It has been known since the pre-chemotherapy era that a minority of Wilms tumours can be cured by nephrectomy alone 1. With the introduction of adjuvant chemotherapy and sometime radiotherapy, survival rates approaching 90% were achieved for localised tumours by the 1970s. Since that time, all trial groups have pursued a strategy of decreasing therapy for stage I tumours that lack high risk features, successfully omitting radiotherapy, then reducing the duration of chemotherapy from a standard 15 months with vincristine-actinomycin D (VA) to as little as 8–10 weeks with VA or vincristine alone 2,3. These excellent survival rates beg the question of how much the chemotherapy is contributing, given the improvements in diagnostic imaging for pre-operative staging, in surgery, supportive care and histological classification. During the 1990s, the fifth trial of the National Wilms Tumor Study Group took the next step of selecting a subgroup of patients with stage I, favourable histology, Wilms tumours for reintroduction of a nephrectomy-only treatment approach 4. This treatment arm was closed prematurely as the stringent stopping rule of no less than 90% relapse free survival was breached. However, the excellent overall survival rate led to reintroduction of the nephrectomy-only strategy in the current Children's Oncology Group trial, but with mandatory multidisciplinary scrutiny of tumour staging by radiologists, surgeons and pathologists (see Table I in Frazier et al. 5). The article by Frazier et al. in this issue of Pediatric Blood & Cancer, uses the theoretical approach of ‘decision tree analysis’ to compare three clinically used treatment regimens for their respective long term outcomes in relation to success of first line therapy for children with very low risk Wilms tumour treated with upfront nephrectomy 5. This carefully argued analysis illustrates a pragmatic approach to comparing treatment options in childhood cancer, especially for patient populations with relatively good outcomes on standard regimens whose short and long term toxicities are well documented. However, such an approach can only inform but never replace careful clinical evaluation of the recommended strategy in real life patients, where the range of variables and what is acceptable to a local or national population may be far outside the parameters set in the theoretical analysis. The conclusion reached by Frazier et al. is that nephrectomy-only is an acceptable strategy for small (< 550 g), favourable histology Wilms tumours in young children (aged < 2 years) in terms of overall survival, recognising that this approach may carry a small increased risk of long term side effects due to the increased proportion exposed to relapse therapy. This calculated risk is based on the assumption that all relapses would be retreated with radiotherapy as well as more toxic drugs than the alternative upfront use of actinomycin D and/or vincristine. The proportion exposed to radiotherapy might be reduced if the response of chemo-naïve relapsed tumour could be factored into the equation. This would mirror the philosophy for treatment of metastatic disease long used by SIOP investigators and more recently adopted by COG. Another aspect not included in the current analysis is the risk of renal insufficiency, which has a major negative impact on quality of life and survival. This can be due to either further surgery for metachronous relapse or underlying genetic predisposition, both of which are enriched in the young population selected for this treatment approach 6,7. Although Frazier et al. do not favour vincristine only, such an approach was very effective in young patients treated with immediate nephrectomy in the United Kingdom national trials over two decades 3. Indeed, in this analysis, tumour size was not an independent prognostic factor and the subgroup of young children aged under two years with the larger tumours (> 550 g), had an event free survival of 95%. This is at the upper limit that would favour this approach in the sensitivity analysis of Frazier et al. However, their model also suggested that the salvage rate for patients relapsing after vincristine only would have to increase to an unlikely 97% for nephrectomy only to lose its top ranking for overall survival. Such discussions illustrate the value of the decision tree sensitivity analysis to highlight which aspects of therapy have the greatest theoretical impact on outcomes. Certainly, vincristine monotherapy should be considered a perfectly reasonable option and either this, or the vincristine-actinomycin D regimen, would be preferable in situations where rigorous and timely review of imaging, pathology and surgery are not practised. Treating any Wilms tumour patient with nephrectomy alone remains an experimental approach to be performed only within the strict confines of a clinical trial. Parents must be fully informed of the balance of risks and feel comfortable with living with a slightly increased risk of relapse. The current parameters used to select patients for this approach take no account of the underlying tumour biology. Recent gene expression profiling and allele loss analyses suggest that there are at least three distinct biological subgroups among the ‘tiny tumours’ that may predict the relapse risk even more precisely 8. Those with epithelial differentiated tubular histology, have a distinctive gene expression signature, no allele loss at 1p, 16q or 11p and none of nine tumours relapsed. By contrast, a second larger subgroup showed mixed histology, an association with nephrogenic rests and decreased WT1 expression and allele loss on 11p: 3 of 13 cases relapsed. As allele loss on 11p is not associated with adverse outcome more generally, this implies that tumours with 11p/WT1 abnormalities may be at higher risk of relapse when treated with minimal therapy, including surgery only. Given that the median age at diagnosis of WT1-associated Wilms tumours is less than two years of age 9, this is an important subgroup to recognise as they are probably disadvantaged by avoidance of chemotherapy. Of similar concern, it is known that the presence of nephrogenic rests in children younger than 12 months is associated with an increased risk of metachronous relapse 6. Furthermore, the same analysis reported that the cumulative incidence of contralateral relapse decreased between the first and subsequent National Wilms Tumour Study Group trials, suggesting a beneficial effect of chemotherapy on the evolution of rest into tumour. Preventing metachronous relapse is an extremely important factor in determining quality of long term survival. Ultimately, biological characterisation of tumours is likely to improve the basis on which the decision to withhold chemotherapy is made. This should identify additional groups of tumours that may be considered for further therapy reduction and enhance or even replace the current age and tumour size criteria. The nephrectomy only approach of the COG cannot take into account the additional risk factor of histological response to chemotherapy that is used by the SIOP investigators for treatment stratification. The SIOP high risk ‘blastemal type’ Wilms tumour does occur in children aged less than 2 years, at a similar frequency to that seen in older children. On the other hand, the current age threshold of up to only 6 months for SIOP investigators to recommend immediate nephrectomy is based solely on minimising the risk of giving chemotherapy to a child with a benign tumour prior to histological confirmation. One can therefore imagine a future where a decision analysis approach, such as used by Frazier et al., could be applied to pull out the best from the experiences of both groups. By making explicit the pros and cons of each approach within defined scenarios, this should bring an end to the age old controversy of which patients benefit most from pre-operative chemotherapy versus immediate nephrectomy. Such an analysis of the complex clinical and biological features that make up Wilms tumour requires carefully orchestrated input from not only childhood cancer specialists in all disciplines, but also biologists, geneticists, nephrologists, long term follow up services and, of course, statisticians. There is also a paucity of data on the burden of psychological stress for families or the health economic benefits created by reducing treatment. Ultimately, to ensure we are not deceived by the tangled web we could weave here, we need accurate, complete data that can be shared on an international platform. The renal tumour clinical trial groups are already on this pathway of cooperation in a spirit of openness. However, they need the continued resources to run clinical studies, even in very good prognosis subgroups, in order to generate the vital input data. Let us hope that they continue to find support for their individual and joint endeavours! The author is Vice-Chair of the SIOP Renal Tumours Study Group.—ed.