External Validation of the Kidney Donor Profile Index (KDPI) to Predict Long-Term Graft Survival in Kidney Transplant Recipients.

Abstract

Background: The Kidney Donor Profile Index (KDPI), developed in the United States, is widely used to predict graft failure after kidney transplantation. However, external validation is crucial to assess its performance in other populations. This study aimed to validate KDPI in Brazilian kidney transplant recipients.

Methods: We conducted a single-center, retrospective cohort study involving 1,807 recipients of deceased-donor kidney transplants performed between January 2013 and December 2015, with a 5-year follow-up. The model's discrimination, overall fitness, and calibration were assessed in predicting graft failure.

Results: Recipients were 49.2 years old, 60.8% were male, on hemodialysis for 40.1 months. The median KDRI and KDPI were 1.2 and 69%, respectively. The AU-ROC for graft survival was highest in the KDPI >85% category [0.576 (95%CI: 0.521-0.631)], followed by the >35-85% [0.504 (95%CI: 0.460-0.549)] and the 0-35% [0.488 (95%CI: 0.408-0.568)] categories. Discrimination improved in the Cox model incorporating recipient variables and KDPI categories [0.801 (95%CI: 0.775-0.827)]. The KDPI as a continuous variable had a C-index of 0.561 (95%CI: 0.533-0.588), similar to donor age alone [0.560 (95%CI: 0.534-0.587)]. The overall fitness assessed by the Brier score was 0.151, and the calibration plot showed a good performance for KDPI. However, it was similar to what was observed using the donor age as a predictor isolated (0.152).

Conclusions: In this transportability external validation study, KDPI performed acceptably as a prediction tool for 5-year graft survival after kidney transplantation. These findings guide organ allocation and support clinical decision-making regarding organ acceptance.