© 2006 American Thoracic Society
From the Authors:We agree with Sweet and Faro that further work is required before making a sweeping recommendation that children with cystic fibrosis (CF) be excluded from lung transplantation. Our recent publication continues efforts to better understand the survival effect of lung transplantation using existing data (1). The tentative verb "may suggest" was intended to stimulate a rethinking of patient selection for lung transplantation. The collegial letter from Sweet and Faro suggests our success. Our study finds no survival benefit for even the most severely ill pediatric patients with CF, in stark contrast with severely ill adult patients. This is not proof that pediatric lung transplantation should be abandoned. However, we argue that the burden of proof has shifted; we should demonstrate that there are pediatric patients with CF who indeed benefit. Sweet and Faro raise two major issues. First, whether any models should be used as the basis for recommending treatment, and second, whether any patients should forgo the potential benefits of transplantation. No model can accurately predict individual mortality. Absence of clinically important variables such as PCO2 may reduce the power of models to predict group survival, but will not produce a systematic bias. However, our model (2) usefully separates patients extremely likely to die from those nearly certain to live. For example, only 20% of patients in our group 1 (predicted 5-year survivorship less than 30%) survive 5 years, while over 98% of group 5 (predicted survivorship greater than 90%) survive. The model was as predictive and statistically valid for adult and pediatric patients in 1998 (unpublished data) as it was in 1993 (2). For the sickest patients, we show a dramatic transplant benefit for adults but no benefit for children (1). For the healthiest patients, lung transplantation is overwhelmingly likely to reduce survival (2–3). These predictions are based, however, on the untestable assumption that listed patients do not differ systematically from retrospectively identified unlisted controls. Other studies have used wait-list patients as controls, and consistently show a benefit of transplantation. These analyses use a statistical model with independent competing risks, an assumption that cannot be tested statistically. The study by Aurora and coworkers (4) is based on the British system, where patients are continually reallocated. This creates dependent competing risks, which make survivorship impossible to analyze (5). In the United States, the former allocation system included reallocation only when patients removed themselves from the list, possibly limiting the magnitude of the dependence. Thus both methods are based on untestable assumptions that could create biases. Does this mean that both methods are useless for patient selection? More specifically, should we ignore our results about CF and lung transplantation? We think not. Despite potential biases, our models corroborate clinical experience and demonstrate that healthy patients of any age should avoid transplantation. For other patients, such as children, however, the decision is not as clear. These problems may suggest that it is time for a randomized controlled trial of lung transplantation in CF. Only a well-designed prospective study may resolve the potential biases that hinder retrospective methods. Such a study would need careful design and patient selection. It is here that our survivorship models could prove most useful, by objectively selecting patients (such as children) for whom survival benefit is most uncertain, and by identifying appropriate controls. We hope to stimulate the CF and transplant communities to discuss whether such a study, fraught with enormous logistical challenges, is worthwhile. In the absence of such a study, decisions must still be made, and models provide the only evidence-based guidance. The data reported here have been supplied by UNOS as the contractor for the Organ Procurement and Transplantation Network (OPTN). The interpretation and reporting of these data are the responsibility of the author(s) and in no way should be seen as an official policy of or interpretation by OPTN or the U.S. Government.
University of Utah, Salt Lake City, Utah
University of California, Los Angeles, California FOOTNOTES Conflict of Interest Statement: None of the authors have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. REFERENCES
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
