*Purpose: Managing complications related to over-immunosuppression represents a growing challenge in post-transplant care, with infections accounting for the second leading cause of death with functioning graft (DWFG) in renal transplant recipients (RTRs) within the first year. At present, there are no clinically validated biomarkers to detect over-immunosuppression. Polyomavirus-associated nephropathy (PVAN) is an opportunistic infection specifically indicative of over-immunosuppression that occurs in 5-10% RTRs leading to graft dysfunction or loss. We leveraged metabolic profiling and machine learning to develop a metabolite-based signature to detect over-immunosuppression in RTRs prior to a PVAN event.

*Methods: Urine metabolites were extracted and analyzed via NMR spectroscopy from 71 RTRs collected 3 months post-transplant including N=39 with a biopsy-proven PVAN (labeled “over”) and N=32 with stable graft function, no histological signs of rejection or indications of over-immunosuppression for 2 years (labeled as “clean”). Differential metabolites between over and clean samples were used to generate cross-validated machine learning models.

*Results: Using a Kruskal-Wallis test for significance we identified 24 differential metabolite resonances between clean and over samples with a fold-change greater than 1.5. The differential resonances were matched by chemical shift to reference libraries for metabolite annotation. A total of 8 machine learning models were built from combinations of features including known metabolites, known and unknown metabolite features, and clinical data. The champion model led to a scoring system (BoR Score 0-1.0) with 3 zones, such that 82.35% of patients with a score higher than 0.6 developed PVAN while 89.47% of patients with a score lower than 0.4 had no adverse events related to under or over-immunosuppression for two years. Patients who scored between 0.4-0.6 (17/71) had indeterminant prediction power.

*Conclusions: The metabolite-based urine signature was able to classify over-immunosuppressed RTRs with PVAN from those with a stable graft with high accuracy. Validation studies in larger cohorts are underway, which could lead to a powerful new biomarker for post-transplant monitoring.

« Back to 2022 American Transplant Congress