Summary of main results

We identified 13 studies of which eight examined the efficacy of therapies to prevent persistent kidney disease in HSP and five examined therapies to treat established severe kidney disease.

Overall completeness and applicability of evidence

Thirteen studies enrolling 1403 participants were included in this review update; four studies only available in the abstract form and several studies included small numbers of patients with incomplete outcome data. Incomplete reporting of these studies may result in incomplete information being included in this systematic review.

Three well‐designed, placebo controlled studies (Dudley 2013; Huber 2004; Ronkainen 2006a) have provided data in over 400 children that there is no significant benefit of prednisone therapy at six to 12 months in children with no or minor kidney involvement at presentation. In addition, eight year follow‐up data in Ronkainen 2006a found no longer term benefit. Therefore further RCTs to evaluate prednisone to prevent kidney disease in this group of children presenting with HSP are unlikely to be justified.

Based on observational studies (Niaudet 1998), intravenous methylprednisolone followed by oral prednisone is commonly used in patients with nephrotic range proteinuria and no reduction in kidney function. However there are no RCTs evaluating methylprednisolone and prednisone compared with placebo or no treatment in patients with nephrotic range proteinuria and normal kidney function at or soon after presentation. Because of the known risk of permanent kidney damage and reduced kidney function, it is unlikely that investigators would be prepared to enrol their patients if such a study was proposed.

At present there are few RCTs with small patient numbers evaluating the treatment of HSPN with immunosuppressive medication to determine the benefits and harms of such treatment for patients with nephrotic syndrome with or without reduced kidney function and with or without more than 50% crescents on biopsy. Nevertheless immunosuppressive agents are commonly used in an attempt to treat established severe kidney disease in HSP with observational studies suggesting benefit of methylprednisolone (Niaudet 1998), azathioprine (Bergstein 1998), mycophenolate mofetil (Du 2012), cyclophosphamide (Flynn 2001; Iijima 1998; Tanaka 2003) and cyclosporin (Ronkainen 2003). However in the four RCTs included in this review, no significant benefits of cyclophosphamide, cyclosporin or mycophenolate mofetil were identified (CESAR Study 2010; Fuentes 2010; Jauhola 2010; Tarshish 2004). Patient numbers in the studies of cyclosporin and mycophenolate were very small resulting in considerable imprecision of results so these treatment modalities need further study. Cyclophosphamide is recommended for patients with crescentic GN (KDIGO 2012). The lack of significant benefit of cyclophosphamide may be related to the types of patients included in the two studies. Although all children included in the Tarshish 2004 study had crescentic GN, the delay between onset of disease and treatment (at least three months) may have allowed the development of chronic changes, which were not amenable to therapy. In addition treatment may have been more effective if methylprednisolone and then oral high dose prednisone had been administered also. In the CESAR Study 2010 only 30% of participants had crescentic disease, which may have influenced why there was no significant additional benefit of cyclophosphamide over methylprednisolone and prednisone alone. However while these issues might suggest that further studies of cyclophosphamide in patients with crescentic GN associated with HSP are indicated, the potential long term adverse effects of cyclophosphamide particularly in children (Coutinho 2001) mean that such studies are unlikely to be performed.

Adverse events of therapies were not all well reported. Limited reporting revealed small numbers of adverse events in all studies with no significant difference between interventions. No studies were identified which evaluated therapy to prevent or treat persistent kidney disease in participants with recurrent episodes of HSP.

In the studies evaluating prednisone therapy, the outcomes reported were poorly defined except for one study (Dudley 2013), which used UPC ratio as the primary outcome. The potential significance for long‐term kidney function of any residual urinary abnormalities could not be assessed in the other studies, since they reported the end point as the presence of haematuria or proteinuria or both without measuring the degree of proteinuria.

No RCTs were identified which examined intravenous immunoglobulin, rituximab or plasma exchange. No studies examining fish oil, ACEi or ARB were identified. No studies specifically addressing whether therapy reduced the risk of recurrent episodes of HSP were identified.

Quality of the evidence

Sequence generation and allocation concealment were at low risk of bias in six (50%) and seven (58%) respectively. This may be attributed to poorer reporting of these parameters in the earlier studies. Blinding of participants, investigators and outcome assessors was only reported in three studies, reflecting a high risk of bias in the remaining studies since knowledge of treatment groups could influence patient management and reporting. Only five studies were at low risk of attrition bias. The otherwise robust study of Dudley 2013, was at high risk of attrition bias due to a significant drop out rate with only 72% (123/171) reporting the primary outcome. Eight studies were at low risk of reporting bias. Studies with a high risk of bias are associated with an increased likelihood of results favouring the study intervention (Schulz 1995; Wood 2008). Exclusion of the study with a high risk of bias removed heterogeneity between studies without altering the overall result, reinforcing the strength of the evidence suggesting that short courses of prednisone do not prevent serious kidney disease in children with HSP.

Only the five studies comparing prednisone with placebo or supportive treatment for the prevention of persistent kidney disease in HSP could be assessed in a summary of findings table (summary of findings Table for the main comparison). The overall quality of the studies was considered moderate for persistence of kidney disease at any time after treatment and for the number of children with continuing kidney disease at varying time points. With the removal of one study, which had inadequate allocation concealment and no blinding, the remaining three studies were graded as of high quality at the three month interval but only moderate at 12 months because of the small number of events and high loss to follow‐up in Dudley 2013. The number developing severe kidney disease was graded as low in two studies, as a result of significant loss to follow‐up in the largest included study and small numbers of events.

The remaining eight studies could not be included into summary of findings tables as they were single studies with small numbers or data could not be included in meta‐analyses.

Potential biases in the review process

A thorough search utilising Cochrane Kidney and Transplant's Specialised Register was completed in July 2015.

The Specialised Register includes published studies and conference abstracts with no restriction of language. The omission of eligible studies was therefore minimised, although potentially eligible studies identified prior to publication have yet to be included (Ding 2014; NCT00301613; Wu 2013b; Wu 2014c). However 40% of study reports in the Specialised Register have been identified by handsearching of conference proceedings so it remains possible that further studies of therapy to prevent or treat serious kidney disease in HSP will be identified as conference proceedings from different congresses are searched.

Four (33%) of the included studies were only available in the abstract form, thus limiting information on study methods and outcomes. Four of the studies were published prior to 2000 before the CONSORT checklist (first published in 1996) would influence trial methodology and reporting. Incomplete reporting of these studies may result in incomplete information being included in this systematic review.

Two authors independently undertook all the steps of this review thereby minimising risks of errors in determining study eligibility, data extraction and risk of bias assessment and data synthesis.

Agreements and disagreements with other studies or reviews

Three systematic reviews have previously assessed the effects of corticosteroid therapy to prevent or alter the course of kidney disease in HSP (Weiss 2007; Wyatt 2001; Zaffanello 2007). All three included data from RCTs and observational studies. Conclusions based on non‐randomised study designs are more likely to be biased towards a benefit of treatment (Chalmers 1983). Two reviews determined that it remained unclear whether corticosteroid therapy prevented or altered the course of HSP‐associated kidney disease (Wyatt 2001; Zaffanello 2007). The third review concluded that corticosteroids decreased the likelihood of developing persistent kidney disease but did not prevent kidney disease (Weiss 2007). This conclusion was based on a meta‐analysis of the ad hoc subgroup of children with kidney abnormalities within one month of presentation from Ronkainen 2006a combined with data from two other studies (Huber 2004; Mollica 1992). With the inclusion in this review of another large study (Dudley 2013), which showed no significant benefit of corticosteroid therapy, the evidence base from RCTs supports the conclusion that corticosteroids given at initial presentation of HSP do not prevent the development or persistence of kidney disease in HSP.

One review also evaluated immunosuppressive and other therapies in HSP (Zaffanello 2007). It concluded based on observational studies that cyclophosphamide was of value in treating HSP‐associated kidney disease. The two studies evaluating cyclophosphamide in this review did not show any benefit of cyclophosphamide (CESAR Study 2010, Tarshish 2004).The studies comparing cyclosporin with methylprednisolone and MMF with azathioprine were too small to establish whether or not these treatments were effective (Fuentes 2010; Jauhola 2010).

The KDIGO guidelines (KDIGO 2012) on HSP recommend treatment with ACEi or ARB in patients with HSPN and persistent low grade proteinuria and recommend corticosteroids after a trial of ACEi or ARB in patients with larger amounts of proteinuria as in patients with IgA nephropathy. However these recommendations have been challenged on the basis that the natural history of HSPN is different from IgA nephropathy with the risk of early severe inflammatory episodes with glomerular crescents followed by a rapid evolution to glomerulosclerosis. Thus delayed treatment may increase the risk of CKD progression so early treatment with methylprednisolone pulses with immunosuppressive drugs, plasma exchange or both is commonly commenced based on observational data. However currently there are no data from RCTs to support or refute this alternative management (Davin 2013).