Chronic Allograft Failure in Human Renal Transplantation

Abstract

Objective

To identify potential risk factors for the development of chronic renal allograft failure.

Summary Background Data

Chronic allograft failure (CAF) is the leading cause of late graft loss in renal transplantation. The authors studied the risk factors for the development of CAF in a single center during a period in which a consistent baseline immunosuppression regimen (cyclosporine, azathioprine, and prednisolone) was used.

Methods

Data from the Oxford Transplant Center Database were assessed on 862 renal allografts during a 10-year period. Risk factors were identified using multivariate logistic regression analysis.

Results

Biopsy-proven CAF occurred in 77 patients (9.2%) in the entire group. Multivariate risk factor analysis revealed that early and late acute rejection episodes, proteinuria, and serum triglycerides were significant factors. Acute rejection after 3 months was more important than early acute rejection. Serum triglyceride level and proteinuria at 1 year were both elevated in the CAF group. Male sex provided a protective effect. Serum creatinine levels at 6 months after the transplant were not predictive of the risk of developing CAF.

Conclusions

These results from the largest single-center review to date suggest that both antigen-dependent and -independent factors are involved in the pathogenesis of CAF. Acute rejection at all time points has a significant impact on the development of CAF.

Renal transplantation is the optimal mode of replacement therapy in most patients with end-stage renal disease. There has been progressive improvement in short-term patient and graft survival rates in the “cyclosporine era.” However, in the longer term, there is a persistent graft loss of 3% to 4% annually. This attrition is due in part to death with a functioning graft, but in most series the most common cause of graft loss is chronic allograft failure (CAF). ¹

The pathogenesis of CAF is unclear. Pathologic changes seen in transplant core biopsies include glomerulosclerosis, tubular atrophy, interstitial fibrosis, and arteriosclerosis, ² all of which are nonspecific, the exception being arteriosclerosis, which may not be seen in all biopsies. In addition, the nonspecific features may even be present in donor biopsies taken before transplantation. ³ Further, the pathologic features of CAF can be detected as early as 6 months after the transplant, well before the clinical features are evident, ^4,5 suggesting that the ability to predict poor outcome at an early stage, whatever the mechanism, will be important in reducing graft loss from this condition.

Theories of CAF pathogenesis suggest that both antigen-dependent and antigen-independent pathways are involved. ^6,7 Antigen-independent injury is thought to arise from inadequate nephron supply, ⁸ ischemia–reperfusion injury, ⁹ and drug toxicity. Antigen-dependent injury is thought to arise from the alloimmune response by both direct and indirect antigen recognition pathways, ¹⁰ in addition to the poorly quantified role of noncompliance.

Although several studies implicate acute rejection as being the major risk factor for CAF, ^11,12 not all reported studies agree. ¹³ Metabolic factors such as serum lipids and hypertension have also been implicated, but their exact role is ill defined. ^14,15

There have been few large-scale studies that have examined multiple risk factors in a stable transplant population followed long term by a single center. We therefore sought to analyze the risk factors for CAF in a single-center, retrospective study in which data were collected prospectively during a period of stable and uniform immunosuppression.

PATIENTS AND METHODS

Pretransplant demographic and clinical and posttransplant variables were collected prospectively, forming the basis of the Oxford Transplant Center database. These included details of recipient age, gender, diabetic and transfusion history, previous transplant, panel reactive antibody, human leukocyte antigen (HLA) typing, immunosuppression, early graft function, levels of serum creatinine, triglyceride, cholesterol, and proteinuria at 1 year, graft and patient survival, along with donor age, gender, and HLA typing. Follow-up was carried out by the Oxford Transplant Center for all patients, with the exception of those who moved to another area, in which case follow-up was actively sought once a year. Data were verified and data on acute rejection episodes, blood pressure control, and serum lipids were collected by retrospective review of the notes and flow charts. Three sets of notes were not found.

The study period covered 1985 to 1996, during which there were 862 transplants, 804 cadaveric and 58 living related. During this time, 26 additional living related grafts were performed on overseas patients; because long-term follow-up was not available, these were excluded. The immunosuppression protocol used was uniform throughout the study and consisted of a triple-therapy regimen of cyclosporine, azathioprine, and prednisolone, as previously described. ¹⁶ A group of patients (n = 17) received mycophenolate in place of azathioprine. Induction therapy with either methylprednisolone or antilymphocyte globulin was used in highly sensitized patients. Prednisolone was started at 20 mg/day, with gradual reduction of the dosage started at 2 months after the transplant, down to 5 mg/day at 1 year, after which it was discontinued in most patients during a 6-month period.

Acute rejection was defined using clinical and biochemical criteria, including an elevation greater than 15% above baseline of serum creatinine and reduction in urine output and response to antirejection therapy. The majority (95%) of acute rejection episodes were confirmed by biopsy. First rejection episodes were treated with pulsed intravenous methylprednisolone (500 mg/day for 3 days); continuing rejection was treated with a further course of methylprednisolone or antilymphocyte agents.

CAF was defined according to the definition established by the Alexis Carrel conferences. ^17,18 This suggests that there should be a progressive decline in renal function during the course of at least 3 months in the absence of another cause (e.g., recurrent glomerulonephritis, renal artery stenosis or obstruction) and supported by a transplant biopsy showing the characteristic histologic features of CAF—namely, glomerulosclerosis, tubular atrophy, interstitial fibrosis, and arteriosclerosis.

The main outcome measure was the development of CAF. Univariate analysis was carried out using the Student t test for continuous data and the chi-square test for categorical data. Multivariate analysis was carried out by logistic regression using forward stepwise selection with entry and exit criteria set at the P = .05 level. Continuous variables were categorized to obviate linearity assumptions. Variables not selected by the initial analyses were reentered individually to check for residual confounding. The multivariate analysis is represented with odds ratios (OR) for individual risk factors, where the odds ratio represents the odds that CAF will occur in an individual with a specific characteristic.

RESULTS

The distribution of various demographic and clinical characteristics for the study population is shown in Table 1. There were 117 failures in the first 6 months after the transplant, and these grafts were excluded from the analysis. Biopsy-proven CAF was present in 77 patients (9.2%), representing 10.4% of living related donor transplants and 9.1% of cadaveric transplants.

Table 1. DEMOGRAPHIC DATA FOR RECIPIENTS, DONORS, AND GRAFTS

Numbers represent frequencies or means (standard deviation) as appropriate.

Multivariate analysis (Table 2) showed that late acute rejection (>6 months after transplantation) was a major risk factor for CAF (OR 5.91;P < .0001), although early acute rejection (within 3 months) was also significant (OR 2.31;P = .0161). Metabolic factors including hypertriglyceridemia at 1 year (>2.7 mmol/L; OR 3.06;P = .0008) and proteinuria at 1 year (>0.1 g/L; OR 5.07;P < .0001) were also highly significant. Younger recipients (age younger than 50 years at the time of transplant) were at higher risk than older recipients (OR 2.51;P = .0132). It also appeared that the male sex had a protective effect (OR 0.35;P = .0015). Renal function, determined by serum creatinine at 6 months, had a minimal impact on the final model (OR 1.0035;P = .032).

Table 2. RISK FACTORS FOR CHRONIC ALLOGRAFT FAILURE IN A MULTIVARIATE ANALYSIS

Considering the number and severity of acute rejection episodes, we found that the number of episodes was a risk factor (16.1% CAF after one or more acute rejection episodes vs. 4.2% after none;P < .0001) (Table 3). Rejection was significant even when it occurred within the first 10 days after transplantation (15.3% CAF;P = .002).

Table 3. IMPACT OF ACUTE REJECTION (AR) EPISODES ON THE LATER DEVELOPMENT OF CHRONIC ALLOGRAFT FAILURE

The severity of rejection episodes was examined by considering a subgroup of grafts that suffered a rejection episode in the first 90 days only (Table 4). If the serum creatinine returned to within 10% of the prerejection baseline levels, there was a late risk of CAF of 14.1% (P < .001). However, grafts where renal function failed to return to baseline levels or that had a steroid-resistant rejection showed an increased incidence of CAF (26.2% and 19.4%, respectively;P < .0001 and P < .001).

Table 4. EFFECT OF SEVERITY OF ACUTE REJECTION ON THE INCIDENCE OF CHRONIC ALLOGRAFT FAILURE

Figures represent a defined subpopulation of grafts where rejection severity could be clearly defined.

There were no significant differences in the degree of HLA matching nor in the amount of immunosuppression to explain this finding, and indeed the steroid withdrawal rate was lower in the CAF group than in controls (21/77 [27.2%] vs. 382/642 [59.5%];P < .0001) (see Table 1), whereas cyclosporine doses and levels were similar (see Table 2).

The strict definition of CAF used to define the study group raised the possibility that some grafts would have been inappropriately excluded from the study group. Similarly, the control group used included all other grafts and therefore included grafts that may have developed CAF given longer follow-up. To control for these confounding factors, the multivariate analysis was repeated first with a control group identified from grafts surviving more than 5 years and second with a less strict definition of CAF (i.e., including grafts where the diagnosis of CAF was not formally proven by core biopsy). Neither of these changes resulted in a significant change in the multivariate model.

DISCUSSION

Clinically, CAF is the leading cause of late graft failure in renal transplantation. Current theories of pathogenesis suggest that antigen-dependent factors interplay with antigen-independent factors. ^6,7 These are supported by reports from animal models of CAF that also implicate immunologic and metabolic factors. ^19–21

Analysis of the risk factors for CAF in our study shows that acute rejection at all time points is important. Late rejection episodes, defined here as those occurring after 3 months, are especially deleterious (OR = 5.3 for late vs. 2.3 for early episodes). This result differs from that found in several studies looking at long-term survival ¹² and CAF, ^13,22 which suggested that early rejection episodes did not have a significant late impact. The explanation for this is not clear, but there are several methodologic possibilities that must be considered.

First, it is important to draw a distinction between CAF and graft loss. This analysis is only of biopsy-proven CAF. Focusing on CAF as an endpoint is more specific than examining risk factors associated with graft or patient survival times.

Second, the definition of CAF used by many studies varies. The internationally accepted definition requires the confirmation of CAF with a transplant core biopsy. This is critical not only for confirmatory purposes but also to exclude recurrent disease or de novo glomerulonephritis as a cause of declining graft function. All patients included in our study group had confirmatory biopsies; only one other study has used similarly stringent criteria. ²³

Third, the immunosuppressive protocols used in different centers vary from that used in Oxford. Studies showing that early acute rejection is not important have used intensive early immunosuppression, often with antilymphocyte agents. ¹² This is likely to alter the quality of the acute rejection episodes experienced by the grafts, if not the frequency. In addition, during the time period of the study, the immunosuppressive regimen was homogeneous in our center.

Fourth, the classic exponential survival curve seen in renal transplantation is formed as a result of a high rate of graft loss in the first few months and a lower but often stable attrition rate thereafter. We have excluded factors that have effects solely confined to the early high-risk period by censoring the data, excluding grafts that survived less than 6 months.

Late acute rejection is acknowledged to be a key risk factor for CAF. Our data confirm this and further illustrate that rejection severity is an important factor. Rejection severity has been implicated as a risk factor for CAF in animal work, where the area under the curve of serum creatinine has been correlated with late histopathologic changes. ²⁴ Possible mechanisms for this include either an increased nephron loss, with late or more severe rejection leading to hyperfiltration injury, or to the development of a self-perpetuating fibrotic repair response to injury, analogous to that seen in animal models of renal damage. ²⁵ Experimental models of CAF have shown that removal of the alloimmune response at an early stage can reverse the pathologic changes of CAF, but not the later stages of the process. ²⁶ This supports the concept of an ultimately self-perpetuating process.

It is also possible that acute rejection results in chronic rejection only when the immune response is severe. This concept fits well with our data. A severe immune response may exceed the required threshold for self-perpetuating injury. Alternatively, acute rejection may cause CAF only when additional risk factors are present, such as those discussed below.

The lack of a correlation between CAF and the degree of HLA matching is surprising, particularly given the strong association between acute rejection and HLA mismatch. ²⁷ An unequivocal association between HLA mismatch and late CAF would strongly support the central role of an immunologic mechanism in its pathogenesis. Our finding is, however, in agreement with results from other centers. ^28,29 One group has reported an association between HLA-DR mismatches and CAF, but this was only in a univariate analysis, and the effect was no longer apparent in a multivariate model. ¹² A potential explanation is that the numbers studied in single-center analyses are insufficient to draw valid conclusions. However, there is little evidence in the data to suggest a trend toward significance. It is also possible that the qualitative effect of acute rejection is more important than a rejection episode per se.

An alternative explanation is that the real significance of acute rejection is not the event itself but the subsequent clinical response to it in terms of antirejection therapy, which may inhibit the development of tolerance. Recent work in experimental transplantation models has shown that cyclosporine inhibits the development of tolerance with T-cell costimulatory blockade, suggesting that cyclosporine and potentially other nonspecific immunosuppressants may block normal regulatory mechanisms controlling T-cell activation and proliferation. ^30–32

There was no convincing evidence in support of the hyperfiltration theory of chronic graft loss. Kidneys from female donors had the same incidence of CAF as those from male donors, and there was no effect of recipient weight. There was, however, a significant effect of donor age.

Our analysis has also shown that proteinuria and triglyceride levels are independent predictors of CAF. These metabolic factors are not simply markers of poor renal function, because creatinine is included in the multivariate model, and they therefore operate independently of renal function. Although proteinuria has long been associated with CAF, it has been difficult to establish whether it is a causative or merely an associated factor. Most renal transplants have some degree of protein leak, ³³ resulting from a failure of permselectivity of the glomerular basement membrane. ³⁴ This is common to most progressive renal diseases and is the strongest single predictor of renal disease progression. ³⁵

There is evidence that protein traffic through the glomeruli and tubules is damaging as a result of nonspecific inflammatory changes. Filtered proteins are usually absorbed by endocytosis in the proximal tubules. ³⁶ Damage to the tubules may occur by several different mechanisms, including lysosomal rupture, ammonia production activating complement, and the release of chemoattractants and cytokines. ³⁷ This results in the accumulation of macrophages, which may themselves recruit and stimulate fibroblasts. The deleterious effects of glomerular macromolecule traffic may also apply to lipids (e.g., triglycerides). The association of triglycerides with CAF is not novel but has been found in only a few studies, ^12,14 and the mechanism of any effect is unclear. The histologic similarities between transplant arteriosclerosis and atherosclerosis ³⁸ and the well-established role of dyslipidemia in atherosclerosis suggest that lipid abnormalities may be involved in the pathogenesis of CAF. Abnormal lipid accumulation in glomeruli is a recognized early event in glomerulosclerosis, and oxidized low-density lipoprotein has been shown to increase the expression of transforming growth factor beta, a known profibrotic cytokine, in human glomerular epithelial cells. ³⁹ Oxidized low-density lipoprotein has also been demonstrated in the thickened intima of arteriosclerotic renal allografts. ⁴⁰

In normal glomeruli, there must exist mechanisms to prevent the accumulation of internalized lipid, because this does not occur even in hyperlipidemic states, ⁴¹ and nontransplant patients with hyperlipidemia do not develop renal insufficiency in the absence of coexistent renal disease. Mechanisms contributing to lipid accumulation may include a failure of the normal clearance pathways or nonreceptor scavenger-type uptake by graft-infiltrating macrophages. Posttransplant hyperlipidemia is common ¹⁴ and resistant to conventional dietary management. Although data on pretransplant lipid levels are not uniformly available in Oxford, previous studies have shown that posttransplant levels correlate well with pretransplant levels and also with the histologic changes of CAF. ⁴² Serum triglyceride would appear from our data to be the most significant lipid group. This is consistent with data from cardiac transplant recipients, where increased triglyceride levels have also been found to be associated more closely with the development of chronic graft arteriosclerosis than cholesterol.

The cause-or-effect question may be answered only when clear evidence emerges from randomized trials investigating the effects of lipid-lowering therapies on the incidence of CAF. As yet, there is no clinically proven therapy for established CAF. Clinical trial design is problematic, particularly in the definition of end points. Surrogate end points may be used, such as the progression of fibrosis ⁵ or the change of slope of glomerular filtration rate with treatment as opposed to graft loss from CAF. ⁴³ Surrogate end points offer the potential to assess treatment effects during a short follow-up period, but there is a risk of type I errors. A further problem is in defining the likely effect of any treatment to generate valid power calculations for a study, because there have been no previous studies demonstrating a treatment effect on which to base a full-scale trial.

Multivariate analyses such as the one presented here offer an alternative approach. Odds ratios can be used to predict the percentage chance of an individual graft developing late CAF. Initial studies could then be performed by entering a patient group with a predicted likelihood of developing CAF over the baseline rate, thus reducing the number of grafts required for any given study. This approach is likely to prove particularly beneficial in CAF because there is no clear idea from the pathogenesis as to which therapeutic agents are likely to provide the greatest benefit. This type of approach will also allow the performance of preliminary studies required to estimate the magnitude of any treatment effect before use in large-scale primary prevention studies.

In conclusion, we have demonstrated in the largest study to date of biopsy-proven CAF that several factors are implicated in its development. Although not all may be causative, they allow prediction of the risk of developing CAF at an early stage before deterioration in graft function. This should allow the identification of a high-risk cohort and the targeting of novel therapeutic agents.