A Novel Genomic Approach for Acute Rejection Across All Age Groups in Kidney Transplantation

Summary

Despite the increase in renal allograft survival rate, acute rejection (AR) remains a common and serious post-transplantation complication (1). The failure to optimize immunosuppression and individual patient nonadherence have been suggested as two factors contribute to AR (2;3).  Over 15% of protocol biopsies done within the first year had histological AR without evidence of kidney dysfunction (4-6). Early detection and treatment of AR prior to the occurrence of irreversible structural lesions would increase graft survival.

Common histological features, such as interstitial fibrosis (IF) and tubular atrophy (TA), correlate with graft dysfunction and risk of progression to failure in transplants (7). However, the simple quantification of IF/TA is insufficient to identify those at greatest risk for long-term graft loss, as at one year post transplant nearly uniformly show IF/TA, reflecting the burden of injury including donor death, organ harvest, and the transplantation process (8;9). Recent studies have proposed that monitoring of renal graft status through peripheral blood (PB) rather than invasive biopsy (kidney tissue) will lessen the risk of infection and other stresses, while reducing the costs of rejection diagnosis (10-13).

Using a large collection of gene expression data obtained from multiple public databases, we developed a peripheral blood gene expression signature of acute rejection that could be used in both pediatric and adult recipients. This peripheral signature of acute rejection will allow for improved monitoring of allografts in pediatric patients as they progress to adulthood allowing for immunosuppression titration and improvement in long-term outcomes.

Question / Proposal

Acute Rejection remains a significant complication in kidney transplantation that affects long-term outcomes. Transcriptional profiling of transplant biopsies has provided useful insights into allograft injury mechanisms such as acute rejection (AR) and chronic rejection (CAN), and offered criteria to refine the histology-based Banff classification (7), which have been used to classify and predict allograft rejection (14-17). However, previously published reports have mainly focused on selected sub-populations of patients, in which, the genome-scale gene expression analysis was performed using relatively small numbers of samples. To capture the heterogeneity allograft rejection, a large population analysis is required. Moreover, there is very limited study on genomic profiles of pediatric kidney transplant, and few prospective study exist to demonstrate if pediatric AR is predictable using genomic features identified from adult samples.

We compiled transplant renal biopsy (n=1091) and peripheral blood cell (n=392) gene expression profiles obtained from 12 independent public databases. We revealed distinct AR-related genomic profiles between adult and pediatric renal samples, suggesting predictor of allograft rejection should consider the genomic stratification between age groups. We developed a peripheral blood gene expression signature of acute rejection that could be used in both pediatric and adult recipients. This peripheral signature of acute rejection will allow for improved monitoring of allografts in pediatric patients as they progress to adulthood allowing for immunosuppression titration and improvement in long-term outcomes. This study serves as a novel model of using public datasets to rapidly screen and validate biomarkers.

 

Research

Although non-invasive profiling of PB provides an opportunity to serially monitor the dynamic process of AR and manage immunosuppression, the transcriptional alterations in response to inflammatory infiltrate resident cells within the kidney showed less diagnostic value related to AR in peripheral blood (18-20). This might also because that previous studies used only renal, the most significant transcriptional alteration in renal samples might be tissue specific, in which the signal reduced to noise ratio in a site remote from the allograft. Similarly, gene signature developed using only peripheral blood samples might not be concordant with the signal revealed in renal samples. In addition, previously published reports have mainly focused on selected sub-populations of patients, in which, the genome-scale gene expression analysis was performed using relatively small numbers of samples. Moreover, there is very limited study on genomic profiles of pediatric kidney transplant, and few prospective study exist to demonstrate if pediatric AR is predictable using genomic features identified from adult samples. More specific and objective diagnostic tools that provide mechanistic information on the underlying pathologic events are needed.

In this study, we first utilized a bioinformatic approach to compile a large collection of gene expression data obtained from multiple public databases. We observed substantial gene expression differences between adult and pediatric cases with acute rejection. After removing genes differentially expressed in children and adults, we developed a peripheral blood gene expression signature of acute rejection that could be used in both pediatric and adult recipients. We further validated this novel acute rejection signature using patient (age 1-78) samples from our institutional biorepository with 1 year follow up. Our finding provides an opportunity to develop strategy for non-invasive measurement of renal allograft rejection for both adult and pediatric samples.

Our study provides an opportunity to delineate the subset of patients who have higher risk of acute transplant rejection that likely will require more aggressive treatment. This peripheral signature of acute rejection will allow for improved monitoring of allografts in pediatric patients as they progress to adulthood allowing for immunosuppression titration and improvement in long-term outcomes. This study will also serve as a novel model of using public datasets to rapidly screen and validate biomarkers.

Method / Testing and Redesign

To capture the heterogeneity renal allograft rejection, we compiled a large collection of either renal or blood gene expression data obtained from public dataset.  A total of 1091 renal gene expression profiles were collected from 7 independent NCBI Gene Expression Omnibus (GEO) datasets: GSE21374, GSE22459, GSE36059, GSE50058, GSE7392, GSE9493, and GSE25902 (pediatric). In addition, we combined a collection of 392 gene expression profiles of blood cells derived from 5 GEO datasets: GSE14346, GSE15296, GSE24223, GSE46474, and GSE20300 (pediatric). As part of the normalization step, VOOM/LIMMA framework was applied on the raw counts. This normalization method transforms the data from a negative binomial to an approximated normal distribution, which will allow us to use such methods as surrogate variable analysis (SVA) to remove any obvious batch effects in the data. As demonstrated in Figure 1, the mixed colors in normalized PCA plots indicated the batch effects among the individual dataset were successfully removed.

To define the genomic differences between age groups, we compared expression profiles between adult and pediatric samples, and identified 25043 probe sets whose expressions in TCMR or CAN samples were significantly different between adult and pediatric samples (P<0.001, MWU test). After removed these 25043 age group related probe sets, the genomic stratification of age groups was successfully minimized Figure 3. As demonstrated in Figure 2, we split our combined dataset into training and validation sets. To develop a gene signature that is significantly associated with AR in both renal and blood sample sets, we conducted a series genome scale expression profile analyses. The Mann-Whitney U Test (MWU) and Cox-regression AR survival analysis were used to determine the significance of predictive score between AR and STA (stable transplantation). For the comparison, all signatures were regenerated using same method, the Shotgun Stochastic Search in Regression software (SSS, developed by Duke University). The receiver operating characteristic curve (ROC curve) was used to access the diagnostic ability of a binary classifier system.

Furthermore, we used patient samples from our institutional Biorepository (42 Rejection, 52 Non-event) age 1-78, and 38(40%) female/56(60%) male to confirm previously genes identified n=76 probes via RT-PCR. This study was carried out under the Immune Development in Pediatric Transplantation (Duke IRB Protocol #00082649) and Immune Monitoring and Assay Development in Organ Transplant Recipients (Duke IRB Protocol #00058027).

Results

We compiled transplant renal biopsy (n=1091) and peripheral blood cell (n=392) gene expression profiles obtained from 12 independent public databases. We removed any obvious batch effects in the data using surrogate variable analysis (Fig. 1).

As demonstrated in Figure 2, we split our combined dataset into training and validation sets.

To define the genomic differences between age groups, we compared expression profiles between adult and pediatric samples, and identified 25043 probe sets whose expressions in TCMR or CAN samples were significantly different between adult and pediatric samples (P<0.001, MWU test) (Fig. 3).

After removing genes differentially expressed in children and adults, we developed a gene signature of acute rejection from renal and peripheral blood cells using adult samples and validated this signature in an independent pediatric dataset (Fig. 4A and 4B). We further validated this novel acute rejection signature using patient (age 1-78) samples from our institutional biorepository with 1 year follow up (Fig. 4C).

 

Conclusion

Using a large collection of gene expression data obtained from multiple public databases, we developed a peripheral blood gene expression signature of acute rejection that could be used in both pediatric and adult recipients. This peripheral signature of acute rejection will allow for improved monitoring of allografts in pediatric patients as they progress to adulthood allowing for immunosuppression titration and improvement in long-term outcomes.

Acute Rejection remains a significant complication in kidney transplantation that affects long-term outcomes. Recent studies have proposed that monitoring of renal graft status through peripheral blood (PB) rather than invasive biopsy will lessen the risk of infection and other stresses, while reducing the costs of rejection diagnosis (10-13). Although non-invasive profiling of PB provides an opportunity to serially monitor the dynamic process of AR and manage immunosuppression, the transcriptional alterations in response to inflammatory infiltrate resident cells within the kidney showed less diagnostic value related to AR in peripheral blood (18-20). This might also because that previous studies used only renal, the most significant transcriptional alteration in renal samples might be tissue specific, in which the signal reduced to noise ratio in a site remote from the allograft. Similarly, gene signature developed using only peripheral blood samples might not be concordant with the signal revealed in renal samples. Although this work has been extended in subsequent studies, each remains an analysis of a limited number of samples that might be insufficient to capture the heterogeneity allograft rejection, in which a large population analysis is required.

Using a large collection of gene expression data obtained from multiple public databases, we observed substantial gene expression differences between adult and pediatric cases with acute rejection. We revealed an age-independent gene set that was associated with acute rejection in both renal and peripheral blood cells and validated in our institutional cohort.  We identified a gene signature that was significantly associated with risk of AR in both of renal and blood sample, which provides an opportunity to develop strategy for non-invasive measurement of renal allograft rejection. Our study provides an opportunity to delineate the subset of patients who have higher risk of acute transplant rejection that likely will require more aggressive treatment. This study serves as a novel model of using public datasets to rapidly screen and validate biomarkers.

About me

My name is Daniel Cheng and I am a junior at the North Carolina School of Science and Mathematics (NCSSM), a two-year residential school for juniors and seniors funded by our state government. I enjoy math, playing the piano and doing scientific research.

My interest in STEM stemmed from my curiosity of the world.  I always wanted to know how things worked and how to solve various problems in society, like monitoring, curing, and preventing people's diseases.  During my summer break, I worked as high school student volunteer at Department of Pediatric, Duke University.  I involved in several research projects at Duke, two of which have been accepted to present at American Transplant Congress (ATC) 2019 in June 1-5 2019, at Boston, MA.

I really admire Marie Curie for her bravery, pursuing her passion in chemistry even though her research was her health into jeopardy.  If it weren't for her work, doctors would not have been able to screen patients' internal injuries with x-rays which would save billions of lives.  Her perseverance reminds me to work hard, no matter what the odds are against me. 

For my future college and career plans, I want to pursue a path in medicine and winning any prize in the Google Science Fair will mean a ton to me because it would show that my hard work on my project paid off and was appreciated.  The scholarship would help pay for my college tuition and help me further my study in science.

Health & Safety

The Institutional Review Board (IRB)

This study was carried out under the Immune Development in Pediatric Transplantation (Duke IRB Protocol #00082649) and Immune Monitoring and Assay Development in Organ Transplant Recipients (Duke IRB Protocol #00058027).

Potential Benefits of the Proposed Research to Human Subjects and Others

There are no direct benefits to the research subjects. The potential benefits are the importance of the knowledge to be gained, in terms of better understanding of evolutionary origins and diversification of neuroendocrine castration-resistant prostate cancer, which will provides a clinical guide to personalized treatment, in doing so reassuring and monitoring those patients who do not need extensive treatment, and acting rapidly and appropriately for those that do.

Data and Safety Monitoring Plan

To complete the clinical correlation data for the specimens in our tissue bank, a REDCap database was developed and housed on a secure server that is password-protected and located in the Duke Department of Pedantic and Surgery. The clinical annotation data was gathered from Epic electronic medical records and/or from the Duke Enterprise Data Unified Content Explorer (DEDUCE) database (after appropriate institutional review board approval). Data was entered into web based, secure case report forms, which was used to generate a patient registry with demographic, clinical, and additional pathologic data for all specimens analyzed. All information connected to individuals was recorded using a unique code. The personal identifiers and links to the unique codes are all stored on secure networks maintained by the Sister Study. Access to this database was limited to key personnel listed on the protocol.

Bibliography, references, and acknowledgements

Reference List

      (1)    Meier-Kriesche HU, Schold JD, Kaplan B. Long-term renal allograft survival: have we made significant progress or is it time to rethink our analytic and therapeutic strategies? Am J Transplant 2004 Aug;4(8):1289-95.

      (2)    Lerut E, Kuypers DR, Verbeken E, Cleutjens J, Vlaminck H, Vanrenterghem Y, et al. Acute rejection in non-compliant renal allograft recipients: a distinct morphology. Clin Transplant 2007 May;21(3):344-51.

      (3)    Morrissey PE, Reinert S, Yango A, Gautam A, Monaco A, Gohh R. Factors contributing to acute rejection in renal transplantation: the role of noncompliance. Transplant Proc 2005 Jun;37(5):2044-7.

      (4)    Rush D, Somorjai R, Deslauriers R, Shaw A, Jeffery J, Nickerson P. Subclinical rejection--a potential surrogate marker for chronic rejection--may be diagnosed by protocol biopsy or urine spectroscopy. Ann Transplant 2000;5(2):44-9.

      (5)    Moreso F, Ibernon M, Goma M, Carrera M, Fulladosa X, Hueso M, et al. Subclinical rejection associated with chronic allograft nephropathy in protocol biopsies as a risk factor for late graft loss. Am J Transplant 2006 Apr;6(4):747-52.

      (6)    Nankivell BJ. Subclinical renal allograft rejection and protocol biopsies: quo vadis? Nat Clin Pract Nephrol 2008 Mar;4(3):134-5.

      (7)    Racusen LC, Solez K, Colvin RB, Bonsib SM, Castro MC, Cavallo T, et al. The Banff 97 working classification of renal allograft pathology. Kidney Int 1999 Feb;55(2):713-23.

      (8)    Nankivell BJ, Borrows RJ, Fung CL, O'Connell PJ, Allen RD, Chapman JR. Natural history, risk factors, and impact of subclinical rejection in kidney transplantation. Transplantation 2004 Jul 27;78(2):242-9.

      (9)    Cosio FG, Grande JP, Larson TS, Gloor JM, Velosa JA, Textor SC, et al. Kidney allograft fibrosis and atrophy early after living donor transplantation. Am J Transplant 2005 May;5(5):1130-6.

    (10)    Li L, Khatri P, Sigdel TK, Tran T, Ying L, Vitalone MJ, et al. A peripheral blood diagnostic test for acute rejection in renal transplantation. Am J Transplant 2012 Oct;12(10):2710-8.

    (11)    Kurian SM, Williams AN, Gelbart T, Campbell D, Mondala TS, Head SR, et al. Molecular classifiers for acute kidney transplant rejection in peripheral blood by whole genome gene expression profiling. Am J Transplant 2014 May;14(5):1164-72.

    (12)    Grigoryev YA, Kurian SM, Avnur Z, Borie D, Deng J, Campbell D, et al. Deconvoluting post-transplant immunity: cell subset-specific mapping reveals pathways for activation and expansion of memory T, monocytes and B cells. PLoS ONE 2010 Oct 14;5(10):e13358.

    (13)    Gunther OP, Shin H, Ng RT, McMaster WR, McManus BM, Keown PA, et al. Novel multivariate methods for integration of genomics and proteomics data: applications in a kidney transplant rejection study. OMICS 2014 Nov;18(11):682-95.

    (14)    Einecke G, Reeve J, Sis B, Mengel M, Hidalgo L, Famulski KS, et al. A molecular classifier for predicting future graft loss in late kidney transplant biopsies. J Clin Invest 2010 Jun;120(6):1862-72.

    (15)    Reeve J, Sellares J, Mengel M, Sis B, Skene A, Hidalgo L, et al. Molecular diagnosis of T cell-mediated rejection in human kidney transplant biopsies. Am J Transplant 2013 Mar;13(3):645-55.

    (16)    Khatri P, Roedder S, Kimura N, De VK, Morgan AA, Gong Y, et al. A common rejection module (CRM) for acute rejection across multiple organs identifies novel therapeutics for organ transplantation. J Exp Med 2013 Oct 21;210(11):2205-21.

    (17)    Saint-Mezard P, Berthier CC, Zhang H, Hertig A, Kaiser S, Schumacher M, et al. Analysis of independent microarray datasets of renal biopsies identifies a robust transcript signature of acute allograft rejection. Transpl Int 2009 Mar;22(3):293-302.

    (18)    Deng MC, Eisen HJ, Mehra MR, Billingham M, Marboe CC, Berry G, et al. Noninvasive discrimination of rejection in cardiac allograft recipients using gene expression profiling. Am J Transplant 2006 Jan;6(1):150-60.

    (19)    Flechner SM, Kurian SM, Head SR, Sharp SM, Whisenant TC, Zhang J, et al. Kidney transplant rejection and tissue injury by gene profiling of biopsies and peripheral blood lymphocytes. Am J Transplant 2004 Sep;4(9):1475-89.

    (20)    Li L, Ying L, Naesens M, Xiao W, Sigdel T, Hsieh S, et al. Interference of globin genes with biomarker discovery for allograft rejection in peripheral blood samples. Physiol Genomics 2008 Jan 17;32(2):190-7.

 

Acknowledgements

                This study is supported by the Translating Duke Health Controlling the Immune System Initiative (TDHI) from Duke University Medical center. I thank the Duke Department of Pediatric for providing me an opportunity to participate in research projects as a high school student volunteer. I sincerely appreciate my supervisor, Dr. Eileen Chambers, M.D. for her guidance and help on my research projects.

                Dr. Chamber’s lab has all equipment for experiments in the areas of biochemistry, molecular biology and cell biology. Adjacent to Dr. Chamber’s lab is a tissue culture room, cold room and shared large equipment such as ultracentrifuges.

                Bioinformatic analyses were conducted using PC workstations and Unix workstation, all of which are connected to collaborators and the laboratory through secure Duke servers. Computing/network infrastructure is provided to the Lab by the Cancer Center Information Systems, including dedicated network data backups.