Risk stratification of lung transplant candidates: implications for organ allocation.

Organ Allocation In response to perceived inequities of the existing timebased system, and under mandate of the federal government, a new lung allocation system was introduced in the United States in 2005. This system allocates organs on the basis of medical urgency (risk for death without transplantation) and “net transplant benefit” (the extent to which survival is increased with, as opposed to without, transplantation). It uses predictive models, incorporating more than a dozen variables, to generate predictions of 1-year survival with and without transplantation for a given patient (1). A raw lung allocation score (LAS) is calculated on the basis of these survival predictions and normalized to a scale of 0 to 100 for ease of use. Because 1-year survival without transplantation is factored into net transplant benefit and medical urgency measures, it affects the LAS more than posttransplantation survival, which is used only in the net transplant benefit calculation. As designed, the system preferentially allocates lungs to sicker patients while attempting to avoid situations in which outcomes are so poor that there would be no meaningful survival benefit. Since its implementation, the LAS system has had a profound and favorable effect on the dynamics of lung transplantation in the United States (2). Because there is no longer an incentive to place patients on the active wait list simply to accrue time, the number of actively listed patients has decreased to approximately one half of the pre-LAS level. Median waiting time, which ranged from 2 to 3 years under the previous allocation system, has decreased to fewer than 200 days, and one quarter of patients wait for fewer than 35 days. Of note, the mortality rate of patients on the wait list has substantially decreased, because those with advanced or rapidly progressive disease who were previously unable to tolerate a protracted wait now receive expedited transplantation. The LAS system is predicated on the notion that we can accurately predict natural history and posttransplantation survival for an array of advanced lung diseases, and that we can do so by using generic as opposed to diseasespecific variables. Questions have been raised about the ability of the LAS system to predict posttransplantation survival exclusively on the basis of pretransplantation variables (3) as well as its ability to predict wait list survival of patients with idiopathic pulmonary arterial hypertension without consideration of hemodynamic measurements reflecting right ventricular performance (4). The Thoracic Organ Transplantation Committee, which performed the Herculean task of developing the LAS system, recognized that developing predictive models is a dynamic job and mandated regular review of the LAS models to ensure that new factors found to predict outcomes are incorporated and current outcomes data are used. As a result of this review, the current PCO2 and change in PCO2 over time were recently added to the calculation of survival without transplantation. In this issue, Tsuang and colleagues (5) identify a new variable that independently predicts survival after lung transplantation: an increase in LAS in the 30 days before transplantation, which presumably reflects deteriorating clinical status. The authors used the Organ Procurement and Transplantation Network national database, restricting their analysis to the 5749 lung transplant recipients with complete data who received organs between May 2005 (the start date of the LAS system) and December 2010 and who were listed for at least 30 days before transplantation. Twelve percent of this cohort experienced an LAS change (LAS ) greater than 5 in the 30 days before receiving an organ. Compared with patients with an LAS 5, these patients had lower survival across the entire range of final LAS values assigned at the time of transplantation. A mixed-effect Cox proportional hazards model was used to adjust for numerous recipient, donor, and procedurerelated characteristics, including the final LAS at the time of transplantation. The authors found that an LAS 5 was independently associated with a significant increase in the hazard of posttransplantation death (hazard ratio, 1.31 [95% CI, 1.11 to 1.54]; P 0.001) compared with an LAS 5. Taken together, these findings show that an LAS 5 is associated with worse posttransplantation outcomes and that the final, higher LAS assigned at the time of transplantation does not explain this association. What are the implications of this study? The predictive variable identified, LAS , cannot simply be incorporated into LAS survivorship models as was done for PCO2. As defined, LAS can be calculated only with knowledge of the actual transplantation date and therefore cannot be used to establish a score for a candidate still on the wait list. What this study offers is a way to identify a highrisk transplant cohort with inferior posttransplantation survival. How the transplant community should respond to this depends largely on the survival metric that drives organ allocation. If the goal of transplantation is to maximize posttransplantation survival, then a logical, albeit controversial, response might be to avoid transplantation in patients with an acute decline in clinical status resulting in an LAS 5, at least until a specified period of subsequent stability has elapsed. However, the current LAS system embraces net survival benefit, rather than absolute posttransplantation survival, as a fundamental governing principle. Compelling evidence already shows that patients assigned the highest LAS values have inferior posttransplantation survival rates (6, 7). Yet these patients, who otherwise face Annals of Internal Medicine Editorial