Wei San Loh, Courtney M Rentas, Seth A Waits, Michael J Englesbe, Kyle H Sheetz
Our findings demonstrate that 90-d graft failure is influenced less by donor attributes alone and more by the dynamic interplay between donor, recipient, and procedural factors. This study further calls into question the ongoing relevance of donor-based risk models in contemporary liver transplantation.
INTRODUCTION: The Liver Donor Risk Index (LDRI) has long served as the principal tool for quantifying donor risk in liver transplantation, integrating donor and graft characteristics to estimate graft failure. However, shifts in donor demographics and the widespread adoption of machine perfusion technologies may have reduced its relevance to contemporary practice. We therefore evaluated the contemporary predictive performance of the LDRI for 90-d graft survival.
METHODS: Using the UNOS registry, we examined 49,220 liver transplant recipients across three time periods, replicating the methodology from the original LDRI study to assess whether donor characteristics from the modern transplant era (2018-2024) retained predictive value for 90-d graft survival.
RESULTS: In modern transplant eras, graft survival differences across LDRI categories were minimal, and multiple modeling strategies using over 250 donor features failed to identify strong predictors of 90-d graft failure, with uniformly limited accuracy and generalizability (AUC = 0.58).
CONCLUSIONS: Our findings demonstrate that 90-d graft failure is influenced less by donor attributes alone and more by the dynamic interplay between donor, recipient, and procedural factors. This study further calls into question the ongoing relevance of donor-based risk models in contemporary liver transplantation.