Outcomes of patients with Wilms' tumour stage III due to positive resection margins only: An analysis of patients treated on the SIOP‐WT‐2001 protocol in the UK‐CCLG and GPOH studies

Stage III Wilms' tumour (WT) represents a heterogeneous group which includes different criteria, but all stage III patients are treated according to the same study regiment. The aim of the study was to retrospectively analyse outcomes in patients with stage III due to positive resection margins (RM) only, sub‐grouped in RM with viable (RM‐v) and nonviable (RM‐nv) tumour. Patients were treated pre‐ and postoperatively according to the SIOP‐WT‐2001 protocol in the UK‐CCLG and GPOH WT trials and studies (2001‐2020). There were 197 patients, including 134 with localised, abdominal stage III and 63 with overall stage IV, but abdominal stage III. Stage III due to RM‐v had 126 patients, and due to RM‐nv 71 patients. The overall 5‐year local‐relapse‐free survival (RFS), event‐free (EFS) and overall survival (OS) estimates for all patients with abdominal stage III RM were 95.7% (±SE1.5%), 85.1 (±SE2.6%) and 90.3% (±SE2.2%), respectively. Patients with stage III RM‐nv had significantly better RFS and EFS than patients with RM‐v (P = .027 and P = .003, respectively). A multivariate analysis showed that RM‐v remained a significant factor for EFS when adjusted for age, presence of metastasis at diagnosis, histological risk group and overall stage in Cox regression analysis (P = .006). Patients with stage III due to RM‐nv only exhibited no local recurrence and have a significantly better RFS and EFS than patients with RM‐v. The results suggest that exclusion of RM‐nv as a stage III criterion in the UMBRELLA staging system and consequent treatment reduction is warranted.

für Pädiatrische Onkologie und Hämatologie and "Deutsche Krebshilfe", Grant/Award Number: 50-2709-Gr2 RFS and EFS than patients with RM-v (P = .027 and P = .003, respectively). A multivariate analysis showed that RM-v remained a significant factor for EFS when adjusted for age, presence of metastasis at diagnosis, histological risk group and overall stage in Cox regression analysis (P = .006). Patients with stage III due to RM-nv only exhibited no local recurrence and have a significantly better RFS and EFS than patients with RM-v. The results suggest that exclusion of RM-nv as a stage III criterion in the UMBRELLA staging system and consequent treatment reduction is warranted.

| INTRODUCTION
International Society of Paediatric Oncology (SIOP) trials/studies include preoperative chemotherapy as the first line of treatment of patients with Wilms' tumour (WT) and its purpose is to reduce surgical complications, down-stage tumours to reduce postoperative treatment and assess the histological response of the primary tumour and pulmonary metastases to determine postoperative treatment. With overall survival rates above 90%, the effort is now focused on identifying positive prognostic factors that can warrant therapy reduction to avoid long-term side effects in these young patients. A possible route is through identifying certain histological features which could create subgroups of WT requiring less treatment but maintaining the excellent outcomes. 1,2 Another is through staging, where certain criteria are revisited to assess whether they could be fine-tuned or changed.
In the early SIOP trials, the presence of chemotherapy-induced changes (CIC) in the renal sinus and its vessels, and in the perirenal fat was regarded as evidence of tumour expansion for staging purposes. 3,4 The SIOP 93-01 study showed that these changes could be ignored for staging, 5 and in the SIOP-WT-2001 trial and study the staging criteria were redefined accordingly, so only the finding of viable tumour in these sites was used as a criterion for stage II. 6 However, the finding of CIC at the resection margins or in the lymph nodes was kept as a criterion for stage III.
The aim of this retrospective study was to analyse the outcomes of patients with abdominal stage III due to positive resection margins alone, and to compare outcomes of those with stage III due to the finding of viable tumour (RM-v) vs CIC at the resection margins (RM-nv), to establish whether the latter could be removed as the criterion for stage III. The rationale for the study was the SIOP experience with excellent outcomes of patients in whom CIC was safely disregarded in other situations: (a) its presence in the renal sinus and its vessels, and in the perirenal fat was not a criterion for upstaging tumour to stage II 6 ; (b) its presence in stage III completely necrotic WT (low-risk, LR-WT) was not indication for radiotherapy 7 ; (c) in stage IV where complete remission of metastasis was achieved after preoperative chemotherapy, radiotherapy was not indicated for the metastatic site for LR-WT and IR-WT and (d) in patients with nonviable pulmonary metastases, radiotherapy was not indicated after complete surgical metastasectomy except for HR-WT. 8 2 | PATIENTS AND METHODS

| Histological assessment
WT were classified and staged according to the SIOP-WT-2001 classification and staging criteria. 6 The criteria for stage III are listed in

| Treatment
All patients were treated according to the SIOP-WT-2001 Trial protocol (Appendix S1). All patients with IR-WT and HR-WTs localised stage III and metastatic stage III were recommended to be treated with flank irradiation.
Follow-up information was obtained from the Study databases containing information documented in case report forms specific to each phase of diagnosis, treatment and follow up and received regularly from the participating centres.

| Patient outcomes
The A multivariate analysis showed that RM-v remained a significant factor for EFS when adjusted for age, presence of metastasis at diagnosis, histological risk group and overall stage in Cox regression analysis (P = .006, Table 3). Multivariate analysis was not reliably possible for RFS due to a small number of local/combined relapses (8 patients).  -v  Overall stage  III, IV  III, IV  III, IV  III, IV  III, IV  III,  and some dependent on the tumour history (preoperative tumour rupture). Therefore, it is important to analyse and stratify them more carefully as they may not require the same treatment. Although stage III criteria and terminology differ between the studies, 15 a significant proportion of stage III were due to either positive resection margins alone (26.1% in the AIEOP study), 16 or combined with other stage III criteria (48% in the NWTS-5 study, and 32.1% in the COG AREN0532 study, respectively). 17,18 In our cohort, positive resection margins as the only criterion comprised 32.5% of stage III, but included localised and metastatic stage III. Other studies have used the NWTS/COG staging criteria, which are different from the SIOP staging criteria, [15][16][17][18] or defined subgroups differently ("microscopic stage III," "macroscopic stage III"), 16 or "lumped" both cases treated with primary surgery and preoperative chemotherapy together. 16,17 The advantage of our study is that all included patients were treated according to the standardised SIOP-WT-2001 protocol which included preoperative chemotherapy and postoperative treatment.
Previous studies have analysed outcomes of patients with stage III WT, looking for subgroups with better or worse outcomes, and considering whether they might need different treatment. 12,[16][17][18] However, none of these studies analysed separately outcomes for patients with stage III due to RM-v vs RM-nv, although they included patients treated with preoperative chemotherapy. In the AREN0532 study, 17  In the AIEOP study, patients with stage III due to positive lymph nodes had the worst outcomes-the 4-year disease-free survival was 73% ± 7% vs 98% ± 2% for patients with negative lymph nodes (P = .001). 16 Similarly, the multivariate analysis of stage III patients in the NWTS-5 study showed that patients with lymph node metastases, and patients with "microscopic residual disease" had worse 8-year EFS in comparison to other stage III patients (P = .005 and .007, respectively). 18 The same studies showed that patients with negative "microscopic residual disease" had excellent 8-year overall survival of