| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Guest Access | Sign In via User Name/Password
|
|
|
|||||||
Guest Access | Sign In via User Name/Password |
|||||||||
* From the Pulmonary and Critical Care Medicine Service (Dr. Shorr), Walter Reed Army Medical Center, Washington, DC; the United Network for Organ Sharing (Dr. Davies), Richmond, VA; and the Inova Fairfax Transplant Center (Dr. Nathan), Fairfax, VA.
Correspondence to: Andrew Shorr, MD, MPH, Pulmonary and Critical Care Medicine, Walter Reed Army Medical Center, 6900 Georgia Ave, NW Washington, DC 20307; e-mail: afshorr{at}dnamail.com
 |
Abstract |
|---|
 TOP Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Design: Retrospective cohort study of patients with sarcoidosis listed for lung transplant in the United States between 1995 and 2000. After identifying important risk factors for death, we developed a mortality prediction model based on an inception cohort of 75% of the subjects. The remaining 25% of the individuals served as a validation cohort for determining the validity of the model.
Setting and patients: All patients with sarcoidosis in the United States irrespective of referral center listed for lung transplantation between 1995 and 2000.
Measurements and main results: Adequate follow-up data were available for 405 patients, and 111 patients (27.4%) died while awaiting lung transplantation. Neither patient age nor gender correlated with mortality. Survivors and nonsurvivors did not differ based on the results of spirometric testing. African Americans faced a significantly increased risk of death, which persisted after controlling for other confounders (odds ratio, 2.5). The amount of supplemental oxygen used and the mean pulmonary artery pressure were the only other variables predictive of mortality. The mean (± SD) pulmonary artery pressure in those who survived was 31.7 ± 11.5 mm Hg, compared to 41.4 ± 14.4 mm Hg in nonsurvivors (p < 0.01). Survivors required 2.2 ± 2.0 L/min of oxygen vs 2.9 ± 1.7 L/min in those who died awaiting transplant (p < 0.01). Differences in pulmonary artery pressures did not reflect differences in cardiac status, as the pulmonary capillary wedge pressure and the cardiac index were similar in survivors and nonsurvivors. The final mortality prediction model included three variables: race, amount of supplemental oxygen needed, and mean pulmonary artery pressure. Based on the validation cohort, the concordance of the model for death within 2 years of listing was 0.61 (95% confidence interval, 0.47 to 0.76), indicating only moderate explanatory power.
Conclusions: Race, pulmonary hypertension, and oxygen use are important factors indicative of mortality in this population. Specific guidelines for determining time of referral for transplantation in advanced sarcoidosis should be developed. Recommendations extrapolated from data for other types of interstitial lung disease may not be applicable in sarcoidosis. The independent effect of race on outcome is troubling.
Key Words: lung transplantation • mortality • pulmonary hypertension • race • sarcoidosis
|
Introduction |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
For patients with end-stage pulmonary sarcoidosis, orthotopic lung transplantation (OLT) represents one alternative.7 Immunosuppresive therapies such as methotrexate are of unproven benefit in cases of end-stage sarcoidosis, while supportive measures often prove frustrating. Organ transplantation is now frequently offered to patients with a number of different chronic respiratory diseases such as cystic fibrosis, COPD, and idiopathic pulmonary fibrosis (IPF).8 9 Additionally, several small series suggest that outcomes for patients with sarcoidosis undergoing OLT are comparable to those transplanted for other diseases.10 11 12 Reflecting these facts is the increasing use of OLT in sarcoidosis. Unfortunately, little information exists to guide clinicians attempting to determine when to place a patient with sarcoidosis on the list for OLT. As a general principle, OLT is considered appropriate when 2-year mortality without OLT is approximately 50%.13 For both cystic fibrosis and IPF, prediction equations have been devised to help estimate intermediate-term mortality.14 15 16 The relative rarity of end-stage sarcoidosis, however, hampers efforts to create mortality prediction models for this disease.
To address this issue, we conducted a retrospective review of all patients who were on the registration list for OLT in the United States between 1995 and 2000 to identify variables associated with 2-year mortality. We then created a model based on this sample to predict the short-term risk for death. We hypothesized that certain factors such as the presence of pulmonary hypertension would be important predictors of mortality. We and others have shown that pulmonary hypertension is frequent in subjects with sarcoidosis awaiting OLT.17 18 The implications of pulmonary hypertension for 2-year mortality in sarcoidosis, though, are unclear.
|
Materials and Methods |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Study Variables
Data regarding patient demographics (age, gender, and race) were recorded. We also noted information concerning pulmonary function and vascular comorbidities. For pulmonary function, both the FEV1 and the FVC were examined. The UNOS does not maintain data regarding the diffusion capacity of the lung for carbon monoxide, and only began collecting the FEV1/FVC ratio in 1999. We therefore did not include these variables in the univariate analysis. The PCO2 and the need for supplemental oxygen were also abstracted. Interestingly, PaO2 is not recorded. For vascular disease, we determined if hypertension, angina, cerebrovascular disease, or peripheral vascular disease were present. When available, information regarding pulmonary artery pressures, the pulmonary capillary wedge pressure, and cardiac index was also collected. For each of the variables, the model inputs reflect information available at time of listing. The analysis fails to account for issues that arise after listing, since this information is not uniformly reported to the UNOS. Cause of death while awaiting OLT is also not regularly reported to the UNOS.
Statistics and Model Building
Univariate analyses were performed on the variables of interest. Continuous data are reported as mean ± SD. Categorical variables were compared with the 
2 test. We utilized the Wilcoxon rank-sum test, as it is a more robust test for analysis of continuous variables. All tests were two tailed; p < 0.05 was assumed to represent statistical significance.
Using a Cox proportional hazard model, potential risk factors were initially analyzed for a significant association with mortality while on the waiting list. In order to obtain a prediction equation and to determine its validity, a split-sample, bootstrapping method was used to identify risk factors.19
For model fitting, a training data set of 300 subjects (approximately 75% of the entire population) was chosen at random from the overall data set. Using all candidate variables, the Cox model was fit in each of 5,000 bootstrapped samples. The significance level for a covariate to be introduced into the model was 0.20; for the covariate to remain in the model, we required a significance level of 
0.05. The significant variables from the Cox model were noted in each of the samples. The six risk factors that were significant with the greatest frequency were chosen for the final model. The model was then fit using the original training set. The remaining 25% of the subjects not examined for the model creation were used for validation. Ninety-five percent confidence intervals (CIs) are reported where appropriate. Analyses were done using the SAS version 8.0 (SAS Institute; Cary, NC) software package.
|
Results |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
As shown in Table 1 , neither age nor gender differed between those who survived and those who died while on the transplant list. For example, the mean age of survivors was 45.6 ± 8.7 years, as compared to 44.7 ± 8.1 year for nonsurvivors (p = 0.47). Similarly, approximately two thirds of both decedents and survivors were female (65.6% vs 67.6%, respectively). Race was the only demographic factor associated with an increased risk of mortality. Although African Americans comprised 70.3% of the entire cohort, 31.9% of these individuals died, while the mortality rate for others was 16.7% (p < 0.01). African Americans faced a more than twofold increase in the risk of mortality (unadjusted odds ratio for death, 2.31; 95% CI, 1.31 to 4.12).
|
For comorbidities, neither the incidence of hypertension nor cerebrovascular disease separated survivors from nonsurvivors, since both of these processes occurred infrequently in the cohort. There was, however, a trend (p = 0.12) toward a difference in the prevalence of angina. Six of 111 persons (5.4%) who died had angina, as compared to 7 of 294 survivors (2.4%). The odds ratio for death in those with angina was 2.34 (95% CI, 0.68 to 7.98).
Table 2 demonstrates findings from invasive hemodynamic monitoring. All parameters (cardiac index, pulmonary artery systolic pressure, pulmonary artery diastolic pressure, mean pulmonary artery pressure, and pulmonary artery wedge pressure) met the threshold for entry into the mortality prediction model. The cardiac index was slightly lower in sarcoidosis patients who died while waiting for transplant (2.6 ± 0.7 L/min/m2 vs 2.8 ± 0.6 L/min/m2, p = 0.11). We observed a similar trend toward a difference in pulmonary artery wedge pressure. The pulmonary artery wedge pressure was 10.0 ± 5.2 mm Hg in survivors, while it measured 11.2 ± 5.7 mm Hg in decedents (p = 0.06). We noted a more striking variation in assessments of pulmonary artery pressures. For example, although the mean pulmonary artery pressure for both groups was elevated, it was nearly 33% higher in individuals with sarcoidosis who eventually died awaiting transplant (41.4 ± 14.4 mm Hg vs 31.7 ± 11.5 mm Hg, p < 0.01). Accounting for the differences seen in mean pulmonary artery pressure, the pulmonary artery systolic and diastolic pressures were 48.3 ± 16.7 mm Hg and 22.1 ± 9.6 mm Hg, respectively, for those surviving and 62.5 ± 20.3 mm Hg and 29.3 ± 22.1 mm Hg, respectively, in subjects who eventually died while listed (p < 0.01 for both comparisons).
|
|
As an example, for an African-American patient with sarcoidosis who required 3.0 L/min of supplemental oxygen and had a mean pulmonary artery pressure of 40.0 mm Hg, x would equal 1.25. This would translate in to a 58% probability of survival. If the same patient were white, the value for x would be 0.35 and yield an 80% probability of living 2 years.
The model had only moderate ability to predict death. Based on the validation cohort, the concordance of the model for death within 2 years of listing is 0.66 (95% CI, 0.52 to 0.81). This implies that the model correctly discriminates between persons with sarcoidosis who die and those who survive in 65% of the cases.
|
Discussion |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
One earlier study18 investigated mortality in persons with sarcoidosis listed for OLT. In an analysis of 43 subjects from one center, Arcasoy and coworkers18 reported a crude mortality rate of 53%. In agreement with our findings, they noted an association between pulmonary hemodynamics and mortality. Specifically, they observed that mean pulmonary artery pressure was the only variable that was independently predictive of death. A right atrial pressure of > 15 mm Hg conferred a 5.2-fold increase in the risk of death. The present investigation, however, builds on this prior effort in several important ways. First, our cohort is significantly larger and representative of a multicenter experience. As such, our findings are more likely to be generalizable to all patients with sarcoidosis awaiting OLT. Second, we were able to identify factors other than pulmonary hemodynamics that were associated with mortality. This likely results from the larger sample size of our investigation, which, in turn, yielded greater statistical power. Third, the model we developed was more rigorous in that we employed a bootstrapping approach rather than a more limited evaluation of the data. We were able to show that although creation of a mortality prediction model is possible for sarcoidosis and that issues not routinely considered in risk stratification may be important. Conversely, since the data set represents reports from multiple, different centers, there is likely variability in practice style. Thus, while the pooling of cases from a large database affords added power to any analysis, it simultaneously allows for potential bias and confounding since patients are not managed in a standard fashion.
In more heterogeneous sarcoidosis populations, radiographic stage, the presence of pulmonary symptoms, and the results from spirometry have been demonstrated to predict outcome.1 2 Likewise, the presence of extrapulmonary sarcoidosis suggests chronic disease progression, while Lofgren syndrome is associated with an excellent prognosis.1 2 Baughman et al20 reported that vital capacity and radiographic evidence of fibrosis increased the risk for respiratory failure and death in a review of 479 patients followed up for at least 7 years. In their analysis, although vital capacity and fibrosis were the only variables that suggested an increased risk of mortality, they had marginal utility as screening tests for mortality. For example, the positive predictive value for death of a vital capacity of < 1.5 L was only 24%.20 We did not observe a relationship between pulmonary function and mortality. The limited predictive value of our model is in agreement with the conclusions of Baughman and colleagues,20 that no one variable measured at a given point in time is completely explanatory.
One troubling finding of our study was the relationship between race and mortality. For several disease states, associations between race and outcome have been found to essentially reflect differences in socioeconomic status and access to care. In other chronic diseases, such as end-stage renal disease, race has an independent impact on mortality separate from possible financial and socioeconomic concerns.21 Race is also associated with outcomes from both renal and liver transplantation.22 23 For sarcoidosis, several factors may explain the link between race and mortality. First, patients with sarcoidosis may be listed for transplant when their disease is more advanced. However, when we controlled for makers of disease severity race remained the most important correlate of mortality. Second, access to care after listing for OLT may play a role. Although as ill as their peers in other races when finally listed for OLT, African Americans may not receive adequate follow-up and may seek medical attention less frequently; therefore, their disease may progress more rapidly. Third, biologically sarcoidosis may behave differently in persons based on their race. Supporting this possibility is the observation that whites are significantly more likely to present with Lofgren syndrome than are African Americans.1 Additionally, several researchers have reported differences in human leukocyte antigen haplotypes in patients with sarcoidosis as a function of race.24 25
Unlike race, however, the implications of our results in terms of pulmonary hemodynamics are clear. The development of secondary pulmonary hypertension in COPD is associated with worse outcomes.26 In primary pulmonary hypertension, hemodynamic parameters also correlate strongly with mortality.27 For interstitial lung disease, the overall importance of pulmonary hypertension is not well studied. It appears evident, however, from our observations and from the report by Arcasoy et al,18 that pulmonary hypertension in subjects with sarcoidosis is an ominous sign. The evolution of pulmonary hypertension is not reflective of either cardiac sarcoidosis or chronic left ventricular failure as evidenced by the nearly normal cardiac indexes and pulmonary wedge pressures among both survivors and nonsurvivors. Therefore, the mechanism of pulmonary hypertension in these patients is possibly related to progressive fibrosis and hypoxemia. However, little is known about the biology of the pulmonary vasculature in sarcoidosis, and one report28 described pulmonary hypertension in the absence of significant parenchymal disease or spirometric limitations in a series of 106 patients with sarcoidosis.
Since pulmonary hypertension adversely affects outcome, physicians should consider prospectively evaluating patients with advanced sarcoidosis for this condition. Noting pulmonary hypertension may prompt earlier referral to a transplant center and, in turn, may improve outcomes. More importantly, treatments for pulmonary hypertension exist. For example, a diagnosis of pulmonary hypertension should prompt an assessment for occult oxygen desaturation and more aggressive use of supplemental oxygen. As a corollary, one possible reason for the disproportionate incidence of pulmonary hypertension in sarcoidosis was that these persons were receiving inadequate supplemental oxygen. This study also indicates that therapies for pulmonary hypertension in sarcoidosis are needed. Two small series suggest that vasodilators may be helpful in pulmonary hypertension associated with sarcoidosis, but studies examining the role for endothelin-receptor antagonists in sarcoidosis have not been performed.29 30 31 Further effort in this area is clearly warranted.
Why did our model explain only approximately 60% of the observed mortality? As stated above, factors after listing are important. Moreover, sarcoidosis has a variable natural history. IPF tends to be a relentless, progressive disease, while even some patients with end-stage sarcoidosis stabilize. Corticosteroids and more aggressive immunosuppression are thought to be more efficacious in sarcoidosis, while their utility in IPF remains less certain. Therapeutic alternatives exist for sarcoidosis that are not available for IPF and other advanced lung diseases. We lacked data on the actual dose of corticosteroids employed and concurrent use of other agents. As such, we could not address the possibility of untreated inflammation or how it might affect intermediate-term mortality in sarcoidosis. Readers should also note that prior efforts to develop predictive models for mortality in chronic lung disease have failed to validate their findings with actual outcomes.14 15 16
Irrespective of the limited predictive power of the model, it demonstrates that current recommendations regarding timing of referral for OLT merit re-evaluation. Because of the paucity of information about prognosis in advanced sarcoidosis, guidelines for the selection of candidates for OLT state that criteria for OLT listing in IPF might be employed for cases of sarcoidosis.13 As the current study highlights, one can now rely on other markers of pulmonary physiology and patient race to guide decisions about listing for OLT in sarcoidosis.
Our study has several limitations. Its retrospective nature raises the potential for certain types of bias. The UNOS database, however, consists of data recorded at time of listing for OLT as opposed to information abstracted in a post hoc fashion. Hence, recall bias is unlikely to be a significant concern. UNOS also relies on referring transplant centers for the quality of the data and lacked values for both the FEV1/FVC ratio and the diffusing capacity of the lung for carbon monoxide. We additionally lacked data on 5% of the eligible study cohort. This degree of lost data, however, is relatively small in light of the many centers that perform OLT and report data to the UNOS. Furthermore, we studied those actually listed for OLT. We did not have information on subjects believed to be ineligible for OLT; our results may not be generalizable to that population. The population of patients awaiting OLT is homogenous in that many with significant comorbidities are excluded. Moreover, as discussed earlier, the UNOS database does not contain data on health insurance or socioeconomic status. Before final listing, though, many institutions strive to guarantee payment coverage for most pre- and post-OLT care. In some sense then, patients awaiting OLT are similar to those with end-stage renal disease, in that they may face less difficulty with access to care because of a lack of health insurance.
In sum, patients with sarcoidosis listed for OLT face significant mortality. Many variables previously shown to predict outcome in interstitial lung disease, generally, and in sarcoidosis, specifically, fail to correlate with risk for death in sarcoidosis. Specific guidelines to aid clinicians in determining time of referral for OLT in advanced sarcoidosis need to be developed. Reliance on recommendations extrapolated from data for other types of interstitial lung disease may not be applicable in sarcoidosis.
|
Footnotes |
|---|
The opinions expressed herein are not be construed as official or as reflecting the policy of either the Department of Defense or the Department of the Army.
Received for publication June 24, 2002. Accepted for publication February 26, 2003.
|
References |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  K. A. Fisher, D. M. Serlin, K. C. Wilson, R. E. Walter, J. S. Berman, and H. W. Farber Sarcoidosis-Associated Pulmonary Hypertension: Outcome With Long-term Epoprostenol Treatment. Chest, November 1, 2006; 130(5): 1481 - 1488. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. Handa, S. Nagai, S. Miki, Y. Fushimi, K. Ohta, M. Mishima, and T. Izumi Incidence of Pulmonary Hypertension and Its Clinical Relevance in Patients With Sarcoidosis Chest, May 1, 2006; 129(5): 1246 - 1252. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. H. Leuchte, R. A. Baumgartner, M. E. Nounou, M. Vogeser, C. Neurohr, M. Trautnitz, and J. Behr Brain Natriuretic Peptide Is a Prognostic Parameter in Chronic Lung Disease Am. J. Respir. Crit. Care Med., April 1, 2006; 173(7): 744 - 750. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. J. Lettieri, S. D. Nathan, S. D. Barnett, S. Ahmad, and A. F. Shorr Prevalence and Outcomes of Pulmonary Arterial Hypertension in Advanced Idiopathic Pulmonary Fibrosis Chest, March 1, 2006; 129(3): 746 - 752. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H Nunes, M Humbert, F Capron, M Brauner, O Sitbon, J-P Battesti, G Simonneau, and D Valeyre Pulmonary hypertension associated with sarcoidosis: mechanisms, haemodynamics and prognosis Thorax, January 1, 2006; 61(1): 68 - 74. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. F. Shorr, D. L. Helman, D. B. Davies, and S. D. Nathan Pulmonary hypertension in advanced sarcoidosis: epidemiology and clinical characteristics Eur. Respir. J., May 1, 2005; 25(5): 783 - 788. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. D. Nathan Lung Transplantation: Disease-Specific Considerations for Referral Chest, March 1, 2005; 127(3): 1006 - 1016. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. H. Leuchte, C. Neurohr, R. Baumgartner, M. Holzapfel, W. Giehrl, M. Vogeser, and J. Behr Brain Natriuretic Peptide and Exercise Capacity in Lung Fibrosis and Pulmonary Hypertension Am. J. Respir. Crit. Care Med., August 15, 2004; 170(4): 360 - 365. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
