Background: Outcomes for children with relapsed and refractory neuroblastoma are dismal. The combination of irinotecan and temozolomide has activity in these patients, and its acceptable toxicity profile makes it an excellent backbone for study of new agents. We aimed to test the addition of temsirolimus or dinutuximab to irinotecan-temozolomide in patients with relapsed or refractory neuroblastoma.

Methods: For this open-label, randomised, phase 2 selection design trial of the Children's Oncology Group (COG; ANBL1221), patients had to have histological verification of neuroblastoma or ganglioneuroblastoma at diagnosis or have tumour cells in bone marrow with increased urinary catecholamine concentrations at diagnosis. Patients of any age were eligible at first designation of relapse or progression, or first designation of refractory disease, provided organ function requirements were met. Patients previously treated for refractory or relapsed disease were ineligible. Computer-based randomisation with sequence generation defined by permuted block randomisation (block size two) was used to randomly assign patients (1:1) to irinotecan and temozolomide plus either temsirolimus or dinutuximab, stratified by disease category, previous exposure to anti-GD2 antibody therapy, and tumour MYCN amplification status. Patients in both groups received oral temozolomide (100 mg/m² per dose) and intravenous irinotecan (50 mg/m² per dose) on days 1-5 of 21-day cycles. Patients in the temsirolimus group also received intravenous temsirolimus (35 mg/m² per dose) on days 1 and 8, whereas those in the dinutuximab group received intravenous dinutuximab (17·5 mg/m² per day or 25 mg/m² per day) on days 2-5 plus granulocyte macrophage colony-stimulating factor (250 μg/m² per dose) subcutaneously on days 6-12. Patients were given up to a maximum of 17 cycles of treatment. The primary endpoint was the proportion of patients achieving an objective (complete or partial) response by central review after six cycles of treatment, analysed by intention to treat. Patients, families, and those administering treatment were aware of group assignment. This study is registered with ClinicalTrials.gov, number NCT01767194, and follow-up of the initial cohort is ongoing.

Findings: Between Feb 22, 2013, and March 23, 2015, 36 patients from 27 COG member institutions were enrolled on this groupwide study. One patient was ineligible (alanine aminotransferase concentration was above the required range). Of the remaining 35 patients, 18 were randomly assigned to irinotecan-temozolomide-temsirolimus and 17 to irinotecan-temozolomide-dinutuximab. Median follow-up was 1·26 years (IQR 0·68-1·61) among all eligible participants. Of the 18 patients assigned to irinotecan-temozolomide-temsirolimus, one patient (6%; 95% CI 0·0-16·1) achieved a partial response. Of the 17 patients assigned to irinotecan-temozolomide-dinutuximab, nine (53%; 95% CI 29·2-76·7) had objective responses, including four partial responses and five complete responses. The most common grade 3 or worse adverse events in the temsirolimus group were neutropenia (eight [44%] of 18 patients), anaemia (six [33%]), thrombocytopenia (five [28%]), increased alanine aminotransferase (five [28%]), and hypokalaemia (four [22%]). One of the 17 patients assigned to the dinutuximab group refused treatment after randomisation; the most common grade 3 or worse adverse events in the remaining 16 patients evaluable for safety were pain (seven [44%] of 16), hypokalaemia (six [38%]), neutropenia (four [25%]), thrombocytopenia (four [25%]), anaemia (four [25%]), fever and infection (four [25%]), and hypoxia (four [25%]); one patient had grade 4 hypoxia related to therapy that met protocol-defined criteria for unacceptable toxicity. No deaths attributed to protocol therapy occurred.

Interpretation: Irinotecan-temozolomide-dinutuximab met protocol-defined criteria for selection as the combination meriting further study whereas irinotecan-temozolomide-temsirolimus did not. Irinotecan-temozolomide-dinutuximab shows notable anti-tumour activity in patients with relapsed or refractory neuroblastoma. Further evaluation of biomarkers in a larger cohort of patients might identify those most likely to respond to this chemoimmunotherapeutic regimen.

Funding: National Cancer Institute.