The clinical outcome and appropriate management for patients showing ‘borderline changes’ on allograft biopsy after renal transplantation is still controversial. In an attempt to identify predictive factors of clinical outcome of patients with such lesions, we reviewed the clinical course of 91 patients with borderline changes. Multivariate analysis revealed significant and independent effects of histological stage (i + t ≤ or > 2) and time to borderline changes (≤ or >3 months after transplant) on serum creatinine levels at 1 year from borderline changes episodes (respectively, p = 0.04 and p = 0.02) and only a significant effect of time to borderline changes on serum creatinine levels at 2 years (p = 0.005). Renal function at 1 year and 2 years as 5- and 8-year graft survival were not significantly different in the group of patients treated with antirejection therapy (T group, n = 49) compared with the untreated group (UT group, n = 42). This study strongly suggests that borderline changes with histological score (i + t) > 2 and late episodes of borderline changes should be considered to be of poor prognosis.
Because they are a crucial determinant of long-term graft function, diagnosis and treatment of acute allograft rejection are essential challenges in the field of kidney transplantation. The Banff 97 working classification of renal allograft pathology was introduced to standardize the histological definition of acute renal allograft rejection, to guide management in transplant patients, and to establish endpoints for clinical trials (1,2). The Banff scheme defines the minimum threshold for rejection as infiltration of 25% or more of the renal cortex with five or more mononuclear cells in a focus of tubulitis (histological indices i2t2). Borderline changes refer to biopsies with changes insufficient for a diagnosis of acute rejection, including mild to moderate (<50%) cortical infiltration and one to four mononuclear cells per tubule cross-section (histological indices i1t1 or i2t1).
While histological criteria clearly differentiate acute rejection and borderline changes, the clinical and immunological relationship between the two events is poorly understood and the pathogenic role of such limited tubulointerstitial mononuclear infiltrates are not well established. The immune activation profile in blood samples and renal biopsies from patients with borderline changes tend to show activity levels that are intermediate between acute rejection and no rejection (3,4). Clinically, borderline changes may be found in the context of renal allograft dysfunction or in biopsy specimens in the setting of stable graft function. In the first context, a few clinical validation studies have demonstrated progression of borderline changes to acute rejection in about one-third of patients (5,6). Several uncontrolled studies have retrospectively addressed the effects of antirejection therapy on these borderline infiltrates. The results are controversial, with some reports suggesting a response rate similar to that of mild acute rejection (6–9) and others suggesting a lower response rate. In contrast, the clinical and histological outcomes of patients exhibiting borderline changes in biopsy specimens of stable renal allografts have been little studied and there are no data yet on the immediate or long-term outcomes of these patients (10).
Taken together, these studies highlight the lack of a clear understanding of the immunological significance and clinical outcomes of borderline changes, and the absence of guidelines in the therapeutic management of these patients. We reviewed the clinical outcome of 91 patients in an attempt to identify predictive factors of clinical outcome of patients with borderline changes diagnosed in the context of renal allograft dysfunction.
Materials and Methods
Kidney pathology records of allograft biopsies obtained over an 8-year period, from January 1996 to September 2003, were reviewed to identify biopsies with borderline changes. Clinical data were obtained by review of patient records and included: recipient age, donor age, gender, HLA status, dates of transplant, immunosuppressive treatment, dates of biopsies, cold ischemia time, incidence of delayed graft function (DGF), clinical course preceding the biopsy, clinical indication for biopsy, baseline serum creatinine, time of biopsy and follow-up at 1 month, 6 months, 1 year and 2 years, incidence of acute rejection episodes and number of deaths and graft loss at 2 years. Multiorgan grafts were excluded from this study. All patients underwent renal allograft biopsy because of acute renal degradation with a 20% or greater rise in serum creatinine or persistent stagnation of serum creatinine level. Patients were divided into two groups according to the therapeutic management of borderline changes (treated patients or T group and untreated patients or UT group). Treatment of borderline changes consisted of intravenous methylprednisolone pulses on 3 subsequent days and antithymocyte globulin or OKT3 in cases of steroid-resistant rejections. DGF was defined by the need for post-transplantation hemodialysis during the first 7 days after transplantation (11). All episodes of acute renal allograft rejection were confirmed by allograft biopsy. Acute rejection episodes were treated with intravenous methylprednisolone pulses on 3 subsequent days and with antithymocyte globulin or OKT3 in cases of steroid-resistant rejections.
All biopsies studied were evaluated and scored using the Banff 97 allograft classification criteria. In this study, we analyzed ‘borderline changes’ defined as i1t1 or i2t1, and biopsies with less than 25% interstitial infiltration in combination with more than four mononuclear cells per tubular cross-section (i1t2). The last category (i1t2) is not clearly defined by the Banff schema, but our interpretation has been to classify them as borderline changes, since they do not reach the threshold for acute rejection grade 1 (5,12). However, in order to take into account this histological heterogeneity, we separated the biopsies into two distinct histological groups: i1t1 and i2t1 or i1t2. Additional histological diagnoses were recorded including urinary outlet obstruction, acute tubular necrosis, thrombotic microangiopathy and cyclosporine or tacrolimus toxicity.
All continuous variables were expressed as mean ± standard deviation; all categorical variables were expressed as N and percentage of total. The outcome evaluated in this study included creatinine serum at 1 month, 6 months, 1 year and 2 years following acute rejection after the ‘borderline’ biopsy and graft survival. To account for factors significantly related to the outcome of patients, univariate analysis was performed. Values in the two groups were compared using a t-test or a Mann–Whitney test when distribution deviated from normal. Categorical values were compared using the Fisher exact test. For multivariate analysis, two-way analysis of variance (ANOVA) was performed to determine the effect of treatment, histological stage and mean time to diagnosis of borderline change on serum creatinine level; p < 0.05 was considered significant.
Characteristics of patients in the T and UT groups
Ninety-one patients with borderline changes on allograft biopsy by the Banff scheme met the inclusion criteria. All patients meeting these histological criteria were included for analysis, excluding the patients with multiorgan transplants. Four patients were excluded from analysis due to the loss to the 2-year follow-up. According to the therapeutic management of borderline changes, patients were included either in the T group (n = 49, 54%) or in the UT group (n = 42, 46%). All patients of the T group were treated with methylprednisolone pulses (500 mg daily for 3 consecutive days) and five patients (10%) received additional therapy, i.e. intravenous antithymocyte gammaglobulins or OKT3. Acute rejection was considered as steroid-resistant if creatinine levels did not return to within 20% of baseline within 3–5 days after the last methylprednisolone pulse. Patient characteristics are reported in Table 1. No statistical differences were observed between the T and the UT groups according to donor and recipient ages, immunological risk factors, cold ischemia time, incidence of DGF, and number and severity of previous episodes of acute graft rejection (Table 1). As shown in Table 1, induction therapy and maintenance immunosuppressive therapy were similar in the two groups.
Table 1. Clinical and histological data
Untreated group (UT) n = 42 (46%)
Treated group (T) n = 49 (54%)
Std: mean values are reported ± standard deviation.
Bitherapy: cyclosporine or azathioprine + steroids
Tritherapy: cyclosporine + azathioprine or MMF + steroids
Tritherapy: tacrolimus + azathioprine or MMF + steroids
Tritherapy: sirolimus + MMF + steroids
Previous episodes of acute graft rejection
Acute cellular rejection
Borderline changes were analyzed according to four criteria (Table 2): Banff histological score (i + t ≤ or > 2), mean interval between transplantation and occurrence of borderline change episodes, serum creatinine at the time of borderline changes, and mean delta of serum creatinine between values at baseline and at the time of borderline changes. The only statistical difference observed was the mean interval which was significantly longer in the UT than the T group, 616 ± 166 and 199 ± 199 days, respectively, p = 0.001.
Table 2. Clinical and histological characteristics of borderline change episodes
Untreated group (UT) n = 42 (46%)
Treated group (T) n = 49 (54%)
Std: mean values are reported ± standard deviation.
Median time to borderline rejection (days)
Mean serum creatinine delta
Mean serum creatinine level at the time of borderline change (μmol/L)
>i1t1 (i1t2, i2t1)
C4d staining was available only in six patients from the T group and four from the UT group, and was negative in all cases. We retrospectively performed C4d staining in 15 new patients from each group. Peritubular capillaries staining was negative in all samples (n = 30). Next, in order to determine the origin of cell infiltrate in BL change, we retrospectively performed CD3 and CD20 staining in 30 patients (15 in the T group and 15 in the UT group). CD3 staining was highly positive in all samples (n = 30). In contrast, CD20 staining was positive in 5/15 patients from the T group and in 7/15 from the UT group (<10% in all cases).
Outcome of patients in the T and UT groups
The 5-year graft survival rate (censored for death) was 85.17% in the T group and 84.2% in the UT group. The 8-year graft survival rate (censored for death) was 72.5% in the T group and 75.9% in the UT group (p = 0.64) (Figure 1). The number of episodes of acute rejection diagnosed on subsequent biopsy occurring after the episode of borderline changes during follow-up was recorded in both groups. Incidence of acute rejection episodes was significantly higher in the T group than in the UT group (respectively, 45% and 21%, p = 0.04). Among the 22 patients from the T group who experienced further episodes of acute rejection, 13 had an initial score of borderline changes (i + t) >2 and 9 patients had <2 (NS). Among the nine patients from the UT group who experienced further episodes of acute rejection, three had an initial score of borderline changes (i+ t) >2 and six patients had ≤2 (NS). Renal function was assessed by means of serum creatinine at 1 month, 6 months, 1 year and 2 years after the episode of borderline changes. No statistical difference was observed between the two groups (Table 3). Mean slope of creatinine clearance improvement 2 years after borderline change diagnosis was 5.64 mL/min/1.73 m2/year in the UT group and 4.92 mL/min/1.73 m2/year in the T group (P = NS). Proteinuria at 1 and 2 years were, respectively, 0.36 and 0.52 g/day in the UT group and 0.58 and 0.63 g/day in the T group (P = NS).
Table 3. Serum creatinine at 1 month, 6 months, 1 year and 2 years after borderline change episode
Untreated group (UT)
Treated group (T)
Mean creatinine serum at 1 month (μmol/L)
Mean creatinine serum at 6 months (μmol/L)
Mean creatinine serum at 1 year (μmol/L)
Mean creatinine serum at 2 years (μmol/L)
Two-way ANOVA were performed to analyze the effect of treatment, histological stage, time to borderline rejection on serum creatinine at 1 year and 2 years (Table 4). At 1 year, ANOVA showed significant and independent effects of histological stage and time to borderline changes (≤ or > 3 months after transplant) on renal function at 1 year (respectively, p = 0.04 and p = 0.02). At 2 years, ANOVA showed a significant effect of time to borderline changes (p = 0.005), but no significant effect of either histological stage (p = 0.06) or treatment (p = 0.08) (Table 5). Previous episode of acute rejection did not significantly influence the 1-year (p = 0.73) and 2-year (p = 0.65) serum creatinine levels. Excluding the 28% of untreated patients and the 16% of treated patients who experimented previous episode of acute rejection, mean serum creatinine levels at 1 year were 169.7 and 157.2 μmol/L in the T (n = 40) and in the UT (n = 29) (p = 0.20) groups, respectively, and mean serum creatinine levels at 2 years were 143 and 169.2 μmol/L, in the T and the UT groups, respectively (p = 0.08).
Table 4. Multivariate analysis using mean serum creatinine at 1 year and 2 years, using two-way ANOVA
Time to borderline rejection
UT = untreated group; T = treated group; ≤3 months = borderline rejection before 3 months, >3 months = borderline rejection after 3 months.
One-year mean creatinine (μmol/L)
Two-year mean creatinine (μmol/L)
Table 5. The ANOVA test p-Values
p-Value (1-year mean creatinine)
p-Value (2-year mean creatinine)
Time to borderline rejection
We analyzed here the clinical course of borderline infiltrates in renal allograft recipients. Our multivariate analysis suggests that both histological stage (i + t ≤ or > 2) and time to borderline changes (< or > 3 months after transplant) are two independent factors that negatively influence the 1-year (for both of them) and 2-year (for time to borderline change) graft function. This confirms the conclusion of Meehan et al. (5) concerning the detrimental effect of histological stage on graft outcome and suggest for the first time that, similar to acute rejection, late episodes of borderline changes are harmful to the long-term graft function. Our findings also suggest that antirejection therapy has no short- or long-term benefits for renal function. We believe that our T and UT groups were appropriate for a comparative statistical analysis since (i) histological criteria including borderline change indices and concurrent diagnosis, (ii) renal function impairment at the time of borderline changes and (iii) other immunological and nonimmunological risk factors associated with unfavorable graft outcome, were not significantly different between the two groups. The only difference observed was the mean time to diagnosis of the borderline infiltrate, which occurred significantly later in the UT group than in the T group. We cannot exclude that this last point, associated with the highest (although not significant) histological score in the T group, might have influenced the decision to treat or not. However, since (i) late immunological events are known to have a more detrimental effect on long-term renal function when compared with early acute rejection (13) and (ii) our multivariate analysis has demonstrated a detrimental effect of late episode of borderline change on renal function, this supports the thesis that greater damage is inflicted by borderline infiltrates in the UT group. Finally, multivariate analysis (ANOVA) confirmed that treatment of borderline changes is not beneficial for graft function at 1 year and 2 years post-transplant, independent of histological stage and time to borderline changes. Because current recommendations to treat or not to treat borderline changes occurring in the context of renal allograft dysfunction are not clearly defined, our study focused on borderline infiltrates identified only in such conditions. This question is clearly correlated with the ability of borderline infiltrates to progress to acute rejection and finally to contribute to chronic allograft nephropathy. The Transplant Group at the University of Chicago reported that only a subset (28%) of patients who had not received antirejection therapy following the diagnosis of borderline changes had a second biopsy within 40 days due to lack of clinical improvement (5). Infiltrates that progressed to rejection had more frequent glomerulitis and Banff acute scores (1 + t + g + v >2). Finally, since most (72%) borderline changes not treated with additional antirejection therapy did not experience further episodes of acute rejection over the 40-day follow-up, the authors suggest that conservative management of these lesions in the short term may be more appropriate than routine treatment. Our study confirms these results, but also suggests that conservative management for graft function is of value up to 2 years of follow-up.
An unexpected result of our analysis was the significantly higher incidence of acute rejection episodes following antirejection therapy of borderline changes in the T group. Occurrence of acute rejection episodes was not statistically correlated with the initial Banff acute score (> or ≤ 2) either in the T or in the UT group. Since controlled studies of the therapeutic management of borderline infiltrates are lacking, this has never been previously reported. In contrast to acute rejection, the immune activation profile of T lymphocytes from borderline infiltrates has been poorly studied. Desvaux et al. demonstrated that intragraft expression of several cytotoxic molecules, including Granzyme B/perforin tends to show activity levels intermediate between acute rejection and nonrejection (3). In another study, intragraft monitoring of cytokines and chemokine gene expression showed that borderline changes enhanced the expression of IL-10 and TGF-β1, two cytokines also known for their immunoregulatory properties (14) but, in contrast with acute rejection, did not increase the expression of TNF-α (15). Taken together, these preliminary results indicate that infiltrates in borderline rejection do not share the same cytotoxic profile as acute rejection. There is a growing belief that the outcome of allograft response—rejection or tolerance—following transplantation is determined by the balance of allo-aggressive to graft-protecting T cells (16,17). The relative expression of tolerance and cytotoxicity markers in borderline infiltrates has yet to be analyzed in order to determine if T cell infiltrates from borderline changes display rather a regulatory or a cytotoxic phenotype.
In summary, this study suggests that borderline changes with histological score (i + t) > 2 and late episodes of borderline changes should be considered to be of poor prognosis. Only the results of a prospective and randomized study will give authority to clinicians to treat or not borderline changes.