Acute renal failure in paediatric systemic lupus erythematosus:
treatment and outcome
S. M. Benseler 1,2,3 , J. M. Bargman 4 , B. M. Feldman 1,2,3 , P. N. Tyrrell 1 , E. Harvey 5 , D. Hebert 5 and E. D. Silverman 1,6 Objective. To determine the outcome of paediatric SLE (pSLE) patients with nephritis who developed acute renal failure (ARF). Efficacy and safety of treatment regimens were compared. Methods. A total of 249 pSLE patients were diagnosed and prospectively followed at a single centre between July 1973 and July 2003; 127 children (51%) had lupus nephritis. ARF was defined as serum creatinine of >250mmol/l or >75% above baseline. Standardized assessments included clinical data and medications, laboratory testing, disease activity and damage scores were obtained. Subsequent renal flares were documented. Primary outcome: renal function at last follow-up. Secondary outcomes: treatment efficacy and safety. AZA- and cyclophosphamide (CYCLO)-treated patients were compared. Propensity score methods were applied to balance covariates. An intention to treat approach was chosen. Results. The ARF study cohort included 50 patients; 13 boys and 37 girls with a median age of 13.2 yrs at diagnosis and a mean follow-up of 45 months. Renal histology: Class III nephritis in 16; Class IV in 34. Dialysis requirement and disease activity were similar in both groups. Treatment: AZA in 33 patients, CYCLO in 9 and corticosteroids only in 8. Outcome: no statistically significant or clinically relevant differences were found for any of the outcome measures including last serum creatinine, time to renal flare, overall renal survival, disease activity over time, disease damage, mean annual corticosteroid dose and rate of infection. Conclusion. The treatment of renal failure in this pSLE cohort was associated with an excellent outcome. AZA and CYCLO were equally efficacious. KEY WORDS: Adolescent rheumatology, Rheumatic diseases, Systematic lupus erythematosus, Autoimmunity, Rheumatic disease, Renal, Tissues, Outcome measures, Health services and practice. Introduction Paediatric SLE (pSLE) accounts for 20% of all cases of SLE, with nephritis occurring in 50% [1]. Renal disease in general, and the development of acute renal failure (ARF) and end-stage renal disease (ESRD) in particular, are important factors contributing to disease morbidity and mortality [27]. Long-term follow-up studies have demonstrated that ESRD may occur in up to 50% of adult SLE patients and is most commonly associated with Class III and Class IV lupus nephritis [3, 8]. Although very few controlled studies have been performed, current management of patients with Class III and Class IV nephritis includes corticoste- roids and another immunosuppressive agent, most commonly cyclophosphamide (CYCLO) [9]. pSLE nephritis is commonly treated similarly to adults [10, 11] and CYCLO is usually considered when children present with Class III and Class IV nephritis. Elevated serum creatinine, histological evidence of proliferative nephritis and disease onset during childhood have been shown to be associated with a poor long-term outcomes [12]. We have previously reported that pSLE patients with Class III and Class IV nephritis treated with a combination of AZA and corticosteroids have a long-term renal survival similar to that reported in adult patients treated with a combination of CYCLO and corticosteroids [13]. However, the role of AZA vs CYCLO for the treatment of the subset of patients with lupus nephritis who present with renal failure remains highly controversial. Moreover, studies examining cytotoxic agents have included patients with only mild renal impairment 1417], which may obscure conclusions drawn about their efficacy. The aim of this study was to analyse and compare the outcome of two different treatment protocols for paediatric patients with lupus nephritis who experienced very significant renal impairment in our centre over the last three decades and to report on the outcome of these patients. Patients and methods Study design and patient search A single centre cohort of consecutively followed children diagnosed with pSLE and followed between July 1973 and July 2003was performed. Approval from the Research Ethics Board at The Hospital for Sick Children, Toronto, was obtained (REB file No.1000000966). All pSLE patients have been followed by a single physician in the paediatric lupus clinic (E.D.S.) utilizing a standardized assessment protocol since 1984. Data before 1984 were obtained from chart review. Patients, cohorts and treatment regimens A total of 249 consecutive pSLE patients were followed. Lupus nephritis was seen in 127/249 children (51%). Fifty of the 127 patients with nephritis met the definition of ARF described below and represented the inception cohort. Based on therapy received, children were assigned to three different treatment cohorts exclusively (i) AZA: 33 patients received daily oral AZA (13 mg/kg/day). The target dose was 3 mg/kg/day (maximum dose of 150 mg/day). Lower doses were used in the face of leucopenia, to maintain total white blood cell 53000, or in case of haepatotoxicty; (ii) CYCLO: nine patients received monthly intravenous CYCLO (5001000 mg/m 2 ) for seven courses. In 2/9 patients, this was followed by six courses of 3 monthly intravenous CYCLO. In the other seven CYCLO-treated patients, the seven courses were followed by AZA maintenance therapy for at least 1 Division of Rheumatology, Department of Paediatrics, 2 Child Health Evaluative Sciences, Research Institute, The Hospital for Sick Children, 3 Department of Health Policy, Management and Evaluation, 4 Division of Nephrology, Department of Medicine, University Health Network, University of Toronto, 5 Division of Nephrology, Department of Paediatrics, The Hospital for Sick Children and 6 Department of Immunology, University of Toronto, Toronto, Canada. Submitted 31 August 2007; revised version accepted 7 November 2008. Correspondence to: E. D. Silverman, Division of Rheumatology, Department of Pediatrics, The Hospital for Sick Children, 555 University Avenue, Toronto, Ontario M5G 1X8, Canada. E-mail: earl.silverman@sickkids.ca Rheumatology 2009;48:176182 doi:10.1093/rheumatology/ken445 176 The Author 2009. Published by Oxford University Press on behalf of the British Society for Rheumatology. All rights reserved. For Permissions, please email: journals.permissions@oxfordjournals.org
b y
g u e s t
o n
J u l y
2 4 ,
2 0 1 4 h t t p : / / r h e u m a t o l o g y . o x f o r d j o u r n a l s . o r g / D o w n l o a d e d
f r o m
18 months. (iii) STE: eight patients were treated with steroids only and did not receive a second-line agent, but received oral prednisone or equivalent (2 mg/kg/day). All AZA and CYCLO patients received the same additional standardized corticosteroid- treatment regime. Prednisone was started at 2 mg/kg for 6 weeks and then slowly tapered over 9 months in 4-weekly intervals. Low-dose corticosteroids were then continued on alternating days. Anti-hypertensive therapy was implemented when required. Choice of treatments varied by time. All patients treated with steroids only were seen prior to 1985. The choice of CYCLO varied by time and whether there was any severe CNS involve- ment. Specifically, the nine patients treated with CYCLO received this therapy for the following reason: (i) severe CNS involvement for five patients; (ii) treated shortly after the publication of the 1986 NIH papertwo patients and (iii) two patients received CYCLO during a time when one of the authors was on sabbatical. All other patients received AZA. The only clinical characteristic which increased the likelihood of receiving CYLCO rather than AZA was the presence of severe CNS disease and not the degree of azotaemia or the need for dialysis. ARF study cohort The ARF study cohort included all pSLE patients, who (i) were diagnosed and followed at the study centre and (ii) had biopsy confirmed lupus nephritis and ARF. The statistical analysis specifically compares the outcome of patients who were treated with either daily oral AZA or monthly intravenous CYCLO as a second-line immunosuppressive agent for nephritis and ARF since this reflects the current standard of care. Definitions Renal failure. Renal failure was defined as an increased serum creatinine >75% above baseline value on at least two consecutive occasions or repeated serum creatinine levels >250 mmol/l. Baseline was defined as the stable creatinine value over the preceding 6 months. Renal flares. A renal flare was defined as increased renal disease activity requiring a significant increase in steroid dosing of >400% or an increased dose to the usual maximum dose of 60 mg/day. Significant infections. Infections were considered significant if the patient (i) required hospitalization and/or (ii) was commenced on oral or intravenous anti-bacterial or anti-viral therapy. Demographics, clinical data and follow-up All patients were followed in the interdisciplinary paediatric lupus clinic at our tertiary paediatric care centre. Patient characteristics including sex, age at diagnosis of pSLE and ARF were recorded. In addition, the time intervals from diagnosis of pSLE to ARF and the follow-up intervals were determined. Measures of nephritis and ARF including the WHO histological classification of lupus nephritis, requirement for dialysis and the overall and renal domains of the SLE disease activity tool and SLEDAI were documented. Hypertension was defined as elevated blood pressure compared with height- and age-adjusted controls and expressed in z-scores [18]. Hypertension was noted when present before start of therapy at the time of diagnosis of ARF. Significant pSLE-related comorbidities including concurrent CNS disease were determined. With each visit all systems were reviewed. A complete physical examination assessed all organ systems potentially affected by pSLE. Associated conditions and complications of both disease and treatment were sought. Treatment protocols, changes in medication dosing, adverse events and intolerance to treatment were documented. Study follow-up assessments were recorded at 0, 3, 6, 9, 12, 18 and 24 months and then yearly. Intervals were re-assigned when a flare occurred. Patient follow-up assessments were then noted analogous to the disease onset. Disease activity and damage scores SLEDAI, ECLAM and SLICC-Damage Index For each visit disease activity was determined utilizing the previously validated SLEDAI [19, 20], the modified version SLEDAI-2K [20] and the ECLAM [21, 22]. Disease activity over time was calculated as mean area under the curve (AUC per time interval) [23]. Disease damage was scored starting at 6 months of disease using the SLICC-Damage Index (SLICC-DI) [24, 25]. Individual SLICC-DI domains including all renal domains (proteinuria >3.5 g/24 h and ESRD), bone damage, cataracts, malignancies, insulin-dependent diabetes mellitus (IDDM) and premature gonadal failure were analysed separately. Laboratory test results Standardized laboratory testing at each visit included: (i) renal function: serum creatinine (CR), urea, albumin (ALB) and urinalysis including dipstick for blood and albumin, microscopy, spot urine for protein:creatinine ratio (P:CR) and protein in 24 h urine collection (uPRO); (ii) inflammatory markers and haema- tology tests: ESR, complete blood count (CBC)/differential including polymorphs, lymphocytes (LYM), haemoglobin (HGB), platelet count (PTL), C3 and C4 complement levels; (iii) liver function: aspartate transaminase (AST) and alanine transaminase (ALT); (iv) SLE serology including ANA, anti- DNA, anti-Ro, anti-La, anti-Sm, anti-RNP and aCL antibodies and (v) pro-thrombotic testing including LAC [thrombin time (TTI), platelet neutralization procedure (PNP) and Russell viper venom (RVV)] and PTT/international normalized ratio (INR) when indicated. Abnormal test results were documented. Infections were determined in viral and bacterial cultures, serology from blood, urine or CSF, when suspected. Imaging Imaging including radiographs and MRI was performed for disease manifestations and complications when indicated. Outcomes Treatment efficacy, damage and safety markers were determined for the study population as well as for each treatment cohort separately. (i) Primary outcome Renal function: (1) overall renal survival at last follow-up (deaths, renal transplants and ESRD); (2) renal function parameters (creatinine and urine protein) at last follow-up; (3) renal flare rates during study interval; and (4) time to renal flare. (ii) Secondary outcomes (a) Disease activity and treatment efficacy: (1) inflammatory markers (ESR and complement) and lupus serology in response to therapy over time; (2) mean disease activity (SLEDAI) over time (AUC); (3) disease activity (SLEDAI) at last follow-up; and (4) mean annual corticosteroid dose. (b) Disease damage: Overall damage score (SLICC-DI) at last follow-up visit and specific SLICC-DI domains including: (1) renal damage: percen- tage of patients with evidence of SLICC-DI proteinuria 53.5 g/ 24 h, reduced estimated glomerular filtration rate of <50% or end Outcome of acute renal failure in pSLE 177
b y
g u e s t
o n
J u l y
2 4 ,
2 0 1 4 h t t p : / / r h e u m a t o l o g y . o x f o r d j o u r n a l s . o r g / D o w n l o a d e d
f r o m
stage renal disease; (2) bone damage: percentage of patients with evidence of SLICC-DI osteoporosis with vertebral fractures, avascular necrosis (AVN) or more than one episode of AVN and (3) eye damage: percentage of patients with cataracts. (c) Treatment safety Percentage of patients with new onset of IDDM; gonadal failure or malignancies. In addition, the rate of significant infections, as defined earlier, was determined. Analysis All clinical, laboratory, radiographic and score data were entered into a designated pSLE Access
database (Microsoft Cor-
poration, Seattle, WA, USA). Baseline demographic data were compared using descriptive statistics. As current therapies of Classes III and IV nephritis include the use of an immunosup- pressive agent, statistical comparisons included patient data of the AZA and CYCLO cohorts exclusively. KaplanMeier survival analysis. Time to renal flare was plotted for the AZA and CYCLO groups using KaplanMeier curves. Both groups were compared in an unadjusted analysis using the log-rank statistic. An adjusted analysis (comparing pairs of patients matched by propensity score, N9 pairs) for time to renal flare was performed using Cox proportional hazards regression. Propensity score matching. A propensity score was generated in order to balance covariates and reduce unobserved selection bias between AZA- and CYCLO-treatment cohorts. The method adjusts for confounding by intent (which, if it occurred, might have led to a group of more severely affected pSLE patients being allocated preferentially to one of the treatment protocols). To determine the propensity score (which is the probability, based on baseline disease and demographic features, that any given patient will be treated with CYCLO), we identified clinically and/or statistically important baseline features that may have differed between the treatment groups. The variables included in the final propensity score equation were: (i) demographics and baseline characteristics: age at pSLE diagnosis, gender, time to nephritis/ ARF, age at ARF pSLE diagnosis; (ii) comorbidities: CNS lupus; (iii) renal function at diagnosis: dialysis requirement, hypertension, creatinine (CRE), ALB, P:CR, uPRO; (iv) disease activity: SLEDAI and ECLAM; (v) laboratory pSLE markers: ESR, CRP, C3, C4, HGB, WBC and PTL; and (vi) SLE serology ANA and dsDNA. Using a logistic regression model with these variables as predictors, we determined the predicted probability that any individual patient would be treated with CYCLO (JMP 5.0.1.2, 2004 SAS Institute Inc. Cary, NC, USA). The logistic model predicted well which treatment group a patient would be in (area under the receiver operating curve of 0.89). Each patient who received CYCLO was matched to the patient in the AZA group with the closest propensity score. For all study outcomes an intention to treat approach was chosen. Results Patients and treatment cohorts The ARF cohort consisted of 50 patients, of whom 33 received AZA plus corticosteroids (AZA cohort), 9 CYCLO plus corticosteroids (CYCLO cohort) and 8 corticosteroids only (STE cohort). There were 13 boys and 37 girls (1: 2.8). The median age at diagnosis of pSLE was 13.4 yrs (range 3.119.3 yrs). ARF and nephritis was present at the time of diagnosis of pSLE in 37 of the 50 patients (74%), 5 patients developed nephritis and renal failure within the first year following diagnosis of pSLE (total of 84% within the first year), another 7 patients (14%) within 25 yrs and 1 patient after 11.7 yrs of disease. The median follow-up time following renal failure for the total cohort was 4.5 yrs (Table 1). Demographics and baseline characteristics All 50 patients had a kidney biopsy performed, which revealed Class IV lupus nephritis (diffuse proliferative lupus nephritis, DPLN) in 68% and Class III lupus nephritis (focal proliferative lupus nephritis, FPLN) in 32%. In the AZA cohort, 79% had Class IV nephritis as compared with 89% of the CYCLO patients (Table 1). Comparison of other baseline demographic characteristics for AZA vs CYCLO cohort revealed no significant differences except for presence of CNS disease: 56% of the CYCLO cohort had concurrent CNS disease as compared with 21% in the AZA cohort (P0.02). The AZA and CYCLO comparison revealed that both cohorts had similar markers of disease activity and severity as measured by SLEDAI score, rate of children requiring dialysis and percentage of patients with hypertension prior to initiation of treatment (Table 1). Overall, 12% of patients required dialysis at the time of diagnosis of ARF. Hypertension was present in 34% of the total ARF cohort prior to treatment (Table 1). Laboratory parameters. At the diagnosis of ARF all patients had a high ESR and low C3 complement levels. Anaemia was found in 100%, leucopenia in 57% and thrombocytopenia in 22% of patients at the time of diagnosis of ARF. Positive anti-DNA antibodies were seen in 97%. There was no statistically significant difference comparing the frequencies of all laboratory baseline parameters between AZA and CYCLO cohorts (Fig. 1). Outcomes (i) Primary outcome Renal function: (1) overall renal survival: overall renal survival of paediatric lupus nephritis patients with ARF was excellent at 92% and similar for the AZA [97%, one death, patient with ESRD] and CYCLO cohort (89%, one patient with ESRD) (P0.38, NS). Two patients (25%) developed ESRD in the STE cohort. Six patients required dialysis at the time of presentation with ARF (three in the AZA cohort, one in the CYCLO cohort and two in the STE cohort). Following treatment, all patients were able to stop dialysis at a mean of 7.7 months (range 313 months). (2) Renal function parameters: normal serum creatinine at last follow-up visit was seen in 43 patients (86%) overall. There was no statically significant differ- ence between AZA patients (91%) and CYCLO patients (78%, P0.28, NS). In addition, there was no difference in time to response for all other renal function markers. At 6 months, overall only 13 patients (26%) continued to demonstrate protei- nuria (27% of AZA and 33% of CYCLO patients). Comparison of renal function parameters is shown in Fig. 1. (3) Renal flares: identical flare rates of 45.4 and 44% were seen for the AZA and CYCLO cohorts. All patients in the STE cohort experienced a renal flare. The median time to flare for the STE cohort was 12.2 months. (4) Time to renal flare: there was no statistically significant difference in time to flare analysis comparing AZA vs CYCLO with the median time to flare being 35.8 months for the AZA cohort and 178 S. M. Benseler et al.
b y
g u e s t
o n
J u l y
2 4 ,
2 0 1 4 h t t p : / / r h e u m a t o l o g y . o x f o r d j o u r n a l s . o r g / D o w n l o a d e d
f r o m
19.6 months for CYCLO (P0.15, NS) (Fig. 2). Interestingly, one of the nine CYCLO patients flared while still on treatment with CYCLO, three patients were off CYCLO at the time of renal flare. After adjusting for the potential confounding variables using the propensity score, the matched time to flare analysis suggested that AZA patients were less likely to flare (risk ratio 0.27; 95% CI 0.038, 0.92; P0.03). (ii) Secondary outcomes (a) Disease activity: (1) Inflammatory markers: comparison of change in inflammatory markers and lupus serology did not reveal a statistically significant difference between AZA and CYCLO cohorts (Fig. 1). Inflammatory markers improved in all treatment cohorts. At 6 months, the ESR remained elevated in 41% of AZA and 44% of CYCLO patients. There was a trend for faster improvement of C3 complement levels in the AZA group. Anti-DNA antibodies remained positive in 68% of AZA patients and 44% of CYCLO patients at 6 months. (2) Disease activity scores: the mean disease activity over time, as measured by mean SLEDAI per day, was similar for all cohorts: 4.0 for AZA, 3.9 for CYCLO and 4.1 for STE. The SLEDAI disease activity score at last follow-up was similar for AZA and CYCLO cohorts at 2.0 and 1.4, respectively. The STE cohort maintained a slightly higher mean SLEDAI of 4.2. (3) Corticosteroids: the mean annual prednisone doses were not statistically significantly different between the groups at 11.8 g for the AZA cohort and 10.8 g for the CYCLO. However, patients in the STE cohort received an average of 16.1 g of prednisone per year. All results are summarized in Table 2. (b) Disease damage Overall disease-related damage, as measured by the SLICC-DI, at last follow-up was similar in all cohorts (0.9 for AZA, 1.8 for CYCLO and 1.5 for STE cohort). (1) Overall renal damage including decreased glomer- ular filtration rate <50%, increased proteinuria on TABLE 1. Characteristics of pSLE patients presenting in ARF ARF study cohort AZA cohort CYCLO cohort STE cohort Number of patients 50 33 9 8 Sex; male : female 13 : 37 10: 23 2: 7 1: 7 Age at diagnosis of pSLE, median (range), yrs 13.4 (3.119.3) 13.0 (3.119.3) 15.1 (9.617.1) 11.9 (6.216.6) Median age at presentation ARF, yrs 13.5 13.2 15.6 13.3 Interval pSLE diagnosis to RF, yrs 011.6 011.6 01.9 04.7 Mean follow-up, months 56.4 47.6 39.6 72.5 Biopsy Class III lupus nephritis a 16 (32%) 9 (21%) 1 (11%) 6 (75%) Biopsy Class IV lupus nephritis a 34 (68%) 24 (79%) 8 (89%) 2 (25%) CNS involvement 17 (34%) 7 (21%) 5 (56%) 5 (63%) SLEDAI at time of RF (SLEDAI range) 20.8 (740) 17.5 (834) 22.9 (1240) 24.9 (734) Dialysis, n (%) 6 (12) 3 (9) 1 (11) 2 (25) Hypertension, n (%) 17 (34) 12 (35) 3 (33) 2 (25) a Histology according to WHO Lupus Nephritis Classification [34]. Months followed 0 12 24 36 48 60 72 84 96 No. at risk 42 39 35 31 29 29 29 29 29 AZA 33 31 29 26 24 23 23 23 23 CYCLO 9 8 6 5 5 5 5 5 5 F l a r e - f r e e
s u r v i v a l
r a t e FIG. 2. KaplanMeier survival analysis of time to renal flare of pSLE patients presenting in renal failure: comparison of AZA vs CYCLO treatment cohorts. Flare- free survival rate is shown on the y-axis and time (in months) on the x-axis. The analysis, adjusted for potential unobserved bias after performing propensity score methods, did not reveal any statistically significant difference in time to renal flare between AZA and CYCLO treatment cohorts. 0 10 20 30 40 50 60 70 80 90 100 RF diagnosis 3 months 6 months last follow-up P e r c e n t a g e
o f
p a t i e n t s P=0.56 P=1.0 P=0.69 P=0.70 P=0.17 P=0.38 P=1.0 P=0.46 P=0.40 % patients with proteinuria AZA % patients with proteinuria CYCLO % patients with elevated ESR AZA % patients with elevated ESR CYCLO % patients with low C3 complement AZA % patients with low C3 complement CYCLO FIG. 1. Resolution of proteinuria, elevated ESR and C3 hypocomplementaemia in the AZA and CYCLO cohorts. Time in months is shown on the x-axis and percentage of patients on the y-axis. Similar response rates for resolution of proteinuria, elevated ESR and C3 hypocomplementaemia were seen for AZA and CYCLO cohorts. Outcome of acute renal failure in pSLE 179
b y
g u e s t
o n
J u l y
2 4 ,
2 0 1 4 h t t p : / / r h e u m a t o l o g y . o x f o r d j o u r n a l s . o r g / D o w n l o a d e d
f r o m
24 h (53.5 g/24 h) or ESRD were seen in 11% of CYCLO and 6% of AZA patients with one patient each in the AZA and CYCLO cohort developing ESRD. Only one patient in each cohort had severe persistent proteinuria (>3.5 g/day). (2) Bone damage was the most common SLICC-DI item detected with AVN in 18% of the total cohort, recurrent AVN in 10% and osteoporosis with vertebral fracture in 12% of patients. The two cohorts had similar rates for all three of these markers of bone damage. (3) Eye damage: cataracts were present in 29% of all of the patients with similar rates of 26 and 33% for the AZA and CYCLO cohorts, respectively (Table 3). (c) Treatment safety None of the patients developed IDDM, malignancies or premature gonadal failure. Significant infections, including herpes zoster and pnaeumonia, were seen overall in nine patients. These were present in 12% of AZA and 23% of CYCLO patients (difference between the AZA and CYCLO cohort was not statistically significant) and in 38% of STE patients (Table 3). Discussion This is the first comprehensive study analysing the outcome of paediatric lupus patients in ARF. We found that the overall renal survival is in fact excellent at 92%. When comparing CYCLO pulse therapy and oral AZA treatment both therapies were equally efficacious. The outcome of both cohorts was similar with regards to renal survival, renal flare rate, control of disease activity over time, cumulative steroid doses and damage. The number of paediatric lupus nephritis studies is limited. There is significantly more information available regarding adult lupus nephritis treatment and outcome. However, one criticism of reports to date is that the majority of patients enrolled in lupus nephritis trials did not, on the whole, have significant impairment of renal function [1417]. Our study has analysed and compared the outcome of two different immunosuppressive treatments for paediatric patients with Class III or IV lupus nephritis who developed renal failure. We found that CYCLO pulse therapy and oral AZA treatment were equally efficacious for treatment of renal failure in pSLE. The outcome of both cohorts was equal with regards to overall renal survival, renal flare rate, control of disease activity over time, cumulative steroid doses and evidence of damage. Comparison of treatment safety in both cohorts did not reveal a statistically significant difference; however, overall there was a trend for an increased rate of infection in the group that received CYCLO. The treatment of severe paediatric lupus nephritis is contro- versial. Treatment decisions for lupus nephritis are generally directed by the histological classification on the kidney biopsy, as proliferative glomerulonephritis has been shown to be associated with the highest risk of renal failure and ESRD [8, 12, 26]. Most paediatric and adult centres have treated patients with Class III or IV nephritis with a 3-yr course of intravenous pulse CYCLO [10, 17]. Recent publications suggest that an induction/ maintenance protocol of CYCLO for 36 months followed by AZA or mycophenolate mofetil for maintenance, lower doses of intravenous CYCLO and mycophenolate mofetil initiation therapy may all be as efficacious and safer than the long-term cyclophosphamide protocol [9, 2628]. Both uncontrolled paedia- tric studies and a recent meta-analysis have shown that AZA is efficacious in the treatment of proliferative lupus nephritis [12, 13, 29, 30]. In our centre, corticosteroids and AZA have been the preferred-treatment regimen for severe lupus nephritis with or without renal failure, while CYCLO therapy is predominantly used in children with concurrent CNS disease or AZA intolerance. This practice explains the greater numbers of patients treated with AZA compared with CYCLO in our study cohort. In order to be able to compare both treatment regimens and to overcome the TABLE 3. Damage and treatment utilizing the SLICC-DI and its domains AZA cohort N33 CYCLO cohort N9 STE cohort N8 AZA vs CYCLO Damage score SLICC DI at last follow-up, mean (range) 0.9 (04) 1.8 (05) 1.5 (03) NA Renal damage Patients with proteinuria, n (%) 1 (3) 1 (11) 1 (12.5) 0.38; NS Patients with overall renal damage, n (%) 2 (6) 3 (33) 1 (12.5) 0.06; NS Bone damage Patients with AVN, n (%) 5 (15) 2 (22) 2 (25) 0.62; NS Patients with more than one episode of AVN, n (%) 3 (9) 1 (11) 1 (12.5) 1.00, NS Patients with osteoporosis and vertebral fracture, n (%) 3 (9) 1 (11) 2 (25) 1.00; NS Other damage Patients with cataracts, n (%) 9 (26) 3 (33) 3 (38) 0.69; NS Patients with IDDM, malignancy or gonadal failure, n 0 0 0 NA Treatment safety Patients with significant infections, n (%) 4 (12) 2 (23) 3 (38) 0.59; NS NA: not applicable; NS: not significant. TABLE 2. Treatment efficacy utilizing the SLEDAI and its domains AZA cohort N33 CYCO cohort N9 STE cohort N8 AZA vs CYCLO* P Overall renal survival, n (%) 32 (97) 8 (89) 6 (75) 0.38; NS Normal creatinine at last follow-up, n (%) 30 (91) 7 (78) 6 (75) 0.28; NS Disease activity, AMS/day, mean (range) 4.0 (0.98.3) 3.9 (0.96.6) 4.1 (2.16.4) 0.89; NS Patients with renal flares, n (%) 15 (45.4) 4 (44.0) 8 (100) 1.00; NS Time RF to renal flare in months (range) 35.8 (7.6126.6) 19.6 (11.526.4) 12.2 (5.016.0) 0.15, NS Disease activity (SLEDAI) at last follow-up (range) 2.0 (04) 1.4 (03) 4.2 (020) 0.36; NS Annual steroid dose, g, mean (range) 11.8 (6.218.11) 10.7 (6.216.6) 16.1 (14.823.8) 0.62; NS *Analysis: Students t-test for mean/median age at diagnosis and follow-up; Fishers exact test for gender, NS not significant. 180 S. M. Benseler et al.
b y
g u e s t
o n
J u l y
2 4 ,
2 0 1 4 h t t p : / / r h e u m a t o l o g y . o x f o r d j o u r n a l s . o r g / D o w n l o a d e d
f r o m
confounding by intent, the propensity score method was utilized. After accounting for all potential differences including CNS disease by propensity score analysis, the data suggest that pSLE patients treated with AZA in fact were less likely to flare compared with those treated with CYCLO. Apart from renal flares, we did not find any clinically relevant or statistically significant difference in the outcome of patients treated with AZA as compared with those treated with CYCLO [31, 32]. The overall renal survival was excellent for patients treated with AZA (97%) and for those treated with CYCLO (89%). However, a significant number of patients flared. CYCLO and AZA treatment regimens of pSLE nephritis and ARF had identical renal flare rates of 45%. This flare rate was identical to the flare rate of a large cohort of adult patients with lupus nephritis treated with the standard regimen of intravenous pulse CYCLO [16]. Comparison of the AZA and CYCLO groups showed that disease activity was equally well controlled over time. All patients who required dialysis at presentation of ARF were able to discontinue dialysis and had a normalization of their renal function. Cumulative annual steroid doses were similar in both groups. These findings demonstrate that both treatment regimens were equally effective for the SLE nephritis patients, who required dialysis at time of renal failure presentation. Both treatment regimens were associated with low rates of severe renal damage. In addition, the rates of bone damage, avascular necrosis and osteoporosis with vertebral fracture, and cataracts were comparable in both cohorts. None of the patients in either group developed diabetes mellitus, a malignancy or premature gonadal failure. The most important limitation of this study is the study design. An observational study in contrast to a randomized controlled trial design is commonly affected by various sources of bias including ascertainment bias. However, as demonstrated by DAgostino [33], propensity score methods are a valid and efficacious way to adjusted for unobserved bias including all patient and disease characteristics that could effect treatment decisions. This technique allows the inclusion of all given consecutive patients into the analysis and may therefore reflect reality and diversity of patient care and reality of treatment efficacy and safety. Propensity score analysis defined the differences between the two treatment groups and adjusted the outcome analysis for these differences. The robustness of our data is supported by our finding of no difference between the treat- ment groups using a standardized set of markers of disease activity and damage. A second limitation of the study is the unbalanced treatment groups as only 9 patients received CYCLO while 33 received AZA. This shortcoming was unavoidable, as the standard of care in our centre for proliferative lupus nephritis is AZA rather than CYCLO. Again, propensity score analysis helped overcome this potential problem. The last limitation is that this is a single centre experience from a large tertiary referral centre. However, it is very likely that all paediatric patients with SLE and renal failure within our catchment area will be seen at our centre. The treatment of ARF in this paediatric lupus nephritis cohort was associated with an excellent outcome. AZA and CYCLO in addition to standard corticosteroid therapy were equally effica- cious for the control of disease activity, overall renal outcome and renal survival; prevention of renal flares and the cohorts had similar disease damage scores. We therefore conclude that AZA and CYCLO are effective in paediatric patients with proliferative lupus nephritis complicated by significant acute renal insuffi- ciency. Given the side-effect profiles of the two agents, risk benefit analysis and the propensity score analysis suggest that AZA should be strongly considered for the management of patients with lupus nephritis, even in patients with significant renal impairment requiring dialysis. Disclosure statement: The authors have declared no conflicts of interest. References 1 Gupta KL. Lupus nephritis in children. Indian J Pediatr 1999;66:21523. 2 Balkaran BN, Roberts LA, Ramcharan J. Systemic lupus erythematosus in Trinidadian children. Ann Trop Paediatr 2004;24:2414. 3 Hu W, Liu Z, Shen S et al. Cyclosporine A in treatment of membranous lupus nephropathy. Chin Med J 2003;116:182730. 4 Huong DL, Papo T, Beaufils H et al. Renal involvement in systemic lupus erythematosus. A study of 180 patients from a single center. Medicine 1999;78:14866. 5 Mosca M, Pasquariello A, Tavoni A et al. Predictors of renal outcome in diffuse proliferative glomerulonephritis in systemic lupus erythematosus. Lupus 1997;6:3718. 6 Sumboonnanonda A, Vongjirad A, Suntornpoch V, Laohapand T, Parichatikanond P. Renal pathology and long-term outcome in childhood SLE. J Med Assoc Thai 1998;81:8304. 7 Wang LC, Yang YH, Lu MY, Chiang BL. Retrospective analysis of mortality and morbidity of pediatric systemic lupus erythematosus in the past two decades. J Microbiol Immunol Infect 2003;36:2038. 8 Weening JJ, DAgati VD, Schwartz MM et al. The classification of glomerulonephritis in systemic lupus erythematosus revisited. J Am Soc Nephrol 2004;15:24150. 9 Yee CS, Gordon C, Dostal C et al. EULAR randomised controlled trial of pulse cyclophosphamide and methylprednisolone versus continuous cyclophosphamide and prednisolone followed by azathioprine and prednisolone in lupus nephritis. Ann Rheum Dis 2004;63:5259. 10 Barbano G, Gusmano R, Damasio B et al. Childhood-onset lupus nephritis: a single- center experience of pulse intravenous cyclophosphamide therapy. J Nephrol 2002;15:1239. 11 Vachvanichsanong P, Dissaneewate P, Winn T. Intravenous cyclophosphamide for lupus nephritis in Thai children. Scand J Rheumatol 2004;33:33942. 12 Bogdanovic R, Nikolic V, Pasic S et al. Lupus nephritis in childhood: a review of 53 patients followed at a single center. Pediatr Nephrol 2004;19:3644. 13 Hagelberg S, Lee Y, Bargman J et al. Longterm followup of childhood lupus nephritis. J Rheumatol 2002;29:263542. 14 Chan TM, Tse KC, Tang CS, Mok MY, Li FK. Long-term study of mycophenolate mofetil as continuous induction and maintenance treatment for diffuse proliferative lupus nephritis. J Am Soc Nephrol 2005;16:107684. 15 Ginzler EM, Dooley MA, Aranow C et al. Mycophenolate mofetil or intravenous cyclophosphamide for lupus nephritis. N Engl J Med 2005;353:221928. 16 Illei GG, Takada K, Parkin D et al. Renal flares are common in patients with severe proliferative lupus nephritis treated with pulse immunosuppressive therapy: long-term followup of a cohort of 145 patients participating in randomized controlled studies. Arthritis Rheum 2002;46:9951002. 17 Lehman TJ, Onel K. Intermittent intravenous cyclophosphamide arrests progression of the renal chronicity index in childhood systemic lupus erythematosus. J Pediatr 2000;136:2437. 18 Rosner B, Prineas RJ, Loggie JM, Daniels SR. Blood pressure nomograms for children and adolescents, by height, sex, and age, in the United States. J Pediatr 1993;123:87186. 19 Bombardier C, Gladman DD, Urowitz MB, Caron D, Chang CH. Derivation of the SLEDAI. A disease activity index for lupus patients. The committee on prognosis studies in SLE. Arthritis Rheum 1992;35:63040. 20 Gladman DD, Ibanez D, Urowitz MB. Systemic lupus erythematosus disease activity index 2000. J Rheumatol 2002;29:28891. 21 Brunner HI, Silverman ED, Bombardier C, Feldman BM. European consensus lupus activity measurement is sensitive to change in disease activity in childhood-onset systemic lupus erythematosus. Arthritis Rheum 2003;49:33541. 22 Vitali C, Bencivelli W, Isenberg DA et al. Disease activity in systemic lupus erythematosus: report of the consensus study group of the European workshop for rheumatology research. II. Identification of the variables indicative of disease activity and their use in the development of an activity score. The European consensus study group for disease activity in SLE. Clin Exp Rheumatol 1992;10:5417. 23 Fortin PR, Abrahamowicz M, Clarke AE et al. Do lupus disease activity measures detect clinically important change? J Rheumatol 2000;27:14218. 24 Gladman DD, Goldsmith CH, Urowitz MB et al. The Systemic Lupus International Collaborating Clinics/American College of Rheumatology (SLICC/ACR) Damage Index for Systemic Lupus Erythematosus International Comparison. J Rheumatol 2000;27:3736. 25 Gladman DD, Urowitz MB, Goldsmith CH et al. The reliability of the systemic lupus international collaborating clinics/American college of Rheumatology Damage Index in patients with systemic lupus erythematosus. Arthritis Rheum 1997;40:80913. 26 Niaudet P. Treatment of lupus nephritis in children. Pediatr Nephrol 2000;14:15866. Rheumatology key messages The long-term outcome of children with severe lupus nephritis and renal failure was excellent. AZA and CYCLO were equally efficacious. Similar flare rates of 45% were observed. Outcome of acute renal failure in pSLE 181
b y
g u e s t
o n
J u l y
2 4 ,
2 0 1 4 h t t p : / / r h e u m a t o l o g y . o x f o r d j o u r n a l s . o r g / D o w n l o a d e d
f r o m
27 Contreras G, Pardo V, Leclercq B et al. Sequential therapies for proliferative lupus nephritis. N Engl J Med 2004;350:97180. 28 Houssiau FA, Vasconcelos C, DCruz D et al. Immunosuppressive therapy in lupus nephritis: the Euro-Lupus Nephritis Trial, a randomized trial of low-dose versus high- dose intravenous cyclophosphamide. Arthritis Rheum 2002;46:212131. 29 Font J, Cervera R, Espinosa G et al. Systemic lupus erythematosus (SLE) in childhood: analysis of clinical and immunological findings in 34 patients and comparison with SLE characteristics in adults. Ann Rheum Dis 1998;57:4569. 30 Bansal VK, Beto JA. Treatment of lupus nephritis: a meta-analysis of clinical trials. Am J Kidney Dis 1997;29:1939. 31 Hullsiek KH, Louis TA. Propensity score modeling strategies for the causal analysis of observational data. Biostatistics 2002;3:17993. 32 Newgard CD, Hedges JR, Arthur M, Mullins RJ. Advanced statistics: the propensity scorea method for estimating treatment effect in observational research. Acad Emerg Med 2004;11:95361. 33 DAgostino RB Jr. Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Stat Med 1998;17:226581. 34 Grishman E, Gerber MA, Churg J. Patterns of renal injury in systemic lupus erythematosus: light and immunofluorescence microscopic observations. Am J Kidney Dis 1982;2(Suppl. 1):13541. 182 S. M. Benseler et al.
b y
g u e s t
o n
J u l y
2 4 ,
2 0 1 4 h t t p : / / r h e u m a t o l o g y . o x f o r d j o u r n a l s . o r g / D o w n l o a d e d