Immunologic Risk Prediction Model for Kidney Graft Function

Clinicians lack appropriate non-invasive methods to be able to predict, diagnose, and reduce the risk of rejection in the years following kidney transplantation. Protocol biopsies and monitoring of serum creatinine levels are the most common methods of monitoring graft function after transplant; however, they have several negative aspects. Use of traditional factors regarding donors and recipients such as Human Leukocyte Antigen (HLA) DNA typing, pre-transplant anti-HLA antibody levels, and basic demographics (age, ethnicity/race, gender), has proved inadequate for post-transplant graft monitoring past the first few years. We propose that by utilizing immunologic factors available to clinicians across the United States, development of a non-invasive model for predicting renal graft outcome will provide a useful tool for post-transplant patient monitoring. We advocate an expanded model which incorporates both the traditional factors, as well as new factors, which have shown promise in predicting kidney outcome and are widely available for testing using commercial kits. These additional factors include major histocompatibility complex class I chain-related gene A (MICA) typing of donor and recipient, degree of matching for killer cell immunoglobulin-like receptors (KIRs) between donor and recipient, detection of MICA antibodies, and soluble CD30 level (sCD30). This proposed graft-function prediction model is the first to include all of these factors.

Using multi-center data from adult recipients of standard-criteria deceased-donor (SCD) kidneys, we were able to construct models, containing the traditional factors only, for prediction of outcome at 1 year and 3 years post-transplant. Using single-center data from adult recipients of standard-criteria deceased-donor kidneys, we developed comparison models containing traditional factors only, as well as, expanded models containing the new suggested variables for prediction of outcome post-transplant. These additional variables, when incorporated into the expanded models provided greater positive predictive values, greater negative predictive values, and lower false negative rates for graft outcome at 1 year and at 3 years post-transplant than the models utilizing traditional factors only. Our results indicate that evaluation of sCD30, MICA and KIR as part of routine protocol testing, is helpful to clinicians for predicting risk of kidney graft rejection.