Logo of pnasPNASInfo for AuthorsSubscriptionsAboutThis Article
Proc Natl Acad Sci U S A. Sep 25, 2007; 104(39): 15448–15453.
Published online Sep 14, 2007. doi:  10.1073/pnas.0705834104
PMCID: PMC2000539

Identification of a peripheral blood transcriptional biomarker panel associated with operational renal allograft tolerance


Long-term allograft survival generally requires lifelong immunosuppression (IS). Rarely, recipients display spontaneous “operational tolerance” with stable graft function in the absence of IS. The lack of biological markers of this phenomenon precludes identification of potentially tolerant patients in which IS could be tapered and hinders the development of new tolerance-inducing strategies. The objective of this study was to identify minimally invasive blood biomarkers for operational tolerance and use these biomarkers to determine the frequency of this state in immunosuppressed patients with stable graft function. Blood gene expression profiles from 75 renal-transplant patient cohorts (operational tolerance/acute and chronic rejection/stable graft function on IS) and 16 healthy individuals were analyzed. A subset of samples was used for microarray analysis where three-class comparison of the different groups of patients identified a “tolerant footprint” of 49 genes. These biomarkers were applied for prediction of operational tolerance by microarray and real-time PCR in independent test groups. Thirty-three of 49 genes correctly segregated tolerance and chronic rejection phenotypes with 99% and 86% specificity. The signature is shared with 1 of 12 and 5 of 10 stable patients on triple IS and low-dose steroid monotherapy, respectively. The gene signature suggests a pattern of reduced costimulatory signaling, immune quiescence, apoptosis, and memory T cell responses. This study identifies in the blood of kidney recipients a set of genes associated with operational tolerance that may have utility as a minimally invasive monitoring tool for guiding IS titration. Further validation of this tool for safe IS minimization in prospective clinical trials is warranted.

Keywords: kidney transplantation, microarray, tolerant, genomics, immunosuppression titration

Despite continuous improvement in renal allograft survival over the last decade, the half-life of renal allografts has increased marginally because of accrual of chronic graft nephropathy from drug-related nephrotoxicity and chronic rejection (1, 2). Patients facing life-long immunosuppression (IS) have increased risk of infection and malignancy (3), whereas insufficient immunosuppressive drug exposure or interruption usually increases rejection risk (4). However, spontaneous and long-term graft acceptance is observed in a small number of patients after solid-organ transplantation (5, 6), years after total withdrawal of immunosuppressive drugs, confirming that a clinical operational state of tolerance to a mismatched graft, described as “a state of quiescence of the transplanted organ, functioning without a destructive immune response” (7), can indeed occur and exist in humans. However, the frequency of this observation in the kidney transplant population is unknown and, currently, we cannot identify patients primed to develop this immune adaptation or monitor for the stability of this status of “operational tolerance.”

The operationally tolerant kidney transplant patients in this study have been previously described (8) and show an altered T cell-receptor repertoire (9) and distinct lymphocyte phenotypes (10). However, none of these observations have been validated as being functionally involved in operational tolerance in humans.

The identification of biomarkers of drug-free operational tolerance in graft recipients is an important and challenging issue allowing for individualized therapy and personalized safe IS minimization (11). In the present study, we studied a cohort of 17 highly informative kidney-transplanted patients presenting clinical operational tolerance years after complete IS withdrawal with the aim of identifying reliable minimally invasive biomarkers diagnostic of clinical operational tolerance in kidney transplantation.

Gene-expression patterns from peripheral blood samples, examined across 91 adults represented by normal adults and five cohorts of renal-transplant recipients in different clinical contexts, identified a minimal set of 49 genes, differentially expressed gene transcripts in drug-free tolerant patients when compared with other patients, with tolerance class prediction scores of >90%. Quantitative RT-PCR across a subset of 33 of these 49 genes can accurately confirm tolerance in an independent-validation group of tolerant patients with a specificity of 99%. Additionally, 1 of 12 stable patients on standard maintenance immunotherapy and 5 of 10 minimally immunosuppressed patients on steroid monotherapy have tolerance class-prediction scores of >80%.

Our data suggest that identification of natural operational tolerance in renal-transplant recipients is possible by gene expression pattern recognition across a modest number of genes in blood. Customized PCR-based gene expression of peripheral blood may target patients on conventional IS as candidates for safe IS minimization.


Biomarker Discovery and Biomarker Validation for a Tolerance Footprint.

Immunosuppressive-drug-free tolerant transplant recipients (TOL) were divided into a training group (n = 5) included in the study between 2000 and 2004 and an independent test group [n = 12; TOL test group (TOL-Test); TOL-Test patients TT1–TT6] included in the study between 2005 and 2006. Microarray analysis was performed on 24 training-group blood samples [5 TOL, 11 chronic rejection (CR), and 8 age-matched healthy volunteers (N)]. Two-class prediction tests using the predictive analysis of microarrays (PAM) class prediction tool (12) was applied between TOL and CR (Fig. 1A) and TOL and N (Fig. 1B). Additional microarrays were next performed on six independent blinded TOL-Test patients (TT1–TT6) but were not used to train the class-prediction algorithm. All patients in the TOL-Test group have tolerance scores of >80% with the exception of patient TT4, showing a strong similarity between the TOL and TOL-Test samples (Fig. 1B). PAM three-class prediction was next applied across TOL, CR, and N blood samples and identified a minimal gene set of 49 unique genes. The expression of these genes distinguishes all N controls and CR patients from tolerance and classifies most tolerant patients accurately. None of the CR or N controls score as tolerant in this analysis, although 10 of 11 TOL patients do (fit-to-phenotype scores of >90%; T5 scored ≈50%; Fig. 1C). These 49 genes have been designated the tolerance footprint [supporting information (SI) Table 2]. Patient T6 classified with highest similarity to CR. Thus, the tolerance footprint has a specificity of 100% and a sensitivity of 90% in the training set samples and the test-set of TOL samples. Given the strength of this 49-gene set for tolerance prediction, microarray experiments were performed on all minimal IS [MIS (steroid monotherapy)] patients (n = 10) and long-term stable test-group (STA) (n = 12) patients, and PAM class prediction of tolerance was applied by using these 49 genes. A surprisingly high (50%; 5 of 10 patients) number of MIS patients as well as 1 of 12 (≈8%) STA patients (Fig. 1C) matched tolerance prediction scores of >90%, highlighting the potential detection of operational tolerance in patients on weaning (one drug) and maintenance (three drugs) IS. Global expression variance across these data, when compared with published peripheral blood data sets (1315), shows that operational tolerance state is not just a paradigm shift toward a normal resting state (SI Fig. 2).

Fig. 1.
Identification of “tolerance genes” in training- and test-set patient samples. (A) Tolerance prediction by two-class comparison of TOL (T) and CR (C). A cross-validated comparison of a training set of 5 TOL (green bars), 11 CR patients ...

“Minimal Tolerance Footprint” Predictive of a Potential Tolerant State in Transplant Patients by Using RT-PCR.

Quantitative RT-PCR for the 49 genes from the tolerance microarray data set and hypoxanthine guanine phosphoribosyl transferase were performed in triplicate on RNA extracted from the peripheral blood mononuclear cells of six independent TOL-Test patients (TT7–TT12) and six independent CR-Test patients (CT6–CT11), none of whom were included in microarray analysis (Table 1 and SI Table 3), and seven stable transplant patients randomly chosen among the 12 STA patients (S2, S3, S7, S8, S9, S10, and S12). Gene candidates described in the literature as being associated with tolerance [FOXP3 (16), GITR (17), and Neuropilin-1 (18)] were also tested. Increased expression of FOXP3 (P = 0.009) in TOL relative to CR was confirmed. Expression of neuropilin-1 and GITR was ≈2- and ≈8-fold greater, respectively, in TOL vs. CR, although statistical significance was not reached (data not shown). Many individual gene-expression measurements, from the 49-gene set, were statistically significant by PCR for the tolerance group when compared with the CR group (P < 0.001 for CCL20, TLE4, CDH2, PARVG, and SPON1; P < 0.006 for RAB30, BTLA, and SMILE; P < 0.03 for SOX3, CHEK1, HBB, and DEPDC1; P = 0.045 for CDC2) (SI Fig. 4A). A composite model of 33 of 49 of the most-abundant PCR gene-expression measurements, used in a blind cross-validated PAM two-class analysis, correctly classified TT7–TT11 as tolerant and CT6–CT11 as CR, with a single misclassification (TT12 as CR) (SI Fig. 4B). Interestingly, although TT12 fulfilled the full clinical description of operational tolerance 2 years before and at the time of harvesting (6 months after testing), decline in renal function was observed (creatinemia: 165 μm per liter; proteinuria: 1g per day), with demonstration of anti-donor class II (anti-HLA DR4) antibodies. Given the small tolerant-sample size, it is difficult to speculate on the status of this patient, but this clinical picture suggests that the operational-tolerance gene-expression signature is likely a metastable rather than a permanent state. Composite PCR expression of the 33 genes was next used to classify seven STA posttransplant patients as TOL or CR. Consistent with the microarray-based classification, a single stable patient (S9; see Fig. 1C) was predicted to share the TOL phenotype with a classification score of >99% (0.996) (SI Fig. 4C). Thus, PCR-based blood expression of operational tolerance in renal-transplant recipients appears robust across a modest number of genes.

Table 1.
Demographic summary of patient groups (median and range)

Global Unsupervised Gene Expression: Immunological Quiescence and Cell Cycling Distinguish Tolerance.

Unsupervised clustering across all 11,820 expressed genes on the lymphochip platform and all patient samples shows that the samples largely segregate by clinical phenotypes, with almost complete segregation of TOL or CR samples across a higher-stringency 2,986-gene list (SI Figs. 2 and 5A). CR patients not under IS (indicated by outlined squares in SI Fig. 5A) coclustered with other CR patients, suggesting that the rejection phenotype was stronger than the effect of treatment. Training-TOL and Test-TOL patients cluster together when combined into a single data set, confirming the homogeneity of the training and test groups.

Three gene clusters were identified to drive the segregation of the tolerant samples from CR and stable blood samples (SI Fig. 5A and SI Table 4). Overall, cluster I suggests reduced immune activation in clinical tolerant patients. Cluster II contains several significantly down-regulated signal transduction genes and RNA binding genes. Cluster III contains the discriminating up-regulated cell cycle regulator genes with a significant enrichment of genes expressed primarily during mitosis and in cellular energetic processes.

Biological Relevance of Tolerance Signature.

To assess potential overlap among acute rejection (AR), chronic rejection (CR), and TOL expression signatures in peripheral blood, 14 blood samples collected from patients at the time of biopsy-proven AR (before treatment intensification) were also examined by microarrays. Statistical analysis of microarrays (SAM) identified 893 genes and 982 genes differentially expressed (q < 0.025) in tolerance when compared with CR (SI Fig. 5B) and AR (SI Fig. 4C), respectively. We observed a <4% overlap between differentially expressed genes in AR and TOL (Fig. 1D) in this analysis. Furthermore, the abundance of several transcripts identified by using four-class discrimination of AR, CR, N, and TOL suggests that the greatest expression differences exist between the CR and TOL groups, regardless of IS usage in the former group (Fig. 1D).

Hypergeometric gene-enrichment analysis shows that the majority of the differentially expressed genes expressed at higher levels in TOL show a strong bias toward cell-cycle regulation (19) (P = 0.00038), T cell-specific expression (20) (P = 0.006), with regulation during T cell suppression and costimulation experiments [T cell α-CD3/α-CD28 in vitro costimulation with and without concurrent calcineurin inhibitor, Prograf (FK506) treatment, P = 1 × 10−8 and P = 0.00002 (21)]. In agreement with the Expression Analysis Systematic Explorer (EASE) analysis, clusters of chemokines and cell-adhesion molecules show reduced expression in TOL relative to CR, whereas several ribosomal protein genes, cell cycle and proliferation markers, and key transcription factors show increased expression. Additionally, ≈25% of the genes that differentiate TOL and CR on SAM two-class analysis overlap with the genes that differentiate TOL from N controls by similar analysis (SI Table 5).

TGF-β Plays a Role in Operational Tolerance.

TGF-β is not differentially expressed between TOL and CR, and this is confirmed by the absence of difference at the protein level in the sera of training-group patients. Indeed, the TGF-β1 serum levels of TOL patients (n = 5) were 39.5 ± 3.7 ng/ml compared with those of the CR patients (27 ± 2 ng/ml; n = 7) and were comparable to the levels observed in the normal healthy controls (n = 8). Interestingly, however, we found that TGF-β regulates the function of 27% of the peripheral blood genes that differentiate tolerance from CR. These TGF-β-regulated genes include latent TGF-β binding protein 4 (LTBP4) (2.6-fold), which functions to convert latent TGF-β protein into the active form; N-cadherin (CDH2) (5-fold), which is known to enhance the ability of TGF-β to induce cell-cycle arrest in the G1 phase (22, 23); and CD9, a surface antigen initiating the TGF-β signaling pathway (24) that was expressed at ≈40% higher levels in tolerance. Additional TGF-β-regulated genes include α-fetoprotein, natural killer cell group 7 (NKG7) (25), connective tissue growth factor (CTGF), and fibronectin (FN1) (26), which are gene markers for apoptosis, immune suppression, growth arrest, and the stress-response, respectively, and are also involved in early T and NK cell activation (27).

Altered Immunological Circuits in Operational Tolerance.

A second important finding in this transcription profiling study was that known costimulatory genes were underexpressed in TOL patients compared with CRs (SAM analysis). This finding is cross-validated by a down-regulation of many other genes previously identified as being differentially expressed during independent T cell costimulation experiments in vitro (13). Mirroring the costimulation response, we also observed an absence of up-regulation of genes generally associated with T cell activation. Indeed, classical markers for early and late T cell activation (CD69, TACTILE, LAG3, or SLAM), expression of cytotoxicity-associated genes such as granzyme, perforin, fas, and granulysin (28), and genes characterizing proinflammatory Th1/Th2 responses (TNFα, IL-4, and IL-10) were consistently reduced in TOL patients. Approximately 90% of known proinflammatory cytokines were reduced in TOL patients, supporting evidence for immune quiescence and ignorance of donor antigen (29, 30). Interestingly, a modest enrichment (P = 0.025) for genes identified in memory T cells (31) was seen in the expression signature from our TOL patients.

Analyses for Potential Clinical Confounders for Gene-Expression Measurements.

SAS multivariate analysis and SAM quantitative analysis were performed to test whether differences between the patient groups, such as type of medication, gender, and prior history of malignancy or CMV infection, could act as confounding factors on the gene-expression measurements. Age-related differences were observed for six genes: IL13RA2 (P = 0.05), RAB30 (P = 0.03), RASGRP1 (P = 0.04), TACC2 (P = 0.03), TBX3 (P = 0.05), and TLE4 (P < 0.001). However, with the exception of the transcription factor TLE4 that was also found to correlate with renal function (as measured by serum creatinine levels; P = 0.01), expression of the six genes correlates more tightly with tolerance than with age and remains significantly differentially expressed between the TOL-phenotype patients and MIS, STA, and CR patients (P = 0.0003, P < 0.0001, and P < 0.015, respectively), as observed by using a multiple linear regression model (P value of the model's global test: 0.0001; adjusted R2 = 0.47). Polymorphonuclear cell, lymphocyte, and monocyte counts were measured to test the hypothesis that cell population differences in the periphery may underlie the expression signature identified. No statistical difference in composition was found between the CR and TOL training groups (SI Table 6). Finally, no statistical correlation was found in the fold expression of 10 randomly chosen genes (TK1, PCP4, Serpina5, DHRS2, CCNB2, CYR61, CDH2, SLC38A6, PARVG, and AFP) and the cell counts (R2 = 0.04 ± 0.06 for leucocytes, R2 = 0.14 ± 0.07 for PNN, R2 = 0.006 ± 0.007 for lymphocytes, and R2 = 0.18 ± 0.12 for monocytes). These data, together with the fact that the tolerance phenotype was identified in most of the TOL-Test patients, reinforces the idea that operational tolerance is associated with a distinct and reproducible transcriptional pattern.


Kidney transplantation remains the major treatment for end-stage renal diseases but is often complicated either by AR or CR or by side effects of the long-term IS. The molecular basis of these processes have been analyzed by gene-expression profiling in various studies focusing on AR (32, 33) and CR (34) demonstrating the potential of this approach to decipher complex pathological processes in human disease. Despite some empirical progress in IS minimization (3), reliable biological markers to identify operational tolerance in kidney transplantation do not exist. We have studied a rare immunosuppressive drug-free cohort of 17 TOL transplant patients from different continents, and these studies have validated a specific biomarker footprint of tolerance where peripheral tolerance is predicted with >99% fit-to-class scores in an independent set of immunosuppressive drug-free TOL patients as well as a subset of stable transplant patients on triple- (8% incidence) and single-drug (50% incidence) immunosuppressive therapy. With the exception of a modest number of genes showing age-associated variation in expression, no association with other potential confounding factors such as immunosuppressive drugs, gender, serum creatinine, and CMV or cancer history was observed. This finding lends credence to the fact that the identified gene signature is directly related to the state of tolerance. For the first time, we may be able to define the patients who could be eligible for a progressive decrease of their immunosuppressive medications and, more importantly, identify the patients who need to stay on their current IS dose.

We previously showed that the TOL patients studied are healthy, free of infection and malignancy, and do not display clinical evidence of immunoincompetency (8). Nevertheless, the fact that the operational tolerance definition refers to a clinical status precludes a possible response of the recipient against his donor, and nothing proves that operational tolerance will be indefinite. Two patients (T6 and TT12) were predicted to resemble chronic rejection. However, their transcriptional profiles and class prediction scoring distinguish them from other TOL patients even before eventual decline in graft dysfunction. Furthermore, the loss of the peripheral signature for tolerance correlates clinically to a change in clinical phenotype from operational tolerance to rejection.

DNA microarrays, which allow detailed measurements of gene expression on an unprecedented global scale (35), have been criticized for variability, lack of reproducibility, and difficult data management. Although more recent technical developments have largely circumvented these issues (36), they remain relevant to the present study because of the rarity of TOL kidney graft recipients and the multiplicity of mechanisms which have been shown to be involved in experimental tolerance models (37). Therefore, several strategies were used to ascertain the robustness and reproducibility of the obtained gene expression profile. (i) Although still a relatively small study because of the extreme scarcity of spontaneous tolerance in kidney transplantation, all 17 patients with operational tolerance and the 74 reference samples were carefully phenotyped and age/gender-matched. As such, the analyses were performed on the largest cohort of TOL kidney graft recipients ever studied. (ii) Gene expression data were tested for potential confounding factors. (iii) Both CR and healthy volunteers were used as reference groups to ascertain that the obtained profile was specific for operational tolerance and not due to an absence of IS or the presence of good renal function. (iv) The gene expression was confirmed on RNA samples with an independent technique by using quantitative RT-PCR. (v) Finally, the peripheral blood-based expression profile was validated on independent samples both by microarrays and RT-PCR.

Although highly convenient for serial analysis in clinics, blood may not be the ideal tissue source for deciphering the underlying molecular mechanisms in graft acceptance. Nevertheless, changes in peripheral blood expression profile have been shown to correlate with biopsy-proven heart rejection (38) and may be useful in immunosuppressive management (39). Moreover, regulatory lymphocytes from peripheral blood transfer long-term survival in a fully mismatched liver allograft rat model (40), suggesting that regulatory mechanisms may be accessible in the periphery. Thus, despite the limitations inherent to peripheral blood sampling, several interesting observations emerge from this study. Although the mRNA levels of TGF-β were not substantially different between patient groups, this chemokine regulates the function of ≈27% of the genes that define the naturally acquired tolerance signature. TGF-β has been shown to be involved in various animal models of tolerance during both the induction (41) and maintenance phases (42). It is also important for homeostasis of CD4+CD25+/CD4+CD25 FOXP3+ regulatory T cells (43) and TGF-β-producing Th3 (44) cells that are known to have suppressive activity. The exact mechanisms driving and maintaining spontaneous tolerance in these patients yet remains unclear; however, there are data emerging about the diverse roles of tissue and peripheral TGF-β in protective (43) and injurious (45) alloresponses. The finding of increased expression of certain molecular markers of regulatory T cells in TOL patients over CR patients suggests differences in T cell subsets between these two groups.

As we have previously shown, no evidence of immunoincompetency was observed among the TOL patients studied (8). Peripheral down-regulation of costimulatory signals and Th1/Th2-related cytokines observed in this study thus suggests that these patients have a normal immune system with donor-specific hyporesponsiveness. FOXP3, as previously shown (46), GITR, and neuropilin were elevated by RT-PCR in the TOL and N controls when compared with CR patients, suggesting a role for an intact T cell regulation in tolerance and, conversely, its loss during CR. We have previously shown decreased numbers of regulatory CD4+CD25+ T cells in the blood of CR patients compared with TOL patients and healthy individuals (46). Phenotyping revealed also an increase of CD8+CD28 cells with higher expression of perforin and granzyme A in CR patients compared with TOL patients and healthy individuals (10). The combination of multiple immune-monitoring parameters such as the tolerance expression footprint identified in this study, T cell receptor repertoire alteration (9, 47), changes in mononuclear cell phenotype (6, 10, 46), or ex vivo inhibition of allospecific responses could increase the accuracy of identifying and predicting the tolerant phenotype.

In summary, we have identified a small biomarker panel using gene-expression profiling of peripheral blood from spontaneously tolerant renal-transplant recipients. The expression signature suggests that TGF-β might contribute to this process, possibly by regulating specific phenotypes of peripheral regulatory T cells or altering the threshold for T cell activation (48). Large-scale clinical studies are now warranted to validate the utility of this tolerance footprint as a means to identify spontaneous clinical operational tolerance in long-term kidney recipients with stable graft function, to determine the timing of appearance of the observed tolerant footprint posttransplantation, and to test the stability of this profile over time. In this context, the present study provides a unique, tolerance-specific, clinical monitoring tool to screen a large cohort of stable patients under conventional IS and identify patients that could be subsequently randomized for progressive and controlled reduction in IS.

Materials and Methods

Patient Selection.

Peripheral blood samples were collected from 75 adult renal-transplant patients (grouped into TOL, STA, AR, and CR; SI Table 2) and 16 N adult controls enrolled in this study at Nantes University (Nantes, France) and Stanford University (Palo Alto, CA). The protocol was approved by each of the university hospital ethical committees and institutional review boards, and required written informed consent was obtained. Samples were separated into Training-group and Test-group cohorts containing patient with different clinical phenotypes (see SI Tables 2 and 3). Apart from tolerant patients for whom biopsy was refused by the patients and the relevant hospital ethical committees, all patients had biopsy-confirmed clinical phenotypes.

To generate informative biomarkers by microarray for operational tolerance, Training-group samples (n = 24) were chosen from three clinical phenotypes.

  1. Immunosuppressive drug-free TOL (n = 5) patients with long-term stable graft function, without IS for at least 2 years (mean duration drug-free, 8.8 ± 4.9 years). Stable graft function was defined as stable Cockcroft-calculated creatinine clearance >60 ml per min per 1.73 m2 with absent or low-grade proteinuria (<1.5 g/day). Causes of IS withdrawal were medical necessity (calcineurin-inhibitor nephrotoxicity, n = 1; posttransplant lymphoproliferative disease, n = 1; complete remission from posttransplant lymphoproliferative disease 7 years before study) and patient treatment nonadherence (n = 3). The clinical and biological characteristics of these patients have been described in detail previously (8), and the most relevant demographic and clinical data of the entire population studied are summarized in SI Tables 2 and 3.
  2. CR patients (n = 11). This group was defined according to clinical and histological criteria. All CR patients had a progressive degradation of their renal function (creatinine clearance of <60 ml per min per 1.73 m2 and/or proteinuria of >1.5 g/day) and histological signs of vascular CR defined as endarteritis and allograft glomerulopathy with basement membrane duplication (49, 50). Because CR patients differ from TOL patients by a “maintenance” IS, 4 of the 11 CR patients chosen had been on dialysis and off immunosuppressive treatment for 1.5 ± 0.5 years and were intentionally enrolled to enable analysis of the confounding effect of treatment. Transplant renal biopsy showed Banff-classified chronic active humoral rejection with C4d deposition in three (undetermined for one) circulating anti-donor class II antibodies. A single patient (TT6; see SI Table 3) was without IS for 19 years with stable function before testing but, 6 months after sampling, was found to have proteinuria above the defined threshold (>1.5 g/day) and graft dysfunction. Because the patient refused a biopsy, he was not assigned to a specific group (labeled # in Fig. 1) and clusters with CR samples on microarray analysis of the sample, even many months before his graft dysfunction event.
  3. N (n = 8) were included as controls. They all had a normal blood formula and no infectious or other concomitant pathology for at least 6 months before the study.

To allow for validation of the discovered biomarkers for operational tolerance, an independent, blinded Test-group of samples (n = 67) from six different posttransplant clinical phenotypes, including patients with AR and stable graft function, were examined by expression profiling (microarray and real-time PCR). Descriptions of these different test-group cohorts are as follows.

  1. TOL-Test (n = 12). All new patients shared the same clinical and pathological criteria as described above. All stopped their IS for nonadherence reasons.
  2. MIS (n = 10). Patients with stable graft function on steroid monotherapy (<10 mg/day) for 4.6 ± 2.6 years. Calcineurin inhibitors and CellCept were removed in these patients because of previous posttransplant lymphoproliferative disease (n = 6), cancer (n = 2), or uncontrolled infectious disease (n = 2).
  3. STA (n = 12). Patients with stable kidney graft function at >5 years posttransplantation while under mycophenolate mofetil or azathioprine and maintenance steroids with (n = 5) or without (n = 7) an associated calcineurin inhibitor.
  4. AR-Test (n = 14). Patients experiencing rapid decline (>20% from baseline) in graft function and biopsy-proven AR (51).
  5. CR-Test (n = 11). This group had biopsy-proven chronic rejection with deteriorating graft function and proteinuria.
  6. N (n = 8). These subjects all had normal blood formulae and no infectious or other concomitant pathology for at least 6 months before the study. To limit the number of array experiments, this group was used for RT-PCR validation only. Detailed demographic and clinical data for all of these patients are shown in SI Tables 2 and 3.

For details on microarray experiments, ELISA for TGF-β, and real-time PCR, please see SI Methods. Raw gene-expression files are available as SI Raw Data.


Wilcoxon rank sum test, logistic regression, and Pearson's correlation test (expressed as R2) were run on the clinical data. SAS multivariate analysis and SAM quantitative analysis were performed to test whether clinical differences between the patient groups, such as type of medication, gender, prior history of malignancy, or CMV infection, could act as confounding factors on the gene-expression measurements assessed as group-specific. Hypergeometric enrichment analysis was run on gene-expression data by using published gene-expression data lists. Cluster (52), Expression Analysis Systematic Explorer, and PAM (12) were used to determine the “expression phenotypes” of the unidentified, independent test-group samples.

Supplementary Material

Supporting Information:


We thank Xin Chen, Balasubramanian Narasimhan, Mei-Sze Chua, Rosa Liu, Christophe Legendre, Eric Thervet, Janie Waskerwitz, Sheryl Shah, Phillip Halloran (University of Alberta, Edmonton, AB, Canada), and the Stanford Microarray Database for sample/data collection and/or analysis; and Patrick Miqueu for editing. This work was supported by grants from the National Institutes of Health (U01 AI55795-02 and R01 AI 61739-01 to M.S.), the Clinical Center for Immunological Studies at Stanford University (to M.S.), the Lucille Packard Foundation (to M.S. and E.S.M.), the “Foundation Progreffe,” the “Establishment Français des Greffes,” Roche Organ Transplantation Research Foundation, the National Institute of Allergy and Infectious Diseases, the National Institute of Diabetes and Digestive and Kidney Diseases, and the Juvenile Diabetes Research Foundation.


immunosuppressive-drug-free tolerant transplant recipients
TOL test group
TOL-Test patient
CR-Test patient
age-matched healthy volunteers
minimal immunosuppression (steroid monotherapy)
MIS test group
chronic rejection
CR test group
long-term stable test-group
acute rejection
AR test group
predictive analysis of microarrays
statistical analysis of microarrays
statistical analysis software.


The authors declare no conflict of interest.

This article contains supporting information online at www.pnas.org/cgi/content/full/0705834104/DC1.


1. Opelz G. Transplantation. 1995;60:1220–1224. [PubMed]
2. Nankivell BJ, Borrows RJ, Fung CL, O'Connell PJ, Allen RD, Chapman JR. N Engl J Med. 2003;349:2326–2333. [PubMed]
3. Dantal J, Hourmant M, Cantarovich D, Giral M, Blancho G, Dreno B, JP S. Lancet. 1998;351:623–628. [PubMed]
4. Iman A, Rao M, Juneja R, Jacob CK. Natl Med J India. 2001;14:75–80. [PubMed]
5. Strober S. J Allergy Clin Immunol. 2000;106:S113–S114. [PubMed]
6. Li J, Spletter ML, Johnson JA. Physiol Genomics. 2004;21:43–58. [PubMed]
7. Waldmann H, Cobbold S. Immunity. 2001;14:399–406. [PubMed]
8. Ballet C, Roussey-Kesler G, Aubin JT, Brouard S, Giral M, Miqueu P, Louis S, van der Werf S, Soulillou JP. Am J Transplant. 2006;6:2796–2801. [PubMed]
9. Brouard S, Dupont A, Giral M, Louis S, Lair D, Braudeau C, Degauque N, Moizant F, Pallier A, Ruiz C, et al. Am J Transplant. 2005;5:330–340. [PubMed]
10. Baeten DLS, Braud C, Braudeau C, Ballet C, Moizant F, Pallier A, Giral M, Brouard S, Soulillou JP. J Am Soc Nephrol. 2006;17:294–304. [PubMed]
11. Martinez-Llordella M, Puig-Pey I, Orlando G, Ramoni M, Tisone G, Rimola A, Lerut J, Latinne D, Margarit C, Bilbao I, et al. Am J Transplant. 2007;7:309–319. [PubMed]
12. Tibshirani R, Hastie T, Narasimhan B, Chu G. Proc Natl Acad Sci USA. 2002;99:6567–6572. [PMC free article] [PubMed]
13. Boldrick JC, Alizadeh AA, Diehn M, Dudoit S, Liu CL, Belcher CE, Botstein D, Staudt LM, Brown PO, Relman DA. Proc Natl Acad Sci USA. 2002;99:972–977. [PMC free article] [PubMed]
14. Alizadeh A, Eisen M, Davis RE, Ma C, Sabet H, Tran T, Powell JI, Yang L, Marti GE, Moore DT, et al. Cold Spring Harbor Symp Quant Biol. 1999;64:71–78. [PubMed]
15. Whitney AR, Diehn M, Popper SJ, Alizadeh AA, Boldrick JC, Relman DA, Brown PO. Proc Natl Acad Sci USA. 2003;100:1896–1901. [PMC free article] [PubMed]
16. Cobbold SP, Castejon R, Adams E, Zelenika D, Graca L, Humm S, Waldmann H. J Immunol. 2004;172:6003–6010. [PubMed]
17. Shimizu J, Yamazaki S, Sakaguchi S. J Immunol. 1999;163:5211–5218. [PubMed]
18. Bruder D, Probst-Kepper M, Westendorf AM, Geffers R, Beissert S, Loser K, Von Boehmer H, Buer J, Hansen W. Eur J Immunol. 2004;34:623–630. [PubMed]
19. Whitfield ML, Sherlock G, Saldanha AJ, Murray JI, Ball CA, Alexander KE, Matese JC, Perou CM, Hurt MM, Brown PO, et al. Mol Biol Cell. 2002;13:1977–2000. [PMC free article] [PubMed]
20. Su AI, Cooke MP, Ching KA, Hakak Y, Walker JR, Wiltshire T, Orth AP, Vega RG, Sapinoso LM, Moqrich A, et al. Proc Natl Acad Sci USA. 2002;99:4465–4470. [PMC free article] [PubMed]
21. Diehn M, Alizadeh AA, Rando OJ, Liu CL, Stankunas K, Botstein D, Crabtree GR, Brown PO. Proc Natl Acad Sci USA. 2002;99:11796–11801. [PMC free article] [PubMed]
22. Satterwhite DJ, Neufeld KL. Cell Cycle. 2004;3:1069–1073. [PubMed]
23. Verrecchia F, Tacheau C, Wagner EF, Mauviel A. J Biol Chem. 2003;278:1585–1593. [PubMed]
24. Hattori T, Kawaki H, Kubota S, Yutani Y, de Crombrugghe B, von der Mark K, Takigawa M. J Cell Physiol. 2003;197:94–102. [PubMed]
25. Bellone G, Aste-Amezaga M, Trinchieri G, Rodeck U. J Immunol. 1995;155:1066–1073. [PubMed]
26. Saed GM, Kruger M, Diamond MP. Wound Repair Regen. 2004;12:557–564. [PubMed]
27. Xu Y, Kapp JA. Invest Ophthalmol Visual Sci. 2002;43:3473–3479. [PubMed]
28. Harris MA, Clark J, Ireland A, Lomax J, Ashburner M, Foulger R, Eilbeck K, Lewis S, Marshall B, Mungall C, et al. Nucleic Acids Res. 2004;32:D258–D261. [PMC free article] [PubMed]
29. Toyo-oka K, Tai XG, Yashiro Y, Ahn HJ, Abe R, Hamaoka T, Kobayashi M, Neben S, Fujiwara H. Immunol Lett. 1997;58:19–23. [PubMed]
30. Park CS, Yashiro Y, Tai XG, Toyo-oka K, Hamaoka T, Yagita H, Okumura K, Neben S, Fujiwara H. J Immunol. 1998;160:5790–5796. [PubMed]
31. Holm AM, Sivertsen EA, Tunheim SH, Haug T, Bjerkeli V, Yndestad A, Aukrust P, Froland SS. Clin Exp Immunol. 2004;138:278–289. [PMC free article] [PubMed]
32. Sarwal M, Chua MS, Kambham N, Hsieh SC, Satterwhite T, Masek M, Salvatierra O., Jr N Engl J Med. 2003;349:125–138. [PubMed]
33. Gimino VJ, Lande JD, Berryman TR, King RA, Hertz MI. Am J Respir Crit Care Med. 2003;168:1237–1242. [PubMed]
34. Flechner SM, Kurian SM, Head SR, Sharp SM, Whisenant TC, Zhang J, Chismar JD, Horvath S, Mondala T, Gilmartin T, et al. Am J Transplant. 2004;4:1475–1489. [PMC free article] [PubMed]
35. Stegall M, Park W, Kim D, Kremers W. Am J Transplant. 2002;2:913–925. [PubMed]
36. Jeon MS, Atfield A, Venuprasad K, Krawczyk C, Sarao R, Elly C, Yang C, Arya S, Bachmaier K, Su L, et al. Immunity. 2004;21:167–177. [PubMed]
37. Ashton-Chess J, Brouard S, Soulillou JP. Transplant Int. 2006;19:539–548. [PubMed]
38. Horwitz PA, Tsai EJ, Putt ME, Gilmore JM, Lepore JJ, Parmacek MS, Kao AC, Desai SS, Goldberg LR, Brozena SC, et al. Circulation. 2004;21:3815–3821. [PubMed]
39. Deng MC, Eisen HJ, Mehra MR, Billingham M, Marboe CC, Berry G, Kobashigawa J, Johnson FL, Starling RC, Murali S, et al. Am J Transplant. 2006;6:150–160. [PubMed]
40. Fujino M, K Y, Kawasaki M, Funeshima N, Kimura H, Nakajima T, Saito H, Li XK. Liver Transplant. 2004;10:379–391. [PubMed]
41. Gagne K, Brouard S, Guillet M, Cuturi MC, Soulillou JP. Eur J Immunol. 2001;31:3111–3120. [PubMed]
42. Knechtle SJ, Burlingham WJ. Transplantation. 2004;77:936–939. [PubMed]
43. Hori S, Nomura T, Sakaguchi S. Science. 2003;299:1057–1061. [PubMed]
44. Weiner HL. Microbes Infect. 2001;3:947–954. [PubMed]
45. Chua MS, Barry C, Chen X, Salvatierra O, Sarwal MM. Am J Transplant. 2003;3:17–22. [PubMed]
46. Louis S, Braudeau C, Giral M, Dupont A, Moizant F, Robillard N, Moreau A, Soulillou JP, Brouard S. Transplantation. 2006;81:398–407. [PubMed]
47. Alvarez CM, Opelz G, Giraldo MC, Pelzl S, Renner F, Weimer R, Schmidt J, Arbelaez M, Garcia LF, Susal C. Am J Transplant. 2005;5:746–756. [PubMed]
48. Graca L, Chen TC, Moine AL, Cobbold SP, Howie D, Waldmann H. Trends Immunol. 2005;26:130–135. [PubMed]
49. Solez K, Colvin RB, Racusen LC, Sis B, Halloran PF, Birk PE, Campbell PM, Cascalho M, Collins AB, Demetris AJ, et al. Am J Transplant. 2007;7:518–526. [PubMed]
50. Colvin RB. J Am Soc Nephrol. 2007;18:1046–1056. [PubMed]
51. Racusen LC, Solez K, Colvin RB, Bonsib SM, Castro MC, Cavallo T, Croker BP, Demetris AJ, Drachenberg CB, Fogo AB, et al. Kidney Int. 1999;55:713–723. [PubMed]
52. Eisen MB, Spellman PT, Brown PO, Botstein D. Proc Natl Acad Sci USA. 1998;95:14863–14868. [PMC free article] [PubMed]

Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


  • Compound
    PubChem Compound links
  • MedGen
    Related information in MedGen
  • PubMed
    PubMed citations for these articles
  • Substance
    PubChem Substance links

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...