Evolution of Endpoints for Renal Transplant Outcome

Short-term graft survival

Table 2 illustrates the immunosuppressive trials with endpoints and outcome used in prospective de novo renal transplant studies during the past two decades. Historically, graft failure at 1 year was used as a definitive endpoint (Figure 1). In a multi-centre European trial, a total of 232 cadaveric recipients were evaluated with CsA (n = 117) therapy and compared with Aza and prednisone (n = 115) treatment (1). After a short follow-up of 11 months, graft survival was found to be superior with CsA therapy (73% vs. 53%). Similarly, short-term graft survival benefit was seen in a Canadian trial with CsA, Aza and prednisone vs. Aza and prednisone therapy (80.4% vs. 64%), p = 0.003 (2). Thus, substantial short-term survival benefit with CsA therapy was detected in a small cohort of patients.

Acute rejection

The benefit of CsA therapy was detected in improvements in 1-year graft survival. For subsequent immunosuppressive therapies, there was a shift from graft failure to acute rejection as an endpoint for measuring transplant results (Figure 1). Acute rejection is an immunological endpoint, which represents under immunosuppression, and it has been shown to adversely influence long-term graft survival (12). During the 1990s, immunosuppressive agents such as MMF, TAC, SR and IL-2 (Interleukin 2) receptor blockers were prospectively evaluated to measure acute rejection rates within 1 year post-transplantation as shown in Table 2 (4–11).

A total of 499 recipients were studied either with MMF 2 or 3 g daily or with Aza therapy (4). Reduced acute rejection rates at 6 months were observed with MMF 2 g daily (19.8%) and 3 g daily (17.5%), compared with Aza (38%) treatment (Table 2). This study used short-term treatment failures (graft loss, death and acute rejection) as an endpoint, rather than acute rejection rates alone, to assess the efficacy of MMF. Treatment failures were substantially lower with MMF (31.1% and 31.3%) compared with Aza therapy (47.6%), p = 0.0015, 0.0021, respectively (4). Similar results were obtained with MMF therapy in the European and Tri-continental trials, and treatment failure was considered as an endpoint to assess the impact of MMF over Aza (5,6). However, pooled analysis of these three studies (n = 1493) revealed significant reductions in short-term acute rejection rates with MMF therapy (13).

Additional trials have used biopsy proven acute rejection as an endpoint. In a randomized trial, TAC (n = 205) was compared with CsA (n = 207) when concomitantly treated with Aza and prednisone. TAC therapy was associated with lower acute rejection rates (30.7%) compared with CsA treatment (46.4%), p = 0.001 (7). In a recent trial, SR 2 and 5 mg daily (n = 284, 274 respectively) were compared with Aza therapy (n = 161) (8). The frequency of biopsy proven acute rejection at 6 months post-transplant was lower with SR 2 mg daily (18.7%) and 5 mg daily (16.8%) therapy compared with Aza (32.3%), p = 0.002, <0.001 (8). Similar results were obtained with SR use in an international study (9). In clinical trials with SR therapy, treatment failures (graft loss, patient loss and acute rejection) were also substantially lower with the trial agent. In the above three trials (FK vs. CsA and SR 2 mg or 5 mg vs. Aza), biopsy proven acute rejection was used as an endpoint. These trials were designed to address short-term acute rejection rates, but not graft survival.

Long-term survival prediction in those who develop acute rejection is dependant on the reversibility of acute rejection. Non-reversibility to baseline results in poor long-term graft survival. A progressive decline in acute rejection rates and absence of perfect correlation with long-term graft failure has limited its usefulness as an endpoint for future trials.

Long-term survival

It would be ideal to consider long-term survival as an endpoint for future studies. The current 5-year renal graft survival rate after cadaveric transplantation is approximately 65%. Functional graft survival (after censoring for death) at 5 years is approximately 75%. Approximately 50% of long-term graft failure during this period is due to death with a functioning kidney, and the remaining 50% due to chronic rejection and recurrent disease (14). However, Table 3 shows the sample size that would have to be studied in each group to detect various differences in 5-year functional survival with various levels of power. Detecting survival improvements by 3% or 5% would require 4195 or 1462 patients in each group.

Patients enrolled in a CsA trial were followed for 4 years (3); those in MMF trials (US, European and Tricontinental) were followed for 3 years (15–17); and those in a TAC trial (TAC vs. CsA) were followed for 5 years (18). The long-term follow-up of the recipients enrolled in these short-term studies has not revealed any survival benefits. Interpretation of intent-to-treat results from a long-term follow-up remains an analytical problem. Measuring projected kidney half-life is an alternative method to evaluate long-term results. However, this approach also requires a large cohort of patients with a minimum follow-up of 3 years. Thus despite its definitiveness as an endpoint, the long follow-up and large cohort size needed for clinical trials preclude long-term survival as a gold standard endpoint. Evaluating alternative short-term endpoints, which can predict long-term survival, offers a more practical solution.

Short-term endpoints predicting long-term survival

Identifying markers within a few months after transplantation that can be used to predict long-term survival is a valuable method for early identification of patients with potentially poorer long-term outcome. Such endpoints can be classified as clinical, histological and immunological types. Potential advantages and disadvantages are shown in Table 1.

Serum creatinine

Post-transplant renal function within a few days after transplantation (19) or at the time of discharge from the hospital (20) has been correlated with long-term survival. Renal function may be estimated by the serum creatinine level. Serum creatinine at 6 and 12 months after renal transplantation has been shown to correlate with long-term graft survival. This has been illustrated in a large database, as well as a single center analysis (21,22). Specifically, analysis of UNOS/OPTN data on 105,742 adult renal transplant recipients has clearly shown that 6-month and 1-year creatinine >1.5 mg/dL correlates with lower long-term graft survival (21). The projected median graft half-life of a cadaveric kidney transplant with creatinine ≤ and >1.5 mg/dL was 15.8 and 8.3 years, respectively (N. Siddiqi, March 2003, unpublished data). For each elevation of serum creatinine by 1.0 mg/dL at 1 year without a change in serum creatinine from 6 months, the risk of graft failure increases by 63%, p < 0.0001. However, when a serum creatinine value increases by 0.5 mg/dL from 6 months to 12 months, the risk of graft failure is increased by 126%, p < 0.0001. The delta creatinine (change in creatinine values from 6 to 12 months) is even more sensitive for its correlation with long-term survival. In the same analysis, when acute rejection and serum creatinine were both included in the model, it was renal function that impacted graft survival, not acute rejection episodes. Thus, renal function served as an important marker, by identifying a group of recipients with elevated serum creatinine, caused by poor response to acute rejection treatment or to other reasons.

To differentiate the efficacy of newer therapy, it is conceivable to use short-term renal function as an endpoint. Table 3 shows the number of patients required for such a study that can evaluate the difference in renal function between the control and the trial group. This sample size is adjusted for 7% graft failure during the first year and assumes an alpha level of 0.05. For example, for 90% power, to detect a 5% and 10% decline in serum creatinine, one will need to evaluate 1977 and 494 patients in each group. Such an evaluation is valid provided that acute rejection rates are similar between groups.

Creatinine clearance

Calculated clearance [(Cockcroft–Gault method, MDRD method) used in clinical practice] or clearance studies, such as 24-h urine collection or nuclear studies (used in clinical trials) have been shown to improve the accuracy of short-term renal function as an endpoint for predicting long-term outcome. In a recent retrospective study, the calculated clearance at 6-months and 1-year has been shown to correlate with long-term survival (23). This justifies the use of serum creatinine as a marker, due to its simplicity over creatinine clearance. Renal function measured by serum creatinine and clearance as an endpoint should be validated in future prospective studies.

Change in 1/creatinine has been evaluated to predict transplant outcome (24). Kasiske et al. evaluated 101 transplant recipients and detected changes in 1/creatinine and creatinine clearance as strong predictors of transplant failure. A decline in 1/creatinine of 40% increases risks of graft failure by 491%, p < 0.0001. However, the authors did not find the slope of 1/creatinine to correlate with transplant failure. It should be noted that these studies were done in patients who were not receiving CsA therapy.

Cystatin C

Estimation of cystatin C is an alternative marker for evaluating renal function. Cystatin C has been compared with serum creatinine and creatinine clearance in renal transplant recipients (25). Plasma cystatin C correlates with plasma creatinine (r = 0.741, p < 0.0001) and creatinine clearance (r = 0.882, p < 0.001). The diagnostic accuracy of cystatin C is slightly better than serum creatinine; however, it is not different from creatinine clearance or β2 microglobulin and does not appear to have any advantage over creatinine clearance. Although it has a marginal benefit over serum creatinine, its cost limits its potential use (26). The use of cystatin C is also dependant on its free availability and validation in large clinical trials.

Elevated serum creatinine is related to immunological events such as acute rejection, which is derived from degree of HLA mismatches, elevated PRA, prior transplantation and prolonged cold ischemia time, and to nonimmunological recipient and donor demographics variables, such as recipient's race (black) and sex (male), and donor age (increasing), race (black) and sex (female). Because these variables influence renal function, short-term renal function is limited as a stand-alone endpoint, in predicting long-term graft survival. In addition, use of noncalcinuerine inhibitor regimen may result in lower serum creatinine, but may result in higher acute rejection rates. Both acute rejection and elevated creatinine are known to impact long-term survival. Thus, a balance between acute rejection rates (which signifies lower immunosuppression) and elevated creatinine (which represents poor long-term survival) is an important measure in evaluating transplant success.

Histological acute rejection

Until recently, renal histological evaluation, a gold standard in evaluating transplant renal dysfunction, was reserved for patients who experienced renal dysfunction. Rush and co-workers have now performed renal biopsies in transplant recipients with stable renal function (27). The diagnosis of subclinical acute rejection has resulted in preservation of renal function without impacting graft success on long-term follow-up (27). However, a recent study revealed a low prevalence of silent acute rejection with MMF and TAC combination therapy (28). Although the additional costs associated with detecting silent acute rejection and the ability to ensure a 100% biopsy rate are a challenge to logistics, future studies should include protocol biopsies to detect silent rejection as an alternative endpoint.

Chronic allograft nephropathy

Chronic allograft nephropathy (CAN) is an important cause of late graft failure that has not been evaluated in clinical trials, mainly because a large patient pool is required to evaluate the difference between protocols with prolonged follow-up (Table 3). Instead, focus should be directed to diagnosing clinical or subclinical histological chronic changes.

Renal scarring has been quantified using chronic indices such as the Banff Score Index and the Chronic Allograft Disease Index (CADI) score, and has been correlated with late graft failure (29,30). The CADI score from protocol renal biopsies at 2 years correlates with graft function at 6 years (r = 0.717, p = 0.0001) with a strong prediction to development of CAN (29). Histological findings, such as macrophages in the renal graft (peritubular capillary lesions detected by electron microscopic evaluation within 1-year post-transplant) also have been correlated with histological chronic rejection (31,32). A computerized histomorphometric assessment of protocol renal biopsies also has correlated with lower glomerular filtration rates (33). The fraction of collagen III immunostain > 40% at 6-months post-transplant correlates with GFR at 2 years. A simple histological evaluation using CAN with and without transplant vasculopathy has been shown to correlate with long-term survival (34). The data suggest that the early histological evaluation after transplantation or perhaps quantifying the change in degree of scarring from transplantation to 3 or 6 months post-transplantation are potential endpoints worth validating in clinical trials. However, the invasive nature of biopsies and associated time, cost and morbidity are the key limiting factors. The ability to assess renal histological parameters also depends on the availability of this information on all patients enrolled in a clinical trial.

Increasing donor age may adversely affect graft survival (35). Differentiating the histological features of donor disease from the early events after transplantation (DGF, acute rejection and drug toxicities) remains a huge challenge. Nevertheless, these histological endpoints such as the Banff chronicity scores are important and should be included in future clinical trials. With limited organ availability and possible future secondary interventional trials, which may delay graft failure, early detection of chronic histological findings is an important marker to be considered for transplant evaluation.

Currently used endpoint such as acute rejection and proposed endpoints such as renal function and scarring have a good correlation, but without satisfactory positive and negative predictive values. Post-transplant renal function predicting long-term survival is also dependant on duration of follow-up and progression of renal disease. Immunosuppressive therapy without a calcinuerine inhibitor and management of hypertension with angiotensin converting enzyme inhibitor or angiotensin receptor blocker can alter the progression of renal failure. Identifying an appropriate surrogate marker with a high positive and negative predictive value will remain a challenge.

Anti-donor antibodies

The degree of HLA mismatching has had a substantial effect on short- and long-term transplant outcome (36). The detection of anti-donor antibodies early after transplantation has been correlated with immunological injury and renal dysfunction (37). Prospective clinical studies are warranted to evaluate the use of anti-donor antibodies as a marker for measuring transplant outcome.

Tissue, blood and urine immune markers

The immunological acceptance of a graft can be evaluated using blood, tissue and urine cytokines. Transforming growth factor-β (TGF-β) has been used as a marker for acute rejection, and has also been correlated with calcineurin inhibitor toxicity and nonspecific fibrosis (38). Markers such as granzyme B and perforin have more specificity for acute rejection. Li et al. have demonstrated that a measurement of perforin and granzyme from urinary cells from transplant recipients is correlated with acute rejection (39). However, these require confirmation and lack very good sensitivity and specificity. Although the reproducibility and availability of such markers in clinical trials and practice will remain a challenge, they are attractive future endpoints, due to their noninvasive nature.

Morbidity and mortality

The risk of death after renal transplantation is higher during the first 100 days post-transplant, when compared with patients who are maintained on dialysis treatment (40). However, after 100 days, the overall risk of death falls to less than that of wait-listed dialysis patients (40). Still, the most important cause of long-term graft failure is death with a functioning graft (41). Mortality after transplantation is a component of composite endpoint; however, it cannot be used as a solitary endpoint, as few deaths occur during the first year.

Immunosuppression cannot be achieved without immunosuppressive side-effects. Overt over- immunosuppressive side-effects seen after transplantation are infections (CMV, EBV, pnuemocystis carinii, polyoma virus) and tumors (lymphomas, skin cancers), which should be included in transplant outcome measurement. Similarly, nonimmunosuppressive-related morbidities, such as cardiovascular risk factors (hypertension, diabetes and hyperlipidemia) are important events, which adversely influence long-term graft and patient survival. As the rates of acute rejection and late graft failure continue to decline, morbidity is becoming increasingly relevant in clinical practice. Measuring morbidity endpoint is not only vital to enhance long-term patient and graft survival, but also provides an opportunity to identify their correlation with newer therapies. The number and percent of patients experiencing efficacy failures (acute rejection, immunological graft loss) and mortality/morbidity endpoints (death, nonimmunological graft loss, infections and tumors) should be measured in future clinical trials. Similarly, morbidities that affect long-term survival (e.g. cardiovascular risk factors – hypertension, hyperlipidemia and diabetes) should be measured separately.