Utilizing Donor Recipient Characteristics To Improve Kidney Allograft Allocation

Findings from a recent study aim to lower the rising rates of donor discards and improve the US kidney allograft allocation system through a post-transplant survival prediction algorithm, or Kidney Allograft Survival Index, to build upon the Kidney Donor Risk Index (KDRI) currently in place.1

The study, which analyzed data from the Organ Procurement and Transplantation Network between March 2016 and December 2021, shows improved allograft allocation through a recipient characteristic-focused model to improve transplant outcomes.

Despite KDRI’s modest predictive accuracy for allograft survival and model calibration problems (C-statistic ≈ 0.61), there is an association between worse KDRI scores and organ discard. In 2016, 3629 of the 18132 (20%) kidneys procured for organ donation were discarded. In 2023, 8182 of 32553 (25%) kidneys were discarded, proving the trend of rising rates and the need for intervention to improve the odds of an allograft for the 90000 individuals waiting for a kidney transplant.1,2

“In the scenario of kidneys from deceased donors <45 years, the KDRI will often predict survival that is 15 – 25% higher or lower than observed survival. Notably, these calibration problems were only apparent through the inspection of graphs that showed observed versus expected survival,” wrote David S. Goldberg, MD MSCE, Division of Digestive Health and Liver Diseases at the University of Miami, and other colleagues, “Inaccurate prediction can cause clinicians to accept kidneys based on flawed projections of post-transplant survival and undermine the informed consent process when discussing post-transplant expectations with patients.”

The primary outcome measured was time to all-cause allograft failure over 3 years, defined as mortality, return to dialysis, or repeat kidney transplantation. The secondary outcome measured was delayed graft function, defined as dialysis in the first post-transplant week, and 1-year graft failure or eGFR ≤20 mL/min/1.73 m². Post-hoc outcomes were five-year all-cause allograft failure, defined as allograft failure by three years.

The retrospective cohort included 75,867 deceased donor kidneys ≥18 years of age. a mean age of 54 years (IQR 44, 64); 40% female; 34% black; and 93% receiving chronic dialysis in concurrence with transplantation. Only the first transplant was analyzed for patients who received >1 kidney transplant.

The study analyzed data from several models with progressively more detailed predictor values. Model A, which used standard KDRI variables, showed limited discrimination (iAUC ≈ 0.61–0.62) and poor calibration. Model F, which included donor, allograft, and recipient variables, performed the best overall with an iAUC of 0.678 for 3-year survival (95% CI 0.677 - 0.679) and an AUC = 0.75 for delayed graft function. Since any additional predictor variables past this point did not impact model performance, Model F was determined as the Kidney Allograft Survival Index.

The inclusion of recipient variables can improve the accuracy of allograft outcomes prediction. The addition of age alone increased the iAUC from 0.62 to 0.63. Contrary to prediction, adding donor-recipient interactions (CMV, HLA, age, sex, and cold ischemia time) did not improve discrimination, longitudinal donor laboratory data did not improve predictive accuracy, and machine learning did not improve recipient donor matching.

“It is possible that replacing the KDRI with a tool with better predictive accuracy such as the Kidney Allograft Survival Index might help clinicians to better identify appropriate kidneys that match their patients’ expectations for post-transplant survival,” investigators concluded. “While the overall rate of organ discard might not change, the Kidney Allograft Survival Index could help clinicians more accurately identify those organs at highest risk for allograft failure for specific recipients.“

1. Potluri VS, Rubin J, Zee J, et al. Assessing Deceased Donor Kidneys Through Post-Transplant Survival Prediction Algorithms. American Journal of Kidney Diseases. Published online October 1, 2025. https://doi.org/10.1053/j.ajkd.2025.07.016

2. Cooper M, Formica R, Friedewald J, et al. Report of National Kidney Foundation Consensus Conference to Decrease Kidney Discards. Clinical Transplantation. 2018;33(1):e13419. https://doi.org/10.1111/ctr.13419