The aim of this study was to examine the clinical characteristics, the histological features and the renal expression of vascular endothelial growth factor (VEGF) of five patients with sirolimus-associated thrombotic microangiopathy (TMA). Sirolimus-induced TMA occurs preferentially in kidneys with concomitant endothelial injury: it was observed in three patients with acute cellular rejection on calcineurin inhibitor-free regimen, in one patient with chronic graft rejection on a calcineurin inhibitor-free protocol and in one patient with chronic calcineurin inhibitor nephrotoxicity. We found that renal VEGF expression during sirolimus-induced TMA was significantly lower than VEGF expression in normal transplanted kidneys (p < 0.01). Decreased expression of VEGF seems to be a consequence of sirolimus treatment since (i) analysis of two biopsies performed after the switch of sirolimus to calcineurin inhibitor showed reappearance of VEGF expression, (ii) no decreased expression of VEGF was found in five kidneys with classical TMA and, (iii) an increased expression of VEGF was observed in seven kidneys with acute cellular rejection on a sirolimus-free immunosuppressive regimen (p < 0.01). The potential role of sirolimus-induced downregulation of VEGF as a predisposing factor to the development of TMA is discussed.
The major reasons for the improvement of kidney transplant outcomes over the past decade include the widespread use of more potent and selective immunosuppressive agents. The recent introduction of sirolimus (rapamycin, Rapamune®) has provided an important evolution in transplantation therapeutics, because this immunosuppressive agent lacks the nephrotoxic profile of the calcineurin inhibitors (1,2). However, data from recent clinical trials suggest that sirolimus may be associated with a high rate of thrombotic microangiopathy (TMA) compared with other immunosuppressive agents (3–8). The incidence of de novo TMA in renal transplantation recipients was 4.9 episodes per 1000 person-years in a historical cohort study of 15 870 patients in the United States Renal Data System (USRDS) (9). Risk factors for de novo TMA included initial use of sirolimus (18.1/1000 person-years in patients on sirolimus therapy). The risk of TMA is significantly increased in patients receiving a combination of ciclosporin and sirolimus (4,9,10). Few patients start transplantation on a calcineurin inhibitor-free sirolimus-based protocol because phase II trials have shown that the lowest acute rejection rates can be obtained by combining sirolimus with ciclosporin. However, two preliminary reports suggested that sirolimus in absence of calcineurin inhibitors may induce TMA in renal transplants (11,12).
Sirolimus forms a complex with the FK-binding protein complex (FKBP-12) that binds with high affinity to the mammalian target of rapamycin (mTOR) (13–15). mTOR controls the phosphorylation of proteins that regulate the cell cycle and is also involved in the regulation of growth-factors production including vascular endothelial growth factor (VEGF), which plays a key role in endothelial survival (16). The mTOR inhibitor, rapamycin, inhibits the production of VEGF in different tumour cell lines in vitro and in vivo (17). Moreover, sirolimus induces endothelial cell death and tumoral vessel thrombosis (18). In the kidney, VEGF is mainly produced in the visceral glomerular epithelial cells (podocytes) (19,20). Reduction of VEGF production in podocytes in transgenic mice leads to proteinuria and endotheliosis, the glomerular lesions occurring in pre-eclampsia and in TMA (21).
In this paper we describe four cases of TMA in patients on a calcineurin inhibitor-free sirolimus-based protocol showing that sirolimus alone, independently of concomitant administration of a calcineurin inhibitor, may increase the risk to develop TMA. We also describe one case of TMA in native kidneys in a patient receiving a combination of tacrolimus and sirolimus. Because VEGF production in podocytes is required for glomerular endothelial cell survival and because sirolimus affects VEGF synthesis in different cell models, we analyzed VEGF expression in podocytes of kidneys from these five patients. In all of them, we found that VEGF expression was considerably reduced. Our results suggest that alteration of VEGF podocyte production is one mechanism by which sirolimus may increase the risk of renal TMA.
Materials and Methods
Patients 1, 2 and 3 received the same immunosuppressive protocol treatment. Initial immunosuppression consisted of a 5-day course of antilymphocyte globuline (Thymoglobulin®, Sangstadt), 1 g twice a day of mycophenolate mofetil (Cellcept®, Hoffman-LaRoche), 15 mg at surgery and then 10 mg orally daily of sirolimus (Rapamune®, Wyeth-Ayerst), plus 250 mg of intravenous (i.v.) methylprednisolone at surgery and then oral prednisone. None of the five patients had anti-HLA antibodies, or personal or familial history of TMA. Table 1 summarizes the laboratory parameters of all patients at the time of biopsies.
Table 1. Laboratory finding at the time of biopsies
A 34-year-old woman with end-stage renal disease due to autosomal-dominant polycystic kidney disease received a cadaveric renal transplant (51-year-old, suicide, 3 mismatches). The allograft functioned immediately and on post-operative day 14, the serum creatinine had fallen from 581 to 111 μmol/L. At 2 months, the serum creatinine was 105 μmol/L and proteinuria was 0.7g/day. Some 3 months later, the serum creatinine increased to 130 μmol/L and proteinuria was 1.7 g/day. The platelet count was 114 000/mm3. There was anemia without biological signs of hemolysis. A renal biopsy was performed. The biopsy revealed acute rejection (Grade IIA) and lesions of TMA (Figure 1A). C4d staining was negative and no anti-HLA antibodies were detected. The patient received i.v. methylprednisolone. The sirolimus treatment was pursued. Twenty days after the first biopsy, the serum creatinine was 145 μmol/L and proteinuria was 3 g/day. The platelet count decreased to 80 000/mm3. A second renal biopsy was therefore performed and confirmed ongoing TMA without histological lesions of acute cellular rejection (Figure 1B). The patient received i.v. methylprednisolone and thymoglobulin for 5 days. The sirolimus treatment was discontinued and replaced by tacrolimus. At 1 year, the TMA had not recurred. The serum creatinine was 163 μmol/L and proteinuria was 0.9 g/day. A third biopsy was performed and showed interstitial fibrosis, nephroangiosclerosis, mild mesangial sclerosis and no sign of cellular rejection or TMA.
A 39-year-old woman with end-stage renal disease due to primary focal segmental glomerular sclerosis received a cadaveric renal transplant (56-year-old, vascular stroke, 4 mismatches) in May 2003 after 4 years of dialysis. There was primary allograft function but only a small fall in plasma creatinine. Serum creatinine level reached 219 μmol/L on post-operative day 14. At day 19, the serum creatinine increased to 288 μmol/L and proteinuria was 1.6 g/day. A renal biopsy was performed and showed acute cellular rejection (Grade IIA) and TMA. C4d staining was negative and no anti-HLA antibodies were detected. At that point, platelet count decreased to 118 000/mm3 and there was hemolytic anemia. The patient received i.v. immunoglobulin administered over 2 days (1gIVIG/Kg/day) and i.v. methylprednisolone. The sirolimus treatment was discontinued and replaced by tacrolimus. At day 28, serum creatinine rose to 465 μmol/L and the patient returned to dialysis. A repeat renal biopsy performed at day 38 showed severe TMA with sign of acute cellular rejection (Grade IA) and interstitial hemorrhage (Figure 1C and D). On post-operative day 55, the patient was detransplanted.
A 37-year-old woman with end-stage renal disease due to IgA nephropathy received a cadaveric renal transplant (31-year-old, AVP, 3 mismatches). The allograft functioned immediately and on post-operative day 15, the serum creatinine had fallen from 686 to 140 μmol/L. Proteinuria was 0.8 g/day. At day 19, the serum creatinine increased to 339 μmol/L and proteinuria was more than 3 g/day. A renal biopsy was performed and showed acute cellular rejection (Grade IB) and TMA with intraluminal thrombi within glomeruli and peritubular capillaries (Figure 1E). C4d staining was negative and no anti-HLA antibodies were detected. The platelet count decreased to 29 000/mm3. There was hemolytic anemia. Oliguria appeared and dialysis was started at day 21. Plasmapheresis was started and continued for 12 days. This treatment was associated with i.v. methylprednisolone. The sirolimus treatment was discontinued and replaced by tacrolimus. Dialysis was stopped at day 38. The serum creatinine began to decline and reached 436 μmol/L at day 50. At 1 year, the TMA had not recurred. The serum creatinine was 156 μmol/L and there was no proteinuria.
A 58-year-old man with end-stage renal disease due to chronic glomerulonephritis received a cadaveric renal transplant at the age of 25. Immunosuppressive treatment consisted in azathioprine and steroids. In February 2003, azathioprine was withdrawn because of recurrent squamous cell carcinoma. In November 2003, sirolimus was introduced because of persistent skin cancer. Serum creatinine was 126 μmol/L and there was no proteinuria. In January 2004, serum creatinine rose to 169 μmol/L and proteinuria appeared (1.5 g/day). In June 2004, serum creatinine was 274 μmol/L and proteinuria 2.2 g/day. A renal biopsy was performed and showed chronic rejection (Grade III g0, i0, t0, v0, ci3, ct3, cg2, cv2, mm3) and histological signs of TMA (Figure 1E). Sirolimus was discontinued. Serum creatinine and proteinuria progressively decreased to 200 μmol/L and 0.5 g/day.
A 53-year-old man was referred to our department in November 2003 for renal failure and proteinuria. He received a liver transplant in 1993 for cirrhosis due to primitive sclerosing cholangitis. The initial immunosuppressive protocol included azathioprine, cyclosporine A and steroids. Azathioprine was withdrawn a few months later. In 1995, cyclosporine was switched for tacrolimus. In 2001, MMF 1 g/day was introduced. In June 2003, sirolimus was added to treat chronic rejection. Serum creatinine was at 150 μmol/L. At admission, serum creatinine was 300 μmol/L and proteinuria was 6 g/day. There was no evidence of mechanical hemolytic anemia, although the haptoglobin rate was low (0.64 g/L). A renal biopsy was performed and showed chronic calcineurin inhibitor nephrotoxicity, TMA and interstitial hemorrhages (Figure 1F). Sirolimus was withdrawn and the renal function stabilized.
Renal biopsy specimens of the patients studied were fixed and prepared for light using standard techniques. Frozen biopsies were used for immunofluorescence. C4d staining was performed using a mouse mAb anti-C4d (Quidel, San Diego, CA) and a sensitive, three-step immunofluorescence technique. Staining for C4d was considered positive when peritubular capillaries were diffusely and brightly stained. Fixed biopsies were used for VEGF immunohistochemical staining. An independent set of 22 archived biopsy samples from native kidneys or from transplanted patients on a sirolimus-free calcineurin inhibitor-based protocol were used as positive controls for VEGF immunostaining: 4 biopsy samples were obtained from normal kidneys, 5 from normal renal allografts; 7 from transplanted patients with acute cellular rejection and 5 from native kidneys with TMA. These biopsy samples were selected from the tissue bank of Maison Blanche Hospital (Reims).
Immunohistochemistry and image cytometry
Antigen retrieval was performed by immersing the 4 μm sections in citric acid buffer (pH 6.0) heated in bath water for 10 min at 95°C. These sections were incubated for 1 h at 32°C with a monoclonal mouse antibody to VEGF (clone C1, Santa Cruz Biotechnology, Santa Cruz, CA) applied to the slide at a 1:500 dilution in antibody diluent (DAKO, Glostrup, Denmark). Immunohistochemical staining was performed using the streptavidin biotin kit (LSAB II, DAKO, Glostrup, Denmark). The peroxidase was visualized using 3-amino-9-ethylcarbazol as chromogen. The slides were rinsed with water, counterstained with hematoxylin and mounted. Negative controls were prepared by replacing the primary antibody with normal mouse serum.
The VEGF immunostaining was quantified by image cytometry with the CAS 200 (Becton Dickinson, Leiden, Netherlands) image analyzer. Staining quantification was performed only on glomeruli because they were the major areas of VEGF secretion in kidneys. This analysis was carried out using two thresholds. The first was performed at 620 nm to determine the total area of a glomerulus. The second threshold was determined at 540 nm for the detection of the VEGF positive immunostaining area. After merging each measured field, the computer calculated the total area of the glomeruli and the percentage of the immunostained area. Immunohistochemistry staining and staining quantification were performed in a blinded fashion.
The results are given as mean values ± 1 standard deviation. Comparison of VEGF expression levels between patients was evaluated by Mann-Whitney U-test. A p-value below 0.05 was considered significant.
All patients presented impaired renal function and proteinuria. In patients 4 and 5, proteinuria and increased creatinine serum were directly related to the introduction of sirolimus. In patients 1, 2 and 3, the potential role of sirolimus in renal failure cannot be analyzed since these patients had acute renal rejection. Two patients presented laboratory signs of a hemolytic syndrome (patients 2 and 3), and one had thrombopenia (patient 1) (Table 1). Sirolimus blood levels were measured low or in the normal range before and at the time of the event.
Histological description of sirolimus-associated TMA
Histological lesions of sirolimus-associated TMA were associated with other renal lesions: acute cellular rejection in patients 1, 2 and 3, chronic rejection in patient 4, and calcineurin inhibitor nephrotoxicity in patient 5. Absence of neutrophilic infiltration of peritubular capillary (PTC), negative C4d immunostaining of PTC, and lack of anti-HLA antibodies in patients 1, 2 and 3 argue against acute humoral renal allograft rejection.
Renovascular lesions observed were characterized primarily by fibrin deposition in the walls of the glomeruli. Glomerular thrombi were segmental and focal (patients 1, 4 and 5) (Figure 1A, F, G) or global and diffuse (patients 2 and 3) (Figure 1D). Thrombi might extend from the glomerular loop to the arteriolar vascular pole (patients 2 and 5) (Figure 1G). In more chronic stages of injury such as patient 1, diffuse glomerular lesions consisting in endothelial cell swelling and a split appearance of the basement membrane occurred (Figure 1B). Interstitial hemorrhage was the second characteristic lesion observed in all biopsies of these five patients. Interstitial hemorrhages were also present in native kidneys (patient 5, Figure 1H). Their localizations were limited (patients 1, 4 and 5) or diffuse (patients 2 and 3). In two patients (patients 2 and 3), interstitial hemorrhages were associated with fibrin thrombi within peritubular capillaries (Figure 1C, E). Figure 1E shows a peritubular capillary occluded by thrombi (arrow) with a large upstream dilation containing packed erythrocytes.
Renal expression of VEGF protein in patients with sirolimus-induced TMA
Because the mTOR inhibitor, sirolimus, is known to inhibit the production of VEGF, a key growth factor for endothelial cell survival, we quantified the expression of VEGF using immunostaining in the renal biopsies of the five patients. In agreement with previous studies, immunohistochemical staining of normal kidneys localized VEGF protein to visceral epithelial cells in glomeruli (Figure 2A). Endothelial cells of renal arteries were also noted but the level of expression was much lower than that found in glomeruli. In contrast with these results, the first biopsy performed in all five patients with sirolimus-associated TMA showed an extremely weak VEGF expression in podocytes (Figure 2B, D and Figure 3) and in arterial endothelial cells. Interestingly, analysis of the biopsies performed after the switch from sirolimus to tacrolimus (patient 1 and 2) showed reappearance of VEGF expression in podocytes (Figure 2C, E). Percentages of glomeruli surfaces immunostained for VEGF were 0%, 0.8% and 0% in the three biopsies from patients 1 and 2 carried out under sirolimus treatment, while they were measured at 13.6% and 8.58% in the two biopsies performed after the switch from sirolimus to tacrolimus. The second biopsy from patient 2 was performed only 14 days after the switch and despite ongoing signs of TMA, there was re-expression of VEGF in podocytes (Figure 2E). These results indicate that sirolimus was involved in the reduction of VEGF expression in podocytes.
Comparison of VEGF expression levels
To determine whether acute rejection or TMA by itself may affect VEGF production in podocytes, we examined VEGF expression in renal biopsies of 7 patients with acute rejection on a sirolimus-free calcineurin inhibitor-based protocol, 5 patients with TMA in native kidneys and one transplanted kidney with a factor-H-associated hemolytic uremic syndrome. VEGF expression was not different between normal transplanted and native kidneys (Figure 3). VEGF expression in the group of 5 patients with sirolimus-associated TMA was significantly lower than VEGF expression in normal transplanted and native kidneys (p < 0.01) (Figure 3). In contrast, renal VEGF expression in transplanted kidneys with acute cellular rejection was significantly higher than normal transplanted and native kidneys (p < 0.01). Finally, VEGF expression in podocytes was normal during TMA in native kidneys (Figure 3) and in one case of transplanted kidney with TMA associated with factor H gene mutation (data not shown). These results show that neither acute rejection nor TMA are involved in the reduction of VEGF expression in podocytes.
The present work suggests that (i) sirolimus may promote TMA in transplanted and native kidneys; (ii) sirolimus-associated TMA occurs preferentially in kidneys with concomitant renal injury; (iii) sirolimus-associated TMA is associated with a decreased expression of VEGF in kidneys.
In a retrospective study of 15 870 renal transplant patients, initial use of sirolimus was independently associated with an increased rate of TMA (9). Recent case reports and case series of renal, bone marrow and intestinal transplant recipients also suggest an increased risk of TMA in transplant kidneys and in native kidneys, in patients receiving a calcineurin inhibitor–sirolimus combination (3–8). Reported incidences of de novo TMA on calcineurin inhibitor–sirolimus combination protocols have varied considerably, ranging from 1.5% to 48%, depending on the type of calcineurin inhibitor drug (ciclosporin or tacrolimus) (4), the dosing of sirolimus (10), and the diagnosis criteria for TMA (biological systemic signs or histological signs on renal biopsies) (4,10). Only few patients start transplantation on a calcineurin inhibitor-free sirolimus-based protocol. Nevertheless, two preliminary reports suggested that sirolimus in the absence of calcineurin inhibitors may induce TMA in renal transplants (11,12). We report 4 cases of TMA on calcineurin inhibitor-free sirolimus-based protocols. Three cases were associated with acute rejection. We cannot eliminate the exclusive role of acute rejection in induction of TMA in these three patients. However, vascular rejection-induced TMA is associated with acute humoral rejection (AHR), and none of the three patients had clinical, biological and pathological criteria for AHR. In patients 4 and 5, TMA occurred after introduction of sirolimus. Patient 4 had chronic graft rejection. He had a transplant for 33 years and had never received calcineurin inhibitors. Patient 5 had chronic calcineurin inhibitor nephrotoxicity. All five patients have a simultaneous cause of endothelium damage. This point suggests that sirolimus-induced TMA may occur mainly in kidneys with concomitant endothelial cell lesions. In these cases, sirolimus may act as a subsequent aggressor, second hit, intensifying endothelial cell damage and thus leading to a much greater risk of TMA.
Our results show that sirolimus-induced TMA is associated with a decreased expression of VEGF in kidneys. Decreased expression of VEGF seems to be a consequence of sirolimus treatment since (i) analysis of the biopsies performed after the switch from sirolimus to tacrolimus (patients 1 and 3) showed reappearance of VEGF expression in podocytes and (ii) there was no decreased expression of VEGF during acute rejection and classical TMA. In contrast, our data show that there is overexpression of VEGF during acute rejection. Expression of VEGF has also been shown to be up-regulated in chronic rejection and in chronic cyclosporine nephrotoxicity (22,23). One can speculate that overexpression of VEGF observed during acute or chronic renal microvascular injury protects endothelial cells from further damage (22,24,25). Indeed, renal VEGF expression has been shown to be crucial to preservation and repair of endothelial glomerular cells and peritubular capillaries in some experimental models (26), and VEGF treatment has been shown to improve recovery in an experimental model of TMA (24). Recent data using the Cre-loxP technology in transgenic mice have demonstrated that there is ongoing requirement for tight regulation of VEGF signalling between the podocyte and glomerular endothelium (21). When VEGF levels in the renal podocytes dropped 50%, there was swelling followed by necrosis of glomerular endothelial cells (21). VEGF has also been shown to support peritubular endothelial cell survival (27). Alteration of VEGF production is therefore one mechanism by which sirolimus may increase the risk of glomerular and peritubular capillary TMA. In this two-hit model, one may speculate that alteration of VEGF renal production induced by sirolimus exacerbates pre-existing endothelial cell lesion. In this hypothesis, downregulation of renal VEGF is not the main cause but a predisposing factor of TMA. This is different from calcineurin inhibitor-induced TMA, which does not appear in the context of rejection or other causes of endothelial cell damage. Cyclosporin has a direct and dose-dependent cytotoxic effect on endothelial cells (28). Decreased VEGF levels may occur in all patients on sirolimus and only patients with a second cause of endothelial injury (such as acute rejection, chronic rejection or calcineurin inhibitor nephrotoxicity) may develop sirolimus-associated TMA. This hypothesis may provide a potential explanation for the high risk of TMA in patients receiving a cyclosporin–sirolimus combination. It may also explain that in patient 2, VEGFre-expression did occur despite ongoing signs of TMA after discontinuation of sirolimus. Demonstrating that alteration of renal VEGF expression is a predisposing factor to the development of TMA requires further studies on animal models. Another possibility is that downregulation of VEGF occurs merely in patients who develop sirolimus-induced TMA. In this option, sirolimus-induced downregulation of VEGF may depend on various factors such as polymorphisms in the VEGF gene, sirolimus blood levels or duration of transplantation. Sequential kidney biopsies of sirolimus-treated patient are needed to determine the conditions leading to sirolimus-induced downregulation of renal VEGF.