SEARCH

SEARCH BY CITATION

Keywords:

  • antibody;
  • Banff classification;
  • macrophages;
  • microarrays;
  • rejection

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Banff defines T-cell-mediated rejection (TCMR) using nonspecific lesions and arbitrary cutoffs, with no external gold standard. We reexamined features of TCMR using exclusively molecular definition independent of histopathology. The definition was derived from mouse kidney transplants with fully developed TCMR, and is based on high expression of transcripts reflecting IFNG effects and alternative macrophage activation. In 234 human kidney transplant biopsies for cause phenotyped by microarrays, we identified 26 biopsies meeting these criteria. After excluding three biopsies with unrelated diseases, all 23 biopsies had typical Banff lesions of TCMR (inflammation, tubulitis), with v lesions in 10/23. Banff histopathology diagnosed 18 as TCMR, 1 as mixed and 4 as borderline. Despite marked changes in transcriptome indicating tissue injury and dedifferentiation, all kidneys with molecularly defined TCMR, even with v lesions or late rejection, demonstrated excellent recovery of function at 6 months with no graft loss (mean follow-up 2.5 years). Thus TCMR defined exclusively by molecules manifests TCMR–related lesions and function impairment, but good recovery and survival, even with late rejection or arteritis. This combination of pathologic, clinical and molecular features constitutes the typical or canonical T-cell-mediated rejection.


Abbreviations: 
ABMR

antibody-mediated rejection

DSA

donor specific HLA antibodies

IFTA

interstitial fibrosis and tubular atrophy

PRA

panel reactive HLA antibodies

TCMR

T-cell-mediated rejection.

Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

In kidney transplantation, T-cell-mediated rejection (TCMR) remains an important disease that must be recognized, treated and distinguished from other conditions. According to the Banff classification, TCMR is defined by histologic features: interstitial inflammation (i-score ≥ 2), tubulitis (t-score ≥ 2) and/or intimal arteritis (v-score > 0) (1). The infiltrate includes effector and effector memory T cells and myeloid cells of the monocyte-macrophage-dendritic cells lineage (2,3). The transcriptome of biopsies with TCMR reveals intense expression of IFNG-induced transcripts (GRITs) (4–6), cytotoxic T-cell-associated transcripts (CATs) (5,7–9) and loss of kidney transcripts (KTs) (10). Although cytotoxic T-cell molecules are strongly expressed, TCMR is not mediated by cytotoxicity: disruption of the potential cytotoxic mechanisms—perforin, granzymes, Fas–Fas ligand—has no effect on the development of the diagnostic TCMR lesions in mouse kidney allografts (11,12).

The histologic lesions used by Banff to diagnose TCMR are nonspecific, and the threshold for diagnosing TCMR was derived empirically (13), leaving considerable potential for errors. Interstitial infiltrate and tubulitis also occurs in renal injury, infection and primary diseases such as glomerulonephritis, and in protocol biopsies of well functioning kidney transplants (14,15). Assessing these lesions semiquantitatively confers some specificity: the Banff consensus selected arbitrary cutoffs for i-score and t-score to permit a diagnosis of TCMR in the context of a kidney transplant biopsy for cause (16–18). Intimal arteritis is also not specific for TCMR because it also occurs in antibody-mediated rejection (ABMR) (19–21). Moreover the manifestations of TCMR may be modified by drugs, as is seen in the monocyte-rich TCMR described in T-cell-depleted patients (22). Such studies emphasize that the monocyte-related population is a robust feature of TCMR, and a heavy T-cell infiltrate is not necessary. Monocytes can also carry a semidirect presentation of endothelial antigens to T cells (23) and are also associated with graft dysfunction (24).

Because transcript changes provide an independent assessment of rejection, we recently compared changes in transcripts to histologic lesions and Banff diagnostic labels in kidney transplant biopsies for cause (5,25). Transcript changes correlated with most Banff consensus features of TCMR, but highlighted three areas for reassessment. First, transcripts indicate that the Banff distinction of tubulitis grade t1 from t2 is not valid and must be reexamined, because t1 and t2 had identical molecular abnormalities. Interestingly, pathologists have also found this distinction poorly reproducible (5,25). Second, transcripts indicated that some biopsies Banff labels TCMR on the basis of intimal arteritis (‘v’) without interstitial inflammation and tubulitis lack molecular features of TCMR (5,25). This has triggered a multicenter study to address the significance of such isolated v lesions (Banff ’09 report, manuscript in preparation). Finally, transcripts indicate that some biopsies labeled ‘borderline’ are actually typical TCMR (5,25).

In the present study we applied a previously developed molecular definition of a canonical TCMR (cTCMR) derived in mouse kidney transplants (Famulski et al., in press) to human kidney biopsies to see the features and outcomes of kidneys with cTCMR. We have reported that when mouse kidney transplants manifest the full histopathology picture of TCMR, particularly tubulitis and intimal arteritis, they express alternative macrophage activation transcripts (AMAT1) such as Arg1, Mgl1, Clec7a, Mmp12, Mmp9, Mrc1, Thsb1, at the same time as they express high levels of IFNG-induced transcripts (GRIT1) such as CXCL9 (4,26). We hypothesized that this combination of high GRIT1 and AMAT1 scores would identify biopsies with cTCMR independent of histopathology. We selected human kidney biopsies for cause with high GRIT1-AMAT1 scores and examined their Banff histopathological lesions, diagnostic labels, function and survival after the biopsy, compared to other biopsies for cause. The goal of this study was to describe the features and behavior of human kidneys with molecular TCMR, but not to develop the diagnostic applications of this knowledge (e.g. specificity, sensitivity), which is the subject of an ongoing study.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Patient population and specimens

The study was approved by the University of Alberta Health Research Ethics Board (Issue no. 5299) and by the University of Illinois, Chicago, Office for the Protection of Research Subjects (protocol no. 2006–0544). After receiving written informed consent, every consenting patient undergoing a transplant biopsy for cause (deterioration in function, proteinuria, stable impaired function) as standard of care between 01/2004 and 01/2008 was included (Table 1). Clinical data was collected for each patient and entered into a Scientific Data Management System. For controls, normal kidney tissue was obtained from histopathologically unaffected areas of the cortex of native nephrectomies performed for renal carcinoma (eight specimens). Biopsies were obtained under ultrasound guidance by spring-loaded needles (ASAP Automatic Biopsy, Microvasive, Watertown, MA).

Table 1.  Demographics and clinical characteristic of 173 patients (234 biopsies for cause)1
Demographics [n = 173]All biopsies (n = 234)Biopsies showing Banff TCMR (n = 45)Biopsies showing other diagnoses (n = 189)p-Value (Banff TCMR vs. all others)
Recipient gender (% male) (n = 173)153 (65%)33 (73%)120 (63%)0.2124
Race [n = 173]
 Caucasian134 (57%)30 (67%)104 (55%)0.1560
 Black19 (8%)1 (2%)18 (10%)0.1349
 Other81 (35%)14 (31%)67 (35%)0.5825
Primary disease [n = 173]
 Diabetic nephropathy32 (14%)3 (7%)29 (15%)0.1526
 Hypertension/large vessel disease24 (10%)4 (9%)20 (11%)1
 Glomerulonephritis/vasculitis90 (38%)17 (38%)73 (39%)0.9165
 Interstitial nephritis/pyelonephritis23 (10%)6 (13%)17 (9%)0.3796
 Polycystic kidney disease32 (14%)9 (20%)23 (12%)0.1694
 Others22 (9%)2 (4%)20 (11%)0.2651
 Unknown etiology11 (5%)4 (9%)7 (4%)0.2296
Previous transplant (n = 173)22 (9%)3 (7%)19 (10%)0.5833
Donor gender (% male) (n = 144)97 (41%)20 (44%)77 (41%)0.6504
Donor type (% deceased donor transplants) (n = 173)128 (55%)24 (53%)104 (55%)0.8375
Mean time from transplantation to biopsy (days)1 (n = 139)1240 (6–11453)707 (8–4445)1111 (6–7649)0.1038
Median time from transplantation to biopsy (days)1 (n = 139)511 (6–11453)241 (8–4445)320 (6–7649)0.1038
Mean time of follow-up after biopsy (days)1 (n = 139)922 (364–1651)893 (364–1588)910 (365–1675)0.4091
Median time of follow-up after biopsy (days)1 (n = 139)887 (364–1651)782 (364–1588)874 (365–1675)0.4091
Clinical characteristics at time of biopsy [n = 234]All biopsies (n = 234)Biopsies taken Banff TCMR (n = 45)Biopsies taken all others (n = 189)p-Value (Banff TCMR vs. all others)
  1. 1Values were calculated after exclusion of failed grafts and deceased patients. Of the 61 repeat biopsies, 213 were the follow-up biopsies from the previous ones to determine the outcome of intervention.

Maintenance immunosuppressive regimens at biopsy
 MMF, Tacrolimus, steroid102 (44%)21 (47%)81 (43%)0.6433
 MMF, Tacrolimus7 (4%)0 (0%)7 (4%)0.3514
 MMF, Cyclosporine, steroid56 (32%)10 (22%)46 (24%)0.7649
 MMF, Steroids5 (3%)0 (0%)5 (3%)0.5862
 Azathioprine, Cyclosporine, Steroids21 (12%)3 (7%)18 (10%)0.7728
 Others43 (25%)11 (24%)32 (17%)0.2422
Indication for biopsy
 Primary nonfunction8 (3%)2 (4%)6 (3%)0.6520
 Rapid deterioration of graft function48 (21%)10 (22%)38 (20%)0.7520
 Slow deterioration of graft function88 (38%)17 (38%)71 (38%)0.9790
 Stable impaired graft function28 (12%)5 (11%)23 (12%)1
 Investigate proteinuria25 (11%)2 (4%)23 (12%)0.1807
 Follow-up from previous biopsy213 (6%)4 (9%)9 (5%)0.2821
 Others10 (4%)0 (0%)10 (5%)0.2158
 Indication unknown14 (6%)5 (11%)9 (5%)0.1529

Diagnostic classifications

Banff histopathologic diagnoses included rejection (T-cell-mediated—TCMR, antibody-mediated—C4d+ABMR, mixed—TCMR/C4d+ABMR, borderline changes suspicious for rejection), BK nephropathy and nonrejecting for example calcineurin inhibitor toxicity, recurrent or de novo glomerulonephritis or ATN (27). Banff consensus criteria were the basis for the diagnoses (1, 28). C4d+ ABMR was diagnosed as follows: Acute C4d+ABMR (n = 3); presence of diffuse C4d staining and donor specific antibody (DSA) and acute tissue injury (ATN-like minimal inflammation and/or peritubular capillaritis and/or glomerulitis and/or thromboses and/or severe arterititis [v3]). Chronic C4d+ABMR (n = 14); presence of diffuse C4d staining and DSA and chronic tissue injury (transplant glomerulopathy and/or peritubular capillary basement membrane multilayering and/or interstitial fibrosis/tubular atrophy and/or fibrous intimal thickening in arteries).

We also inspected biopsies for C4d negative ABMR, a new category we recently defined, previously designated in Banff as suspicious for ABMR in Banff (28). We defined C4d negative ABMR as follows: DSA positive and negative or minimal focal or focal C4d immunofluorescence staining and either of the following: peritubular capillaritis (ptc > 0) and/or glomerulitis (g > 0) and/or thromboses and/or arteritis (v > 0) with no tubulointerstitial TCMR (I < 2 and t < 2). Of 30 C4d negative ABMR, 13 had morphologic evidence of acute and 13 had chronic tissue injury (n = 13) (27,29).

Biopsies were assessed and graded by a pathologist (BS), blinded to the results of molecular studies. In addition to Banff lesions, total interstitial inflammation (ti score) was scored in both nonscarred and fibrotic cortical parenchyma either excluding or including immediate subcapsular cortex and adventitia around large vessels (ti1 and ti2 respectively). The diagnosis of TCMR and borderline was constructed using ‘i’ score in both unscarred and scarred cortex.

Renal function

Renal function was defined by estimated creatinine clearance (eGFR, Cockcroft-Gault) at the time of biopsy and 6 month after the biopsy. We also calculated the change in eGFR (delta eGFR) at 6 month after the biopsy.

Follow-up

The follow-up time is defined as time between the last biopsy of each patient and end of follow-up. Death censored graft survival is defined as the time to return to dialysis. The mean time of follow-up of patients alive with functioning grafts is 922 days.

RNA extraction and microarrays

As previously reported (5), in addition to cores for diagnostic histopathology, one 18-gauge biopsy core was collected for gene expression analysis and placed immediately in RNA. Later, kept at 4°C for 4–24 h, then stored at −20°C. RNA extraction, quality control and HG_U133_Plus_2.0 GeneChip (Affymetrix, Santa Clara, CA) processing were described previously (5). Detailed protocols are available in the Affymetrix Technical Manual (http://www.affymetrix.com).

Analysis of microarray data

Human Genome U133 plus 2.0 arrays (Affymetrix) were preprocessed in one batch using robust multichip averaging (RMA, Bioconductor version 1.9, R version 2.4). The microarray expression data is available in the supplemental material. Gene expression is given as fold change versus controls (nephrectomies). Following transcript sets representing various biological processes were included in the analyses (Table 2): features of alternative macrophage activation (AMAT1), transcripts of IFNG effects (GRIT1), CTL-associated transcripts (QCAT), macrophage-associated transcripts (QCMAT, see below), kidney parenchymal function (KT2) and parenchymal injury and repair (IRIT) (see results section). The AMAT1 set contains human orthologs (Affymetrix) of mouse transcript markers of alternative macrophage activation (26). We added CD163, the prototypical marker of AMA in humans. Thus AMAT1 set contains the following genes: ARG1, CD163, CD163L1, CLEC7A, CLEC10A, M160, MMP9, MMP12, RNASE3 and THBS1. Gene expression within each transcript set was summarized as the gene/transcript set score: the geometric mean of fold changes across all probe sets in the transcript set. Alternatively, gene set scores can be expressed as the z-scores after median sweep normalization (Bioconductor) of 234 microarrays alone. This approach gives exactly the same relationships between the summarized gene expression, histology and function across the biopsies for cause.

Table 2.  Pathogenesis based transcript sets (PBTs)
PBT abbreviationPBT nameBiological descriptionReference
QCATQuantitative cytotoxic T- cell-associated transcriptsBurden of T cells(2)
QCMATQuantitative constitutive macrophage-associated transcriptsBurden of macrophagesthis paper
GRIT1γ-interferon and rejection induced transcripts set 1Interferon γ effects on the tissue(4,5)
AMAT1γ-interferon suppressed transcripts—alternative macrophage activation set 1Transcripts associated with alternatively associated macrophages based on literature(26)
ENDATEndothelium-associated transcriptsTranscripts associated with endothelium based on literature(29)
IRITD (I)Injury and repair induced transcripts day 3Transcripts induced by a nonimmune kidney injury in isografts peaking around day 3 post transplant(5,38)
IRITD (L)Injury and repair induced transcripts day 5Transcripts induced by a nonimmune kidney injury in isografts peaking around day 5 post transplant(5,38)
KT2Kidney transcripts—set 2Epithelial solute carriers showing high expression in control mouse kidney and reduced after injury or rejection(5,10)

Gene set scores were compared using the one way ANOVA with Dunnett Multiple Comparison Test after log2 transformation.

Expression values for individual AMAT1 and QCMAT transcripts were calculated as fold change versus eight nephrectomies that served as controls (5). In case of multiple probe sets per gene, one with the lowest p value in comparison of high GRIT1-AMAT1 group versus remaining TCMR group was selected by the Kruskal–Wallis t-test with Dunn's Multiple Comparison Test.

Real time RT-PCR

Details pertaining to RT-PCR analysis can be found in the supplemental section.

Quantitative constitutive macrophage-associated transcripts (QCMAT)

Monocytes were purified from PBMCs from healthy volunteers by immunomagnetic positive selection of CD14+ cells to a purity ≥99% according to manufacturer's instructions (StemCell Technologies, Vancouver, British Columbia, Canada). Purified monocytes were resuspended in complete RPMI (Invitrogen, Carlsbad, CA), and allowed to adhere on 100 mm plates (BD Falcon, Missisauga, Ontario, Canada) for 24 h in 5% CO2 at 37°C. Plates were rinsed with HBSS (Invitrogen) and adherent cells were recovered following treatment with Trypsin (Invitrogen). RNA was extracted from cultured macrophages using TRIzol (Invitrogen) according to manufacturer's instructions. Macrophage cRNA was diluted with kidney cRNA in increasing ratios (1:1; 1:4; 1:8; 1:16 and 1:32) and each dilution was hybridized onto Affymetrix human microarrays. Transcripts with expression levels >1000 (raw values) in macrophage cultures were selected as long as expression was <200 in normal kidney, purified effector CD4 and CD8 CTL, B cells, NK cells, human umbilical vein endothelial cells and renal proximal tubule epithelial cells. All samples were preprocessed in one batch using RMA. The transcripts that met the above criteria and also showed a Pearson correlation coefficient >0.9 with the serially diluted macrophage cRNA were selected. The list of transcripts belonging to the QCMAT set is presented in Supporting Table S1.

Molecular classification of rejection

We previously described a molecular classifier for all rejection, including both TCMR and ABMR (25). We rederived this equation using the present 234 biopsies. We defined two classes based on the Banff diagnostic labels: rejecting (TCMR, n = 45, and ABMR, n = 17, and mixed, n = 4, total = 66); and nonrejecting (BK virus, n = 7; borderline, n = 34 and ‘other’ e.g. ATN, CNIT, n = 127, total n = 168). Classifier statistics were calculated using a multiple resampling method (30). The 234 samples were split 50:50 into training and test sets of size 117. Stratified sampling (31) was used to maintain equal proportions of rejecting biopsies in each set. This splitting was repeated 1000 times.

Supplemental materials and methods

  • 1
    Real time RT-PCR of the AMA markers in the biopsies for cause.
  • 2
    Microarray expression data and its description.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Demographics of biopsied patients

We studied 234 consecutive renal transplant biopsies for cause from 173 consenting patients, biopsied between 1 week and 31 years post transplant (Table 1), as detailed elsewhere (27). All consenting patients were included without exclusions or selection, in order to represent the complete spectrum of diagnoses in renal transplants presenting with indications for biopsy for cause. The biopsies were given a code based on their Banff histopathology label, indicating A for C4d positive ABMR (A1-A17; T for TCMR (T1-45); M for mixed (M1-4); P for polyoma virus nephropathy (P1-7); B for borderline (B1-34) and O for others that is nonrejecting, including recurrent/de novo glomerulonephritis (O1-127). The mean follow-up time post biopsy for patients who have not experienced graft failure or death is 922 days (range 364–1651).

Identifying human kidney biopsies with molecularly defined cTCMR changes

We examined all biopsies for the expression of GRIT1 and AMAT1 in a two dimensional plot (Figure 1; colors are substituted for the Banff label; e.g. red = Banff TCMR. Thus T1 is represented as ‘1’ in red). We identified 26 biopsies with high GRIT1-AMAT1 scores, arbitrarily selecting the 80th percentile as the definition of ‘high’ and compared them with their previously assigned Banff diagnostic labels (Table 3). This exercise was designed to select pure cTCMR cases for study, but probably leaves some similar but milder cases behind by using these thresholds. By histopathology, 18 of the high GRIT1-AMAT1 biopsies had previously been labeled TCMR (T3, 4, 7, 8, 9, 10, 12, 15, 18, 19, 20, 21, 23, 25, 35, 40, 41 and 45); one (M4) had been labeled mixed and four (B13, 16, 19, 33) had been labeled borderline. The four borderline cases all had interstitial infiltration (I 1–3) and tubulitis (t1–2) but failed to meet the arbitrary Banff threshold (i2 t2 and/or v lesions) required to assign the TCMR label. The 23 biopsies with high GRIT1-AMAT1 scores were taken from 18 patients, while the remaining 27 biopsies with Banff TCMR were from 25 patients.

image

Figure 1. Scatter plot of 234 biopsies for cause ordered by the summarized expression scores for GRIT1 and AMAT1 transcript sets. The scores are expressed as log2 values of fold change versus control nephrectomies. Colors codes for Banff diagnoses are shown in the inset and numbers indicate individual biopsies. Thus, for example green numbers 1–7 indicate all biopsies with polyoma virus nephropathy, while red numbers 1–45 represent all biopsies with Banff TCMR. The top righthand quadrant represents the 80th percentile for both GRIT1 and AMAT1 scores. Other—Banff nonrejecting, BK virus—polyoma virus nephropathy.

Download figure to PowerPoint

Table 3.  High GRIT1-AMAT1 biopsies classified according to Banff 2007 categories (28)
Banff diagnosis (no.)Biopsies with high AMAT1 and GRIT1 scoresRemaining biopsies
TCMR (45)18 (40%) 27 (60%)
C4d+ABMR (17) 1 (A13) (6%) 16 (94%)
Mixed (4) 1 (25%)  3 (75%)
Borderline (34) 4 (12%) 30 (88%)
Polyoma virus nephropathy (7) 0 (0%)  7 (100%)
Other (127) 2 (O63, O115) (2%)125 (98%)
Total: 23426 (11%)208 (89%)

The high GRIT1-AMAT1 phenotype occurred in only 3/151 (2%) of biopsies lacking TCMR lesions, and all had other reasons for interstitial inflammation. A13 had recurrent lupus; O63 had severe epithelial changes typical of virus infection despite reportedly negative polyoma virus immunohistochemistry; O115 had recurrent membranoproliferative glomerulonephritis with marked interstitial inflammation (Table 3).

Thus with three biopsies with other diseases excluded, every high GRIT1-AMAT1 biopsy had TCMR histologic lesions, and had been labeled TCMR, mixed and borderline by Banff. Importantly, no biopsy with C4d positive ABMR had high GRIT1-AMAT1 scores (except A13, which also had lupus).

Biopsies selected for high GRIT1-AMAT1 show severe histologic lesions of TCMR

High GRIT1-AMAT1 biopsies displayed higher i-scores and ti scores than other Banff diagnostic groups, including the remaining Banff TCMR biopsies (Table 4). This also was true when only one (last) biopsy per patient was analyzed (not shown). High GRIT1-AMAT1 biopsies also had high t scores, as did the other biopsies labeled TCMR by Banff as expected, given the importance of t scores in the Banff TCMR label. V lesions occurred in 10/23 cTCMR biopsies: T7, 9, 12, 15, 18, 20, 21, 35, 41, 45 and in 11/27 of remaining Banff TCMR. Cases T7, 9 and 15 also had C4d negative ABMR, which could have produced the v lesions, but the others did not.

Table 4.  Comparison of scores for Banff lesions in high GRIT1-AMAT1 biopsies versus remaining Banff diagnoses. Lesions of TCMR are in bold
LesionHigh GRIT1- AMAT N = 23Remaining TCMR* N = 27Remaining C4d+ ABMR* N = 16Remaining borderline* N = 30Remaining nonrejecting* N = 125
  1. *In each group the high GRIT1-AMAT1 cases have been removed. Mean scores are shown. Statistical test: Kruskal–Wallis with Dunn's Multiple Comparison Test. bxTx indicates time from transplant to biopsy.

  2. 1p < 0.0001.

  3. 2p < 0.01.

  4. 3p < 0.001.

  5. 4p < 0.05.

g0.560.330.880.200.28
cg0.300.331.5310.200.41
i1.910.6310.1310.3310.221
ti (1)2.441.8521.4731.2731.133
ti (2)2.521.7041.3131.3730.711
ci1.001.5241.7141.470.201
t2.221.890.4711.2320.201
ct1.301.441.591.501.17
v0.640.520.4100
cv1.321.521.411.101.25
ah0.570.891.4141.2741.274
mm0.390.441.6510.530.62
ptc0.830.411.9420.530.28
ptcml4.134.437.6443.233.612
bxTx (in days)1182058912498237

Conversely, double contours (cg), mesangial matrix increase, capillaritis and capillary multilayering were increased in ABMR but not in high GRIT1-AMAT1 biopsies. High GRIT1-AMAT1 biopsies had less fibrosis (ci) and hyalinosis (ah), probably because they were taken a shorter time post transplant than biopsies in other groups.

Thus biopsies with high GRIT1-AMAT1 scores had histopathology findings lesions of TCMR and lacked the lesions of ABMR.

Every biopsy with high GRIT1-AMAT1 scores is identified as rejection by the classifier

We tested the 23 biopsies with high GRIT1-AMAT1 scores with our rejection–nonrejection classifier based on predictive analysis of microarrays (PAM) (25). PAM uses a weighted assessment of molecular features to assign the probability of rejection (25). Every biopsy labeled high GRIT1-AMAT1 was also labeled rejection by the PAM classifier (Figure 2).

image

Figure 2. PAM rejection probabilities. Each number represents individual biopsy, same as in Figure 1 and its location on the graph is dictated by the predicted rejection probability. The vertical doted lines separate their Banff histopathology diagnoses. Thus for example biopsy no. 4 in the C4d+ABMR category is the same as blue 4 in Figure 1, while biopsy no. 4 in TCMR category is the same as red 4 in Figure 1. Circled numbers indicate biopsies with high GRIT1-AMAT1 expression. Probabilities over 0.5 are predicted to be rejecting. PVN—polyoma virus nephropathy.

Download figure to PowerPoint

Other transcript sets are strongly disturbed in high GRIT1-AMAT1 biopsies

We previously defined pathogenesis-based transcript sets (PBTs) reflecting the major biological processes in rejecting transplants (5). Details of PBTs are provided on the home page http://transplants.med.ualberta.ca/. We compared expression of PBTs in high GRIT1-AMAT1 biopsies to other groups classified by their Banff labels (Table 5). High GRIT1-AMAT1 biopsies had more severe disturbances in CTL-associated transcripts (QCAT), constitutive macrophage transcripts (QCMAT), KTs induced by injury (intermediate IRITs, late IRITS), and KTs decreased by injury (KT2). The exception was endothelial transcripts (ENDATs), which were most increased in ABMR as previously reported (29).

Table 5.  Comparison of the PBT scores in high GRIT1-AMAT1 biopsies versus other Banff diagnoses (main diagnostic label only)*
PBTHigh GRIT1-AMAT1 N = 23Remaining TCMR* N = 27Remaining C4d+ ABMR* N = 16Remaining borderline* N = 30Remaining nonrejecting* N = 125
  1. *In each group the high GRIT1-AMAT1 cases have been removed. Statistical test: one way ANOVA with Dunnett Multiple Comparison Test.

  2. 1p < 0,0001.

  3. 2p < 0.001.

  4. 3p < 0.05.

  5. 4p < 0.01.

AMAT11.71.211.311.111.11
GRIT12.91.912.221.611.51
QCMAT2.61.511.611.311.31
QCAT5.72.412.511.911.61
IRIT (I)1.11.131.11.011.14
IRIT (L)1.41.241.231.211.21
ENDAT1.11.11.511.01.0
KT20.60.810.820.810.81

Expression of individual macrophage-associated transcripts in high GRIT1-AMAT1 biopsies

We compared expression of individual AMATs in high GRIT1-AMAT1 biopsies to the remaining Banff diagnoses. Seven out of ten AMAT1 had higher expression in GRIT-AMAT1 biopsies than in the remaining Banff TCMR and other diagnoses (Table 6).

Table 6.  Comparison of fold increase values for individual AMAT1 and QCMAT genes in high GRIT1-AMAT1 biopsies versus remaining TCMR and other Banff diagnoses
Gene abbreviationHigh GRIT1-AMAT1 N = 23Remaining TCMR* N = 27Remaining C4d+ABMR* N = 16Remaining borderline* N = 30Remaining nonrejecting* N = 125
  1. *In each group the high GRIT1-AMAT1 cases have been removed. Expression values were calculated as fold change versus eight nephrectomies. In case of multiple probe sets per gene, one with the lowest p-value in comparison of high GRIT1-AMAT1 versus remaining TCMR was selected. Statistical test: Kruskal–Wallis with Dunn's Multiple Comparison Test. Forty-seven out of 71 QCMAT were significantly different in the high GRIT1-AMAT1 group versus remaining TCMR. The top 10 transcripts, ordered by their p-values, are shown.

  2. 1p < 0.001.

  3. 2p < 0.0001.

  4. 3p < 0.05.

  5. 4p < 0.01.

Significant AMAT1s
 CD16310.44.314.012.622.52
 CLEC7A6.11.822.321.321.22
 CLEC10A3.91.821.721.521.32
 M160 (CD163B)1.41.131.131.141.02
 MMP94.01.911.511.521.62
 MRC14.72.422.941.721.62
 THBS11.10.710.740.740.83
Significant QCMAT (top 10 by p-value)
 IGSF63.91.421.621.321.22
 CENTA22.01.421.321.221.22
 DMXL21.41.021.521.121.02
 CCRL21.61.121.221.021.12
 HK31.61.221.241.121.12
 MREG1.91.321.221.331.21
 IL1RN1.71.221.241.141.22
 ECGF12.81.421.431.211.22
 PLA2G73.31.521.331.311.32
 MS4A75.12.424.041.721.42

The real time RT-PCR verified that the expression of prototypical AMA markers, CD163 and MRC1 calculated as percentage of HPRT was increased in the high GRIT1-AMAT1 group compared to the remaining TCMR biopsies (894.7 vs. 236.3 and 6.5 vs. 1.7, respectively; Supporting Table S2).

We also analyzed expression of individual constitutive macrophage transcripts highly expressed in adherent human macrophages, excluding overlap with AMAT1 transcripts. High GRIT1-AMAT1 biopsies displayed higher expression of 47 of these other macrophage transcripts than other biopsy groups (Table 6). Thus biopsies selected for high GRIT1-AMAT1 scores also had high expression of transcripts associated with adherent macrophages in culture.

Renal function recovers in high GRIT1-AMAT1 cases

We studied the estimated glomerular filtration rate (eGFR) at the time of biopsy and where available 6 months after biopsy. Kidneys with high GRIT1-AMAT1 scores had decreased function at the time of biopsy (Table 7) but recovery of function appeared 6 months post biopsy. This was particularly striking in cases presenting late (>1 year post transplant), which had better recovery of function than any other late biopsy group. This contrasts with previous experience that kidneys with late rejection have a poor prognosis (32,33).

Table 7.  Comparison of the estimated GFR (eGFR) in high GRIT1-AMAT1 biopsies versus other Banff diagnoses*
 High GRIT1-AMAT1Remaining TCMRRemaining C4d+ABMRRemaining borderlineRemaining nonrejecting
  1. *GFR values are mean ± SEM. Values in parentheses represent the number of biopsies. Statistical test: one way ANOVA with Dunnett Multiple Comparison Test. In each group the high GRIT1-AMAT1 cases have been removed. na = not available.

  2. 1p < 0.05.

  3. 2p < 0.0001.

  4. 3p < 0.01.

All biopsies
 eGFR@bx40 ± 3.8 (23)49 ± 3.5 (27)45 ± 4.2 (16)47 ± 3.1 (30) 43 ± 1.7 (125)
 eGFR 6 month55 ± 5.1 (19)60 ± 4.6 (26)43 ± 3.8 (15)54 ± 4.4 (28) 52 ± 2.2 (121)
 Delta eGFR13 ± 2.9 (19)10 ± 4.1 (26) −3.1 ± 2.41 (15) 6.1 ± 2.9 (28) 9.3 ± 1.9 (121)
Early biopsies (<12 month)
 eGFR@bx38 ± 4.6 (14)45 ± 5.0 (15)21 (1)45 ± 3.5 (15)39 ± 2.4 (53)
 eGFR 6 month47 ± 6.4 (12)62 ± 7.5 (14)na57 ± 4.4 (14)59 ± 3.2 (52)
 Delta eGFR 9 ± 3.7 (12)16 ± 7.2 (14)na11 ± 3.6 (14)21 ± 2.9 (52)
Late biopsies (>12 month)
 eGFR@bx43 ± 6.6 (9) 54 ± 5.0 (12)45 ± 4.2 (15)49 ± 5.0 (15)45 ± 2.0 (72)
 eGFR 6 month67 ± 7.0 (7) 57 ± 5.0 (12)43 ± 3.8 (15)51 ± 7.8 (14)47 ± 3.9 (69)
 Delta eGFR20 ± 2.9 (7) 3.1 ± 2.01 (12)  −3.1 ± 2.42 (15)  1.3 ± 4.23 (14) 1.0 ±.1.62 (69)

Excellent graft survival after biopsy in high GRIT1-AMAT1 cases

No death censored graft losses subsequent to the last biopsy have yet been recorded in 17 patients whose biopsy showed high GRIT1-AMAT1 (excluding three with other major diseases), despite v lesions and late rejections. In contrast, many losses were recorded in kidneys manifesting ABMR (either C4d positive or C4d negative) or glomerulonephritis (Figure 3).

imageimage

Figure 3. Death-censored kidney graft survival in relationship to high GRIT1-AMAT1 molecular criteria and major Banff diagnoses (A) or C4d-nedABMR and glomerulonephritis (GN, B). In panel B, C4d-negABMR and GN were excluded from the Banff nonrejecting diagnosis.

Patients with Banff TCMR but not having high GRIT1-AMAT1 also had a good prognosis. Only two kidneys failed: T29 and T39-T42. T29 subsequently developed C4d negative ABMR. The patient having biopsies T39 and T42 became nonadherent and discontinued immunosuppression after biopsy and was lost to follow-up, eventually returning with end stage kidney failure.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Because the Banff definition of TCMR relies on nonspecific lesions and arbitrary cutoffs, we used an independently derived molecular definition to identify biopsies with fully developed TCMR and to study their features and behaviors. Our definition was derived from mouse kidney allografts with severe TCMR and typical histologic lesions (26,34) (Famulski et al., in press), which develop high scores for GRIT1 very early then express high AMAT1 scores as tubulitis and arteritis lesions develop. The GRIT1-AMAT1 combination provided us with a definition of the cTCMR independent of knowledge of histology. The high cutoffs used for the definition ensured that every case was molecularly typical. (We did not define diagnostic cutoffs, which must await ongoing analyses, as outlined below.) The cTCMR biopsies were all heavily inflamed with the typical lesions of TCMR and expressed transcripts for effector T cells, constitutive macrophage markers and injury. The cTCMR pattern did not occur in C4d positive ABMR, and was rare in other biopsies. Paradoxically, kidneys with cTCMR had the most intense inflammation but the best prognosis in terms of recovery of function and survival, despite features previously considered high risks for loss—intimal arteritis (v scores) and late presentation, reflecting the striking sensitivity of TCMR to current therapy when uncomplicated by ABMR.

The fact that the cTCMR biopsies were the most severely inflamed but had the best prognosis invites us to reexamine the belief that inflammation per se is a bad prognostic factor in kidney biopsies. Biopsy studies have shown that inflammation is an adverse factor for outcome in late kidney biopsies with scarring (14,15). However, this probably reflects the characteristics of the underlying disease, rather than inflammation per se. The inflammation of TCMR is highly sensitive to conventional management after biopsy. The policy in our centers is to treat TCMR with steroids and to add ATG in cases that are steroid resistant or have v lesions, but the treatment is determined by the transplant nephrologist. In contrast, inflammation in scarred biopsies is probably the harbinger of a poor prognosis because it often reflects ongoing diseases that are resistant to treatment, particularly ABMR (27). Thus the significance of inflammation depends on the disease process. We find no evidence that TCMR leads to late deterioration, at least within the limits of this observation period, even after late rejection episodes or when v lesions are present.

Ironically, myeloid cells of the monocyte-macrophage-dendritic series and their transcripts may prove to be a more reliable feature of the molecular TCMR process than T-cell infiltration, particularly in cell-depleted patients with TCMR. As in mouse, both QCMATs and AMATs were strongly associated with cTCMR. The macrophages could be either a cause or an effect of the parenchymal deterioration, and possibly both. Macrophages and related dendritic cells are intrinsic to antigen presentation and immunoregulation and have many potential roles as effector cells. Macrophage infiltration and particularly alternative macrophage activation are strongly associated with tissue injury and remodeling (35–37), and thus participate in response to severe tissue deterioration.

The present data are consistent with the proposed model of TCMR as a cognate noncytotoxic interaction between a relatively small number of antigen-specific effector T cells and APCs, in which the antigen-specific interaction triggers the creation of an inflammatory compartment attracting noncognate T cells and macrophages. This in turn causes a stereotyped dedifferentiation of the renal epithelium, manifested first by the appearance of injury genes and the loss of kidney function molecules such as epithelial solute carriers (10). In rejecting mouse transplants macrophage infiltration and particularly the progressively increasing AMAT1 transcripts are strongly associated with tubulitis, and like tubulitis are later developments in the time course of TCMR than are IFNG effects or T-cell infiltration.

The ability to define cases with severe TCMR independent of histology allows us to evaluate the strengths and weaknesses of the existing histology-based definition of TCMR. The absence of TCMR lesions—tubulitis, infiltrate and intimal arteritis—virtually excluded the possibility of cTCMR as defined by molecules. In 151 biopsies for cause without t or i or v scores (and thus not labeled TCMR or borderline or mixed), only 3 (2%) had the cTCMR pattern, all with other inflammatory diseases (lupus, glomeronephritis and viral nephropathy). On the other hand, histology interpreted 4 of the 23 biopsies with cTCMR as borderline, which is unsatisfactory. Moreover, histopathology diagnosed TCMR or borderline changes in many cases that the PAM classifier (Figure 2) called nonrejection, as we reported previously (25), probably reflecting the inherently nonspecific aspect of the lesions.

The emerging molecular definition of cTCMR will allow us to revisit the Banff lesions and diagnostic labels such as ‘TCMR’ and ‘borderline’ and iteratively refine the diagnostic groups until the labels reflect the molecular and histologic realities and the underlying disease mechanisms, eliminating the category ‘borderline’. An early estimate, based on Figure 2 and our previous publication (25), indicates that biopsies with histologic lesions of TCMR and labeled either TCMR or borderline actually contains three states: cTCMR, milder TCMR below the cTCMR thresholds and nonTCMR (i.e. other states). As estimated from Figure 2, the 45 biopsies that Banff labels TCMR include 18 cTCMR (40%) cTCMR; 15 mild-TCMR (33%) and 12 non-TCMR that is 27% false positives. The false positives include treated TCMR; C4d negative ABMR and injury. We can further estimate that the 34 biopsies labeled borderline include 4 cTCMRs (12%), 9 mild-TCMR (26%), and 21 non-TCMR (62%). We believe that a new Banff classification of TCMR can emerge from such analyses to provide better support for treatment decisions. Elimination of the borderline category will be a step in that direction.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

The authors thank Dr. Zija Jacaj for help with collection of the clinical data; and Vido Ramassar, Anna Hutton and Kara Allanach for technical support. This research has been supported by funding and/or resources from Genome Canada, Genome Alberta, the University of Alberta, the University of Alberta Hospital Foundation, Alberta Health Services, Roche Molecular Systems, Hoffmann-La Roche Canada Ltd., Alberta Ministry of Advanced Education and Technology, the Roche Organ Transplant Research Foundation, the Kidney Foundation of Canada and Astellas Canada. Dr. Halloran also holds the Muttart Chair in Clinical Immunology.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Table S1. Quantitative constitutive macrophage-associated transcripts.

Table S2. Real-time RT-PCR analysis of the prototypical AMA markers, CD163 and MRC1 in biopsies with high GRIT1/AMAT1 score and in remaining Banff TCMR biopsies.

Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. any queries (other than missing material) should be directed to the corresponding author of the article.

FilenameFormatSizeDescription
AJT_3007_sm_TableS1.doc86KSupporting info item
AJT_3007_sm_TableS2.doc31KSupporting info item
AJT_3007_sm_Methods.doc35KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.