Chronic Progressive Calcineurin Nephrotoxicity: An Overstated Concept

Prospective randomized studies

There are major problems with the data purported to show progressive CNI nephrotoxicity after kidney transplantation. First, there are no prospective randomized studies that clearly demonstrate CNI nephrotoxicity to be responsible for a significant proportion of late graft dysfunction. Just the opposite—most studies have shown that CNI-free immunosuppression provides no long-term benefit (5–11). Yes, initial eGFR is better in recipients not taking CNIs. But there is no difference in the slope of eGFR versus time in those taking or not taking CNIs. In addition, CNI-free protocols have their own drug-specific complications and limitations.

Overdiagnosis of ‘CNI nephrotoxicity’

Second, there are no clinical or histologic parameters that are diagnostic of chronic CNI nephrotoxicity (1); as a consequence, CNI nephrotoxicity may be overdiagnosed. For CNI-immunosuppressed kidney transplant recipients who develop slow deterioration of graft function (or extrarenal-renal transplant recipients who develop native kidney dysfunction), a kidney biopsy is often not done and the clinical diagnosis of ‘CNI nephrotoxicity’ is made. This diagnosis is then entered into the recipient's chart or into a database (including registry databases). Retrospective analyses of these databases attribute dysfunction and kidney failure to CNI nephrotoxicity. Alternatively, if a biopsy is done and shows fibrosis and atrophy, the pathologist, in the absence of any other specific diagnosis, often interprets the biopsy as consistent with CNI nephrotoxicity. This was seen in the deterioration of the kidney allograft function (DeKAF) study in which patients with new onset late kidney allograft dysfunction underwent a percutaneous allograft biopsy (12). In 30% of the cases, the biopsy was interpreted as being consistent with CNI nephrotoxicity. For these recipients, if there was no circulating donor-specific antibody (DSA) and if histology showed no inflammation and was not C4d positive, prognosis was excellent.

Concerns regarding late graft dysfunction and graft loss after kidney transplantation

It has been noted that although CNI-based immunosuppression has resulted in significant improvement of short-term outcome, there has been little parallel improvement in long-term outcome. This observation has been interpreted as suggesting that any early survival gain is countered by CNI nephrotoxicity. However, there is an alternate explanation. Half-lives (the length of time until 50% of grafts surviving 1 year subsequently fail) have not changed (or have increased) since the introduction of CNIs (13). Therefore, for recipients on CNIs whose grafts survive 1 year, there is no decrease in long-term graft survival (vs. historical CNI-free protocols)—suggesting that chronic progressive CNI nephrotoxicity is not affecting the grafts.

Why then might CNIs result in improving early but not long-term graft survival? One possibility is that although CNIs decrease acute rejection rates (and increase early graft survival), they have no, or minimal, impact on other factors responsible for late graft dysfunction and late graft loss (e.g. noncompliance, recurrent disease, chronic antibody-mediated rejection).

Progression of histologic lesions on protocol biopsy after kidney transplant

Major support for the concept of chronic CNI nephrotoxicity after kidney transplantation came from Nankivell et al. who reported on sequential biopsies in 120 simultaneous kidney-pancreas recipients immunosuppressed with cyclosporine (CSA), prednisone, and azathioprine (AZA) (14). Nankivell et al. described the progressive development of glomerulosclerosis, periglomerular fibrosis, and totally sclerosed glomeruli. Of note, the 10-year death-censored graft survival for their patient population was 95%; mean serum creatinine level was 1.6 ± 0.5 mg/dL. Therefore, although histologic abnormalities certainly developed, the long-term outcome was excellent. Importantly, there are numerous problems with Nankivell et al.'s interpretation of their data. First, this was not a randomized series; all recipients were on CNIs. The development of histologic lesions may have been due to CNIs but also could have been due to other common factors. Second, there were relatively few late observations; the median histologic follow-up was 3.9 +/− 3.3 years. Third, and perhaps most important, subclinical rejection remained a significant clinical problem in their series, occurring in 19.5% of biopsies done 2–5 years post-transplant, and 12.3% of biopsies done 6–10 years post-transplant. Thus, it is quite plausible that the development of progressive histologic lesions was due to ongoing subclinical rejection. Finally, two-thirds of the fibrosis that was present at 10 years had already appeared by 1 year; there was little progression beyond this point.

Perhaps the most powerful observation challenging Nankivell et al.'s interpretation comes from a subsequent paper by the same group in which they reported on sequential biopsies in a second cohort of recipients (15). This cohort was different in that mycophenolate mofetil (MMF) replaced AZA in the immunosuppressive protocol. The authors noted that the MMF-treated recipients had decreased acute rejection and decreased need for OKT3 treatment (vs. the earlier cohort). In addition, this was associated with ‘limited chronic interstitial fibrosis, striped fibrosis, and peri-glomeruli fibrosis (p < 0.05 to.001), mesangial matrix accumulation (p < 0.01), chronic glomerulopathy scores (p < 0.05), and glomerulosclerosis (p < 0.05)’. Nankivell et al. reported that the ‘MMF-treated patients had reduced arterial hyalinosis, striped fibrosis, and tubular microcalcification’. Therefore, they reported a significant minimization of the lesions they associated with CNI nephrotoxicity by using MMF instead of AZA while maintaining CSA use (and decreasing their rates of rejection, and presumably of subclinical rejection).

In contrast to Nankivell et al., other clinical series have not shown CNI nephrotoxicity to be responsible for a significant proportion of late graft dysfunction. Humar et al. reported graft survival for CNI-immunosuppressed kidney recipients after excluding those with death with function, technical failure, primary nonfunction and recurrent disease (16). The actuarial 10-year graft survival for those with no rejection was 91% (vs. 45% for those with ≥1 rejection) (p = 0.001). CNI toxicity was a rare cause of graft loss in either group. El-Zoghby et al. reported on 330 graft losses in 1317 recipients (17). Of the 330, 138 (43.4%) were lost due to death with function, 39 (11.8%) were lost due to primary nonfunction, and 156 (46.3%) were lost due to other causes (18). The latter group was subdivided by cause: glomerular disease (37%), fibrosis and atrophy (31%), medical or surgical causes (16%), acute rejection (12%) and unclassifiable (5%). Of those with fibrosis and atrophy (representing 15% of the total population of graft loss), only one case was attributed to CNI nephrotoxicity. Other reports have shown no subsequent progressive deterioration of function when transplant biopsies showed only fibrosis and atrophy (and no inflammation). Only those grafts with evidence of active inflammation at the time of graft biopsy progressed to graft failure (18,19).

Concerns regarding development of renal dysfunction after extrarenal transplants

A major argument for the existence of chronic CNI nephrotoxicity has been the development of kidney dysfunction in extrarenal transplants. Myers et al. first reported that 12-month post-transplant eGFR was significantly lower in 17 CSA-immunosuppressed heart transplant recipients than in CNI-free historical controls (20). However, the CSA doses used by Myers et al. were extraordinarily high—the dose at transplant was 17.5 mg/day and the trough levels ranged from 300 to 350 ng/dL for the first 4 months post-transplant and were still 164±18 ng/mL at 2 years post-transplant (21). Even with these trough levels, the renal function in the majority of patients remained stable.

More recently, Ojo et al. described the cumulative incidence of chronic renal failure among 69 000 patients receiving extrarenal organ transplant and reported to the OPTN database (22). Multivariate analysis showed the important risk factors for renal dysfunction to be increasing age, pretransplant hepatitis C, diabetes mellitus, postoperative acute renal failure, female gender and hypertension. Use of a CNI was not significant. Also of note was that the rate of development of chronic renal failure was not related to organ-specific CNI target levels. For example, liver transplant recipients who are targeted for relatively low short- and long-term levels had a much higher rate of chronic renal failure over the first 10 years than did lung, heart, or heart-lung transplant recipients (who are targeted for higher CNI levels).

Recent single-center analyses have similarly noted that CNI use was not a risk factor for chronic renal dysfunction. In heart transplant recipients (n = 352), Hamour et al. reported that risk factors for low eGFR were: need for postoperative renal replacement therapy, pretransplant diabetes, increasing recipient age and female recipient or donor gender (23). CNI use was not shown to have a long-term impact on renal function. Navarro-Manchon et al. reported that elevated pretransplant serum creatinine, CMV infection and diabetes were risks for post-transplant renal dysfunction after heart transplantation; interestingly, use of an interleukin-2 receptor inhibitor and MMF (vs. AZA) were protective (24).

Liver failure, itself, is associated with renal dysfunction. Numerous series have shown glomerular lesions at autopsy (12%–100%, with an overall rate of 45%) in patients with liver cirrhosis (25,26). In two liver transplant studies, a kidney biopsy was done at the time of transplantation (26,27). Both studies showed significant rates of histologic abnormalities. In Axelson et al.'s study (n = 23), 8 had glomerular lesions, 2 IgA nephropathy, 1 angio capillary GN type 1 and 12 had minor glomerular abnormalities (26). In McGuire et al.'s study (n = 30 with hepatitis C-induced cirrhosis), only 1 had no histologic abnormalities (27). Of the 30, 12 had MPGN type I, 7 had IgA nephropathy, 6 had angio glomerulonephritis and 4 had minor glomerular abnormalities.

In other liver transplant studies, a kidney biopsy has been done at the time of new-onset late renal dysfunction. Phillebout et al. reported on 26 patients with post-transplant kidney biopsies done at a mean of 4.8 ± 0.7 years post-transplant; of these, 5 had renal failure pretransplant, 8 had pre-existing diabetes, and 9 had hypertension (28). The authors noted multiple significant histologic lesions including severe arterial lesions in 65%, hydroxy starch nephropathy in 50%, thrombotic microangiopathy in 46%, diabetic lesions in 34% and FSGS in 34%. Those with hepatitis C recurrence had worse glomerular lesions. More recently, Kim et al. reported on 80 kidney biopsies done at a mean of 5 years post-transplant; all biopsies showed glomerular lesions (29). Kim et al. concluded, ‘Our findings suggest that chronic kidney disease post orthotopic liver transplant may only rarely be ascribed to CNI toxicity and instead has a complex and varied pathologic basis’. Sanchez et al. followed long-term (15-year) renal function after liver transplantation (30). They showed that the lower the GFR is after transplant, the sooner renal failure develops. However, for most recipients, function was stable for the 15-year follow-up.

In CNI-treated pancreas recipients, long-term histologic follow-up of native kidneys showed that between 5 and 10 years post-transplant there was significant improvement in the histology (31).

The strongest data to support CNI nephrotoxicity are the data from CNI-treated patients with autoimmune diseases (32). However, even here there is controversy: (a) some autoimmune disease are associated with kidney lesions; (b) when used, CNIs are often associated with reduced renal function in a small percent of patients; and (c) usually renal function returns to normal if CNIs are stopped or reduced (32–34).

CNI-free protocols have not shown an advantage

There is considerable data that refute chronic CNI nephrotoxicity. First, in prospective randomized studies of CNI-containing vs. CNI-free protocols, there has been no long-term benefit for the CNI-free groups (5–11). A number of studies have randomized recipients on antibody, MMF, and prednisone to CNI vs. sirolimus. Larson et al. reported no difference between groups at 12 or at 24 months in patient survival, graft survival, acute rejection rates, measured GFR and no difference in histology on protocol biopsies (5). Buchler et al. reported no difference in 12-month patient survival, graft survival, acute rejection rates or eGFR (6). Similarly, Glotz et al. reported lower 12-month graft survival in the sirolimus group and no significant difference in GFR (7). In the Symphony study, Ekberg et al. randomized patients treated with MMF and prednisone to four arms: standard dose cyclosporine, or IL-2R induction with low-dose cyclosporine, low-dose tacrolimus, or low-dose sirolimus (8). The worst results were seen in the CNI-free arm. More recently, two large trials randomized recipients on antibody, MMF, and prednisone to CNI vs. Belatacept (9,10). The CNI-treated recipients had higher serum creatinine levels and lower GFR. However, in both groups, creatinine and GFR were stable over 2 years.

Stable renal function in recipients on long-term CNIs

A second line of evidence refuting chronic CNI nephrotoxicity is that many recipients have done well on long-term CNIs. Kandaswamy et al. reported on 1263 patients with graft survival ≥1 year and remaining on CNIs (3). In this group, the mean serum creatinine level and calculated creatinine clearance were stable through 20 years. Thus, CNI nephrotoxicity, if real, certainly does not affect all grafts.

Alternative explanations for chronic graft dysfunction

Perhaps most important is that there are alternative explanations for most late renal dysfunction. The concept of chronic CNI nephrotoxicity evolved in an era where our diagnostic armamentarium was not nearly as sophisticated as it is today. It was known that acute CNI nephrotoxicity existed. Therefore, it is not surprising that when CNI-treated transplant recipients developed late renal dysfunction, it was attributed to CNIs. However, even during that era, alternative explanations existed. Those for extrarenal transplant recipients are described above.

In kidney transplantation, it has been known since the early 1990s that recipients having ≥1 acute rejection episodes were at increased risk for late graft dysfunction and graft loss (35). Humar et al. showed that 10-year graft survival, in the absence of an acute rejection episode, was excellent (16); Nankivell et al. minimized ‘CNI nephrotoxicity’ by using a more powerful (CNI-based) immunosuppressive regime and decreasing subclinical rejection (15). In the 1990s, Rush et al. documented the negative impact of untreated subclinical rejection on long-term graft outcome, and this has been confirmed by others (36). And patients having protocol biopsies or biopsies for dysfunction have had no subsequent decline in renal function if the biopsies showed no evidence of active inflammation (or recurrent disease) (18,19). Of importance is that since by Banff criteria inflammation in the area of scarring was not scored, the degree of inflammation in most studies may have been underreported. Mannon et al. recently showed that inflammation and tubulitis in regions of fibrosis and atrophy were strongly correlated with each other (p < 0.0001), and that inflammation solely in these areas was strongly associated with death-censored graft failure when compared to recipients whose biopsies had no inflammation (even after adjusting for the presence of interstitial fibrosis, tubular atrophy, serum creatinine at the time of biopsy, time to biopsy and i-score (37). Mengel et al. reported that scoring of total inflammation (i.e. in both scarred and non-scarred areas) was a better predictor of graft survival than the Banff i-score and all current diagnostic Banff categories (reviewed in 38). Thus, historically, immune-mediated damage may have been underappreciated, and the progressive graft dysfunction erroneously attributed to CNIs.

Recently, with the development of sensitive technology to measure development of donor-specific antibody and to diagnose antibody-mediated rejection (AbMR), there has been a revolution in our understanding of chronic graft dysfunction after kidney transplantation. Studies by Worthington et al. and Terasaki et al. have shown an association of circulating anti-HLA antibodies and chronic graft loss (39,40) and Terasaki and Cai have suggested that donor-specific anti-HLA antibody (DSA) is responsible for most cases of late graft loss (40). Kidney transplant recipients with capillaritis (seen in AbMR, and not scored in the original Banff classification) have worse long-term post-transplant outcome (41) as do recipients who are C4d positive (12,42). A series of studies by the Edmonton group suggested that ABMR was the major cause of graft loss: (1) de novo DSA was associated with microcirculatory changes (in the biopsy) and subsequent graft failure; and (2) a significant subset of ABMB was C4d negative (reviewed in 38). Separately, and similar to the studies on subclinical cell-mediated rejection (36), Loupy et al. reported that subclinical AbMR (in 3-month protocol biopsies) was associated with increased IFTA and lower GFR at 1 year (43). As noted above, this and other types of chronic progressive graft functional deterioration may not have been affected by CNIs, and, in fact, erroneously attributed to CNIs. Hopefully, with better characterization of the many causes of chronic progressive kidney dysfunction, future studies will better delineate the role (or lack of) chronic CNI toxicity.

Commercial Organizations

This manuscript was not prepared in any part by a commercial organization.

This manuscript was not funded in any part by a commercial organization, including educational grants.

Conflict of Interest

The author of this manuscript has the following conflicts of interest to disclose as described by the American Journal of Transplantation:

Research support from Bristol Myers Squibb (BMS), Pfizer Inc. (formerly Wyeth Pharmaceuticals), Genentech Pharma (formerly Roche) and Genzyme Corporation.

The author of this manuscript serves as a consultant for BMS and has received speaking honorariums from Genzyme Corporation and Astellas Pharma US, Inc.