Molecular assessment of antibody-mediated rejection in human pancreas allograft biopsies

Pancreas transplant longevity is limited by immune rejection, which is diagnosed by graft biopsy using the Banff Classification. The histological criteria for antibody-mediated rejection (AMR) are poorly reproducible and inconsistently associated with outcome. We hypothesized that a 34-gene set associated with antibody-mediated rejection in other solid organ transplants could improve diagnosis in pancreas grafts. The AMR 34-gene set, comprising endothelial, natural killer cell and inflammatory genes, was quantified using the NanoString platform in 52 formalin-fixed, paraffin-embedded pancreas transplant biopsies from 41 patients: 15 with pure AMR or mixed rejection, 22 with T cell-mediated rejection/borderline and 15 without rejection. The AMR 34-gene set was significantly increased in pure AMR and mixed rejection (P = .001) vs no rejection. The gene set predicted histological AMR with an area under the receiver operating characteristic curve (ROC AUC) of 0.714 (P = .004). The AMR 34-gene set was the only biopsy feature significantly predictive of allograft failure in univariate analysis (P = .048). Adding gene expression to DSA and histology increased ROC AUC for the prediction of failure from 0.736 to 0.770, but this difference did not meet statistical significance. In conclusion, assessment of transcripts has the potential to improve diagnosis and outcome prediction in pancreas graft biopsies.