Composite Lung Transplant Suitability Score (CLaSS): A Novel Predictor of Survival Following Lung Transplantation

Purpose Lung transplantation is indicated in patients with respiratory failure who progress despite optimal therapy. The selection of suitable transplant recipients considers internationally accepted absolute and relative contra-indications that are known to negatively influence survival, and typically involves a multi-disciplinary assessment. We assessed the impact on routinely collected pre-transplant variables on post-transplant survival and created a composite model that predicted post-transplant survival. Methods Adult patients (n = 916) undergoing lung transplant assessment at a single center between 2013 and 2021 had the following variables analyzed: age, gender, pre-transplant diagnosis, BMI, eGFR, HbA1c, albumin, 6-minute walk distance (6MWD) and Stanford Integrated Psychosocial Assessment for Transplant (SIPAT) score. Logistic regression was performed for post-transplant survival. Multiple imputation was used for missing data. Variables for inclusion in the final model were selected using backward elimination and confirmed using forward elimination (p-value <0.1). Internal validation of the model was performed using-one-out cross validation and 70/30 data split. Results Variables as assessed by univariate analysis predicting worse survival included pre-transplant diagnosis (CLAD), limited renal reserve and reduced 6MWD, with CLAD and 6MWD remaining significant predictors of death in the multivariate analysis. All variables were initially included in the predictive model, however the model performance improved with the selective inclusion of pre-transplant diagnosis, 6MWD and HbA1c (ROC 0.672, sensitivity 98.7%). Case examples demonstrated the practical utility of the model. Scenario #1. CF, HbA1c 5, Albumin 40 and 6MWD 400m; predicted probability of survival is 85%. Scenario #2. CLAD, HbA1c 5, albumin 20, 6MWD 250m; predicted probability of survival is 36%. Conclusion Using variables that are commonly collected during a pre-transplant assessment, we have created a clinically applicable model that predicts post-transplant survival. The model can aid in recipient selection and inform discussion with potential transplant candidates on expected post-transplant outcomes, including survival. Lung transplantation is indicated in patients with respiratory failure who progress despite optimal therapy. The selection of suitable transplant recipients considers internationally accepted absolute and relative contra-indications that are known to negatively influence survival, and typically involves a multi-disciplinary assessment. We assessed the impact on routinely collected pre-transplant variables on post-transplant survival and created a composite model that predicted post-transplant survival. Adult patients (n = 916) undergoing lung transplant assessment at a single center between 2013 and 2021 had the following variables analyzed: age, gender, pre-transplant diagnosis, BMI, eGFR, HbA1c, albumin, 6-minute walk distance (6MWD) and Stanford Integrated Psychosocial Assessment for Transplant (SIPAT) score. Logistic regression was performed for post-transplant survival. Multiple imputation was used for missing data. Variables for inclusion in the final model were selected using backward elimination and confirmed using forward elimination (p-value <0.1). Internal validation of the model was performed using-one-out cross validation and 70/30 data split. Variables as assessed by univariate analysis predicting worse survival included pre-transplant diagnosis (CLAD), limited renal reserve and reduced 6MWD, with CLAD and 6MWD remaining significant predictors of death in the multivariate analysis. All variables were initially included in the predictive model, however the model performance improved with the selective inclusion of pre-transplant diagnosis, 6MWD and HbA1c (ROC 0.672, sensitivity 98.7%). Case examples demonstrated the practical utility of the model. Scenario #1. CF, HbA1c 5, Albumin 40 and 6MWD 400m; predicted probability of survival is 85%. Scenario #2. CLAD, HbA1c 5, albumin 20, 6MWD 250m; predicted probability of survival is 36%. Using variables that are commonly collected during a pre-transplant assessment, we have created a clinically applicable model that predicts post-transplant survival. The model can aid in recipient selection and inform discussion with potential transplant candidates on expected post-transplant outcomes, including survival.