Challenges Posed by the Banff Classification: Diagnosis and Treatment of Chronic Active T-Cell-Mediated Rejection

The Japanese Society for Transplantation Research publishes a Transplant Fact Book on its website; this is an annual statistical survey on transplantation [1]. The survival rate of kidney transplant patients from living donors since 2001 is 91% at 10 years and 86.3% at 15 years. The leading causes of death are malignancy (15.7%), infectious diseases (12.9%), and cardiovascular disease (12.0%). These conditions increase as kidney dysfunction develops. For example, in terms of malignancies, cancers of the urinary, endocrine, and gastrointestinal systems become more common with the development of kidney dysfunction [2], as do the risks of various infectious diseases. The concept of secondary immunodeficiency related to kidney disease, which affects the immunological responses to vaccines, has been proposed [3]. We previously found that the antibody response to SARS-CoV-2 mRNA vaccines is extremely low in kidney transplant recipients [4]. Similarly, it has long been known that cardiovascular death and events increase with development of kidney dysfunction [5]. Thus, prevention of kidney function deterioration and increasing the kidney allograft survival rate are extremely important. The kidney allograft survival rates from living donors since 2001 are 83.7% at 10 years and 70.5% at 15 years. Most kidney transplant rejections in Japan are chronic rejections (24.2%), followed by acute rejections (7.0%) and disease recurrence (3.9%) [1, 6, 7]. The Australia and New Zealand Dialysis and Transplant Registry includes the causes of kidney allograft loss and ranks chronic rejection first. The second most common cause is death with a functioning allograft, and the third is recurrent nephritis/nephropathy [8]. To prevent such conditions, diagnoses of rejection (acute and chronic) and recurrent nephritis/nephropathy are essential. In this review, we overview the Banff 2019 classification that contains the global standard criteria for diagnosis of transplanted kidney conditions and emphasizes the importance of T-cell-mediated rejection (TCMR), which has recently received much attention. We discuss changes in the Banff classification of TCMR, treatment, and the future prospects.

The year 2021 marked the 30th anniversary of the Banff Conference, which was first held in Banff, Canada, in 1991, when Professor Kim Solez (Alberta University) and Professor Lorraine Racusen (John Hopkins University) led a discussion of international standards for diagnosis of allograft pathology. The Banff meeting is now held every 2 years, and the proceedings are published in top journals such as the American Journal of Transplantation (AJT) and Kidney International (KI). The latest edition of the Banff 2019 classification was discussed in Pittsburgh, USA, in September 2019. There are currently 11 working groups on TCMR. New findings in the decade since the 2009 update include the following: microvascular injury is associated with graft loss and donor-specific antibodies (DSAs); thrombotic microangiopathy and intimal arteritis (v) lesions are more common in antibody-mediated rejection (ABMR) than TCMR patients; C4d-negative ABMR exists; caution should be exercised when evaluating early lesions in terms of transplant glomerulopathy (TG) (via cg1a scoring); tubulitis may develop in the interstitial fibrosis and tubular atrophy (IFTA) lesions of patients with chronic active TCMR; stratification methods may predict the prognosis of polyomavirus nephropathy (PVN); and molecular pathology is becoming ever more important [9]. The Banff classification is divided into five major categories: normal; ABMR; borderline (BL) change (suspicious) for TCMR; TCMR; and PVN. There are three key considerations when seeking to understand the Banff classification: lesion definitions and grading, the presence or absence of complement activity (C4d), and evaluation of DSA. Diagnosis is now automated by reference to these factors [10]. Five lesional sites must be scored: the acronyms are i for interstitial, t for tubular, v for vascular endothelium, g for glomerular, and ptc for the peritubular capillaries; chronic lesions are abbreviated ci, ct, cv, cg, and ptcml, respectively. Former Banff meeting reports were not easy to read because they often dealt only with revisions, but the Banff 2019 report is very user-friendly, with the entire scoring system on one page and the entire Banff classification on another.

TCMR is an inflammatory reaction principally of the kidney tubules and arterioles caused by CD8-positive T cells. It peaks about 3 months postoperatively, followed by a gradual decline, and is rarely encountered after 10 years [11]. Pure TCMR has a good prognosis, but TCMR featuring antibody-related rejection and mixed-type TCMR have poor prognoses [12]. It has been suggested that TCMR may trigger ABMR. For example, one study found that the appearance of de novo DSA, which causes ABMR, is more common in cases experiencing TCMR within 6 months [13]. Another study demonstrated that TCMR cases experience more TG than others; this pathology is characteristic of chronic active ABMR [14]. If such rejection targets tubular epithelial cells via CD8-positive T cells, thus in patients who are BL/acute TCMR, DSA and ABMR may develop.

TCMR is divided into acute and chronic active. The t, i, and v scores are those of acute TCMR lesions and are graded from 1 to 3 when present. Tubulitis is graded by the number of lymphocytes among epithelial cells of tubular slices. The “i” score reflects “inflammation of the non-scarred cortex” and is the areal percentage of inflammatory cell infiltration. The “v” score reflects vascular endarteritis. The most affected vessels are graded in terms of the percentage leukocyte adhesion to the endothelium. A “v3” grade is the most serious, indicating fibrinoid necrosis and/or transmural injury. Chronic active TCMR lesions are graded by the presence or absence of i-IFTA, ti, t or t-IFTA, and chronic allograft arteriopathy (cv). The “cv” score is graded by evaluating the most affected vessels in terms of the percentage of luminal narrowing. Normal atherosclerotic lesions evidence multiple layers of fibrous intima, but “cv” lesions are characterized by intimal thickening with inflammatory cells.

One issue that has been of long-term concern among those engaged in Banff classification is the etiological classification of IFTA. For example, for ABMR, the pathological changes and the pathophysiologies of glomerulitis and TG are easy to understand; these are characteristic chronic lesions developing from persistent acute lesions. In the TCMR context, however, it is difficult to define the relationship between tubulitis and tubular atrophy as the development of a characteristic chronic lesion from a persistent acute lesion. This is because tubular atrophy is associated with more than just tubulitis, exhibiting several additional etiologies. Specifically, the tubular atrophy of TCMR is indistinguishable from that of PVN, pyelonephritis, ABMR, recurrent glomerulonephritis, and urinary obstruction. It is important to distinguish the various forms of IFTA, thus the final outcomes of various kidney diseases, by their etiologies. Category 5 of the Banff classification used to be IFTA, a nonspecific diagnosis, but it is essential to find a more etiologically specific name that includes i-IFTA, t-IFTA, or total inflammation (ti) in the definition of chronic active TCMR. When tubulitis persists in smoldering form, this is not simply IFTA, rather than reflecting infiltration of a few or many inflammatory cells, or tubular atrophy. Efforts are being made to fine-grain the IFTA lesional grades by the etiologies. The Banff 2015 classification defined only “BL” and acute TCMR; chronic active TCMR was defined for the first time in 2017.

In the current revision (Banff 2019), there are three principal changes. First, given the better prognosis of “i0 t1” patients, this class has been removed from the BL diagnostic criteria because isolated “t” status was associated with extremely favorable allograft survival of 98.8% and 92.7% at 1 and 5 years, suggesting that the “i” score rather than the “t” score may determine prognosis [15]. Second, BL is now defined as “BL change (suspicious) for TCMR.” It has long been argued that BL status accompanied by an elevated creatine level should be viewed as TCMR. For example, in 1996, the significance of the Banff “BL” definition was already under discussion. In this report, 59 patients of “i1, t1, v0” status and 22 of “i2, t1, v0” status were treated as were acute TCMR patients (PSL pulse 48%, anti-thymocyte globulin 52%) and the average creatine elevation was 1.1 ± 0.1-fold in 96% of cases; the complete remission rate was 43%, the partial remission rate 28%, and the no-remission rate 30% [16]. A recent molecular approach showed that 13 of 40 BL cases were TCMR-like and 27 non-rejection-like. If “i” was below 27% and the extent of tubulitis was less than 3%, there was no rejection [17]. This indicates that “i0” may be associated with a benign prognosis but “i2” and “i3” with tubulitis that may be TCMR-like and treatment should be considered. Third, although “i-IFTA” and “ti” were added to the Banff 2017 criteria, “t-IFTA” was added to the diagnostic criteria for chronic active TCMR in Banff 2019. The idea in fact emerged around 2010 that the morphological profiles of “i-IFTA,” “t-IFTA,” and “ti,” were important. In the past, inflammatory cell infiltration of IFTA sites and tubulitis were difficult to evaluate. However, one report found that grading of inflammatory cell infiltration into the IFTA lesions allows stratification of kidney allograft survival [18]. Another study examined “i-IFTA” and “t-IFTA” status in first-year protocol biopsies of 1,539 transplant patients and found that IFTA was present in 61.5%, of whom 41.6% had i-IFTA and acute TCMR, thus exacerbated IFTA. The 8-year graft survival rate was 70.8% compared to 83.5% (p = 0.01) in i-IFTA(−) patients [19]. TCMR preceded i-IFTA and in fact caused IFTA in 429 of the indication biopsies and 2,052 of the protocol biopsies, and i-IFTA in the first year contributed not only to IFTA development but also to arterial intima-media thickening, TG, and kidney dysfunction [20]. However, it is debatable whether the etiology of i-IFTA is active TCMR [21, 22]. For example, another study examined i-IFTA cases using a microarray-based molecular diagnostic system (MMDx). Totals of 108 i-IFTA cases, 73 IFTA-only cases, and 53 no-IFTA cases were compared. In all patients, the expression levels of genes involved in ABMR/TG were elevated regardless of the i-IFTA grade. i-IFTA was clearly associated with enhanced expression of TCMR-associated genes in year 1, but this later became less significant [23]. A recent report found that the t-IFTA score could be used to stratify kidney allograft survival [24]. Another report showed that the prognoses of kidney allografts could be precisely stratified by the percentage of total inflammatory cell infiltration into both the atrophic and non-atrophic lesions of the IFTA site, and proposed a “ti” score [25].

The goal of TCMR treatment is to stop rejection and thus return kidney allograft function to as close to baseline (before acute TCMR) as possible. When Banff grade IA rejection is encountered, glucocorticoids alone are the mainstay of treatment [26]. For patients with Banff grade IB rejection, rabbit ATG (rATG) is added to glucocorticoid pulse therapy. The most important predictor of acute TCMR outcome is the arterial lesion status [27]. Thus, for patients with Banff grade II or III rejection (endarteritis), rATG plus glucocorticoids is recommended. However, 39% of such patients exhibited persistent TCMR that exceeded the Banff “BL” criterion 2–9 months after anti-rejection therapy [26]. There is no standard treatment for chronic active TCMR; the protocols vary among centers [10]. A recent study showed that chronically active TCMR responded variably to immunosuppressive therapy [28]; further research is required. There is no consensus treatment for BL patients. Some centers do not use higher tacrolimus trough concentrations to treat rejection. Other centers view BL status as acute TCMR because untreated BL cases can progress to frank rejection [29].

Molecular pathology has become increasingly important. C4d-negative ABMR was described in 2013; the TCMR and ABMR gene sets were defined in 2015; the nCounter technique was presented in 2017. Pathological diagnoses are limited in terms of mechanism and function. For example, as we have discussed above, the mechanisms behind IFTA are diverse, including antibody-related rejection, cellular rejection, CNI toxicity, BK virus nephropathy, reflux nephropathy, and so on. RT-PCR data from raw kidney biopsy specimens are poorly reproducible. nCounter does not require the conventional exhaustive analysis methods of “target gene refinement by DNA microarray” and “confirmation by real-time PCR,” eliminating technical complications such as enzymatic reactions and repetitive reactions. In addition, since gene expression can be analyzed from paraffin sections, it is possible to add immunostaining and other analyses to the same tissue, which has the advantage of being highly reproducible and reliable. nCounter explores gene expression in 20-μm paraffin sections of renal biopsy specimens in FFPE blocks. The sections are placed in 1.5-mL tubes, mixed with a deparaffinized RNA isolation solution and evaluated. ABMR specimens are used for validation; RT-PCR data correlated with the nCounter data. Only the “ptc” score correlated with the Banff histological lesion on qRT-PCR, but the nCounter data correlated with the “g,” “cg,” “ptc,” “i,” “t,” “ti,” “ci,” and “ct” scores [30]. The three genes encoding the von Willebrand factor, the Duffy antigen receptor for chemokines, and caveolin-1 were overexpressed during chronic antibody-associated rejection in nonhuman primates [31], consistent with our previous reports [32‒34]. nCounter is unique in that [1] it analyzes the same samples used for optical microscopy and [2] it can be used to verify long-term allograft survival in large retrospective studies. In the recent Banff meeting, 1,749 genes (excluding duplicates) from 2,521 publications were identified by a PubMed search, of which 1,050 had been described by Banff members; the affected cellular pathways were extensively discussed. The Banff-Human Organ Transplant (B-HOT) panel listed 770 genes involved in rejection, tolerance, and other processes, and will be validated at the next Banff conference [35]. Another new technology is deep learning that digitizes major structures in PAS-stained specimens. In a recent study, features of kidney allograft biopsy specimens were correlated with the Banff score. The “%Interstitium” area correlated with the pathologists’ scores (0.81, p < 0.001) as did “ci,” “ti,” and “IFTA,” although there is room for improvement [36]. With advances in deep learning, it may be possible to obtain a highly accurate diagnosis simply by digitally capturing PAS staining.

We first emphasize the importance of kidney allograft pathology, followed by an overview of the Banff 2019 classification, the importance of TCMR, changes in the Banff classification of TCMR, TCMR treatment, and future prospects involving molecular pathology and deep learning. It is expected that the Banff classification will evolve further, incorporating new technologies but continuing to define lesions, establish working groups, validate the findings, and revise the classification. This will ensure that the Banff classification remains the global diagnostic standard.

The authors have no conflicts of interest to declare.

This work was not supported by any funding sources.

Izumi Yamamoto and Mayuko Kawabe drafted and revised the manuscript. Izumi Yamamoto, Mayuko Kawabe, Hiroyasu Yamamoto, Ayaka Hayashi, Akimitsu Kobayashi, and Takashi Yokoo contributed to the preparation of the manuscript and approved the final version.