FC031: Validation of a Prediction System for Risk of Allograft Loss (IBOX) in Pediatric Kidney Transplant Recipients

Abstract

Abstract BACKGROUND AND AIMS Kidney allograft loss is a common cause of end-stage renal disease but accurate prediction models of kidney allograft loss are lacking in children. The iBOX system has been broadly validated among adults. We aimed to validate the iBOX system in a large international cohort of pediatric kTx recipients. METHOD In this observational study, we used data from pediatric (<21) patients transplanted between 2005 and 2017 from 20 institutions in Europe and the USA. Patients with functional parameters (eGFR and UPCR), donor specific antibody and biopsy results (Banff scores g, ptc, cg, i, t and IFTA) were included. Individual predictions of allograft loss were obtained by applying the iBOX score on our data. The prediction performances of the model in our population were assessed via discrimination (c-statistics) and calibration. RESULTS A total of 573 kTx recipients were included. Median time from transplantation to evaluation was 1.0 (0.5–2.0) year with a mean age at evaluation at 12.1 (5.5) years and mean follow-up after transplantation 5.1 (2.8) years. Five-year death-censored graft survival from evaluation was 95%. At the time of evaluation, mean eGFR and uPCR were 65.5 (29.6) mL/min/1.73 m2 and 0.25 (1.2) g/g, respectively. A total of 118 (20.6%) of the patients had DSA. The iBOX system showed good discrimination with a c-statistic of 0.81 and good calibration (Figure 1). CONCLUSION The iBOX system demonstrated high accuracy in predicting kidney allograft loss in children with performances similar to those reported in adults.