Development and Validation of a Multidimensional Score to Predict Long-term Kidney Transplant Outcomes

Overview

To further develop personalized medicine in kidney transplantation and improve transplant patient outcomes, attention has been given to define early surrogate endpoints that might aid therapeutic interventions, clinical trials and clinical decision-making. Despite a clear pressing need, no population-scale prognostication system exists that will combine traditional factors and biomarker candidates to represent the complete spectrum of risk predicting parameters. To adequately predict transplant patients' individual risks of allograft loss, this would require a complex integration of data, including: donor data, recipient characteristics, transplant characteristics, allograft precision phenotypes, ethnicity, immunosuppressive regimen monitoring, allograft infections, acute kidney injuries, and recipient immune profiles. This project aims: 1. To develop a generalizable, transportable, mechanistically and data driven composite surrogate end point in kidney transplantation; 2. To validate several risk scores to predict kidney allograft survival and response to treatment of individual patients; Eventually, it will provide an easily accessible tool to calculate individual patients' risk profiles after kidney transplantation, by using datasets from prospective cohorts and post hoc analysis of randomized control trial datasets.

Background The field of kidney transplantation currently lacks robust models to predict long-term allograft failure, which represents a major unmet need in clinical care and clinical trials. This study aims to generate and validate an accessible scoring system that predicts individual patients' risk of long-term kidney allograft failure. Main Outcome(s) and Measure(s) A score based on classical statistical approaches to model determinants of allograft and patient survival (Cox model, multinomial regression). These models will be further completed with statistical approaches derived from artificial intelligence and machine learning.

Conditions