Development and Validation of a Comprehensive Classification Automation System for Kidney Allograft Biopsies

Overview

Since 1991, the Banff classification has been the gold standard for defining antibody-mediated rejection (AMR) and T-cell mediated rejection (TCMR), thereby guiding the treatment and management of transplant recipients. Starting from a pure histological approach, the classification has moved over the past three decades towards an integrated precision diagnosis system, which encompasses other expertise, such as immunology, immunogenetic, other basic sciences, biostatistics, data science, and artificial intelligence The counterpart of this constant refinement is that Banff rules are becoming complex to follow, with numerous possible scenarios leading to a high degree of inter-observer variability and misclassifications, which may lead to therapeutic consequences. The aims of this study are: 1. To integrate and decode all Banff rules and develop a computer-based application - the Banff Automation System - which provides automated and reproducible diagnoses 2. To validate the ability of the Banff Automation System to reclassify rejection diagnoses in multicenter cohort studies and clinical trials.

Despite considerable advances in the development of effective immunosuppressive therapies, allograft rejection remains the main cause of graft loss after kidney transplantation. Since 1991, the Banff classification has been the gold standard for defining antibody-mediated rejection (AMR) and T-cell mediated rejection (TCMR), thereby guiding the treatment and management of transplant recipients. Starting from a pure histological approach, the classification has moved over the past three decades towards an integrated precision diagnosis system, which encompasses other expertise, such as immunology, immunogenetic, other basic sciences, biostatistics, data science, and artificial intelligence The counterpart of this constant refinement is that Banff rules are becoming complex to follow, with numerous possible scenarios leading to a high degree of inter-observer variability and misclassifications, which may lead to therapeutic consequences. Hence, international transplant societies and regulatory agencies urgently appealed for a more comprehensible and reproducible classification, required for decision-making process and reliable surrogate endpoints, to further improve patients care and drug development. The aims of this study are: 1. To integrate and decode all Banff rules and develop a computer-based application - the Banff Automation System - which provides automated and reproducible diagnoses 2. To validate the ability of the Banff Automation System to reclassify rejection diagnoses in multicenter cohort studies and clinical trials. Based on the results, the investigators will provide a comprehensive, user-friendly, and open-access online application which might improve reproducibility and precision of biopsies' diagnoses, thereby reducing misclassifications, and path the way to standardize histology-based endpoints in observational studies and clinical trials, and post-transplant diagnostic and therapeutic's management of kidney transplant recipients.