Abstract Background and Aims BK virus-associated nephropathy (BKvN) is an important complication occurring after kidney transplantation. Its treatment mainly relies on lowering immune suppression. BKvN’s histological aspect can mimic T cell-mediated rejection (TCMR) and other inflammatory conditions featuring tubulitis and interstitial infiltrate. At present, the screening strategy consists of measuring urinary and blood viral loads of BKv. Upon rising BK viremia, BKvN diagnosis is established by a graft biopsy. Biopsies are invasive and are not suitable for repeated screening. Therefore, the aim of this case-control study was to establish a urinary proteome-based test for BKvN diagnosis. Method Urine was obtained from 700 allograft recipients prior to either protocol or „for cause“ biopsies during the first year of post-transplantation surveillance. Aside from normal biopsy findings (N=294), the histological diagnoses included BKvN (N=50), IFTA II-III (N=145), glomerulonephritis (N=17) and T cell-mediated, antibody-mediated, mixed or borderline rejection episodes (N=194). From all patients, relevant clinical and demographic data including Banff classification scores, HLA mismatches and the presence of donor-specific antibodies was collected at the date of sample collection and in further clinical follow-up. BKvN cases were defined as having SV40-positive immunostaining on allograft biopsy. BKvN patients were considered as cases and all other recipients were considered as controls. Patient’s urinary peptide profiles were generated using capillary electrophoresis coupled to mass spectrometry consisting of 9430 different peptides in the mass range of 0.8 to 20 kDa. Results The CE-MS peptide profiles of randomly selected 30 BKvN cases and 307 kidney allograft controls (one sample per patient) were statistically analyzed to identify the most discriminative peptide markers for BKvN. In total, there were 117 peptides detected to have significantly different urinary excretion rates between both patient groups after false discovery rate adjustment for multiple testing. Thirty-two of these peptides were selected by cross-validated variable selection to establish a Support Vector Machine-based multi-marker model. Applying the 32-peptide marker model to an independent validation cohort consisting of 20 BKvN cases and 343 controls resulted in an area under the receiver operating characteristic curve of 0.86 and a 95% confidence interval from 0.82 to 0.89 with the p-value below 0.0001 and sensitivities and specificities for BKvN diagnosis of 85 and 76 %, respectively. Most notably, distribution of the classification values (figure 1) in the different patient groups of the validation set indicated specificities of the peptide marker model for TCMR/borderline of 85% (17 of 20 true negative classifications) and for viruria/viremia of 73% (8 of 11 true negative classifications) Conclusion In conclusion, the established urinary peptide marker model might serve as a non-invasive diagnostic test for BKvN and its differentiation from TCMR and states of isolated viruria or viremia.
Read full abstract