Journal Logo

Clinical Transplantation

EFFECT OF HISTOLOGICAL DAMAGE ON LONG-TERM KIDNEY TRANSPLANT OUTCOME

Nankivell, Brian J.1,2; Fenton-Lee, Carol A.1; . Kuypers, Dirk R.J1,3; Cheung, Elaine1,4; Allen, Richard D.M.5; O’Connell, Philip J.1; Chapman, Jeremy R.1

Author Information
  • Free

Abstract

INTRODUCTION

Chronic renal allograft failure, especially chronic rejection, remains an important long-term problem in kidney transplantation despite improvements in transplant immunosuppression. A number of clinical risk factors such as donor quality, ischemic time, delayed graft function, lymphocytotoxic cross-match, HLA mismatch, immunosuppression, and acute rejection have been correlated with renal allograft failure (1–7). The characteristic interstitial fibrosis, tubular atrophy, and glomerulosclerosis present in chronically failing kidney allografts (8) have also been correlated with adverse graft outcomes. In studies of protocol biopsies taken between 3 months and 2 years after transplantation, histological damage has been associated with impaired renal function (9–11) or with graft failure (9,12–14). Other studies have suggested that “subclinical rejection” may reduce graft survival (15). The mechanism of how these clinical risk factors act at histological level to cause nephron damage is poorly understood.

We studied a longitudinal cohort study of patients treated with cyclosporine, azathioprine, and prednisolone and correlated histology of the 3-month biopsy with outcome for a median follow-up of 9.3 years (range, 1–13.9 years). Endpoints were a decline in transplant renal function, subsequent histological damage on repeat protocol biopsy at 12 months, onset of chronic allograft nephropathy (CAN), and long-term graft survival, censored for death. The aims of this study were to determine the features present on a 3-month protocol biopsy that were associated with chronic graft damage and poor long-term transplant outcome. The effects of individual Banff qualifiers such as residual interstitial lymphocyte infiltrate and tubulitis (termed “subclinical rejection”) was assessed and correlated with chronic interstitial damage and long-term transplant outcome. Sequential examination of the implantation, 3-month, and 12-month histology were studied to determine the persistence of chronic interstitial and vascular damage (such as those observed in older donor kidneys), to delineate if accumulation of renal damage occurred, and to assess the transplanted kidney’s ability to recover from pathological insults. Multivariate analysis of proteinuria, hypertension, 3-month histological damage and other clinical risk factors was used to evaluate their effects and interactions on the occurrence of CAN, decline in renal functional decline, and graft failure.

PATIENTS AND METHODS

Transplant population.

Patients were initially immunosuppressed using nonmicroemulsion cyclosporine (CsA, Sandimmune Sandoz, Basel, Switzerland [12.5 tapered to 5 mg/kg/day from 3 to 6 months]), azathioprine (1.5 mg/kg/day), and prednisolone (20 mg/day for 3 months, then tapered to 10 mg/day). Induction therapy with antithymocyte globulin (ATG, Fresenius, Germany) or OKT3 (Orthoclone OKT3, Ortho Biotech) was reserved for highly sensitized patients, and those with initial nonfunction (n=28). Rejection episodes were defined by clinical parameters supported by histology, and treated by high-dose intravenous methylprednisolone for 3 days and/or 10 days of antilymphocyte therapy (OKT3 or ATG) for steroid-resistant or vascular rejection. Delayed graft function was defined as dialysis requirement after transplantation or a serum creatinine greater than 150 μmol/L at day 8. Proteinuria was categorized by spot early morning urinary concentration into five groups of nil, 0–0.3 g/L, 0.3–1 g/L, 1–3 g/L, and >3 g/L. The effect of clinical CsA exposure was estimated from the mean CsA levels and dose, the presence and number of episodes of CsA toxicity within the first 3 months, and the time-averaged dose of CsA for the study duration.

Study design and outcome measures.

Of 163 consecutive renal allografts performed at Westmead Hospital from January 1986 that were functioning at 3 months, 112 underwent a 3-month protocol biopsy, as described previously (16). The study group (n=102) was demarcated by those biopsies with adequate tissue on the 3-month biopsy, as defined by Banff criteria (17).

This cohort of patients was then followed for up to 13.9 years after transplantation, with a median follow-up time of 9.3 years, and evaluated for several markers of functional and structural outcomes. Histological injury was assessed by repeat protocol kidney biopsies 1 year after transplantation (n=39). Renal functional decline was estimated from the change in serum creatinine (calculated from the difference of the nadir and final serum creatinine, divided by follow-up time) and the change in yearly Tc99m diethylenetriaminepentaacetic acid GFR estimations (calculated from the slope of linear regression line against time after transplantation). CAN was defined by clinical features of declining renal function supported by a subsequent transplant histology (8). Graft survival was defined as the resumption of dialysis, achievement of an irreversible serum creatinine of 500 μmol/L, or death resulting from uremia. Patients who died of unrelated causes were censored at that point and analyzed by actuarial survival methodology as a graft functioning up to the time of death (18).

Renal transplant biopsy assessment.

Routine implantation donor biopsies were undertaken after release of vascular cross-clamps. Needle-core protocol biopsies at 3 and 12 months were obtained, unless contraindicated or consent withheld, and histopathology was evaluated by blinded observers, as described previously (16). The Banff schema was used throughout the analysis, based on the original descriptive paper (17) and subsequently supplemented by the chronic Banff schema (19,20). The “borderline rejection” classification from Banff 93–95 schema (17) was retained for this paper for the purposes of continuity with previous studies and because of its distinction from the acute rejection grade I. Chronic nephropathy grade was classified by the chronic Banff schema (20). As there can be inter-observer variation in the interpretation of the Banff schema, to facilitate translation of these findings into the practice of others, we would refer readers to illustrations on the Web, such as those found at http://tpis.upmc.edu. As late rejection because of noncompliance caused subsequent histological damage consistent with chronic rejection, these patients were included in the chronic allograft nephropathy group. Medication noncompliance was assessed by historical confirmation by the patient.

Statistical analysis.

An unpaired Student’s t test or a Wilcoxon test was used for nominal data, as appropriate. Pearson’s and Spearman’s tests were applied for comparative data. Categorical data were examined by a chi-square test or by conditional binomial test. Multiple groups were compared using the least-significant-differences or Newman-Keuls’ method, as appropriate, after one-way analysis of variance. Multiple linear regression was used to determine the predictive factors of individual Banff histological parameters, after backward elimination to determine the best-fitting model. Collinearity diagnostics were applied to detect significant interactions. Cox regression and logistic regression were used where appropriate. Multivariate models were examined using initially histological predictors to evaluate the effect of structural damage on outcome, with analyses supplemented by clinical predictors, adjusted when necessary for confounding variables. Input variables assessed (n=179) included donor histology, ischemic times, HLA mismatch, blood group, cytomegalovirus status, 3- and 12-month histology, posttransplant events such as DGF, acute rejection, immunosuppressive therapy, and delayed factors including late rejection, noncompliance, persistent hypertension, and proteinuria. Within a group of possible predictors (e.g., donor variables), the best predictor of outcome was carried forward and tested against other factors in subsequent models. Approximately 1,000 models were tested for parsimony. Data are expressed as mean ± SD unless otherwise stated. A probability of less than 0.05 was considered significant.

RESULTS

Transplant and Clinical Characteristics

Patients (n=112) were 43.6±13.6 years old at transplantation; 48% were male with a mean time on dialysis of 29.9±32.5 months. The total HLA mismatch score was 2.7±1.3, and 15% of patients received repeat grafts. Donors were 33.8±15.8 years old; 69% male and 17% were of living-related origin. In donor biopsies taken at implantation, mild (ci grades I to II) fibrosis was present in 57%, and vascular changes (intimal fibrosis or hyalinosis) were present in 24% of implantation biopsies. DGF occurred in 49% of recipients and was associated with an increased number of acute rejection episodes (P <0.001). The mean numbers of acute cellular and vascular rejections were 1.36±0.86 and 0.40±0.55 episodes, respectively. Of those 102 patients with adequate tissue at 3-month follow-up, 56 were still alive with functioning grafts (median follow-up of this subgroup was 10.3 years), 26 patients died with a functioning transplant (12 cardiovascular deaths, 7 late infections, 6 cancer, 1 suicide), and 20 had graft failure necessitating return to dialysis.

Beyond 3 months after transplantation, hypertension was present in 79.4%, proteinuria in 38.2% and was mild (less than 1 g/L) in 17.6%, moderate in 13.7% and severe (>3 g/L) in 5%. Late rejection occurred in 11.8% of patients, declining renal function in 23.5% and chronic allograft nephropathy in 20.6% of patients. Nadir serum creatinine was 143±54 μmol/L, and the time-averaged change of serum creatinine was −26.0±60.6 μmol/year. Follow-up was 100%.

Acute Banff Qualifiers at 3 Months

Acute inflammatory activity in the protocol biopsies 3 months after transplantation was present in a surprisingly large number of biopsies, with Banff borderline changes present in 49% (Table 1). Subclinical rejection, defined as histological evidence of acute rejection but without associated acute functional decline, was present in 29% (11 with grade 1 acute Banff changes and 7 with Banff 2 or greater). By multivariate analysis, the presence of subclinical rejection at 3 months was independently predicted by previous cellular rejection (odds ratio 2.24 per rejection, P <0.05), increased HLA mismatch (odds ratio 1.77 per mismatch, P <0.05), and a shorter time on dialysis (odds ratio 0.58 per year, P <0.05). The presence of any residual rejection (including both Banff borderline and subclinical acute rejection on the 3-month biopsy) was predicted by prior rejection (odds ratio 39.4, P <0.01) and increased HLA mismatch (odds ratio 2.87 per mismatch, P <0.05). Banff chronic nephropathy was present in 24% of 3-month biopsies. Predictors of chronic tubulointerstitial damage at 3 months included DGF, microvascular disease in the donor, cold ischemic time, and acute vascular rejection (P <0.001), as described previously in detail (16).

Table 1
Table 1:
Summary data from protocol histology at 3 and 12 months after transplantation (mean ± SD)

Study Outcomes

Protocol histology at 12 months.

Acute Banff rejection grade at 3 months was correlated with chronic Banff grade at 12 months (r =0.36, P <0.05). The best correlations were between tubulitis and tubular atrophy (r =0.32), interstitial mononuclear cell infiltration and chronic interstitial fibrosis (r =0.36), and acute vascular intimal inflammation and chronic intimal thickening (r =0.66), at 3 and 12 months, respectively (P <0.05 - 0.001, Table 2). Hence, persistent subacute injury at 3 months, localized within a histological compartment, was associated with chronic histological damage at 12 months, within that same compartment.

Table 2
Table 2:
Spearman correlation coefficients ( r ) between acute Banff qualifiers at 3 and chronic tubulointerstitial damage at 12 months

Some of the chronic tubulointerstitial damage observed at 12 months had already occurred by 3 months, as shown by the correlation of acute interstitial inflammation with chronic fibrosis (r =0.31, P <0.01) and tubular atrophy (r =0.30, P <0.01) in the 3-month biopsy. There was a poor correlation between acute mononuclear cell inflammation at 3 months and additional damage (calculated by the differences between the 3- and 12-month Banff scores) for ci and ct. Subclinical rejection, however, was associated with progressive chronic intimal thickening between 3 and 12 months (P <0.01). The acute inflammatory activity had settled by 12 months with only 6 of 39 biopsies (15.4%) showing borderline changes, and none with subclinical rejection. These changes occurred without specific antirejection therapy, and there was no bias to interval graft loss (only 4 grafts that failed between 3 and 12 months and their 3-month acute Banff scores were evenly distributed). Hence, the effect of subclinical rejection causing tubulointerstitial damage peaked by 3 months, whereas microvascular injury was persistent for at least 1 year.

The mean Banff chronic fibrosis score (ci) increased from 0.62±0.60, observed in implantation biopsies, to 1.01±0.69 at 3 months and stabilized at 1.03±0.74 at 12 months after transplantation (both P <0.001 vs. donor, Table 2). The finding of a poor individual correlation between the 3-month biopsy ci and either the donor or 12-month biopsy (r =0.194, P =NS) indicates that either measurement of interstitial fibrosis lacked reproducibility or more consistent with the data, was that variation between time points was primarily determined by intervening events (such as DGF, vascular rejection, and subclinical rejection). Chronic interstitial fibrosis (ci) at 12 months, however, strongly correlated with chronic tubular atrophy (r =0.90, P <0.001), and also both with 3-month histology of acute mononuclear cell interstitial infiltration (r =0.39, P <0.05) and subclinical rejection (P <0.05, Fig. 1).

Figure 1
Figure 1:
The relationship between 12-month histology, stratified according to the level of acute rejection on the 3-month protocol biopsy, including Banff “borderline rejection” and subclinical rejection. Figure shows mean ± SEM. *, P <0.05 vs. nil rejection.

In contrast, vascular changes were more consistent and predictable, with 12-month chronic intimal vascular thickening (cv) strongly correlated with 3-month cv (r =0.70, P <0.001), acute vascular changes (v, r =0.66) and acute glomerulitis (g, r =0.47, Table 2). At 12 months, chronic fibrointimal thickening correlated with tubular atrophy (r =0.34, P <0.05). Arteriolar hyalinosis score (ah) peaked at 3 months, correlated with donor biopsy vascular disease (r =0.54, P <0.001), and fell to minimal levels by 12 months. This fall was in parallel with reduction in CsA doses, but there was no correlation with CsA dose, levels, or episodes of CsA nephrotoxicity. Hence, vascular intimal fibrosis was related to acute vascular injury and preexisting microvascular damage, which did not recover by 1 year after transplantation.

The 3-month glomerulitis (g) score was abnormal in 4/102 (4%) and was correlated with 12-month cv (r =0.47, P <0.01), mesangial matrix increase (r =0.54, P <0.01), segmental sclerosis (r =0.86, P <0.001), number of sclerosed glomeruli (r =0.50, P <0.01), and the diagnosis of recurrent glomerulonephritis (r =0.62, P <0.001).

Renal impairment.

Decline in renal function, as assessed by change in serum creatinine over time, was a unimodal and positive-skewed histogram. The mean change of the overall group was −26.0±90.7 μmol/year over the study duration. The majority of patients (77%) showed no change (normal range 95% confidence interval of −13.7 μmol/year was calculated from the patients alive without CAN or recurrent glomerulonephritis). In the group of patients alive with graft function at follow-up, there was no significant change over baseline of serum creatinine of 0.16±4.5 μmol/year (P =NS vs. nil change). Deteriorating serum creatinine was correlated with chronic tubulointerstitial injury on the 3-month biopsy including Banff ci (r =0.30, P <0.01), ct (r =0.23, P <0.05), and cv (r =0.44, P <0.001). Acute Banff qualifiers had little effect on the rate of deterioration of serum creatinine.

The nadir serum creatinine reflected chronic tubulointerstitial damage on the 3-month biopsy, and also correlated with ci (r =0.21, P <0.05), ct (r =0.29, P <0.01), and cv (r =0.20, P =0.054). In contrast, isotopic GFR at 3 months correlated best with the percentage of sclerosed glomeruli (r =0.21, P <0.05), but not with Banff ci, cv, and ct. A possible explanation in that chronic tubulointerstitial damage may cause impaired tubular creatinine secretion and elevated serum creatinine, but be insufficient to influence isotopic GFR. By multivariate analysis, decline in serum creatinine was greatest when abnormal 3-month chronic intimal thickening or persistent proteinuria was present, when adjusted for body weight (Table 3). The nadir serum creatinine was associated with increased recipient body weight (r =0.23, P <0.05) and DGF (r =0.47, P <0.001), which enhanced the decline in serum creatinine when adjusted by multivariate analysis. Serum creatinine at 3 months predicted long-term graft failure (P <0.001) and could be substituted for chronic interstitial fibrosis in the multivariate model. It was correspondingly excluded as a predictive factor from further survival analysis, to avoid the circular logic that impaired kidneys tend to fail more often and more quickly.

Table 3
Table 3:
Clinical and 3-month histological predictors of subsequent decline of renal function assessed by change in serum creatinine using multivariate analysis ( R 2=0.367, df =94)

The decline in serial Tc99m diethylenetriaminepentaacetic acid GFR was exacerbated by chronic interstitial fibrosis (ci) or chronic Banff nephropathy grade (P <0.05) and by persistent proteinuria (P =0.07) when adjusted for recipient body weight by multivariate analysis. Nadir GFR was decreased by acute interstitial mononuclear cell infiltration (P <0.05) or acute vascular changes (P <0.01), which correspondingly reduced the decline in the slope of serial isotopic GFR measurements.

Chronic allograft nephropathy.

CAN was present in 21.6% of the total study group by the end of follow-up, and in 75% of those patients with graft failure. It was associated with late rejection from suboptimal immunosuppression in 32%, persistent chronic cellular and vascular rejection despite standard immunosuppression in 10% (“chronic rejection”), no clear immunological precedent in the remaining 36% (presumptive nonimmunological causes including CsA), and mixed causal pathology in 23%. It did not always result in graft loss at the time of study follow-up, being present in 23.1% of those who died with a functioning graft and 2% of those alive with graft function at follow-up (P <0.001). By univariate analysis, the clinical risk factors for chronic allograft nephropathy included early and late rejection, hypertension, persistent proteinuria, and impaired renal function (Table 4). Noncompliance accounted for 27% of patients with CAN and correlated with the onset of chronic allograft nephropathy (r =0.38, P <0.01). When multivariate logistic regression incorporating only the 3-month histological parameters was used, the predictors of CAN included residual interstitial cellular infiltration (Banff i, P <0.05), chronic interstitial fibrosis (Banff ci, P <0.05), and tubular injury (P <0.05). By logistic regression analysis incorporating clinical and histological predictors, the risk factors for CAN included kidney anastomosis time, 3-month chronic intimal thickening (ci) and tubular injury, late rejection, and proteinuria, but not hypertension (Table 5).

Table 4
Table 4:
Clinical and histological findings in patients who developed chronic allograft nephropathy (CAN) and excluding those with recurrent glomerulonephritis: mean ± SD (%)
Table 5
Table 5:
Multivariate predictors of chronic allograft nephropathy by logistic regression analysis, incorporating clinical and histological Banff qualifiers of the 3-month protocol biopsies (n=102)

Actuarial graft failure.

Of the 20 patients whose graft failed, they were 32.5±12.2 years old and had spent 2.09±1.78 years on dialysis. Late rejection occurred in 45% of these patients (P <0.01), noncompliance in 30% (P <0.001), and persistent proteinuria in 80% (P <0.001 vs. other groups). Noncompliance correlated strongly with late rejection (r =0.57, P <0.001) and transplant failure (hazard ratio 5.61, P <0.001) (Fig. 2). Mean urinary protein concentration measured 1.30±1.23 g/L (P <0.001 vs. other groups). Chronic allograft nephropathy was responsible in 75% of graft losses, and the remainder failed because of recurrent glomerulonephritis (n=3), postpartum hemolytic uremic syndrome (n=1), and kidney transplant trauma (n=1). The rate of decline in renal function of serum creatinine in this group was −114.8±179.2 μmol/year and exceeded the other groups (P <0.001).

Figure 2
Figure 2:
Instantaneous hazard plots for graft failure for individual Banff qualifiers present on the 3-month protocol biopsy, for up to 10 years after transplantation.

The univariate clinical and histological predictors of graft failure are presented in Table 6. Of the 3-month histology examined by Cox modelling (excluding clinical factors), acute glomerular abnormalities (Banff g), interstitial inflammation and chronic intimal fibrosis (Banff i and ci, respectively), and vascular intimal disease (cv) were independent predictors of the risk of graft failure (P <0.05-0.001, data not shown). When clinical factors were incorporated into the model (Table 7), only chronic tubulointerstitial damage and chronic fibrointimal vascular thickening remained significant (Banff ci and cv), when adjusted for recipient age and late rejection. Younger age correlated with 3-month glomerulitis (r =0.22, P <0.05) and subsequent recurrent glomerulonephritis (r =0.31, P <0.01), and could be substituted for glomerulitis in the model (Table 7). This reflects the correlation of young living-related transplant recipients with recurrent glomerulonephritis. DGF had a substantial effect on 3-month histology and was associated with significantly increased chronic interstitial fibrosis, tubular atrophy, and chronic Banff grades (P <0.05-0.001). However, by multivariate analysis, DGF did not add anything to the effect of 3-month chronic interstitial fibrosis on long-term survival.

Table 6
Table 6:
Univariate histological and selected clinical predictors of transplant graft failure, censored for death (n = 102)a
Table 7
Table 7:
Composite multivariate model of transplant failure by Cox regression modeling, incorporating 3-month histology and adjusted for late rejection (n=102)

Chronic arterial fibrointimal thickening detected in the renal biopsy at 12 months was one of the strongest and most consistent predictors of outcome (Fig. 3) leading to 14.2% reduction of 10-year graft survival (P <0.05). The absence of chronic interstitial fibrosis at 12 months was associated with outstanding long-term graft survival, with 90% of those grafts functioning between 10 and 15 years (Fig. 4). In contrast, the presence of chronic interstitial fibrosis (chronic Banff grade ci2 or greater) was a portent of graft failure with only 60.4% of grafts functioning at 10 years (P <0.01 vs. ci0). The survival of grafts with an intermediate grade of fibrosis (ci1) was not different from ci1 until 8 years, although a longer duration of follow-up will be required to determine if there is an impact beyond 15 years. The 10-year survival rates for Banff chronic nephropathy grade 0 was 90.4%, grade 1 81.0%, and grades 2 or greater was 57.9% (P <0.01). Hence, early tubulointerstitial damage at 3 months profoundly influenced subsequent graft survival at 10 years and beyond.

Figure 3
Figure 3:
Actuarial graft survival stratified according to the grade of chronic fibrointimal thickening (cv) in 102 patients; cv1 represents chronic intimal thickening causing vascular narrowing of the small arteries of up to 25% of the luminal cross-sectional area. *, P <0.05.
Figure 4
Figure 4:
Actuarial graft survival according to the grade of chronic interstitial fibrosis (ci) in 102 patients; ci2 represents moderate interstitial fibrosis of 26-50% fibrosis of the cortical area. **, P <0.01.

DISCUSSION

The pathogenesis of chronic allograft nephropathy is uncertain, in part because of a relative paucity of histological studies of nephron damage that have been linked to graft survival. This paper examines the long-term importance of chronic tubulo-interstitial damage present at 3 months and detected histologically on routine protocol biopsy together with other clinical risk factors. Common motifs include the effects of late rejection, proteinuria and early tubulo-interstitial damage on graft survival, onset of CAN, and decline in renal function. Acute cellular rejection treated successfully within the first 3 months did not result in any measurable damage, but did identify individuals who are at greater immunological risk for CAN. Rejection caused histological damage if it occurred unexpectedly (late), was difficult to control (such as early vascular rejection), or remained subclinical and thus undetected because of a lack of elevation of serum creatinine. Subclinical rejection present on the 3-month biopsy resulted in chronic tubulointerstitial damage at 1 year. Early risk factors for chronic damage at 3 months included DGF, cold ischemia and donor factors, and vascular rejection. Subsequent damage leading to graft failure was associated with late rejection, proteinuria, and chronic tubulointerstitial damage. The profound effects of chronic tubulointerstitial damage present by 3 months on long-term graft survival emphasize the importance of early posttransplant injury. Chronic vascular intimal thickening and chronic interstitial fibrosis both independently reduced graft survival and were associated with chronic tubular atrophy, suggesting a role for ongoing tubular ischemia and permanent nephron loss in functional graft failure.

Increased rates of graft loss were associated with acute Banff qualifiers of glomerulitis and acute mononuclear cell interstitial infiltrate on the 3-month biopsy; however, these had differential effects and clinical outcomes. Acute glomerulitis (g) was highly predictive of subsequent recurrent glomerulonephritis, segmental glomerulosclerosis at 12 months, persistent proteinuria, and high instantaneous hazard rates for graft loss within the first 2 years after transplantation. Recurrence of disease accounts for only about 2% of all graft failures (21), but its overall incidence is probably much greater (22). Although uncommon, any early glomerular abnormality should thus raise the possibility of recurrent glomerulonephritis and prompt histological evaluation by immunofluorescence and electron microscopy.

Subclinical rejection, defined as Banff acute rejection (17) without functional deterioration (11), was present in 29% of biopsies at 3 months, and Banff “borderline” changes in an additional 49%. Subclinical and borderline rejection were associated with increased HLA mismatch and with prior acute rejection, especially if the later occurred near the time of 3-month protocol biopsy. By 12 months, subclinical rejection, acute tubulitis, and arteriolar hyalinosis had largely resolved, and Banff borderline rejection was reduced to 15% of biopsies. Acute inflammatory abnormalities may be seen at the end of treatment of acute rejection (23), and in our study were associated with subsequent chronic tubulointerstitial damage. Using the acute Banff scoring, i correlated with ci within the 3-month biopsy, indicating that much of the chronic interstitial damage was already established. Subclinical rejection was associated with progressive vascular fibrointimal thickening. By multivariate analysis, chronic interstitial fibrosis (ci) at 3 months accounted for much of the variance and hazard for graft loss, and acute interstitial inflammation added little to the overall predictive power of the model. The resolution of tubulitis, arteriolar hyalinosis, and subclinical rejection at 12 months, and the falling instantaneous hazard rates for acute Banff qualifiers by 2 years on survival analysis, indicated that the effect of subclinical rejection had largely abated by 1 year. Other risk factors such as late rejection, proteinuria, and established chronic interstitial fibrosis became more important at 1 year and beyond.

Early acute rejection functioned as a composite risk factor, representing a mixture of immunological risk factors, such as panel-reactive antibody levels, HLA mismatch, and repeat graft etc., as well as nonimmunological factors, such as poor CsA absorption. The multifactorial nature of this adverse risk factor probably explains its predictive power in many studies. Acute rejection has been associated with a poorer graft outcome (5,7,12) and may be the harbinger of chronic rejection (10,24–26). In our study, early acute cellular rejection, if promptly diagnosed and treated, did not result in any direct histological graft damage at 3 months (16). Although more common in the CAN group, early acute rejection may not progress directly to chronic rejection (26) nor reduce actuarial survival when late rejection and established histological damage were accounted for by multivariate analysis. Hence, its effect may predominantly represent “graft intolerance,” reflecting a capability of the recipient to mount a rejection response when faced with standard immunosuppression. It may contribute indirectly to chronic tubulointerstitial damage through the damaging effect of persistent subclinical rejection. Direct and widespread immunological damage occurred both with vascular rejection despite rapid diagnosis and treatment and with late rejection when longer surveillance intervals allowed substantial histological injury to occur before diagnosis. Late rejection was usually secondary to patient noncompliance (27) or clinician-directed reduction of immunosuppression for another reason, such as the diagnosis of neoplasia. The very high hazard rates of noncompliance and late rejection reflect a substantial risk and poor final outcome in our study and others (27). Hence, rejection-mediated graft damage is dependent on type, timing, and severity of the episode(s).

This study, and that of others (9,13,14,28), have demonstrated the adverse long-term consequences of tubulo-interstitial damage in renal transplantation. Our study used the Banff schema to detail more precisely the differential effects of chronic interstitial fibrosis, tubular atrophy, chronic vascular changes, and acute inflammatory changes on long-term outcome. The 3-month Banff ci was predominantly the result of DGF, vascular rejection, and older donor age (16) and represented established nephron damage. It correlated strongly with tubular atrophy and was independently predictive of progressive renal dysfunction and graft failure. The effect of ci on decline of isotopic GFR and graft loss was most pronounced when there was at least moderate interstitial fibrosis affecting 26–50% of the cortical area (ci2). Pathology studies of native kidney disease have demonstrated that interstitial damage and fibrosis was the most reliable adverse risk factor, irrespective of the etiology of the renal disease (29,30). In our study, absence of chronic interstitial fibrosis at 3 months predicted an excellent 14-year graft survival exceeding 90% (censored for death). Even when minor fibrosis was present (ci1: interstitial fibrosis tissue of 6–25% of cortical area), graft survival was indistinguishable from those grafts without fibrosis until 8 years, after which survival curves appeared to separate. As the majority of pathological insults to a transplanted kidney occur in the tubulo-interstitial compartment, this is likely that interstitial fibrosis represents a common end-pathway of inflammation with nephron destruction resulting in eventual decline in renal function.

Chronic vascular thickening (Banff cv) was a consistent and important feature of chronic rejection, and predicted the onset of CAN, renal function decline, and ultimate graft failure. By 3 months, chronic intimal thickening was the cumulative result of preexisting vascular disease in the donor, prolonged cold ischemia, and episodes of acute vascular rejection. Intimal thickening, once established on the implantation or 3-month biopsy, did not recover despite the absence of additional insults. This suggests the renal transplant microvascular compartment may have a reduced capacity to remodel either intrinsically or under the influence of immunosuppression, or was more sensitive to the stresses and injuries of the transplant process, including ischemia and CsA nephrotoxicity. The potential role of CsA nephrotoxicity causing chronic tubulointerstitial and vascular damage was difficult to ascertain and distinguish from CAN and chronic rejection in our study, but other studies has been shown it to be common and important (31). The temporal precedence of chronic intimal thickening, its independent, strong and consistent prediction of adverse outcomes by differing measures, and relative specificity for CAN and graft failure, make it a likely causal factor for CAN. Chronic vascular narrowing appeared to be an additive or exacerbating factor, increasing the separation of long-term actuarial survival. This suggests that tubular ischemia from an impaired microvascular circulation may have a role in the development of interstitial fibrosis and tubular atrophy in the pathogenesis of CAN.

In summary, this study and its companion paper (16) have identified a number of discrete insults that are capable of inflicting permanent histological damage on the transplanted kidney, including DGF, prolonged ischemic time, recurrent glomerulonephritis, acute vascular rejection, subclinical rejection and late rejection. These events appear to be exacerbated by older donor age and donor vascular disease, and subsequently by persistent proteinuria. These insults resulted in tubular atrophy, chronic interstitial fibrosis, and vascular changes, which in turn progressed to eventual decline in renal function and graft failure. Chronic allograft nephropathy can be conceptualized as the sequelae of incremental and cumulative damage to the transplanted kidney. The duration of graft survival is dependent and predicted by the quality of the transplanted donor kidney combined with the intensity, frequency, and irreversibility of these damaging insults.

Acknowledgment.

The authors thank Dr. Theresa Yung for histological advice and help with methodological validation.

REFERENCES

1. Gjertson DW. Survival trends in long-term first cadaver-donor kidney transplants. Clin Transplant 1991; 225.
2. Matas A. Chronic rejection in renal transplant recipients: risk factors and correlates. Clin Transplant 1994; 8 (3 Pt 2): 332.
3. Matas AJ, Gillingham KJ, Payne WD, Najarian JS. The impact of an acute rejection episode on long-term renal allograft survival (t1/2). Transplantation 1994; 57 (6): 857.
4. Association of chronic kidney graft failure with recipient blood pressure. Collaborative Transplant Study Kidney Int 1998; 53 (1): 217.
5. Pirsch JD, Ploeg RJ, Gange S, et al. Determinants of graft survival after renal transplantation. Transplantation 1996; 61 (11): 1581.
6. Massy ZA, Guijarro C, Kasiske BL. Clinical predictors of chronic renal allograft rejection. Kidney Int 1995; 52 (suppl): S85.
7. Ojo AO, Wolfe RA, Held PJ, Port FK, Schmouder RL. Delayed graft function: risk factors and implications for renal allograft survival. Transplantation 1997; 63 (7): 968.
8. Mihatsch MJ, Nickeleit V, Gudat F. Morphologic criteria of chronic renal allograft rejection. Transplant Proc 1999; 31 (1–2): 1295.
9. Nicholson ML, McCulloch TA, Harper SJ, et al. Early measurement of interstitial fibrosis predicts long-term renal function and graft survival in renal transplantation. Br J Surg 1996; 83 (8): 1082.
10. Nickerson P, Jeffery J, Gough J, et al. Identification of clinical and histopathologic risk factors for diminished renal function 2 years posttransplant. J Am Soc Nephrol 1998; 9 (3): 482.
11. Rush D, Jeffery J, Trpkov K, Solez K, Gough J. Effect of subclinical rejection on renal allograft histology and function at 6 months. Transplant Proc 1996; 28 (1): 494.
12. Seron D, Moreso F, Bover J, et al. Early protocol renal allograft biopsies and graft outcome. Kidney Int 1997; 51 (1): 310.
13. Isoniemi H, Taskinen E, Hayry P. Histological chronic allograft damage index accurately predicts chronic renal allograft rejection. Transplantation 1994; 58 (11): 1195.
14. Dimeny E, Wahlberg J, Larsson E, Fellstrom B. Can histopathological findings in early renal allograft biopsies identify patients at risk for chronic vascular rejection? Clin Transplant 1995; 9 (2): 79.
15. Rush D, Jeffery J, Gough J. Sequential protocol biopsies in renal transplant patients: clinico-pathological correlations using the Banff schema. Transplantation 1995; 59 (4): 511.
16. Kuypers DR, Chapman JR, O’Connell PJ, Allen RD, Nankivell BJ. Predictors of renal transplant histology at three months. Transplantation 1999; 67 (9): 1222.
17. Solez K, Axelsen RA, Benediktsson H, et al. International standardization of criteria for the histologic diagnosis of renal allograft rejection: the Banff working classification of kidney transplant pathology. Kidney Int 1993; 44 (2): 411.
18. Matas AJ, Gillingham KJ, Sutherland DE. Half-life and risk factors for kidney transplant outcome: importance of death with function. Transplantation 1993; 55 (4): 757.
19. Solez K, Benediktsson H, Cavallo T, et al. Report of the Third Banff Conference on Allograft Pathology (July 20-24, 1995) on classification and lesion scoring in renal allograft pathology. Transplant Proc 1996; 28 (1): 441.
20. Racusen LC, Solez K, Colvin RB, et al. The Banff 97 working classification of renal allograft pathology. Kidney Int 1999; 55 (2): 713.
21. Mathew TH. Recurrence of disease following renal transplantation. Am J Kidney Dis 1988; 12 (2): 85.
22. Andresdottir MB, Hoitsma AJ, Assman KJ, Koene RA, Wetzels JF. The impact of recurrent glomerulonephritis on graft survival in recipients of human histocompatibility leucocyte antigen-identical living related donor grafts. Transplantation 1999; 68 (5): 623.
23. Gaber LW, Moore LW, Gaber AO, et al. Correlation of histology to clinical rejection reversal: a thymoglobulin multicenter trial report. Kidney Int 1999; 55 (6): 2415.
24. Flechner SM, Modlin CS, Serrano DP, et al. Determinants of chronic allograft rejection in cyclosporine-treated recipients. Transplantation 1996; 62 (9): 1235.
25. Knight RJ, Kerman RH, Welsh M, et al. Chronic rejection in primary allograft recipients under cyclosporine-prednisone immunosuppressive therapy. Transplantation 1991; 51 (2): 355.
26. Ashraf S, Parrott NR, Dyer P, Roberts I, Johnson RW. Clinical response and temporal patterns of acute cellular rejection: relationship to chronic transplant nephropathy. Transplant Int 1998; 11 (Suppl 1): S5.
27. De Geest S, Borgermans L, Gemoets H, et al. Incidence, determinants, and consequences of subclinical noncompliance with immunosuppressive therapy in renal transplant recipients. Transplantation 1995; 59 (3): 340.
28. Kasiske BL, Kalil RS, Lee HS, Roa KV. Histopathologic findings associated with a chronic, progressive decline in renal allograft function. Kidney Int 1991; 40 (3): 514.
29. Bohle A, Wehrmann M, Bogenschultz O, et al. The long-term prognosis of the primary glomerulonephritides: a morphological and clinical analysis of 1747 cases. Pathol Res Pract 1992; 188 (7): 908.
30. Bohle A, Makensen-Haen S, von Gise H, et al. The consequences of tubulo-interstitial changes for renal function in glomerulopathies: a morphometric and cytological analysis. Pathol Res Pract 1990; 186 (1): 135.
31. Benigni A, Bruzzi I, Mister M, et al. Nature and mediators of renal lesions in kidney transplant patients given cyclosporine for more than one year. Kidney Int 1999; 55 (2): 674.
© 2001 Lippincott Williams & Wilkins, Inc.