# External validation of a prognostic model predicting renal graft function one year after brain-dead donor kidney transplantation

## TL;DR

This study validates a model predicting kidney function one year after transplant, showing it can reliably identify poor outcomes for better donor-recipient matching.

## Contribution

The model was externally validated and recalibrated for broader clinical applicability.

## Key findings

- The model predicted eGFR categories G4 and G5 with an AUC > 0.700 after recalibration.
- Recalibration was necessary for full validation across all eGFR categories.
- The model helps avoid poor donor-recipient pairings by identifying marginal graft function.

## Abstract

A German transplant center recently published a prognostic model predicting graft function one year after deceased donor kidney transplantation (KTx) relying on pre-transplant variables. The aim of this study is to externally validate this model.

We retrospectively analyzed clinical data from deceased donor KTx recipients undergoing transplantation between January 2007 and December 2023 at University Hospital Rheinisch-Westfälische Technische Hochschule (RWTH) Aachen. Receiver operating characteristics (ROC) curves were analyzed to validate the prognostic model based on donor age, donor serum creatinine, recipient body mass index, re-transplantation > 2nd KTx, and cold ischemia time. Glomerular filtration rates were categorized using the Kidney Disease: Improving Global Outcomes (KDIGO) categories G1 – G5.

A total of 494 kidney transplantations were performed at our institution, 350 (70.9%) thereof from donation after brain death. The median one-year estimated glomerular filtration rate (eGFR) was 42 [12–94] mL/min/1.73 m2. A validation for all eGFR categories was only possible with recalibration of the constant and coefficients of the original model, whereas without recalibration it could only be validated for KDIGO G2 and G4. Unfavourable recipient/donor-pairings with eGFR categories G4 and G5 one year after KTx could be predicted with an area under the ROC curve (AUC) > 0.700 in the validation and the original study cohort.

We successfully validated the prognostic model for prediction of eGFR categories G4 and G5, which is of high clinical importance to identify outcomes with marginal graft function one year after KTx, thereby facilitating the avoidance of futile recipient/donor-pairings during allocation.

The online version contains supplementary material available at 10.1007/s00423-025-03962-8.