- Top of page
Multiple cell types infiltrate acutely rejecting renal allografts. Typically, monocytes and T cells predominate. Although T cells are known to be required for acute rejection, the degree to which monocytes influence this process remains incompletely defined. Specifically, it has not been established to what degree monocytes impact the clinical phenotype of rejection or how their influence compares to that of T cells. We therefore investigated the relative impact of T cells and monocytes by correlating their presence as measured by immunohistochemical staining with the magnitude of the acute change in renal function at the time of biopsy in 78 consecutive patients with histological acute rejection. We found that functional impairment was strongly associated with the degree of overall cellular infiltration as scored using Banff criteria. However, when cell types were considered, monocyte infiltration was quantitatively associated with renal dysfunction while T-cell infiltration was not. Similarly, renal tubular stress, as indicated by HLA-DR expression, increased with monocyte but not T-cell infiltration. These data suggest that acute allograft dysfunction is most closely related to monocyte infiltration and that isolated T-cell infiltration has less acute functional impact. This relationship may be useful in assigning acute clinical relevance to biopsy findings.
- Top of page
Renal allograft infiltration by mononuclear leukocytes is a histological hallmark of acute renal allograft rejection (AR). Using the Banff classification scheme for AR (1), pathologists score the intensity and distribution of infiltrating cells to assess the severity of a rejection episode specifically noting the degree of interstitial cell infiltration, the translocation of infiltrating cells across the tubular basement membrane, tubulitis and the invasion of blood vessel walls, vasculitis. T cells are known to be important in the pathogenesis of AR, and indeed this lymphocyte often constitutes a predominant infiltrating cell type. Still, multiple other mononuclear cell sub-populations, including B cells, NK cells, plasma cells and monocytes/macrophages also variably contribute to the composition of infiltrates such that a histological diagnosis of AR can result from widely heterogenous infiltrating cell populations. However, the phenotype of infiltrating cells is not formally taken into consideration in the histological assessment of AR. Although necessary for the histological diagnosis of AR, cellular infiltration per se is not sufficient to influence functional parameters such as azotemia and proteinuria. Indeed, studies on biopsies taken for protocol surveillance of stable patients report a 10–30% incidence of sub-clinical rejection (SCR), that being histological infiltration and AR scored by the Banff criteria (IA or higher) without significant allograft dysfunction (2–4). The clinical entity of SCR thus demonstrates that a significant inflammatory infiltrate reaching the histological threshold for rejection (Banff IA or higher) is not always associated with clinically detectable allograft dysfunction. Thus, despite general associations between T-cell tubulitis and clinical rejection, T-cell infiltration and tubulitis can occur without any allograft dysfunction.
The lack of strict cause–effect relationship between lymphocytic infiltration and allograft dysfunction suggests that lymphocytes may not be the sole or even main culprit of renal dysfunction during a rejection episode, regardless of their role in initiating rejection. Effector mechanisms other that contact dependent T cellular cytotoxicity may be important or perhaps even predominant in mediating renal tubular dysfunction and reduced creatinine clearance. Consistent with this, we have recently noted that early clinical rejection-like episodes occurring in patients receiving vigorous T-cell depleting induction therapies are associated with substantial renal dysfunction despite being associated with minimal lymphocytic infiltrates that fail to reach the diagnostic threshold spelled out in the Banff criteria (5). These patients have, however, had monocytic infiltrates. We speculated that the composition of the infiltrate may be as important as its intensity in determining its acute functional consequences. In particular, we hypothesized that macrophages were primarily responsible for the effector elements mediating acute allograft dysfunction during AR. We therefore studied renal allograft biopsies exhibiting various degrees of AR according to Banff criteria, taken from patients with varying degrees of lymphocyte depletion, and associated with varying degrees of renal dysfunction with the aim of investigating the relationship between monocytes and T cells within a cellular infiltrate and their relative impact on clinical allograft dysfunction. We find that the monocyte/macrophage component of an infiltrate is more closely associated with the degree of renal dysfunction than the T-cell infiltrate, thus implicating monocytes as a critical effector in clinical AR.
- Top of page
According to the Banff classification, 23 biopsies (29.5%) were classified as borderline (BR), 32 were IA (41%), 16 were IB (20.5%), six were IIA (8%) and one was IIB (1%) (Figure 1). The distribution of immunohistochemistry scores is reported in Figure 2. All patients had cellular infiltration although the degree and type of infiltration varied substantially across the population. Similarly, HLA-DR expression was widely variable throughout the studied biopsies. The degree of functional impairment was similarly wide with the PFC ranging from 0.82 to 6.95 (median 1.09). This wide range of T cell and monocyte prevalence combined with the broad range of clinical dysfunction facilitated the analysis of the import of particular cell type on PFC.
Figure 1. Distribution of Banff classification. Shown are the number of biopsies (y-axis) achieving a particular Banff rejection score (x-axis).
Download figure to PowerPoint
As expected, the degree of allograft dysfunction at the time of biopsy, as measured by logPFC, was directly correlated with the Banff classification severity (Table 2; Figure 3). The median logPFC increased from 0.037 (range −0.183–0.560) in the borderline class to 0.654 (0.143–1.943) in the IIA/IIB class, and this correlation was highly significant (Pearson correlation coefficient = 0.45; p < 0.0001) (Figure 3, Tables 2 and 3).
Table 2. LogPFC according to Banff classification
|IIA/IIB|| 7||0.654||0.821||0.694|| 0.143||1.943|
Figure 3. Correlation between Banff class and logPFC. A statistically significant correlation (p < 0.0001) between Banff score and the degree of dysfunction as measured by PFC was found.
Download figure to PowerPoint
Table 3. Correlation between LogPFC and Banff or immunohistochemistry scores
| ||N||Pearson correlation|
|CD68||68|| 0.29|| 0.0159|
On immunohistochemistry, both scores of CD68 positive cell infiltration and HLA-DR expression were significantly correlated (p = 0.0159 and <0.0001 respectively) with the degree of functional impairment of the graft (Table 3; Figures 4A and B). Thus, as macrophage infiltration increased, allograft function worsened, and as HLA-DR expression on the renal tubules intensified, allograft function worsened. However, this was remarkably and unexpectedly not the case when the degree of lymphocytic infiltration was considered. Indeed, although all biopsies were infiltrated with T cells, T-cell infiltration as measured by CD3 expression was significantly (Correlation Coefficient −0.29, p = 0.0101) inversely correlated with allograft dysfunction (Table 3; Figure 5) suggesting that when rejection was present, a prominent T-cell infiltrate had a mitigating effect on allograft dysfunction. Consistent with this, tubular DR expression was strongly correlated with CD68 expression (Correlation Coefficient 0.42, p = 0.0004), and negatively correlated with CD3 expression (Correlation Coefficient −0.38, p = 0.0009). Similar inverse relationships were seen as trends between logPFC and CD4 and CD8 that failed to reach statistical significance (Table 3). Thus, while T-cell infiltration clearly was, as expected, a hallmark of AR in these patients, once AR was established, the magnitude of the dysfunction was not a product of intensifying T-cell infiltration, and objectively, T-cell intensity was correlated with less change in renal function.
Figure 4. Correlation between logPFC and CD68 and HLA-DR scores. A statistically significant correlation between the degree of dysfunction as measured by logPFC and both the intensity of (A) the monocyte infiltrate (CD68, p = 0.0159), and (B) tubular expression of HLA-DR (p < 0.0001).
Download figure to PowerPoint
Figure 5. Correlation between logPFC (y-axis) and lymphocyte immunohistochemical scores (x-axis). A statistically significant negative correlation was present between the degree of dysfunction as measured by logPFC and (A) the intensity of the T-cell infiltrate (p = 0.0101). Negative trends between PFC and (B) CD4 and (C) CD8 were also seen that failed to reach statistical significance (p > 0.2).
Download figure to PowerPoint
Interestingly, when the composition of the infiltrate was compared to the Banff score, CD68 expressing cell infiltration was positively correlated with an advancing Banff score while CD3 expression was not (Table 4). This was indicative of a rather constant presence of CD3 positive T cells that did not advance as the Banff score advanced, and was not due to an absence of T cells. CD4 expressing cell infiltration was correlated with Banff score, but this was related to the presence of CD4 on monocytes and not due to an increasing number of CD4 expressing CD3 cells. There was, however, a significant shift of the T-cell infiltrate towards CD8 expression as the Banff score worsened.
Table 4. Correlation between Banff grade and immunohistochemistry scores
| ||N||Pearson correlation|
When biopsies were considered according to the type of induction immunosuppression (depletion versus non-depletion), there was no difference in the Banff scores between depleted and non-depleted patients and further there was no histological difference in the severity of AR as expressed by the individual ‘t’ (tubular),‘i’ (interstitial) and ‘v’ (vascular) scores of the Banff classification (Table 5). As would be predicted, there were fewer T cells in the biopsies from depleted patients with significantly lower scores related to lymphocytic infiltration (CD3, CD4 and CD8) in the depletion group compared to non-depletion (p = 0.0002, 0.0508 and 0.0139 respectively). On the contrary, there was no significant difference in CD68 score between depleted and non-depleted patients (Table 5). Thus, the degree of monocytic infiltration was relatively consistent amongst rejecting biopsies regardless of the induction regimen, while the T-cell infiltrate varied with peripheral T-cell depletion. While the majority of both depleted and non-depleted biopsies in our data set were within the first year posttransplant, the depleted group was numerically higher in the first year (88% versus 64% of biopsies studied). LogPFC was not different between cases prior to or after 1 year but monocytic biopsies were more prevalent in the depleted group and in the first transplant year. However, when the analysis between immunohistochemical scores and logPFC were performed excluding all biopsies beyond 1 year of transplant, all relationships remained statistically significant and similar in magnitude (not shown, supplementary data provided).
Table 5. Histological and immunohistochemistry comparison between depleted and non-depleted patients
|N||Median||Mean||Std Dev||N||Median||Mean||Std Dev|
The positive correlation between allograft dysfunction expressed by logPFC and the Banff classification, CD68 score and HLA-DR score remained statistically highly significant in depleted patients (p = 0.0007, 0.0419 and 0.0024 respectively; Table 6). Interestingly, in non-depleted patients, these relationships were not significant (Table 6). Also, there was no significant correlation between CD3, CD4, CD8 scores and logPFC in either depleted or non-depleted patients. Of note, T-cell infiltration and logPFC remained inversely related in both depleted and non-depleted patients. No associations between logPFC and Banff or immunohistochemical score were observed based on time posttransplant although the T-cell infiltrate was significantly reduced in depleted patients within the first 6 months posttransplant (not shown). There was no difference in Banff score, immunohistochemical scoring or logPFC that was referable to the cause of renal failure, specifically considering immune versus non-immune causes of renal failure (supplemental data provided).
Table 6. Correlation between LogPFC and histology and immunohistochemistry scores, by depletion
|n||Pearson correlation coefficients||p||n||Pearson correlation coefficients||p|
|Banff||45|| 0.49||0.0007||33|| 0.43||0.0136|
|CD68||43|| 0.31||0.0419||25|| 0.13||0.5514|
|HLA_DR||42|| 0.46||0.0024||32|| 0.29||0.1044|
- Top of page
T cells have long been known to be necessary for renal AR, and this is supported by, among other things, the fact that athymic animals do not reject allografts. Accordingly, T cells have also been presumed to be the predominant mediator of the renal dysfunction associated with acute rejection. However, direct evidence for this is sparse. It is clear that T-cell-mediated cellular cytotoxicity is a means of parenchymal cell destruction in vitro and in vivo. However, it has been equally well established that T cells can and often do infiltrate an allograft without causing any clinically detectable phenotype (2,3). Furthermore, the degree of allograft dysfunction seen in rejection can be disproportionately high relative to the T-cell content of an infiltrate, particularly in patients that have undergone aggressive T-cell depletion (5). Furthermore, recent serial observations in mice (9) and protocol biopsy evidence in humans (5) have shown that tubulitis is a rather late manifestation of rejection and renal dysfunction can develop before tubulitis or evidence of renal cell necrosis. In addition, following the treatment of an acute rejection episode, the T-cell infiltrate persists long after the graft has resumed normal function. These observations suggest that the clinical phenotype of rejection is determined through the effects of intragraft elements other than T cells.
In this study, we have investigated the potential role of monocytes in mediating the clinical dysfunction associated with histological acute rejection. We have done so because monocytes infiltrate allografts as do T cells and have not been studied in experimental isolation, mainly because there are no robust animal or human models of monocyte deficiency. It has also been demonstrated experimentally that parenchymal injury can occur merely by proximity to an inflammatory nidus (10), likely through the effects of cytokines, and many monocyte-derived cytokines are markedly amplified within the allograft during rejection episodes (7). We have included biopsies from patients with and without profound T-cell depletion and those with normal and impaired function to be able to assess variables in infiltrate magnitude, content and relevance. Importantly, the patient population involved has been under close surveillance to ensure that the dysfunction under evaluation is acute. Furthermore, all rejections were initial presentations of rejection. Thus, the findings speak to acute episodes of rejection and not chronic changes. We find that CD68 mononuclear cells constitute a significant component of the inflammatory infiltrate during acute AR and that the intensity of the monocytic infiltrate, but surprisingly not that of the T-cell infiltrate, is closely related to the degree of allograft dysfunction. These observations have several notable implications.
The current Banff classification system for AR assesses the degree of cellular infiltration without differentiating the phenotype of the infiltrate or considering clinical dysfunction. The utility of this approach with respect to assessing the clinical importance of the histological process is borne out in this study by the direct correspondence between Banff score and the degree of allograft dysfunction. Even in a population such as ours chosen to include a significant amount of SCR the Banff scoring system allows the correlation between general infiltrate and graft dysfunction to be recognized. However, it has led to a situation whereby the pathologist's diagnosis cannot directly assess the clinical relevance of a particular rejection episode since the correspondence between the general cellular infiltrate and allograft dysfunction is not direct. Indeed, interstitial cellular infiltrates have long been recognized to be present in early biopsies of normally functioning renal allografts (11), and have been demonstrated in tolerant patients and experimental animals (12,13). Thus, assessment of other aspects of the infiltrate could provide a more accurate assessment of the clinical situation.
The heterogeneous composition of the cellular infiltrate during acute AR has long been recognized to include monocytes (14–16) and non-specific responses have been considered to contribute to the complex host immune response to the allograft (17). Furthermore, the aggregate contribution of different cell types has been investigated in previous studies, and these studies have correlated increasing heterogeny with worsening outcome (15,18). More recently, both experimental (19,20) and clinical studies (21–26) have reported on the role of macrophages during AR (review in 27). In particular, Grimm et al. (28). showed that activated macrophages and their products characterized acute dysfunction during rejection compared to normal histology and to subclinical rejection, and Kajiwara et al. reported increased CD68/leukocyte common antigen ratio in addition to increased HLA DR expression and increased granulocyle-monocyte colony-stimulating factor during acute rejection compared to borderline changes and chronic rejection (29). In our study, we have endeavored to distinguish the relative contribution of T cells or monocytes considered independently, and to do so in acute terms, assigning a dynamic measure, the PFC, as opposed to a static measure of function to each biopsy. With this approach, we have not only corroborated the importance of infiltrate heterogeny seen in previous studies, but have found that it is the CD68 monocyte component of the infiltrate that most closely predicts the acute clinical scenario. Strikingly, CD3 T cells had no demonstrable acute association with worsening renal function beyond their ubiquitous presence in rejection, but rather showed an inverse relationship with rising creatinine, and an inverse association with tubular DR expression. These data are consistent with a view of T cells as both instigating and controlling elements in rejection and fit well with the observations that regulatory expression of genes, such as CD152 and FoxP3 are associated with clinical but not sub-clinical rejection episodes (7,30). As parenchymal injury increases, the prevailing need is to limit immune damage to maintain homeostasis.
These observations suggest that while acute cellular cytotoxicity can occur during rejection, a predominant cause of acute renal dysfunction during rejection may be related to monocyte-derived cytokines. We have previously reported that clinically relevant rejection episodes in depleted and non-depleted patients are, compared to quiescent biopsies, transcriptionally rich in monocyte-associated cytokines, such as TNF-α (5,7). We have also reported that macrophage infiltration distinguishes sub-clinical from clinical infiltrates in non-human primates treated with CD40:CD154 blockade (13), a critical pathway in monocyte activation and cytokine release (31). These monocytic infiltrates have been noted recently to be surrounded by radial patterns of tubular HLA-DR expression suggesting a diffusible mediator (5). Many investigators have reported increased parenchymal DR expression during AR (15,16,32). This study now correlates tubular HLA-DR expression with CD68 monocyte content and directly associates this process with allograft dysfunction. Given that tubular HLA-DR expression was the parameter most strongly associated with the degree of renal dysfunction, one is drawn to consider the potential implications of broader DR expression. The increased expression of DR antigens on renal tubular cells during rejection could broaden the repertoire of T cells capable of eliciting chronic delayed-type cellular cytotoxicity responses, and could similarly broaden the influence of class II specific alloantibody. Thus, the increased risk for AR associated with donor class II specific allosensitization, could be driven in large part by increased antigen expression in addition to augmented antibody or effector cell production.
It is important to underscore that these data do not speak to the prognosis of histological findings with regard to response to therapy. All rejections in this study were reversible, however, the treatments were individualized. Treatment therefore was not a standardized variable stratified by histology. It will be interesting to determine whether monocyte content identifies a patient group more likely to respond to a particular therapeutic regimen, but such an assessment will require an additional prospective study.
Based on these studies and the others cited herein, we favor a model in which a relatively small number of T cells are necessary to induce a rejection episode. Their acute effects are locally amplified through activated monocytes and the release of soluble elements like cytokines that impair parenchymal cell function. Inhibition of monocyte activation, either through cytokine sequestration, blockade of the CD40 pathway or drugs with prominent monocyte effects, such as corticosteroids, limit the acute effects of a rejection episode but may not necessarily prevent ongoing T-cell effects. This may explain the exceptionally preserved function in anti-CD154 treated animals despite T-cell infiltration that then gives way to chronic graft fibrosis, and monocyte inhibition in the absence of concomitant T-cell inhibition may be a more general mechanism of chronic allograft nephropathy. Left unchecked, T-cell cytotoxicity can lead to insidious renal tubular destruction that in and of itself has little acute functional consequence, but over time, even in the absence of acute dysfunction, produces the late immune sequelae of tubular drop out and eventual fibrosis.