*Purpose: Monitoring kidney transplant rejection by repeated biopsies might result in increased procedural complications and cost. Exosomes are nanometer-sized vesicles released by cells carrying trans membrane, nucleic acids and intracellular proteins that mirror the parental cells function and may carry parent cell’s mRNA of the transplanted kidney and originate from the immune cells during active rejection.

*Methods: We collected 190 urine samples from adult renal transplant recipients at the time of clinically indicated kidney transplant biopsies. We measured mRNA directly from urinary exosomes to identify a specific exosome nucleic signature for allograft rejection. Exosomes were isolated from the urine supernatant using Exosome Diagnostics’ EXOPRO Urine Clinical Sample Concentrator Kit. A panel of 586 than of 112 TaqMan assays (TaqMan® OpenArray® Human Inflammation) were used to analyze the urine samples. We used the Boruta to identify 16 important relevant genes of the training data set by performing feature selection. In order to enhance the prediction accuracy and interpretability in generating the rejection probabilities, a LASSO logistic regression model was developed and allowed us to identify an 8-gene signature that distinguished biopsy with any cause rejection from no rejection in the training and validation cohorts. Furthermore, using the same approach, we have identified a 13 genes signature that is associated with T cell-mediated rejection (TCMR) that included borderline cellular rejection.

*Results: The 8 genes signature helped in differentiating biopsies with any-cause rejection from no-rejection, with ROC-AUC of 0.851 (95% CI 0.768 to 0.934) for the training set and 0.756 (95% CI 0.645-0.867) in the validation set. Negative predictive values (NPV) for active rejection with this 8 genes signature were 94.44% in the training set and 95.65% in validation set, respectively. The AUC for T cell-mediated rejection (TCMR), using a 13 genes signature, was 0.851 (95% CI 0.768 to 0.934) and 0.756 (95% CI 0.645 to 0.867) for the training and validation data set, repsectively . The NPV of TCMR using a defined cutoff of the 13 genes signature were 96.61% in the training set and 94.12% in the validation set.

*Conclusions: Urinary exosomes mRNA represents a robust and non-invasive way to detect active kidney transplant rejection. An 8 genes signature was able to predict any cause active rejection. Whereas, a13 gens signature was associated with cell-mediated rejection. These diagnostic signatures showed good NPV and sensitivity, and provide a helpful tool to rule out unnecessary biopsies.

« Back to 2020 American Transplant Congress