Correspondence
Lung Transplantation and Survival in Children with Cystic Fibrosis
N Engl J Med 2008; 358:1753-1755April 17, 2008
- Article
To the Editor:
The analysis by Liou et al. (Nov. 22 issue)1 is a step toward a better understanding of the risk–benefit ratio of lung transplantation in cystic fibrosis. The study, however, has several limitations.
First, the authors have neither validated their model in an independent data set nor provided an indication of its discriminative ability. Unless the concordance statistic2 for the model was consistent with good discrimination of patients who survived from those who died, the study conclusions might not be applicable to a significant proportion of patients. Second, the model does not include any donor factors or surrogates for center experience, both of which were previously shown to be associated with a poor outcome.3,4
Third, the authors have handled missing covariate values by excluding patients with missing values from the analysis, an approach known to result in biased estimates.5 Furthermore, their article does not state whether patients who were excluded — a not insignificant proportion of the cohort (15%) — had characteristics and, above all, mortality that were similar to those in the group of patients who were not excluded. Given the above limitations, we believe caution should be exercised in interpreting the results of this study.
5 ReferencesMuhammad F. Dawwas, M.R.C.P.
Christopher J. Watson, M.D.
Alexander E. Gimson, F.R.C.P.
Addenbrooke's Hospital, Cambridge CB2 2QQ, United Kingdom
drdawwas@gmail.com 1
Liou TG ,Adler FR ,Cox DR ,Cahill BC . Lung transplantation and survival in children with cystic fibrosis. N Engl J Med 2007;357:2143-2152
Full Text | Web of Science | Medline2
Harrell FE Jr. Regression modeling strategies: with applications to linear models, logistic regression, and survival analysis. New York: Springer-Verlag, 2001.
3
Orens JB ,Boehler A ,de Perrot M , et al. A review of lung transplant donor acceptability criteria. J Heart Lung Transplant 2003;22:1183-1200
CrossRef | Web of Science | Medline4
Garrity ER ,Moore J ,Mulligan MS ,Shearon TH ,Zucker MJ ,Murray S . Heart and lung transplantation in the United States, 1996-2005. Am J Transplant 2007;7:1390-1403
CrossRef | Web of Science | Medline5
Ambler G ,Omar RZ ,Royston P . A comparison of imputation techniques for handling missing predictor values in a risk model with a binary outcome. Stat Methods Med Res 2007;16:277-298
CrossRef | Web of Science | Medline
To the Editor:
Liou et al. conclude that most children with cystic fibrosis who were selected for lung transplantation during the period from 1992 through 2002 did not have a significant survival benefit from transplantation.
Until 2005, the allocation of lungs for transplantation was based on patients' accrued time on the waiting list, which sometimes exceeded 2 years. Therefore, many clinicians referred patients for placement on the waiting list long before the need for transplantation was imminent. Evidence for early placement of children on the waiting list includes the high 5-year predicted survival rate among patients at the time of listing, reported as 57%. In addition, some patients had improvement while on the waiting list.1 If patients were offered lungs when their clinical condition did not warrant transplantation, they could temporarily inactivate their listing status, while their accrued time was maintained.
The result of this allocation policy was that healthier patients waited longer for transplantation, which, as Liou et al. acknowledge, biases their analysis against transplantation. Thus, the analysis by Liou et al. does not warrant the conclusion that lung transplantation for children with cystic fibrosis is unlikely to improve survival.
1 ReferencesRan D. Anbar, M.D.
State University of New York Upstate Medical University, Syracuse, NY 13210
anbarr@upstate.edu 1
Allen J ,Visner G . Lung transplantation in cystic fibrosis -- primum non nocere? N Engl J Med 2007;357:2186-2188
Full Text | Web of Science | Medline
To the Editor:
The assessment by Liou and colleagues of the survival benefit in U.S. children undergoing lung transplantation for cystic fibrosis appears to be statistically sound. However, covariate data were collected up to 2 years before patients were placed on the waiting list, and there was a median wait of 15 months for transplantation. We question the predictive value of data obtained 3 years or more before transplantation. The assumption that healthier patients did not wait longer from listing to transplantation is inconsistent with historical practice. The observed 5-year cohort survival is more than 15% lower than expected. Each issue introduces bias against transplantation.
In 2005, the Organ Procurement and Transplantation Network (OPTN) adopted a new allocation policy for lung transplantation, which was intended to maximize the survival benefit and minimize the number of deaths of patients on the waiting list.1 With the use of OPTN models for waiting-list mortality and post-transplantation survival and patient data obtained at the time of transplantation, a survival benefit at 3 years is predicted for 69% of adolescents with cystic fibrosis who underwent lung transplantation after the new allocation policy went into effect (Edwards L: personal communication). Therefore, continuing to perform lung transplantation will not “harm” children with cystic fibrosis.
1 ReferencesStuart C. Sweet, M.D., Ph.D.
Washington University in St. Louis, St. Louis, MO 63110
sweet@kids.wustl. edu Christian Benden, M.D.
University Hospital Zurich, CH-8091 Zurich, SwitzerlandOkan Elidemir, M.D.
Baylor College of Medicine, Houston, TX 770301
Egan TM ,Murray S ,Bustami RT , et al. Development of the new lung allocation system in the United States. Am J Transplant 2006;6:1212-1227
CrossRef | Web of Science | Medline
Author/Editor Response
Readers question our use of data obtained up to 2 years before patients were placed on the waiting list and our exclusion of transplantation-center experience and the 15% of patients with missing data. The median time from clinical measurements to placement on the waiting list was 183 days and was uncorrelated with the outcome. To avoid selection effects associated with the length of time since listing, our analyses implicitly compared patients who had undergone transplantation with patients who had been on the waiting list for about the same time and who had not received a transplant. We excluded patients with missing data instead of imputing values or attempting a review of original charts. Of the 88 patients who were excluded, 70 had missing clinical data during the 2 years before placement on the waiting list. Excluded patients were slightly younger but otherwise indistinguishable from included patients for modeled covariates. They were more likely to undergo transplantation, but had no difference in mortality after listing or transplantation. A preliminary analysis of pediatric lung transplantation showed reduced harm but no benefit at large centers.
In reply to Dawwas et al., we used all patients in our model development to maximize statistical power. We used bootstrapping to rebuild the model with randomly resampled populations and found the reported model to be robust. The high significance and magnitude of the effects establish discriminatory power. This is our second study with the use of a different patient population and markedly different methods to show a lack of survival benefit associated with lung transplantation in children with cystic fibrosis.1 To assist clinicians in listing patients for transplantation, we excluded factors from analysis that were unknowable at the time of decision making; these factors include donor characteristics and the future experience of transplantation centers.
Anbar and Sweet et al. suggest that healthier patients waited longer than sick patients for transplantation, perhaps biasing the results against transplantation. However, waiting times relative to health have not been carefully studied. Transplantation may also be deferred in patients who are deemed to be too sick to undergo the surgery, resulting in a different bias. Preliminary examination of center-specific data suggests that healthier patients at large centers had shorter times to wait before transplantation than sicker patients at small centers.
Sweet et al. report that the current lung-allocation score predicts that 69% of patients benefit from transplantation. This assertion of realized benefit constitutes a hypothesis that cannot be tested retrospectively. Under the lung-allocation scoring system, patient selection is based on the estimated survival benefit, confounding an analysis of the survival benefit of transplantation.
The question of the benefit of transplantation for children with cystic fibrosis cannot be resolved by retrospective study because of unobserved biases. The best solution is to positively answer our call for a prospective, randomized trial of lung transplantation.
1 ReferencesTheodore G. Liou, M.D.
Frederick R. Adler, Ph.D.
University of Utah, Salt Lake City, UT 84132
ted.liou@utah. edu David R. Cox, Ph.D.
Nuffield College, Oxford OX1 1NF, United Kingdom1
Liou TG ,Adler FR ,Huang D . Use of lung transplantation survival models to refine patient selection in cystic fibrosis. Am J Respir Crit Care Med 2005;171:1053-1059
CrossRef | Web of Science | Medline
- Citing Articles (1)
Citing Articles
1
Markus Hofer, Christian Benden, Ilhan Inci, Christoph Schmid, Sarosh Irani, Rudolf Speich, Walter Weder, Annette Boehler. (2009) True Survival Benefit of Lung Transplantation for Cystic Fibrosis Patients: The Zurich Experience. The Journal of Heart and Lung Transplantation 28:4, 334-339
CrossRef
