Background Improved predictive models are needed in lung transplantation in the setting of a proposed allocation system that incorporates longer-term post-transplant survival in the United States. Allocation systems require accurate mortality predictions to justly allocate organs. Methods Utilizing the United Network for Organ Sharing database (2005-2017), we fit models to predict 1-year mortality based on the Lung Allocation Score (LAS), the Chan, et al, 2019 model, a novel “clinician” model (a priori clinician selection of pre-transplant covariates), and two machine learning models (Least Absolute Shrinkage and Selection Operator; LASSO and Random Forests) for predicting 1-year and 3-year post-transplant mortality. We compared predictive accuracy among models. We evaluated the calibration of models by comparing average predicted probability vs observed outcome per decile. We repeated analyses fit for 3-year mortality, disease category, including donor covariates, and LAS era. Results The area under the cure for all models was low, ranging from 0.55 to 0.62. All exhibited reasonable negative predictive values (0.87-0.90), but the positive predictive value for was poor (all <0.25). Evaluating LAS calibration found 1-year post-transplant estimates consistently overestimated risk of mortality, with greater differences in higher deciles. LASSO, Random Forests, and clinician models showed no improvement when evaluated by disease category or with the addition of donor covariates and performed worse for 3-year outcomes. Conclusions The LAS overestimated patients’ risk of post-transplant death, thus underestimating transplant benefit in the sickest candidates. Novel models based on pre-transplant recipient covariates failed to improve prediction. There should be wariness in post-transplant survival predictions from available models. Improved predictive models are needed in lung transplantation in the setting of a proposed allocation system that incorporates longer-term post-transplant survival in the United States. Allocation systems require accurate mortality predictions to justly allocate organs. Utilizing the United Network for Organ Sharing database (2005-2017), we fit models to predict 1-year mortality based on the Lung Allocation Score (LAS), the Chan, et al, 2019 model, a novel “clinician” model (a priori clinician selection of pre-transplant covariates), and two machine learning models (Least Absolute Shrinkage and Selection Operator; LASSO and Random Forests) for predicting 1-year and 3-year post-transplant mortality. We compared predictive accuracy among models. We evaluated the calibration of models by comparing average predicted probability vs observed outcome per decile. We repeated analyses fit for 3-year mortality, disease category, including donor covariates, and LAS era. The area under the cure for all models was low, ranging from 0.55 to 0.62. All exhibited reasonable negative predictive values (0.87-0.90), but the positive predictive value for was poor (all <0.25). Evaluating LAS calibration found 1-year post-transplant estimates consistently overestimated risk of mortality, with greater differences in higher deciles. LASSO, Random Forests, and clinician models showed no improvement when evaluated by disease category or with the addition of donor covariates and performed worse for 3-year outcomes. The LAS overestimated patients’ risk of post-transplant death, thus underestimating transplant benefit in the sickest candidates. Novel models based on pre-transplant recipient covariates failed to improve prediction. There should be wariness in post-transplant survival predictions from available models. 

The lung allocation score and other available models lack predictive accuracy for post-lung transplant survival.

High Lung Allocation Scores and Lung Transplantation: Is There a Ceiling?

While the field of lung transplantation has evolved considerably since the first lung transplant with long-term success was performed in 1983,1 the need for donated lungs to help patients with progressive end-stage lung diseases that do not respond to non-transplant therapies continues to outstrip the availability of lungs from organ donors.2 Many patients with certain types of lung disease or other characteristics such as short stature or ABO blood type O may be less likely to receive an organ offer than those with different profiles and are, therefore, more at risk to die while waitlisted and awaiting lung transplantation.3 4 Organ allocation systems need to evolve (along with improvements in technical and medical aspects of lung transplantation) to optimally balance fairness and efficacy in donor lung allocation while also attempting to minimise the risk of death without transplant for waitlisted patients, optimise post-transplant survival and quality of life and avoid unsustainable resource utilisation.

In this issue of the journal, Kourliouros et al 5 examined outcomes emanating from the donor lung allocation system used in the UK by examining National Health Service UK Transplant Registry data for a patient cohort (n=2213) comprised of all adult patients (≥16 years old) registered for lung transplantation between January 2004 and March 2014. Donor lung allocation to listed patients during this period of time was at the discretion of the specific transplant centre (there are five designated centres in the UK for adult lung transplantation) at which candidates were listed for transplantation, and decisions for which patient to transplant with available organs were based on clinical assessments that would often take into account both waiting times and the degree of acuity for a specific patient. Statistical analyses of the available data showed that disease-specific and key patient-specific factors had a significant impact on length of time from …

https://thorax.bmj.com/content/thoraxjnl/early/2018/11/16/thoraxjnl-2018-212634.full.pdf

Balancing fairness and efficacy in lung allocation for transplantation: unfinished business

This chapter reviews the history, development, and institution of the lung allocation score (LAS) in the United States Lung transplantation can be a lifesaving therapy for patients suffering with end-stage lung diseases for which there are no remaining therapeutic options available Unfortunately, there remains an imbalance between the number of patients in need of this therapy and the availability of suitable donor organs for transplantation The LAS was developed in the spirit of the important ethical principles in transplantation of utility and justice More specifically, the objective of the LAS is to decrease waitlist time (and thus risk for waitlist mortality) and to allocate organs based upon medical urgency so as to balance predicted urgency with predicted benefit The institution of the LAS for the allocation of suitable donor lungs to appropriate candidates on May 4, 2005, literally changed the landscape of lung transplantation overnight This chapter also explores the ongoing evolution of the LAS as advancements continue to be made in medical therapies and technologies available for the support and management of end-stage lung diseases

M. Patricia George

Papers

Prognostic Markers and the LAS for Lung Transplantation: Impact of New Revisions for Successful Outcome