35–8.76 1.14–5.93
0.0001 0.0003 0.02 0.03
0.35 2.32 1.96 0.56 0.38 0.58 0.67
0.10–1.21 0.76–7.08 0.70–5.48 0.21–1.48 0.08–1.95 0.21–1.65 0.18–2.44
0.09 0.14 0.20 0.24 0.25 0.31 0.55
*Age at referral indicates increased risk for earlier age at referral (hazard ratio 0.90/year).
ity rate of 45% (19 of 42 patients) for an index ⱖ3.0 mV. Complex anatomy was not found to be a significant multivariate predictor for poor outcome. However, a relation was present between age at presentation and patient anatomy, with complex patients presenting early (23.7 ⫾ 7.2 years) versus those with simple anatomy (31.6 ⫾ 11.8 years), p ⫽ 0.001.
DISCUSSION
Estimate of survival: In this study of adult patients with the Eisenmenger syndrome we found a predicted median survival of 52.6 years based on a population of 109 patients. There were 33 deaths at an average age of 37.0 ⫾ SD 13.3 years. In previous studies, Wood7 reported 56 deaths with an average age at time of death of 33 years for aortopulmonary and ventricular defects and 26 years for atrial septal defects. Three other studies have observed a mortality rate of 10% to 44%, with an average age at death ranging from 25 to 32 years.8 –10 The lower age at death in 2 of these studies9,10 may be a result of the inclusion of pediatric patients. Our population has a substantially higher age at death than these previous reports, including the adult subpopulation of Young and Mark’s publication.8 Lung and heart-lung transplantations are the most significant changes to the management of the Eisenmenger patient in the 25 years that have passed since Young’s study. However, this does not appear to be the reason that our population has an improved survival over Young’s study (the average age at death or transplantation was 37.7 ⫾ 12.1 years), which still exceeds his observed age at death). Our clinic specializes in the adult patient (age ⬎18 years); therefore, part of our higher age at death may reflect a referral bias. However, we postulate that some of the improved survival reflects changes in contemporary management that may include more appropriate use of phlebotomy, improved perioperative anesthetic management,11 and appropriate counseling regarding pregnancy, all of which may contribute to improved survival in Eisenmenger patients. 680 THE AMERICAN JOURNAL OF CARDIOLOGY姞
VOL. 84
Cause of death: Cause of death for the adult patient is also slightly different from that reported in previous studies. Wood2,7 stressed the potential risk of hemoptysis, the cause of death in 29% of the deaths in his population. Death due to hemoptysis is similarly reported in 11% to 15% of other studies.9,10 In contrast, we found only 1 of 33 deaths (3%) to be attributable to massive hemoptysis, despite a high prevalence of hemoptysis within the population (31%). In many previous studies, the most common mode of death is sudden (range from 19% to 47% of reported deaths).8 –10 Sudden death represented 21% of the deaths in our study, but in contrast to other studies congestive heart failure was the predominant mode of death (observed in 42%). Mortality risk factors: Earlier age at presentation, the electrocardiogram index for right ventricular hypertrophy, a history of supraventricular arrhythmia, and poor functional class were significant predictors of mortality. The observation that patients who presented to the clinic at a younger age were at increased risk for early mortality is presumably a function of both a referral bias for complex patients obtaining early referral and for patient referral being biased to the sicker patient. We found the modified electrocardiogram index for right ventricular hypertrophy to be predictive of decreased survival. We believe that the index reflects the degree and duration of pulmonary hypertension. However, it is possible that the index may reflect right ventricular dilation, altered conduction in a failing ventricle (a proportion of these patients had right bundle branch block and were included with no adjustment of the index), or reflect underlying complex anatomy. The presence of supraventricular arrhythmia as a significant predictor of mortality may be a result of a failing right ventricle or could have contributed to clinical deterioration through the loss of preload. Previous studies have found higher mortality rates in patients with ventricular or aortopulmonary shunts than with atrial shunts.3,8 We could not confirm this because of the small patient numbers. However a similar age at death was detected for patients with atrial septal defect (45.3 ⫾ 15.8), ventricular septal defect (43.3 ⫾ 12.4), and patent ductus arteriosus (42.4 ⫾ 20.1), all of whom appeared to be older than those with atrioventricular septal defect (27.8 ⫾ 7.9) or other more complex anatomy (28.2 ⫾ 7.9). Other potential risk factors for death that have been identified include: syncope,10 noncardiac surgery,12 trisomy 21,9 hemoptysis,2,13 elevated right atrial pressure, and low arterial oxygen saturation.10 The variation of factors between these studies likely reflects the relatively small sample sizes of each study, the different management periods under observation, variation of study population age, and different center referral and follow-up bias. There is a need for a prospective evaluation of previously identified predictors of mortality optimize patient selection for intervention such as lung or heart-lung transplantation. The substantial mortality rate of our patients after transplantation SEPTEMBER 15, 1999
highlights the importance of selection of only patients at highest risk for death from Eisenmenger syndrome. Study limitations: As with all retrospective studies our results are limited by variation in patient care, follow-up, and investigation. However, most patients were followed through our center so that consistent clinical data are available. In addition, mortality, the primary outcome of the study, was available through provincial registries. Eisenmenger syndrome is infrequent; therefore, we have a small sample size that reduces the power of the study to recognize all potential important risk factors. However, only 1 study has a larger sample size.10 Our center is a tertiary care facility so that referral bias affects the composition of our population and the time of patient presentation (as is recognized by the importance of age at presentation as an important predictor for survival). However, as the major site for evaluation of adults with congenital heart disease in our region, we expect a large proportion of patients with this unusual problem to have been reviewed here. As a retrospective study we were not able to obtain hemodynamic data to confirm Eisenmenger physiology in 50 patients who were included based on clinical diagnosis. The inclusion of these patients had the potential to alter the survival of the total population. However, baseline characteristics between these 2 groups was minor, with a possibly higher proportion of complex patients and trisomy 21 patients in the group lacking catheterization data (potentially biasing
the results to a worse prognosis for patients without catheterization data). Analysis revealed that the outcome for these 2 groups was not different by either direct comparison or by multivariate analysis.
1. Eisenmenger V. Die angeborenen defecte der kammerscheidewand des herzen. Z Klin Med 1897;32:1–28. 2. Wood P. The Eisenmenger syndrome or pulmonary hypertension with reversed central shunt. Br Med J 1958;2:701–709. 3. Touzet AD, Corone A, Jullien JL, Corone P. Evolution spontanee et complication des syndromes d’Eisenmenger. Arch Mal Coeur Vaiss 1983;5:563–567. 4. Hallidie-Smith KA, Tortoledo R, Goodwin JF. Presentation and natural history of Eisenmenger syndrome. Br Heart J 1971;33:610. 5. Hosenpud JD, Bennett LE, Keck BM, Fiol B, Boucek MM, Novick RJ. The registry of the international society for heart and lung transplantation: fifteenth official report–1998. J Heart Lung Transplant 1998;17:656 – 668. 6. Chou T. Right ventricular hypertrophy. In: Electrocardiography in Clinical Practice. Philadelphia: WB Saunders, 1996:53– 68. 7. Wood P. The Eisenmenger syndrome or pulmonary hypertension with reversed central shunt. Br Med J 1958;2:755–762. 8. Young D, Mark H. Fate of the patient with Eisenmenger syndrome. Am J Cardiol 1971;28:658 – 669. 9. Corone S, Davido A, Lang T, Corone P. Outcome of patients with Eisenmenger syndrome. Apropos of 62 cases followed-up for an average of 16 years. Arch Mal Coeur Vaiss 1992;85:521–526. 10. Saha A, Balakrishnan KG, Jaiswal PK, Venkitachalam CG, Tharakan J, Titus T, Kutty R. Prognosis for patients with Eisenmenger syndrome of various aetiology. Int J Cardiol 1994;45:199 –207. 11. Ammash NM, Connolly HM, Abel MD, Warnes CA. Noncardiac surgery in Eisenmenger syndrome. J Am Coll Cardiol 1999;33:222–227. 12. Hallidie-Smith KA, Goodwin JF. The Eisenmenger syndrome. Prog Cardiol 1974;3:211–225. 13. Clarkson PM, Frye RL, Dushane JW, Burcheu HB, Wood EH, Weiman WH. Prognosis of patients with ventricular septal defect and severe pulmonary vascular obstructive disease. Circulation 1968;38:129 –135.
CONGENITAL HEART DISEASE/SURVIVAL IN THE ADULT EISENMENGER
681