Pre-transplant TCR Network Topology Predicts Kidney Allograft Rejection Independent of HLA Mismatch

Abstract

Despite extensive pretransplant serological screening and HLA matching, 10-15% of kidney allografts experience acute rejection within the first year. Currently, risk stratification for transplantation relies primarily on antibody reactivity to HLA molecules, with no assessment of the T cell compartment before or after transplantation. In our previously established longitudinal cohort of 54 patients, T cell receptor {beta} (TCR{beta}) sequencing was performed on paired kidney biopsy and peripheral blood samples. Here, we further analyzed the data to construct a comprehensive set of sequence-similarity networks and quantify over 30 network metrics. After adjusting for repertoire size, graft status was the strongest signal for the underlying differences in network metrics. Individuals who rejected the kidney graft generally exhibited more fragmented and less connected networks at baseline, with fewer interconnect T cell clones and more isolated sequences. Notably, pre-transplant peripheral blood mononuclear cell (PBMC) network topology alone predicted non-stable outcomes with an area under the curve (AUC) of 0.81, sensitivity of 76%, and specificity of 76%. The performance of this prediction model was independent of HLA mismatch, while changes in network topology at three months post-transplantation further improved prediction to an AUC of 0.88 (permutation p = 0.009). Collectively, TCR sequencing and network analysis represent a potential novel, non-invasive approach for pre-transplant risk stratification and immune monitoring, capturing functional immunological risk that may not be accessible through HLA genotyping or serology.