Infiltrates in Protocol Biopsies from Renal Allografts


* Corresponding author: Michael Mengel,

†Both authors contributed equally to this work.


In renal transplantation, clinical decisions are based primarily on the Banff classification of biopsies. However, the incorporation of `minor or nonspecific' cellular infiltrates into the Banff classification and their interpretation is uncertain.

We analyzed 833 protocol and 306 indicated biopsies to test whether such infiltrates are harmless or whether they have a bearing on outcomes. We characterized morphology, localization and cellular composition of infiltrates, and correlated these findings to the Banff classification and allograft outcome.

We found that protocol biopsies had the same prevalence of infiltrates as indication biopsies (87% vs. 87%). Diffuse cortical infiltrates, the hallmark of cellular rejection were more common in indication biopsies and related to tubulitis and a rise in serum creatinine. However, in biopsies with cellular rejection according to Banff criteria, we observed an increase in all infiltrate types (specific and nonspecific) and all cell types (T cells, B cells, histiocytes). The only predictor of allograft function outcome was persistent inflammation in sequential biopsies, irrespective of type, localization and composition of the cellular infiltrates.

As detected by sequential biopsies, persistence of any inflammation including those infiltrates currently not considered by the Banff classification should be regarded as a morphological correlate of ongoing allograft damage.


Long-term renal transplant results have not changed considerably in the past two decades (1,2), even though the incidence of acute rejection episodes in the first year has continuously decreased (3). However, such episodes are still a major risk factor for long-term allograft dysfunction (4,5). Protocol biopsies that are obtained at predetermined intervals, could uncover subclinical rejection and thereby improve long-term outcomes (6–14). Such biopsies could reveal hidden rejection that is missed by `indicated' biopsies obtained solely because of deteriorating renal function. Transplant renal biopsies are generally evaluated by the Banff classification, a grading scale established by experienced nephropathologists and transplant clinicians (15,16). The Banff classification system is universally applied in the setting of multicentric drug trials and according to the Banff criteria, cellular infiltrates are considered to be specific for acute cellular rejection in terms of localization (within the cortex) and pattern (infiltrative in between nonatrophic tubules). Infiltrates in areas of tubular atrophy and interstitial fibrosis, around large vessels, in the subcapsular zone of the renal cortex, or circumscribed nodular cell aggregates are regarded as nonspecific. Hence, these infiltrates should not be considered for the Banff i-score and thus for the diagnosis of acute cellular rejection. Furthermore, the infiltrative cell type is not considered in the diagnosis. However, at least 10% of the cortical area must show infiltrate and tubulitis (1–4 cells/tubular cross section) for a borderline (`suspicious for') rejection diagnosis. More than 25% of specific interstitial infiltrate and at least moderate tubulitis (5–10 cells/tubular cross-section) are required for the diagnosis of acute cellular rejection. Minimal infiltrates being quantitatively below these thresholds are usually ignored and regarded as having no clinical implications. However, other classification systems suggested applying even lower cut-offs to make the diagnosis of acute rejection (17). Nonetheless, protocol biopsy centers maintain that so-called `nonspecific' infiltrates in areas of cortical atrophy may not be harmless (18,19). We aimed to elucidate the role of these nonspecific and minimal (i.e. below the Banff threshold to be diagnostic) cellular infiltrates in allograft renal biopsies.

Material and Methods


The Hannover Protocol Biopsy Program was begun in December, 2000 (9,20). The program is part of the routine medical care following renal transplantation. The University of Hannover Ethics Committee approved the protocol, data collection and analyses. Patients were enrolled after written informed consent was obtained. Protocol biopsies were carried out at 6, 12 and 26 weeks after transplantation as an out-patient procedure using 18- or 16-gauge automated biopsy needles guided by ultrasound (9,21).

All other biopsies not taken at the above-mentioned protocol biopsy time points were regarded as indication biopsies and taken either due to allograft dysfunction (i.e. any rise in the serum creatinine) or from initially nonfunctioning transplants as determined by the respective physician. Demographic and clinical data as well as routine laboratory results were collected corresponding to the biopsy time points and yearly posttransplantation. Biopsies were rapidly processed and embedded in paraffin on the same day. Serial sections stained for hematoxylin and eosin and periodic acid–Schiff were reviewed according to the updated Banff classification system (15,16). A preliminary diagnosis was transmitted to the physician within 4 h and had direct therapeutic consequences as described in detail elsewhere in those cases with respective biopsy findings (9,22). C4d stain was done overnight and reviewed the next day leading to the final written report. In the present study, biopsies taken during the period between December, 2000 and March 2004 in the setting of the Hannover biopsy program were included.

Infiltrate categories

One thousand four hundred and fifteen renal allograft biopsies comprising 1020 protocol and 395 indication biopsies from 428 patients were taken during the study period. Three hundred forty-one protocol biopsies were obtained at 6 weeks, 355 protocol biopsies at 12 weeks and 324 protocol biopsies at 26 weeks after transplantation. From 1139 biopsies (833 protocol and 306 indication biopsies) sufficient material (comprising renal cortical tissue with at least one artery) was available for a re-review according to the following descriptive characteristics of the interstitial infiltrates: (1) infiltrate pattern, i.e. infiltrate present, unifocal infiltrate, multifocal infiltrates (2) infiltrate type, i.e. diffuse (Figure 1A) in >50% of nonfibrotic cortical area with infiltration in between nonatrophic tubules, nodular (Figure 1B) well circumscribed without infiltration in between adjacent tubules, focal but raggedly (Figure 1C) outlined (i.e. focal infiltrate with infiltration in between nonatrophic tubules at its periphery), between/adjacent to atrophic (Figure 1D) tubules and/or fibrotic areas; (3) localization of the infiltrate, i.e. subcapsular with direct contact to the organ capsule, perivascular adjacent to larger arteries or venules, cortical within the renal cortex. In biopsies where several infiltrate characteristics were present, all were documented in a SPSS database for further analysis.

Figure 1.

(A) Photomicrograph of a typical diffuse cortical cellular infiltrate with tubulitis representing the morphological correlate of acute cellular rejection according to the current Banff classification. (B) Example of nodular cortical infiltrate with numerous B cells in an immunohistochemical CD20 stain. (C) Raggedly outlined cortical infiltrate with cellular invasion in between adjacent nonatrophic tubules and tubulitis. (D) Cellular infiltration in an area of tubular atrophy and interstitial fibrosis. Tubulitis can be observed. Such infiltrates are currently ignored according to the Banff classification.


From 126 patients with a total of 322 protocol biopsies (126 6-week biopsies, 115 12-week biopsies, 80 26-week biopsies) further material of paraffin embedded specimens was available for immunohistochemistry. Paraffin sections were stained for cytotoxic T cells (Granzyme B, clone GrB-7; Chemicon International, Temecula, CA), B cells (anti-CD20, clone L26; DAKO), histiocytes (anti-CD68, clone KP1; DAKO) and activated T-lymphocytes (anti-ZAP70, clone 2F3.2; Upstate Inc., Charlottesville, VA). Stained slides (n = 1258) were evaluated in areas of most dense infiltrates by counting the number of positive cells in three high power fields (400× magnification) and calculating the mean number of positive cells per biopsy. Results were correlated to Banff grade of rejection and the morphological infiltrate category.

Statistical analysis

We used the SPSS statistical software package, version 12.0.1 (SPSS Inc., Chicago, IL). Comparisons of categorical data between different groups and biopsy time points were performed with Fischer's exact test or the chi-square test. Numerical data were compared with the U-test and, for multiple comparisons, with the Kruskal–Wallis test. A general linear model (SPSS-GLM procedure: univariate analysis of variance) was applied, with the calculated creatinine clearance (Crockroft–Gault formula) at 1 year after transplantation as the dependent variable, the different combinations of infiltrate types in each patient as fixed categorical factors and the total number of infiltrates in each patient as a covariate factor. In addition, we calculated linear regression to examine the relationship between the `sum of infiltrates' (for definition see last paragraph of results) and graft function (calculated creatinine clearance) at 1 and 2 years after transplantation. p-values of <0.05 were considered as significant.


Histological classification according to Banff

The majority (65.3%) of protocol biopsies were taken from clinically stable allografts that showed no signs of acute rejection according to the Banff classification. 10.5% of protocol biopsies (30.4% of indication biopsies) were taken from allografts with an increase in serum creatinine concentration at the time point of biopsy but showed histologically changes being below the Banff threshold for the diagnosis of acute cellular rejection. Subclinical (without concomitant creatinine increase) borderline changes were found in 13.9% of protocol biopsies. 10.8% of the indication biopsies were obtained from allografts with initially delayed graft function and revealed borderline changes at the histological review. Subclinical Banff grade I rejection episodes were found in 5.4% of protocol biopsies. Clinically, rejection episodes with a creatinine increase were more frequent in indication biopsies (19.6%) than in protocol biopsies (2.0%). Vascular cellular rejection episodes (Banff grade II) and active humoral rejection (C4d-positive) were found in a low number of indication biopsies and very rarely in protocol biopsies.

Infiltrate categories

More than 85% of all analyzed protocol and indication biopsies showed cellular infiltrates (Table 1). Diffuse interstitial infiltrates were significantly more common in indication biopsies (25.2%) than in protocol biopsies (11.8%; p = 0.001). In contrast, nodular cellular aggregates were significantly more frequently observed in protocol biopsies (29.1%) (indication biopsies 17.0%; p = 0.001). Nodular infiltrates were detected significantly more often in late protocol biopsies (12 and 26 weeks after transplantation) than in early protocol biopsies (6 weeks, p = 0.032). All other infiltrate types were equally prevalent in both biopsy types.

Table 1.  Infiltrate pattern and type in protocol and indication biopsies
InfiltrateProtocol biopsies (n = 833)Indication biopsies (n = 306)p-value
  1. ns = not significant.

Infiltrate present86.8%87.3%ns

Infiltrate localization

Except for perivascular nodular infiltrates in indication biopsies, all documented infiltrate types were most frequently found within the cortex, followed by the perivascular localization (Table 2). The subcapsular localization was rare. A significantly higher percentage of protocol biopsies (15.3%) displayed cortical nodular infiltrates compared to indication biopsies (9.5%; p = 0.01). Raggedly outlined infiltrates were less frequently found around blood vessels in protocol biopsies (37.4%) compared to indication biopsies (46.7%; p = 0.004). All other infiltrate types were equally localized in both biopsy types.

Table 2.  Infiltrate localization in protocol and indication biopsies
InfiltrateProtocol biopsies (n = 833)Indication biopsies (n = 306)p-value
  1. ns = not significant.


Infiltrates and Banff classification of acute cellular rejection

82.5% of protocol biopsies (Table 3A) had interstitial infiltrates that were quantitatively and/or qualitatively not sufficient to make the diagnosis of `borderline' or cellular rejection according to the Banff classification system. This percentage was similar in indication biopsies (81.1%, Table 3B).

Table 3A.  Infiltrates and Banff classification in protocol biopsies
InfiltrateProtocol biopsies no rejection (n = 619)Protocol biopsies borderline (n = 149)Protocol biopsies rejection1 (n = 65)p-value2p-value3
  1. 1Grade I and II cellular rejection according to the Banff classification.

  2. 2No rejection vs. borderline.

  3. 3No rejection vs. rejection.

  4. ns = not significant.

Infiltrate present82.5%100%100%0.00010.0001
Table 3B.  Infiltrates and Banff classification in protocol biopsies
InfiltrateIndication biopsies no rejection (n = 190)Indication biopsies borderline (n = 49)Indication biopsies rejection1 (n = 67)p-value2p-value3p-value4
  1. 1According to the Banff classification; left column grade I, right column grade II cellular rejection.

  2. 2No rejection vs. borderline.

  3. 3No rejection vs. Banff grade I rejection.

  4. 4No rejection vs. Banff grade II rejection.

  5. ns = not significant.

Infiltrate present81.1%100%100%100%0.00020.0002ns

A diffuse infiltration pattern was significantly more frequently observed in biopsies with a diagnosis of acute rejection according to the Banff classification (protocol biopsies: 6.6% vs. 40.8%, p = 0.005; indication biopsies 8.9% vs. 51.0%, p = 0.0001). Similar results were obtained for nodular infiltrates, which were significantly more frequent in protocol (p = 0.0001) and in indication biopsies with signs of cellular rejection (p = 0.0008). Nodular infiltrates were rarely present before the onset of borderline rejection, but significantly (p = 0.0006) more common after a borderline episode. No statistical correlation between nodular infiltrates and acute humoral rejection episodes (C4d-positive) or a specific immunosuppressive medication at the time point of biopsy was found.

Biopsies (protocol and indication) with signs of cellular rejection showed also a significant increase in raggedly outlined infiltrates as well as infiltrates in areas of tubular atrophy (p < 0.05). In protocol biopsies with and without rejection subcapsular localized infiltrates of any pattern revealed no significant difference between these two groups, while in indication biopsies only subcapsular localized raggedly outlined infiltrates were significantly more frequent in biopsies with rejection (p < 0.05).

Infiltrates and tubulitis, polyoma, calcineurin inhibitor toxicity

In nearly 70% of protocol biopsies from stable allografts infiltrates were present without concomitant tubulitis (Figure 2A). Sixty percent of protocol biopsies from allografts with an incidental rise in serum creatinine at the time point of the biopsy had infiltrates without tubulitis. This constellation (infiltrate without tubulitis) was found in 56.3% of indication biopsies from allografts with dysfunction (Figure 2A). Diffuse infiltrates (Figure 2B) were significantly (p < 0.0001) more frequent in biopsies with tubulitis. Higher grades of tubulitis were seen in biopsies (protocol and indication) from allografts with a simultaneous rise in serum creatinine. Nodular infiltrates were more frequently seen with concomitant tubulitis as well as with higher grades of tubulitis in biopsies from allografts with dysfunction (Figure 2C). Most biopsies with raggedly outlined infiltrates had no simultaneous tubulitis (Figure 2D). For infiltrates in areas of tubular atrophy (Figure 2E) higher frequencies and grades of tubulitis were observed in indication biopsies compared to protocol biopsies with and without simultaneous rise in the serum creatinine. It should be borne in mind that, according to the current Banff classification we recorded only tubulitis in nonatrophic tubuli.

Figure 2.

(A–E) The relationship between the Banff grades of tubulitis and the analyzed infiltrate types is shown separately for protocol biopsies with and without a rise in the serum creatinine at the time point of the biopsies as well as for indication biopsies.

Only six of the 1139 biopsies showed polyomavirus nephropathy by immunohistochemistry. All six had infiltrates in areas of atrophic tubules as well as raggedly outlined infiltrates, none had a diffuse infiltrate. Two biopsies showed additional nodular infiltrates.

No statistically significant relationship between any infiltrate type and the presence of histological signs of calcineurin inhibitor toxicity was found.


A diffuse infiltration pattern (protocol and indication biopsies grouped together) was dominated by histiocytes (CD68-positive) and was accompanied by high median numbers of Granzyme B-positive cytotoxic T-lymphocytes. Biopsies with nodular infiltrates showed the highest median density of infiltrating B cells (CD20-positive), accompanied by a comparable number of activated T cells (ZAP 70-positive). A detailed re-review of biopsies with nodular infiltrates revealed that within the nodular cell aggregates dense B-cell groups were accompanied by nearly equal numbers of T cells in the immediate vicinity. In addition, scattered T cells and histiocytes were always intermixed with B cells in the nodules. Biopsies with raggedly outlined infiltrates and infiltrates in areas of atrophic tubules displayed an almost equal cellular composition with a slight predominance of ZAP 70-positive activated T cells besides high median numbers of B cells and histiocytes.

Analyzing the cellular composition and density according to the Banff classification, biopsies with infiltrates not sufficient to make the diagnosis of rejection generally displayed lower median numbers of all investigated cell types. A continuous increase in the median number of T cells, histiocytes and B cells was observed from biopsies where the diagnosis was `no rejection' to biopsies with borderline changes and even further to biopsies with cellular rejection.

Infiltrates and allograft function

The whole biopsy collective comprised protocol biopsies from stable allografts as well as from organs which had an incidentally significant rise in the serum creatinine at the predetermined time point of the protocol biopsy. Analyzing the influence of infiltrate type and localization on allograft function, we compared these biopsies separately with indication biopsies from allografts where a significant rise in serum creatinine was documented and led to a nonpredetermined biopsy (Table 4).

Table 4.  Infiltrates and allograft function
InfiltrateProtocol biopsies without rise in creatinine (n = 630)Protocol biopsies with rise in creatinine1 (n = 113)Indication biopsies with rise in creatinine1 (n = 53)p-value
  1. 1Rise in serum of >20% over baseline creatinine at the time point of biopsy.

  2. ns = not significant.

Infiltrate present86.7%84.1%90.6%ns

86.7% of biopsies from clinically stably functioning renal allografts displayed interstitial cellular infiltrates. This percentage was the same as that observed in protocol biopsies (84.1%) from allografts with overt dysfunction at the time point of biopsy (rise in serum creatinine of at least 20% compared to previous determinations) and not significantly different from indication biopsies. For the vast majority of analyzed infiltration patterns, types and localizations the prevalence was similar in biopsies from stably functioning allografts and allografts with dysfunction (p-values >0.05, Table 4). An exception was seen for diffuse infiltrates (p = 0.0001) and perivascular localized, raggedly outlined infiltrates (p = 0.013), which were found significantly more often in indication biopsies from allografts with an increased serum creatinine at the time of biopsy. The prevalence of these infiltrates was not increased in protocol biopsies with a coincidental rise in serum creatinine compared to those from clinically stable allografts.

Infiltrate combinations and allograft function

In approximately 75% of all analyzed biopsies, more than one infiltrate type and infiltrate localization were present (Table 5). The combination of raggedly outlined infiltrates and infiltrates in areas of atrophic tubules was the most frequent in protocol as well as in indication biopsies. No significant differences in the incidence of analyzed infiltrate combinations between protocol and indication biopsies were found. The highest increase in serum creatinine at the time of biopsy was observed in biopsies (protocol and indication biopsies together) with a combination of raggedly outlined and diffuse infiltrates (median rise in creatinine 13.4% compared to biopsies without infiltrates 4.9%; p = 0.046). All other infiltration combinations (n = 10) showed no statistically significant differences concerning the median rise of the serum creatinine at the time of biopsy.

Table 5.  Incidence of combinations of infiltrate types in protocol and indication biopsies
Infiltrate combinationProtocol biopsies (n = 833)Indication biopsies (n = 306)
Raggedly + atrophic20.5%22.0%
No infiltrate13.1%13.1%
Nodular + raggedly + atrophic14.4%5.6%
Diffuse + raggedly + atrophic4.4%11.1
Nodular + raggedly5.3%2.6%
Diffuse + other types except raggedly3.1%5.5%
Diffuse + raggedly2.6%5.6%
Nodular + atrophic2.7%2.6%
Diffuse + atrophic1.7%3.0%

Infiltrates and allograft outcome

In the entire patient collective (protocol and indication biopsies together), an univariate analysis of variance (GLM procedure) revealed no significant impact of any specific infiltrate type or infiltrate combination on the creatinine clearance at 1 year after transplantation (p < 0.05; not shown). However, with this statistical approach the number of infiltrates per patients was significantly associated with the clearance 1 year posttransplantation (for further analysis see last paragraph of results).

To exclude any bias due to lacking biopsy results we defined a subcollective of patients having all three protocol biopsies (160 patients with 480 protocol biopsies and 140 indication biopsies = 620 biopsies) and carried out additional outcome analysis with two different end points.

First, for the three protocol biopsy time points we analyzed separately the relationship between the different infiltrate types and the creatinine clearance at 2 years after transplantation. Patients showing in their 6-month protocol biopsy infiltrates in areas of tubular atrophy had a significantly lower (50.6 mL/min) median creatinine clearance than those without such infiltrates (59.2 mL/min; p = 0.05). All other tested biopsy time points (6 weeks and 3 months) and infiltrate types revealed no statistically significant relationship with the 2 year creatinine clearance.

Second, we analyzed as another end point the presence of chronic tubulo-interstitial changes in the third protocol biopsy taken 6 months after transplantation. From 71 patients having chronic tubulo-interstitial changes of at least Banff grade I in their 6-month protocol biopsy, 74.6% showed infiltrates in areas of tubular atrophy in their prior protocol biopsy taken 3 months earlier. In contrast, only 25.4% of patients without inflammation in areas of tubular atrophy in the second protocol biopsy showed Banff grade I chronic changes at 6 months (p = 0.001). None of the other infiltrate types revealed statistically significant differences concerning the onset of chronic tubulo-interstitial changes in the 6-month protocol biopsy and their presence at 3 months. Looking at the first protocol biopsy taken 6 weeks after transplantation, no infiltrate type or localization significantly predicted the presence of chronic tubulo-interstitial changes at 6 months.

Infiltrate number and allograft function

Looking for the number of different infiltrate types present simultaneously in one biopsy revealed no significant differences between protocol and indication biopsies. The same percentages of protocol as well indication biopsies had no, one, two, three and four different infiltrate types. Analyzing the number of different infiltrate types per biopsy in relation to the serum creatinine at the time point of biopsy showed no significant differences between protocol and indication biopsies with no, one, two or three different infiltrates. Only biopsies with simultaneously four different infiltrate types (nodular + raggedly + atrophic +diffuse) had a significant increase in the serum creatinine at the time point of the biopsy (p = 0.03).

Infiltrate number and allograft outcome

Stimulated by the results of the univariate GLM-procedure (see above) we included for the analysis of outcome at 1 and 2 years after transplantation again only patients with all three protocol biopsies were examined (160 patients with 480 protocol biopsies and 140 indication biopsies = 620 biopsies). Analysis of the total number of different infiltrate types per patient, namely all different infiltrate types one patient had in his/her sequential protocol biopsies and, if carried out, indication biopsies within the first year after transplantation (e.g. first protocol biopsy nodular + raggedly outlined infiltrate; second protocol biopsy diffuse infiltrate; indication biopsy atrophic + nodular; third protocol biopsy atrophic: sum of infiltrates = six infiltrates), revealed a weak but significant inverse correlation between the sum of infiltrates and the creatinine clearance recorded at 1 and 2 years after transplantation in a linear regression model (Figure 3; r = 0.28, p = 0.0001). Patients with one or no infiltrate during the first year after transplantation had the best creatinine clearance at 1 and 2 years after transplantation.

Figure 3.

Linear regression between creatinine clearance at 1 and 2 years after transplantation in relation to the sum of infiltrates per patient (for detailed definition see last paragraph of results) present in all biopsies (protocol and indication) within the first year. Median values, quartiles and 95% confidence intervals are shown; r = 0.283, p = 0.0001.


The important finding in our study was that more than 85% of the protocol biopsies taken from clinically stable renal allografts had cellular infiltrates. Also, in more than 80% of all biopsies, either protocol or indication, the infiltrates were quantitatively and/or qualitatively below the current threshold of the Banff criteria to diagnose rejection. These findings support the conclusion from other protocol biopsy studies that an acute change in the serum creatinine is not a reliable marker for what is really going on inside an allograft (12,14,18). Furthermore, our findings will fuel a heated discussion on whether or not such `minor' or `nonspecific' and thus currently ignored infiltrates are harmless or harmful.

No significant differences were found in the frequency of infiltrate types between protocol and indication biopsies, except for nodular and diffuse infiltrates. Nodular infiltrates were more frequent in protocol biopsies while diffuse infiltrates were more common in indication biopsies. That diffuse infiltrates are the morphological hallmark in indication biopsies with the diagnosis of acute cellular rejection according to the current Banff classification is not surprising. The Banff classification system for renal allograft pathology is made for, and predominately based on, data from biopsies taken from allografts with clinical overt dysfunction, namely indication biopsies. The basic concept of the Banff classification is that an increasing interstitial infiltrate accompanied by increasing tubulitis indicates increasing grades of severity of cellular rejection (23–25). Our data support this concept. However, in biopsies that were classified as rejection according to Banff, besides the typical diffuse infiltrate, all types of infiltrates (including the `nonspecific') can be observed significantly more often than in nonrejecting allograft biopsies.

Indication biopsies from allografts with clinically suspected rejection (i.e. rise in creatinine) had besides diffuse infiltrates, perivascular raggedly outlined infiltrates which are currently not considered by the Banff classification as a significant morphological correlate. Interestingly, this was not observed in protocol biopsies with an incidental concomitant rise in the serum creatinine. However, diffuse infiltrates were those which were related to a higher frequency as well as higher grades of tubulitis, especially in indication biopsies as well as in protocol biopsies with an incidental concomitant rise in the serum creatinine. These findings suggest that a diffuse cellular infiltration with tubulitis (both hallmarks of cellular rejection according to Banff) might rather represent a more progressed, more severe rejection episode leading to allograft dysfunction and therefore to a clinically indicated biopsy. This supports the concept that rejection is a linear process with the clinically detectable rise in creatinine being the latest event. Furthermore, as shown in animal models, changes in the transcriptome might even precede minor subclinical infiltrates and subclinical tubulitis detected by protocol biopsies (26). However, diffuse infiltrates displayed a high density of CD68-positive macrophages and frequently B cells besides T cells indicating that cellular rejection is not merely a T-cell-dependent process. Nodular B-cell-rich aggregates were significantly more frequent in protocol biopsies from stable allografts than in indication biopsies. This type of infiltrate of uncertain pathogenic significance is currently not considered by the Banff classification system, although recent data suggest a potential pathogenic role for B cells in the rejection process (27–30). Our finding that nodular infiltrates are more frequent in rejecting biopsies than in nonrejecting (according to the Banff classification) supports a pathogenic role for B cells in acute cellular rejection. We were not able to demonstrate any significant prognostic impact of nodular infiltrates per se in our large cohort. Nevertheless, several possible B-cell functions in cellular rejection are conceivable. Nodular B cell aggregates might represent secondary lymphoid organs in the transplant where B cells can function as antigen-presenting cells for T cells (31–36). Recently, IL6 production by B cells was described as a crucial event in liver fibrosis (37,38). Since we observed nodular infiltrates more commonly after acute rejection episodes and at later time points after transplantation, a potential pathogenic role of B cells in the chronic rejection process requires further elucidation. Most current immunosuppressive therapy protocols do not cover B cells. However, we recently observed that nodular infiltrates resolved in a patient with refractory rejection after rituximab was given (39).

Nodular cell aggregates in the immediate vicinity to the organ capsules, as all other subcapsular infiltrate types, were without any impact on allograft function and were similarly distributed in all diagnoses, types of biopsies and at all time points. Besides nodular or other perivascular infiltrate types, which are all not taken into account by the current Banff classification, recent data from protocol biopsies suggest a relevant role for infiltrates in areas of tubular atrophy (18). The simultaneous presence of infiltrates in areas of nonatrophic and atrophic tubules seems to particularly indicate ongoing immunological damage to the allograft associated with functional impairment and reduced graft survival (19,40,41). This hypothesis is further supported by our observation that patients with chronic tubulo-interstitial changes (i.e. at least Banff CAN grade I) in their 6-month protocol biopsies significantly more often show signs of inflammation in areas of tubular atrophy already in their prior protocol biopsy at 3 months after transplantation. Furthermore, patients with infiltrates in areas of tubular atrophy in their 6-month protocol biopsy had a significantly lower creatinine clearance at 2 years compared to those who lacked this type of inflammation.

Most biopsies (75%) showed combinations of different types of infiltrates, but no significant prognostic impact of any specific infiltrate combination was found in a multivariate analysis. In contrast, in a linear regression model, continuous detection of inflammation (in our study = high sum of infiltrates, irrespective of a specific infiltrate or cell type or combination of infiltrates) in sequential biopsies taken during the first year of transplantation was significantly related to an inferior allograft function 1 and 2 years after transplantation. In this setting, not all infiltrates and cell types can be regarded as a sign of rejection or as the cause of allograft damage. Nonimmunological insults can also lead to cellular infiltration and inflammation. Furthermore, even tubulitis might be the consequence rather than the cause of tubular epithelial cell damage (42). However, failure of resolution obviously perpetuates a continuous process of inflammation and leads to chronic-irreversible allograft pathology (43). Therefore, we do not believe that all subclinical infiltrates should be regarded as rejection and lead to immediate anti-rejection treatment. Nevertheless, the present findings challenge immunosuppressive therapy protocols that merely cover T cells. A prospective randomized trial that includes anti-B-cell therapy in patients with nodular B-cell aggregates would be welcome. Furthermore, histiocytes should also be considered as targets in future therapeutic regiments.

We believe our results have important clinical implications. Our findings are in accordance with the Banff concept that cellular rejection/allograft inflammation is a continuous process that becomes more severe with increasing amounts of infiltrates. The current Banff classification system does not take into account all the infiltrate types (nodular, perivascular, in areas of tubular atrophy) that might participate in chronic allograft damage. Which of these infiltrate types are signs of cellular rejection and which are a consequence of nonimmunological insults needs to be elucidated. We are not suggesting a new classification system for protocol biopsies from renal allografts. However, the presented findings might be a basis for future adaptation of the Banff classification. Based on our findings and those of other protocol biopsy programs, a consideration of nodular infiltrates and inflammation in areas of tubular atrophy seem to be especially reasonable (18,19,40,41). Reproducibility and therapeutic consequences of documenting currently not considered types of inflammation have to be addressed in prospective multicentric trials including protocol biopsies. From our experience with a comprehensive protocol biopsy program after renal transplantation, a meaningful evaluation of an allograft can only be achieved by analyzing sequential protocol biopsies. In general, biopsies enforced by clinical dysfunction reveal far progressed stages of acute or chronic allograft damage. Analyzing only one time point after transplantation would not allow detection of success or failure of treatment. This is especially important in randomized drug trials, which do not usually include sequential protocol biopsies.

In summary, one specific cell type or morphological pattern of inflammation in one biopsy is not decisive for the fate of a renal allograft. Persistence of inflammation or failure of clearance by therapy can cause progressive allograft damage. From our findings, all cellular infiltrates, except those in the immediate vicinity of the organ capsule might contribute to this process.


Results of the study have been previously published in abstract form (Journal of the American Society of Nephrology, Supplement, Volume 15, 2004, Abstract #F-FC089 and Journal of the American Society of Nephrology, Supplement, Volume 16, 2005, Abstract #F-FC158).

Michael Mengel currently receives a grant from the `Deutsche Forschungs Gemeinschaft' (ME 2097/3-1), a nonindustrial, governmental funding source without any influence on the conduct of the present study or interpretation of results.

We are grateful to Friedrich C. Luft for critically reviewing and editing the manuscript.