Journal Logo

Original Clinical Science—General

Calcineurin Inhibitor Nephrotoxicity Through the Lens of Longitudinal Histology

Comparison of Cyclosporine and Tacrolimus Eras

Nankivell, Brian J. MD, MSc, PhD, MRCP (UK), FRACP; P'Ng, Chow H. FRCPA; O'Connell, Philip J. FRACP; Chapman, Jeremy R. FRACP

Author Information
doi: 10.1097/TP.0000000000001243

The introduction of cyclosporine (CSA) revolutionized solid organ transplantation, reducing rejection rates and improving early graft survival. Although the calcineurin inhibitors (CNIs), which include tacrolimus (TAC), now form the backbone of current immunosuppression, their long-term nephrotoxicity has long been a clinical concern.1-5 Declining renal function from CSA nephrotoxicity contributes to end-stage renal failure in 6.9% to 28.3% of recipients of a nonrenal organ transplant after 5 years.6 Our initial study identified frequently occurring and progressive CSA nephrotoxicity 10 years after kidney transplantation and questioned the suitability of CNI as maintenance agents.7 Since then, transplant practice has evolved: transitioning from high-dose CSA therapy to lower-dose TAC, and supplemented by use of mycophenolate, robust induction regimes, antiviral prophylaxis, and better technologies to detect donor-specific antibodies (DSA). These have reduced acute rejection rates and early immune-mediated allograft injury. Yet, chronic functional deterioration and late attrition rates have not greatly changed.8 The question posed is whether these high rates nephrotoxicity previously observed with CSA, still threaten the transplanted kidney in the modern TAC era.

The aims of this longitudinal histological cohort study were to: (i) evaluate and compare the incidence of CNI nephrotoxicity in the TAC and CSA eras; (ii) contrast the incidence and prevalence of individual histological markers of nephrotoxicity, including de novo arteriolar hyalinosis, striped cortical fibrosis, tubular microcalcification, peritubular and glomerular capillary congestion, and diffuse interstitial fibrosis with both CNI; and (iii) determine the pathological sequelae of arteriolar hyalinosis with respect to subsequent glomerulosclerosis using serial surveillance histopathology and morphometric analysis of vascular lumen to estimate blood flow.

We evaluated a well-characterized cohort of combined renal-pancreas transplant recipients for evidence of nephrotoxicity using a programme of regular, prospective surveillance kidney biopsies taken up to 10 years after transplantation and beyond. Routine initial immunosuppression at transplantation transitioned from CSA to TAC in 1999. We excluded diabetic patients from pancreas graft failure and those with donor hyalinosis: selecting a study cohort with pristine implanted kidneys to observe the gradual development of CNI nephrotoxicity over a decade.

MATERIALS AND METHODS

Study Population

The screened population comprised consecutive type I diabetic recipients of a kidney-pancreas transplanted at the Australian National Pancreas Transplant Unit at Westmead from 1988 until 2012, where a minimum of 3 years histological follow-up was thus available, and where donor arteriolar hyalinosis and recurrent diabetes mellitus had been excluded. The institutional ethics approval was HREC LNR/12/WMEAD/114.

Calcineurin-Based Immunosuppression Doses and Levels

Triple therapy immunosuppression comprised oil-based CSA (Sandimmune; Novartis, Basel, Switzerland) initiated at 12.5 mg/kg per day until 1996, when replaced by microemulsion CSA (Neoral; Novartis) starting at 8 mg/kg per day, and adjusted to trough levels of 150 to 450 ng/mL. Tacrolimus was used from 1999 (Prograf; Fujisawa, Osaka, Japan) initiated at 0.15 mg/kg per day with target trough levels of 5 to 15 ng/mL. Calcineurin inhibitor doses were delivered twice daily and slowly reduced over time, adjusted to clinical events, acute rejection episodes, functional nephrotoxicity, and side effects. Calcineurin-sparing agents, such as diltiazem, were avoided. Mycophenolate mofetil (MMF) (Cellcept; Roche, Nutley, NJ) replaced azathioprine (AZA, at 1.5 mg/kg per day) as the antiproliferative agent in 1996. Prednisolone was routinely used in all, but antilymphocyte induction was reserved for highly sensitized patients.

Predose 12-hour CSA trough blood concentrations were measured by whole-blood radioimmunoassay (Cyclo-Trac SP; DiaSorin, Stillwater, MN). Tacrolimus trough levels were measured by a whole-blood chemiluminescent immunoassay (Architect c4000; Abbott Laboratories, Abbott Park, IL).

Histological Assessment of CNI Nephrotoxicity

Sequential histopathology from needle-core protocol biopsy samples was evaluated by a specialist nephropathologist and classified by the Banff schema.9 Blinded duplicate scores were used for biopsies until 2002, whereas limited clinical data were provided to subsequent interpreting pathologists. Protocol biopsies were taken at defined intervals including implantation, 1 week (in the early era); 1, 3, 6, and 12 months and then annually or second/third yearly for 10 years and beyond, as previously described.10 The first occurrence of a histological lesion on serial pathology was used for actuarial analysis.

The primary study endpoint was structural CNI nephrotoxicity, defined a priori by appearance of de novo arteriolar hyalinosis. Nodularity of hyalinosis was not used to define nephrotoxicity because vascular wall cross-section varies by section plane, the lack of a rigorous definition of nodularity, our prior findings that early circumferential hyalinosis can evolve into a nodular pattern, and uncertainty over its specificity.11 The same definition of nephrotoxicity was used for both TAC and CSA groups.5 Other features of CNI nephrotoxicity including striped cortical fibrosis, tubular microcalcification, and diffuse interstitial fibrosis were recorded. Juxtaglomerular hyperplasia and isometric tubular vacuolisation were not routinely scored. Clinically significant nephrotoxicity was defined as CNI-mediated dysfunction confirmed by histopathology where the clinician had altered the CNI. C4d was assessed from paraffin-embedded samples using rabbit polyclonal anti-C4d antibody (Cell Marque, Rocklin, CA).12

To estimate the effect of hyalinosis on glomerular blood flow, we measured the long and short axes of the outer circumference and inner lumen dimensions of arterioles from the PAS sections containing a spectrum of hyalinosis severity. Luminal and vascular wall area was calculated using the ellipse formula (π × ab where a and b are the major and minor axes, respectively), and estimated reduction in blood flow was calculated using Poiseuille's equation (volumetric flow rate = π × vessel radius4 × pressure difference / dynamic fluid viscosity × vessel length × 8) and compared with vessels without hyalinosis.

Data Analysis

All biopsies were scored and included where possible. Missing data were omitted. Cox regression was used for survival analysis, preceded by backward elimination. Linear or binomial generalized estimating equations were used for repeated measurements, as appropriate. Potential risk factors tested in models are listed (Table S1, SDC, http://links.lww.com/TP/B277). Analysis was by intention-to-treat as classified by initial CNI therapy. Data are expressed as mean ± SD unless stated. All tests were 2-tailed, and a probability below 0.05 was considered significant.

RESULTS

Study Group Selection and Excluded Patients

From 342 consecutive type 1 diabetic recipients of combined kidney-pancreas transplant screened for study suitability, 142 recipients were excluded because of preexisting donor arteriolar hyalinosis (n = 27, 7.9%). Donor arteriolar hyalinosis correlated with donor age (r = 0.28, P < 0.001) in screened patients. Other reasons for exclusion were: posttransplant diabetes mellitus (n = 47, 13.7%) from early pancreas thrombosis or pancreas rejection or superimposed type 2 diabetes mellitus; insufficient protocol biopsy follow-up (97, 28.4%); or combinations of these factors (n = 29). The final study group comprised 200 nondiabetic study recipients with pristine donor kidneys at transplantation and complete longitudinal histology (detailed Table S2, SDC,http://links.lww.com/TP/B277).

Demographic and Clinical Results

Study recipients were 53.3% men and 38.0 ± 6.8 years old, with diabetes for 25.3 ± 7.3 years before transplantation. Sustained normoglycemia was achieved after transplantation with a mean HBA1C of 5.6 ± 0.9% (2104 samples, 6 months to 25 years). Donors were 26.8 ± 9.9 years old, and 62.0% were men. The renal cold ischemic time was 11.6 ± 3.0 hours, initial hemodialysis was required in 6 patients (3%), the HLA mismatch score was 4.6 ± 1.2 (of 6), and hypertension occurred in 44.7%. Treatment crossovers from CSA-to-TAC comprised 36 of 463 biopsies (7.8% in the CSA arm); and for TAC-to-CSA were 34 of 855 (3.9%), mostly occurring late after transplantation. Seven patients underwent cessation of CNI at some point during follow-up (3.5%).

CNI Doses and Levels

The mean dose of CSA over the study period was 4.6 ± 1.7 mg/kg per day (309 ± 118 mg/day) and TAC was 0.11 ± 0.069 mg/kg per day (6.8 ± 3.6 mg/day), yielding trough levels of 191 ± 111 ng/mL and 9.4 ± 5.7 ng/mL, respectively. Calcineurin inhibitor doses were slowly reduced after transplantation in all (Figure 1, Table S3a and S3b, SDC,http://links.lww.com/TP/B277), and substantial dose reductions undertaken in 61%.

FIGURE 1
FIGURE 1:
A, Cyclosporine daily dose (bars) and trough levels (line) according to time after transplantation. B, Tacrolimus dose and trough levels. C, Extent of interstitial mononuclear inflammation, chronic interstitial fibrosis (D), arteriolar hyalinosis (E), and glomerulosclerosis (F) after transplantation according to calcineurin inhibitor, scored by the Banff schema. Mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001.

Calcineurin inhibitor nephrotoxicity was diagnosed by histopathology in 60.8% of 200 patients. Clinically significant nephrotoxicity occurred in 77 of 200 (38.5%), diagnosed by histopathology alone (42.9%), renal dysfunction (15.6%), or both abnormalities (41.6%). Resultant medication changes in 60 patients included CNI dose minimization in 55 (91.7%) of 60 patients, CNI cessation in 3 (5%) patients, and/or mammalian target of rapamycin substitution in 4 (6.7%) patients. Medications were unchanged in 17 patients after diagnosis because of coexistent rejection or previously completed minimization. To facilitate safe CNI minimisation in CSA-treated patients, 35.9% were eventually converted to low-dose TAC, and 51.3% were converted from AZA to MMF. Renal function remained stable with serum creatinine levels of 133 ± 37 μmol/L before, and 135 ± 56 μmol/L after intervention. Acute clinical rejection or clinically relevant DSA formation (MFI > 3000) consequent to medication adjustment driven by toxicity occurred in 5% and 3.4%.

Renal Transplant Biopsies

From 1669 available biopsies, 47 samples were excluded because of inadequate tissue, leaving 1622 evaluable samples, treated with CSA (n = 712, 43.8%) or TAC (n = 910, 56.1%). Suboptimal samples with fewer than 7 glomeruli or no artery9 that contained arterioles were scored and comprised 22 (1.3%) of 1622 samples and 4 (0.2%) of 1622 samples of evaluable samples in CSA and TAC groups, respectively. Each patient underwent 8.1 ± 4.1 biopsies (range, 3-24) to 7.4 ± 4.4 years posttransplant (median, 5; interquartile range [IQR], 5-10; range, 3-20). Samples contained 12.2 ± 8.3 glomeruli and 2.0 ± 1.1 arteries. The intraobserver and interobserver κ values for arteriolar hyalinosis were 0.82 and 0.46, respectively.

Time-zero biopsies (taken before implantation) or within the first week did not contain arteriolar hyalinosis (by exclusion criteria). Chronic fibrosis and tubular atrophy scores were 0.02 ± 0.08 and 0.03 ± 0.17, respectively, and 1.2 ± 3% of glomeruli were sclerosed, reflecting the absence of preexisting chronic damage of implanted kidneys.

CNI Nephrotoxicity Markers: CSA Versus TAC

The prevalence of CNI-associated lesions increased after transplantation. For CSA-treated patients, the 1-, 5-, and 10-year cumulative Kaplan-Meier prevalence for mild de novo arteriolar hyalinosis (Banff ah grade 1) was 45.6%, 86.0%, and 95.6%, respectively; for moderate arteriolar hyalinosis (Banff ah grade 2 or greater), 5.4%, 38.4%, and 79.1%; for striped fibrosis, 35.1%, 71.0%, and 71.0%; and for tubular microcalcification, 43.9%, 65.3% and 73.5%, respectively (Figure 2). For TAC-treated patients, the corresponding time-dependent prevalence of mild arteriolar hyalinosis was 29.9%, 77.9%, and 93.7% (Peto = 10.54, P < 0.001 vs CSA); for moderate arteriolar hyalinosis was 4.3%, 33.6%, and 77.2% (Peto = 0.892, P = 0.35); for striped fibrosis was 5.7%, 15.1%, and 17.9% (Peto = 63.87, P < 0.001); and tubular microcalcification was 8.6%, 18.7%, and 28.8% (Peto = 41.3, P < 0.001), respectively (Figure 2). Glomerular congestion was observed in 165 (23.6%) of 712 CSA samples compared with 91 (10.1%) of 910 for TAC (P < 0.001). In biopsies taken from 3 months and 10 years posttransplant, arteriolar hyalinosis occurred in 56.3% of CSA compared with 39.6% for TAC (χ2 = 30.1, P < 0.001), 20.2% and 3.6% for striped fibrosis (χ2 = 76.7, P < 0.001), and 24.0% and 7.3% for microcalcification (χ2 = 59.9, P < 0.001), respectively.

FIGURE 2
FIGURE 2:
Kaplan-Meier plots of histological lesions of calcineurin inhibitor nephrotoxicity according to calcineurin inhibitor use. A, Early mild de novo arteriolar hyalinosis (Banff ah score 1). B, Moderate severity arteriolar hyalinosis (Banff ah score 2). C, Striped interstitial fibrosis. D, Tubular microcalcification.

Mild arteriolar hyalinosis occurred earlier in CSA patients (median, 2 years; IQR, 0.25-3 years; 95% confidence interval [95%CI], 0.5-2) compared wiht 3 years for TAC (IQR, 1-5 years; 95%CI, 3-4; P < 0.001). The onset of moderate arteriolar hyalinosis was 7 years (IQR, 3.5-10 years; 95%CI, 5-9) for CSA and 10 years for TAC (IQR, 5-10 years; 95% CI, 6-10; P < 0.001). The median onset of striped fibrosis was 2 years (IQR, 0.5-15 years) and 3 years for tubular calcification (IQR, 0.5-1 year) for the CSA group, although corresponding results for TAC were not calculated because 86.4% and 80.6% of the results were censored because of insufficient events. High instantaneous hazard rates occurred for mild arteriolar hyalinosis, striped fibrosis, and tubular calcification occurred within the first year in the CSA group. In contrast, the instantaneous hazard for moderate arteriolar hyalinosis remained relatively constant over the study duration and was indistinguishable between CNI therapies (Figure S1, SDC,http://links.lww.com/TP/B277).

Using Cox regression, mild arteriolar hyalinosis was greater with CSA use (hazard ratio [HR], 1.70 vs TAC; 95%CI, 1.21-2.39; P = 0.002), when adjusted for donor age (Figure 2, Table S4, SDC,http://links.lww.com/TP/B277); as was the adjusted risk of striped fibrosis (HR, 9.35 vs TAC; 95%CI, 4.93-17.54; P < 0.001). Tubular calcification increased with CSA use (HR, 3.78; 95%CI, 2.19-6.51; P < 0.001 vs TAC) and acute rejection (HR, 1.76; 95% CI, 1.03-3.03; P = 0.041).

Correlations of arteriolar hyalinosis with glomerulitis (r = 0.035, P = 0.165), peritubular capillaritis (r = 0.035, P = 0.165), and C4d scores (r = 0.036, P = 0.327) were poor. The lack of association of moderate hyalinosis with pretransplant DSA (HR, 1.221; 95% CI, 0.778-1.917; P = 0.385) or de novo DSA (HR, 1.031; 95% CI, 0.555-1.915; P = 0.924) by Cox regression refutes a humoral etiology. Class I HLA mismatch showed a modest relationship, which was lost by multivariate analysis (adjusted HR, 1.296; 95% CI, 1.004-1.673; P = 0.924).

CNI Features of Nephrotoxicity and Drug Exposure

A major analytical challenge was correlating transplant histology with the CNI dose and contemporaneous blood concentrations: complicated by clinician-instigated dose reductions after fixed structural nephrotoxicity was detected, which then obscured causal relationships. Early CSA dosage used from 1 to 3 months predicted striped fibrosis (HR, 1.27 per mg/kg; 95% CI, 1.041-1.524; P = 0.017) when MMF (vs AZA) was included in the model (HR, 0.120; 95%CI, 0.036-0.401; P < 0.001; Table S7, SDC, http://links.lww.com/TP/B277). Tubular calcification was predicted by averaged CSA dose using Cox regression (HR, 1.42 per mg/kg; 95%CI, 1.02-1.96; P < 0.05). Mean trough TAC levels, averaged from 1 month to 10 years, correlated with increased striped fibrosis (HR, 1.023 per ng/mL; 95% CI, 1.004-1.043; P = 0.021) and tubular calcification (HR, 1.023; 95%CI, 1.005-1.042; P = 0.011), by univariate Cox regression. Other histological markers failed to demonstrate significant correlations with CNI doses (Tables S5 to S8, SDC,http://links.lww.com/TP/B277). Glomerular congestion was a subjective attribute systemically recorded (interobserver κ, 0.36). It was consistently more prevalent with CSA use (P < 0.001 vs TAC), correlated with TAC dose (P < 0.001) and CSA levels (P = 0.003), and gradually declined with time (Figure 3D). The limited or absent relationship between prescribed CNI dose or blood concentrations against histological markers13 can be attributed to suboptimal correlation between measured CNI trough level and pharmacokinetic exposure or local tissue concentration,14 by temporal discrepancies between early exposure and later appearance of fixed hyalinosis, followed by dose reductions.

FIGURE 3
FIGURE 3:
A, The relationship of arteriolar hyalinosis with glomerulosclerosis on the same tissue sample (r = 0.44,P < 0.001), and with the subsequent sample of 1356 sequential biopsy pairs (B). C, The relationship of arteriolar hyalinosis and chronic interstitial fibrosis by calcineurin inhibitor used. D, Point prevalences of glomerular congestion according to calcineurin inhibitor. E, Morphometric microvascular changes with advancing hyalinosis, with very modest reductions of the vascular circumference, wall thickness and wall cross-sectional area. Measures are the outer arteriolar circumference, vascular wall area and wall thickness. F, Decreased internal luminal area as arteriolar hyalinosis becomes more advanced with calculated reduction in arteriolar blood flow (Q) compared to normal. Mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001.

Arteriolar hyalinosis score was associated with tubular atrophy (Banff ct score, P < 0.001) and arterial thickening (Banff cv score, P < 0.001), when adjusted for time after transplantation (Table S9a, SDC,http://links.lww.com/TP/B277). Moderate arteriolar hyalinosis (Banff ah score 2) was predicted by the tubular atrophy score (odds ratio, 1.778; 95% CI, 1.348-2.346; P < 0.001) and arterial thickening (Banff cv score: odds ratio, 1.485; 95% CI, 1.004-2.198;, P < 0.001), independent of T-cell rejection (Table S9b, SDC,http://links.lww.com/TP/B277). Arteriolar hyalinosis did not correlate with kidney anastomosis time (r = 0.00, P = NS), or total renal ischemic times (r = 0.034, P = NS) in early biopsy specimens to 1 year, nor was it associated with oral glucose tolerance tests (P = NS), or HBA1C levels (5.7 ± 0.13% and 5.5 ± 0.07%, for biopsies with and without hyalinosis, respectively, P = NS).

Arteriolar Hyalinosis and Progressive Glomerulosclerosis

Arteriolar hyalinosis and glomerulosclerosis both increased after transplantation (Figures 1E and F) and were correlated with each other (r = 0.46, P < 0.001) in CSA and in TAC patients (r = 0.42, P < 0.001). On contemporaneous samples, glomerulosclerosis was greater for CSA for hyalinosis categories (Figure 3A, P < 0.001 for group), likely from formation of atubular glomeruli from immune-mediated tubular injury.15 Biopsy pairs were formed to investigate causal sequences of histological abnormalities from sequential kidney biopsies obtained from the same patient (n = 1356 pairs), and the increase in glomerulosclerosis was compared with severity of arteriolar hyalinosis from the preceding biopsy. Without arteriolar hyalinosis, glomerulosclerosis increased by 1.34 ± 13.6% to 1.94 ± 9.93% for CSA and TAC, respectively, to 4.27 ± 17.9% and 5.29 ± 16.2% after mild arteriolar hyalinosis, and 6.82 ± 27.5% and 8.53 ± 23.1% after moderate-to-severe arteriolar hyalinosis (both P < 0.001 vs nil hyalinosis), without difference between CNI used.

Arteriolar morphometry demonstrated transition from normal vessels to hyalinosis resulting in slight shrinkage of the outer circumference, an unchanged vascular wall thickness and area, but a marked reduction in the vascular lumenal cross-sectional area (Figure 3). The resultant flow (compared with normal) was calculated at 39 ± 59% and 20 ± 34% for moderate and severe hyalinosis, respectively (P < 0.05 and <0.01, Table S13, SDC,http://links.lww.com/TP/B277), due to a reduction in flow of 61 ± 59% and 80 ± 34%, compared with normal vessels (Figure 3).

Chronic interstitial fibrosis increased not only with time (Figure 1D) and correlated with acute interstitial infiltrate score (coefficient ± SE, 0.120 ± 0.031; P < 0.001), which fell (Figure 1C), but also with arteriolar hyalinosis score (0.201 ± 0.034, P < 0.001) and was greater for CSA therapy (vs. TAC, 0.349 ± 0.048, P < 0.001), using linear generalized estimating equation from 1622 evaluable biopsies in 200 patients (Table S10, SDC,http://links.lww.com/TP/B277).

Isotopic 99mTc DTPA glomerular filtration rate (GFR) correlated weakly with Banff ah score (r = −0.18, P < 0.001 using all time points). Measured GFR was independently reduced by inflow vascular disease (Banff cv score, coefficient −4.409 ± 1.260, P < 0.001) and arteriolar hyalinosis (−4.271 ± 1.189, P < 0.001), percentage glomerulosclerosis (−1.496 ± 0.705, P = 0.034), and functioning tubules to process the glomerular ultrafiltrate (Banff ct score, −0.117 ± 0.041, P = 0.004, Table S11, SDC,http://links.lww.com/TP/B277). Tacrolimus era patients displayed greater GFR compared with CSA (P < 0.001, Table 1).

TABLE 1
TABLE 1:
Demographic characteristics and key clinical outcomes of study patients, classified by calcineurin inhibitor at transplantation

DISCUSSION

This unique, longitudinal cohort study evaluated the histological development of CNI nephrotoxicity in the modern TAC era compared with the older CSA era, and found both similarities and differences (Table 2). Although the incidence of acute mild arteriolopathy, striped interstitial fibrosis, glomerular congestion, and tubular microcalcification were all ameliorated in TAC biopsies, the inexorable progression of chronic arteriolar hyalinosis was unchanged from the CSA era. Other smaller studies have not shown histological differences between CNI. The US multicenter kidney transplant trial compared TAC with CSA in 114 patients and found new-onset arteriolar hyalinosis in 24.1% and 16.9%, respectively, from 2-year protocol samples.5 A retrospective single-center study of 81 indication biopsies at about 3 months after transplant found arteriolar hyalinosis in 23% and 40% on CSA and TAC and striped fibrosis in 23% and 35%, respectively.16 Our larger study showed an increased prevalence according to time posttransplant, and differences favoring TAC. The same markers of CNI nephrotoxicity were expressed at differing proportions according to CNI era: a quantitative rather than qualitative distinction.

TABLE 2
TABLE 2:
Summary histological features of all evaluable biopsies according to cyclosporine and tacrolimus treatment

Tacrolimus therapy reduced mild arteriolar hyalinosis, striped fibrosis, and tubular microcalcification, possibly attributable to CSA-mediated microvascular toxicity. Early hyalinosis has been correlated with high CSA blood concentrations,7 and is potentially reversible. Acute arteriolopathy confined to afferent vessels is characterized by vacuolation and swelling of endothelial cells and apoptosis or necrosis of individual myocytes of the tunica media.17 In patients with CNI nephrotoxicity, endothelial cell mitochondria are ultrastructurally abnormal and sparse, with reduced protective vascular endothelial growth factor, and increased nitrotyrosine immunohistochemical staining, a peroxynitrite formation marker.18 Cyclosporine produces greater vasoconstriction in both animal19 and human20,21 studies, compared with TAC. Cyclosporine activates the renin-angiotensin system, increases secretion of vasoconstrictors including endothelin and thromboxane, while reducing vasodilatory prostacyclin, prostaglandin E2, and nitric oxide synthase.4,22 Cyclosporine increased rates of acute nephrotoxicity (HR, 1.49; 95%CI, 1.04-2.14) compared with TAC in a large indication biopsy study.23 In our study, the instantaneous hazard rates of nephrotoxicity were greatest early after transplantation, correlating to the period of high CNI exposure. A vulnerable endothelium is more susceptible to additional CSA-induced microvascular toxicity relative to TAC.

Striped fibrosis was commonly reported when high-dose CSA was first introduced to kidney transplantation, but is also observed with TAC.2,3,24,25 Striped fibrosis correlated with early CSA dosage and averaged TAC blood levels in our study. Stripes occur in the thick ascending limb traversing the inner stripe of outer medulla and medullary ray in rodent CSA nephrotoxicity.25 The sharply demarcated area with localized tubular atrophy adjacent to normal cortex7,17 suggests an ischemic etiology. The same pattern also occurs in hyperhomocysteinemic weanling rats with arteriolar/arterial wall thickening,26 after microsphere injection of normal kidneys,27 and is abrogated in experimental CSA nephrotoxicity by trimetazidine, an anti-ischemic agent.28 Hence, microvascular injury causing luminal narrowing may result in downstream localized infarction.

Calcineurin inhibitors are directly cytotoxic to tubular cells at high concentrations.18,29-32 Proximal tubular cells exposed to CSA display isometric vacuolation, abnormal mitochondria, enlarged autolysosomes filled with distorted mitochondrial fragments, and dilated endoplasmic reticulum—leading to cellular death with dystrophic microcalcification.2,3,33,34 Tubular atrophy correlated with arteriolar hyalinosis, and interstitial fibrosis was more pronounced with CSA compared with TAC, even when the effects of immune injury were controlled by multivariate analysis. Hence, arteriolar hyalinosis (as a marker of nephrotoxicity) independently correlated with interstitial fibrosis (IF) and tubular atrophy (TA). Human studies of CNI-related interstitial fibrosis showed upregulated profibrotic genes including TGF-ß, collagen, fibronectin, matrix metalloproteinase-2, tissue inhibitor of metalloproteinase 2, and osteopontin, but not clearly favoring either CNI.31,35,36 In experimental models of ischemic injury37 and in vitro isolated renal cells, CSA produced more profibrotic molecules and IF/TA.29-31 Tubular microcalcification increased with CSA therapy and acute rejection, correlating with CSA dose and averaged TAC levels in respective cohorts.

In contrast to the reduction of early microvascular toxicity with TAC, the incidence of moderate hyalinosis was unchanged compared with CSA, with overlapping Kaplan-Meier and point-prevalence curves. Tacrolimus concentrations averaged 7.9 ± 3.9ng/mL, comparable to the median 7ng/mL achieved by the Symphony study.38 Chronic hyalinosis was not explained by donor disease or recurrent diabetes—both excluded at study entry, nor by multiple other tested risk factors: including demographic factors, ischemia-reperfusion injury, hypertension, immune response, and humoral rejection (by microvascular inflammation, C4d staining and circulating DSA). By exclusion, cumulative nephrotoxicity from prolonged CNI exposure over years-to-decades remains the best explanation in this selected cohort.

Progressive arteriolar hyalinosis correlated with developing glomerular abnormalities.7 Arterioles are defined by 1 or 2 smooth muscle cell layers, and absent or incomplete internal elastic lamina. Injured or apoptotic medial smooth muscle cells are replaced by hyaline deposits, which can produce a “nodular” appearance. Morphometric analysis showed wall thickness actually remained constant as arteriolar hyalinosis developed, leading to vascular narrowing with substantial luminal encroachment. Arteriolopathy features vacuolation, apoptosis or necrosis of tunica media myocytes, and replacement by protein “insudates” without intrinsic mechanical strength.17 Loss of supportive myocytes leads to inward collapse of a weakened arteriolar wall. Although small arteries mediate resistance, arterioles regulate and control flow by vasodilatation and vasoconstriction. Laminar blood flow in small arterioles broadly obeys Poiseuille law, where flow is proportional to the fourth power of the vessel radius. Blood flow was reduced by 80% with severe hyalinosis calculated from the cross-sectional luminal area. Greater severity of hyalinosis increased numbers of sclerosed glomeruli on sequential biopsy pairs irrespective of the CNI used, supporting a hemodynamic causality. Hence, severe narrowing of distal interlobular arteries and afferent arterioles supplying the glomerulus causes ischemic glomerulosclerosis.7,17

The study strengths include its large size, use of histology to define chronic nephrotoxicity and availability of repeated protocol samples across the transplant's lifespan, allowing accurate estimates of incidence and point’prevalence of individual lesions. Functional impairment is insensitive and nonspecific marker of CNI nephrotoxicity, overlooking subclinical disease, unless an indication biopsy was provoked. We selected only histologically normal donor kidneys in nondiabetic recipients for inclusion to eliminate confounding etiological factors for arteriolar hyalinosis. Comprehensive clinical data allowed evaluation of putative risk factors. Intention-to-treat analysis was used, and treatment crossovers between CNI were modest and late, without affecting biopsy results. One limitation was that our study was not randomized, but a historical cohort study and intrinsic bias limit our ability to prove causality. Although patient demographics and implanted kidneys were constant and comparable between groups, other changes alter with era, including reduced immune-mediated nephron injury. However, CNI nephrotoxicity was distinguishable from rejection by histology.2,4 Although the analyses were retrospective, kidney samples and all other data were prospectively collected as per protocol.

Our study compared CSA and TAC, but lacked a CNI-free control arm. A retrospective cohort study of 141 kidney transplant recipients by Snanoudj et al found arteriolar hyalinosis in 92% of 10-year protocol biopsies in CSA-treated patients. The CSA group encountered greater progression of hyalinosis, fibrointimal thickening, IF/TA, glomerulosclerosis, and graft failure. Interestingly, arteriolar hyalinosis (including muscular deposits) also occurred in 65% of azathioprine/prednisolone patients, although 32.2% were already present at 3 months from likely donor origin.39 The suboptimal diagnostic specificity of hyalinosis was highlighted, and unanswered questions of etiology were raised.11 Differential diagnoses of hyaline arteriolopathy include intrinsic vascular aging, diabetes mellitus (present in 30-37% after transplantation),40 and severe hypertension. Similarly, striped fibrosis is neither sensitive nor specific, having been reported in chronic obstruction, reflux nephropathy, and allograft pyelonephritis.41 Tubular microcalcification can follow acute tubular necrosis or severe rejection,7,34 and interstitial calcification may arise from posttransplant hyperparathyroidism.42 Unambiguously, CNI exposure to patients with autoimmune diseases or nonrenal transplants, or rodent models of nephrotoxicity—all produce a stereotypical constellation of renal histological abnormalities. The difficulty is extracting and separating these variably specific markers of nephrotoxicity from the background noise of immune and nonimmune mediated damage. Overall, de novo or progressive arteriolar hyalinosis remains the most reliable diagnostic marker of nephrotoxicity with the above caveats. Progression, verified by serial biopsy or comparison to implantation pathology, is important to distinguish an aged vasculature from extended-criteria donor kidneys.39

The debate over rejection versus nephrotoxicity should not, in our view, be seen as an either/or argument. Neither factor is mutually exclusive: both immune and nonimmune risk factors operating over the lifespan can injure the allograft.10,43 Hence, assigning a single causation of failure is logically unsound. The role of CNI nephrotoxicity can be conceptualized as a background “environmental toxin” insidiously inflicting chronic parenchymal deterioration. The concept of a therapeutic window is illusionary: to be effective immunosuppressants, CNI must operate within their nephrotoxic therapeutic range. The immediate increase in GFR after CNI withdrawal illustrates acute functional nephrotoxicity. The histological evidence indicates that chronic CNI nephrotoxicity remains an important contributor to ongoing graft damage—particularly in the microvascular and glomerular compartments of stable protocol studies of compliant transplant populations, whereas rejection constitutes the most important cause of acute or subacute graft failure, from the subset of patients where an indication biopsy is provoked.44 Isolated CNI nephrotoxicity has a relatively good outcome compared with late rejection associated with noncompliance.45 Therapeutic options for CNI nephrotoxicity include minimization (sometimes aided by use of mycophenolate and/or corticosteroids), CNI elimination, and/or substitution by mammalian target of rapamycin inhibitor.4 However, none of these approaches provide immunosuppressive control that can equal the CNIs. When nephrotoxicity and rejection coexist, immune control takes precedence, and CNI discontinuation is usually unsafe. The treatment choice depends on recipient immune risk (rejection occurs in 5% to 15% of conversions) and side effects.

In conclusion, although recognizing the immune system remains the primary threat to the allograft and requires ongoing suppression, the dream of “one kidney for life”46 will remain largely unrealized with CNI dependent therapy—as nephrotoxicity becomes marked and histologically important a decade after transplantation and beyond, even with low-dose TAC therapy. Our results favor the use of TAC over CSA with less early microvascular nephrotoxicity, comparable chronic toxicity, and less rejection and immune mediated graft damage—producing a wider therapeutic margin.

ACKNOWLEDGMENTS

The authors thank Ms Julie McKelvey for organizational help, our nursing colleagues in Westmead Hospital, our referring medical practitioners, and acknowledge Drs. Caroline L-S. Fung and Richard J. Borrows who participated in early histology scoring.

REFERENCES

1. Bennett WM, DeMattos A, Meyer MM, et al. Chronic cyclosporine nephropathy: the Achilles' heel of immunosuppressive therapy. Kidney Int. 1996;50:1089–1100.
2. Mihatsch MJ, Ryffel B, Gudat F. The differential diagnosis between rejection and cyclosporine toxicity. Kidney Int Suppl. 1995;52:S63–S69.
3. Mihatsch MJ, Thiel G, Ryffel B. Histopathology of cyclosporine nephrotoxicity. Transplant Proc. 1988;20(3 Suppl 3):759–771.
4. Naesens M, Kuypers DR, Sarwal M. Calcineurin inhibitor nephrotoxicity. Clin J Am Soc Nephrol. 2009;4:481–508.
5. Solez K, Vincenti F, Filo RS. Histopathologic findings from 2-year protocol biopsies from a U.S. multicenter kidney transplant trial comparing tarolimus versus cyclosporine: a report of the FK506 Kidney Transplant Study Group. Transplantation. 1998;66:1736–1740.
6. Ojo AO, Held PJ, Port FK, et al. Chronic renal failure after transplantation of a nonrenal organ. N Engl J Med. 2003;349:931–940.
7. Nankivell BJ, Borrows RJ, Fung CL, et al. Calcineurin inhibitor nephrotoxicity: longitudinal assessment by protocol histology. Transplantation. 2004;78:557–565.
8. Lamb KE, Lodhi S, Meier-Kriesche HU. Long-term renal allograft survival in the United States: a critical reappraisal. Am J Transplant. 2011;11:450–462.
9. Racusen LC, Solez K, Colvin RB, et al. The Banff 97 working classification of renal allograft pathology. Kidney Int. 1999;55:713–723.
10. Nankivell BJ, Borrows RJ, Fung CL, et al. The natural history of chronic allograft nephropathy. N Engl J Med. 2003;349:2326–2333.
11. Snanoudj R, Royal V, Elie C, et al. Specificity of histological markers of long-term CNI nephrotoxicity in kidney-transplant recipients under low-dose cyclosporine therapy. Am J Transplant. 2011;11:2635–2646.
12. Haas M, Sis B, Racusen LC, et al. Banff 2013 meeting report: inclusion of c4d-negative antibody-mediated rejection and antibody-associated arterial lesions. Am J Transplant. 2014;14:272–283.
13. Kuypers DR, Naesens M, de Jonge H, et al. Tacrolimus dose requirements and CYP3A5 genotype and the development of calcineurin inhibitor-associated nephrotoxicity in renal allograft recipients. Ther Drug Monit. 2010;32:394–404.
14. Knops N, van den Heuvel LP, Masereeuw R, et al. The functional implications of common genetic variation in CYP3A5 and ABCB1 in human proximal tubule cells. Mol Pharm. 2015;12:758–768.
15. Gibson IW, Downie TT, More IA, et al. Atubular glomeruli and glomerular cysts—a possible pathway for nephron loss in the human kidney? J Pathol. 1996;179:421–426.
16. Randhawa PS, Shapiro R, Jordan ML, et al. The histopathological changes associated with allograft rejection and drug toxicity in renal transplant recipients maintained on FK506. Clinical significance and comparison with cyclosporine. Am J Surg Pathol. 1993;17:60–68.
17. Liptak P, Ivanyi B. Primer: histopathology of calcineurin-inhibitor toxicity in renal allografts. Nat Clin Pract Nephrol. 2006;2:398–404.
18. Sharma A, Jain S, Gupta R, et al. Ultrastructural alterations in endothelial mitochondria are associated with enhanced nitrotyrosine accumulation and progressive reduction of VEGF expression in sequential protocol renal allograft biopsies with calcineurin inhibitor toxicity. Transpl Int. 2010;23:407–416.
19. Bagnis C, Deray G, Dubois M, et al. Comparative acute nephrotoxicity of FK-506 and ciclosporin in an isolated in situ autoperfused rat kidney model. Am J Nephrol. 1997;17:17–24.
20. Klein IH, Abrahams A, van Ede T, et al. Different effects of tacrolimus and cyclosporine on renal hemodynamics and blood pressure in healthy subjects. Transplantation. 2002;73:732–735.
21. Nankivell BJ, Chapman JR, Bonovas G, et al. Oral cyclosporine but not tacrolimus reduces renal transplant blood flow. Transplantation. 2004;77:1457–1459.
22. 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;55:674–685.
23. Jacobson PA, Schladt D, Israni A, et al. Genetic and clinical determinants of early, acute calcineurin inhibitor-related nephrotoxicity: results from a kidney transplant consortium. Transplantation. 2012;93:624–631.
24. Bennett WM. Insights into chronic cyclosporine nephrotoxicity. Int J Clin Pharmacol Ther. 1996;34:515–519.
25. Rosen S, Greenfeld Z, Brezis M. Chronic cyclosporine-induced nephropathy in the rat. A medullary ray and inner stripe injury. Transplantation. 1990;49:445–452.
26. Kumagai H, Katoh S, Hirosawa K, et al. Renal tubulointerstitial injury in weanling rats with hyperhomocysteinemia. Kidney Int. 2002;62:1219–1228.
27. Suzuki T, Kimura M, Asano M, et al. Role of atrophic tubules in development of interstitial fibrosis in microembolism-induced renal failure in rat. Am J Pathol. 2001;158:75–85.
28. Satyanarayana PS, Chopra K. Oxidative stress-mediated renal dysfunction by cyclosporine A in rats: attenuation by trimetazidine. Ren Fail. 2002;24:259–274.
29. Bing P, Maode L, Li F, et al. Comparison of expression of TGF-beta1, its receptors TGFbeta1R-I and TGFbeta1R-II in rat kidneys during chronic nephropathy induced by cyclosporine and tacrolimus. Transplant Proc. 2006;38:2180–2182.
30. Esposito C, Foschi A, Parrilla B, et al. Effect of calcineurin inhibitors on extracellular matrix turnover in isolated human glomeruli. Transplant Proc. 2004;36:695–697.
31. Khanna A. Tacrolimus and cyclosporinein vitro and in vivo induce osteopontin mRNA and protein expression in renal tissues. Nephron Exp Nephrol. 2005;101:e119–e126.
32. Moutabarrik A, Ishibashi M, Fukunaga M, et al. FK506-induced kidney tubular cell injury. Transplantation. 1992;54:1041–1047.
33. Davies DR, Bittmann I, Pardo J. Histopathology of calcineurin inhibitor-induced nephrotoxicity. Transplantation. 2000;69(12 Suppl):SS11–SS13.
34. Kim JY, Suh KS. Light microscopic and electron microscopic features of cyclosporine nephrotoxicity in rats. J Korean Med Sci. 1995;10:352–359.
35. Baboolal K, Jones GA, Janezic A, et al. Molecular and structural consequences of early renal allograft injury. Kidney Int. 2002;61:686–696.
36. Roos-van Groningen MC, Scholten EM, Lelieveld PM, et al. Molecular comparison of calcineurin inhibitor-induced fibrogenic responses in protocol renal transplant biopsies. J Am Soc Nephrol. 2006;17:881–888.
37. Jain S, Bicknell GR, Nicholson ML. Tacrolimus has less fibrogenic potential than cyclosporin A in a model of renal ischaemia-reperfusion injury. Br J Surg. 2000;87:1563–1568.
38. Ekberg H, Tedesco-Silva H, Demirbas A, et al. Reduced exposure to calcineurin inhibitors in renal transplantation. N Engl J Med. 2007;357:2562–2575.
39. Bröcker V, Schubert V, Scheffner I, et al. Arteriolar lesions in renal transplant biopsies: prevalence, progression, and clinical significance. Am J Pathol. 2012;180:1852–1862.
40. First MR, Dhadda S, Croy R, et al. New-onset diabetes after transplantation (NODAT): an evaluation of definitions in clinical trials. Transplantation. 2013;96:58–64.
41. Kobayashi A, Yamamoto I, Ito S, et al. Medullary ray injury in renal allografts. Pathol Int. 2010;60:744–749.
42. Gwinner W, Suppa S, Mengel M, et al. Early calcification of renal allografts detected by protocol biopsies: causes and clinical implications. Am J Transplant. 2005;5:1934–1941.
43. Nankivell BJ, Chapman JR. Chronic allograft nephropathy: current concepts and future directions. Transplantation. 2006;81:643–654.
44. El-Zoghby ZM, Stegall MD, Lager DJ, et al. Identifying specific causes of kidney allograft loss. Am J Transplant. 2009;9:527–535.
45. Gourishankar S, Leduc R, Connett J, et al. Pathological and clinical characterization of the ‘troubled transplant’: data from the DeKAF study. Am J Transplant. 2010;10:324–330.
46. Montgomery RA. One kidney for life. Am J Transplant. 2014;14:1473–1474.

Supplemental Digital Content

Copyright © 2016 Wolters Kluwer Health, Inc. All rights reserved.