Inflammation Lesions in Kidney Transplant Biopsies: Association with Survival Is Due to the Underlying Diseases

Authors

  • J. Sellarés,

    1. Alberta Transplant Applied Genomics Centre, University of Alberta, Edmonton, AB, Canada
    2. Department of Medicine, Division of Nephrology and Transplant Immunology, University of Alberta, Edmonton, AB, Canada
    Search for more papers by this author
  • D. G. de Freitas,

    1. Alberta Transplant Applied Genomics Centre, University of Alberta, Edmonton, AB, Canada
    2. Department of Medicine, Division of Nephrology and Transplant Immunology, University of Alberta, Edmonton, AB, Canada
    Search for more papers by this author
  • M. Mengel,

    1. Alberta Transplant Applied Genomics Centre, University of Alberta, Edmonton, AB, Canada
    2. Department of Laboratory Medicine and Pathology, University of Alberta, Edmonton, AB, Canada
    Search for more papers by this author
  • B. Sis,

    1. Alberta Transplant Applied Genomics Centre, University of Alberta, Edmonton, AB, Canada
    2. Department of Laboratory Medicine and Pathology, University of Alberta, Edmonton, AB, Canada
    Search for more papers by this author
  • L. G. Hidalgo,

    1. Alberta Transplant Applied Genomics Centre, University of Alberta, Edmonton, AB, Canada
    2. Department of Medicine, Division of Nephrology and Transplant Immunology, University of Alberta, Edmonton, AB, Canada
    Search for more papers by this author
  • A. J. Matas,

    1. Department of Surgery, University of Minnesota, Minneapolis, MN
    Search for more papers by this author
  • B. Kaplan,

    1. Department of Medicine (Division of Nephrology), University of Arizona, Tucson, AZ
    Search for more papers by this author
  • P. F. Halloran

    Corresponding author
    1. Alberta Transplant Applied Genomics Centre, University of Alberta, Edmonton, AB, Canada
    2. Department of Medicine, Division of Nephrology and Transplant Immunology, University of Alberta, Edmonton, AB, Canada
      Corresponding author: Philip F. Halloran, Phil.halloran@ualberta.ca
    Search for more papers by this author

Corresponding author: Philip F. Halloran, Phil.halloran@ualberta.ca

Abstract

Assessment of kidney transplant biopsies relies on nonspecific inflammatory lesions: Interstitial infiltrates (i), tubulitis (t) and intimal arteritis (v). We studied the relationship between inflammation and prognosis in biopsies for clinical indications from 314 patients (median follow-up 25 months). We used a modified Banff classification, separately assessing inflammation (i-) in nonscarred (i-Banff), scarred (i-IFTA) and whole cortex (i-total), plus tubulitis and intimal arteritis. In early biopsies (<1 year), i- and t-lesions had no association with graft survival. In late (>1 year) biopsies, all i-scores correlated with progression to failure, due to the association of these infiltrates with progressive diseases: antibody-mediated rejection (ABMR) and glomerulonephritis. Tubulitis in nonscarred areas had no impact on survival. Severe tubulitis including scarred areas (tis3) was associated with worse survival, but reflected polyoma virus nephropathy or ABMR, not T-cell-mediated rejection. Intimal arteritis (v-lesions) had no association with allograft loss in early or late biopsies. In multivariate analysis, outcome was better predicted by the presence of progressive disease than by inflammation. Thus inflammation in late kidney transplants has no inherent prognostic impact, but predicts reduced survival because inflammation indicates actively progressing diseases. The most important predictor of outcome is the diagnosis of a progressive disease.

Abbreviations: 
ABMR

antibody-mediated rejection

ATN

acute tubular necrosis

BFC

biopsy for clinical indications

GN

glomerulonephritis

IFTA

interstitial fibrosis and tubular atrophy

TCMR

T-cell-mediated rejection

Introduction

Assessment of biopsies of troubled kidney transplants relies heavily on inflammatory lesions, which are the basis of diagnosing rejection and have implications for prognosis. The lesions used by the current criteria (1) for the diagnosis of T-cell-mediated rejection (TCMR) are interstitial inflammation (i-Banff score ≥2), tubulitis (t-score ≥2) and/or intimal arteritis (v-score > 0). We recently confirmed that these lesions are highly associated with TCMR by showing their presence in TCMR defined exclusively by molecular features (2). However, these lesions are inherently nonspecific: i-lesions and t-lesions are associated with various conditions in transplants and in native kidneys: acute tubular necrosis (ATN) (3), interstitial nephritis and polyoma virus nephropathy (PVN) (4), primary diseases such as glomerulonephritis (GN) (5,6), antibody-mediated rejection (ABMR) (7) and protocol biopsies with no known diseases (8,9). V-lesions are also ambiguous, since they can be induced either by TCMR or ABMR (10–12). However, the Banff consensus thresholds (1) permit us to use i-, t- and v-lesions to diagnose TCMR (13) when the prior probabilities are supportive.

In addition to diagnostics, some lesions have prognostic associations. Inflammation in areas of scarring (i-IFTA) correlates with poor allograft outcome (14). ‘Vascular rejection’ (i.e. with intimal arteritis) is reported to have a worse prognosis than ‘tubulo-interstitial rejection’, i.e. TCMR with no intimal arteritis (15,16). However, these conclusions must be revisited since ABMR was not excluded in these older studies. The i-total score (i.e. all inflammation in the cortex including both scarred and nonscarred areas) correlates better with the molecular phenotype than other histologic features do and predicts outcome, although it lacks diagnostic specificity (17). Our recent analyses have found that a molecular risk score is the strongest known predictor of outcome, but this leaves open the question of why some inflammatory lesions are associated with outcome (18).

The significance of inflammation lesions has been difficult to dissect because the disease processes were not well understood, especially in biopsies taken late (>1 year) posttransplantation. This led to the belief that mysterious and poorly characterized inflammation itself caused progression. Previous efforts to understand renal allograft deterioration were frustrated because ∼50% of ABMR was missed and attributed inappropriately to progressive scarring or calcineurin inhibitor toxicity (CNIT) (19). The recent identification of C4d negative ABMR (20–22) changes this situation, permitting a re-examination of earlier assumptions about i-, t- and v-lesions. Recent data-driven analyses indicate that ABMR and recurrent/de novo glomerular diseases explain the majority of renal allograft losses following biopsies for clinical indications (BFC) (21,23,24). Moreover, inflammation is a feature of progressive renal diseases and may be associated with survival because disease injured tissue evokes inflammation as part of the injury–repair response, analogous to attempted wound healing, rather than a mechanism of progression (25). The present study aimed to critically reassess the prognostic associations of interstitial infiltrates (i-scores: i-Banff, i-IFTA, i-total), tubulitis (t-scores) and intimal arteritis (v-scores) across a spectrum of well-defined disease entities in a BFC population.

Materials and Methods

Patients and sample collection

The study was approved by the University of Alberta Health Research Ethics Board (Issue # 5299), by the University of Illinois, Chicago, Office for the Protection of Research Subjects (protocol # 2006-0544) and by the University of Minnesota (protocol # 0606M87646). Written informed consent was obtained from all study patients. Patients were recruited from three different centers: the University of Alberta (n = 180), the University of Minnesota (n = 99) and the University of Illinois (n = 35). Renal transplant patients undergoing a transplant biopsy for clinical indication (deterioration in function, proteinuria and stable but impaired function) as standard of care between 09/2004 and 10/2009 were included. Indications for biopsies were comparable between centers. Biopsies were obtained under ultrasound guidance by spring-loaded needles (ASAP Automatic Biopsy, Microvasive, Watertown, MA, USA). Only the last biopsy available for each patient (n = 314) was included in the analysis. One patient whose graft failed due to fibrillary GN was not considered for this analysis since the tissue obtained from the biopsy was entirely submitted for electron microscopy only.

Histopathology

Paraffin sections were prepared according to Banff criteria. All samples were adequate, with the exception of seven biopsies lacking arteries, and were graded by the current classification (1). In addition, interstitial infiltrates were assessed using three different methods: i-Banff, i-IFTA and i-total scores. The i-Banff score was defined as the interstitial inflammation only in nonscarred cortex (excluding perivascular, nodular and subcapsular infiltrates), and was assessed as an absolute percentage of the unscarred compartment only, as per Banff convention (1). The i-IFTA score, which was defined as the interstitial inflammation only in scarred cortex (excluding perivascular, nodular and subcapsular infiltrates), and the i-total score, defined as all cortical inflammation, were assessed as an absolute the percentage of the whole cortex (17) (Figure 1). Consensus rules permit tubulitis in non- or mildly atrophic tubules to be scored in two ways: both excluding and including such tubules in scarred areas (26). Thus we did not score moderate-severely atrophic tubules but separately scored tubulitis in non- or mildly atrophic tubules excluding scarred areas (t-score) and including scarred areas (tis score). C4d staining was performed on frozen sections (available in 282/314 biopsies) using a monoclonal anti-C4d antibody (Quidel, San Diego, CA, USA) by indirect immunoflourescence. Diffuse linear C4d staining (i.e. in >50% of peritubular capillaries) was interpreted as positive. The criteria for C4d negative ABMR (designated by Banff as suspicious for ABMR (1, 27)) were based on our previous description (21): DSA positive, nondiffuse C4d staining (negative or minimal focal or focal) and any of the following microcirculation lesions: Peritubular capillaritis (ptc > 0) and/or glomerulitis (g > 0) and/or thromboses and/or transplant glomerulopathy (cg > 0) and no criteria for TCMR.

Figure 1.

Scoring of interstitial infiltrates in renal allograft biopsies. The extent of the unscarred and scarred compartments was assessed as absolute percentage in terms of 100% cortex present. The extent of infiltrates was semi-quantitatively assessed as absolute percentages of the whole cortex as 100% in i-IFTA and i-total, whereas the i-Banff was scored not considering scarred areas, i.e. nonscarred areas were set as 100%.

HLA antibody screening

Of 314 renal allograft biopsies, 271 had available serum at time of biopsy for HLA antibody testing. Antibody screening was performed using FlowPRA® beads. Antibody specificities of patient sera were determined by FlowPRA® specific class I and/or II and/or FlowPRA® single antigen I and II beads (One Lambda, Canoga Park, CA, USA). Further testing for specificities was done if the screen was positive (≥5% PRA) (10).

Data analysis

Data analyses were performed using Graph Pad Prism 5 statistical software and R version 2.10. Chi-square or Fisher's exact tests were used for comparing counts between groups. Continuous variables were correlated by Spearman correlation.

Graft failure was defined as return to dialysis. Death censored graft survival analysis was performed using the Kaplan–Meier method. Patients were censored either for the end of the study (December 2009), death with functioning graft (n = 15) or loss of follow-up (n = 35). Significance was assessed by the log rank test. The influence of histopathological lesions and progressive diseases on graft survival was analyzed by univariable and multivariable Cox regression.

Results

Study population and diagnostic classification

We analyzed 314 transplant BFC from 314 kidney transplant recipients, taken between 2004 and 2009, from 6 days to 32 years posttransplant (median 21 months) (Table 1). There were 129 early biopsies (<1 year posttransplant) and 185 late biopsies (>1 year posttransplant). The median time posttransplant of early biopsies was 1.8 months (6 days–10 months), and of late biopsies was 69 months (13–381 months).

Table 1.  Patient demographics
Patient demographics [n = 314]All biopsies (n = 314)Patients biopsied <12 months (n = 129)Patients biopsied >12 months (n = 185)p-Value (early vs. late)
Mean recipientage43.4 (5–76)48.4 (14–76)39.8 (5–73)<0.001
Recipient Gender (% male) [n = 314]196 (62%)82 (64%)114 (62%)0.726
Recipient ethnicity [n = 314]
 Caucasian200 (64%)76 (59%)124 (67%)0.141
 Black33 (11%)17 (13%)16 (9%)0.198
 Other81 (26%)36 (28%)45 (24%)0.475
Primary disease [n = 314]
 Diabetic nephropathy65 (21%)29 (22%)36 (19%)0.516
 Hypertension/large vessel disease29 (9%)18 (14%)11 (6%)0.016
 Glomerulonephritis/vasculitis97 (31%)29 (22%)68 (37%)0.007
 Interstitial nephritis/pyelonephritis19 (6%)9 (7%)10 (5%)0.566
 Polycystic kidney disease46 (15%)22 (17%)24 (13%)0.314
 Others45 (14%)17 (13%)28 (15%)0.626
 Unknown etiology13 (4%)5 (4%)8 (4%)0.844
Mean donor age40.5 (2–70)43.7 (13–70)38.2 (2–68)0.002
Donor gender (% male) [n = 314]120 (38%)46 (36%)74 (40%)0.436
Donor ethnicity
 Caucasian139 (44%)60 (47%)79 (43%)0.504
 Black2 (1%)1 (1%)1 (1%)1
 Other16 (5%)9 (7%)7 (4%)0.206
 Unknown157 (50%)59 (46%)98 (53%)0.183
Donor type (% deceased donor transplants) [n = 314]150 (48%)59 (46%)91 (49%)0.547
Number of transplants per patient
 Patients with one transplant256 (82%)108 (84%)148 (80%)0.403
 Patients with multiple transplants41 (13%)17 (13%)24 (13%)0.958
Median time from transplantation to biopsy (months)211.868.9
Median duration follow up post-biopsy (months)2527.623
Clinical characteristics at time of biopsy [n = 314]All biopsies (n = 314)Patients biopsied <12 months (n = 129)Patients biopsied >12 months (n = 185)p-Value (early vs. late)
Indication for biopsy
 Delayed graft function6 (2%)6 (5%)00.004
 Rapid deterioration of graft function77 (25%)24 (19%)53 (29%)0.042
 Slow deterioration of graft function114 (36%)45 (35%)69 (37%)0.662
 Stable impaired graft function53 (17%)32 (25%)21 (11%)0.002
 Investigate proteinuria32 (10%)11 (9%)21 (11%)0.416
 Follow-up from previous biopsy10 (3%)4 (3%)6 (3%)1
 Others7 (2%)1 (1%)6 (3%)0.247
 Indication unknown15 (5%)6 (5%)9 (5%)0.930
Maintenance immnnosuppressive regimens at biopsy
 MMF, Tacrolimus, steroid87 (28%)43 (33%)44 (24%)0.063
 MMF, Tacrolimus48 (15%)31 (24%)17 (9%)<0.001
 MMF, Cyclosporine, steroid51 (16%)21 (16%)30 (16%)0.988
 MMF, Cyclosporine31 (10%)16 (12%)15 (8%)0.209
 MMF, Steroids10 (3%)5 (4%)5 (3%)0.746
 Steroids, Tacrolirnus14 (4%)4 (3%)10 (5%)0.412
 Azathioprine, Cyclosporine, Steroids11 (4%)0 (0%)11 (6%)0.003
 Others62 (20%)9 (7%)53 (29%)<0.001

Biopsies were classified by a modified Banff schema incorporating C4d negative ABMR (21). Of 314 biopsies, 24 were diagnosed as C4d positive ABMR, 46 C4d negative ABMR, 1 mixed TCMR plus C4d positive ABMR (mixed), 20 TCMR, 46 borderline, 27 glomerulonephritis (GN), 6 polyoma virus nephropathy (PVN), 6 transplant glomerulopathy (TG), 17 ATN, 37 interstitial fibrosis and tubular atrophy (IFTA) not otherwise specified, 31 CNIT and 53 others (Table 2). The median duration of follow-up postbiopsy was 25 months. During the follow-up time, 72 out of 314 (23%) allografts failed. Fifteen (20.8%) patients died with functioning graft and 57 were graft failures. There were 7 failures after early biopsies and 50 failures after late biopsies. Of the early biopsies, two grafts failed at 8 and 10 months post-transplant due to PVN (one of these was attributed to PVN based on highly suggestive histological changes but not confirmed by immunohistochemistry or in situ hybridization), while the remainder all failed after the first year (median 5.1 years). Most progression to failure in the late biopsy group (34/50) occurred in those with progressive diseases diagnosed in the biopsy: ABMR and GN. This confirms our earlier observation that early BFC are not associated with increased risk of later graft failure (21), and that failure after late BFC is often due to ABMR or GN.

Table 2.  Histopathological diagnoses (progressive diseases are in bold)
Histopatbology diagnoses [n = 314]All biopsies (n= 314)Patients biopsied <12 months (n= 129)Patients biopsied >12 months (n= 185)Failures <12 months (n= 7)Failures >12 months (n= 50)
  1. 1Cases with transplant glomerulopathy that did not fulfill the criteria for C4d negative ABMR [n = 6]: n = 3 were PRA positive but no identified DSA and n = 3 were PRA negative.

  2. 2Others distribution [n = 53]: n = 14 minor changes, n = 12 normal, n = 5 C4d positive with no pathology associated, n = 4 diabetic nephropathy, n = 4 donor origin vascular disease, n = 4 acute tubular injury, n = 3 suspicious of GN, n = 2 suspicious of PVN, n = 1 suspicious of viral infection, n = 1 pyelonephritis, n = 1 oxalosis, n = 1 tubulo-interstitial nephritis, n = 1 posttransplant lymphoproliferative disorder.

C4d positive ABMR24 (8%)3 (2%)21 (11%)010
C4d negative ABMR46 (15%)3 (2%)43 (23%)015
Mixed TCMR plus ABMR1 (1%)01 (1%)01
TCMR20 (6%)15 (12%)5 (3%)22
Borderline46 (15%)23 (13%)23 (12%)05
Glomeıulonephritis27 (9%)4 (3%)23 (12%)08
Polyomavirus nephropathy6 (2%)4 (3%)2 (1%)10
Transplant glomerulopathy16 (2%)06 (3%)02
Acute tubular necrosis17 (5%)17 (13%)000
Calcineurin inhibitortoxicity37 (11%)17 (13%)20 (11%)03
Interstitial fibrosis and tubular atrophy NOS31 (10%)3 (2%)23 (15%)03
Others253 (17%)40 (31%)13 (7%)41
Total314 (100%)129 (100%)185 (100%)750

The types of GN that were associated with subsequent progression to failure were membranoproliferative GN (five failures out of nine cases), focal and segmental glomerulosclerosis (two failures out of two), and one kidney that failed with crescentic GN of unknown etiology. None of the kidneys with IgA nephropathy failed (n = 8).

Interstitial infiltrates and allograft survival

Three types of interstitial infiltrates were evaluated: i-Banff, i-IFTA and i-total. None of these inflammation scores was associated with decreased graft survival after early BFC: i-Banff (p = 0.66), i-IFTA (p = 0.97) and i-total (p = 0.23).

In late biopsies, all i-scores > 0 (i-Banff, i-IFTA and i-total) were associated with increased future graft loss. Grafts with i-Banff > 0 had worse outcome than those with no inflammation (i-Banff0) (p = 0.01), and the analysis of individual grades showed no differences between i-Banff1 and i-Banff > 1 (Figure 2A).

Figure 2.

Allograft survival and interstitial infiltrates. Kaplan–Meier curves for i-Banff (A), i-IFTA (B) and i-total scores (C) in late (n = 185) biopsies for clinical indications.

The i-IFTA score predicted worse outcome, with 82% allograft survival in i-IFTA0 and 51.3% in i-IFTA > 0 (p = 0.0001). Similarly, individual i-IFTA scores in late biopsies correlated with poor allograft survival (Figure 2B): 74% in i-IFTA1 (i-IFTA0 versus i-IFTA1: p = 0.04; i-IFTA1 versus i-IFTA ≥ 2: p = 0.03), 39% in i-IFTA ≥ 2 (i-IFTA0 versus i-IFTA ≥2: p < 0.0001).

The i-total score (ti > 0) (which includes i-Banff and i-IFTA) in late biopsies was also associated with worse graft survival: 59% in ti > 0 compared to 82% for noninflamed grafts (ti0) (p = 0.009). The risk increased with higher ti-scores (Figure 2C): 49% in ti ≥ 2 (ti0 versus ti ≥ 2: p = 0.001) and 67% in ti1 (ti1 versus ti ≥ 2: p = 0.03). Moreover, i-total and i-IFTA were associated with interstitial fibrosis (r = 0.7 and r = 0.85 respectively: p < 0.001), whereas i-Banff was not (p = 0.3). Thus i-total and i-IFTA represented a surrogate for scarring, and this accounts for much of the association of these scores with outcome.

Interstitial fibrosis and i-IFTA

We analyzed the impact of significant inflammation within scarred areas. We selected a subset of biopsies with interstitial fibrosis with no/minimal inflammation in the nonscarred area (i-Banff < 2). The i-IFTA score increased in direct proportion with the degree of interstitial fibrosis (ci) (Table 3), but the percentage of scarred area showing infiltrates was almost constant—60–70%.

Table 3.  Percentage of scarring with inflammation according to the extent of interstitial fibrosis in biopsies with minimal inflammation in the nonscarred compartment (i-Banff < 2) (n = 236)
Scarred biopsies with little inflammation in the nonscarred area
 
% Interstitial fibrosisnMean % i-IFTAMean % of scarred areas with inflammation1
  1. 1One-way ANOVA and Bonferroni's Multiple Comparison Test between the groups: p = NS.

1–5%432.4168.6
6–25%1089.3262.3
26–50%6027.172.5
50%2549.963.1

We divided these selected biopsies with scarring into those with ≥50% or <50% of the scar showing inflammation. We used two different definitions of scarring: >5% (ci > 0) of the biopsy (n = 196) (Figure 3A) or >25% (ci > 1) (n = 88) of the biopsy (Figure 3B). In both analyses the difference in survival between highly inflamed and less inflamed was not significant.

Figure 3.

Allograft survival and scarring with infiltrates. Kaplan–Meier curves for the different extent of scarring with inflammation in biopsies with i-Banff < 2 either in those with scarring >5% (ci > 0) (n = 196) (A) or >25% (ci > 1) (n = 88) (B) of interstitial fibrosis.

These analyses are limited by the fact that 90% of BFC with scarring had some i-IFTA (236 i-IFTA > 0%/262 interstitial fibrosis >0%). Thus high degrees of scarring without inflammation were uncommon. Poor survival was associated with inflamed scars, but the association between the degree of scarring and the degree of inflammation was too tight to permit us to define an independent impact of either inflammation or scarring in this BFC cohort.

Association of interstitial infiltrates with diseases and failures

As shown in Table 4, of the 50 kidneys that progressed to failure after late BFC, 42 (84%) had i-total, 40 (80%) had i-IFTA and 17 (34%) had i-Banff. Of the 42 kidneys that failed after a late BFC with i-total > 0, 30 had identifiable progressive diseases: ABMR 21 (50%), mixed 1 (3%) or GN 8 (19%). Thus graft losses after BFCs showing i-total > 0 were mainly due to ABMR, mixed or GN.

Table 4.  Distribution of individual interstitial infiltrates grades according to histopathological diagnosis and allograft failure in late biopsies
DiagnosesLate biopsies (>1 yr)
ni-Banff0i-Banff>0i-total0i-total>0i-IFTA0i-IFTA>0
  1. Shaded area: Progressive diseases with inflammation: Antibody-mediated rejection, mixed TCMR plus ABMR, glomerulonephritis. The number of failures is shown in parenthesis ().

  2. 1Late TCMR failures n = 2: 1 medical causes, 1 missing data: TCMR plus IFTA grade 2 plus DSA. No response to treatment. Failure 8 months after the biopsy.

  3. 2Late borderline failures n = 3: 2 nonadherent, 1 PVN.

  4. 3Late others distribution: n = 5 minor changes, n = 3 diabetic nephropathy, n = 2 normal histology, n = 1 tubulo-interstitial nephritis, n = 1 posttransplant lymphoproliferative disorder, n = 1 suggestive of PVN (failed).

C4d positive ABMR21 (10)11 (5)10 (5)219 (10)3 (1)18 (9)
C4d negative ABMR43 (15)34 (11)9 (4)11 (4)32 (11)17 (5)26 (10)
MiKedTCMR plus ABMR1 (1)01 (1)01 (1)01 (1)
TCMR5 (2)11 (1)4 (1)05 (2)23 (2)
Borderline23 (5)214 (4)9 (1)4 (2)19 (3)7 (2)16 (3)
Glomerulonephritis23 (8)18 (4)5 (4)320 (8)617 (8)
Polyoma virus nephropathy2201111
Transplant glomerulopathy6 (2)6 (2)006 (2)06 (2)
Calcineurin inhibitor toxicity28 (3)28 (3)016 (1)12 (2)19 (1)9 (2)
interstitial fibrosis and tubular atrophy NOS20 (3)20 (3)010 (1)10 (2)12 (1)8 (2)
Others133 (1)85 (1)76 (1)85 (1)
Total185 (50)142 (33)43 (17)54 (8)131 (42)75 (10)110 (40)
% of failures27%23.2%39.5%18.4%32%13.3%36.3%

Late biopsies showing little inflammation had lower rates of subsequent progression to failure: 8/54 with i-total = 0 failed, and 10/75 with i-IFTA = 0 failed. These biopsies also had a low incidence of progressive diseases. For example, only four ABMRs had i-total = 0 and six ABMRs had i-IFTA = 0.

Although i-Banff was less common than i-IFTA, such kidneys usually had progressive diseases: 14 of 17 failures (76.5%) after late biopsies with i-Banff > 0 had progressive diseases: 9 ABMR, 4 GN and 1 mixed.

Thus in late biopsies i-total and i-IFTA scores were associated with increased risk of failure because of the association of these lesions with active disease progression, mainly ABMR and GN. Although not associated with scarring, i-Banff score was also associated with ABMR and GN. Thus the impact of inflammation lesions in late biopsies on outcome is largely due to the tendency of progressive diseases to display both inflamed scarred areas and inflammation in unscarred areas.

Tubulitis per se has no effect but severe tubulitis (tis3) scored including scarred areas is associated with poor prognosis

Tubulitis is a nonspecific lesion, seen in TCMR but also common in other types of renal injury. Previously we have demonstrated that tubulitis was not related to poor graft outcomes in BFC (21). We confirmed this in the present study when tubulitis was scored excluding scarred areas, both in early (Figure 4A) and in late biopsies (Figure 4B).

Figure 4.

Allograft survival and tubulitis. Kaplan–Meier curves for the individual grades of tubulitis scored in nonscarred areas (t) in early (n = 129) (A) and late biopsies (n = 185) (B) and for the individual grades of tubulitis scored in scarred and nonscarred areas (tis) in early (n = 129) (C) and late biopsies (n = 185) (D).

When tubulitis was scored including scarred areas, low and intermediate grades (tis = 1, tis = 2) had no survival effect compared to tis = 0. Severe tubulitis including scarred areas (tis3), although uncommon, was associated with more subsequent failure in early (Figure 4C) and late biopsies (Figure 4D).

We compared biopsies with tis3 that subsequently failed (F, n = 10) to those that did not fail (W, n = 9) (Table 5). Failures were attributed to diseases by reviewing histopathologic and clinical information as discussed elsewhere (manuscript in preparation). The 9 cases with tis = 3 that did not fail (6 TCMR, 3 borderline) lacked DSA and PVN. The 10 kidneys with tis = 3 that failed had evidence for conditions that increased their risk of failure, usually ABMR or PVN. Four failed due to ABMR; three were suspicious for ABMR due to glomerulitis plus peritubular capillaritis in the biopsy (F2, F1, F4), and one (F5) had DSA at biopsy and was nonadherent. Three kidneys with tis = 3 that progressed to failure had either PVN (F10) or probable PVN (F8 and F9) based on histology but not confirmed by immunohistochemistry or in situ hybridization. Three were diagnosed as TCMR (F3, F6 and F7) but had evidence for ABMR: two (F3, F6) had DSA at biopsy and one (F7) was highly suspicious for ABMR, having TG and PRA but no identified DSA.

Table 5.  Histological diagnoses and clinical aspects of the failed (F, n = 10) and nonfailed (W, n = 9) cases with severe tubulitis (tis3)
Graft statusBiopsyRevised histological diagnosisTime posttransplantAntibody statusAttribution of failures
  1. 1Excellent response to antirejection treatment at the time of biopsy. At that time it had DSA and proteinuria. Failure one year after the biopsy.

  2. 2At the time of biopsy had TCMR plus IFTA grade 2 plus DSA. No response to treatment. Failure 8 months after the biopsy.

Working grafts n = 9W1TCMREarlyPRA neg
W2TCMREarlyPRA neg
W3TCMREarlyPRA neg
W4TCMREarlyPRA neg
W5BorderlineLatePRA neg
VV6TCMRLatePRA neg
W7TCMREarlyNDSA
W8BorderlineLateNDSA
W9BorderlineLateNDSA
Failed grafts n = 10FlC4d neg ABMREarlyDSAABMR
F2MixedEarlyDSAABMR
F3TCMREarlyDSAMissing data1
F4C4d pos ABMRLateDSANonadherent: probable ABMR
F5BorderlineLateDSANonadherent: probable ABMR
F6TCMRLateDSAMissing data2
F7TCMRLateNDSAMedical causes
F8OtherLateNot donePVN
F9OtherEarlyNot donePVN
F10PVNEarlyPRA negPVN

Thus tubulitis has little impact overall but in a few cases a previously unrecognized feature—tis = 3 lesions including scarred areas—was associated with increased progression to failure due to diseases other than TCMR.

Intimal arteritis is not associated with increased risk of graft failure

Biopsies lacking arteries (n = 7) were excluded. Table 6 shows that v-lesions were uncommon, being present in 19 grafts (14% of all rejecting grafts): 11 TCMR and 8 ABMR (5 C4d positive and 3 C4d negative). Thirteen were v1, five v2 and one v3 with transmural arteritis and foci of fibrinoid necrosis. Whether analyzed as v0 versus v > 0, or as individual grades of intimal arteritis, v-lesions had no correlation with subsequent risk of failure (Figures 5A and B).

Table 6.  Histopathological diagnosis, individual v-score grade distribution, number of failures and attribution on patients whose grafts had intimal arteritis
Histological diagnosisnv1v2v3# FailuresAttribution of failures
  1. The number of failures in each individual score is shown in parenthesis ().

C4d positive ABMR54 (2)102ABMR (×2)
C4d negative ABMR31 (1)201ABMR
TCMR118 (1)2 (1)12Medical causes, Nonadherent (probable ABMR)
Total1913 (4)5 (1)15 
Figure 5.

Allograft survival and v-score. Kaplan–Meier curves for the v-scores in all biopsies with arteries (n = 307). Either v0 versus v > 1 (A) or comparing the individual v-scores (B).

The lack of influence of v-lesions on survival was not necessarily dependent on aggressive treatment. Despite the presence of v-lesions in their biopsy, 9/19 cases were treated only with steroids and all but one of these was TCMR; two patients received no treatment. Treatment choices may often have been made prior to identification of the v-lesions.

Of five failures after BFC showed v > 0, three had ABMR. One of these did not receive treatment, and two were treated with steroids plus IVIG. The other two failures occurred in biopsies diagnosed as TCMR but both failed with evidence suggesting ABMR: one patient was nonadherent and the graft failed one year later with DSA, and the other was suspicious for ABMR (TG and PRA without identifiable donor specificity) and failed in the context of a severe medical illness.

While intimal arteritis did not confer an identifiable risk of graft failure, the limited number of cases (19 in 314 biopsies) limits the strength of these conclusions.

Multivariate assessment of relative impact of inflammation versus progressive diseases on allograft loss

We assessed the association of inflammation lesions, scarring and the diagnosis of progressive diseases with graft survival by univariable and multivariable analysis (Table 7). The i-total was not included in the model because it is essentially represented by i-Banff and i-IFTA separately. Progressive diseases were assigned according to histopathological diagnoses and included GN, ABMR and mixed. Given the rarity of graft loss after early biopsies, we limited this analysis to late biopsies (n = 185).

Table 7.  Multivariate analysis
Late biopsies (> 1 yr) [n = 135]
 
FeatureHazard ratio (95%Cl)p-Value
  1. Progressive diseases: Antibody-mediated rejection plus mixed plus glomerulonephritis; t = tubulitis scored in nonscarred areas; tis = tubulitis scored in both scarred and nonscarred areas v = endothelialitis; ci = interstitial fibrosis.

Univariable analysis
 i-IFTA1.8 (1.4–2.4)<0.001
 i-Banff1.7 (1.1–2.6)0.01
 tis1.6 (1.2–2.0)0.001
 t1.21 (0.9–1.7)0.25
 v1.2 (0.6–2.5)0.55
 ci1.9 (1.4–2.7)<0.001
 Progressive diseases2.6 (1.4–4.6)0.002
Multivariable analysis
 Progressive diseases2.05 (1.1–3.9)0.02

In the univariate analysis all features evaluated were associated with graft loss, with the exception of intimal arteritis and tubulitis scored in nonscarred areas. In the multivariate analyses only the presence of progressive diseases remained an independent predictor of graft loss.

Discussion

In 314 patients undergoing biopsies, we analyzed the impact of the inflammatory lesions, interstitial inflammation, tubulitis and intimal arteritis on the probability of future failure. Biopsies were analyzed separately according to time posttransplant as early (<1 year) or late, in order to identify those conditions that operate in each time period, as they represent different populations with different prognosis (21). In early biopsies no lesions were associated with future failure, because they were due to treatable diseases or self-limiting conditions. However in late BFC, interstitial inflammation, and severe tubulitis scored in both scarred and nonscarred areas, was associated with reduced survival, due to the occurrence of these lesions in progressive diseases like ABMR or GN. Intimal arteritis had no impact on graft survival, and T cell-mediated chronic arteriopathy was never observed in this cohort. In multivariate analyses the only significant predictor of graft loss was the presence of progressive diseases. Thus the impact of inflammation on survival reflects the association of progressing diseases with inflammation, both in the scarred areas and in unscarred areas, and is not due to inflammation per se. We propose that this is not because inflammation causes progression, but because progressive diseases evoke inflammation as part of the injury–repair response, similar to wound healing, i.e. severe nephron injury evokes an inflammatory reaction as the nephron and the stroma remodel and attempt to heal. The design of this study did not intend to investigate the effects of antirejection treatment and to answer this question a prospective trial would be required.

The current Banff criteria for diagnosing TCMR are not working in late biopsies. The inflammation lesions used as diagnostic features of TCMR need to be revisited now that many kidney transplants present for late BFCs with complex scarring patterns. The Banff criteria for TCMR were generated when early TCMR was common—the prior probability was high—and most biopsies showed little scarring (28). In such biopsies, interstitial inflammation (i-Banff i-total), tubulitis and intimal arteritis work well for defining probable TCMR, but the differential diagnosis was simple: the prior probability of TCMR was high and PVN, ABMR and GN were uncommon. The situation now in transplant diagnostics has changed. Indeed, once severe scarring and atrophy are present, the diagnosis of TCMR becomes nearly impossible because the unscarred compartment is too small, and the inflammation due to the injury–repair response is extensive. While some may argue that this means that TCMR is universal, the Banff process does not accept this opinion, particularly since similar inflammation including tubulitis occurs in primary renal diseases in native kidneys. New rules for diagnosing late TCMR in kidneys with extensive scarring are needed, especially since this is critical for managing these patients.

In late biopsies, the association of inflammation in scarred and unscarred areas with progressing diseases is probably due to the inflammatory reaction to nephron loss, and explains why inflammation in late biopsies correlates with inferior outcomes (14). In the present BFC population, scarring was almost always inflamed, making it impossible to separate the impact of scarring from the impact of inflammation in the scarred areas. Both i-IFTA and i-total are associated with increased risk of graft loss because they are strongly influenced by the degree of atrophy–scarring, i.e. irreversible nephron loss (17). In one recent study, inflammation seemed to add to the risk of scarring alone (14,29). We find that the amount of inflammation increases as the scarring increases, making dissociation of these processes unreliable. The most important variable was not inflammation but progressing diseases, for which inflammation and scarring serve as markers. The i-Banff score is per definition not associated with scarring but was associated with poor graft survival, supporting the idea that progressive nephron injury induces inflammation in the unscarred as well as the scarred areas. This is also consistent with our understanding of progressive primary kidney diseases, where inflammation often accompanies disease activity (5,6).

A surprising new finding is that severe tubulitis scored including non- or mildly atrophic tubules in scarred and nonscarred areas (tis) is associated with poor outcome, highlighting the importance of attention to detail when scoring tubulitis (21). Differences in such details probably explain why studies relating severe tubulitis to graft survival find conflicting results, ranging from no association (30–32) to a strong association (33). According to consensus (26), tubulitis should only be scored in non- or mildly atrophic tubules. However, it is not clear whether mildly atrophic tubules at the interface between IFTA and non-IFTA areas should be included. Some pathologists ignore tubules in areas of interstitial fibrosis, while others grade tubulitis in areas of transition between ‘normal’ to ‘IFTA’. The tis = 3 lesions identified in the present study were often in mildly atrophic tubules in scarred areas. The biopsies with tis = 3 lesions which fail were apparently due to entities other than TCMR: PVN, rejection in the presence of DSA and nonadherence. An analysis of the impact of details of scoring of tubulitis and interstitial inflammation on diagnosis will be discussed elsewhere (Sis and Mengel, manuscript in preparation).

Intimal arteritis conferred no additional risk when it occurred in TCMR, and probably only impacts on survival through ABMR. The impact of intimal arteritis is difficult to interpret because of the under-recognition of ABMR in the past, making discussions of ‘vascular rejection’ largely meaningless. V-lesions, including intimal and transmural arteritis and fibrinoid necrosis have often been associated with worse outcomes (34–36). While early studies often showed worse survival in kidneys with intimal arteritis (15,16,37–39)—especially v3 cases with fibrinoid necrosis—(32,40), few studies used C4d staining or sensitive DSA testing (10–12). The lack of association between intimal arteritis and survival in the present study probably reflects the fact that many cases with TCMR and intimal arteritis (v-lesions) were treated effectively. Our data also question whether the presence of v-lesions justifies automatic use of heavy immunosuppression (e.g. rabbit anti-thymocyte globulin treatment) in TCMR cases: some were successfully treated with steroids alone, and the failures (5/19) were attributed to ABMR or nonadherence, not uncomplicated TCMR. Because v-lesions were uncommon, even in kidneys that progressed to failure after BFC, the relatively small number of biopsies with v-lesions limits the power of our study, and makes us cautious about our conclusions. However, the low incidence of v-lesions in this large cohort of BFC is an interesting finding per se, reflecting the infrequency of arteritis in the current era, probably due to improved immunosuppression. While further studies are needed, we suggest that the prognostic significance of intimal arteritis has been overestimated if ABMR is excluded (2). This is consistent with recent clinical trials, where early TCMR with v-lesions had little measurable impact on graft survival compared to rejection episodes lacking v-lesions (41).

Thus the inflammatory lesions are associated with poor outcomes in the late BFC population because of their association with the real risk: the presence of untreatable ongoing diseases like ABMR and GN. The present findings do not support the existence of mysterious ‘chronic inflammation’ that drives deterioration independent of disease. The lesions themselves should suggest a search for an ongoing disease causing nephron loss, which should create a new understanding of the significance of i-, t- and v-lesions, and of allograft prognosis.

Acknowledgments

The authors thank Dr. Zija Jacaj for help with collection of the clinical data, and Kara Allanach, Vido Ramassar and Anna Hutton for technical support.

Funding: 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 Advanced Education and Technology, Roche Molecular Systems, Hoffmann-La Roche Canada Ltd., the Alberta Ministry of Advanced Education and Technology, the Roche Organ Transplantation Research Foundation, the Kidney Foundation of Canada and Astellas Canada. Dr. Halloran also holds a Canada Research Chair in Transplant Immunology and the Muttart Chair in Clinical Immunology. The authors have no competing financial interests.

Disclosure

The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.

Ancillary