The Banff Conferences on renal allograft pathology – the latest 2013 report

Abstract

The Banff Conference diagnostic categories and their criteria for renal biopsy interpretation were created in 1991 by a group comprising nephrologists, pathologists, transplant surgeons and immunologists. These guidelines are widely used in many countries. Every two years, participants at these meetings present and discuss research findings that have added to our knowledge of allograft pathology. From the start, features of T-cell mediated rejection were established. This was followed by discovery of C4d staining in biopsy tissue and better characterisation of antibody mediated rejection. The formation of working groups to look into problematic areas has allowed better refinements to be made to the classification scheme. The latest Banff 2013 report is significant for the inclusion of a C4d-negative category under humoral rejection. Together with the realisation that endarteritis may be antibody mediated, this latest report will greatly impact how pathologists interpret the allograft biopsy.

Introduction – evolution of a classification system

The Banff Conferences on allograft pathology are biennial events that began in 1991 in Canada. The impetus to standardise renal allograft biopsy interpretation came from consensus grading systems for heart and lung transplants around the same time.^1,2 Six diagnostic categories established at the first meeting were: ‘normal’, ‘hyperacute rejection’, ‘borderline changes’, ‘acute rejection grades I to III’, ‘chronic allograft nephropathy/chronic rejection grades I to III’ and ‘other’. This laid the foundation for uniformity in reporting by pathologists and helped guide therapy as well as future research and clinical trials. ³ The role of molecular pathology was presented at the second conference. At the third meeting, Banff lesions were harmonised with the chronic allograft damage index (CADI) scoring schema. Efforts to remove glomerulitis (g lesion) from the classification were unsuccessful in the 1995 meeting, ⁴ which proved fortuitous as glomerulitis, together with peritubular capillaritis (ptc), is now regarded as one of the key histological features of microvascular inflammation (MI) supporting the presence of antibody mediated rejection (ABMR). ⁵

By the mid-1990s, the Banff system had gained acceptance by most physicians and pathologists, with a number of studies confirming its reproducibility and validity.^{6 –8} The 1997 meeting saw the incorporation of the United States National Institutes of Health Cooperative Clinical Trials in Transplantation (CCTT) criteria into the Banff system, creating the so called Banff 97 diagnostic categories (Table 1). ⁹ The classification schema thus emerged formed the cornerstone upon which subsequent modifications could be made. During this period, histological changes of ABMR were increasingly being recognised and the use of C4d immunohistochemistry on allografts also became available. ¹⁰ In 2003, criteria for diagnosing ABMR were incorporated into the Banff system, ¹¹ after minor modifications of the original proposal by Mauiyyedi et al. published a year earlier. ¹² These criteria were:

1. Histologic evidence of acute tissue injury such as (a) acute tubular injury, or (b) neutrophils or mononuclear cells in peritubular capillaries or glomeruli, and/or capillary thrombosis, or (c) fibrinoid necrosis/intramural or transmural arteritis;

2. Immunopathological evidence of antibody interaction with tissue, such as (a) C4d and/or immunoglobulin in peritubular capillaries (uncommon), or (b) immunoglobulin and complement in arterial fibrinoid necrosis;

3. Serologic evidence of circulating antibodies to donor endothelial antigens.

Table 1.

Banff 97 diagnostic categories for renal allograft biopsies.

1. Normal
2. Antibody mediated rejection  A. Immediate (hyperacute)  B. Delayed (accelerated acute)
3. Borderline changes ‘suspicious’ for acute rejection	Foci of mild tubulitis (1–4 mononuclear cells/tubular cross section) and interstitial inflammation (10–25% of parenchyma affected)
4. Acute/active rejection  Type IA  Type IB  Type IIA  Type IIB  Type III	Significant interstitial inflammation (>25% of parenchyma affected) and foci of moderate tubulitis (>4 mononuclear cells/tubular cross section or group of 10 tubular cells)Significant interstitial inflammation and foci of severe tubulitis (>10 mononuclear cells/tubular cross section or group of 10 tubular cells)Mild to moderate intimal arteritis (v1)Severe intimal arteritis comprising >25% of luminal area (v2)Transmural arteritis and/or arterial fibrinoid change and necrosis of medial smooth muscle cells (v3) with accompanying lymphocytic inflammation
5. Chronic/sclerosing allograft nephropathy  Grade I  Grade II  Grade III	Mild interstitial fibrosis and tubular atrophy without (a) or with (b) specific changes suggesting chronic rejectionModerate interstitial fibrosis and tubular atrophy (a) or (b)Severe interstitial fibrosis and tubular atrophy and tubular loss (a) or (b)
Changes unrelated to rejection

All three of the above numbered features must be present for diagnosis but a case is considered ‘suspicious’ for acute ABMR if only two of the three criteria were met. On occasions only one criterion is fulfilled, most commonly that of acute tissue injury. In this situation, other non-immunologically mediated processes causing acute tubular injury, MI and/or vascular damage must be carefully excluded by clinical correlation. Rarely, standalone positive C4d in peritubular capillaries or serological detection of anti-donor antibodies is seen. This requires repeat evaluation of the tests to ensure that false positive results have not occurred, especially if C4d by immunohistochemistry has produced high or non-specific background staining.

Back in 1993, the Banff classification had also adopted the term ‘chronic allograft nephropathy’ (CAN) to include four entities not always distinguishable from one another. They were chronic rejection, chronic calcineurin inhibitor toxicity, hypertensive vascular disease and chronic infection and/or reflux nephropathy. ³ However, ‘CAN’ soon became an umbrella or wastebasket term that is often used rather loosely by pathologists, lacking specificity in itself. Subsequent meetings in 2005 and 2007 agreed by consensus to abolish this term and replace it with more precise diagnostic categories or descriptive phrases like ‘interstitial fibrosis and tubular atrophy’.^13,14 Currently, ‘CAN’ is used only when biopsy findings are entirely non-specific, though personally I have avoided this term altogether. In the 2007 meeting, findings on the role of protocol biopsies were presented by a number of investigators. Scoring of ptc and C4d staining were introduced and published. The latter, based on the percentage of stained tissue or five high-power fields, is scored as: C4d0 (negative by immunofluorescence and immunohistochemistry), C4d1 (<10%, minimal staining), C4d2 (10–50%, focal staining) and C4d3 (>50%, diffuse staining). C4d1 by immunofluorescence is interpreted as negative. Considered to be of unknown significance in the 2007 report, C4d1 on immunohistochemistry is now interpreted as positive (Table 2) as this detection technique is known to be less sensitive than direct immunofluorescence. A new lesion termed ‘total interstitial inflammation’ or ‘ti’ was added to quantify the severity of interstitial mononuclear cell infiltration including areas with tubular atrophy/interstitial fibrosis, subcapsular cortex and perivascular cortex, using the same semi-quantitative criteria as the Banff 97 ‘i’ score. An alternative quantitative scoring for arteriolar hyalinosis, which may indicate calcineurin inhibitor arteriolopathy, was proposed but not incorporated into the classification. ¹⁴ Another major improvement to the Banff conferences since 2009 was the formation of international working groups, tasked to address problematic issues and validate research findings. Some of these issues include quality assurance in histology and immunohistochemistry, ABMR diagnosis independent of C4d in the kidney, isolated endarteritis without tubulointerstitial inflammation (isolated ‘v’ lesion), a separate new classification system for BK polyomavirus nephropathy and inclusion of molecular pathology into diagnostic categories.^15,16

Table 2.

Revised (Banff 2013) classification of antibody-mediated rejection in renal allografts.

Acute/Active ABMR; all three features must be present for diagnosis

1. Histologic evidence of acute tissue injury, including one or more of the following: Microvascular inflammation (g > 0 and/or ptc > 0).Intimal or transmural arteritis (v > 0).Acute thrombotic microangiopathy, in the absence of any other cause.Acute tubular injury, in the absence of any other apparent cause.2. Evidence of current/recent antibody interaction with vascular endothelium, including at least one of the following: Linear C4d staining in peritubular capillaries (C4d2 or C4d3 by IF on frozen sections, or C4d > 0 by IHC on paraffin sections).At least moderate microvascular inflammation (g + ptc ⩾ 2).Increased expression of gene transcripts in the biopsy tissue indicative of endothelial injury, if thoroughly validated.3. Serologic evidence of DSAs (HLA or other antigens).

Chronic, active ABMR; all three features must be present for diagnosis

1. Morphologic evidence of chronic tissue injury, including one or more of the following: Transplant glomerulopathy (cg > 0), if no evidence of chronic thrombotic microangiopathy.Severe peritubular capillary basement membrane multilayering (requires EM).Arterial intimal fibrosis of new onset, excluding other causes.2. Evidence of current/recent antibody interaction with vascular endothelium, including at least one of the following: Linear C4d staining in peritubular capillaries (C4d2 or C4d3 by IF on frozen sections, or C4d > 0 by IHC on paraffin sections).At least moderate microvascular inflammation (g + ptc ⩾ 2).Increased expression of gene transcripts in the biopsy tissue indicative of endothelial injury, if thoroughly validated.3. Serologic evidence of DSAs (HLA or other antigens).

C4d staining without evidence of rejection; all three features must be present for diagnosis

1. Linear C4d staining in peritubular capillaries (C4d2 or C4d3 by IF on frozen sections, or C4d > 0 by IHC on paraffin sections).2. g = 0, ptc = 0, cg = 0 (by light microscopy and by EM if available), v = 0; no TMA, no peritubular capillary basement membrane multilayering, no acute tubular injury (in the absence of another apparent cause for this).3. No acute cell mediated rejection (Banff 97 type IA or greater) or borderline changes.

g: glomerulitis; ptc: peritubular capillaritis; IF: immunofluorescence; IHC: immunohistochemistry; EM: electron microscopy; TMA: thrombotic microangiopathy.

As mentioned above, one of the major highlights of the 11^th Banff Conference was the recognition of C4d-negative ABMR. ¹⁶ Two major phenotypes of ABMR were identified: (1) ABMR developing in the pre-sensitised patient early post transplantation, and (2) ABMR following de novo formation of donor specific antibodies (DSAs) late in the transplant course, mostly due to noncompliance. The second phenotype is a significant contributor to late kidney allograft loss. ¹⁷ Participants from Paris presented data showing that the risk of progression to chronic ABMR in pre-sensitised patients is driven primarily by MI and DSA. The same groups also furnished evidence supporting the association between arterial sclerosis and ABMR or DSA, which is independent of other risk factors such as hypertension and aging.^18,19 Limitations of C4d as a diagnostic marker were discussed. C4d has limited sensitivity due to differences in methodology and inherent biological factors like non-complement fixing antibodies, vagaries of C4d deposition in renal tissue or loss of peritubular capillaries in scarred parenchyma. Conversely, C4d can be positive in ABO-incompatible allografts (Figure 1) and those treated with eculizumab without graft dysfunction or histologic evidence of MI.

Figure 1.

Diffuse positive staining for C4d in peritubular capillaries. This patient received an ABO-incompatible kidney transplant. (Anti-C4d immunoperoxidase, original magnification ×200)

The Banff 2013 meeting report – its impact on how the biopsy is interpreted

The 12^th Banff Conference on Allograft Pathology was held in Comandatuba, Brazil, from 19 to 23 August 2013. It was attended by more than 200 transplant pathologists, clinicians, immunologists and researchers from five continents. A major focus of the conference was presentation of findings of organ-specific Banff Working Groups (BWGs) initially formed at the 2009 and 2011 Banff meetings. These findings have resulted in new or modified criteria and specific recommendations for the diagnosis and reporting of a number of lesions, including ABMR and T-cell mediated rejection (TCMR) in renal and other solid organ allografts. The conference was preceded by a two-day Latin American Symposium on Transplant Immunobiology and Immunopathology. ²⁰

One of the findings of the BWGs was that isolated ‘v’ (intimal/transmural arteritis or mural fibrinoid necrosis) lesions show comparable response to treatment and graft survival as ‘v’ lesions with tubulointerstitial inflammation. As such, most isolated ‘v’ lesions should be reported as type II or type III acute TCMR. However, arterial inflammation may also represent acute ABMR or mixed ABMR/TCMR, the latter more common than pure ABMR. ²¹ Thus intimal arteritis or endarteritis is now regarded as a histologic criterion for ABMR (Table 2 and Figure 2). This is an important development in the Banff classification scheme because intimal arteritis or arteriolitis (v1) has all along implied a T-cell mediated process, and reported by pathologists as such since the 1^st Banff meeting. ²² Intuitively, one could reason that a less severe manifestation of vascular inflammation may share a similar immunologic aetiology as higher grade lesions along the same spectrum. Based on studies by Lefaucheur et al., ²¹ intimal arteritis with positive DSA is uncommon in the absence of glomerulitis and/or ptc. Similarly, arterial intimal fibrosis is accepted as a feature of chronic ABMR in this meeting.

Figure 2.

Intimal arteritis: mononuclear cells beneath the endothelial lining of an artery. (Periodic acid-Schiff (PAS), original magnification ×400).

Definitions and thresholds for transplant glomerulitis (g) and transplant glomerulopathy (cg) were revised to improve inter-observer agreement in scoring. Changes made were based on two independent sets of 30 and 17 biopsies, respectively. Cases were scanned and scored as virtual slides by a group of pathologists who were blinded to clinical and serological data and C4d results. Only those definitions or thresholds showing the highest kappa scores were used for the second slide circulation. For glomerulitis, complete or partial occlusion of ⩾1 glomerular capillary by leukocyte infiltration and endothelial cell enlargement is the new definition that gives the best inter-observer agreement as well as correlation with C4d scores and gene transcript profiles (Figure 3). Glomerulitis severity (g) score is the same as the original Banff 97 proposal, ⁹ determined by the percentage of involved glomeruli rather than the number of inflammatory cells per glomerulus, as suggested by Batal et al. ²³ It is noteworthy that congested glomeruli often contain circulating leukocytes which should be ignored. Strict adherence to this definition will prevent over-diagnosing glomerulitis or MI in the biopsy. Nevertheless, inter-observer variability remains an issue (kappa = 0.31 in the second slide circulation for g score). The usefulness of adding CD68 staining to assess glomerulitis remains unresolved but unpublished data from Willicombe et al., presented at the 2013 American Transplant Congress, has shown a strong correlation with DSA and development of transplant glomerulopathy in cases containing a pure population of CD68+ infiltrating monocytes/histiocytes within glomeruli. Another hallmark of MI is ptc (Figure 4) and its scoring had already been defined in the 2007 update, without any further recent modification. ¹⁴ Biopsies with inflammatory cells in <10% of cortical peritubular capillaries are scored as ptc0, regardless of the number of cells in the most severely involved capillary. Once ⩾10% of capillaries are inflamed, the ptc score is based upon the greatest number of luminal inflammatory cells in the mostly affected cross-sectioned cortical peritubular capillary. Ptc1 has up to four luminal cells, ptc2 up to 10 cells and ptc3 > 10 cells. The composition (mononuclear cells vs. neutrophils) and extent (focal ⩽ 50% vs. diffuse > 50%) of ptc should be commented. There are several situations where ptc scoring should be avoided altogether: capillaries cut in longitudinal plane, in the medulla due to the association of acute tubular injury with vasa recta infiltrates, vessels surrounding nodular lymphoid aggregates, areas of pyelonephritis and adjacent to infarcts. Listed in table 2 are two more features of antibody mediated acute tissue injury, namely acute thrombotic microangiopathy (Figure 3) and acute tubular injury. It is important that other causes be excluded as they are much less specific findings compared to MI or arteritis.

Figure 3.

Transplant glomerulitis and acute thrombotic microangiopathy. Endothelial cell swelling, leukocyte infiltration and luminal thrombus (asterisk) resulting in capillary luminal occlusion. Leukocyte margination is more prominent in other glomeruli from the same biopsy (PAS, original magnification ×400).

Figure 4.

Peritubular capillaritis. Only cross-sectioned capillaries (arrows) are scored. (PAS, original magnification ×400).

Transplant glomerulopathy carries a poor prognosis, especially if there is evidence of concurrent antibody interaction with vascular endothelium. A significant proportion of such grafts fail eventually and have to be removed. The threshold for cg has been lowered to double contours in ⩾1 capillary loops in a single glomerulus, resulting in better inter-observer agreement (kappa = 0.47). In addition, there were better correlations with anti-class II DSA and gene transcripts of endothelial activation and injury. Previously, a threshold of ⩾10% of capillary loops in the most severely affected glomeruli was used. Furthermore, cg scores relied entirely on light microscopy (LM) features. However, new evidence emerged from studies by Wavamunno and others that electron microscopy (EM) can be helpful in documenting early changes of chronic ABMR. ²⁴ These investigators found that glomerular basement membrane (GBM) duplication (Figure 5(a)), endothelial cell activation/swelling and subendothelial electron-lucent widening detected within the first three months post transplantation were highly predictive of later development of overt transplant glomerulopathy (visible on LM) and correlated well with DSA.^24,25 Changes typifying endothelial cell activation include hypertrophy, cytoplasmic vacuolation, transformation from fenestrated to continuous endothelium, as well as serration and expansion of lamina rara interna. Increased numbers of mitochondria, Golgi and ribosomes can also be seen. Endothelial cell activation involves peritubular capillaries as well, followed by peritubular capillary basement membrane multilayering. As a result of these findings, the Banff work group members have recommended ultrastructural studies to be performed in all biopsies from patients who are sensitised, have documented DSA any time post transplantation and/or who have had a prior biopsy showing positive C4d and/or MI. EM should also be considered for all biopsies performed ⩾6 months post transplantation and for indication biopsies done ⩾3 months post transplantation to determine if early changes of CG are present, prompting DSA testing. This has implications for transplant centres without EM facilities as it means this new set of criteria cannot be applied to all laboratories. On the other hand, most centres with EM capability perform EM on renal allograft biopsies only when recurrent or de novo glomerulonephritis is suspected. With these new recommendations, more laboratories will incorporate ultrastructural studies in their transplant biopsy workup, increasing the cost and technical demands of a biopsy. Transplant glomerulopathy is now defined as follows:

cg0 – no GBM double contours by LM or EM

cg1a – no GBM double contours by LM but double contours (circumferential or incomplete) in at least three glomerular capillaries by EM with associated endothelial swelling and/or subendothelial electron-lucent widening

cg1b – one or more glomerular capillaries with GBM double contours in ⩾1 non-sclerotic glomerulus by LM; EM confirmation is recommended if available

cg2 – double contours affecting up to 26–50% of peripheral capillary loops in most severely affected glomeruli

cg3 – double contours affecting >50% of peripheral capillary loops in most severely affected glomeruli

Figure 5.

Transplant glomerulopathy showing GBM duplication (a). Multilayering of peritubular capillary basement membrane (b). (Uranyl acetate and lead citrate, original magnification ×12,000 left panel and ×10,000 right panel).

As alluded to above, multilayering of peritubular capillary basement membrane (Figure 5(b)) is a sign of chronic ABMR. According to the findings of Liapis et al., ²⁶ and adopted by Banff meeting participants, peritubular capillary basement membrane multilamination which has ⩾7 layers in one cortical capillary and ⩾5 in two additional capillaries is considered severe and included as a feature of chronic antibody mediated injury. Stringent criteria are necessary because basement membrane multilayering is not pathognomonic of any specific disease entity. In allografts, it can be demonstrated in acute/chronic T-cell rejection and calcineurin inhibitor toxicity. It is also not an uncommon occurrence in native kidneys especially those with chronic thrombotic microangiopathy.

Recognised since the preceding conference, C4d-negative ABMR is formally adopted as a distinct diagnostic category in the 12^th Banff meeting. This is supported by many studies done on dysfunctional grafts and grafts with stable function from protocol biopsies.^{5,18,19,25,27 –30} In this meeting, Robert Colvin reported work from his laboratory indicating that about 20% of for-cause biopsies with acute microvascular injury and DSA showed negative C4d staining with even the most sensitive indirect immunofluorescence technique. ²⁰ Detection of C4d in allograft tissue, previously needed for diagnosis of ABMR, has been replaced by a new category known as ‘current/recent antibody interaction with vascular endothelium’. Evidence for the latter can take the form of C4d positivity, at least moderate MI or increased expression of gene transcripts in the biopsy tissue indicative of endothelial injury. Moderate MI is defined as g + ptc ⩾ 2, and has been shown to be associated with transplant glomerulopathy even in C4d-negative cases in the presence of DSA. ²⁵ For the first time, molecular evidence of active endothelial injury in tissue is accepted as a marker of ABMR provided thorough validation of the test has been done. Sis and co-workers were the first to validate ENDAT (endothelial activation and injury transcript) expression in C4d-negative grafts, finding correlation with development of transplant glomerulopathy and graft survival. ²⁷

Future directions – perfecting a classification system

Now into its 12^th revision, the Banff meeting participants have made great strides in creating and continuously improving on a pathology-based classification scheme that has clinical relevance. Arising from discussions at the 2013 conference, new BWGs were formed to address problematic areas. ²⁰ A work group was tasked specifically to look into issues pertaining to TCMR. One area is the clinical significance of borderline rejection, whether it should be eliminated or modified to better select lesions that truly represent mild rejection. Tubulitis and interstitial inflammation are fundamentally quite subjective and difficult to score when scarring (tubular atrophy/interstitial fibrosis) is extensive. Tubulointerstitial inflammation is also a feature of BK polyomavirus nephropathy, and can be difficult to differentiate from rejection. In fact, both can coexist. Should there be emphasis on the extent of tubular injury, interstitial oedema and tubulitis? New techniques like molecular phenotyping have the potential to reclassify borderline cases into TCMR and nonrejection, ³¹ but questions regarding validity and the types of transcripts to use remain. The concept that interstitial inflammation in scarred areas of the biopsy is detrimental to the graft is known for some time (see inclusion of ‘ti’ lesion in 2007 above). Published data lend support to this idea.^32,33 The same group will be studying if the ‘ti’ score should be included in the classification for TCMR or as part of a new category of chronic active TCMR. Another working group will investigate and report on evidence-based clinical and laboratory assessment of highly sensitised patients. A third working group was formed to evaluate the role of molecular diagnostics as a complementary tool to DSA testing. Finally, transplant centres in Singapore not infrequently encounter patients who have received a kidney transplant from abroad, such as China and India. In these cases, the donor HLA phenotype is often unknown, rendering DSA detection for managing humoral rejection difficult. A group from Hong Kong has previously reported on the feasibility of using DNA extract from renal allograft biopsy tissue to obtain information on donor HLA phenotypes. ³⁴ Another initiative worth exploring is the use of surveillance protocol biopsies to detect subclinical rejection and other unexpected pathologies like calcineurin inhibitor (CNI) toxicity and recurrent disease. ³⁵ These biopsies can be very informative for following high-risk patients (e.g. cross-match positive, ABO-incompatible recipients) and during periods of reduced immunosuppression. Individual centres should weigh available resources against early detection of reversible lesions which prolongs graft survival.