*Purpose: Current prediction systems of kidney-graft loss do not integrate the dynamic effect of parameters assessed over the time course of kidney recipients. Recent dynamic approaches could enhance risk stratification in kidney recipients by constructing a predictive system that could continuously be updated over time thereby improving patient care and treatment management.

*Methods: International population-based study involving 20 transplant centers and 6 randomized controlled trials (RCT). Patients were divided into a derivation cohort consisting of 4 French centers with patients transplanted between 2000 and 2014 and validation cohorts from 7 centers in Europe, 5 in the US, 4 in South-America, and 6 RCTs with patients transplanted between 2000 and 2016. Patients underwent assessment of clinical, functional, histological and immunological parameters, together with prospective, protocol-based estimated glomerular filtration rate (eGFR) and proteinuria repeated measures after transplantation. With the use of joint modelling, the iBox score previously published by our team (BMJ 2019) was combined to eGFR and proteinuria repeated measurements to derive a kidney-graft survival dynamic prediction system.

*Results: A total of 12,683 patients were included (3,774 patients in the derivation cohort and 8,909 patients in the validation cohorts). After a median follow-up of 7.42 years (IQR 5.21-10.07) post transplantation, 1,408 allograft failures occurred. A total of 416,510 eGFR and proteinuria repeated measures were assessed. With multivariable joint modeling analysis, the iBox score and the eGFR and proteinuria repeated measurements were independently associated with graft loss. Based on the final multivariable model, we derived a dynamic prediction model that demonstrated accurate calibration and very high discrimination in the derivation cohort (AUC= 0.857). The performance of the model was confirmed in the six validation cohorts from Europe (AUC= 0.833), the USA (AUC= 0.897), South-America (AUC= 0.891), and the RCTs (AUC= 0.922). We also validated the dynamic model in a large series of clinical scenarios and subpopulations.

*Conclusions: We developed for the first time an integrative dynamic system that accurately predicts the risk of long-term allograft failure and outperforms any current prediction models in kidney transplantation based on classical statistical approaches. This dynamic system shows generalisability across centers and countries worldwide. This original dynamic approach may help adjusting prognostic judgements of clinicians in everyday practice and improve the design of future clinical trials.

« Back to 2020 American Transplant Congress