Background. The Banff classification for assessment of renal allograft biopsies was introduced as a standardized international classification of renal allograft pathology and acute rejection. Subsequent debate and evaluation studies have attempted to develop and refine the classification. A recent alternative classification, known as the National Institutes of Health Collaborative Clinical Trials in Transplantation (NIH-CCTT) classification, proposed three distinct types of acute rejection. The 1997 Fourth Banff meeting appeared to move towards a consensus for describing transplant biopsies, which incorporated both approaches. Patients who received a renal allograft at the Oxford Transplant Centre were managed by a combination of protocol and clinically indicated biopsies. We have undertaken a retrospective analysis of the biopsies correlated with the clinical outcome to test the prognostic value of the original Banff (Banff 93–95) and NIH-CCTT classifications.
Methods. Three hundred and eighty-two patients received renal allografts between May 1985 and December 1989, and were immunosuppressed using a standard protocol of cyclosporine, azathioprine and steroid. Adequate 5-year follow-up data were available on 351 patients, and of these, 293 had at least one satisfactory biopsy taken between days 2 and 35 after transplantation, the latter patients forming the study group. The D2–35 biopsies taken from these patients, which were not originally reported according to the Banff classification, were re-examined and classified according to the Banff 93–95 protocols. For each patient the biopsy found to be the most severely abnormal was selected, and the Banff and NIH-CCTT grading compared with the clinical outcome.
Results. Seven hundred and forty-three biopsies taken from 293 patients between days 2 and 35 after transplantation were examined and the patients categorized on the basis of the `worst' Banff grading as follows. Normal or non-rejection, 20%; borderline, 34%; acute rejection grade I (AR I), 18%; AR IIA, 6%; AR IIB, 14%; AR III, 1%; AR IIIC, 3%; widespread necrosis 3%. The clinical outcome for the last two groups combined was very poor with 18% of grafts functioning at 3 months and 6% at 5 years. The other groups with vascular rejection (AR IIB and AR III) had an intermediate outcome, graft survival being 78% at 3 months and 61% at 5 years. The remaining four groups (normal, borderline, cellular AR I and AR IIA) had the best outcome: graft survival 95% at 3 months and 78% at 5 years with virtually no difference between the four groups. Three forms of acute rejection, namely tubulo-interstitial, vascular and transmural vascular, were identified, but only the latter two categories were associated with a poor outcome.
Conclusions. The eight sub-categories of the Banff classification of renal allograft biopsies are associated with three different prognoses with respect to graft survival in the medium term. These three prognostic groups correspond to the three NIH-CCTT types. The data provide support for the consensus developed at Banff 97 separating tubulo-interstitial, vascular and transmural vascular rejection (types I, II and III acute rejection).
Rejection remains a significant problem following renal allotransplantation and although powerful anti-rejection therapy is available, its unguided use is associated with significant morbidity and mortality. The needle core biopsy for conventional histological examination remains the technique of choice for diagnosis of rejection , but in the past interpretation of the histological appearance was an individual art. A series of meetings at Banff introduced an international standard for evaluation of biopsies after renal transplantation which allowed classification of renal allograft pathology and acute rejection in particular. The goal of the Banff classification of renal allograft rejection was a schema in which a given biopsy grading would imply a prognosis for a therapeutic response or long-term function . It was emphasized from the outset that the development of the standardized schema was only the first step, with the clinical implications to be proven through further studies. However, such studies remain sparse. Dooper and co-workers  found that the Banff criteria correlated with the retrospective clinical diagnosis in 77% of cases of renal dysfunction, in a study not including protocol biopsies. Gaber and co-workers  found that the response to rejection treatment correlated well with Banff categories with complete reversal of rejection in all eight cases defined as borderline rejection, 93% of 14 grade I (mild cellular rejection) and 79% of 19 defined as grade II (moderate acute rejection, either cellular or vascular), while only 47% of cases of grade III rejection (severe vascular) were reversible. More recently Gaber et al.  presented multicentre data supporting the use of Banff criteria. Nickerson and co-workers  focused their study on the ability of protocol biopsies to predict diminished function at 24 months post transplant. Rush and colleagues [7,8] classified up to 30% of stable protocol biopsies in the first 3 months as acute rejection, including a small group with vascular rejection. When regular protocol biopsies were performed, borderline changes were frequent.
Another important clinical question is the significance of any given Banff category of rejection in terms of the prognosis for long-term graft function. Ideally, the answer to that question would be derived from a prospective clinical trial in which a large number of consecutive transplants would be managed similarly using a common immunosuppressive regimen and performing both routine and clinically indicated biopsies regularly on every transplant. The morphological findings could then be correlated with the short, medium and long-term outcome. However the logistics of such a study are considerable since results would only become available after 5 years or more. Between 1985 and 1995 the Oxford Transplant Centre used a standardized triple therapy regimen (cyclosporine, azathioprine and prednisolone) for treatment of renal transplant patients and the management was undertaken by the same senior staff, providing uniformity and continuity. During this time it was routine policy at this centre to undertake biopsies of the donor kidney before transplantation as well as protocol biopsies at 7, 21, 28, 90 days and approximately 1 year post transplant. In addition biopsies were always taken when there was clinical suspicion of rejection or other cause of impaired function. The biopsies had not been originally reported using the Banff criteria. Importantly, the follow up of virtually all the patients remained directly under the control of the transplant centre with duplicate record keeping and computerized data entry and retrieval.
In the absence of prospective data these patients and their available renal biopsies provided the next best opportunity to test the correlation between the various classifications of acute rejection and other morphological subgroups defined by the Banff criteria and the short-, medium- and long-term outcome. Because of the likely influence of changes in immunosuppression the period chosen for study was selected because the drug regimen used had remained standardized, with a written protocol agreed by senior staff. The triple therapy regimen used is similar to that used in many centres throughout the world, and is probably the closest approach to an internationally standardized regimen.
Materials and methods
All patients undergoing renal transplantation at the Oxford Transplant Centre between May 1985 and December 1989 inclusive were considered for the study. Patients were included in the study if they had evaluable biopsies available taken between days 2 and 35 after transplantation. Subsequently, all biopsies taken from those patients up to the first 18 months after transplantation were re-evaluated, these later biopsies being used in part (along with clinical data) to define the transplant outcome. All patients received a standard triple therapy immunosuppression protocol consisting of daily cyclosporine 10 mg/kg (subsequently adjusted on the basis of trough cyclosporine levels), azathioprine 1.5 mg/kg and prednisolone 20 mg (tapered to 10 mg over the first 6 months). Antilymphocyte preparations were not routinely used for prophylaxis, but were used to treat steroid-resistant or very severe rejection, starting after a biopsy had been taken.
To prevent knowledge of the patient outcome influencing the evaluation, the histological slides were identified by histological number only, before re-examination and classification. Furthermore, the two pathologists (WDB and DRD) had no knowledge of whether biopsies were protocol biopsies or had been taken for any clinical indication.
For all biopsies there were usually at least three haematoxylin and eosin sections, three PAS and four trichrome-stained sections. Where more sections had originally been cut the slides evaluated always included the first and last sections as well as a slide of the middle section. The adequacy of the specimen as a whole was assessed using the standard Banff criteria. Adequate biopsies contained seven or more glomeruli and at least one artery. Unsatisfactory biopsies contained no glomeruli or arteries. Marginal biopsies contained between one and six glomeruli with an artery. Those specimens categorized as marginal were used in the study in addition to the adequate group.
The features scored on each biopsy were first the number of glomeruli, to assess the adequacy of the biopsy, the presence of glomerulitis, interstitial infiltrate, tubulitis, intimal arteritis and arteriolar hyalinosis was then scored on a scale of 0–3 according to criteria laid down in the Banff classification. If a biopsy was given the score 3 for intimal arteritis, two subgroups were identified as either grade III or grade IIIC, using the criteria suggested at the Third Banff Conference in 1995 . Biopsies that showed any combination of the following changes: transmural necrotizing arteritis, fibrinoid vascular change or medial smooth muscle necrosis, were categorized as Banff AR IIIC while those with severe vasculitis without these features were categorized as AR III.
The Banff category for each biopsy was then determined so that they could be placed in one of the following nine groups.These categories correspond to those described in the original Banff 93–95 criteria with the exception of group 8, which is an additional group where there was widespread necrosis or infarction and no viable tissue or no arteries were identified (a situation not specifically allowed for in the original Banff criteria). It was only possible to categorize the various forms of vascular rejection (AR IIB, AR III and AR IIIC) when viable vessels were present. The Banff classification describes grade III, severe acute rejection in the following terms: `Recent focal infarction and interstitial haemorrhage without obvious cause are also regarded as evidence for Grade III rejection' . While accepting the usual validity of that statement we felt our approach fewer assumptions or inferences.
Normal and other (including non-specific elements of tubular necrosis or possible cyclosporine toxicity).
Borderline (mild tubulitis).
Acute rejection grade I (moderate tubulitis).
Acute rejection, grade IIA (severe tubulitis).
Acute rejection, grade IIB (mild or moderate arteritis).
Acute rejection, grade III (severe intimal arteritis without vessel changes described in grade IIIC).
Acute rejection, grade IIIC (severe intimal arteritis with transmural necrotizing arteritis, fibrinoid change or medial smooth muscle necrosis).
Widespread necrosis or infarction (but no evidence of vascular rejection seen).
Unsatisfactory specimen, inadequate number of glomeruli or arterioles.
An initial pilot study on 20 biopsies was undertaken by WDB and DRD to ensure acceptable application of the Banff criteria. Areas of uncertainty and difference were discussed and where possible clarified. These were then included in the main study, where the evaluation and scoring of the biopsies was carried out essentially by one pathologist (WDB). During the study certain problem areas and issues appeared and were discussed by WDB and DRD. It was reassuring to confirm that many of these difficulties had also been raised and discussed at the Third Banff Conference on Allograft Pathology [9–12]. Only when all the biopsies had been evaluated and scored on a chart were the names, chronological transplant numbers and times from transplantation added.
To improve the accuracy and quality of the observations on which the scoring was performed and to audit possible errors, the original biopsy reports, which did not use the Banff criteria, were reviewed at this stage. Reports which appeared to be at variance with the Banff score were then re-evaluated by WDB and changes made where they seemed appropriate. After re-evaluation, 46/743 (6.2%) of the biopsies were altered to another grade. Of the 46, 37 were given scores of increased severity, four were decreased and five were placed in the widespread necrosis category, a non-BANFF group. The most frequent category revisions were normal revised to borderline (18 cases) and borderline revised to AR I (12 cases), due to identification of grades of tubulitis not recognized at initial examination.
It should be emphasized that although clinical data would have influenced the categorization of the original biopsy reports, changes made to the Banff categories were dependent only on morphological features. The definition of rejection or otherwise in this biopsy study is based on histology alone.
When patients had more than one satisfactory biopsy taken between days 2 and 35 after transplantation, the one showing the most severe changes according to the Banff criteria was selected as predictive for that patient. The clinical outcome and number of rejection episodes for each patient was then determined by a survey of the notes, charts and computer files, defining a rejection episode as a sustained rise in creatinine of >15%, supported by any combination of the following features: biopsy evidence, treatment of rejection, response to treatment and the lack of another explanation for the creatinine rise (such as high cyclosporine levels or urinary obstruction). The number of actual courses of methyl prednisolone given for treatment of rejection was retrieved from a computer database, which has been used to prospectively record the number of treatments of methyl prednisolone given to each patient following transplantation (among other data). The long-term outcome for the groups showing each Banff grade was expressed in terms of actual 3-month, 1-year, 3-year and 5-year graft survival.
Three hundred and eighty-three consecutive patients underwent renal transplantation at the Oxford Transplant Centre between May 1985 and December 1989. Thirty-two patients (8%) were excluded because of lack of full clinical information or follow-up (mostly these were visitors returning to other countries). Of the remaining 351 patients, 293 (83%) had at least one satisfactory biopsy taken between days 2 and 35 after transplantation and so were included in the study. A total of 743 biopsies taken between days 2 and 35 post transplantation were evaluated according to the Banff criteria and the patients allocated to Banff categories according to the `worst' biopsy for each patient. The distinction between protocol and non-protocol biopsies (the latter presumably mostly for dysfunction) was not recorded prospectively, but 194 biopsies (66% of 293) in total were taken between days 6–8 (153 for that period alone) and 26–32 (41 for that period alone), which correspond to the chosen protocol timings.
The percentage of patients allocated to each Banff category is shown graphically in Figure 1. Of the patients, 60/293 (20.4%) had their `worst' biopsy categorized as normal, or as showing non-specific changes or non-rejection changes such as tubular necrosis or cyclosporin toxicity. The borderline category introduced by Banff was the largest single group, 100/293 (34.1%). The two variants of cellular tubulo-interstitial rejection were less common: groups AR I 52/293 (17.7%), AR IIA, 18/293 (6.1%); and combined 70/93 (23.8%). Vascular rejection was even less common: groups AR IIB 42/293 (14.3%), AR III 4/293 (1.3%) and AR IIIC 8/293 (3.0%); but if the three vascular rejection groups were combined, 54/293 (18.4%). Widespread necrosis, the extra group added to the original Banff classification, was seen in 9/293 (3.0%). Four of these cases were subsequently proven to have been due to renal vein thrombosis of the graft.
The outcome in terms of graft survival for patients with biopsies in each Banff category is shown graphically in Figure 2. Patients whose biopsies showed the most severe changes had a very poor outcome. None of the eight graded as Banff AR IIIC had a functioning graft beyond 3 years, and only one patient of nine whose graft showed necrosis without clear evidence of vascular rejection had a graft that survived 5 years. Indeed, 14 of the 17 grafts with this type of biopsy were lost before 3 months.
Figure 2 shows that two diagnostic categories of biopsy were associated with an intermediate level of graft survival, these two being the groups of mild, moderate and severe vasculitis (Banff classifications AR IIB and AR III), which together numbered 46 patients, the majority being in the mild/moderate group AR IIB (42/46). The combined group had a 5-year kidney graft survival of 60.8% (28/46).
A further grouping were the grafts that showed only normal or non-specific changes (60 patients), the borderline group (100 patients), and the mild (AR I, 52 patients) and moderate cellular rejection groups (AR IIA, 18 patients). These four categories had the best overall prognosis, together having a graft survival of 95.2% at 3 months, 93.5% at 1 year, 86.1% at 3 years and 78.2% at 5 years. Furthermore, instead of a prognostic gradient favouring normal and declining towards the AR IIA group, the outcome at 5 years was remarkably similar at 78.3, 78, 78.8 and 77.8% for the normal, borderline, AR I and AR IIA groups, respectively. As expected, the number of recorded rejection episodes and the total methyl prednisolone given tended to increase from normal to severe cellular rejection as shown on early biopsy (Figures 3 and 4).
The study of renal transplant pathology presented here is unusual in having such a comprehensive set of biopsies, including protocol biopsies, allied to excellent clinical and follow-up data. The uniform immunosuppression protocol used and the stability of patient management allows a level of confidence in the results which may approach that of a true prospective study. When the Banff classification was first developed it was acknowledged that the categorization was only the initial step in a process of evaluation of the scheme. Further clinical, diagnostic and prognostic studies would be needed to assess, and if need be, refine the schema . Our study was undertaken in the hope that it would contribute to this continuing development. Our findings appear to support the view that there are three separate prognostic groups of renal transplant pathology associated with differing short- and long-term outcomes.
The first grouping of Banff categories are those described as normal or non-specific changes, borderline, mild (AR I) and moderate (AR IIA) cellular rejection. In fact at 3 years the Banff AR IIA group had the best graft survival at 17/18 (94%), although not statistically different from the other groups. The fact that the normal or non-specific change group had an excellent outcome is unsurprising, but the absence of any deleterious effect of the increasing severity of rejection found in the other three groups is perhaps surprising, and must indicate the reversibility of the pathological processes observed with adequate treatment. It is also reasonable to question whether the cellular infiltrate, which is largely the basis for categorizing the latter three groups is actually not just damaging to the graft but may have some protective or beneficial role. The finding of a large number of biopsies being graded as borderline by the Banff criteria (34.1%) confirmed the impression from current biopsies at the time of the study (1996) that a large proportion of both protocol and clinically indicated biopsies have changes that place them in this group. As was noted at the Third Banff meeting, this group presents particular challenges in terms of management decisions . Our study suggests that despite these difficulties the outcome is excellent. Whether this is because of the management decisions taken, or in spite of them, is unclear from this study. Rush et al.  also documented the frequency of borderline changes in protocol biopsies and suggested that persistent borderline changes were associated with decreased function.
The second main grouping of Banff criteria in our study consists of those patients with biopsies showing the features of mild, moderate and severe vasculitis (Banff AR IIB and AR III), which together were seen in 15.6% of patients, the majority being in the mild/moderate group AR IIB. Graft survival at 5 years was significantly reduced to 60.8% for the combined group, supporting the view that vasculitis is a more significant feature in terms of progressive damage than cellular infiltration or tubulitis.
The third and last main grouping of Banff criteria consists of the group with necrotizing vasculitis, AR IIIC and the extra group that we had categorized as widespread necrosis. These two subgroups showed a similar poor outcome with only 18% (3/17) of the combined group surviving at 3 months and 6% (1/17) at 5 years. This does not necessarily imply similar pathogenesis but does support the view that severe parenchymal damage in a biopsy usually implies a serious, probably vascular, underlying cause, unless there is a good explanation for local damage such as a biopsy from an adjacent area.
In the intervening period since the introduction of the original Banff criteria, an alternative classification, the NIH-CCTT system was developed. In this system three histopathological categories of acute rejection were recognized: Type I, tubulo-interstitial with mononuclear infiltrate and tubulitis; type II, arterial or arteriolar endothelialitis; and type III, arterial fibinoid necrosis or transmural inflammation with or without thrombosis, parenchymal necrosis or haemorrhage . At the Banff meeting in 1997 the NIH-CCTT scoring system was presented as an alternative to the Banff criteria. There was agreement that there was substantial correspondence between the Banff schema 93–95 and the NIH-CCTT scoring system and Banff 97 was developed accepting the following groupings of acute rejection: type I, tubulo-interstitial rejection; type II, vascular rejection; and type III, transmural vascular rejection .
Although our study was originally designed to test the prognostic value of the Banff 93–95 categorization alone it is possible to draw some inferences regarding the NIH-CCTT system of classification and the combined Banff 97. There is support in the present study for choosing just three diagnostic categories, but our findings would seem to question the necessity of continued emphasis on the degree of infiltrate or tubulitis, as different degrees of severity within the types gave similar outcomes, at least to 5 years. However, it should be noted that outcomes may result from different treatments for rejection as determined by the clinical state and biopsy findings.
The clinical diagnosis of rejection in our study was to some extent based on the biopsy report and is therefore something of a circular argument, nevertheless we would contend that the retrospective evaluation of the overall clinical picture, biopsy report, level of nephrotoxic drugs and response to therapy gives the most accurate assessment of the presence of rejection. The use of methyl prednisolone courses alone as a measure of rejection may also be criticized because it does not include treatment with anti-lymphocytic preparations. However most patients given anti-lymphocytic preparations also received at least one course of methyl prednisolone before starting antibody treatment.
A recent study from The Netherlands provides useful data for comparison to our study . In their cohort of 482 transplant patients, 76 (15,8%) developed vascular rejection in the first 3 months and 115 (23,9%) interstitial rejection. One- and 5-year graft survival for the groups without rejection, with interstitial rejection and vascular rejection were respectively: 87.8 and 73.5%; 87 and 71.4%; and 48.7 and 34.3%. It should be noted that those with no clinical rejection (n=245) were not biopsied and some (n=46) were treated for rejection without a biopsy. The conclusions were that early vascular rejection (and/or widespread parenchymal necrosis) occurring within 3 months after transplantation, was the most important predictor of both early and late graft loss. Kidney allografts described as having early rejection (appropriately managed) had the same favourable prognosis as those without early rejection. This is a finding which we have confirmed and extended at a morphological level as we have shown that it holds for Banff subgroups with increasing degrees of cellular rejection. It is important to emphasize that these data should not be interpreted as implying that transplant biopsies or Banff categories have no influence on 5-year outcome, rather that appropriate treatment of cellular rejection may result in excellent graft survival, equivalent to that of grafts with no histological evidence of rejection in the first 35 days after transplantation.
The findings of our study support the integration with the NIH-CCTT approach that occurred at Banff 97 by showing that the seven Banff categories described in the 1993–1995 era correspond well with the three patterns of acute rejection emphasized by the NIH-CCTT approach in terms of prognostic significance at 5 years. Type I, tubulo-interstitial rejection (when treated) is associated with a similar outcome to those grafts with normal and borderline changes. Type II, intimal arteritis, has an intermediate outcome with moderate graft loss in the first 3 months after transplantation and type III, transmural vasculitis, has a poor outcome with grafts only rarely surviving beyond 3 months. Banff 97 vindicates our approach to widespread necrosis by suggesting that such biopsies should be classified as v0*, as haemorrhage or infarction alone is no longer considered adequate to presumptively diagnose vascular rejection.
These findings, namely the successful outcome of most grafts in which borderline or tubulointerstitial rejection was defined and treated, explain the observation in many multicentre trials of new immunosuppressive agents in recent years (where biopsy proven rejection was a primary end-point) that a significant reduction in biopsy-proven rejection is not necessarily associated with improved graft survival.