A Peripheral Blood Gene Expression Signature for Subclinical Acute Rejection and Associated Long-Term Graft Injury.

Abstract

Currently, the diagnosis of acute rejection requires renal allograft biopsy triggered by an elevation of creatinine. Data on the impact of subclinical rejection and borderline changes on graft outcomes are conflicting, and diagnosis is impacted by subjective reporting. In our prospective genomics of chronic allograft rejection (GoCAR) cohort, we examined the impact of these diagnoses on 3-mth protocol biopsy on outcomes in 199 patients on whom concurrent peripheral blood transcriptome analysis was also performed. Twenty-eight of 199 patients had subclinical acute or borderline rejection (SCR) diagnosed by the central histology core which was blinded to clinical data. Allografts with SCR had greater renal function decline by delta eGFR-creatinine [p<0.05], and sustained significantly higher risks of subsequent acute rejection [OR - 3.4; CI - 1.25 to 9.14 & OR=7.8; CI - 1.47 to 40.89 at 1- and 2-years]. In 36-mths of follow-up, allografts with acute rejection or with borderline changes alone had significantly reduced death-censored, graft survival (Log-rank p = 0.01). Notably, 60% of borderline rejections were not diagnosed on locally reported pathology and were untreated. Transcriptome analysis was performed on 3-mths blood for the 199 recipients (RNAseq and microarray). Using RNAseq data of 66 patients with 3-mth renal biopsies as training set, we identified a 9-gene signature that accurately diagnosed SCR (AUC =0.96) with 80% cross-validation accuracy from 100 iterations. The geneset was technically confirmed using an independent cohort of GOCAR patients by microarray (n=39; AUC=0.835) and a publicly available dataset GSE 14346 ( n=75; AUC=0.865). At 1- and 2-years, allografts identified by this signature had a persistent inflammatory phenotype [i-scores - 0.9±1.0 vs 0.4±0.7 & 1.0±1.1 vs 0.3±0.7; p <0.01] with greater histological damage than allografts without this signature [CADI-scores - 3.46±2.78 vs 1.81± 2.19 & 3.80±1.62 vs 1.53±2.31; p <0.001]. In conclusion, we have identified a 9-gene peripheral blood signature that accurately diagnoses SCR and identifies allografts at risk for subsequent histological, functional decline and graft loss.