To review event-free (EFS) and overall survival (OS) from publications describing outcome for children with relapsed Wilms' tumour. Comparisons are made between those receiving myeloablative high dose chemotherapy with autologous stem-cell rescue (HDT) and those not (NoHDT).

Relevant information was extracted from individual patient or summary data and 3-year EFS and OS rates established. These rates were combined in a weighted manner to derive hazard ratios (HRs).

Nineteen publications were identified (5 HDT, 6 NoHDT, 8 both). Pooling all studies suggested an advantage to HDT with a hazard ratio (HR) for EFS of 0.87 (95% confidence interval (CI) 0.67-1.12) and 0.94 (0.71-1.24) for OS. A stratified analysis confined to studies that provided individual patient data on both HDT and NoHDT gave HRs of 0.83 (0.56-1.24) and 0.92 (0.59-1.41). Further, analyses of risk groups, defined by treatment and/or histology prior to first relapse, suggested a HR for EFS of 0.90 (95% CI 0.62-1.31) for those of high and 0.50 (CI 0.31-0.82) for the very high risk patients.

The evidence suggests, although there are many caveats since the information summarised here is not from randomised trials, a great deal of uncertainty concerning the role of HDT in patients following relapse after treatment for their Wilms' tumour. For each risk group we propose a randomised trial comparing a standard with a more intensive therapy with specific choice of regimen tailored to the risk group (and co-operative groups) concerned. A synthesis of updated evidence from studies in this overview together with any emerging studies and future trial information will form the basis for future evidence-based clinical decision-making.