Henoch-Schönlein purpura (HSP) is the most common vasculitis in children, with an incidence of 10–20 per 100 000 children.1 2 Clinical manifestations include purpuric skin lesions, abdominal pain, gastrointestinal bleeding, arthropathy and renal involvement. Renal involvement is the most serious long-term complication of HSP. In a systematic review of studies of unselected patients, renal involvement occurred in 34% of children; 80% had isolated haematuria and/or proteinuria, and 20% had acute nephritis or nephrotic syndrome.3 Renal complications, if they did occur, developed early—by 4 weeks in 85% and by 6 months in nearly all children. Persistent renal involvement (hypertension, reduced renal function, nephrotic or nephritic syndrome) occurred in 1.8% of children overall, but the incidence varied with the severity of the kidney disease at presentation. It occurred in 5% of children with isolated haematuria and/or proteinuria, but in 20% who had acute nephritis and/or nephrotic syndrome in the acute phase. In a study of 78 children with HSP and renal involvement presenting to two paediatric nephrology services,4 44% of children presenting with acute nephritis and/or nephrotic syndrome compared with 13% presenting with haematuria and/or proteinuria had hypertension or impaired renal function at a mean follow-up of 23.4 years.

Corticosteroid treatment is commonly used in the acute phase of HSP, particularly for abdominal pain. Controversy remains as to whether corticosteroids can prevent the development of renal involvement and/or reduce its severity in HSP. One systematic review concluded that early corticosteroid treatment may reduce the risk of developing persistent renal abnormalities.5 However, two systematic reviews concluded that the benefit of corticosteroids in preventing persistent renal involvement remained unproven.6 7 Similarly, there remains considerable uncertainty about the efficacy of treatments to prevent progression to chronic or end stage renal failure in children with HSP-associated acute nephritis or nephrotic syndrome. Corticosteroid therapy,8 azathioprine,9 10 cyclophosphamide,11 cyclosporin,12 antiplatelet therapy with anticoagulants13 and plasmapheresis14 have been used in such patients with variable results. The aims of this systematic review were to determine the benefits and harms of different interventions used to prevent or treat persistent renal involvement in HSP. The scope was deliberately broad because randomised trials in HSP are few and variability in the spectrum of renal disease included in the relevant trials was very likely.

METHODS

Literature search

Randomised and quasi-randomised controlled trials (RCTs) were identified from the Cochrane Renal Group’s specialised register (www.cochrane.org), the Cochrane Central Register of Controlled Trials (Cochrane Library Issue 4, 2007), Medline (1966 to November 2007) and EMBASE (1980 to November 2007) without language restriction. The optimally sensitive strategies developed for the Cochrane Collaboration for the identification of RCTs were used for Medline15 and EMBASE16 searches. These search strategies were combined with the following terms: Schoenlein-Henoch purpura, Henoch-Schönlein purpura, anaphylactoid purpura, allergic purpura, Henoch purpura, non-thrombocytopenic purpura, peliosis purpura, purpura rheumatica, leukocytoclastic vasculitis and Schönlein disease. Reference lists of nephrology textbooks, review articles and relevant trials were also searched. The Cochrane Renal Group’s specialised register includes RCTs identified by hand-searching conference proceedings.

Study eligibility

All RCTs of interventions (including corticosteroids, anticoagulants, antiplatelet agents, immunosuppressive agents, angiotensin-converting enzyme inhibitors, plasma exchange) compared with placebo, no specific treatment or another intervention in patients with HSP with or without renal manifestations (haematuria, proteinuria, acute nephritis or nephrotic syndrome) were included. Studies that included other forms of vasculitis were excluded. The primary outcomes sought were the number of children who developed persistent renal involvement or who had a change in existing renal disease (haematuria, proteinuria, hypertension, nephrotic syndrome, acute nephritis, renal dysfunction, need for dialysis or transplantation), mortality and adverse effects of drugs.

Study selection, data extraction and assessment of methodological quality

All titles and abstracts were screened independently by two reviewers (WC and SO), and clearly irrelevant studies were discarded. The full text of the remaining studies was assessed to determine if the inclusion criteria were met. Data on trial characteristics and results were extracted independently from included studies by three reviewers (WC, SO and EH). Methodological quality (randomisation method, allocation concealment, blinding of participants, investigators, outcome assessors and data analysers, intention-to-treat analysis, and completeness of follow-up) of included studies was assessed by the same reviewers, without blinding to author or source. Any discrepancies in methodological quality assessment or in data extraction were resolved by discussion.

Statistical analysis

For dichotomous outcomes describing benefits (preventing the development or progression of renal involvement in children), the estimate of effect was calculated in terms of relative risks (RRs) with 95% CI in Review Manager V4.2.8 for individual studies. For continuous outcomes (duration of haematuria or proteinuria), the weighed mean difference (WMD) with 95% CI was calculated. The summary statistics were calculated using a random effects model, which takes into account the between-study variability as well as the within-study variability. Statistical significance was tested for using the test z statistic. Heterogeneity between studies was analysed by Cochran’s Q statistic, with α = 0.05 for statistical significance, and by the statistic I2, which is derived from Q and describes the proportion of total variance that is due to heterogeneity beyond chance.17 We hypothesised that certain between-study differences in participants (severity of renal disease, renal pathology), interventions (agent, dose and duration of treatment) and trial quality may explain any observed heterogeneity of treatment effects. Examination of these possible between-study differences by subgroup analysis was attempted but was limited by insufficient trial data.

RESULTS

Literature search

Ten trials involving 1230 children were identified from the literature search. Seven trials were identified by full-text review to be RCTs18^–24 from 907 studies obtained from the electronic search. Two RCTs25 26 were identified from reference lists of review articles. The 10th trial was identified from the National Research Register, NHS, UK, and further information on trial methodology and results was obtained directly from the triallists and from a published abstract.27 28 Four trials were available in abstract form only.19 21 22 24 All included trials were published in English. One trial21 compared two different interventions with the control intervention and was treated as two trials, so effectively data from 11 trials were included.

Characteristics of included trials (table 1)

Five trials18–20 25 27 28 (789 children) examined the effects of short-duration corticosteroids (14–28 days) on preventing persistent HSP nephritis at 6–12 months after presentation. Three of these trials18 20 27 28 included children with renal involvement at randomisation. Children considered to have established HSP nephritis (proteinuria exceeding 300 mg/l or haematuria exceeding 10 red blood cells per high power field) were excluded from one trial,20 while the other two trials included children with any severity of renal involvement.18 27 28 Two trials21 26 (138 children) evaluated antiplatelet agents, and one trial22 (228 children) compared heparin with placebo. One trial26 provided outcome data separately for children with and without renal involvement at presentation, whereas the other trials only included children without renal involvement at randomisation. Two trials examined the treatment of severe HSP nephritis (nephrotic range proteinuria, International Study of Kidney Disease in Children grade III–IV changes on biopsy): one trial23 (56 children) compared cyclophosphamide with no specific treatment and one trial24 (19 children) compared cyclosporin with methylprednisolone. One trial27 28 used the urinary protein/creatinine ratio as the primary end point, with dipstick proteinuria and haematuria as secondary end points. In the remaining trials, the primary end point of renal involvement was defined by a composite of haematuria, proteinuria, hypertension and reduced renal function.

Quality of design and reporting

Four trials18 20 23 27 28 reported adequate allocation concealment, and three18 20 27 28 of these also reported a computer-generated random allocation sequence (table 2). Allocation concealment was inadequate in one study25 and was unclear in the remaining studies. Blinding of participants, investigators and outcome assessors occurred in three trials,18 20 27 28 and blinding of participants was reported in one trial.22 No trial reported an intention-to-treat analysis. Loss to follow-up ranged from 0% to 10%.

Effects of corticosteroid treatment

In meta-analyses of four of the five RCTs that evaluated prednisone therapy at presentation of HSP, there was no significant difference in the risk of development or persistence of renal involvement at 1, 3, 6 and 12 months (fig 1) with prednisone compared with placebo or no specific treatment. There was substantial variability in trial outcomes at all time points, except at 6 months, which was largely due to one trial.25 This trial, which had inadequate allocation concealment and was therefore at high risk of bias,29 showed a large benefit of prednisone in contrast with the other three studies. Exclusion of this trial eliminated the heterogeneity except at 1 month. The fifth trial19 evaluating prednisone therapy was not included in the meta-analyses as it did not report the time at which the end point was measured. This trial found no significant difference in the risk of developing renal involvement with prednisone compared with no specific treatment (RR 0.60, 95% CI 0.33 to 1.06) and no significant difference in the duration of haematuria (WMD −1.00, 95% CI −10.26 to +8.26) or proteinuria (WMD −1.60, 95% CI −15.62 to +12.42). In one trial,20 post hoc subgroup analysis of 71 children with renal involvement at or within 1 month of randomisation found that renal involvement was significantly less common at 6 months after prednisone therapy compared with placebo (RR 0.45, 95% CI 0.21 to 0.98). No adverse effects related to prednisone were reported in three trials18 20 27; adverse effects were not discussed in the other two trials.19 25

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Meta-analyses of the relative risk (95% CI) for the presence of persistent renal involvement in children with Henoch-Schönlein purpura at 1, 3, 6 and 12 months after short-course prednisone therapy compared with placebo or supportive therapy. Results are shown ordered by trial weights. The tests for overall effect (z statistic) indicate no evidence of benefit of prednisone over placebo or supportive therapy. There is significant heterogeneity, as indicated by the tests for heterogeneity (χ²and I2). This could be eliminated by excluding the trial by Mollica et al.25

Effects of antiplatelet agents or heparin

In children treated with antiplatelet agents (dipyridamole with/without cyproheptadine, aspirin) or no specific therapies, there was no significant difference in the risk of renal disease during follow-up in children without (two trials; 101 children; RR 1.16, 95% CI 0.46 to 2.95)21 26 or with (one trial; 19 children; RR 0.92, 95% CI 0.23 to 3.72)26 renal involvement at presentation. In one trial, there was no significant difference in the risk of renal involvement with aspirin compared with no specific treatment (one trial; 18 children; RR 0.14; 95% CI 0.01 to 2.42)21; duration of follow-up was not reported. Adverse effects were not reported in these trials. Heparin significantly reduced the risk of renal involvement compared with no treatment (one trial; 228 patients; RR 0.27, 95% CI 0.14 to 0.55) at 3 months or more after onset or relapse of HSP.22 No child developed severe bleeding.

Effect of treatments in severe HSP nephritis

In a single trial of 56 children with severe renal disease, there was no significant difference in the risk of persistent renal disease (RR 1.07, 95% CI 0.65 to 1.78), severe renal disease (nephrotic range proteinuria, reduced renal function) (RR 0.88, 95% CI 0.37 to 2.09) or end stage renal failure (RR 0.75, 95% CI 0.18 to 3.05) with cyclophosphamide treatment compared with no specific treatment.23 In children with severe renal disease, seven of 10 children treated with cyclosporin were in remission at final follow-up compared with two of nine treated with methylprednisolone. However, the difference was not significant (one trial; 19 children; RR 0.39, 95% CI 0.14 to 1.06).24 Neither trial reported on adverse effects.

DISCUSSION

This systematic review of RCTs identified 10 trials, of which eight examined the efficacy of treatments to prevent persistent renal disease in HSP and two examined treatments for established severe renal disease. Meta-analyses of four RCTs revealed no significant difference in the number of children with persistent renal disease at 3, 6 and 12 months after short-term prednisone therapy at presentation of HSP compared with placebo or no specific treatment. Sensitivity analysis, in which the trial with a high risk of bias was excluded, removed heterogeneity between studies and tended to make the summary result less favourable towards corticosteroids. Studies with a high risk of bias (inadequate or unclear allocation concealment, no blinding of participants, investigators or outcome assessors) are associated with an increased likelihood of results favouring the trial intervention.29^–31 Thus three well-designed placebo-controlled RCTs18 20 27 28 have demonstrated no significant benefit of short-term prednisone therapy in preventing persistent renal disease in children with HSP. The three trials included children with and without renal involvement at presentation. In the largest of these trials,27 28 which enrolled children between January 2001 and January 2005, the primary outcome (urinary protein/creatinine ratio at 1 year) was measured in 290 children. This is the largest study to date showing no significant benefit of prednisone over placebo in preventing persistent renal disease. Preliminary results27 have been presented, but full publication is awaited. Data from these three trials indicate that there remains considerable residual imprecision in the results at 6 months (RR 0.59, 95% CI 0.23 to 1.50) and 1 year (RR 1.28, 95% CI 0.70 to 2.62) as indicated by wide confidence intervals. Thus at 1 year after presentation, prednisone could reduce the risk of kidney disease or increase the risk by over 2.5-fold. In addition, the potential significance for long-term renal function of residual urinary abnormalities could not be assessed in three of four trials, which reported the end point as the presence of haematuria and/or proteinuria without any measure of the degree of proteinuria.

One trial20 presented a post hoc analysis of 71 children, who had renal involvement at or within 1 month of presentation. Prednisone therapy for 28 days significantly reduced the number of children with persistent renal disease at 6 months. The trial was not stratified before randomisation for the presence or absence of renal involvement, and the sample size was small so the results can only be considered as hypothesis-generating. As the trial only provided outcome data to 6 months after randomisation, it is unclear whether prednisone treatment reduced the number of patients with persistent HSP nephritis overall or promoted more rapid resolution of renal disease compared with placebo. In addition, children considered to have established nephritis at randomisation were not included in this trial. In the other well-designed trials of prednisone therapy,18 27 28 children with any severity of renal involvement at presentation were potentially included, and a meta-analysis of these trials showed no significant difference in the risk for persistent HSP nephritis at 12 months. A further trial of short-term prednisone therapy compared with placebo should be considered in children who have or develop renal disease with HSP or have risk factors for developing renal disease (older age,32 severe abdominal pain,20 32 persistent20 32 or recurrent32 purpura). However, recruitment may be difficult, as it has been shown in an RCT20 that short-course prednisone significantly reduces the severity and duration of abdominal pain in children with HSP, making it unlikely that clinicians will be prepared to withhold prednisone from children with severe HSP-associated abdominal pain.

Three systematic reviews have previously assessed the effects of corticosteroid therapy in preventing or altering the course of renal disease in HSP.5^–7 All three included data from RCTs and case series; conclusions based on non-randomised study designs are more likely to be biased towards a benefit of treatment.33 Two reviews6 7 determined that it remained unclear whether corticosteroid therapy prevented or altered the course of HSP nephritis. The third review5 concluded that corticosteroids decreased the likelihood of developing persistent renal disease but did not prevent renal involvement. This conclusion was based on a meta-analysis of the ad hoc subgroup of children with renal abnormalities within 1 month of presentation from the Ronkainen trial20 combined with data from the trials of Huber et al18 and Mollica et al25. The inclusion of the most recent large trial,27 28 which showed no benefit of corticosteroid treatment, in the current systematic review increases the evidence base from RCTs to support the conclusion that corticosteroids do not prevent or alter the course of renal involvement in HSP.

No significant benefit of antiplatelet agents was shown in two small trials,21 26 suggesting that these agents have no role in preventing renal involvement in HSP. One trial,22 available in abstract only, showed that heparin reduced the number of children with renal involvement. No details on trial methodology were available. Heparin or placebo was administered to children at the onset of disease and at relapse and it was not possible to determine how many children in each group received more than one period of treatment. Although bleeding was not reported in this study, the use of such a potentially dangerous treatment is not justified when only about a third of children with HSP develop renal involvement and the majority of those recover completely.

Only two RCTs that examined the treatment of severe established renal disease in HSP were identified. In one trial23 in children with HSP and nephrotic-range proteinuria, no significant benefit of cyclophosphamide alone compared with no specific treatment could be demonstrated. A small trial24 suggested that cyclosporin may be more effective than methylprednisolone and prednisone in inducing remission in children with HSP and nephrotic-range proteinuria, but numbers were too small to achieve statistical significance. No trials examining intravenous methylprednisolone alone or in combination with cyclophosphamide, azathioprine or plasma exchange were identified.

Adequately powered well-designed RCTs with at least 5-year follow-up periods are particularly needed in children with HSP-associated nephritic syndrome and/or nephrotic syndrome. Therefore a multicentre trial comparing, for example, a 6-month course of corticosteroids with placebo or short-duration corticosteroids (28 days) in children with moderately severe renal disease (acute nephritis or nephrotic syndrome with normal renal function and less than 50% crescents or sclerosing lesions on biopsy) should be considered. Clinicians are likely to be reluctant to enter children with HSP and crescentic glomerulonephritis involving more than 50% of glomeruli into RCTs with a placebo arm even though currently no treatment has been shown to be effective in an RCT in such children. On the basis of data from a trial24 comparing cyclosporin and methylprednisolone, a multicentre trial should be set up to evaluate further the role of cyclosporin in treating children with HSP and severe nephritis.