Matthay 1999.

Methods	Setting Multicentre study denoted CCG‐3891. The abbreviation CCG derives from the Children's Cancer Group that merged in 2000 with other groups to form the Children's Oncology Group United States Duration of enrollment From 1991 to 1996 Randomisation A permuted‐block design was used for the random assignment of approximately equal numbers of participants from each of 2 strata (those with and those without metastatic disease) The study design included 2 randomisations. The 2nd randomisation was similarly balanced with respect to the numbers of participants from each group of the first randomisation and non‐randomised patients who were ineligible for transplantation First randomisation: Participants without progressive disease after 2 of 5 cycles of initial chemotherapy were randomly allocated to cytotoxic therapy group 1 that consisted of myeloablative therapy and autologous bone marrow transplantation or to cytotoxic therapy group 2 that consisted of 3 cycles of continuation chemotherapy. Participants with progressive disease were non‐randomly assigned to continuation chemotherapy 2nd randomisation: Participants without progressive disease after consolidation therapy were randomly assigned to 13‐cis‐retinoic acid or to no further therapy Median follow‐up time After diagnosis: median of 43 months (range 2 to 89) for 539 participants eligible for initial chemotherapy; not reported separately for the 98 participants eligible for this review From transplantation to start of 13‐cis‐retinoic acid median of 97 days; no information on time from transplantation to 2nd randomisation for the 98 participants eligible for this review Length of follow‐up from 2nd randomisation until end of study not reported for the 98 participants eligible for this review
Participants	Eligibility criteria Newly‐diagnosed high‐risk neuroblastoma patients; high‐risk neuroblastoma was defined as: stage IV neuroblastoma; stage III disease with 1 or more of the following: amplification of the MYCN oncogene, a serum ferritin level of at least 143 ng per mL, and unfavourable histopathological findings; stage II disease with amplification of MYCN (age > 1 year); stage I or II disease with bone metastases before therapy other than surgery; and stage IV disease with MYCN amplification for < 1 year. Staging was done using the Evans staging criteria (Evans 1971). Unfavourable histopathological findings were based on the Shimada classification (Shimada 1984) 1 to 18 years of age Number of patients eligible for this review 98 participants after high‐dose chemotherapy followed by bone marrow transplantation: 50 received retinoic acid and 48 received no further therapy Age Median and range not reported for 98 participants eligible for this review Gender Not reported Stage of disease Not reported for 98 participants eligible for this review Remission status None of the 98 participants eligible for this review had progressive disease at the time of the 2nd randomisation; no further information provided Compliance with randomisation Not reported for the 2nd randomisation Previous treatment, except initial and consolidation chemotherapy Not reported Comorbidity Not reported
Interventions	All participants Initial chemotherapy for 5 cycles at 28‐day intervals. 1 cycle consisted of: cisplatin (60 mg/m2 total dose), doxorubicin (30 mg/m2 total dose), etoposide (200 mg/m2 total dose), and cyclophosphamide (2000 mg/m2 total dose). Following induction chemotherapy patients with gross residual disease received surgery and radiotherapy (number of participants not reported) First randomisation BMT arm (patients in the continuation chemotherapy arm were not eligible for this review) Carboplatin (1000 mg/m2 total dose), etoposide (640 mg/m2 total dose), melphalan (210 mg/m2 total dose), total‐body irradiation (1000 cGy), purged bone marrow harvested at end of cycle 2, and granulocyte‐macrophage colony‐stimulating factor (250 microgram/m2 per day, number of days not reported) 2nd randomisation Retinoic acid arm 50 of 98 participants after high‐dose chemotherapy followed by autologous bone marrow transplantation were randomised to the retinoic acid arm to receive 6 cycles of retinoic acid. 1 cycle consists of: 13‐Cis retinoic acid at a dose of 160 mg/mg/m2 per day for 14 consecutive days in a 28‐day cycle (14 days on, 14 days off), resulting in a cumulative dose after 28 days of 2240 mg/m2 Control arm 48 of 98 participants after high‐dose chemotherapy followed by autologous bone marrow transplantation were randomised to the no‐further‐therapy arm
Outcomes	Primary outcomes Overall survival (definition not provided, but measured from 2nd randomisation) Secondary outcomes Event‐free survival. According to Matthay 1999: "The primary end point, prespecified by the protocol, was event‐free survival calculated from the time of second randomisation. The events considered were relapse, disease progression, death from any cause, and a second neoplasm, whichever occurred first."
Notes	No competing interest reported, funding, grants, and awards received from not‐for‐profit organizations. Supported in part by the National Cancer Institute; Neil Bogart Memorial Laboraties of the T.J. Martell Foundation for Leukemia, Cancer, and AIDS Research; American Institute for Cancer Research. The authors of Matthay 1999 concluded in their follow‐up paper in 2009: "Myeloablative therapy and autologous hematopoietic cell rescue result in significantly better 5‐year EFS and OS than nonmyeloablative chemotherapy; cis‐RA given after consolidation independently results in significantly improved OS." We have questioned the results of the statistical analyses and the authors eventually initiated a revision and published an erratum (Erratum 2014) and revised their conclusion as follows: "Myeloablative therapy and autologous hematopoietic cell rescue result in significantly better 5‐year EFS than nonmyeloablative chemotherapy; neither myeloablative therapy with autologous hematopoietic cell rescue nor cis‐RA given after consolidation therapy significantly improved OS.”
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	A permuted‐block design was used for the random assignment of approximately equal numbers of participants from each of 2 strata (those with and those without metastatic disease) to transplantation or continuation chemotherapy. The 2nd randomisation was similarly balanced with respect to the numbers of participants from each group of the first randomisation and non‐randomised participants who were ineligible for transplantation
Allocation concealment (selection bias)	Unclear risk	Allocation concealment was not described
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Blinding of participants, physicians and nurses was not reported
Blinding of outcome assessment (detection bias) All outcomes	Low risk	The study committee and investigators were unaware of participants' treatment assignments, and the study was monitored by an independent committee according to a group sequential monitoring plan
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	It was not reported if all reported outcomes for all participants were assessed
Selective reporting (reporting bias)	Low risk	Reporting was in agreement with the protocol with regard to the outcome measures. We were concerned with the possibility that data for the same participants were included in several publications. However, we did not include duplicate data in the review
Other bias	Unclear risk	Analysis was conducted using the data of the randomised participants, which should be consistent with the ITT principle. It was unclear if all 98 participants received the treatment to which they were randomised. Also, not all participants treated with transplantation in the first randomisation and without progressive disease afterwards were included in the 2nd randomisation. It is unclear how many eligible patients did not undergo the 2nd randomisation. The consequences of 2 randomisations in 1 study for the risk of bias are unclear.

AIDS: acquired immunodeficiency syndrome; BMT: bone marrow transplantation; cGy: centi‐Gray; CR: complete response; N: number.