Development and validation of primary graft dysfunction predictive algorithm for lung transplant candidates.

Area of Science:

• Transplantation Science
• Medical Informatics
• Predictive Analytics

Background:

• Primary graft dysfunction (PGD) is the primary cause of early complications and death following lung transplantation.
• Accurate PGD risk prediction is crucial for optimizing donor selection and perioperative management.
• A generalizable and clinically applicable PGD prediction model is needed to support transplant decision-making.

Purpose of the Study:

• To develop a clinically useful and generalizable prediction model for primary graft dysfunction (PGD) risk.
• To create a user interface for real-time PGD risk assessment in lung transplantation.
• To evaluate the clinical utility and net benefit of the PGD prediction model.

Main Methods:

• A prospective cohort study (2012-2018) was used to derive the PGD prediction model.
• Regularized (lasso) logistic regression was employed to identify significant PGD predictors.
• External validation was performed at a single center, with decision curve analysis assessing model utility.

Main Results:

• The PGD prediction model incorporates donor and recipient factors, including distance, age, lung capacity, LAS, BMI, pulmonary artery pressure, sex, transplant indication, donor characteristics, and interactions.
• A user interface enables real-time PGD risk assessment for donor-recipient pairs.
• The model demonstrated decision-making net benefit across specific PGD risk ranges in derivation and validation cohorts.

Conclusions:

• A clinically valuable PGD predictive algorithm was developed to aid transplant decision-making.
• The model supports post-transplant care planning and can enrich patient cohorts for PGD treatment trials.
• This tool enhances the management of lung transplant recipients by providing accurate PGD risk stratification.