Author + information
- Received June 17, 1996
- Revision received March 26, 1997
- Accepted April 17, 1997
- Published online August 1, 1997.
- Martha L Clabby, MDA,
- Charles E Canter, MD, FACCA,
- James H Moller, MD, FACCB and
- Nancy D Bridges, MD, FACCC,* ()
- ↵*Dr. Nancy D. Bridges, Cardiology Division, Children’s Hospital of Philadelphia, 34th Street and Civic Center Boulevard, Philadelphia, Pennsylvania 19104.
Objectives. Using data from a multi-institutional data base, we sought to determine whether hemodynamic data predict duration of survival in children with primary or secondary pulmonary hypertension.
Background. Lung transplantation is a therapeutic option for children with pulmonary hypertension. Appropriate timing of lung transplantation requires reliable methods of predicting duration of survival in potential candidates.
Methods. A regional data base was used to obtain cardiac catheterization data on 50 children with mean pulmonary artery pressure (mPAP) >25 mm Hg and indexed pulmonary resistance (Rp) >4.5 Wood units. Data on survival were obtained from the participating centers.
Results. There were 15 patients without congenital heart disease (group 1) and 35 patients with congenital heart disease (group 2) for analysis. Actuarial survival at 1, 2 and 5 years was 86%, 69% and 69% in group 1 and 88%, 88% and 77% in group 2, respectively (p = NS). Hemodynamic variables that predicted survival on univariate analysis were mean right atrial pressure (mRAP) (p < 0.0001), mPAP (p = 0.034), Rp (p < 0.0001) and pulmonary flow (p = 0.003), as well as a variable that we generated—mRAP × Rp (p < 0.0001). On multivariate stepwise logistic regression analysis, mRAP × Rp was independently related to survival. A model using mRAP × Rp allows for the estimation of probability of death at 1 and 2 years after catheterization.
Conclusions. Hemodynamic variables can predict survival in children with pulmonary hypertension in the presence or absence of congenital heart defects. This information can be used to determine the optimal timing of listing for lung transplantation.
Severe pulmonary hypertension may occur in children in association with congenital heart disease or as an isolated disease process. Previous reports suggest the prognosis for such patients is poor, but specific predictors of survival have not been defined [1–4]. Hemodynamic variables including lower mean right atrial pressure (mRAP), lower mean pulmonary artery pressure (mPAP) and a higher cardiac index have been shown to predict survival in adults with primary pulmonary hypertension . Lower mRAP correlated with survival in a small group of children with primary pulmonary hypertension evaluated in Mexico City . It is currently unknown whether hemodynamic variables can predict survival in pediatric patients with irreversible pulmonary hypertension secondary to congenital heart disease. Lung transplantation is now a therapeutic option for this condition; however, optimal utilization of this procedure requires that we reliably estimate an individual’s chance of survival without transplantation. We report a retrospective analysis of survival among children with pulmonary hypertension and elevated pulmonary vascular resistance, with or without congenital heart disease, who underwent cardiac catheterization. The purpose of this study was to determine whether hemodynamic variables can be used to predict survival in such patients.
1.1 Patient selection.
Patient histories and catheterization data were obtained from the Pediatric Cardiac Care Quality Assurance Consortium (PCCC), a data base of cardiac catheterization data and surgical procedures on pediatric patients from 28 medical centers (Appendix A). The data base, for the period 1982–1992, was searched for four diagnosis codes: fixed pulmonary vascular disease, pulmonary hypertension, primary pulmonary hypertension or secondary pulmonary hypertension. Patients were eligible for the study if they met the inclusion criteria listed in Table 1. The hemodynamic inclusion criteria were mPAP >25 mm Hg, indexed pulmonary flow (Qp) >4.5 Wood units, pulmonary artery wedge pressure ≤12 mm Hg if left atrial or left ventricular end-diastolic pressure had not been obtained and a complete set of hemodynamic data collected while the patient breathed room air. The clinical inclusion criteria were age at cardiac catheterization >120 days and <18 years, absence of left heart obstructive lesions, absence of dilated cardiomyopathy, absence of lesions with inadequate pulmonary vascular bed (i.e., tetralogy of Fallot with pulmonary atresia and hypoplastic pulmonary arteries), absence of surgical intervention after the most recent cardiac catheterization, absence of lung disease as stated by the contributing center and at least one point of follow-up available.
1.2 Cardiac catheterization data.
Cardiac catheterizations were performed at 15 different centers, and cardiac catheterization reports were analyzed retrospectively. For the purposes of this study, a complete and accurate hemodynamic evaluation was defined as follows: 1) arterial pH maintained between 7.30 and 7.50; 2) hemoglobin measured and recorded; 3) pressure measurements made in the right atrium, main pulmonary artery, pulmonary capillary wedge (or left atrium), left ventricle (if pulmonary artery wedge pressure was >12 mm Hg) and aorta; 4) oxygen saturation measured in the superior vena cava, right atrium, pulmonary artery, either pulmonary veins, left atrium or left ventricle and aorta; 5) cardiac index determined in one of the following ways: the Fick method, using a value for oxygen consumption that was measured in a flowthrough hood or that could be estimated using the patient’s age, heart rate and gender using the tables of LaFarge and Miettinen ; or indicator dye or thermodilution in children in whom no intracardiac shunt was demonstrated by oximetry.
Cardiac anatomic diagnoses were accepted as stated in the catheterization report; however, if the hemodynamic data were in conflict with the anatomic diagnosis, the patient was excluded (on the basis of inadequate catheterization data). Cardiac catheterization data and registry reports contained date of birth, date of catheterization and date(s) of surgical repair, as well as information on systemic diseases and chromosomal abnormalities. For hemodynamic variables in which cardiac catheterization data are used to calculate values, all values were recalculated and adjusted if there was a discrepancy. Among patients who had undergone multiple catheterizations, data from the single most recent complete catheterization were used to correlate clinical outcome with the most recent hemodynamic data. Duration of survival was measured from the time of catheterization to the point of last follow-up or death. Patients who underwent lung transplantation were censored on the date of transplantation. Although many patients had evaluation of their response to pulmonary vasodilators, including oxygen, this sort of testing was not standardized and thus no attempt has been made to investigate the significance of reactivity of the pulmonary vascular bed. In patients with complete atrioventricular canal defects (CAVCD), pulmonary artery wedge pressures were not obtained. The left atrial pressure was used in the calculation of their indexed pulmonary arteriolar resistance (Rp); pulmonary vein stenosis has not been ruled out in these patients.
1.3 Follow-up information.
Follow-up information was requested from the participating centers (Appendix A). Patients were identified by a registry identification number, which was sent to participating centers along with a form asking for the following information: date of last clinic visit; transplantation status (i.e., had the patient received a transplant and if so what type and on what date?); and for patients who had died, date of death and cause of death. In some cases, this information was gathered by a representative of the participating center. In others, one of the investigators (M.L.C.) reviewed clinic charts and hospital records to complete the follow-up form. Information on medications was not obtained.
1.4 Statistical methods.
Statistical analysis was performed using STATA 4.0 software (Stata Corporation). Analysis of demographic and hemodynamic characteristics of patients with or without congenital heart disease was performed using the Wilcoxon rank-sum test with the Bonferroni correction. Univariate and multivariate analyses were performed using the Cox proportional hazards model . Duration of survival was entered as the dependent variable, and the following were entered as independent variables—age, gender, hemoglobin, group (i.e., 1 or 2), presence of trisomy 21, mRAP, mPAP, mean aortic pressure (mAOP), mPAP/mAOP, Qp, indexed systemic flow (Qs), Rp, indexed systemic resistance (Rs) Qp/Qs and Rp/Rs. In addition, we created another variable, the product of the mRAP and the Rp (mRAP × Rp), as a measure of both the pressure load on the right ventricle and how well the right ventricle handles that load. Results were expressed as odds ratios with 95% confidence intervals (CIs). For continuous variables, the odds ratios were expressed per unit of measurement. Forward stepwise logistic regression was used to generate the equation used in the predictive model . The Kaplan-Meier method was used to estimate actuarial survival. The log-rank test was used to compare survival curves of two or more groups. The Fisher exact test was used to assess the significance of frequencies. Throughout the statistical analysis, p <0.05 was considered significant, using the Bonferroni correction where appropriate.
2.1 Patient group.
The search of the PCCC data base yielded 158 patients who initially appeared to meet the inclusion criteria. At this point, follow-up information was requested from the participating centers, and the catheterization reports were further scrutinized. Patients were excluded for the following reasons: pH not reported in the catheterization report or pH outside the specified range (n = 46), incomplete catheterization report (n = 39), absence of data on room air (n = 15), lung disease (n = 3), age outside the specified range (n = 2) and lack of follow-up (n = 3). Thus, 50 patients were suitable for further analysis. Patients were classified into two groups: group 1 consisted of 15 patients without congenital heart disease. Within group 1, there were four patients who had other medical problems—one each of the following: mosaic triploidy, type I hyperlipidemia and primary hyperaldosteronism, insulin-dependent diabetes mellitus with seizure disorder and mental retardation with seizure disorder. Group 2 consisted of 35 patients with congenital heart disease. The patient profiles are shown in Table 2. The two groups were comparable with respect to age, hemoglobin, Qs, mRAP, mPAP, mAOP, Rp, Rs, Qp/Qs, Rp/Rs and gender. They differed in the incidence of trisomy 21 (43% in group 2 and 0% in group 1, p = 0.002) and duration of follow-up (5.1 ± 3.6 years in group 2 compared with 2.6 ± 2.8 years in group 1, p = 0.009).
The patients in group 2 had a spectrum of congenital heart defects: ventricular septal defect (VSD) in eight patients, CAVCD in six, CAVCD with patent ductus arteriosus in four, VSD with patent ductus arteriosus in four, VSD with atrial septal defect in three, atrial septal defect with patent ductus arteriosus in one, patent ductus arteriosus only in one and atrial septal defect only in one. In addition, another seven patients had one each of the following complex lesions: VSD with patent ductus arteriosus and pulmonary stenosis, “corrected transposition” with VSD, double-inlet left ventricle with pulmonary stenosis, interrupted aortic arch type A with patent ductus arteriosus and VSD, truncus arteriosus, transposition of the great vessels with VSD and transposition of the great vessels with an intact ventricular septum. Among these patients with congenital heart disease, 14 (40%) were unrepaired, 5 (14%) were palliated, 9 (26%) had undergone surgical repair but had residual defects on angiography as reported in the catheterization summary and 7 (20%) were completely repaired. The mean age at complete repair was 4.1 ± 3.7 years (range 1.4 to 12.1). The spectrum of congenital heart defects in the patients with trisomy 21 was CAVCD in six patients, CAVCD with patent ductus arteriosus in three, VSD with patent ductus arteriosus in three, VSD in one and atrial septal defect with patent ductus arteriosus in one. Of the patients with trisomy 21, nine (60%) were unrepaired, one (7%) was palliated, four (26%) had undergone surgical repair but had residual defects and only one (7%) had been completely repaired. The low rate of complete repair in the patients with trisomy 21 compared with those without trisomy 21 was significant (p = 0.04). Right to left shunting was present in four patients (27%) from group 1 and 10 patients (29%) from group 2. No patients in group 2 had systemic disease apart from trisomy 21.
Complete follow-up was defined as knowledge of whether the patient was alive or had survived to lung transplantation. Patients were censored at the time of transplantation for calculation of actuarial survival. Among patients in group 1, complete follow-up was available in 93% at 1 year, in 73% at 2 years and in 47% at 3 years. Among patients in group 2, complete follow-up was available in 97% at 1 year, in 94% at 2 years and in 77% at 3 years. Actuarial survival for all patients is shown in Fig. 1and for each of the two groups in Fig. 2. The actuarial survival for group 1 is 86% (95% CI 55% to 96%) at 1 year, 69% (95% CI 35% to 87%) at 2 years and 69% (95% CI 35% to 87%) at 5 years. The actuarial survival for group 2 is 88% (95% CI 72% to 95%) at 1 year, 88% (95% CI 72% to 95%) at 2 years and 77% (95% CI 57% to 88%) at 5 years. Group 2 was further divided into patients with or without trisomy 21. Actuarial survival for 1 year, 2 years and 5 years was 93%, 93% and 93% (95% CI 61% to 99%) for those with trisomy 21 and 85% (95% CI 60% to 95%), 85% (95% CI 60% to 95%) and 79% (95% CI 53% to 92%) for those without trisomy 21, respectively. The differences in survival are not significant.
There were four deaths in group 1 (27%) and seven deaths in group 2 (20%). The causes of death included sudden death at home in three patients, progressive right heart failure in two, sudden death in hospital in one, procedure-related in four, (two after cardiac catheterization, one during lung biopsy and one after hemothorax during Swan placement to assess response to pulmonary vasodilators) and circumstances unknown in one.
For the purposes of further analysis, the two groups were combined. The results of the univariate regression analysis are summarized in Table 3. For continuous variables, the odds ratios are expressed per unit of measurement, with ranges for the individual measurements listed. The factors that were found to be significant predictors of survival were lower mRAP (p < 0.0001), lower mPAP (p = 0.034), lower Rp (p < 0.0001), lower Qp (p = 0.003) and lower mRAP × Rp (p < 0.0001). Taking into account colinearity, permutations of these variables were entered into a Cox multivariate analysis. The variables Qp and PAP did not retain significance. The variable found to have the best maximal likelihood of fit was mRAP × Rp (p < 0.0001). This variable retained significance in the multivariate regression analysis when group 1 and group 2 were analyzed separately. Fig. 3demonstrates the effect of mRAP × Rp on survival. Patients with mRAP × Rp >160 mm Hg2/liter per min per m2have a significantly lower predicted survival than those with a lower mRAP × Rp (p < 0.0001). Patients with a mRAP × Rp of 161 to 260 mm Hg2/liter per min per m2had an estimated 2-year survival rate of 65% (95% CI 25% to 87%), and those with mRAP × Rp >260 mm Hg2/liter per min per m2had an estimated 2-year survival rate of 35% (95% CI 5% to 68%). In Fig. 3, superimposed on the survival curves from patients in this study is the actuarial survival curve of pediatric patients with cardiovascular disease undergoing lung transplantation from a previously reported study .
A model was generated using patients with complete follow-up at 1 and 2 years after catheterization and the variable mRAP × Rp in the logistic regression equation to predict the probability of death (p):
At 1 year, a = −3.8, b = 0.010 and p = 0.0006; at 2 years, a = −3.4, b = 0.011 and p = 0.0003.
For example, a patient with an mRAP of 5 mm Hg and Rp of 8 Wood units has a probability of death at 1 year after catheterization of 3.2%, and at 2 years 4.9%. Similarly, a patient with an mRAP of 12 mm Hg and Rp of 32 Wood units has a 58% probability of death at 1 year and 70% at 2 years.
Of the 47 patients with complete follow-up at 1 year, the model estimated the probability of death at 1 year to be >50% in four patients, three (75%) of whom were dead at 1 year. Of 43 patients in whom the model predicted the probability of death at 1 year to be <50%, 39 (93%) were alive at 1 year. The model predicted the probability of death to be >50% at 2 years in four patients, three (50%) of whom were dead at 2 years. In the 40 patients in whom the model predicted the probability of death to be <50% at 2 years, 36 (90%) were alive at 2 years.
3.1 Predicting survival in children with pulmonary hypertension.
This retrospective analysis represents the largest published series of pediatric patients with pulmonary hypertension in whom complete cardiac catheterization data are available for analysis, and it demonstrates that hemodynamic variables (lower mRAP, lower mPAP, lower Qp, lower Rp and lower mRAP × Rp) can estimate predicted survival in these patients. When multivariate analysis was applied to the data, low mRAP × Rp emerged as a factor independently related to survival. This variable, generated by multiplying the mRAP times the Rp, was formulated by us to represent the pressure load on the right ventricle (Rp) and how well the right ventricle handles that load (mRAP). Use of this model allows for the estimation of the probability of death 1 and 2 years after catheterization. This information could be particularly useful in the management of children being evaluated for lung transplantation.
Previous studies of children with pulmonary hypertension and congenital heart disease have correlated histologic or angiographic findings with hemodynamic variables [11–14], but none have examined the relation between hemodynamic data and survival. Sandoval et al. reported on 18 children with primary pulmonary hypertension who had undergone cardiac catheterization. Their patients had a median survival of 4.1 years from the time of diagnosis; lower mRAP and higher stroke volume index emerged as predictors of survival. Other investigators have quoted a mean survival of 1 year from the time of diagnosis among children with primary pulmonary hypertension [1, 15]. In a prospective study of adults with primary pulmonary hypertension through the National Institutes of Health registry, lower mRAP, lower PAP and a higher cardiac index correlated with survival .
3.2 Implications for listing for lung transplantation.
Although medical therapy, including calcium channel blockers , anticoagulation and continuous intravenous infusion of prostacyclin [18, 19], have been shown to prolong survival in patients with primary pulmonary hypertension, there is no definitive therapy for this progressive disease. Lung transplantation has become a therapeutic option for children with pulmonary hypertension, with or without congenital heart disease, when medical management fails [10, 20, 21]. The optimal time to list patients who choose this therapy is unclear; it will be influenced by predicted survival without lung transplantation, the waiting time for donor organs and the predicted survival after lung transplantation. Information on waiting times for donor organs and actuarial survival after lung transplantation is available from the International Society for Heart and Lung Transplantation . That information, combined with hemodynamic data, can be used to make rational decisions regarding the timing of listing patients for lung transplantation.
Fig. 3shows the actuarial survival among the study patients, stratified by the product variable mRAP × Rp. Superimposed on these survival curves is the actuarial survival among children with cardiovascular disease undergoing lung transplantation . Patients with a mRAP × Rp <160 mm Hg2/liter per min per m2would be expected to survive longer without lung transplantation than with lung transplantation, and thus this intervention would be unwarranted. Patients with a mRAP between 161 and 260 mm Hg have survival comparable to children undergoing lung transplantation; those with a mRAp × Rp >260 mm Hg2/liter per min per m2have a worse survival than those undergoing lung transplantation. Current experience indicates that patients who survive the perioperative period of lung transplantation have better exercise capacity and subjectively improved quality of life than the pretransplant patients with severe pulmonary hypertension [10, 23]. Thus, a reasonable strategy would be to list a patient for lung transplantation when the mRAP × Rp exceeds 160 mm Hg2/liter per min per m2.
The rate of disease progression among children with pulmonary hypertension remains unknown. Until recently, cardiac catheterization has been used to confirm the diagnosis of irreversible pulmonary hypertension, with subsequent cardiac catheterization studies considered to carry substantial risk with little benefit . However, as hemodynamic variables obtained with cardiac catheterization can help predict survival probability, the current availability of lung transplantation for previously untreatable disease(s) implies that serial cardiac catheterizations should play a role in managing these patients.
3.3 Primary versus secondary pulmonary hypertension.
Although it is generally recognized that patients with primary pulmonary hypertension have a shorter life expectancy than those with pulmonary hypertension in association with congenital heart disease, our analysis did not reveal a survival difference based on diagnosis. Survival in our study was measured from the time of cardiac catheterization, not from the time of disease onset; the latter is unknowable in this group of patients. Our findings indicate that elevated RAP and Rp predict a shorter duration of survival, regardless of the diagnosis. It may very well be the case that it takes longer to reach a given RAP and Rp when the underlying disease is congenital heart disease rather than primary pulmonary hypertension. Our study does not and cannot address this question.
3.4 Study limitations.
A retrospective study using patients from a registry has inherent shortcomings. The use of precisely defined inclusion criteria, particularly with respect to cardiac catheterization data, selected a specific target patient group but also reduced the number of patients available for study. We did not detect a difference in the variables that predict survival in group 1 versus group 2; the number of patients in this study may have been too small to detect differences. However, in subgroup analysis, all the stated predictors retained their significance. There are a number of important characteristics that we were unable to study. These include the predictive value of New York Heart Association functional class, the effects of medical treatment on survival and the predictive value of response to vasodilators. Because they have been shown to increase survival, calcium channel blockers and prostacyclin are used with increasing frequency; thus, a study of the “natural history” of pulmonary hypertension (i.e., without vasodilator therapy) is not possible. For this reason, the largest and most complete available study of survival in patients with primary pulmonary hypertension was carried out in a patient group that was heterogeneous with regard to medical management , and we have had to do the same. Whether the predictive value of hemodynamic variables is changed by such therapy is unknown.
In children with pulmonary hypertension, in the presence or absence of congenital heart disease, hemodynamic variables can estimate survival probability. This information may be helpful to physicians caring for children with severe pulmonary hypertension in whom lung transplantation is being considered. Serial cardiac catheterizations in regional centers specializing in the evaluation of these high risk patients may now yield specific information with which to estimate survival and may serve to define the rate of disease progression. Further studies that address functional status and response to pulmonary vasodilators are needed. Prospective application of this model to children with pulmonary hypertension will be necessary to validate this preliminary study.
Centers Participating in the Pediatric Cardiac Care Quality Assurance Consortium That Contributed Data for This Study
All Children’s Hospital, St. Petersburg, Florida; Arkansas Children’s Hospital, Little Rock, Arkansas; Cardinal Glennon Hospital, St. Louis, Missouri; Children’s Health Care–Minneapolis, Minneapolis, Minnesota; Children’s Health Care–St. Paul, St. Paul, Minnesota; Children’s Medical Center of Dallas, Dallas, Texas; Children’s Mercy Hospital, Kansas City, Missouri; Columbus Children’s Hospital, Columbus, Ohio; Children’s National Medical Center, Washington, DC; Fargo Clinic, Fargo, North Dakota; LeBonheur Children’s Medical Center, Memphis, Tennessee; Mayo Clinic, Rochester, Minnesota; Mercy Hospital Medical Center, Des Moines, Iowa; Oschner Foundation Hospital, New Orleans, Louisiana; St. Louis Children’s Hospital, St. Louis, Missouri; St. Luke’s Regional Medical Center, Boise, Idaho; University of Arizona Health Sciences Center, Tucson, Arizona; University of Iowa Hospitals and Clinic, Iowa City, Iowa; University of Kentucky Medical Center, Lexington, Kentucky; University of Miami School of Medicine, Miami, Florida; University of Minnesota Hospital, Minneapolis, Minnesota; University of Mississippi Medical Center, Jackson, Mississippi; University of Nebraska Medical Center, Omaha, Nebraska; and West Virginia University, Morgantown, West Virginia.
☆ Dr. Clabby is supported by the American Academy of Pediatrics through the Pediatric Scientist Development Program.
- complete atrioventricular canal defect
- mean aortic pressure
- mean pulmonary artery pressure
- mean right atrial pressure
- Pediatric Cardiac Care Quality Assurance Consortium
- indexed pulmonary flow
- indexed systemic flow
- indexed pulmonary arteriolar resistance
- indexed systemic resistance
- ventricular septal defect
- Received June 17, 1996.
- Revision received March 26, 1997.
- Accepted April 17, 1997.
- The American College of Cardiology
- Thilenius OG,
- Nadas AS,
- Jockin H
- Cartmill TB, DuShane JW, McGoon DC, Kirklin JW. Results of repair of ventricular septal defect. J Thorac Cardiovasc Surg 1966;52:486–50.
- Clarkson PM,
- Frye RL,
- DuShane JW,
- Burchell HB,
- Wood EH,
- Weidman WH
- Sandoval J,
- Bauerle O,
- Gomez A,
- Palomar A,
- Guerra MLM,
- Furuya ME
- LaFarge CG,
- Miettinen OS
- Cox DR
- Pagano M,
- Gauvreau K
- Rabinovitch M,
- Haworth SG,
- Castenada AR,
- Nadas AS,
- Reid LM
- Rabinovitch M,
- Keane JF,
- Fellows K,
- Castaneda AR,
- Reid LM
- Rabinovitch M,
- Keane JF,
- Norwood WI,
- Castenada AR,
- Reid LM
- Fuster V,
- Steele PM,
- Edwards WD,
- Gersh BJ,
- McGoon MD,
- Frey RL
- Rhodes J,
- Barst RJ,
- Garofano RP,
- Thoele DG,
- Gersony WM