High Sirolimus Levels May Induce Focal Segmental Glomerulosclerosis De Novo

Clinical Data

Since 1997, at least 263 renal transplant recipients have received sirolimus either de novo (n = 76) or after conversion from CNI (cyclosporine or tacrolimus; n = 187) at the Necker and Saint-Louis hospitals in Paris. In this study, we include eight patients who met the following criteria: (1) development of biopsy-proven FSGS lesions during sirolimus treatment; (2) absence of primary FSGS; and (3) for switched patients, (a) proteinuria <0.5 g/24 h before the conversion from CNI to sirolimus, (b) absence of FSGS lesions on renal biopsy performed <3 mo before conversion, and (c) CNI/sirolimus overlap that did not exceed 4 wk.

The clinical characteristics and baseline immunosuppression of the eight patients are detailed in Table 1. Patients 1 to 3 received sirolimus de novo, from the first day of the graft, and patients 4 to 8 received sirolimus after a conversion from CNI. The indication for switching from CNI to sirolimus was the presence of CAN lesions on routine biopsy. We identified a fourth de novo patient with nephrotic-range proteinuria and typical FSGS lesions, but the diagnosis was made retrospectively in a nephrectomy. This case therefore was excluded.

In both centers, dosages of sirolimus were usually adjusted to achieve serum levels of 10 to 15 ng/ml. We calculated the mean residual sirolimus serum level during the months that preceded the biopsy. Proteinuria was measured by 24-h urine collection. Patients neither were hypertensive nor received angiotensin-converting enzyme inhibitors before biopsy, and no episode of acute vascular rejection was recorded. Patients did not exhibit anti-HLA antibodies. Graft biopsies were performed on the day of the transplantation in all de novo patients. They displayed only mild lesions according to the Banff classification: “cv1” lesions in patient 2 and “ah1, cv1” lesions in patient 3.

All of the biopsies analyzed herein were performed while patients were receiving sirolimus. A biopsy that was performed in a tacrolimus-treated patient who developed typical FSGS lesions was used as control for vascular endothelial growth factor (VEGF) expression in glomeruli. This patient received a transplant because of nephroangiosclerosis and never received sirolimus.

Renal Biopsies

The renal biopsies were fixed in alcoholic Bouin’s solution and embedded in paraffin for histology and immunohistochemistry. Multiple sections were stained with Masson’s trichrome, periodic acid-Schiff, or silver staining for histopathology. Using immunohistochemistry, podocytes were characterized using an anti-synaptopodin mAb, clone G1D4 (Biogen and Biotechnik, Heidelberg, Germany) and an anti-CKI p57 C-20 polyclonal antibody and an anti-human VEGF mAb, clone C1 (all from Santa Cruz Biotechnology, Santa Cruz, CA). Cytokeratin was labeled using C2562 mAb cocktail (Sigma Aldrich, Saint Quentin Fallavier, France) directed against nine cytokeratin types. Normal adult podocytes are not labeled by C2562 mAb, because fetal podocytes lose cytokeratin expression from the capillary loop stage onward. PAX2 was characterized by an anti-PAX2 polyclonal antibody (Zymed, San Francisco, CA). Normal adult podocytes do not express PAX2 because it disappears from podocytes at the point where the S-shaped bodies transform into the capillary loop stage. Immunohistochemistry procedures were performed as described previously (7,8). FSGS lesions were analyzed according to the recently proposed pathologic classification (9,10).

Clinical Evolution

The evolution of proteinuria and renal function, assessed by serum creatinine level, is detailed in Figure 1 for patients 1 to 3 (de novo sirolimus) and in Figure 2 for patients 4 to 8 (switched patients), providing evidence for a severe and progressive increase in urinary protein excretion rate. In the de novo group, all patients exhibited nephrotic-range proteinuria (maximum from 6.8 to 13.24 g/24 h). In the switched group, mean proteinuria increased from 0.28 ± 0.14 g/24 h before conversion to 2.54 ± 0.99 g/24 h after (n = 5; P = 0.0068, bilateral paired t test). We observed that mean sirolimus trough levels were markedly superior to those currently recommended (Table 1). All de novo patients and two of the switched group were changed from sirolimus to tacrolimus in view of increased proteinuria and FSGS lesions on biopsies, leading to a decrease in proteinuria. No patient was switched back to sirolimus.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Proteinuria and renal function in de novo patients. Urinary protein excretion (black lines), serum creatinine level (gray lines), and sirolimus trough levels (bars) are shown for patients 1 to 3, who received sirolimus de novo. Time after renal transplantation is expressed in months (mo). Dotted lines correspond to the time after sirolimus withdrawal, and time of conversion to tacrolimus is indicated by an arrow.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Proteinuria and renal function in switched patients. Mean urinary protein excretion (black line), mean serum creatinine level (gray line), and mean sirolimus trough level (bar) are shown for patients 4 to 8 before and 3 to 6 mo after a switch from CNI to sirolimus. Error bars correspond to SEM.

Renal Biopsies

The indications for biopsy and the histologic data are detailed in Tables 2 and 3. FSGS lesions involved only a few glomeruli (Table 2). They belonged to the classic type (not otherwise specified) variant in all patients but seemed really recent in de novo patients because scar lesions were uncommon, unlike in switched patients, whose lesions were sclerotic (Figure 3). Immunohistochemistry study was incomplete because in some biopsies, previous histologic analysis had exhausted the blocks, especially in patients 3 and 6. Furthermore, FSGS was not observed in all of the sections performed for immunohistochemistry, because of the focal and segmental features of the disease.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Histopathology. (A through C) Classical FSGS (or variant not otherwise specified) lesions in renal biopsies from patients in the de novo group. Note the podocyte hyperplasia (arrows) and detached floating podocytes (arrowhead). (D) Classical FSGS lesion in renal biopsy from a patient in the switched group showing scarring fibrosis with a clear halo (arrow). Masson’s trichrome stain in A, B, and D; periodic acid-Schiff stain in C. Magnifications: ×700 in A through C; ×400 in D.

In FSGS lesions, podocytes had lost synaptopodin and acquired new epitopes, cytokeratin and PAX2 (Figure 4). To a lesser extent, a decrease of p57 expression was observed in podocytes in some glomeruli. Staining for VEGF was lost in some FSGS lesions. VEGF expression was also studied in a patient who developed CNI-induced FSGS lesions. Although a strong VEGF signal was observed in normal glomeruli, unfortunately no FSGS lesions were present on immunohistochemistry sections (Figure 4).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Immunohistochemistry on renal biopsies from the de novo group of patients. (A) Abnormal cytokeratin expression in hyperplastic visceral podocytes (arrows) in an FSGS lesion. (B) Abnormal PAX2 expression in hyperplastic visceral podocytes (arrows) in an FSGS lesion. (C) Focal loss of synaptopodin expression in an FSGS lesion (arrow). (D) Focal loss of VEGF expression in hyperplastic visceral podocytes in an FSGS lesion (arrows). (E) In an adjacent normal glomerulus on the same biopsy, the VEGF expression is normal. (F) Normal VEGF expression in a normal glomerulus on the biopsy from the tacrolimus-treated patient. (G) In a rejecting graft without FSGS, visceral podocytes do not express cytokeratin. (H) In a rejecting graft without FSGS, visceral podocytes do not express PAX2. Magnifications: ×600 in A, B, and D; ×400 in C, E, F, G, and H.

Sirolimus-Induced Proteinuria and Glomerular Lesions

CNI withdrawal per se may unmask underlying glomerular damage that results in proteinuria. Ducloux et al. (17) reported that in 31 renal transplant recipients whose cyclosporine (CsA) was withdrawn because of cyclosporine toxicity or CAN, baseline proteinuria was 0.79 ± 0.6 g/d and increased to 1.79 ± 0.8 g/d 24 mo after CsA withdrawal. Because CNI reduce renal blood flow, CNI withdrawal may reveal existing CAN lesions that result in proteinuria, and an increase in intraglomerular pressure has been described after the switch (3,5,18).

Nevertheless, several lines of evidence argue for a specific role of sirolimus in the development of glomerular lesions. First, we previously observed a dramatic increase in urinary protein excretion and the existence of nephrotic-range proteinuria after the conversion from CNI to sirolimus in individual patients who had no significant proteinuria before (5). Conversion in stable renal transplant recipients (without evidence for renal lesions) is associated with the development of proteinuria in 30% of the patients (6).

Second, the kinetics of the increase in proteinuria raise questions: the incidence and the severity of proteinuria increase gradually from month 3 to months 12 to 24 after conversion, suggesting that a process is at work and that the acute increase in intraglomerular pressure at the time of the conversion does not totally explain the proteinuria (5). In a similar way, a progressive increase of proteinuria was observed in de novo patients in this study.

Third, cases of proteinuria and microalbuminuria in patients who were treated with sirolimus after clinical islet transplantation have been reported (19). Proteinuria and microalbuminuria resolved after both sirolimus discontinuation and simultaneous increase in tacrolimus dose, suggesting an effect of sirolimus itself on native kidneys in patients with diabetic. In a similar way, a recent report demonstrated that conversion from azathioprine to sirolimus in renal transplant patients may cause an increase of proteinuria that cannot be ascribed to hemodynamic renal effects of CNI withdrawal (20).

It has been recognized since 2003 that some patients develop FSGS after the switch (1). However, these data did not prove that sirolimus was involved, because all patients received CNI previously. It has been observed that 10% of patients who are on a CNI-based immunosuppressive regimen may develop progressive FSGS lesions and that CsA induces glomerular injury in cardiac transplant recipients (21,22). Nevertheless, “symptomatic” de novo FSGS lesions in renal transplantation have been essentially described in a context of de novo collapsing glomerulopathy (23). This rare case is characterized by nephrotic-range proteinuria and podocyte injury, but rapid decline in renal function and capillary collapse distinguish it from our cases. The development of FSGS in renal transplant recipients who never received CNI and did not have a medical history of FSGS lesions in native kidneys strongly support the hypothesis that sirolimus exerts a direct effect on podocytes. The exact prevalence of sirolimus-induced FSGS lesions is difficult to assess because borderline cases may be underscored, but an estimation of 3 or 4% at least seems likely in our units.

Evidence for FSGS and Podocyte Phenotypic Changes

Histologic analyses showed that the first detectable lesions are the cellular type of FSGS and that scar formation and tuft contraction follow (24). It was previously demonstrated that podocyte phenotypic changes accompany or follow the podocyte injury that is observed in recurrent FSGS in renal transplant recipients (7). The origin of these cells remains controversial, however (25,26). Similar to the initial lesions that are observed in recurrence of primary FSGS, here we observed few scar lesions and focal changes in podocyte phenotype, suggesting podocyte dedifferentiation because synaptopodin expression was lost and a fetal phenotype pattern developed, with podocyte expression of cytokeratin and PAX2 in some FSGS cellular lesions. We previously observed exactly the same pattern in recurrent FSGS and showed that adult podocytes in control biopsies do not express PAX2 and cytokeratin but are labeled by synaptopodin mAb (7). The small sample size and the lack of sections with FSGS lesions are a limitation of the study, but, in our experience, this pattern is specific of phenotype dysregulation.

Potential Mechanisms of Sirolimus-Induced Toxicity

A recent article noted no podocytes foot process effacement and the absence of albumin tubular reabsorption in a case of sirolimus-associated proteinuria, suggesting that tubular injury explains proteinuria (27). Nevertheless, the present study provides evidence for podocyte injury, and the existence of a nephrotic-range proteinuria suggests that a glomerular mechanism is involved. Further studies may assess the respective participation of tubular and glomerular mechanism in the development of proteinuria.

Sirolimus is known to decrease VEGF synthesis, making it an interesting drug for transplant patients who develop Kaposi or malignant diseases (28). Moreover, in a recent case report of sirolimus-induced thrombotic microangiopathy, a decreased expression of VEGF in glomeruli seemed to be the consequence of sirolimus treatment (29). Experimental data suggest that podocytes have a functional autocrine VEGF-A system that promotes podocyte survival and differentiation (30). It is interesting that we found a focal decrease of VEGF expression in podocytes in FSGS lesions. By comparison, we observed an intense expression of VEGF in podocytes of a tacrolimus-treated patient who developed FSGS lesions. Unfortunately, no typical lesions of FSGS but only abnormally enlarged podocytes were seen on VEGF-stained sections in this patient. Therefore, it cannot be determined whether sirolimus causes VEGF decrease or this decrease is a consequence of podocyte dedifferentiation. Our data contrast with those of a recent report that described a collapsing glomerulopathy associated with an increased expression of VEGF in glomeruli in a renal transplant recipient who converted to sirolimus because of an human herpesvirus 8–related Kaposi (31). These lesions were attributed to sirolimus. Collapsing glomerulopathy is the classic lesion seen in HIV-associated nephropathy and is experimentally induced when podocyte-specific overexpression of the VEGF164 isoform is promoted (32). Because we observed a clearly different variant of FSGS lesions and because sirolimus does not increase VEGF synthesis but rather decreases it, the responsibility of sirolimus seems doubtful in this observation.

Beyond its role in VEGF synthesis, sirolimus blocks mammalian target of rapamycin (mTOR) activity, which itself regulates Akt, an essential pathway for cell survival, differentiation, and adhesion. It was recently shown that prolonged sirolimus treatment reduces mTORC2, a multiprotein complex that contains mTOR, to levels below those needed to maintain Akt signaling (33). Moreover, recent reports emphasized the role of integrin-linked kinase, an activator of the Akt pathway, in podocyte survival and cytoskeleton maintenance (34).

Because residual concentrations of sirolimus were higher than those currently recommended, sirolimus may have exerted direct toxicity on podocytes. Nevertheless, many patients were exposed to similar concentrations of sirolimus until 2003 and did not develop nephrotic-range proteinuria or FSGS lesions. In the same way, we did not find any correlation between sirolimus levels and urinary protein excretion level (personal data).

It seems likely that FSGS lesions occur only in particularly susceptible patients, but we did not identify any demographic factor that could explain this point. Similarly, sirolimus enhances ischemia-reperfusion injury in renal tubular cells but does not alter renal function in stable renal transplant recipients (35). The underlying mechanisms of this individual susceptibility are unknown. Renal biopsies that were performed on the day of transplantation in patients 1 to 3 did not show significant lesions. Donors were young. Patients did not receive CNI or have hypertension. There was no acute vascular rejection, but the rapid development of vascular lesions and especially arteriolar hyalinosis in the de novo group is noteworthy. Similar processes might underlie glomerular and vascular lesions.