In total, EpSSG RMS 2005 included 1,674 patients aged < 18 years at diagnosis. Of 1,270 IRS group III patients, 613 (48%) had three-dimensional measurements available at diagnosis and after three cycles of neoadjuvant chemotherapy and were included for this analysis (Fig. 1).

Consort diagram. EpSSG RMS 2005 European paediatric Soft tissue sarcoma Study Group Rhabdomyosarcoma 2005, IRS group Intergroup Rhabdomyosarcoma Studies post-surgical group

When comparing IRS group III patients included in the analysis to those excluded (i.e. without three-dimensional measurements available at any of the timepoints), patients with all measurements available after the third cycle were younger (P < 0.001), had more tumours at unfavourable sites (P < 0.001), and had larger tumours (P < 0.001) (Supplementary Material 3).

Of the 613 patients, 267 (44%) were female and the median age at diagnosis was 4.2 years (range 6 days – 17.9 years). Most patients, 426 of 613 patients (69%), had a tumour with embryonal histology; and, of the patients of whom the fusion status was known (82%), 398 of 505 patients (79%) had a tumour with negative PAX-FOXO1 fusion status (Table 1).

At radiological tumour response evaluation after the third cycle of chemotherapy, 48 of 613 patients (7.8%) achieved complete response. Most patients, 371 patients (60.5%), had a partial response, 139 patients (22.7%) a minor partial response, 48 patients (7.8%) stable disease, and seven patients (1.1%) showed progressive disease. Supplementary Material 4 gives a detailed overview of the seven patients with progressive disease in this study. Younger age (P = 0.0089), tumour histology (P = 0.039), unfavourable tumour site (P < 0.001), large tumour size (P < 0.001), and clinical risk groups (P = 0.0075) were significantly associated with poorer tumour volume response (Table 1). PAX-FOXO1-fusion status, gender, tumour invasiveness, and loco-regional lymph node involvement were not significantly associated with any of the different response groups.

Furthermore, we observed that 90 of the 613 patients included (15%) scored a different response group when the tumour response was calculated based on the radiological measurements available, compared to the response group that was reported in the EpSSG central database, as is shown in Supplementary Material 5.

The EpSSG RMS 2005 protocol advised a switch to second-line chemotherapy when less than minor partial response was reached after three cycles of neoadjuvant chemotherapy. However, in 22 of 48 patients (46%) with stable disease and in one of seven patients (14%) with progressive disease, no changes in chemotherapy regimen were made after response evaluation. However, for the patients with calculated progressive disease, the reported response group within the EpSSG database was stable disease. Furthermore, two of 371 patients (< 1%) with partial response and five of 139 patients (4%) with minor partial response did switch to second-line chemotherapy, aiming to increase response to allow for the application of adequate local therapy.

Radiotherapy was performed in 35 of 48 patients (73%) with complete response, 304 of 371 patients (82%) with partial response, 117 of 139 patients (84%) with minor partial response, 34 of 48 patients (71%) with stable disease, and three of seven patients (43%) with progressive disease. Reasons not to perform radiotherapy were the decision of the centre (33%), a very young age (32%), early progressive disease (13%), parental refusal (6%), amputation before radiotherapy (3%), and optional according to risk group or site (13%).

Delayed surgery of primary tumour or lymph nodes was performed in five of 48 patients (10%) with complete response, 221 of 371 patients (60%) with partial response, 92 of 139 patients (66%) with minor partial response, 33 of 48 patients (69%) with stable disease, and two of seven patients (29%) with progressive disease. Of 353 patients who received delayed surgery, a complete resection (R0) could be achieved in 220 patients (62%). In patients with radiological complete response, this was achieved in five of five patients (100%), with partial response in 146 of 221 patients (66%), with minor partial response in 50 of 92 patients (54%), with stable disease in 18 of 33 patients (55%) and with progressive disease in one of two patients (50%).

Of all 613 patients, 267 patients (44%) received both radiotherapy and delayed surgery, 86 patients (14%) received delayed surgery only, 226 patients (37%) received radiotherapy only, and 34 patients (5%) received no local therapy.

Overall, 219 of 613 patients (36%) experienced an event, and 158 (26%) died. Three hundred and ninety-four patients (64%) are alive in first complete response at last follow-up. The median follow-up time for alive patients was 117 months from the start of chemotherapy (interquartile range 90.0–144.7). Of the 219 patients who experienced an event, 92 patients (42%) had a local relapse, 12 (5%) had a regional lymph node relapse, nine (4%) had a lymph node and local relapse, 50 (23%) had a metastatic relapse with or without local or nodal relapse, 39 (18%) had progressive disease with or without nodal or metastatic relapse, 11 (5%) developed a second tumour, four (1%) died of toxicity (i.e. one of post-surgery septic shock, two of cardiac arrest in the context of a septic process, and one suspension of the protocol due to meningoencephalitis), and two (< 1%) died of other non-oncological reasons.

The estimated 5-year failure-free survival and overall survival for all 613 included patients were 66.5% (95% confidence interval (CI) 62.6%−70.1%) and 77.0% (95% CI 73.4%−80.1%), respectively. No significant differences were observed between the different response groups for failure-free survival or overall survival, P = 0.11 and P = 0.12, respectively (Figs. 2 and 3).

Kaplan–Meier estimates of failure-free survival based on early tumour volume response of 606 IRS group III patients with complete response, partial response, minor partial response, and stable disease included in the EpSSG RMS 2005 protocol. 5-yr 5-year, CI confidence interval, CR complete response, EpSSG European paediatric Soft tissue sarcoma Study Group, FFS failure-free survival, IRS Intergroup Rhabdomyosarcoma Studies, mPR minor partial response, PR partial response, SD stable disease

Kaplan–Meier estimates of overall survival based on early tumour volume response of 606 IRS group III patients with complete response, partial response, minor partial response, and stable disease included in the EpSSG RMS 2005 protocol. 5-yr 5-year, CI confidence interval, CR complete response, EpSSG European paediatric Soft tissue sarcoma Study Group, IRS Intergroup Rhabdomyosarcoma Studies, mPR minor partial response, OS overall survival, PR partial response, SD stable disease

Univariate Cox proportional hazards model did not show a significant difference for failure-free survival and overall survival between the different response groups. Hazard ratios (HR) for partial response, minor partial response, and stable disease were 1.02 (95% CI 0.61–1.72), 1.13 (95% CI 0.64–1.98), and 1.75 (95% CI 0.93–3.29) (P = 0.11), respectively, for failure-free survival and 1.33 (95% CI 0.67–2.65), 1.63 (95% CI 0.79–3.37), and 2.23 (95% CI 1.00–4.96) (P = 0.13), respectively, for overall survival. The multivariate Cox regression analysis identified site (unfavourable; HR 1.49, 95% CI 1.04–2.13, P = 0.03), histology (unfavourable; HR 1.49, 95% CI 1.11–1.99, P = 0.01), and loco-regional lymph node involvement (N1; HR 1.47, 95% CI 1.07–2.03, P = 0.02) as independent prognostic factors for failure-free survival. For overall survival, age at diagnosis (≥ 10 years; HR 1.49, 95% CI 1.00–2.20, P = 0.05), site (unfavourable; HR 3.26, 95% CI 1.87–5.67, P < 0.001), and loco-regional lymph node involvement (N1; HR 2.33, 95% CI 1.66–3.26, P < 0.001) were identified as independent prognostic factors as well. For both failure-free survival and overall survival, tumour volume response was not identified as an independent prognostic factor.