Proteinuria Following a Switch from Calcineurin Inhibitors to Sirolimus

Abstract

Background. 

Sirolimus is an alternative option for kidney transplant patients treated with calcineurin inhibitors (CNI) when renal function is deteriorating. However, the incidence of proteinuria following a switch from CNI to sirolimus has caused concern, and was therefore investigated here.

Methods. 

In a retrospective study, 68 renal transplant recipients were switched from CNI to sirolimus. Proteinuria was measured using 24-hour urine collection before the switch and collections 3, 6, 12, and 24 months thereafter. In addition, proteinuria was measured in patients who had to be switched back to CNI due to side effects. Survival analyses were performed.

Results. 

Baseline proteinuria was 0.39±0.69 g/day in all 68 patients. It increased to a mean 1.44±1.90 g/day at 3 months (P<0.001) and remained elevated at 6, 12 and 24 months. When sirolimus was withdrawn after the CNI-sirolimus switch for 19 patients, proteinuria decreased from 1.95±2.06 g/day to 0.9±1.4 g/day (P<0.05). Proteinuria above 0.3 g/day before the CNI-sirolimus switch correlated significantly with the decrease of renal function thereafter.

Conclusion. 

CNI-treated kidney transplant recipients may develop reversible proteinuria when switched to sirolimus.

Nephrotoxicity and hypertension are associated with calcineurin blockade (^{1, 2}). Because renal function and blood pressure control are important predictors of graft survival (^{3, 4}), new immunosuppressive drugs are considered to constitute an alternative to cyclosporine (CsA) and tacrolimus, and may prevent arteriolar hyalinosis, glomerulosclerosis and tubulointerstitial damage which are liable to occur with CNI (⁵). The absence of hemodynamic-induced nephrotoxicity of sirolimus (rapamycin) coupled with its immunosuppressive properties, make it an interesting alternative to CNI.

According to a recent study, withdrawing CsA from a Sirolimus-CsA-steroid regimen 3 months after transplantation results in long-term benefits for renal transplant recipients (⁶).

Although switching from CNI to sirolimus when chronic allograft nephropathy (CAN) is suspected, with or without evidence of CNI nephrotoxicity on renal biopsy, is an everyday practice in many units, it might be complicated by the development of proteinuria and nephrotic syndrome (⁷). The incidence of such proteinuria and its signifiance need to be determined, because proteinuria has a significant prognostic value and is an accurate predictor of graft loss (⁸). The study was therefore designed to clarify these questions.

PATIENTS AND METHODS

Study Design

A retrospective study was conducted at the Necker and Saint-Louis Hospitals in Paris (France), two centers which have used sirolimus since 1997 and where 24-hour urine collection of urine and routine biopsies are standard practice. The inclusion criteria were:

• Treatment with CNI (i.e. CsA or tacrolimus) from the time of transplantation,
• At least one measurement of urinary protein excretion during the 3 months preceding the switch.
• At least 3 months of follow-up after the switch and at least one measurement of urinary protein excretion 3 or 6 months after the switch.

Overall, 105 patients were switched to sirolimus. Thirteen patients were monitored less than 3 months because they were back to hemodialysis (n=1) or because of adverse events due to sirolimus and proteinuria (n=12). Twenty-four patients did not meet the other criteria. Sixty-eight patients finally met these inclusion criteria and were included.

The indications for switching from CNI to sirolimus were:

1. The existence of CAN lesions on routine renal biopsies in 54 of the 68 patients, with histologic evidence of calcineurin inhibitor nephrotoxicity in at least 43. Patients with Banff grade I, grade II or grade III CAN were respectively 15 (28%), 27 (50%) and 12 (22%). Unfortunately, the level of glomerular damage (cg) was not available for all patients.
2. A progressive worsening of renal function in 11 patients: serum creatinine level increased from 135±36 μmol/L to 167±35 μmol/L (+24%) the year before the switch. These patients were not selected because they already had proteinuria and they were on a similar baseline immunosuppression than patients of the biopsy-proven CAN group.
3. Skin, kidney and breast cancer respectively in the remaining three cases, in view of the presumed antitumoral properties of sirolimus.

Nevertheless, 9 of the 14 last patients had a biopsy within the year after calcineurin inhibitor withdrawal. CAN lesions were seen in all cases.

The switch was guided by a previously described protocol (⁷), so that the period of overlap did not exceed 4 weeks. All patients were switched from their calcineurin inhibitor regimen to sirolimus without changing the other immunosuppressants. The doses of sirolimus were adjusted to achieve serum levels of 10 to 15 ng/ml.

Proteinuria was measured by 24 hr urine collection during the 3 months before the switch and 3, 6, 12 and 24 months thereafter (i.e. between months 2 and 4, 5 and 7, 10 and 14 and 20 and 28 respectively). When several measurements of proteinuria were available, the one closest to the switch was retained. Proteinuria was also assessed when sirolimus had to be stopped and calcineurin inhibitors reintroduced. In the 19 patients concerned, it was measured during the 6 months before sirolimus withdrawal and 12 months thereafter. We used serum creatinine rather than an estimation of the glomerular filtration rate because weight and urine creatinine values were sometimes not recorded at the same time as proteinuria.

Patients

Sixty-eight patients comprising 32 females and 36 males (mean age: 44±12 years,) were included in the study. Of these, 59 had been grafted with a cadaveric kidney. The switch from CNI to sirolimus was performed after a mean graft duration of 55±52 months (range of times: 2-197 months). Baseline population characteristics are summarized in Table 1. Three patients had biopsy-proven FSGS lesions before the switch and one patient had a recurrence of membranoproliferative nephropathy before the switch. Twenty-seven patients experienced at least one biopsy-proven acute rejection episode before switching (seven patients experienced 2 episodes and one patient 3 episodes). Subclinical acute cellular rejection (Banff classification: 1a), associated to CAN, was detected in three biopsies performed within 3 months before the switch: one patient received IV solumedrol and one was switched from CsA to tacrolimus less than 3 months before the switch. There were antibodies to donor HLA in the sera of nine patients with biopsy-proven CAN before the switch. Eighteen patients had a biopsy after the switch that revealed CAN lesions (n=18), focal and segmental glomerulosclerosis (FSGS) lesions (n=4, of which de novo FSGS: n=3) and acute rejection (n=1). The mean follow-up duration of the 68 patients in this study was 12±8 months. The fall in numbers for whom complete data is available corresponds to the end of follow-up and not to an artefact (such as measuring proteinuria only in patients with significant urinary protein excretion). Sirolimus was stopped in 29 patients, because of adverse events or because they were back to hemodialysis (Table 1). The mean duration of treatment of the 29 patients was 10.6±8.2 months. 19 of these patients were switched back to CNI.

TABLE 1: Patient characteristics

Statistical Methods

Proteinuria and serum creatinine level before and after the switch and proteinuria of patients with or without a past history of acute rejection were compared by the Mann Whitney and the Wilcoxon tests.

The role of the CAN in the development of proteinuria was assessed by the Kruskal-Wallis test, according to the proteinuria of patients with Banff classification grade I, grade II or grade III CAN lesions.

Patients were divided into 2 groups: non-progressors, with improved or stable renal function (serum creatinine value always below or equal to the preswitch value + 10%) and <<progressors>> (serum creatinine value above the preswitch value +10% at least once during follow-up). A survival analysis was performed to evaluate the impact of different variables on the decrease in renal function. For univariate analysis, we used the Cox model, which incorporates the quantitative variables assumed to determine renal outcome. These were proteinuria before the switch and 3 months thereafter, serum creatinine before the switch, age at the time of the switch, the serum creatinine level of the kidney donor, donor age, duration of cold ischemia and graft duration. The validity of the proportional hazard hypothesis was assessed by graphic methods. The following qualitative variables were studied by an actuarial method, using the log rank test to calculate signifiance: proteinuria before the switch above 0.3g/24h, proteinuria 3 months after the switch below or above 1g/ day, sex, presence of a nephrotic syndrome before transplantation, anti-HLA immunization, delayed graft function, donor death due to a cardiovascular event. All statistical tests were two-sided at the 5% signifiance level. Results are expressed as means ± SD as indicated. Statistical analyses were performed with SAS system software.

RESULTS

Baseline proteinuria was 0.39±0.69 g/day (68 patients). Proteinuria was significantly different before and after the switch (Fig. 1), and we observed that it increased to a mean of 1.44±1.90 g/day at 3 months (P<0.0001, 56 patients), 1.63±2.14 g/day at 6 months (P<0.0001, 46 patients,), 1.41±1.39 g/day at 12 months (P<0.0001, 41 patients) and 1.64±1.90g/day at 24 months (P<0.0001, 20 patients). Of 68 patients, 10.3% had a proteinuria over a gram/24h before the conversion and respectively 41.1%, 45.7%, 46.3% and 40% at 3, 6, 12 and 24 months (Fig. 2). Sixty-five patients had an increase in proteinuria after the switch and 3 patients had a (slight) decrease. We noted that some individual patients had no significant proteinuria before the switch but developed nephrotic-range proteinuria after the switch (for instance: 0.11 g/24h before the switch and 4.74 g/24h 3 months thereafter or 0.27 g/24h before the switch and 3.82 g/24h 3 months thereafter). When urinary proteins were analyzed, the main protein found was albumin, thus indicating the glomerular origin of the proteinuria.

FIGURE 1.: Mean proteinuria and serum creatinine in 68 renal transplant recipients before and after a switch from calcineurin inhibitors to sirolimus. Error bars represent SEM.

FIGURE 2.: percentage of patients with urinary protein excretion over 1g/24h.

Baseline serum creatinine was 172±65 μmol/L (68 patients). Mean serum creatinine decreased to 157±65 μmol/L (56 patients) 3 months after the switch but this decrease was not significant (P=0.08). Serum creatinine was 180±109 μmol/L at 6 months (46 patients), 171±82 μmol/L at 12 months (41 patients), and 152±52 μmol/L at 24 months (20 patients, P=0,20, nonsignificant).

When the 54 patients with biopsy-proven CAN before the switch were studied separately, we observed a similar course of proteinuria and serum creatinine than in the whole group. Baseline proteinuria was 0.362±0.068 g/day (54 patients) and increased to 1.35±1.87 g/day at 3 months (P<0.0001, 43 patients), 1.67±2.31 g/day at 6 months (P<0.0001, 39 patients), 1.27±1.20 g/day at 12 months (P<0.0001, 31 patients) and 1.14±1.40 g/day at 24 months (P=0.0003, 16 patients). Serum creatinine was 179±67 μmol/L before the switch and decreased to 160±61 μmol/L at 3 months (P=0.0516). Serum creatinine was 187±117 at 6 months and 177±90 μmol/L at 12 months but there was a significant decrease at 24 months (141±42 μmol/L, P=0.02).

The increase of proteinuria was not explained by a lower use of Angiotensin-Converting Enzyme Inhibitors (ACEI) or Angiotensin Receptor Blockers (ARB): 34% patients received ACEI or ARB before the switch, 34% at 3 months, 37% at 6 months, 46% at 12 months and 45% at 24 months. Proteinuria was assessed in eleven patients that were given ACEI or ARB for the first time while they received sirolimus: it decreased from 2.29±2.27 g/24h to 1.26±1.42 g/24h. Baseline proteinuria of the 27 patients that experienced biopsy-confirmed and treated acute rejection was not significantly different from the proteinuria of the other patients: 0.53±0.80 g/24h vs. 0.32±0.61 g/24h (P=0.06). There was no difference between the two groups at 3, 6, 12 and 24 months (P=0.78, 0.53, 0.76 and 0.84 respectively). In a similar way, there was no significant difference between the proteinuria level of patients with Banff grade I, grade II or grade III CAN (data not shown).

Proteinuria was reversible at least partially: thus, when 19 patients had to be switched back from sirolimus to CNI due to side effects, their proteinuria decreased significantly from 1.95±2.06 g/day to 0.9±1.41 (P=0.001), while the serum creatinine increased in a non significant manner from 181±83 to 204±111 μmol/L (Fig. 3). The proteinuria of these 19 patients was 0.33±0.44 g/24h before the switch to sirolimus (comparison preswitch to (¹) sirolimus: P=0.003, and (²) after return to CNI: P=0.11, ns).

FIGURE 3.: Mean proteinuria and serum creatinine in 19 transplant recipients before and after a switch from calcineurin inhibitors to sirolimus and after a second switch from sirolimus to calcineurin inhibitors. Error bars represent SEM.

Data at 12 months (±1 month) before the switch were available in 46 patients. Twenty-seven of these 46 patients were classified as non-progressors and 19 were classified as progressors according to the evolution of the renal function after the switch. The rate of decline of renal function the year before the switch was 16.5% in the non-progressors group (serum creatinine: 135±43 μmol/L to 157±52 μmol/L) and 16.7% in the progressors group (serum creatinine: 146±37 to 170±40 μmol/L).

Survival analyses revealed that some variables were similar in the 2 groups: graft duration at the time of the switch (P=0.3655), donor age (P=0.1105), duration of cold ischemia (P=0.6942) and sex (P=0.2169). Proteinuria above 1g/24h 3 months after the switch might be harmful but the results in this respect were not significant (P=0.0817).

However, proteinuria above 0.3 g/24h before the switch was significantly related to a decrease in renal function after the switch (P=0.0049). Other variables could not be used because the validity of the proportional hazard hypothesis was not fulfilled.

Baseline proteinuria was 0.19±0.27 g/24 hr in non-progressors (n=43) and 0.76±0.99 g/24 hr in progressors (n=25). Proteinuria remained higher after the switch in the progressors group than in the non progressors group (P<0.001). Serum creatinine level was similar in progressors and non-progressors groups before the switch (186±78 vs. 163±54, P=0.15).

DISCUSSION

This study shows that in renal transplant recipients with CAN, a switch from CNI to sirolimus is significantly associated with the development of proteinuria during the 3 months after the switch.

De novo treatment by sirolimus has not been so far associated with the development of proteinuria (^{6, 9}). Peddi et al. recently reported that proteinuria was declared in 3 patients (5%) only after that 60 established renal allograft recipients with moderate renal insufficiency were switched from CNI to sirolimus after a mean time from transplantation of 60.8 months (¹⁰). Nevertheless, proteinuria was not systematically recorded in these studies.

In contrast, Butani et al. recently described that, out of 13 pediatric renal transplant recipients, 12 developed heavy proteinuria after a mean interval of 1 month when sirolimus was used as a rescue therapy and Diekmann et al. reported that proteinuria increased significantly in patients whom renal function deteriorated after conversion from CNI to sirolimus (^{11, 12}).

The role of sirolimus in the development of glomerular lesions remains debated. FSGS and posttransplantation glomerulonephritis occurred during the months after the switch from CNI to sirolimus (^{7, 13}) but there are few experimental data supporting the hypothesis of a direct toxicity of sirolimus for the glomeruli (¹⁴).

In our study, at least 54/68 patients had documented CAN before the switch, the most frequent form of nephropathy involved in posttransplant nephrotic syndrome (¹⁵). According to the results of Nankivell et al., it is difficult to separate out CNI toxicity from other causes of CAN but there is evidence that 10% of the patients on a CNI-based immunosuppressive regimen may develop FSGS lesions and that CsA induces glomerular injury in cardiac transplant recipients (^{5, 16, 17}).

Because CNI reduce renal blood flow, CNI withdrawal might reveal existing glomerular lesions that result in proteinuria. Incidentally, “improved” renal function (i.e. lower serum creatinine within weeks after the switch) might be a marker of increased renal blood flow. The reversibility of proteinuria after withdrawal of sirolimus and the reintroduction of CNI supports the hypothesis that hemodynamic mechanisms play an important role in the development of proteinuria. Myers et al. described that the excretion rate of albumin increased significantly in heart transplant recipients after CsA elimination without replacement by sirolimus, from an initial value of 162±70 μg/min (0.23±0.1 g/24h) to 546±300 μg/min (0.79±0.43 g/24h) at 48 months (¹⁷). In a prospective study involving 212 stable renal recipients randomized to stop either CsA, prednisone or continue triple drug therapy at 6 months after transplantation, Smak Gregoor et al. did not observe a significant change in proteinuria in any group (¹⁸). In contrast, Ducloux et al. have reported that in 31 renal transplant recipients whose CsA was withdrawn because of cyclosporine toxicity or CAN, baseline proteinuria was 0.79±0.6 g/day and increased to 1.79±0.8 g/day 24 months after CsA withdrawal (¹⁹). The increase of urinary protein excretion after CsA withdrawal reported by Ducloux and Myers was however noticed 24 or 48 months after CsA withdrawal, so that the progression of underlying disease could have taken part in the increase of proteinuria.

Nevertheless, the dramatic increase of proteinuria in the short term after conversion to sirolimus and the development of nephrotic range proteinuria in some individual patients without significant proteinuria before argues in favor of a role of the sirolimus itself.

The harmful effect of proteinuria on the kidney is well established and proteinuria above 1g/24 hr 3 months after a switch from CNI to sirolimus might be predictive of a decline in renal function, but our results in this respect failed to reach significance maybe owing to the lack of power and the short duration of the follow-up.

Here, the existence of significant proteinuria before the switch correlated closely with a decline in renal function after the switch, suggesting that the existence of glomerular lesions such as CAN is predictive of renal outcome after the switch. Diekmann et al. recently reported that a proteinuria below 800 mg/day at conversion from CNI to sirolimus is the only independent predictor for positive outcome in chronic allograft dysfunction (¹²). Fervenza et al. revealed that standard doses of sirolimus can induce acute and partially reversible nephrotoxicity in native kidneys of patients with chronic glomerular disease, proteinuria ≥1.0 g/24h and moderate renal dysfunction (²⁰). These data suggest that the combination of proteinuria and sirolimus can cause acute nephrotoxicity. In vitro studies have shown that sirolimus induces apoptosis and cell-cycle arrest in mouse proximal tubular cells (²¹). Since sirolimus binds to the albumin in serum, the increase in glomerular permeability, and the resulting proteinuria and albumin uptake by proximal tubular cells, might induce in vivo exposure of these cells to sirolimus.

In the absence of a control group, it seems difficult to determine whether the evolution of patients exhibiting proteinuria would have been more favorable if CNI treatment had been maintained. Weir et al. previously reported that CNI treated recipients with CAN deteriorate at a faster rate when CNI drugs are maintained (^{22, 23}). Nevertheless, these authors considered that reduction or withdrawal of CNI did not help all patients. At least, there is some concern about the high rate of other sirolimus-related adverse events in this study (Table 1).

In conclusion, switching renal transplant recipients with CAN from CNI to sirolimus is associated with high levels of proteinuria, sometimes in the nephrotic range. It is likely that CNI withdrawal might reveal underlying glomerular disease and thus contribute to the development of proteinuria. However the dramatic increase in proteinuria after the switch, and the existence of nephrotic range proteinuria after the conversion in individual patients without significant proteinuria before suggest that sirolimus is involved too. Moreover, patients exhibiting mild proteinuria, and by implication those with glomerular disease before the switch, constitute a subgroup likely to experience an accelerated decrease in renal function after the switch. The safety of sirolimus use is questionable when chronic allograft nephropathy is suspected, and earlier conversion from CNI to sirolimus in stable renal transplant recipients might be safer and more effective.