Measurement of a 5-gene panel in whole blood in kidney transplant recipients with acute rejection and stable controls

Introduction

Further improvements are required to monitor rejection of kidney allografts post-transplant. Current methodology to detect rejection by kidney biopsy is invasive and presents a risk to the patient, whilst an inadequate biopsy sample may prevent an accurate diagnosis. The patient risk associated with biopsy deters some transplant centers from performing protocol biopsies, instead using clinical signs of allograft damage to inform a decision to perform an indication biopsy. Significant immune mediated damage to the allograft may occur before rejection is clinically visible. Therefore new biomarkers that can be repeatedly measured by minimally invasive sampling that identifies damaging alloreactivity prior to significant pathology and assesses the whole organ are required¹.

A 5-gene panel (DUSP1; NKTR; PBEF1; MAPK9; and PSEN1) in whole blood has been shown to distinguish stable allografts from those with acute rejection (AR) in paediatric kidney transplant recipients (KTRs) with mixed ethnicity. Expression of each gene was significantly different between kidneys with AR vs stable function in a separate paediatric validation population (of which 60% had received a deceased-donor allograft)². Interestingly, expression of these genes appeared to identify borderline rejection as AR and may therefore predict AR before it has become clinically evident.

The same panel was recently validated in a Korean adult KTR population, although only PSEN1 and MAPK9 were significantly different between AR and control groups³. The lack of differential expression between stable and AR groups for genes DUSP1, NKTR and PBEF1 might demonstrate a variability in gene expression between KTR populations; however these three genes did contribute to the ability of the 5-gene panel to distinguish AR from non-AR using multivariate logistic regression analysis.

The expression data presented in our study suggests that these 5 genes, measured in whole blood of KTRs, do not distinguish between stable function and AR in Caucasian KTRs in the UK. Our data, provided with complete clinical and raw gene expression data, will be available to be incorporated into larger studies in the future, providing an important resource.

Methods

7 KTRs with clinical evidence of AR confirmed by allograft biopsy (5 AR, 2 borderline changes suspicious for AR) and 5 control KTRs with stable graft function closely matched for Human Leukocyte Antigen mismatch, race, initial graft function, induction and maintenance therapy were studied (Supplementary material Table 1; REC number 07/H0603/26, following informed consent). KTRs were adult Caucasians of which 20% had received a deceased-donor allograft. All patients were negative for donor specific antibodies pre-transplantation and therefore did not receive desensitization. There was no difference in panel reactive antibodies between control and AR groups. Blood samples were collected during AR episodes occurring on days 4, 6, 6, 41, 48, 210 and 393 in individual KTRs post transplant.

cDNA was produced from whole blood RNA collected in BD tempus tubes and isolated using the Tempus Spin RNA Isolation kit (BD). Expression of a 5-gene panel comprising DUSP1, PBEF1, PSEN1, MAPK9, NKTR (normalized with HPRT expression; Supplementary material Table 2) was accessed by qPCR using TaqMan Gene Expression Assays (Life Technologies) and measured using a Stratagene Mx3000P qPCR machine (Agilent Technologies). Kits were used according to manufacturers instructions.

Each AR episode was treated successfully with three i.v. daily doses of 500-mg methylprednisolone, as measured by restored eGFR to baseline. Two patients with AR additionally received 30mg i.v alemtuzumab. All patients with AR episodes were re-bled at 1 year post-transplantation, and maintained stable graft function and eGFR, except 1 KTR with AR at day 393 and a second patient that declined follow-up at 1 year.

Univariate logistic regression analysis was performed using IBM SPSS version 22. All other statistics were performed in GraphPad Prism version 5.0c.

Results

Univariate logistic regression analysis was performed on 2^-ΔΔCt values calculated using either the pre-operative time point ΔCt of each KTR⁴ (Figure 1; DUSP1: p=0.772; MAPK9: p=0.733; PBEF1: p=0.525; NKTR: p=0.698; PSEN1: p=0.935) or an average of all stable KTR ΔCt values as a calibrator for each gene. With both methods of 2^-ΔΔCt calculation, univariate logistic regression analysis showed that no single gene was a significant predictor of AR. The size of our cohort was insufficient to allow for multivariate analysis. Expression of each gene showed no significant difference when AR and one year samples (DUSP1: p=0.44; MAPK9: p=0.44; PBEF1: p=0.44; NKTR: p=0.31; PSEN1: p=0.44) or AR and stable KTR (DUSP1: p=0.88; MAPK9: p=0.76; PBEF1: p=0.76; NKTR: p=0.52; PSEN1: p=1.0) samples were compared using Wilcoxon matched-pairs signed rank test and Mann-Whitney test respectively.

Figure 1. ΔCt values were calibrated using each patient’s pre-operative sample ΔCt value to produce 2^-ΔΔCt values, shown here and depicting the mean and SEM.

Conclusion

This study indicates that the expression of this 5-gene panel is unable to distinguish AR from stable allograft function in this adult Caucasian population, which might be explained by differences in expression of these genes between different populations. Importantly, the raw data provided in supplemental tables here are available to supplement future studies involving larger cohorts of patients.